{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.92295014421096,
  "eval_steps": 500,
  "global_step": 7000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002746875429199286,
      "grad_norm": 0.1937730759382248,
      "learning_rate": 0.0,
      "loss": 1.8294,
      "step": 1
    },
    {
      "epoch": 0.0005493750858398572,
      "grad_norm": 0.15454284846782684,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.7116,
      "step": 2
    },
    {
      "epoch": 0.0008240626287597857,
      "grad_norm": 0.14069359004497528,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.3158,
      "step": 3
    },
    {
      "epoch": 0.0010987501716797144,
      "grad_norm": 0.13770702481269836,
      "learning_rate": 1.2e-05,
      "loss": 1.2236,
      "step": 4
    },
    {
      "epoch": 0.0013734377145996428,
      "grad_norm": 0.15386351943016052,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.3798,
      "step": 5
    },
    {
      "epoch": 0.0016481252575195715,
      "grad_norm": 0.21265602111816406,
      "learning_rate": 2e-05,
      "loss": 1.8568,
      "step": 6
    },
    {
      "epoch": 0.0019228128004395001,
      "grad_norm": 0.20548376441001892,
      "learning_rate": 1.999725161467638e-05,
      "loss": 1.5556,
      "step": 7
    },
    {
      "epoch": 0.002197500343359429,
      "grad_norm": 0.2386006861925125,
      "learning_rate": 1.999450322935276e-05,
      "loss": 1.7897,
      "step": 8
    },
    {
      "epoch": 0.002472187886279357,
      "grad_norm": 0.17586326599121094,
      "learning_rate": 1.9991754844029133e-05,
      "loss": 1.3609,
      "step": 9
    },
    {
      "epoch": 0.0027468754291992857,
      "grad_norm": 0.19550961256027222,
      "learning_rate": 1.998900645870551e-05,
      "loss": 1.4681,
      "step": 10
    },
    {
      "epoch": 0.0030215629721192143,
      "grad_norm": 0.209793359041214,
      "learning_rate": 1.998625807338189e-05,
      "loss": 1.6005,
      "step": 11
    },
    {
      "epoch": 0.003296250515039143,
      "grad_norm": 0.2240128368139267,
      "learning_rate": 1.9983509688058267e-05,
      "loss": 1.4801,
      "step": 12
    },
    {
      "epoch": 0.0035709380579590716,
      "grad_norm": 0.22508294880390167,
      "learning_rate": 1.9980761302734648e-05,
      "loss": 1.3613,
      "step": 13
    },
    {
      "epoch": 0.0038456256008790003,
      "grad_norm": 0.1746370941400528,
      "learning_rate": 1.997801291741102e-05,
      "loss": 1.2613,
      "step": 14
    },
    {
      "epoch": 0.004120313143798929,
      "grad_norm": 0.2583545744419098,
      "learning_rate": 1.9975264532087398e-05,
      "loss": 1.4942,
      "step": 15
    },
    {
      "epoch": 0.004395000686718858,
      "grad_norm": 0.18254704773426056,
      "learning_rate": 1.997251614676378e-05,
      "loss": 1.2358,
      "step": 16
    },
    {
      "epoch": 0.004669688229638786,
      "grad_norm": 0.23128043115139008,
      "learning_rate": 1.9969767761440156e-05,
      "loss": 1.4434,
      "step": 17
    },
    {
      "epoch": 0.004944375772558714,
      "grad_norm": 0.1845933049917221,
      "learning_rate": 1.9967019376116533e-05,
      "loss": 1.2212,
      "step": 18
    },
    {
      "epoch": 0.005219063315478643,
      "grad_norm": 0.17907685041427612,
      "learning_rate": 1.996427099079291e-05,
      "loss": 1.2381,
      "step": 19
    },
    {
      "epoch": 0.005493750858398571,
      "grad_norm": 0.19670948386192322,
      "learning_rate": 1.9961522605469287e-05,
      "loss": 1.2741,
      "step": 20
    },
    {
      "epoch": 0.0057684384013185,
      "grad_norm": 0.19699521362781525,
      "learning_rate": 1.9958774220145667e-05,
      "loss": 1.6691,
      "step": 21
    },
    {
      "epoch": 0.006043125944238429,
      "grad_norm": 0.18602783977985382,
      "learning_rate": 1.9956025834822044e-05,
      "loss": 1.3378,
      "step": 22
    },
    {
      "epoch": 0.006317813487158357,
      "grad_norm": 0.1472821831703186,
      "learning_rate": 1.995327744949842e-05,
      "loss": 1.4843,
      "step": 23
    },
    {
      "epoch": 0.006592501030078286,
      "grad_norm": 0.13644804060459137,
      "learning_rate": 1.9950529064174798e-05,
      "loss": 1.4285,
      "step": 24
    },
    {
      "epoch": 0.006867188572998215,
      "grad_norm": 0.13536669313907623,
      "learning_rate": 1.9947780678851175e-05,
      "loss": 1.5478,
      "step": 25
    },
    {
      "epoch": 0.007141876115918143,
      "grad_norm": 0.11944937705993652,
      "learning_rate": 1.9945032293527552e-05,
      "loss": 1.4516,
      "step": 26
    },
    {
      "epoch": 0.007416563658838072,
      "grad_norm": 0.10020400583744049,
      "learning_rate": 1.9942283908203933e-05,
      "loss": 1.3101,
      "step": 27
    },
    {
      "epoch": 0.007691251201758001,
      "grad_norm": 0.09920955449342728,
      "learning_rate": 1.993953552288031e-05,
      "loss": 1.548,
      "step": 28
    },
    {
      "epoch": 0.00796593874467793,
      "grad_norm": 0.09061945229768753,
      "learning_rate": 1.9936787137556687e-05,
      "loss": 1.4037,
      "step": 29
    },
    {
      "epoch": 0.008240626287597858,
      "grad_norm": 0.10895679891109467,
      "learning_rate": 1.9934038752233064e-05,
      "loss": 1.3453,
      "step": 30
    },
    {
      "epoch": 0.008515313830517787,
      "grad_norm": 0.09087394922971725,
      "learning_rate": 1.993129036690944e-05,
      "loss": 0.8304,
      "step": 31
    },
    {
      "epoch": 0.008790001373437715,
      "grad_norm": 0.10762802511453629,
      "learning_rate": 1.992854198158582e-05,
      "loss": 1.1338,
      "step": 32
    },
    {
      "epoch": 0.009064688916357644,
      "grad_norm": 0.0972258597612381,
      "learning_rate": 1.9925793596262198e-05,
      "loss": 1.346,
      "step": 33
    },
    {
      "epoch": 0.009339376459277572,
      "grad_norm": 0.0915093943476677,
      "learning_rate": 1.9923045210938575e-05,
      "loss": 1.2229,
      "step": 34
    },
    {
      "epoch": 0.009614064002197501,
      "grad_norm": 0.1097780242562294,
      "learning_rate": 1.9920296825614952e-05,
      "loss": 1.3209,
      "step": 35
    },
    {
      "epoch": 0.009888751545117428,
      "grad_norm": 0.09500670433044434,
      "learning_rate": 1.991754844029133e-05,
      "loss": 1.1753,
      "step": 36
    },
    {
      "epoch": 0.010163439088037357,
      "grad_norm": 0.11108588427305222,
      "learning_rate": 1.991480005496771e-05,
      "loss": 1.4102,
      "step": 37
    },
    {
      "epoch": 0.010438126630957285,
      "grad_norm": 0.08854890614748001,
      "learning_rate": 1.9912051669644087e-05,
      "loss": 1.1796,
      "step": 38
    },
    {
      "epoch": 0.010712814173877214,
      "grad_norm": 0.09179817140102386,
      "learning_rate": 1.9909303284320464e-05,
      "loss": 1.0448,
      "step": 39
    },
    {
      "epoch": 0.010987501716797143,
      "grad_norm": 0.08385608345270157,
      "learning_rate": 1.990655489899684e-05,
      "loss": 1.4231,
      "step": 40
    },
    {
      "epoch": 0.011262189259717071,
      "grad_norm": 0.08239055424928665,
      "learning_rate": 1.9903806513673218e-05,
      "loss": 1.0382,
      "step": 41
    },
    {
      "epoch": 0.011536876802637,
      "grad_norm": 0.0889146700501442,
      "learning_rate": 1.9901058128349595e-05,
      "loss": 1.1364,
      "step": 42
    },
    {
      "epoch": 0.011811564345556929,
      "grad_norm": 0.10676855593919754,
      "learning_rate": 1.9898309743025975e-05,
      "loss": 1.5221,
      "step": 43
    },
    {
      "epoch": 0.012086251888476857,
      "grad_norm": 0.09268155694007874,
      "learning_rate": 1.9895561357702352e-05,
      "loss": 1.3128,
      "step": 44
    },
    {
      "epoch": 0.012360939431396786,
      "grad_norm": 0.09370176494121552,
      "learning_rate": 1.989281297237873e-05,
      "loss": 1.4961,
      "step": 45
    },
    {
      "epoch": 0.012635626974316715,
      "grad_norm": 0.083768330514431,
      "learning_rate": 1.9890064587055106e-05,
      "loss": 1.105,
      "step": 46
    },
    {
      "epoch": 0.012910314517236643,
      "grad_norm": 0.10379151999950409,
      "learning_rate": 1.9887316201731483e-05,
      "loss": 0.9532,
      "step": 47
    },
    {
      "epoch": 0.013185002060156572,
      "grad_norm": 0.0819801390171051,
      "learning_rate": 1.9884567816407864e-05,
      "loss": 1.1663,
      "step": 48
    },
    {
      "epoch": 0.0134596896030765,
      "grad_norm": 0.07967396825551987,
      "learning_rate": 1.988181943108424e-05,
      "loss": 1.3103,
      "step": 49
    },
    {
      "epoch": 0.01373437714599643,
      "grad_norm": 0.08881644904613495,
      "learning_rate": 1.9879071045760618e-05,
      "loss": 1.2531,
      "step": 50
    },
    {
      "epoch": 0.014009064688916358,
      "grad_norm": 0.08425169438123703,
      "learning_rate": 1.9876322660436995e-05,
      "loss": 1.1376,
      "step": 51
    },
    {
      "epoch": 0.014283752231836287,
      "grad_norm": 0.09973838925361633,
      "learning_rate": 1.9873574275113372e-05,
      "loss": 1.019,
      "step": 52
    },
    {
      "epoch": 0.014558439774756215,
      "grad_norm": 0.09437184035778046,
      "learning_rate": 1.9870825889789752e-05,
      "loss": 1.1693,
      "step": 53
    },
    {
      "epoch": 0.014833127317676144,
      "grad_norm": 0.08502259105443954,
      "learning_rate": 1.986807750446613e-05,
      "loss": 1.3755,
      "step": 54
    },
    {
      "epoch": 0.015107814860596072,
      "grad_norm": 0.07461732625961304,
      "learning_rate": 1.9865329119142506e-05,
      "loss": 1.4107,
      "step": 55
    },
    {
      "epoch": 0.015382502403516001,
      "grad_norm": 0.07736615091562271,
      "learning_rate": 1.9862580733818883e-05,
      "loss": 1.2233,
      "step": 56
    },
    {
      "epoch": 0.015657189946435928,
      "grad_norm": 0.09092046320438385,
      "learning_rate": 1.985983234849526e-05,
      "loss": 1.1681,
      "step": 57
    },
    {
      "epoch": 0.01593187748935586,
      "grad_norm": 0.07622669637203217,
      "learning_rate": 1.9857083963171637e-05,
      "loss": 1.4459,
      "step": 58
    },
    {
      "epoch": 0.016206565032275785,
      "grad_norm": 0.1118270754814148,
      "learning_rate": 1.9854335577848018e-05,
      "loss": 1.1701,
      "step": 59
    },
    {
      "epoch": 0.016481252575195716,
      "grad_norm": 0.1012958288192749,
      "learning_rate": 1.9851587192524395e-05,
      "loss": 1.1292,
      "step": 60
    },
    {
      "epoch": 0.016755940118115643,
      "grad_norm": 0.09912410378456116,
      "learning_rate": 1.984883880720077e-05,
      "loss": 1.4852,
      "step": 61
    },
    {
      "epoch": 0.017030627661035573,
      "grad_norm": 0.08557552844285965,
      "learning_rate": 1.984609042187715e-05,
      "loss": 1.3164,
      "step": 62
    },
    {
      "epoch": 0.0173053152039555,
      "grad_norm": 0.0909140557050705,
      "learning_rate": 1.9843342036553526e-05,
      "loss": 0.8671,
      "step": 63
    },
    {
      "epoch": 0.01758000274687543,
      "grad_norm": 0.08416318148374557,
      "learning_rate": 1.9840593651229906e-05,
      "loss": 1.4408,
      "step": 64
    },
    {
      "epoch": 0.017854690289795357,
      "grad_norm": 0.07851261645555496,
      "learning_rate": 1.9837845265906283e-05,
      "loss": 1.254,
      "step": 65
    },
    {
      "epoch": 0.018129377832715288,
      "grad_norm": 0.10467074811458588,
      "learning_rate": 1.9835096880582657e-05,
      "loss": 1.0836,
      "step": 66
    },
    {
      "epoch": 0.018404065375635215,
      "grad_norm": 0.09777582436800003,
      "learning_rate": 1.9832348495259037e-05,
      "loss": 0.9239,
      "step": 67
    },
    {
      "epoch": 0.018678752918555145,
      "grad_norm": 0.09177341312170029,
      "learning_rate": 1.9829600109935414e-05,
      "loss": 1.2187,
      "step": 68
    },
    {
      "epoch": 0.018953440461475072,
      "grad_norm": 0.08834896981716156,
      "learning_rate": 1.9826851724611795e-05,
      "loss": 1.0094,
      "step": 69
    },
    {
      "epoch": 0.019228128004395002,
      "grad_norm": 0.0814296156167984,
      "learning_rate": 1.9824103339288168e-05,
      "loss": 1.099,
      "step": 70
    },
    {
      "epoch": 0.01950281554731493,
      "grad_norm": 0.09274858981370926,
      "learning_rate": 1.9821354953964545e-05,
      "loss": 1.1168,
      "step": 71
    },
    {
      "epoch": 0.019777503090234856,
      "grad_norm": 0.07761111855506897,
      "learning_rate": 1.9818606568640926e-05,
      "loss": 1.006,
      "step": 72
    },
    {
      "epoch": 0.020052190633154787,
      "grad_norm": 0.08699138462543488,
      "learning_rate": 1.9815858183317303e-05,
      "loss": 1.239,
      "step": 73
    },
    {
      "epoch": 0.020326878176074713,
      "grad_norm": 0.08698683232069016,
      "learning_rate": 1.981310979799368e-05,
      "loss": 1.3336,
      "step": 74
    },
    {
      "epoch": 0.020601565718994644,
      "grad_norm": 0.08187371492385864,
      "learning_rate": 1.9810361412670057e-05,
      "loss": 0.9883,
      "step": 75
    },
    {
      "epoch": 0.02087625326191457,
      "grad_norm": 0.07891972362995148,
      "learning_rate": 1.9807613027346434e-05,
      "loss": 1.2826,
      "step": 76
    },
    {
      "epoch": 0.0211509408048345,
      "grad_norm": 0.07186971604824066,
      "learning_rate": 1.9804864642022814e-05,
      "loss": 1.0775,
      "step": 77
    },
    {
      "epoch": 0.021425628347754428,
      "grad_norm": 0.07952643930912018,
      "learning_rate": 1.980211625669919e-05,
      "loss": 1.236,
      "step": 78
    },
    {
      "epoch": 0.02170031589067436,
      "grad_norm": 0.09609956294298172,
      "learning_rate": 1.9799367871375568e-05,
      "loss": 1.4409,
      "step": 79
    },
    {
      "epoch": 0.021975003433594285,
      "grad_norm": 0.07306283712387085,
      "learning_rate": 1.9796619486051945e-05,
      "loss": 0.9703,
      "step": 80
    },
    {
      "epoch": 0.022249690976514216,
      "grad_norm": 0.09465730935335159,
      "learning_rate": 1.9793871100728322e-05,
      "loss": 1.3007,
      "step": 81
    },
    {
      "epoch": 0.022524378519434143,
      "grad_norm": 0.08982721716165543,
      "learning_rate": 1.97911227154047e-05,
      "loss": 1.182,
      "step": 82
    },
    {
      "epoch": 0.022799066062354073,
      "grad_norm": 0.08521714061498642,
      "learning_rate": 1.978837433008108e-05,
      "loss": 1.2381,
      "step": 83
    },
    {
      "epoch": 0.023073753605274,
      "grad_norm": 0.11300639808177948,
      "learning_rate": 1.9785625944757457e-05,
      "loss": 1.2941,
      "step": 84
    },
    {
      "epoch": 0.02334844114819393,
      "grad_norm": 0.09087110310792923,
      "learning_rate": 1.9782877559433834e-05,
      "loss": 1.1269,
      "step": 85
    },
    {
      "epoch": 0.023623128691113857,
      "grad_norm": 0.08665549755096436,
      "learning_rate": 1.978012917411021e-05,
      "loss": 1.2973,
      "step": 86
    },
    {
      "epoch": 0.023897816234033788,
      "grad_norm": 0.06901514530181885,
      "learning_rate": 1.9777380788786588e-05,
      "loss": 1.2491,
      "step": 87
    },
    {
      "epoch": 0.024172503776953715,
      "grad_norm": 0.11828906834125519,
      "learning_rate": 1.9774632403462968e-05,
      "loss": 1.2151,
      "step": 88
    },
    {
      "epoch": 0.024447191319873645,
      "grad_norm": 0.09510165452957153,
      "learning_rate": 1.9771884018139345e-05,
      "loss": 1.0363,
      "step": 89
    },
    {
      "epoch": 0.024721878862793572,
      "grad_norm": 0.21834878623485565,
      "learning_rate": 1.9769135632815722e-05,
      "loss": 1.1662,
      "step": 90
    },
    {
      "epoch": 0.024996566405713502,
      "grad_norm": 0.110747791826725,
      "learning_rate": 1.97663872474921e-05,
      "loss": 1.0725,
      "step": 91
    },
    {
      "epoch": 0.02527125394863343,
      "grad_norm": 0.09425097703933716,
      "learning_rate": 1.9763638862168476e-05,
      "loss": 1.1552,
      "step": 92
    },
    {
      "epoch": 0.02554594149155336,
      "grad_norm": 0.08260218054056168,
      "learning_rate": 1.9760890476844857e-05,
      "loss": 0.854,
      "step": 93
    },
    {
      "epoch": 0.025820629034473287,
      "grad_norm": 0.10150811821222305,
      "learning_rate": 1.9758142091521234e-05,
      "loss": 1.1776,
      "step": 94
    },
    {
      "epoch": 0.026095316577393217,
      "grad_norm": 0.09899923950433731,
      "learning_rate": 1.975539370619761e-05,
      "loss": 1.1015,
      "step": 95
    },
    {
      "epoch": 0.026370004120313144,
      "grad_norm": 0.09975653886795044,
      "learning_rate": 1.9752645320873988e-05,
      "loss": 0.9539,
      "step": 96
    },
    {
      "epoch": 0.02664469166323307,
      "grad_norm": 0.0907176062464714,
      "learning_rate": 1.9749896935550365e-05,
      "loss": 0.9953,
      "step": 97
    },
    {
      "epoch": 0.026919379206153,
      "grad_norm": 0.07899327576160431,
      "learning_rate": 1.9747148550226742e-05,
      "loss": 0.8356,
      "step": 98
    },
    {
      "epoch": 0.027194066749072928,
      "grad_norm": 0.0981535017490387,
      "learning_rate": 1.9744400164903122e-05,
      "loss": 1.238,
      "step": 99
    },
    {
      "epoch": 0.02746875429199286,
      "grad_norm": 0.1263914406299591,
      "learning_rate": 1.97416517795795e-05,
      "loss": 1.1938,
      "step": 100
    },
    {
      "epoch": 0.027743441834912785,
      "grad_norm": 0.08866863697767258,
      "learning_rate": 1.9738903394255876e-05,
      "loss": 1.0056,
      "step": 101
    },
    {
      "epoch": 0.028018129377832716,
      "grad_norm": 0.09463110566139221,
      "learning_rate": 1.9736155008932253e-05,
      "loss": 1.0881,
      "step": 102
    },
    {
      "epoch": 0.028292816920752643,
      "grad_norm": 0.07883048802614212,
      "learning_rate": 1.973340662360863e-05,
      "loss": 0.7137,
      "step": 103
    },
    {
      "epoch": 0.028567504463672573,
      "grad_norm": 0.08269808441400528,
      "learning_rate": 1.973065823828501e-05,
      "loss": 1.0279,
      "step": 104
    },
    {
      "epoch": 0.0288421920065925,
      "grad_norm": 0.10365591198205948,
      "learning_rate": 1.9727909852961388e-05,
      "loss": 1.2886,
      "step": 105
    },
    {
      "epoch": 0.02911687954951243,
      "grad_norm": 0.09271346777677536,
      "learning_rate": 1.9725161467637765e-05,
      "loss": 0.9809,
      "step": 106
    },
    {
      "epoch": 0.029391567092432357,
      "grad_norm": 0.10162105411291122,
      "learning_rate": 1.9722413082314142e-05,
      "loss": 0.9802,
      "step": 107
    },
    {
      "epoch": 0.029666254635352288,
      "grad_norm": 0.10606992989778519,
      "learning_rate": 1.971966469699052e-05,
      "loss": 1.1599,
      "step": 108
    },
    {
      "epoch": 0.029940942178272215,
      "grad_norm": 0.12414596974849701,
      "learning_rate": 1.97169163116669e-05,
      "loss": 1.0514,
      "step": 109
    },
    {
      "epoch": 0.030215629721192145,
      "grad_norm": 0.11226197332143784,
      "learning_rate": 1.9714167926343276e-05,
      "loss": 1.231,
      "step": 110
    },
    {
      "epoch": 0.030490317264112072,
      "grad_norm": 0.10513854771852493,
      "learning_rate": 1.9711419541019653e-05,
      "loss": 0.998,
      "step": 111
    },
    {
      "epoch": 0.030765004807032002,
      "grad_norm": 0.08851984143257141,
      "learning_rate": 1.970867115569603e-05,
      "loss": 0.9635,
      "step": 112
    },
    {
      "epoch": 0.03103969234995193,
      "grad_norm": 0.09589681029319763,
      "learning_rate": 1.9705922770372407e-05,
      "loss": 1.1727,
      "step": 113
    },
    {
      "epoch": 0.031314379892871856,
      "grad_norm": 0.10749652236700058,
      "learning_rate": 1.9703174385048784e-05,
      "loss": 1.0073,
      "step": 114
    },
    {
      "epoch": 0.03158906743579179,
      "grad_norm": 0.10644505172967911,
      "learning_rate": 1.9700425999725165e-05,
      "loss": 1.1448,
      "step": 115
    },
    {
      "epoch": 0.03186375497871172,
      "grad_norm": 0.09398119151592255,
      "learning_rate": 1.969767761440154e-05,
      "loss": 1.014,
      "step": 116
    },
    {
      "epoch": 0.032138442521631644,
      "grad_norm": 0.1006399393081665,
      "learning_rate": 1.969492922907792e-05,
      "loss": 1.1963,
      "step": 117
    },
    {
      "epoch": 0.03241313006455157,
      "grad_norm": 0.11721698194742203,
      "learning_rate": 1.9692180843754296e-05,
      "loss": 1.2536,
      "step": 118
    },
    {
      "epoch": 0.032687817607471505,
      "grad_norm": 0.1279844492673874,
      "learning_rate": 1.9689432458430673e-05,
      "loss": 1.1334,
      "step": 119
    },
    {
      "epoch": 0.03296250515039143,
      "grad_norm": 0.10548660904169083,
      "learning_rate": 1.9686684073107053e-05,
      "loss": 1.21,
      "step": 120
    },
    {
      "epoch": 0.03323719269331136,
      "grad_norm": 0.12169504910707474,
      "learning_rate": 1.968393568778343e-05,
      "loss": 1.1411,
      "step": 121
    },
    {
      "epoch": 0.033511880236231285,
      "grad_norm": 0.09394995123147964,
      "learning_rate": 1.9681187302459804e-05,
      "loss": 1.2519,
      "step": 122
    },
    {
      "epoch": 0.03378656777915121,
      "grad_norm": 0.09256689995527267,
      "learning_rate": 1.9678438917136184e-05,
      "loss": 1.1683,
      "step": 123
    },
    {
      "epoch": 0.034061255322071146,
      "grad_norm": 0.10530166327953339,
      "learning_rate": 1.967569053181256e-05,
      "loss": 0.8727,
      "step": 124
    },
    {
      "epoch": 0.03433594286499107,
      "grad_norm": 0.10331820696592331,
      "learning_rate": 1.967294214648894e-05,
      "loss": 1.2499,
      "step": 125
    },
    {
      "epoch": 0.034610630407911,
      "grad_norm": 0.09267503023147583,
      "learning_rate": 1.967019376116532e-05,
      "loss": 1.0475,
      "step": 126
    },
    {
      "epoch": 0.03488531795083093,
      "grad_norm": 0.10376419872045517,
      "learning_rate": 1.9667445375841692e-05,
      "loss": 1.034,
      "step": 127
    },
    {
      "epoch": 0.03516000549375086,
      "grad_norm": 0.10059747099876404,
      "learning_rate": 1.9664696990518073e-05,
      "loss": 0.8422,
      "step": 128
    },
    {
      "epoch": 0.03543469303667079,
      "grad_norm": 0.11446807533502579,
      "learning_rate": 1.966194860519445e-05,
      "loss": 1.0946,
      "step": 129
    },
    {
      "epoch": 0.035709380579590715,
      "grad_norm": 0.10400739312171936,
      "learning_rate": 1.9659200219870827e-05,
      "loss": 1.1966,
      "step": 130
    },
    {
      "epoch": 0.03598406812251064,
      "grad_norm": 0.09485658258199692,
      "learning_rate": 1.9656451834547207e-05,
      "loss": 1.0134,
      "step": 131
    },
    {
      "epoch": 0.036258755665430575,
      "grad_norm": 0.10425304621458054,
      "learning_rate": 1.965370344922358e-05,
      "loss": 0.9866,
      "step": 132
    },
    {
      "epoch": 0.0365334432083505,
      "grad_norm": 0.07792259752750397,
      "learning_rate": 1.965095506389996e-05,
      "loss": 0.6403,
      "step": 133
    },
    {
      "epoch": 0.03680813075127043,
      "grad_norm": 0.10967417806386948,
      "learning_rate": 1.9648206678576338e-05,
      "loss": 0.8981,
      "step": 134
    },
    {
      "epoch": 0.037082818294190356,
      "grad_norm": 0.12117212265729904,
      "learning_rate": 1.9645458293252715e-05,
      "loss": 1.1396,
      "step": 135
    },
    {
      "epoch": 0.03735750583711029,
      "grad_norm": 0.10437469184398651,
      "learning_rate": 1.9642709907929092e-05,
      "loss": 0.9927,
      "step": 136
    },
    {
      "epoch": 0.03763219338003022,
      "grad_norm": 0.1386258453130722,
      "learning_rate": 1.963996152260547e-05,
      "loss": 1.3475,
      "step": 137
    },
    {
      "epoch": 0.037906880922950144,
      "grad_norm": 0.12044299393892288,
      "learning_rate": 1.9637213137281846e-05,
      "loss": 1.1889,
      "step": 138
    },
    {
      "epoch": 0.03818156846587007,
      "grad_norm": 0.1298886388540268,
      "learning_rate": 1.9634464751958227e-05,
      "loss": 1.2637,
      "step": 139
    },
    {
      "epoch": 0.038456256008790005,
      "grad_norm": 0.09976986795663834,
      "learning_rate": 1.9631716366634604e-05,
      "loss": 1.0974,
      "step": 140
    },
    {
      "epoch": 0.03873094355170993,
      "grad_norm": 0.11049885302782059,
      "learning_rate": 1.962896798131098e-05,
      "loss": 1.1213,
      "step": 141
    },
    {
      "epoch": 0.03900563109462986,
      "grad_norm": 0.112995445728302,
      "learning_rate": 1.9626219595987358e-05,
      "loss": 1.0659,
      "step": 142
    },
    {
      "epoch": 0.039280318637549785,
      "grad_norm": 0.11282811313867569,
      "learning_rate": 1.9623471210663735e-05,
      "loss": 1.1644,
      "step": 143
    },
    {
      "epoch": 0.03955500618046971,
      "grad_norm": 0.11578016728162766,
      "learning_rate": 1.9620722825340115e-05,
      "loss": 1.0966,
      "step": 144
    },
    {
      "epoch": 0.039829693723389646,
      "grad_norm": 0.10546927154064178,
      "learning_rate": 1.9617974440016492e-05,
      "loss": 1.1513,
      "step": 145
    },
    {
      "epoch": 0.04010438126630957,
      "grad_norm": 0.12422440946102142,
      "learning_rate": 1.961522605469287e-05,
      "loss": 0.8248,
      "step": 146
    },
    {
      "epoch": 0.0403790688092295,
      "grad_norm": 0.10891063511371613,
      "learning_rate": 1.9612477669369246e-05,
      "loss": 1.0009,
      "step": 147
    },
    {
      "epoch": 0.04065375635214943,
      "grad_norm": 0.12664256989955902,
      "learning_rate": 1.9609729284045623e-05,
      "loss": 1.0582,
      "step": 148
    },
    {
      "epoch": 0.04092844389506936,
      "grad_norm": 0.10750885307788849,
      "learning_rate": 1.9606980898722004e-05,
      "loss": 1.0251,
      "step": 149
    },
    {
      "epoch": 0.04120313143798929,
      "grad_norm": 0.10955626517534256,
      "learning_rate": 1.960423251339838e-05,
      "loss": 1.008,
      "step": 150
    },
    {
      "epoch": 0.041477818980909215,
      "grad_norm": 0.14049407839775085,
      "learning_rate": 1.9601484128074758e-05,
      "loss": 1.1506,
      "step": 151
    },
    {
      "epoch": 0.04175250652382914,
      "grad_norm": 0.11375827342271805,
      "learning_rate": 1.9598735742751135e-05,
      "loss": 0.9165,
      "step": 152
    },
    {
      "epoch": 0.042027194066749075,
      "grad_norm": 0.13538746535778046,
      "learning_rate": 1.9595987357427512e-05,
      "loss": 1.2151,
      "step": 153
    },
    {
      "epoch": 0.042301881609669,
      "grad_norm": 0.1271492838859558,
      "learning_rate": 1.959323897210389e-05,
      "loss": 1.2073,
      "step": 154
    },
    {
      "epoch": 0.04257656915258893,
      "grad_norm": 0.15812204778194427,
      "learning_rate": 1.959049058678027e-05,
      "loss": 1.0088,
      "step": 155
    },
    {
      "epoch": 0.042851256695508856,
      "grad_norm": 0.1072746217250824,
      "learning_rate": 1.9587742201456646e-05,
      "loss": 0.9323,
      "step": 156
    },
    {
      "epoch": 0.04312594423842879,
      "grad_norm": 0.13435253500938416,
      "learning_rate": 1.9584993816133023e-05,
      "loss": 1.0096,
      "step": 157
    },
    {
      "epoch": 0.04340063178134872,
      "grad_norm": 0.1116880401968956,
      "learning_rate": 1.95822454308094e-05,
      "loss": 1.0373,
      "step": 158
    },
    {
      "epoch": 0.043675319324268644,
      "grad_norm": 0.1288655400276184,
      "learning_rate": 1.9579497045485777e-05,
      "loss": 0.9866,
      "step": 159
    },
    {
      "epoch": 0.04395000686718857,
      "grad_norm": 0.15545062720775604,
      "learning_rate": 1.9576748660162158e-05,
      "loss": 1.0436,
      "step": 160
    },
    {
      "epoch": 0.044224694410108505,
      "grad_norm": 0.14241527020931244,
      "learning_rate": 1.9574000274838535e-05,
      "loss": 1.2257,
      "step": 161
    },
    {
      "epoch": 0.04449938195302843,
      "grad_norm": 0.137925386428833,
      "learning_rate": 1.9571251889514912e-05,
      "loss": 0.8741,
      "step": 162
    },
    {
      "epoch": 0.04477406949594836,
      "grad_norm": 0.11884178221225739,
      "learning_rate": 1.956850350419129e-05,
      "loss": 0.9186,
      "step": 163
    },
    {
      "epoch": 0.045048757038868285,
      "grad_norm": 0.1497151404619217,
      "learning_rate": 1.9565755118867666e-05,
      "loss": 1.0099,
      "step": 164
    },
    {
      "epoch": 0.04532344458178822,
      "grad_norm": 0.11900920420885086,
      "learning_rate": 1.9563006733544046e-05,
      "loss": 0.9404,
      "step": 165
    },
    {
      "epoch": 0.045598132124708146,
      "grad_norm": 0.11962069571018219,
      "learning_rate": 1.9560258348220423e-05,
      "loss": 0.7316,
      "step": 166
    },
    {
      "epoch": 0.04587281966762807,
      "grad_norm": 0.11308661848306656,
      "learning_rate": 1.95575099628968e-05,
      "loss": 1.2878,
      "step": 167
    },
    {
      "epoch": 0.046147507210548,
      "grad_norm": 0.15555961430072784,
      "learning_rate": 1.9554761577573177e-05,
      "loss": 1.0916,
      "step": 168
    },
    {
      "epoch": 0.04642219475346793,
      "grad_norm": 0.1287151426076889,
      "learning_rate": 1.9552013192249554e-05,
      "loss": 1.0111,
      "step": 169
    },
    {
      "epoch": 0.04669688229638786,
      "grad_norm": 0.10680960863828659,
      "learning_rate": 1.954926480692593e-05,
      "loss": 0.8627,
      "step": 170
    },
    {
      "epoch": 0.04697156983930779,
      "grad_norm": 0.1455630511045456,
      "learning_rate": 1.9546516421602312e-05,
      "loss": 1.1644,
      "step": 171
    },
    {
      "epoch": 0.047246257382227715,
      "grad_norm": 0.15489095449447632,
      "learning_rate": 1.954376803627869e-05,
      "loss": 0.8835,
      "step": 172
    },
    {
      "epoch": 0.04752094492514764,
      "grad_norm": 0.09784316271543503,
      "learning_rate": 1.9541019650955066e-05,
      "loss": 0.7948,
      "step": 173
    },
    {
      "epoch": 0.047795632468067575,
      "grad_norm": 0.15262404084205627,
      "learning_rate": 1.9538271265631443e-05,
      "loss": 1.2137,
      "step": 174
    },
    {
      "epoch": 0.0480703200109875,
      "grad_norm": 0.16134193539619446,
      "learning_rate": 1.953552288030782e-05,
      "loss": 0.9787,
      "step": 175
    },
    {
      "epoch": 0.04834500755390743,
      "grad_norm": 0.2693291902542114,
      "learning_rate": 1.95327744949842e-05,
      "loss": 1.1603,
      "step": 176
    },
    {
      "epoch": 0.048619695096827356,
      "grad_norm": 0.11801369488239288,
      "learning_rate": 1.9530026109660577e-05,
      "loss": 0.9708,
      "step": 177
    },
    {
      "epoch": 0.04889438263974729,
      "grad_norm": 0.1580204963684082,
      "learning_rate": 1.9527277724336954e-05,
      "loss": 1.0453,
      "step": 178
    },
    {
      "epoch": 0.04916907018266722,
      "grad_norm": 0.16258390247821808,
      "learning_rate": 1.952452933901333e-05,
      "loss": 1.002,
      "step": 179
    },
    {
      "epoch": 0.049443757725587144,
      "grad_norm": 0.13782678544521332,
      "learning_rate": 1.9521780953689708e-05,
      "loss": 1.0402,
      "step": 180
    },
    {
      "epoch": 0.04971844526850707,
      "grad_norm": 0.15491430461406708,
      "learning_rate": 1.951903256836609e-05,
      "loss": 1.1229,
      "step": 181
    },
    {
      "epoch": 0.049993132811427005,
      "grad_norm": 0.1389324963092804,
      "learning_rate": 1.9516284183042466e-05,
      "loss": 1.0141,
      "step": 182
    },
    {
      "epoch": 0.05026782035434693,
      "grad_norm": 0.11529399454593658,
      "learning_rate": 1.9513535797718843e-05,
      "loss": 1.1073,
      "step": 183
    },
    {
      "epoch": 0.05054250789726686,
      "grad_norm": 0.16894353926181793,
      "learning_rate": 1.951078741239522e-05,
      "loss": 1.1254,
      "step": 184
    },
    {
      "epoch": 0.050817195440186785,
      "grad_norm": 0.14054784178733826,
      "learning_rate": 1.9508039027071597e-05,
      "loss": 0.7504,
      "step": 185
    },
    {
      "epoch": 0.05109188298310672,
      "grad_norm": 0.1289864331483841,
      "learning_rate": 1.9505290641747974e-05,
      "loss": 1.1245,
      "step": 186
    },
    {
      "epoch": 0.051366570526026646,
      "grad_norm": 0.11992117762565613,
      "learning_rate": 1.9502542256424354e-05,
      "loss": 1.1431,
      "step": 187
    },
    {
      "epoch": 0.05164125806894657,
      "grad_norm": 0.12022904306650162,
      "learning_rate": 1.9499793871100728e-05,
      "loss": 0.838,
      "step": 188
    },
    {
      "epoch": 0.0519159456118665,
      "grad_norm": 0.1361938714981079,
      "learning_rate": 1.9497045485777108e-05,
      "loss": 1.1262,
      "step": 189
    },
    {
      "epoch": 0.052190633154786434,
      "grad_norm": 0.14422768354415894,
      "learning_rate": 1.9494297100453485e-05,
      "loss": 0.8646,
      "step": 190
    },
    {
      "epoch": 0.05246532069770636,
      "grad_norm": 0.11403167992830276,
      "learning_rate": 1.9491548715129862e-05,
      "loss": 0.9033,
      "step": 191
    },
    {
      "epoch": 0.05274000824062629,
      "grad_norm": 0.1277257353067398,
      "learning_rate": 1.9488800329806243e-05,
      "loss": 1.0048,
      "step": 192
    },
    {
      "epoch": 0.053014695783546215,
      "grad_norm": 0.14546413719654083,
      "learning_rate": 1.9486051944482616e-05,
      "loss": 1.2398,
      "step": 193
    },
    {
      "epoch": 0.05328938332646614,
      "grad_norm": 0.12667320668697357,
      "learning_rate": 1.9483303559158993e-05,
      "loss": 0.741,
      "step": 194
    },
    {
      "epoch": 0.053564070869386075,
      "grad_norm": 0.14974722266197205,
      "learning_rate": 1.9480555173835374e-05,
      "loss": 1.0986,
      "step": 195
    },
    {
      "epoch": 0.053838758412306,
      "grad_norm": 0.12947455048561096,
      "learning_rate": 1.947780678851175e-05,
      "loss": 1.0561,
      "step": 196
    },
    {
      "epoch": 0.05411344595522593,
      "grad_norm": 0.1456575244665146,
      "learning_rate": 1.947505840318813e-05,
      "loss": 1.1502,
      "step": 197
    },
    {
      "epoch": 0.054388133498145856,
      "grad_norm": 0.14889703691005707,
      "learning_rate": 1.9472310017864505e-05,
      "loss": 0.9891,
      "step": 198
    },
    {
      "epoch": 0.05466282104106579,
      "grad_norm": 0.12183894962072372,
      "learning_rate": 1.9469561632540882e-05,
      "loss": 0.8823,
      "step": 199
    },
    {
      "epoch": 0.05493750858398572,
      "grad_norm": 0.12501539289951324,
      "learning_rate": 1.9466813247217262e-05,
      "loss": 1.0282,
      "step": 200
    },
    {
      "epoch": 0.055212196126905644,
      "grad_norm": 0.126509428024292,
      "learning_rate": 1.946406486189364e-05,
      "loss": 1.1267,
      "step": 201
    },
    {
      "epoch": 0.05548688366982557,
      "grad_norm": 0.15080179274082184,
      "learning_rate": 1.9461316476570016e-05,
      "loss": 0.8591,
      "step": 202
    },
    {
      "epoch": 0.055761571212745505,
      "grad_norm": 0.24349328875541687,
      "learning_rate": 1.9458568091246393e-05,
      "loss": 1.1298,
      "step": 203
    },
    {
      "epoch": 0.05603625875566543,
      "grad_norm": 0.15895332396030426,
      "learning_rate": 1.945581970592277e-05,
      "loss": 0.9345,
      "step": 204
    },
    {
      "epoch": 0.05631094629858536,
      "grad_norm": 0.14921225607395172,
      "learning_rate": 1.945307132059915e-05,
      "loss": 0.8803,
      "step": 205
    },
    {
      "epoch": 0.056585633841505285,
      "grad_norm": 0.16234958171844482,
      "learning_rate": 1.9450322935275528e-05,
      "loss": 1.0274,
      "step": 206
    },
    {
      "epoch": 0.05686032138442522,
      "grad_norm": 0.1281847357749939,
      "learning_rate": 1.9447574549951905e-05,
      "loss": 0.7726,
      "step": 207
    },
    {
      "epoch": 0.057135008927345146,
      "grad_norm": 0.1627439707517624,
      "learning_rate": 1.9444826164628282e-05,
      "loss": 1.0628,
      "step": 208
    },
    {
      "epoch": 0.05740969647026507,
      "grad_norm": 0.12735991179943085,
      "learning_rate": 1.944207777930466e-05,
      "loss": 0.9859,
      "step": 209
    },
    {
      "epoch": 0.057684384013185,
      "grad_norm": 0.17674946784973145,
      "learning_rate": 1.9439329393981036e-05,
      "loss": 0.9574,
      "step": 210
    },
    {
      "epoch": 0.057959071556104934,
      "grad_norm": 0.20028840005397797,
      "learning_rate": 1.9436581008657416e-05,
      "loss": 1.2196,
      "step": 211
    },
    {
      "epoch": 0.05823375909902486,
      "grad_norm": 0.1389719843864441,
      "learning_rate": 1.9433832623333793e-05,
      "loss": 0.8134,
      "step": 212
    },
    {
      "epoch": 0.05850844664194479,
      "grad_norm": 0.17926430702209473,
      "learning_rate": 1.943108423801017e-05,
      "loss": 0.9373,
      "step": 213
    },
    {
      "epoch": 0.058783134184864715,
      "grad_norm": 0.172418013215065,
      "learning_rate": 1.9428335852686547e-05,
      "loss": 1.0027,
      "step": 214
    },
    {
      "epoch": 0.05905782172778464,
      "grad_norm": 0.18690133094787598,
      "learning_rate": 1.9425587467362924e-05,
      "loss": 1.2091,
      "step": 215
    },
    {
      "epoch": 0.059332509270704575,
      "grad_norm": 0.13616186380386353,
      "learning_rate": 1.9422839082039305e-05,
      "loss": 0.8798,
      "step": 216
    },
    {
      "epoch": 0.0596071968136245,
      "grad_norm": 0.1806904673576355,
      "learning_rate": 1.9420090696715682e-05,
      "loss": 1.024,
      "step": 217
    },
    {
      "epoch": 0.05988188435654443,
      "grad_norm": 0.15543559193611145,
      "learning_rate": 1.941734231139206e-05,
      "loss": 0.6903,
      "step": 218
    },
    {
      "epoch": 0.060156571899464356,
      "grad_norm": 0.16417598724365234,
      "learning_rate": 1.9414593926068436e-05,
      "loss": 1.0613,
      "step": 219
    },
    {
      "epoch": 0.06043125944238429,
      "grad_norm": 0.17087729275226593,
      "learning_rate": 1.9411845540744813e-05,
      "loss": 0.8139,
      "step": 220
    },
    {
      "epoch": 0.06070594698530422,
      "grad_norm": 0.16241183876991272,
      "learning_rate": 1.9409097155421193e-05,
      "loss": 1.2418,
      "step": 221
    },
    {
      "epoch": 0.060980634528224144,
      "grad_norm": 0.19880400598049164,
      "learning_rate": 1.940634877009757e-05,
      "loss": 0.9504,
      "step": 222
    },
    {
      "epoch": 0.06125532207114407,
      "grad_norm": 0.1443425714969635,
      "learning_rate": 1.9403600384773947e-05,
      "loss": 0.7887,
      "step": 223
    },
    {
      "epoch": 0.061530009614064005,
      "grad_norm": 0.13300739228725433,
      "learning_rate": 1.9400851999450324e-05,
      "loss": 1.0692,
      "step": 224
    },
    {
      "epoch": 0.06180469715698393,
      "grad_norm": 0.17682743072509766,
      "learning_rate": 1.93981036141267e-05,
      "loss": 1.2767,
      "step": 225
    },
    {
      "epoch": 0.06207938469990386,
      "grad_norm": 0.2004268914461136,
      "learning_rate": 1.939535522880308e-05,
      "loss": 1.0078,
      "step": 226
    },
    {
      "epoch": 0.062354072242823785,
      "grad_norm": 0.15890991687774658,
      "learning_rate": 1.939260684347946e-05,
      "loss": 0.7408,
      "step": 227
    },
    {
      "epoch": 0.06262875978574371,
      "grad_norm": 0.13084553182125092,
      "learning_rate": 1.9389858458155836e-05,
      "loss": 0.8581,
      "step": 228
    },
    {
      "epoch": 0.06290344732866364,
      "grad_norm": 0.1990857720375061,
      "learning_rate": 1.9387110072832213e-05,
      "loss": 1.0633,
      "step": 229
    },
    {
      "epoch": 0.06317813487158358,
      "grad_norm": 0.16882210969924927,
      "learning_rate": 1.938436168750859e-05,
      "loss": 1.0038,
      "step": 230
    },
    {
      "epoch": 0.0634528224145035,
      "grad_norm": 0.2074211835861206,
      "learning_rate": 1.9381613302184967e-05,
      "loss": 1.1391,
      "step": 231
    },
    {
      "epoch": 0.06372750995742343,
      "grad_norm": 0.16279420256614685,
      "learning_rate": 1.9378864916861347e-05,
      "loss": 0.9447,
      "step": 232
    },
    {
      "epoch": 0.06400219750034336,
      "grad_norm": 0.12484488636255264,
      "learning_rate": 1.9376116531537724e-05,
      "loss": 0.8492,
      "step": 233
    },
    {
      "epoch": 0.06427688504326329,
      "grad_norm": 0.17254042625427246,
      "learning_rate": 1.93733681462141e-05,
      "loss": 1.0012,
      "step": 234
    },
    {
      "epoch": 0.06455157258618321,
      "grad_norm": 0.13968612253665924,
      "learning_rate": 1.937061976089048e-05,
      "loss": 0.8029,
      "step": 235
    },
    {
      "epoch": 0.06482626012910314,
      "grad_norm": 0.14487391710281372,
      "learning_rate": 1.9367871375566855e-05,
      "loss": 1.2325,
      "step": 236
    },
    {
      "epoch": 0.06510094767202307,
      "grad_norm": 0.18000279366970062,
      "learning_rate": 1.9365122990243236e-05,
      "loss": 0.9215,
      "step": 237
    },
    {
      "epoch": 0.06537563521494301,
      "grad_norm": 0.15903238952159882,
      "learning_rate": 1.9362374604919613e-05,
      "loss": 0.6348,
      "step": 238
    },
    {
      "epoch": 0.06565032275786294,
      "grad_norm": 0.3841058015823364,
      "learning_rate": 1.935962621959599e-05,
      "loss": 1.2468,
      "step": 239
    },
    {
      "epoch": 0.06592501030078286,
      "grad_norm": 0.16707414388656616,
      "learning_rate": 1.9356877834272367e-05,
      "loss": 1.0285,
      "step": 240
    },
    {
      "epoch": 0.06619969784370279,
      "grad_norm": 0.16990646719932556,
      "learning_rate": 1.9354129448948744e-05,
      "loss": 0.9581,
      "step": 241
    },
    {
      "epoch": 0.06647438538662272,
      "grad_norm": 0.19277583062648773,
      "learning_rate": 1.935138106362512e-05,
      "loss": 1.0062,
      "step": 242
    },
    {
      "epoch": 0.06674907292954264,
      "grad_norm": 0.20673274993896484,
      "learning_rate": 1.93486326783015e-05,
      "loss": 1.1598,
      "step": 243
    },
    {
      "epoch": 0.06702376047246257,
      "grad_norm": 0.17789694666862488,
      "learning_rate": 1.9345884292977878e-05,
      "loss": 1.16,
      "step": 244
    },
    {
      "epoch": 0.0672984480153825,
      "grad_norm": 0.17580854892730713,
      "learning_rate": 1.9343135907654255e-05,
      "loss": 0.8938,
      "step": 245
    },
    {
      "epoch": 0.06757313555830242,
      "grad_norm": 0.169183611869812,
      "learning_rate": 1.9340387522330632e-05,
      "loss": 1.0482,
      "step": 246
    },
    {
      "epoch": 0.06784782310122237,
      "grad_norm": 0.16836467385292053,
      "learning_rate": 1.933763913700701e-05,
      "loss": 1.0321,
      "step": 247
    },
    {
      "epoch": 0.06812251064414229,
      "grad_norm": 0.16751019656658173,
      "learning_rate": 1.933489075168339e-05,
      "loss": 1.0547,
      "step": 248
    },
    {
      "epoch": 0.06839719818706222,
      "grad_norm": 0.1935374140739441,
      "learning_rate": 1.9332142366359767e-05,
      "loss": 0.8726,
      "step": 249
    },
    {
      "epoch": 0.06867188572998215,
      "grad_norm": 0.1703059822320938,
      "learning_rate": 1.932939398103614e-05,
      "loss": 0.8531,
      "step": 250
    },
    {
      "epoch": 0.06894657327290207,
      "grad_norm": 0.15444602072238922,
      "learning_rate": 1.932664559571252e-05,
      "loss": 1.2211,
      "step": 251
    },
    {
      "epoch": 0.069221260815822,
      "grad_norm": 0.1815018504858017,
      "learning_rate": 1.9323897210388898e-05,
      "loss": 0.9448,
      "step": 252
    },
    {
      "epoch": 0.06949594835874193,
      "grad_norm": 0.1607893705368042,
      "learning_rate": 1.9321148825065278e-05,
      "loss": 1.1993,
      "step": 253
    },
    {
      "epoch": 0.06977063590166185,
      "grad_norm": 0.12963637709617615,
      "learning_rate": 1.9318400439741655e-05,
      "loss": 0.7567,
      "step": 254
    },
    {
      "epoch": 0.0700453234445818,
      "grad_norm": 0.16476108133792877,
      "learning_rate": 1.931565205441803e-05,
      "loss": 0.8441,
      "step": 255
    },
    {
      "epoch": 0.07032001098750172,
      "grad_norm": 0.16153515875339508,
      "learning_rate": 1.931290366909441e-05,
      "loss": 1.0119,
      "step": 256
    },
    {
      "epoch": 0.07059469853042165,
      "grad_norm": 0.16923516988754272,
      "learning_rate": 1.9310155283770786e-05,
      "loss": 0.9391,
      "step": 257
    },
    {
      "epoch": 0.07086938607334158,
      "grad_norm": 0.1405295580625534,
      "learning_rate": 1.9307406898447163e-05,
      "loss": 1.0393,
      "step": 258
    },
    {
      "epoch": 0.0711440736162615,
      "grad_norm": 0.15492890775203705,
      "learning_rate": 1.930465851312354e-05,
      "loss": 1.3664,
      "step": 259
    },
    {
      "epoch": 0.07141876115918143,
      "grad_norm": 0.18521930277347565,
      "learning_rate": 1.9301910127799917e-05,
      "loss": 0.9557,
      "step": 260
    },
    {
      "epoch": 0.07169344870210136,
      "grad_norm": 0.17794646322727203,
      "learning_rate": 1.9299161742476298e-05,
      "loss": 0.7601,
      "step": 261
    },
    {
      "epoch": 0.07196813624502128,
      "grad_norm": 0.17436501383781433,
      "learning_rate": 1.9296413357152675e-05,
      "loss": 0.8434,
      "step": 262
    },
    {
      "epoch": 0.07224282378794121,
      "grad_norm": 0.1715608537197113,
      "learning_rate": 1.9293664971829052e-05,
      "loss": 1.2185,
      "step": 263
    },
    {
      "epoch": 0.07251751133086115,
      "grad_norm": 0.16292747855186462,
      "learning_rate": 1.929091658650543e-05,
      "loss": 0.8552,
      "step": 264
    },
    {
      "epoch": 0.07279219887378108,
      "grad_norm": 0.18009597063064575,
      "learning_rate": 1.9288168201181806e-05,
      "loss": 0.9166,
      "step": 265
    },
    {
      "epoch": 0.073066886416701,
      "grad_norm": 0.1907026767730713,
      "learning_rate": 1.9285419815858183e-05,
      "loss": 0.9343,
      "step": 266
    },
    {
      "epoch": 0.07334157395962093,
      "grad_norm": 0.19873075187206268,
      "learning_rate": 1.9282671430534563e-05,
      "loss": 1.1497,
      "step": 267
    },
    {
      "epoch": 0.07361626150254086,
      "grad_norm": 0.16478480398654938,
      "learning_rate": 1.927992304521094e-05,
      "loss": 0.8748,
      "step": 268
    },
    {
      "epoch": 0.07389094904546079,
      "grad_norm": 0.19074921309947968,
      "learning_rate": 1.9277174659887317e-05,
      "loss": 0.8424,
      "step": 269
    },
    {
      "epoch": 0.07416563658838071,
      "grad_norm": 0.1560112088918686,
      "learning_rate": 1.9274426274563694e-05,
      "loss": 0.7739,
      "step": 270
    },
    {
      "epoch": 0.07444032413130064,
      "grad_norm": 0.14705249667167664,
      "learning_rate": 1.927167788924007e-05,
      "loss": 0.8875,
      "step": 271
    },
    {
      "epoch": 0.07471501167422058,
      "grad_norm": 0.14907163381576538,
      "learning_rate": 1.9268929503916452e-05,
      "loss": 0.7408,
      "step": 272
    },
    {
      "epoch": 0.0749896992171405,
      "grad_norm": 0.25197142362594604,
      "learning_rate": 1.926618111859283e-05,
      "loss": 0.9836,
      "step": 273
    },
    {
      "epoch": 0.07526438676006043,
      "grad_norm": 0.15481406450271606,
      "learning_rate": 1.9263432733269206e-05,
      "loss": 1.0301,
      "step": 274
    },
    {
      "epoch": 0.07553907430298036,
      "grad_norm": 0.17336656153202057,
      "learning_rate": 1.9260684347945583e-05,
      "loss": 1.0967,
      "step": 275
    },
    {
      "epoch": 0.07581376184590029,
      "grad_norm": 0.19636762142181396,
      "learning_rate": 1.925793596262196e-05,
      "loss": 0.8243,
      "step": 276
    },
    {
      "epoch": 0.07608844938882021,
      "grad_norm": 0.2009076625108719,
      "learning_rate": 1.925518757729834e-05,
      "loss": 0.8461,
      "step": 277
    },
    {
      "epoch": 0.07636313693174014,
      "grad_norm": 0.16950824856758118,
      "learning_rate": 1.9252439191974717e-05,
      "loss": 0.865,
      "step": 278
    },
    {
      "epoch": 0.07663782447466007,
      "grad_norm": 0.17805103957653046,
      "learning_rate": 1.9249690806651094e-05,
      "loss": 0.9746,
      "step": 279
    },
    {
      "epoch": 0.07691251201758001,
      "grad_norm": 0.15453767776489258,
      "learning_rate": 1.924694242132747e-05,
      "loss": 0.8678,
      "step": 280
    },
    {
      "epoch": 0.07718719956049994,
      "grad_norm": 0.23246809840202332,
      "learning_rate": 1.924419403600385e-05,
      "loss": 1.1294,
      "step": 281
    },
    {
      "epoch": 0.07746188710341986,
      "grad_norm": 0.254901260137558,
      "learning_rate": 1.9241445650680225e-05,
      "loss": 1.107,
      "step": 282
    },
    {
      "epoch": 0.07773657464633979,
      "grad_norm": 0.18279704451560974,
      "learning_rate": 1.9238697265356606e-05,
      "loss": 1.0445,
      "step": 283
    },
    {
      "epoch": 0.07801126218925972,
      "grad_norm": 0.24285514652729034,
      "learning_rate": 1.9235948880032983e-05,
      "loss": 1.1957,
      "step": 284
    },
    {
      "epoch": 0.07828594973217964,
      "grad_norm": 0.1509721279144287,
      "learning_rate": 1.923320049470936e-05,
      "loss": 0.7003,
      "step": 285
    },
    {
      "epoch": 0.07856063727509957,
      "grad_norm": 0.19022563099861145,
      "learning_rate": 1.9230452109385737e-05,
      "loss": 0.9512,
      "step": 286
    },
    {
      "epoch": 0.0788353248180195,
      "grad_norm": 0.18767860531806946,
      "learning_rate": 1.9227703724062114e-05,
      "loss": 1.1765,
      "step": 287
    },
    {
      "epoch": 0.07911001236093942,
      "grad_norm": 0.19026300311088562,
      "learning_rate": 1.9224955338738494e-05,
      "loss": 0.9973,
      "step": 288
    },
    {
      "epoch": 0.07938469990385937,
      "grad_norm": 0.18111121654510498,
      "learning_rate": 1.922220695341487e-05,
      "loss": 1.0408,
      "step": 289
    },
    {
      "epoch": 0.07965938744677929,
      "grad_norm": 0.1737125664949417,
      "learning_rate": 1.921945856809125e-05,
      "loss": 1.1665,
      "step": 290
    },
    {
      "epoch": 0.07993407498969922,
      "grad_norm": 0.21849822998046875,
      "learning_rate": 1.9216710182767625e-05,
      "loss": 1.2456,
      "step": 291
    },
    {
      "epoch": 0.08020876253261915,
      "grad_norm": 0.2132040560245514,
      "learning_rate": 1.9213961797444002e-05,
      "loss": 0.7612,
      "step": 292
    },
    {
      "epoch": 0.08048345007553907,
      "grad_norm": 0.18481497466564178,
      "learning_rate": 1.9211213412120383e-05,
      "loss": 1.0744,
      "step": 293
    },
    {
      "epoch": 0.080758137618459,
      "grad_norm": 0.23042362928390503,
      "learning_rate": 1.920846502679676e-05,
      "loss": 1.1812,
      "step": 294
    },
    {
      "epoch": 0.08103282516137893,
      "grad_norm": 0.19078391790390015,
      "learning_rate": 1.9205716641473137e-05,
      "loss": 0.7947,
      "step": 295
    },
    {
      "epoch": 0.08130751270429885,
      "grad_norm": 0.17270620167255402,
      "learning_rate": 1.9202968256149514e-05,
      "loss": 0.934,
      "step": 296
    },
    {
      "epoch": 0.0815822002472188,
      "grad_norm": 0.20097284018993378,
      "learning_rate": 1.920021987082589e-05,
      "loss": 0.793,
      "step": 297
    },
    {
      "epoch": 0.08185688779013872,
      "grad_norm": 0.2355027198791504,
      "learning_rate": 1.9197471485502268e-05,
      "loss": 1.086,
      "step": 298
    },
    {
      "epoch": 0.08213157533305865,
      "grad_norm": 0.15790455043315887,
      "learning_rate": 1.9194723100178648e-05,
      "loss": 0.7594,
      "step": 299
    },
    {
      "epoch": 0.08240626287597858,
      "grad_norm": 0.18923120200634003,
      "learning_rate": 1.9191974714855025e-05,
      "loss": 0.905,
      "step": 300
    },
    {
      "epoch": 0.0826809504188985,
      "grad_norm": 0.19738522171974182,
      "learning_rate": 1.9189226329531402e-05,
      "loss": 0.8632,
      "step": 301
    },
    {
      "epoch": 0.08295563796181843,
      "grad_norm": 0.20312167704105377,
      "learning_rate": 1.918647794420778e-05,
      "loss": 1.0733,
      "step": 302
    },
    {
      "epoch": 0.08323032550473836,
      "grad_norm": 0.19975174963474274,
      "learning_rate": 1.9183729558884156e-05,
      "loss": 0.9486,
      "step": 303
    },
    {
      "epoch": 0.08350501304765828,
      "grad_norm": 0.10008151829242706,
      "learning_rate": 1.9180981173560537e-05,
      "loss": 0.5193,
      "step": 304
    },
    {
      "epoch": 0.08377970059057822,
      "grad_norm": 0.2182292640209198,
      "learning_rate": 1.9178232788236914e-05,
      "loss": 0.8774,
      "step": 305
    },
    {
      "epoch": 0.08405438813349815,
      "grad_norm": 0.17745904624462128,
      "learning_rate": 1.917548440291329e-05,
      "loss": 0.8935,
      "step": 306
    },
    {
      "epoch": 0.08432907567641808,
      "grad_norm": 0.2336822748184204,
      "learning_rate": 1.9172736017589668e-05,
      "loss": 1.1938,
      "step": 307
    },
    {
      "epoch": 0.084603763219338,
      "grad_norm": 0.23907525837421417,
      "learning_rate": 1.9169987632266045e-05,
      "loss": 1.0945,
      "step": 308
    },
    {
      "epoch": 0.08487845076225793,
      "grad_norm": 0.16967865824699402,
      "learning_rate": 1.9167239246942425e-05,
      "loss": 0.9525,
      "step": 309
    },
    {
      "epoch": 0.08515313830517786,
      "grad_norm": 0.18005500733852386,
      "learning_rate": 1.9164490861618802e-05,
      "loss": 0.8939,
      "step": 310
    },
    {
      "epoch": 0.08542782584809779,
      "grad_norm": 0.25240203738212585,
      "learning_rate": 1.9161742476295176e-05,
      "loss": 1.0182,
      "step": 311
    },
    {
      "epoch": 0.08570251339101771,
      "grad_norm": 0.19365176558494568,
      "learning_rate": 1.9158994090971556e-05,
      "loss": 1.1076,
      "step": 312
    },
    {
      "epoch": 0.08597720093393764,
      "grad_norm": 0.17365913093090057,
      "learning_rate": 1.9156245705647933e-05,
      "loss": 0.8023,
      "step": 313
    },
    {
      "epoch": 0.08625188847685758,
      "grad_norm": 0.14304108917713165,
      "learning_rate": 1.915349732032431e-05,
      "loss": 0.5628,
      "step": 314
    },
    {
      "epoch": 0.0865265760197775,
      "grad_norm": 0.24889570474624634,
      "learning_rate": 1.915074893500069e-05,
      "loss": 0.999,
      "step": 315
    },
    {
      "epoch": 0.08680126356269743,
      "grad_norm": 0.22276797890663147,
      "learning_rate": 1.9148000549677064e-05,
      "loss": 1.1354,
      "step": 316
    },
    {
      "epoch": 0.08707595110561736,
      "grad_norm": 0.208163782954216,
      "learning_rate": 1.9145252164353445e-05,
      "loss": 1.0199,
      "step": 317
    },
    {
      "epoch": 0.08735063864853729,
      "grad_norm": 0.17377251386642456,
      "learning_rate": 1.9142503779029822e-05,
      "loss": 0.7484,
      "step": 318
    },
    {
      "epoch": 0.08762532619145721,
      "grad_norm": 0.180647611618042,
      "learning_rate": 1.91397553937062e-05,
      "loss": 0.9377,
      "step": 319
    },
    {
      "epoch": 0.08790001373437714,
      "grad_norm": 0.21593716740608215,
      "learning_rate": 1.913700700838258e-05,
      "loss": 0.8456,
      "step": 320
    },
    {
      "epoch": 0.08817470127729707,
      "grad_norm": 0.22909027338027954,
      "learning_rate": 1.9134258623058953e-05,
      "loss": 1.0863,
      "step": 321
    },
    {
      "epoch": 0.08844938882021701,
      "grad_norm": 0.1879461407661438,
      "learning_rate": 1.913151023773533e-05,
      "loss": 0.7656,
      "step": 322
    },
    {
      "epoch": 0.08872407636313694,
      "grad_norm": 0.15301813185214996,
      "learning_rate": 1.912876185241171e-05,
      "loss": 0.8398,
      "step": 323
    },
    {
      "epoch": 0.08899876390605686,
      "grad_norm": 0.14185811579227448,
      "learning_rate": 1.9126013467088087e-05,
      "loss": 0.6995,
      "step": 324
    },
    {
      "epoch": 0.08927345144897679,
      "grad_norm": 0.20367728173732758,
      "learning_rate": 1.9123265081764464e-05,
      "loss": 0.9025,
      "step": 325
    },
    {
      "epoch": 0.08954813899189672,
      "grad_norm": 0.20401909947395325,
      "learning_rate": 1.912051669644084e-05,
      "loss": 0.9581,
      "step": 326
    },
    {
      "epoch": 0.08982282653481664,
      "grad_norm": 0.1789512038230896,
      "learning_rate": 1.911776831111722e-05,
      "loss": 0.7546,
      "step": 327
    },
    {
      "epoch": 0.09009751407773657,
      "grad_norm": 0.1964261829853058,
      "learning_rate": 1.91150199257936e-05,
      "loss": 0.9651,
      "step": 328
    },
    {
      "epoch": 0.0903722016206565,
      "grad_norm": 0.2441503256559372,
      "learning_rate": 1.9112271540469976e-05,
      "loss": 1.1417,
      "step": 329
    },
    {
      "epoch": 0.09064688916357644,
      "grad_norm": 0.14766603708267212,
      "learning_rate": 1.9109523155146353e-05,
      "loss": 0.6347,
      "step": 330
    },
    {
      "epoch": 0.09092157670649637,
      "grad_norm": 0.18716420233249664,
      "learning_rate": 1.910677476982273e-05,
      "loss": 0.8922,
      "step": 331
    },
    {
      "epoch": 0.09119626424941629,
      "grad_norm": 0.24644790589809418,
      "learning_rate": 1.9104026384499107e-05,
      "loss": 0.8423,
      "step": 332
    },
    {
      "epoch": 0.09147095179233622,
      "grad_norm": 0.17163582146167755,
      "learning_rate": 1.9101277999175484e-05,
      "loss": 1.1647,
      "step": 333
    },
    {
      "epoch": 0.09174563933525615,
      "grad_norm": 0.20418354868888855,
      "learning_rate": 1.9098529613851864e-05,
      "loss": 0.957,
      "step": 334
    },
    {
      "epoch": 0.09202032687817607,
      "grad_norm": 0.24328942596912384,
      "learning_rate": 1.909578122852824e-05,
      "loss": 0.8754,
      "step": 335
    },
    {
      "epoch": 0.092295014421096,
      "grad_norm": 0.30439668893814087,
      "learning_rate": 1.909303284320462e-05,
      "loss": 1.1421,
      "step": 336
    },
    {
      "epoch": 0.09256970196401593,
      "grad_norm": 0.20734351873397827,
      "learning_rate": 1.9090284457880995e-05,
      "loss": 0.7701,
      "step": 337
    },
    {
      "epoch": 0.09284438950693585,
      "grad_norm": 0.1469442993402481,
      "learning_rate": 1.9087536072557372e-05,
      "loss": 0.9125,
      "step": 338
    },
    {
      "epoch": 0.0931190770498558,
      "grad_norm": 0.2230120152235031,
      "learning_rate": 1.9084787687233753e-05,
      "loss": 0.9293,
      "step": 339
    },
    {
      "epoch": 0.09339376459277572,
      "grad_norm": 0.17525744438171387,
      "learning_rate": 1.908203930191013e-05,
      "loss": 0.877,
      "step": 340
    },
    {
      "epoch": 0.09366845213569565,
      "grad_norm": 0.24115079641342163,
      "learning_rate": 1.9079290916586507e-05,
      "loss": 1.1363,
      "step": 341
    },
    {
      "epoch": 0.09394313967861558,
      "grad_norm": 0.15906444191932678,
      "learning_rate": 1.9076542531262884e-05,
      "loss": 0.8805,
      "step": 342
    },
    {
      "epoch": 0.0942178272215355,
      "grad_norm": 0.26078560948371887,
      "learning_rate": 1.907379414593926e-05,
      "loss": 1.0682,
      "step": 343
    },
    {
      "epoch": 0.09449251476445543,
      "grad_norm": 0.18845054507255554,
      "learning_rate": 1.907104576061564e-05,
      "loss": 0.8613,
      "step": 344
    },
    {
      "epoch": 0.09476720230737536,
      "grad_norm": 0.17851805686950684,
      "learning_rate": 1.906829737529202e-05,
      "loss": 0.8075,
      "step": 345
    },
    {
      "epoch": 0.09504188985029528,
      "grad_norm": 0.14846324920654297,
      "learning_rate": 1.9065548989968395e-05,
      "loss": 0.8094,
      "step": 346
    },
    {
      "epoch": 0.09531657739321522,
      "grad_norm": 0.21913903951644897,
      "learning_rate": 1.9062800604644772e-05,
      "loss": 0.9514,
      "step": 347
    },
    {
      "epoch": 0.09559126493613515,
      "grad_norm": 0.27384698390960693,
      "learning_rate": 1.906005221932115e-05,
      "loss": 1.0756,
      "step": 348
    },
    {
      "epoch": 0.09586595247905508,
      "grad_norm": 0.17414698004722595,
      "learning_rate": 1.9057303833997526e-05,
      "loss": 0.8022,
      "step": 349
    },
    {
      "epoch": 0.096140640021975,
      "grad_norm": 0.2148537039756775,
      "learning_rate": 1.9054555448673907e-05,
      "loss": 0.7785,
      "step": 350
    },
    {
      "epoch": 0.09641532756489493,
      "grad_norm": 0.20703645050525665,
      "learning_rate": 1.9051807063350284e-05,
      "loss": 0.9234,
      "step": 351
    },
    {
      "epoch": 0.09669001510781486,
      "grad_norm": 0.32048484683036804,
      "learning_rate": 1.904905867802666e-05,
      "loss": 1.0091,
      "step": 352
    },
    {
      "epoch": 0.09696470265073479,
      "grad_norm": 0.1910296231508255,
      "learning_rate": 1.9046310292703038e-05,
      "loss": 0.7743,
      "step": 353
    },
    {
      "epoch": 0.09723939019365471,
      "grad_norm": 0.23881374299526215,
      "learning_rate": 1.9043561907379415e-05,
      "loss": 1.2118,
      "step": 354
    },
    {
      "epoch": 0.09751407773657465,
      "grad_norm": 0.23802803456783295,
      "learning_rate": 1.9040813522055795e-05,
      "loss": 0.9756,
      "step": 355
    },
    {
      "epoch": 0.09778876527949458,
      "grad_norm": 0.16748611629009247,
      "learning_rate": 1.9038065136732172e-05,
      "loss": 0.7039,
      "step": 356
    },
    {
      "epoch": 0.0980634528224145,
      "grad_norm": 0.20205305516719818,
      "learning_rate": 1.903531675140855e-05,
      "loss": 0.8259,
      "step": 357
    },
    {
      "epoch": 0.09833814036533443,
      "grad_norm": 0.23501551151275635,
      "learning_rate": 1.9032568366084926e-05,
      "loss": 1.121,
      "step": 358
    },
    {
      "epoch": 0.09861282790825436,
      "grad_norm": 0.2610239088535309,
      "learning_rate": 1.9029819980761303e-05,
      "loss": 1.1473,
      "step": 359
    },
    {
      "epoch": 0.09888751545117429,
      "grad_norm": 0.17410385608673096,
      "learning_rate": 1.9027071595437684e-05,
      "loss": 1.0664,
      "step": 360
    },
    {
      "epoch": 0.09916220299409421,
      "grad_norm": 0.16954925656318665,
      "learning_rate": 1.902432321011406e-05,
      "loss": 0.7869,
      "step": 361
    },
    {
      "epoch": 0.09943689053701414,
      "grad_norm": 0.22946037352085114,
      "learning_rate": 1.9021574824790438e-05,
      "loss": 0.9426,
      "step": 362
    },
    {
      "epoch": 0.09971157807993407,
      "grad_norm": 0.21892409026622772,
      "learning_rate": 1.9018826439466815e-05,
      "loss": 1.0728,
      "step": 363
    },
    {
      "epoch": 0.09998626562285401,
      "grad_norm": 0.16896452009677887,
      "learning_rate": 1.9016078054143192e-05,
      "loss": 0.9081,
      "step": 364
    },
    {
      "epoch": 0.10026095316577394,
      "grad_norm": 0.14271801710128784,
      "learning_rate": 1.901332966881957e-05,
      "loss": 0.7545,
      "step": 365
    },
    {
      "epoch": 0.10053564070869386,
      "grad_norm": 0.18783532083034515,
      "learning_rate": 1.901058128349595e-05,
      "loss": 0.6591,
      "step": 366
    },
    {
      "epoch": 0.10081032825161379,
      "grad_norm": 0.17143316566944122,
      "learning_rate": 1.9007832898172326e-05,
      "loss": 0.985,
      "step": 367
    },
    {
      "epoch": 0.10108501579453372,
      "grad_norm": 0.2065405696630478,
      "learning_rate": 1.9005084512848703e-05,
      "loss": 0.9448,
      "step": 368
    },
    {
      "epoch": 0.10135970333745364,
      "grad_norm": 0.20083104074001312,
      "learning_rate": 1.900233612752508e-05,
      "loss": 0.8976,
      "step": 369
    },
    {
      "epoch": 0.10163439088037357,
      "grad_norm": 0.25744059681892395,
      "learning_rate": 1.8999587742201457e-05,
      "loss": 1.0222,
      "step": 370
    },
    {
      "epoch": 0.1019090784232935,
      "grad_norm": 0.23219414055347443,
      "learning_rate": 1.8996839356877838e-05,
      "loss": 0.9667,
      "step": 371
    },
    {
      "epoch": 0.10218376596621344,
      "grad_norm": 0.17629918456077576,
      "learning_rate": 1.8994090971554215e-05,
      "loss": 0.8444,
      "step": 372
    },
    {
      "epoch": 0.10245845350913337,
      "grad_norm": 0.2130831480026245,
      "learning_rate": 1.899134258623059e-05,
      "loss": 0.7289,
      "step": 373
    },
    {
      "epoch": 0.10273314105205329,
      "grad_norm": 0.2470535933971405,
      "learning_rate": 1.898859420090697e-05,
      "loss": 0.7429,
      "step": 374
    },
    {
      "epoch": 0.10300782859497322,
      "grad_norm": 0.20799966156482697,
      "learning_rate": 1.8985845815583346e-05,
      "loss": 0.8345,
      "step": 375
    },
    {
      "epoch": 0.10328251613789315,
      "grad_norm": 0.16663435101509094,
      "learning_rate": 1.8983097430259726e-05,
      "loss": 0.8363,
      "step": 376
    },
    {
      "epoch": 0.10355720368081307,
      "grad_norm": 0.15801647305488586,
      "learning_rate": 1.89803490449361e-05,
      "loss": 0.7495,
      "step": 377
    },
    {
      "epoch": 0.103831891223733,
      "grad_norm": 0.2167907953262329,
      "learning_rate": 1.8977600659612477e-05,
      "loss": 0.936,
      "step": 378
    },
    {
      "epoch": 0.10410657876665293,
      "grad_norm": 0.2024611234664917,
      "learning_rate": 1.8974852274288857e-05,
      "loss": 1.1056,
      "step": 379
    },
    {
      "epoch": 0.10438126630957287,
      "grad_norm": 0.22914043068885803,
      "learning_rate": 1.8972103888965234e-05,
      "loss": 1.062,
      "step": 380
    },
    {
      "epoch": 0.1046559538524928,
      "grad_norm": 0.21754172444343567,
      "learning_rate": 1.896935550364161e-05,
      "loss": 0.913,
      "step": 381
    },
    {
      "epoch": 0.10493064139541272,
      "grad_norm": 0.19849437475204468,
      "learning_rate": 1.896660711831799e-05,
      "loss": 1.1552,
      "step": 382
    },
    {
      "epoch": 0.10520532893833265,
      "grad_norm": 0.26813170313835144,
      "learning_rate": 1.8963858732994365e-05,
      "loss": 0.9604,
      "step": 383
    },
    {
      "epoch": 0.10548001648125258,
      "grad_norm": 0.2021423727273941,
      "learning_rate": 1.8961110347670746e-05,
      "loss": 1.1084,
      "step": 384
    },
    {
      "epoch": 0.1057547040241725,
      "grad_norm": 0.238947331905365,
      "learning_rate": 1.8958361962347123e-05,
      "loss": 0.9358,
      "step": 385
    },
    {
      "epoch": 0.10602939156709243,
      "grad_norm": 0.22574155032634735,
      "learning_rate": 1.89556135770235e-05,
      "loss": 0.725,
      "step": 386
    },
    {
      "epoch": 0.10630407911001236,
      "grad_norm": 0.19930416345596313,
      "learning_rate": 1.8952865191699877e-05,
      "loss": 0.9979,
      "step": 387
    },
    {
      "epoch": 0.10657876665293228,
      "grad_norm": 0.2722059488296509,
      "learning_rate": 1.8950116806376254e-05,
      "loss": 1.0608,
      "step": 388
    },
    {
      "epoch": 0.10685345419585222,
      "grad_norm": 0.19103911519050598,
      "learning_rate": 1.894736842105263e-05,
      "loss": 0.9474,
      "step": 389
    },
    {
      "epoch": 0.10712814173877215,
      "grad_norm": 0.20642293989658356,
      "learning_rate": 1.894462003572901e-05,
      "loss": 0.8737,
      "step": 390
    },
    {
      "epoch": 0.10740282928169208,
      "grad_norm": 0.20131151378154755,
      "learning_rate": 1.894187165040539e-05,
      "loss": 0.9362,
      "step": 391
    },
    {
      "epoch": 0.107677516824612,
      "grad_norm": 0.28266796469688416,
      "learning_rate": 1.8939123265081765e-05,
      "loss": 0.8557,
      "step": 392
    },
    {
      "epoch": 0.10795220436753193,
      "grad_norm": 0.22416886687278748,
      "learning_rate": 1.8936374879758142e-05,
      "loss": 1.1964,
      "step": 393
    },
    {
      "epoch": 0.10822689191045186,
      "grad_norm": 0.2844870686531067,
      "learning_rate": 1.893362649443452e-05,
      "loss": 1.0948,
      "step": 394
    },
    {
      "epoch": 0.10850157945337179,
      "grad_norm": 0.20682677626609802,
      "learning_rate": 1.89308781091109e-05,
      "loss": 0.9233,
      "step": 395
    },
    {
      "epoch": 0.10877626699629171,
      "grad_norm": 0.2839471697807312,
      "learning_rate": 1.8928129723787277e-05,
      "loss": 1.1683,
      "step": 396
    },
    {
      "epoch": 0.10905095453921165,
      "grad_norm": 0.22104988992214203,
      "learning_rate": 1.8925381338463654e-05,
      "loss": 0.8203,
      "step": 397
    },
    {
      "epoch": 0.10932564208213158,
      "grad_norm": 0.1830834299325943,
      "learning_rate": 1.892263295314003e-05,
      "loss": 0.8508,
      "step": 398
    },
    {
      "epoch": 0.1096003296250515,
      "grad_norm": 0.25269556045532227,
      "learning_rate": 1.8919884567816408e-05,
      "loss": 1.0976,
      "step": 399
    },
    {
      "epoch": 0.10987501716797143,
      "grad_norm": 0.17195792496204376,
      "learning_rate": 1.891713618249279e-05,
      "loss": 0.7126,
      "step": 400
    },
    {
      "epoch": 0.11014970471089136,
      "grad_norm": 0.2326473891735077,
      "learning_rate": 1.8914387797169165e-05,
      "loss": 0.7134,
      "step": 401
    },
    {
      "epoch": 0.11042439225381129,
      "grad_norm": 0.1769459992647171,
      "learning_rate": 1.8911639411845542e-05,
      "loss": 0.863,
      "step": 402
    },
    {
      "epoch": 0.11069907979673121,
      "grad_norm": 0.2001204788684845,
      "learning_rate": 1.890889102652192e-05,
      "loss": 0.9319,
      "step": 403
    },
    {
      "epoch": 0.11097376733965114,
      "grad_norm": 0.2530067265033722,
      "learning_rate": 1.8906142641198296e-05,
      "loss": 0.9768,
      "step": 404
    },
    {
      "epoch": 0.11124845488257108,
      "grad_norm": 0.2522168457508087,
      "learning_rate": 1.8903394255874673e-05,
      "loss": 0.9447,
      "step": 405
    },
    {
      "epoch": 0.11152314242549101,
      "grad_norm": 0.23499180376529694,
      "learning_rate": 1.8900645870551054e-05,
      "loss": 1.0682,
      "step": 406
    },
    {
      "epoch": 0.11179782996841094,
      "grad_norm": 0.25343775749206543,
      "learning_rate": 1.889789748522743e-05,
      "loss": 0.881,
      "step": 407
    },
    {
      "epoch": 0.11207251751133086,
      "grad_norm": 0.23713713884353638,
      "learning_rate": 1.8895149099903808e-05,
      "loss": 1.027,
      "step": 408
    },
    {
      "epoch": 0.11234720505425079,
      "grad_norm": 0.2176835983991623,
      "learning_rate": 1.8892400714580185e-05,
      "loss": 0.8581,
      "step": 409
    },
    {
      "epoch": 0.11262189259717072,
      "grad_norm": 0.20576606690883636,
      "learning_rate": 1.8889652329256562e-05,
      "loss": 1.0259,
      "step": 410
    },
    {
      "epoch": 0.11289658014009064,
      "grad_norm": 0.17979589104652405,
      "learning_rate": 1.8886903943932942e-05,
      "loss": 0.7127,
      "step": 411
    },
    {
      "epoch": 0.11317126768301057,
      "grad_norm": 0.2202794849872589,
      "learning_rate": 1.888415555860932e-05,
      "loss": 0.902,
      "step": 412
    },
    {
      "epoch": 0.1134459552259305,
      "grad_norm": 0.3301778733730316,
      "learning_rate": 1.8881407173285696e-05,
      "loss": 1.0581,
      "step": 413
    },
    {
      "epoch": 0.11372064276885044,
      "grad_norm": 0.2804981470108032,
      "learning_rate": 1.8878658787962073e-05,
      "loss": 0.857,
      "step": 414
    },
    {
      "epoch": 0.11399533031177037,
      "grad_norm": 0.2614142596721649,
      "learning_rate": 1.887591040263845e-05,
      "loss": 0.9306,
      "step": 415
    },
    {
      "epoch": 0.11427001785469029,
      "grad_norm": 0.2429119050502777,
      "learning_rate": 1.887316201731483e-05,
      "loss": 0.8721,
      "step": 416
    },
    {
      "epoch": 0.11454470539761022,
      "grad_norm": 0.22157345712184906,
      "learning_rate": 1.8870413631991208e-05,
      "loss": 0.9658,
      "step": 417
    },
    {
      "epoch": 0.11481939294053015,
      "grad_norm": 0.21497415006160736,
      "learning_rate": 1.8867665246667585e-05,
      "loss": 0.9262,
      "step": 418
    },
    {
      "epoch": 0.11509408048345007,
      "grad_norm": 0.2083192765712738,
      "learning_rate": 1.8864916861343962e-05,
      "loss": 0.9667,
      "step": 419
    },
    {
      "epoch": 0.11536876802637,
      "grad_norm": 0.24023711681365967,
      "learning_rate": 1.886216847602034e-05,
      "loss": 0.9016,
      "step": 420
    },
    {
      "epoch": 0.11564345556928993,
      "grad_norm": 0.22409336268901825,
      "learning_rate": 1.8859420090696716e-05,
      "loss": 1.0159,
      "step": 421
    },
    {
      "epoch": 0.11591814311220987,
      "grad_norm": 0.23109453916549683,
      "learning_rate": 1.8856671705373096e-05,
      "loss": 0.8823,
      "step": 422
    },
    {
      "epoch": 0.1161928306551298,
      "grad_norm": 0.27504128217697144,
      "learning_rate": 1.8853923320049473e-05,
      "loss": 0.8851,
      "step": 423
    },
    {
      "epoch": 0.11646751819804972,
      "grad_norm": 0.29683804512023926,
      "learning_rate": 1.885117493472585e-05,
      "loss": 0.9419,
      "step": 424
    },
    {
      "epoch": 0.11674220574096965,
      "grad_norm": 0.26477527618408203,
      "learning_rate": 1.8848426549402227e-05,
      "loss": 0.7529,
      "step": 425
    },
    {
      "epoch": 0.11701689328388958,
      "grad_norm": 0.2923823595046997,
      "learning_rate": 1.8845678164078604e-05,
      "loss": 0.8767,
      "step": 426
    },
    {
      "epoch": 0.1172915808268095,
      "grad_norm": 0.2167349010705948,
      "learning_rate": 1.8842929778754985e-05,
      "loss": 0.9381,
      "step": 427
    },
    {
      "epoch": 0.11756626836972943,
      "grad_norm": 0.2131974846124649,
      "learning_rate": 1.8840181393431362e-05,
      "loss": 0.7897,
      "step": 428
    },
    {
      "epoch": 0.11784095591264936,
      "grad_norm": 0.22184516489505768,
      "learning_rate": 1.8837433008107736e-05,
      "loss": 1.0772,
      "step": 429
    },
    {
      "epoch": 0.11811564345556928,
      "grad_norm": 0.22165638208389282,
      "learning_rate": 1.8834684622784116e-05,
      "loss": 0.9561,
      "step": 430
    },
    {
      "epoch": 0.11839033099848922,
      "grad_norm": 0.25748538970947266,
      "learning_rate": 1.8831936237460493e-05,
      "loss": 0.9188,
      "step": 431
    },
    {
      "epoch": 0.11866501854140915,
      "grad_norm": 0.22586983442306519,
      "learning_rate": 1.8829187852136873e-05,
      "loss": 0.8344,
      "step": 432
    },
    {
      "epoch": 0.11893970608432908,
      "grad_norm": 0.24904757738113403,
      "learning_rate": 1.882643946681325e-05,
      "loss": 1.0379,
      "step": 433
    },
    {
      "epoch": 0.119214393627249,
      "grad_norm": 0.2936582565307617,
      "learning_rate": 1.8823691081489624e-05,
      "loss": 0.8611,
      "step": 434
    },
    {
      "epoch": 0.11948908117016893,
      "grad_norm": 0.26008275151252747,
      "learning_rate": 1.8820942696166004e-05,
      "loss": 1.0692,
      "step": 435
    },
    {
      "epoch": 0.11976376871308886,
      "grad_norm": 0.20124831795692444,
      "learning_rate": 1.881819431084238e-05,
      "loss": 0.9004,
      "step": 436
    },
    {
      "epoch": 0.12003845625600879,
      "grad_norm": 0.20466113090515137,
      "learning_rate": 1.881544592551876e-05,
      "loss": 0.8621,
      "step": 437
    },
    {
      "epoch": 0.12031314379892871,
      "grad_norm": 0.25696465373039246,
      "learning_rate": 1.881269754019514e-05,
      "loss": 0.8615,
      "step": 438
    },
    {
      "epoch": 0.12058783134184865,
      "grad_norm": 0.2348812073469162,
      "learning_rate": 1.8809949154871513e-05,
      "loss": 0.7684,
      "step": 439
    },
    {
      "epoch": 0.12086251888476858,
      "grad_norm": 0.2591266632080078,
      "learning_rate": 1.8807200769547893e-05,
      "loss": 0.8197,
      "step": 440
    },
    {
      "epoch": 0.1211372064276885,
      "grad_norm": 0.2707699239253998,
      "learning_rate": 1.880445238422427e-05,
      "loss": 1.061,
      "step": 441
    },
    {
      "epoch": 0.12141189397060843,
      "grad_norm": 0.2097112536430359,
      "learning_rate": 1.8801703998900647e-05,
      "loss": 0.7911,
      "step": 442
    },
    {
      "epoch": 0.12168658151352836,
      "grad_norm": 0.2747368812561035,
      "learning_rate": 1.8798955613577027e-05,
      "loss": 0.9617,
      "step": 443
    },
    {
      "epoch": 0.12196126905644829,
      "grad_norm": 0.2582624852657318,
      "learning_rate": 1.87962072282534e-05,
      "loss": 1.1505,
      "step": 444
    },
    {
      "epoch": 0.12223595659936821,
      "grad_norm": 0.23905441164970398,
      "learning_rate": 1.8793458842929778e-05,
      "loss": 0.726,
      "step": 445
    },
    {
      "epoch": 0.12251064414228814,
      "grad_norm": 0.23514749109745026,
      "learning_rate": 1.879071045760616e-05,
      "loss": 0.768,
      "step": 446
    },
    {
      "epoch": 0.12278533168520808,
      "grad_norm": 0.4417974650859833,
      "learning_rate": 1.8787962072282535e-05,
      "loss": 0.987,
      "step": 447
    },
    {
      "epoch": 0.12306001922812801,
      "grad_norm": 0.2699570655822754,
      "learning_rate": 1.8785213686958912e-05,
      "loss": 0.9399,
      "step": 448
    },
    {
      "epoch": 0.12333470677104794,
      "grad_norm": 0.20059294998645782,
      "learning_rate": 1.878246530163529e-05,
      "loss": 0.6581,
      "step": 449
    },
    {
      "epoch": 0.12360939431396786,
      "grad_norm": 0.19415035843849182,
      "learning_rate": 1.8779716916311667e-05,
      "loss": 0.8291,
      "step": 450
    },
    {
      "epoch": 0.12388408185688779,
      "grad_norm": 0.25061920285224915,
      "learning_rate": 1.8776968530988047e-05,
      "loss": 0.7513,
      "step": 451
    },
    {
      "epoch": 0.12415876939980772,
      "grad_norm": 0.2325860559940338,
      "learning_rate": 1.8774220145664424e-05,
      "loss": 1.0016,
      "step": 452
    },
    {
      "epoch": 0.12443345694272764,
      "grad_norm": 0.279107004404068,
      "learning_rate": 1.87714717603408e-05,
      "loss": 1.0162,
      "step": 453
    },
    {
      "epoch": 0.12470814448564757,
      "grad_norm": 0.2581306993961334,
      "learning_rate": 1.8768723375017178e-05,
      "loss": 1.1968,
      "step": 454
    },
    {
      "epoch": 0.1249828320285675,
      "grad_norm": 0.17765311896800995,
      "learning_rate": 1.8765974989693555e-05,
      "loss": 0.9275,
      "step": 455
    },
    {
      "epoch": 0.12525751957148742,
      "grad_norm": 0.25398772954940796,
      "learning_rate": 1.8763226604369935e-05,
      "loss": 0.9433,
      "step": 456
    },
    {
      "epoch": 0.12553220711440735,
      "grad_norm": 0.20465533435344696,
      "learning_rate": 1.8760478219046312e-05,
      "loss": 0.7373,
      "step": 457
    },
    {
      "epoch": 0.12580689465732728,
      "grad_norm": 0.1799350082874298,
      "learning_rate": 1.875772983372269e-05,
      "loss": 0.7277,
      "step": 458
    },
    {
      "epoch": 0.1260815822002472,
      "grad_norm": 0.34671053290367126,
      "learning_rate": 1.8754981448399066e-05,
      "loss": 0.9391,
      "step": 459
    },
    {
      "epoch": 0.12635626974316716,
      "grad_norm": 0.2670315206050873,
      "learning_rate": 1.8752233063075444e-05,
      "loss": 1.0569,
      "step": 460
    },
    {
      "epoch": 0.1266309572860871,
      "grad_norm": 0.3131506145000458,
      "learning_rate": 1.874948467775182e-05,
      "loss": 0.8661,
      "step": 461
    },
    {
      "epoch": 0.126905644829007,
      "grad_norm": 0.2903030812740326,
      "learning_rate": 1.87467362924282e-05,
      "loss": 0.8896,
      "step": 462
    },
    {
      "epoch": 0.12718033237192694,
      "grad_norm": 0.28127631545066833,
      "learning_rate": 1.8743987907104578e-05,
      "loss": 0.9746,
      "step": 463
    },
    {
      "epoch": 0.12745501991484687,
      "grad_norm": 0.19400393962860107,
      "learning_rate": 1.8741239521780955e-05,
      "loss": 0.9649,
      "step": 464
    },
    {
      "epoch": 0.1277297074577668,
      "grad_norm": 0.2746599316596985,
      "learning_rate": 1.8738491136457332e-05,
      "loss": 0.9909,
      "step": 465
    },
    {
      "epoch": 0.12800439500068672,
      "grad_norm": 0.294656902551651,
      "learning_rate": 1.873574275113371e-05,
      "loss": 0.8708,
      "step": 466
    },
    {
      "epoch": 0.12827908254360665,
      "grad_norm": 0.298637330532074,
      "learning_rate": 1.873299436581009e-05,
      "loss": 0.7627,
      "step": 467
    },
    {
      "epoch": 0.12855377008652658,
      "grad_norm": 0.2807645797729492,
      "learning_rate": 1.8730245980486466e-05,
      "loss": 0.8775,
      "step": 468
    },
    {
      "epoch": 0.1288284576294465,
      "grad_norm": 0.21301205456256866,
      "learning_rate": 1.8727497595162843e-05,
      "loss": 1.0353,
      "step": 469
    },
    {
      "epoch": 0.12910314517236643,
      "grad_norm": 0.2104647010564804,
      "learning_rate": 1.872474920983922e-05,
      "loss": 0.9076,
      "step": 470
    },
    {
      "epoch": 0.12937783271528636,
      "grad_norm": 0.21317437291145325,
      "learning_rate": 1.8722000824515598e-05,
      "loss": 0.9567,
      "step": 471
    },
    {
      "epoch": 0.12965252025820628,
      "grad_norm": 0.21871301531791687,
      "learning_rate": 1.8719252439191978e-05,
      "loss": 0.906,
      "step": 472
    },
    {
      "epoch": 0.1299272078011262,
      "grad_norm": 0.1788519322872162,
      "learning_rate": 1.8716504053868355e-05,
      "loss": 0.5625,
      "step": 473
    },
    {
      "epoch": 0.13020189534404614,
      "grad_norm": 0.3006768822669983,
      "learning_rate": 1.8713755668544732e-05,
      "loss": 1.0502,
      "step": 474
    },
    {
      "epoch": 0.13047658288696606,
      "grad_norm": 0.2541774809360504,
      "learning_rate": 1.871100728322111e-05,
      "loss": 0.8167,
      "step": 475
    },
    {
      "epoch": 0.13075127042988602,
      "grad_norm": 0.18099336326122284,
      "learning_rate": 1.8708258897897486e-05,
      "loss": 0.6602,
      "step": 476
    },
    {
      "epoch": 0.13102595797280595,
      "grad_norm": 0.23710748553276062,
      "learning_rate": 1.8705510512573863e-05,
      "loss": 0.9111,
      "step": 477
    },
    {
      "epoch": 0.13130064551572587,
      "grad_norm": 0.26162609457969666,
      "learning_rate": 1.8702762127250243e-05,
      "loss": 0.9481,
      "step": 478
    },
    {
      "epoch": 0.1315753330586458,
      "grad_norm": 0.2598304748535156,
      "learning_rate": 1.870001374192662e-05,
      "loss": 1.2586,
      "step": 479
    },
    {
      "epoch": 0.13185002060156573,
      "grad_norm": 0.23699040710926056,
      "learning_rate": 1.8697265356602997e-05,
      "loss": 0.6682,
      "step": 480
    },
    {
      "epoch": 0.13212470814448565,
      "grad_norm": 0.21577335894107819,
      "learning_rate": 1.8694516971279374e-05,
      "loss": 0.7578,
      "step": 481
    },
    {
      "epoch": 0.13239939568740558,
      "grad_norm": 0.24274682998657227,
      "learning_rate": 1.869176858595575e-05,
      "loss": 0.8331,
      "step": 482
    },
    {
      "epoch": 0.1326740832303255,
      "grad_norm": 0.252348929643631,
      "learning_rate": 1.8689020200632132e-05,
      "loss": 0.9805,
      "step": 483
    },
    {
      "epoch": 0.13294877077324543,
      "grad_norm": 0.23001180589199066,
      "learning_rate": 1.868627181530851e-05,
      "loss": 0.7122,
      "step": 484
    },
    {
      "epoch": 0.13322345831616536,
      "grad_norm": 0.2690826654434204,
      "learning_rate": 1.8683523429984886e-05,
      "loss": 1.1991,
      "step": 485
    },
    {
      "epoch": 0.1334981458590853,
      "grad_norm": 0.37251466512680054,
      "learning_rate": 1.8680775044661263e-05,
      "loss": 0.9591,
      "step": 486
    },
    {
      "epoch": 0.13377283340200521,
      "grad_norm": 0.19780333340168,
      "learning_rate": 1.867802665933764e-05,
      "loss": 1.0179,
      "step": 487
    },
    {
      "epoch": 0.13404752094492514,
      "grad_norm": 0.3556842803955078,
      "learning_rate": 1.867527827401402e-05,
      "loss": 0.8388,
      "step": 488
    },
    {
      "epoch": 0.13432220848784507,
      "grad_norm": 0.2255561500787735,
      "learning_rate": 1.8672529888690397e-05,
      "loss": 0.7686,
      "step": 489
    },
    {
      "epoch": 0.134596896030765,
      "grad_norm": 0.2652037441730499,
      "learning_rate": 1.8669781503366774e-05,
      "loss": 0.9311,
      "step": 490
    },
    {
      "epoch": 0.13487158357368492,
      "grad_norm": 0.2757638394832611,
      "learning_rate": 1.866703311804315e-05,
      "loss": 0.8642,
      "step": 491
    },
    {
      "epoch": 0.13514627111660485,
      "grad_norm": 0.19367769360542297,
      "learning_rate": 1.866428473271953e-05,
      "loss": 0.7407,
      "step": 492
    },
    {
      "epoch": 0.1354209586595248,
      "grad_norm": 0.2865351438522339,
      "learning_rate": 1.8661536347395906e-05,
      "loss": 1.1408,
      "step": 493
    },
    {
      "epoch": 0.13569564620244473,
      "grad_norm": 0.31574440002441406,
      "learning_rate": 1.8658787962072286e-05,
      "loss": 1.0371,
      "step": 494
    },
    {
      "epoch": 0.13597033374536466,
      "grad_norm": 0.32998353242874146,
      "learning_rate": 1.8656039576748663e-05,
      "loss": 0.9868,
      "step": 495
    },
    {
      "epoch": 0.13624502128828458,
      "grad_norm": 0.24652743339538574,
      "learning_rate": 1.865329119142504e-05,
      "loss": 0.9781,
      "step": 496
    },
    {
      "epoch": 0.1365197088312045,
      "grad_norm": 0.26320403814315796,
      "learning_rate": 1.8650542806101417e-05,
      "loss": 0.861,
      "step": 497
    },
    {
      "epoch": 0.13679439637412444,
      "grad_norm": 0.30732235312461853,
      "learning_rate": 1.8647794420777794e-05,
      "loss": 1.0222,
      "step": 498
    },
    {
      "epoch": 0.13706908391704437,
      "grad_norm": 0.31340765953063965,
      "learning_rate": 1.8645046035454174e-05,
      "loss": 1.1556,
      "step": 499
    },
    {
      "epoch": 0.1373437714599643,
      "grad_norm": 0.31222808361053467,
      "learning_rate": 1.8642297650130548e-05,
      "loss": 1.1087,
      "step": 500
    },
    {
      "epoch": 0.13761845900288422,
      "grad_norm": 0.24171403050422668,
      "learning_rate": 1.8639549264806925e-05,
      "loss": 0.7595,
      "step": 501
    },
    {
      "epoch": 0.13789314654580415,
      "grad_norm": 0.4138529598712921,
      "learning_rate": 1.8636800879483305e-05,
      "loss": 1.1153,
      "step": 502
    },
    {
      "epoch": 0.13816783408872407,
      "grad_norm": 0.275631844997406,
      "learning_rate": 1.8634052494159683e-05,
      "loss": 0.7963,
      "step": 503
    },
    {
      "epoch": 0.138442521631644,
      "grad_norm": 0.2577419877052307,
      "learning_rate": 1.8631304108836063e-05,
      "loss": 0.8805,
      "step": 504
    },
    {
      "epoch": 0.13871720917456393,
      "grad_norm": 0.2504889965057373,
      "learning_rate": 1.8628555723512437e-05,
      "loss": 1.1017,
      "step": 505
    },
    {
      "epoch": 0.13899189671748385,
      "grad_norm": 0.29448580741882324,
      "learning_rate": 1.8625807338188814e-05,
      "loss": 0.9054,
      "step": 506
    },
    {
      "epoch": 0.13926658426040378,
      "grad_norm": 0.23002640902996063,
      "learning_rate": 1.8623058952865194e-05,
      "loss": 0.8638,
      "step": 507
    },
    {
      "epoch": 0.1395412718033237,
      "grad_norm": 0.27383124828338623,
      "learning_rate": 1.862031056754157e-05,
      "loss": 0.7134,
      "step": 508
    },
    {
      "epoch": 0.13981595934624363,
      "grad_norm": 0.3101145625114441,
      "learning_rate": 1.8617562182217948e-05,
      "loss": 0.8292,
      "step": 509
    },
    {
      "epoch": 0.1400906468891636,
      "grad_norm": 0.26085567474365234,
      "learning_rate": 1.8614813796894325e-05,
      "loss": 0.9792,
      "step": 510
    },
    {
      "epoch": 0.14036533443208352,
      "grad_norm": 0.3337884545326233,
      "learning_rate": 1.8612065411570702e-05,
      "loss": 1.0475,
      "step": 511
    },
    {
      "epoch": 0.14064002197500344,
      "grad_norm": 0.29944074153900146,
      "learning_rate": 1.8609317026247082e-05,
      "loss": 0.7612,
      "step": 512
    },
    {
      "epoch": 0.14091470951792337,
      "grad_norm": 0.18158505856990814,
      "learning_rate": 1.860656864092346e-05,
      "loss": 0.7925,
      "step": 513
    },
    {
      "epoch": 0.1411893970608433,
      "grad_norm": 0.26668816804885864,
      "learning_rate": 1.8603820255599837e-05,
      "loss": 1.0461,
      "step": 514
    },
    {
      "epoch": 0.14146408460376322,
      "grad_norm": 0.27571502327919006,
      "learning_rate": 1.8601071870276214e-05,
      "loss": 0.9462,
      "step": 515
    },
    {
      "epoch": 0.14173877214668315,
      "grad_norm": 0.27844569087028503,
      "learning_rate": 1.859832348495259e-05,
      "loss": 0.9894,
      "step": 516
    },
    {
      "epoch": 0.14201345968960308,
      "grad_norm": 0.3059858977794647,
      "learning_rate": 1.8595575099628968e-05,
      "loss": 0.8796,
      "step": 517
    },
    {
      "epoch": 0.142288147232523,
      "grad_norm": 0.26401078701019287,
      "learning_rate": 1.8592826714305348e-05,
      "loss": 1.0515,
      "step": 518
    },
    {
      "epoch": 0.14256283477544293,
      "grad_norm": 0.24275195598602295,
      "learning_rate": 1.8590078328981725e-05,
      "loss": 0.8407,
      "step": 519
    },
    {
      "epoch": 0.14283752231836286,
      "grad_norm": 0.3123442232608795,
      "learning_rate": 1.8587329943658102e-05,
      "loss": 1.0378,
      "step": 520
    },
    {
      "epoch": 0.14311220986128279,
      "grad_norm": 0.273445725440979,
      "learning_rate": 1.858458155833448e-05,
      "loss": 0.9748,
      "step": 521
    },
    {
      "epoch": 0.1433868974042027,
      "grad_norm": 0.27950888872146606,
      "learning_rate": 1.8581833173010856e-05,
      "loss": 0.9384,
      "step": 522
    },
    {
      "epoch": 0.14366158494712264,
      "grad_norm": 0.2604884207248688,
      "learning_rate": 1.8579084787687236e-05,
      "loss": 0.9113,
      "step": 523
    },
    {
      "epoch": 0.14393627249004257,
      "grad_norm": 0.24848584830760956,
      "learning_rate": 1.8576336402363613e-05,
      "loss": 0.8174,
      "step": 524
    },
    {
      "epoch": 0.1442109600329625,
      "grad_norm": 0.36588430404663086,
      "learning_rate": 1.857358801703999e-05,
      "loss": 1.1302,
      "step": 525
    },
    {
      "epoch": 0.14448564757588242,
      "grad_norm": 0.27987441420555115,
      "learning_rate": 1.8570839631716368e-05,
      "loss": 0.9502,
      "step": 526
    },
    {
      "epoch": 0.14476033511880237,
      "grad_norm": 0.26083073019981384,
      "learning_rate": 1.8568091246392745e-05,
      "loss": 0.8542,
      "step": 527
    },
    {
      "epoch": 0.1450350226617223,
      "grad_norm": 0.3805755078792572,
      "learning_rate": 1.8565342861069125e-05,
      "loss": 0.9536,
      "step": 528
    },
    {
      "epoch": 0.14530971020464223,
      "grad_norm": 0.30252256989479065,
      "learning_rate": 1.8562594475745502e-05,
      "loss": 0.8612,
      "step": 529
    },
    {
      "epoch": 0.14558439774756216,
      "grad_norm": 0.2916070520877838,
      "learning_rate": 1.855984609042188e-05,
      "loss": 0.9729,
      "step": 530
    },
    {
      "epoch": 0.14585908529048208,
      "grad_norm": 0.3079086244106293,
      "learning_rate": 1.8557097705098256e-05,
      "loss": 0.8514,
      "step": 531
    },
    {
      "epoch": 0.146133772833402,
      "grad_norm": 0.2451677769422531,
      "learning_rate": 1.8554349319774633e-05,
      "loss": 0.8516,
      "step": 532
    },
    {
      "epoch": 0.14640846037632194,
      "grad_norm": 0.21859918534755707,
      "learning_rate": 1.855160093445101e-05,
      "loss": 0.9723,
      "step": 533
    },
    {
      "epoch": 0.14668314791924186,
      "grad_norm": 0.24152825772762299,
      "learning_rate": 1.854885254912739e-05,
      "loss": 1.1494,
      "step": 534
    },
    {
      "epoch": 0.1469578354621618,
      "grad_norm": 0.2671487331390381,
      "learning_rate": 1.8546104163803768e-05,
      "loss": 0.9946,
      "step": 535
    },
    {
      "epoch": 0.14723252300508172,
      "grad_norm": 0.26619958877563477,
      "learning_rate": 1.8543355778480145e-05,
      "loss": 0.9284,
      "step": 536
    },
    {
      "epoch": 0.14750721054800164,
      "grad_norm": 0.27628570795059204,
      "learning_rate": 1.854060739315652e-05,
      "loss": 1.1446,
      "step": 537
    },
    {
      "epoch": 0.14778189809092157,
      "grad_norm": 0.30042463541030884,
      "learning_rate": 1.85378590078329e-05,
      "loss": 0.9798,
      "step": 538
    },
    {
      "epoch": 0.1480565856338415,
      "grad_norm": 0.2423877716064453,
      "learning_rate": 1.853511062250928e-05,
      "loss": 0.8297,
      "step": 539
    },
    {
      "epoch": 0.14833127317676142,
      "grad_norm": 0.29137903451919556,
      "learning_rate": 1.8532362237185656e-05,
      "loss": 0.7448,
      "step": 540
    },
    {
      "epoch": 0.14860596071968135,
      "grad_norm": 0.19852769374847412,
      "learning_rate": 1.8529613851862033e-05,
      "loss": 0.5587,
      "step": 541
    },
    {
      "epoch": 0.14888064826260128,
      "grad_norm": 0.3014676570892334,
      "learning_rate": 1.852686546653841e-05,
      "loss": 0.6472,
      "step": 542
    },
    {
      "epoch": 0.14915533580552123,
      "grad_norm": 0.375203400850296,
      "learning_rate": 1.8524117081214787e-05,
      "loss": 0.9647,
      "step": 543
    },
    {
      "epoch": 0.14943002334844116,
      "grad_norm": 0.3352760672569275,
      "learning_rate": 1.8521368695891167e-05,
      "loss": 0.9213,
      "step": 544
    },
    {
      "epoch": 0.1497047108913611,
      "grad_norm": 0.2952688932418823,
      "learning_rate": 1.8518620310567544e-05,
      "loss": 0.8027,
      "step": 545
    },
    {
      "epoch": 0.149979398434281,
      "grad_norm": 0.31546926498413086,
      "learning_rate": 1.851587192524392e-05,
      "loss": 0.9997,
      "step": 546
    },
    {
      "epoch": 0.15025408597720094,
      "grad_norm": 0.2087523639202118,
      "learning_rate": 1.85131235399203e-05,
      "loss": 0.7749,
      "step": 547
    },
    {
      "epoch": 0.15052877352012087,
      "grad_norm": 0.35741350054740906,
      "learning_rate": 1.8510375154596676e-05,
      "loss": 1.0029,
      "step": 548
    },
    {
      "epoch": 0.1508034610630408,
      "grad_norm": 0.255536824464798,
      "learning_rate": 1.8507626769273053e-05,
      "loss": 0.6903,
      "step": 549
    },
    {
      "epoch": 0.15107814860596072,
      "grad_norm": 0.23056337237358093,
      "learning_rate": 1.8504878383949433e-05,
      "loss": 0.7713,
      "step": 550
    },
    {
      "epoch": 0.15135283614888065,
      "grad_norm": 0.25519782304763794,
      "learning_rate": 1.850212999862581e-05,
      "loss": 0.8315,
      "step": 551
    },
    {
      "epoch": 0.15162752369180058,
      "grad_norm": 0.2869972288608551,
      "learning_rate": 1.8499381613302187e-05,
      "loss": 0.9231,
      "step": 552
    },
    {
      "epoch": 0.1519022112347205,
      "grad_norm": 0.3669261634349823,
      "learning_rate": 1.8496633227978564e-05,
      "loss": 0.7836,
      "step": 553
    },
    {
      "epoch": 0.15217689877764043,
      "grad_norm": 0.32073289155960083,
      "learning_rate": 1.849388484265494e-05,
      "loss": 0.8545,
      "step": 554
    },
    {
      "epoch": 0.15245158632056036,
      "grad_norm": 0.26832064986228943,
      "learning_rate": 1.849113645733132e-05,
      "loss": 1.0733,
      "step": 555
    },
    {
      "epoch": 0.15272627386348028,
      "grad_norm": 0.2760774493217468,
      "learning_rate": 1.84883880720077e-05,
      "loss": 0.9903,
      "step": 556
    },
    {
      "epoch": 0.1530009614064002,
      "grad_norm": 0.2336362600326538,
      "learning_rate": 1.8485639686684072e-05,
      "loss": 0.7916,
      "step": 557
    },
    {
      "epoch": 0.15327564894932014,
      "grad_norm": 0.20207110047340393,
      "learning_rate": 1.8482891301360453e-05,
      "loss": 0.694,
      "step": 558
    },
    {
      "epoch": 0.15355033649224006,
      "grad_norm": 0.226780503988266,
      "learning_rate": 1.848014291603683e-05,
      "loss": 0.8028,
      "step": 559
    },
    {
      "epoch": 0.15382502403516002,
      "grad_norm": 0.35357192158699036,
      "learning_rate": 1.847739453071321e-05,
      "loss": 0.9702,
      "step": 560
    },
    {
      "epoch": 0.15409971157807995,
      "grad_norm": 0.19573193788528442,
      "learning_rate": 1.8474646145389587e-05,
      "loss": 0.8277,
      "step": 561
    },
    {
      "epoch": 0.15437439912099987,
      "grad_norm": 0.27062803506851196,
      "learning_rate": 1.847189776006596e-05,
      "loss": 1.0763,
      "step": 562
    },
    {
      "epoch": 0.1546490866639198,
      "grad_norm": 0.24779543280601501,
      "learning_rate": 1.846914937474234e-05,
      "loss": 0.6068,
      "step": 563
    },
    {
      "epoch": 0.15492377420683973,
      "grad_norm": 0.25140437483787537,
      "learning_rate": 1.8466400989418718e-05,
      "loss": 0.9198,
      "step": 564
    },
    {
      "epoch": 0.15519846174975965,
      "grad_norm": 0.283179372549057,
      "learning_rate": 1.8463652604095095e-05,
      "loss": 1.0677,
      "step": 565
    },
    {
      "epoch": 0.15547314929267958,
      "grad_norm": 0.2933553457260132,
      "learning_rate": 1.8460904218771472e-05,
      "loss": 0.8258,
      "step": 566
    },
    {
      "epoch": 0.1557478368355995,
      "grad_norm": 0.24335375428199768,
      "learning_rate": 1.845815583344785e-05,
      "loss": 0.7699,
      "step": 567
    },
    {
      "epoch": 0.15602252437851943,
      "grad_norm": 0.23881500959396362,
      "learning_rate": 1.845540744812423e-05,
      "loss": 1.2124,
      "step": 568
    },
    {
      "epoch": 0.15629721192143936,
      "grad_norm": 0.20414380729198456,
      "learning_rate": 1.8452659062800607e-05,
      "loss": 1.0291,
      "step": 569
    },
    {
      "epoch": 0.1565718994643593,
      "grad_norm": 0.258512407541275,
      "learning_rate": 1.8449910677476984e-05,
      "loss": 0.9382,
      "step": 570
    },
    {
      "epoch": 0.15684658700727921,
      "grad_norm": 0.3129018247127533,
      "learning_rate": 1.844716229215336e-05,
      "loss": 0.9572,
      "step": 571
    },
    {
      "epoch": 0.15712127455019914,
      "grad_norm": 0.24401775002479553,
      "learning_rate": 1.8444413906829738e-05,
      "loss": 0.8822,
      "step": 572
    },
    {
      "epoch": 0.15739596209311907,
      "grad_norm": 0.2823740839958191,
      "learning_rate": 1.8441665521506115e-05,
      "loss": 1.1145,
      "step": 573
    },
    {
      "epoch": 0.157670649636039,
      "grad_norm": 0.23666821420192719,
      "learning_rate": 1.8438917136182495e-05,
      "loss": 0.6785,
      "step": 574
    },
    {
      "epoch": 0.15794533717895892,
      "grad_norm": 0.4498724043369293,
      "learning_rate": 1.8436168750858872e-05,
      "loss": 0.853,
      "step": 575
    },
    {
      "epoch": 0.15822002472187885,
      "grad_norm": 0.28817686438560486,
      "learning_rate": 1.843342036553525e-05,
      "loss": 0.8096,
      "step": 576
    },
    {
      "epoch": 0.1584947122647988,
      "grad_norm": 0.27882474660873413,
      "learning_rate": 1.8430671980211626e-05,
      "loss": 0.7945,
      "step": 577
    },
    {
      "epoch": 0.15876939980771873,
      "grad_norm": 0.2442457675933838,
      "learning_rate": 1.8427923594888003e-05,
      "loss": 0.9249,
      "step": 578
    },
    {
      "epoch": 0.15904408735063866,
      "grad_norm": 0.3320907652378082,
      "learning_rate": 1.8425175209564384e-05,
      "loss": 0.8767,
      "step": 579
    },
    {
      "epoch": 0.15931877489355858,
      "grad_norm": 0.3132637143135071,
      "learning_rate": 1.842242682424076e-05,
      "loss": 0.7401,
      "step": 580
    },
    {
      "epoch": 0.1595934624364785,
      "grad_norm": 0.3782828152179718,
      "learning_rate": 1.8419678438917138e-05,
      "loss": 0.8497,
      "step": 581
    },
    {
      "epoch": 0.15986814997939844,
      "grad_norm": 0.26013556122779846,
      "learning_rate": 1.8416930053593515e-05,
      "loss": 0.7716,
      "step": 582
    },
    {
      "epoch": 0.16014283752231837,
      "grad_norm": 0.382529079914093,
      "learning_rate": 1.841418166826989e-05,
      "loss": 0.9789,
      "step": 583
    },
    {
      "epoch": 0.1604175250652383,
      "grad_norm": 0.22469781339168549,
      "learning_rate": 1.8411433282946272e-05,
      "loss": 0.7262,
      "step": 584
    },
    {
      "epoch": 0.16069221260815822,
      "grad_norm": 0.23357030749320984,
      "learning_rate": 1.840868489762265e-05,
      "loss": 0.766,
      "step": 585
    },
    {
      "epoch": 0.16096690015107815,
      "grad_norm": 0.2392839938402176,
      "learning_rate": 1.8405936512299026e-05,
      "loss": 0.6975,
      "step": 586
    },
    {
      "epoch": 0.16124158769399807,
      "grad_norm": 0.3278224468231201,
      "learning_rate": 1.8403188126975403e-05,
      "loss": 0.7475,
      "step": 587
    },
    {
      "epoch": 0.161516275236918,
      "grad_norm": 0.2898288667201996,
      "learning_rate": 1.840043974165178e-05,
      "loss": 0.7987,
      "step": 588
    },
    {
      "epoch": 0.16179096277983793,
      "grad_norm": 0.34589558839797974,
      "learning_rate": 1.8397691356328157e-05,
      "loss": 0.887,
      "step": 589
    },
    {
      "epoch": 0.16206565032275785,
      "grad_norm": 0.24613259732723236,
      "learning_rate": 1.8394942971004538e-05,
      "loss": 0.8931,
      "step": 590
    },
    {
      "epoch": 0.16234033786567778,
      "grad_norm": 0.29850825667381287,
      "learning_rate": 1.8392194585680915e-05,
      "loss": 0.82,
      "step": 591
    },
    {
      "epoch": 0.1626150254085977,
      "grad_norm": 0.24874812364578247,
      "learning_rate": 1.838944620035729e-05,
      "loss": 0.7909,
      "step": 592
    },
    {
      "epoch": 0.16288971295151766,
      "grad_norm": 0.3203515410423279,
      "learning_rate": 1.838669781503367e-05,
      "loss": 0.8834,
      "step": 593
    },
    {
      "epoch": 0.1631644004944376,
      "grad_norm": 0.21578067541122437,
      "learning_rate": 1.8383949429710046e-05,
      "loss": 0.636,
      "step": 594
    },
    {
      "epoch": 0.16343908803735752,
      "grad_norm": 0.3225516378879547,
      "learning_rate": 1.8381201044386426e-05,
      "loss": 1.0414,
      "step": 595
    },
    {
      "epoch": 0.16371377558027744,
      "grad_norm": 0.3256843388080597,
      "learning_rate": 1.8378452659062803e-05,
      "loss": 1.0784,
      "step": 596
    },
    {
      "epoch": 0.16398846312319737,
      "grad_norm": 0.2766369581222534,
      "learning_rate": 1.837570427373918e-05,
      "loss": 0.777,
      "step": 597
    },
    {
      "epoch": 0.1642631506661173,
      "grad_norm": 0.24921578168869019,
      "learning_rate": 1.8372955888415557e-05,
      "loss": 0.7235,
      "step": 598
    },
    {
      "epoch": 0.16453783820903722,
      "grad_norm": 0.29504579305648804,
      "learning_rate": 1.8370207503091934e-05,
      "loss": 0.6029,
      "step": 599
    },
    {
      "epoch": 0.16481252575195715,
      "grad_norm": 0.27804693579673767,
      "learning_rate": 1.8367459117768315e-05,
      "loss": 0.6821,
      "step": 600
    },
    {
      "epoch": 0.16508721329487708,
      "grad_norm": 0.3019275963306427,
      "learning_rate": 1.836471073244469e-05,
      "loss": 0.734,
      "step": 601
    },
    {
      "epoch": 0.165361900837797,
      "grad_norm": 0.26158076524734497,
      "learning_rate": 1.836196234712107e-05,
      "loss": 0.7609,
      "step": 602
    },
    {
      "epoch": 0.16563658838071693,
      "grad_norm": 0.21532940864562988,
      "learning_rate": 1.8359213961797446e-05,
      "loss": 0.7901,
      "step": 603
    },
    {
      "epoch": 0.16591127592363686,
      "grad_norm": 0.2506101429462433,
      "learning_rate": 1.8356465576473823e-05,
      "loss": 0.9156,
      "step": 604
    },
    {
      "epoch": 0.16618596346655679,
      "grad_norm": 0.34476438164711,
      "learning_rate": 1.83537171911502e-05,
      "loss": 0.7018,
      "step": 605
    },
    {
      "epoch": 0.1664606510094767,
      "grad_norm": 0.3356035649776459,
      "learning_rate": 1.835096880582658e-05,
      "loss": 0.8648,
      "step": 606
    },
    {
      "epoch": 0.16673533855239664,
      "grad_norm": 0.30837973952293396,
      "learning_rate": 1.8348220420502957e-05,
      "loss": 0.7861,
      "step": 607
    },
    {
      "epoch": 0.16701002609531657,
      "grad_norm": 0.2889169752597809,
      "learning_rate": 1.8345472035179334e-05,
      "loss": 0.8746,
      "step": 608
    },
    {
      "epoch": 0.1672847136382365,
      "grad_norm": 0.3531164228916168,
      "learning_rate": 1.834272364985571e-05,
      "loss": 0.9603,
      "step": 609
    },
    {
      "epoch": 0.16755940118115645,
      "grad_norm": 0.298837810754776,
      "learning_rate": 1.8339975264532088e-05,
      "loss": 1.0424,
      "step": 610
    },
    {
      "epoch": 0.16783408872407637,
      "grad_norm": 0.2255786657333374,
      "learning_rate": 1.833722687920847e-05,
      "loss": 0.5799,
      "step": 611
    },
    {
      "epoch": 0.1681087762669963,
      "grad_norm": 0.22079892456531525,
      "learning_rate": 1.8334478493884846e-05,
      "loss": 0.9109,
      "step": 612
    },
    {
      "epoch": 0.16838346380991623,
      "grad_norm": 0.3381830155849457,
      "learning_rate": 1.8331730108561223e-05,
      "loss": 0.9535,
      "step": 613
    },
    {
      "epoch": 0.16865815135283616,
      "grad_norm": 0.3505811095237732,
      "learning_rate": 1.83289817232376e-05,
      "loss": 0.9544,
      "step": 614
    },
    {
      "epoch": 0.16893283889575608,
      "grad_norm": 0.3550000786781311,
      "learning_rate": 1.8326233337913977e-05,
      "loss": 0.9735,
      "step": 615
    },
    {
      "epoch": 0.169207526438676,
      "grad_norm": 0.30557557940483093,
      "learning_rate": 1.8323484952590357e-05,
      "loss": 0.8587,
      "step": 616
    },
    {
      "epoch": 0.16948221398159594,
      "grad_norm": 0.3053141236305237,
      "learning_rate": 1.8320736567266734e-05,
      "loss": 1.1434,
      "step": 617
    },
    {
      "epoch": 0.16975690152451586,
      "grad_norm": 0.3229580223560333,
      "learning_rate": 1.8317988181943108e-05,
      "loss": 0.8749,
      "step": 618
    },
    {
      "epoch": 0.1700315890674358,
      "grad_norm": 0.22379010915756226,
      "learning_rate": 1.8315239796619488e-05,
      "loss": 0.8752,
      "step": 619
    },
    {
      "epoch": 0.17030627661035572,
      "grad_norm": 0.3309415876865387,
      "learning_rate": 1.8312491411295865e-05,
      "loss": 0.7022,
      "step": 620
    },
    {
      "epoch": 0.17058096415327564,
      "grad_norm": 0.19306893646717072,
      "learning_rate": 1.8309743025972242e-05,
      "loss": 0.5886,
      "step": 621
    },
    {
      "epoch": 0.17085565169619557,
      "grad_norm": 0.25453290343284607,
      "learning_rate": 1.8306994640648623e-05,
      "loss": 0.8534,
      "step": 622
    },
    {
      "epoch": 0.1711303392391155,
      "grad_norm": 0.3024492561817169,
      "learning_rate": 1.8304246255324996e-05,
      "loss": 0.9007,
      "step": 623
    },
    {
      "epoch": 0.17140502678203542,
      "grad_norm": 0.279197096824646,
      "learning_rate": 1.8301497870001377e-05,
      "loss": 0.9165,
      "step": 624
    },
    {
      "epoch": 0.17167971432495535,
      "grad_norm": 0.30749595165252686,
      "learning_rate": 1.8298749484677754e-05,
      "loss": 1.1949,
      "step": 625
    },
    {
      "epoch": 0.17195440186787528,
      "grad_norm": 0.3016921877861023,
      "learning_rate": 1.829600109935413e-05,
      "loss": 1.1691,
      "step": 626
    },
    {
      "epoch": 0.17222908941079523,
      "grad_norm": 0.29726776480674744,
      "learning_rate": 1.829325271403051e-05,
      "loss": 0.7634,
      "step": 627
    },
    {
      "epoch": 0.17250377695371516,
      "grad_norm": 0.24442657828330994,
      "learning_rate": 1.8290504328706885e-05,
      "loss": 0.9061,
      "step": 628
    },
    {
      "epoch": 0.1727784644966351,
      "grad_norm": 0.3107200264930725,
      "learning_rate": 1.828775594338326e-05,
      "loss": 0.9297,
      "step": 629
    },
    {
      "epoch": 0.173053152039555,
      "grad_norm": 0.27311021089553833,
      "learning_rate": 1.8285007558059642e-05,
      "loss": 0.9829,
      "step": 630
    },
    {
      "epoch": 0.17332783958247494,
      "grad_norm": 0.28850460052490234,
      "learning_rate": 1.828225917273602e-05,
      "loss": 1.0178,
      "step": 631
    },
    {
      "epoch": 0.17360252712539487,
      "grad_norm": 0.23773612082004547,
      "learning_rate": 1.82795107874124e-05,
      "loss": 0.7666,
      "step": 632
    },
    {
      "epoch": 0.1738772146683148,
      "grad_norm": 0.26956796646118164,
      "learning_rate": 1.8276762402088773e-05,
      "loss": 0.8042,
      "step": 633
    },
    {
      "epoch": 0.17415190221123472,
      "grad_norm": 0.33031603693962097,
      "learning_rate": 1.827401401676515e-05,
      "loss": 1.0109,
      "step": 634
    },
    {
      "epoch": 0.17442658975415465,
      "grad_norm": 0.1961609572172165,
      "learning_rate": 1.827126563144153e-05,
      "loss": 0.5814,
      "step": 635
    },
    {
      "epoch": 0.17470127729707458,
      "grad_norm": 0.23148740828037262,
      "learning_rate": 1.8268517246117908e-05,
      "loss": 1.1016,
      "step": 636
    },
    {
      "epoch": 0.1749759648399945,
      "grad_norm": 0.33485981822013855,
      "learning_rate": 1.8265768860794285e-05,
      "loss": 0.7394,
      "step": 637
    },
    {
      "epoch": 0.17525065238291443,
      "grad_norm": 0.3546698987483978,
      "learning_rate": 1.826302047547066e-05,
      "loss": 0.8423,
      "step": 638
    },
    {
      "epoch": 0.17552533992583436,
      "grad_norm": 0.23223799467086792,
      "learning_rate": 1.826027209014704e-05,
      "loss": 0.6932,
      "step": 639
    },
    {
      "epoch": 0.17580002746875428,
      "grad_norm": 0.3167639970779419,
      "learning_rate": 1.8257523704823416e-05,
      "loss": 0.7991,
      "step": 640
    },
    {
      "epoch": 0.1760747150116742,
      "grad_norm": 0.2726075053215027,
      "learning_rate": 1.8254775319499796e-05,
      "loss": 0.6298,
      "step": 641
    },
    {
      "epoch": 0.17634940255459414,
      "grad_norm": 0.3439108729362488,
      "learning_rate": 1.8252026934176173e-05,
      "loss": 0.6915,
      "step": 642
    },
    {
      "epoch": 0.1766240900975141,
      "grad_norm": 0.28645139932632446,
      "learning_rate": 1.824927854885255e-05,
      "loss": 0.9419,
      "step": 643
    },
    {
      "epoch": 0.17689877764043402,
      "grad_norm": 0.2851347327232361,
      "learning_rate": 1.8246530163528927e-05,
      "loss": 0.9286,
      "step": 644
    },
    {
      "epoch": 0.17717346518335395,
      "grad_norm": 0.3159637451171875,
      "learning_rate": 1.8243781778205304e-05,
      "loss": 1.0365,
      "step": 645
    },
    {
      "epoch": 0.17744815272627387,
      "grad_norm": 0.35052335262298584,
      "learning_rate": 1.8241033392881685e-05,
      "loss": 0.8516,
      "step": 646
    },
    {
      "epoch": 0.1777228402691938,
      "grad_norm": 0.2679232060909271,
      "learning_rate": 1.823828500755806e-05,
      "loss": 0.8337,
      "step": 647
    },
    {
      "epoch": 0.17799752781211373,
      "grad_norm": 0.3186495900154114,
      "learning_rate": 1.823553662223444e-05,
      "loss": 0.7104,
      "step": 648
    },
    {
      "epoch": 0.17827221535503365,
      "grad_norm": 0.28310537338256836,
      "learning_rate": 1.8232788236910816e-05,
      "loss": 0.9734,
      "step": 649
    },
    {
      "epoch": 0.17854690289795358,
      "grad_norm": 0.28515127301216125,
      "learning_rate": 1.8230039851587193e-05,
      "loss": 0.7335,
      "step": 650
    },
    {
      "epoch": 0.1788215904408735,
      "grad_norm": 0.23006613552570343,
      "learning_rate": 1.8227291466263573e-05,
      "loss": 0.7567,
      "step": 651
    },
    {
      "epoch": 0.17909627798379343,
      "grad_norm": 0.3793664872646332,
      "learning_rate": 1.822454308093995e-05,
      "loss": 0.9324,
      "step": 652
    },
    {
      "epoch": 0.17937096552671336,
      "grad_norm": 0.29529643058776855,
      "learning_rate": 1.8221794695616327e-05,
      "loss": 0.993,
      "step": 653
    },
    {
      "epoch": 0.1796456530696333,
      "grad_norm": 0.3387726843357086,
      "learning_rate": 1.8219046310292704e-05,
      "loss": 0.9085,
      "step": 654
    },
    {
      "epoch": 0.17992034061255321,
      "grad_norm": 0.3196391463279724,
      "learning_rate": 1.821629792496908e-05,
      "loss": 0.9593,
      "step": 655
    },
    {
      "epoch": 0.18019502815547314,
      "grad_norm": 0.24776925146579742,
      "learning_rate": 1.8213549539645458e-05,
      "loss": 0.9938,
      "step": 656
    },
    {
      "epoch": 0.18046971569839307,
      "grad_norm": 0.3105149269104004,
      "learning_rate": 1.821080115432184e-05,
      "loss": 0.7691,
      "step": 657
    },
    {
      "epoch": 0.180744403241313,
      "grad_norm": 0.29543790221214294,
      "learning_rate": 1.8208052768998216e-05,
      "loss": 0.8799,
      "step": 658
    },
    {
      "epoch": 0.18101909078423292,
      "grad_norm": 0.29265451431274414,
      "learning_rate": 1.8205304383674593e-05,
      "loss": 1.1637,
      "step": 659
    },
    {
      "epoch": 0.18129377832715288,
      "grad_norm": 0.24893561005592346,
      "learning_rate": 1.820255599835097e-05,
      "loss": 1.1774,
      "step": 660
    },
    {
      "epoch": 0.1815684658700728,
      "grad_norm": 0.24477779865264893,
      "learning_rate": 1.8199807613027347e-05,
      "loss": 0.8307,
      "step": 661
    },
    {
      "epoch": 0.18184315341299273,
      "grad_norm": 0.293477326631546,
      "learning_rate": 1.8197059227703727e-05,
      "loss": 0.9774,
      "step": 662
    },
    {
      "epoch": 0.18211784095591266,
      "grad_norm": 0.31927239894866943,
      "learning_rate": 1.8194310842380104e-05,
      "loss": 0.9355,
      "step": 663
    },
    {
      "epoch": 0.18239252849883258,
      "grad_norm": 0.23482821881771088,
      "learning_rate": 1.819156245705648e-05,
      "loss": 0.8314,
      "step": 664
    },
    {
      "epoch": 0.1826672160417525,
      "grad_norm": 0.23910771310329437,
      "learning_rate": 1.8188814071732858e-05,
      "loss": 0.8261,
      "step": 665
    },
    {
      "epoch": 0.18294190358467244,
      "grad_norm": 0.2613186836242676,
      "learning_rate": 1.8186065686409235e-05,
      "loss": 0.9062,
      "step": 666
    },
    {
      "epoch": 0.18321659112759237,
      "grad_norm": 0.37584325671195984,
      "learning_rate": 1.8183317301085616e-05,
      "loss": 0.9488,
      "step": 667
    },
    {
      "epoch": 0.1834912786705123,
      "grad_norm": 0.34097862243652344,
      "learning_rate": 1.8180568915761993e-05,
      "loss": 0.9981,
      "step": 668
    },
    {
      "epoch": 0.18376596621343222,
      "grad_norm": 0.25237393379211426,
      "learning_rate": 1.817782053043837e-05,
      "loss": 0.9564,
      "step": 669
    },
    {
      "epoch": 0.18404065375635215,
      "grad_norm": 0.29327550530433655,
      "learning_rate": 1.8175072145114747e-05,
      "loss": 0.8953,
      "step": 670
    },
    {
      "epoch": 0.18431534129927207,
      "grad_norm": 0.35314878821372986,
      "learning_rate": 1.8172323759791124e-05,
      "loss": 1.0202,
      "step": 671
    },
    {
      "epoch": 0.184590028842192,
      "grad_norm": 0.32555294036865234,
      "learning_rate": 1.81695753744675e-05,
      "loss": 0.8521,
      "step": 672
    },
    {
      "epoch": 0.18486471638511193,
      "grad_norm": 0.3486904501914978,
      "learning_rate": 1.816682698914388e-05,
      "loss": 1.0171,
      "step": 673
    },
    {
      "epoch": 0.18513940392803185,
      "grad_norm": 0.22449781000614166,
      "learning_rate": 1.8164078603820258e-05,
      "loss": 0.7546,
      "step": 674
    },
    {
      "epoch": 0.18541409147095178,
      "grad_norm": 0.365487277507782,
      "learning_rate": 1.8161330218496635e-05,
      "loss": 0.958,
      "step": 675
    },
    {
      "epoch": 0.1856887790138717,
      "grad_norm": 0.26869526505470276,
      "learning_rate": 1.8158581833173012e-05,
      "loss": 1.1516,
      "step": 676
    },
    {
      "epoch": 0.18596346655679166,
      "grad_norm": 0.3736107349395752,
      "learning_rate": 1.815583344784939e-05,
      "loss": 0.9342,
      "step": 677
    },
    {
      "epoch": 0.1862381540997116,
      "grad_norm": 0.41668611764907837,
      "learning_rate": 1.815308506252577e-05,
      "loss": 0.9423,
      "step": 678
    },
    {
      "epoch": 0.18651284164263152,
      "grad_norm": 0.29717308282852173,
      "learning_rate": 1.8150336677202147e-05,
      "loss": 0.6291,
      "step": 679
    },
    {
      "epoch": 0.18678752918555144,
      "grad_norm": 0.3586972653865814,
      "learning_rate": 1.814758829187852e-05,
      "loss": 0.8914,
      "step": 680
    },
    {
      "epoch": 0.18706221672847137,
      "grad_norm": 0.33082374930381775,
      "learning_rate": 1.81448399065549e-05,
      "loss": 0.8223,
      "step": 681
    },
    {
      "epoch": 0.1873369042713913,
      "grad_norm": 0.30926334857940674,
      "learning_rate": 1.8142091521231278e-05,
      "loss": 1.082,
      "step": 682
    },
    {
      "epoch": 0.18761159181431122,
      "grad_norm": 0.2450881153345108,
      "learning_rate": 1.8139343135907658e-05,
      "loss": 0.8811,
      "step": 683
    },
    {
      "epoch": 0.18788627935723115,
      "grad_norm": 0.3527241349220276,
      "learning_rate": 1.8136594750584035e-05,
      "loss": 1.0029,
      "step": 684
    },
    {
      "epoch": 0.18816096690015108,
      "grad_norm": 0.4031405746936798,
      "learning_rate": 1.813384636526041e-05,
      "loss": 0.8452,
      "step": 685
    },
    {
      "epoch": 0.188435654443071,
      "grad_norm": 0.3567340672016144,
      "learning_rate": 1.813109797993679e-05,
      "loss": 0.8448,
      "step": 686
    },
    {
      "epoch": 0.18871034198599093,
      "grad_norm": 0.2990186810493469,
      "learning_rate": 1.8128349594613166e-05,
      "loss": 0.8187,
      "step": 687
    },
    {
      "epoch": 0.18898502952891086,
      "grad_norm": 0.3368537127971649,
      "learning_rate": 1.8125601209289543e-05,
      "loss": 0.8869,
      "step": 688
    },
    {
      "epoch": 0.18925971707183079,
      "grad_norm": 0.3107588589191437,
      "learning_rate": 1.812285282396592e-05,
      "loss": 0.9863,
      "step": 689
    },
    {
      "epoch": 0.1895344046147507,
      "grad_norm": 0.26889729499816895,
      "learning_rate": 1.8120104438642297e-05,
      "loss": 1.1085,
      "step": 690
    },
    {
      "epoch": 0.18980909215767064,
      "grad_norm": 0.30325961112976074,
      "learning_rate": 1.8117356053318678e-05,
      "loss": 1.1098,
      "step": 691
    },
    {
      "epoch": 0.19008377970059057,
      "grad_norm": 0.4157354533672333,
      "learning_rate": 1.8114607667995055e-05,
      "loss": 0.9305,
      "step": 692
    },
    {
      "epoch": 0.1903584672435105,
      "grad_norm": 0.24400031566619873,
      "learning_rate": 1.811185928267143e-05,
      "loss": 0.7941,
      "step": 693
    },
    {
      "epoch": 0.19063315478643045,
      "grad_norm": 0.2999146282672882,
      "learning_rate": 1.810911089734781e-05,
      "loss": 0.8275,
      "step": 694
    },
    {
      "epoch": 0.19090784232935037,
      "grad_norm": 0.2653135657310486,
      "learning_rate": 1.8106362512024186e-05,
      "loss": 0.7797,
      "step": 695
    },
    {
      "epoch": 0.1911825298722703,
      "grad_norm": 0.2722197473049164,
      "learning_rate": 1.8103614126700563e-05,
      "loss": 1.1072,
      "step": 696
    },
    {
      "epoch": 0.19145721741519023,
      "grad_norm": 0.24562907218933105,
      "learning_rate": 1.8100865741376943e-05,
      "loss": 0.5924,
      "step": 697
    },
    {
      "epoch": 0.19173190495811016,
      "grad_norm": 0.34976616501808167,
      "learning_rate": 1.809811735605332e-05,
      "loss": 0.9135,
      "step": 698
    },
    {
      "epoch": 0.19200659250103008,
      "grad_norm": 0.3236413300037384,
      "learning_rate": 1.8095368970729697e-05,
      "loss": 0.7156,
      "step": 699
    },
    {
      "epoch": 0.19228128004395,
      "grad_norm": 0.2706535756587982,
      "learning_rate": 1.8092620585406074e-05,
      "loss": 0.8043,
      "step": 700
    },
    {
      "epoch": 0.19255596758686994,
      "grad_norm": 0.3679579198360443,
      "learning_rate": 1.808987220008245e-05,
      "loss": 0.7272,
      "step": 701
    },
    {
      "epoch": 0.19283065512978986,
      "grad_norm": 0.3032127916812897,
      "learning_rate": 1.808712381475883e-05,
      "loss": 1.1396,
      "step": 702
    },
    {
      "epoch": 0.1931053426727098,
      "grad_norm": 0.25677740573883057,
      "learning_rate": 1.808437542943521e-05,
      "loss": 0.9611,
      "step": 703
    },
    {
      "epoch": 0.19338003021562972,
      "grad_norm": 0.35270893573760986,
      "learning_rate": 1.8081627044111586e-05,
      "loss": 0.9556,
      "step": 704
    },
    {
      "epoch": 0.19365471775854964,
      "grad_norm": 0.2739832401275635,
      "learning_rate": 1.8078878658787963e-05,
      "loss": 1.0006,
      "step": 705
    },
    {
      "epoch": 0.19392940530146957,
      "grad_norm": 0.3328511714935303,
      "learning_rate": 1.807613027346434e-05,
      "loss": 0.8947,
      "step": 706
    },
    {
      "epoch": 0.1942040928443895,
      "grad_norm": 0.3556508719921112,
      "learning_rate": 1.807338188814072e-05,
      "loss": 0.8954,
      "step": 707
    },
    {
      "epoch": 0.19447878038730942,
      "grad_norm": 0.24262696504592896,
      "learning_rate": 1.8070633502817097e-05,
      "loss": 0.67,
      "step": 708
    },
    {
      "epoch": 0.19475346793022935,
      "grad_norm": 0.28940925002098083,
      "learning_rate": 1.8067885117493474e-05,
      "loss": 0.8972,
      "step": 709
    },
    {
      "epoch": 0.1950281554731493,
      "grad_norm": 0.25126582384109497,
      "learning_rate": 1.806513673216985e-05,
      "loss": 0.7562,
      "step": 710
    },
    {
      "epoch": 0.19530284301606923,
      "grad_norm": 0.38338369131088257,
      "learning_rate": 1.8062388346846228e-05,
      "loss": 0.8842,
      "step": 711
    },
    {
      "epoch": 0.19557753055898916,
      "grad_norm": 0.3380014896392822,
      "learning_rate": 1.8059639961522605e-05,
      "loss": 0.6961,
      "step": 712
    },
    {
      "epoch": 0.1958522181019091,
      "grad_norm": 0.20760098099708557,
      "learning_rate": 1.8056891576198986e-05,
      "loss": 0.8008,
      "step": 713
    },
    {
      "epoch": 0.196126905644829,
      "grad_norm": 0.3137963116168976,
      "learning_rate": 1.8054143190875363e-05,
      "loss": 0.9515,
      "step": 714
    },
    {
      "epoch": 0.19640159318774894,
      "grad_norm": 0.2968253493309021,
      "learning_rate": 1.805139480555174e-05,
      "loss": 0.7941,
      "step": 715
    },
    {
      "epoch": 0.19667628073066887,
      "grad_norm": 0.30517134070396423,
      "learning_rate": 1.8048646420228117e-05,
      "loss": 1.0076,
      "step": 716
    },
    {
      "epoch": 0.1969509682735888,
      "grad_norm": 0.2658628821372986,
      "learning_rate": 1.8045898034904494e-05,
      "loss": 0.6945,
      "step": 717
    },
    {
      "epoch": 0.19722565581650872,
      "grad_norm": 0.35380855202674866,
      "learning_rate": 1.8043149649580874e-05,
      "loss": 0.7839,
      "step": 718
    },
    {
      "epoch": 0.19750034335942865,
      "grad_norm": 0.316192626953125,
      "learning_rate": 1.804040126425725e-05,
      "loss": 1.0564,
      "step": 719
    },
    {
      "epoch": 0.19777503090234858,
      "grad_norm": 0.34393835067749023,
      "learning_rate": 1.8037652878933628e-05,
      "loss": 0.884,
      "step": 720
    },
    {
      "epoch": 0.1980497184452685,
      "grad_norm": 0.28059515357017517,
      "learning_rate": 1.8034904493610005e-05,
      "loss": 0.8919,
      "step": 721
    },
    {
      "epoch": 0.19832440598818843,
      "grad_norm": 0.3476724326610565,
      "learning_rate": 1.8032156108286382e-05,
      "loss": 0.9464,
      "step": 722
    },
    {
      "epoch": 0.19859909353110836,
      "grad_norm": 0.2473224550485611,
      "learning_rate": 1.8029407722962763e-05,
      "loss": 0.8355,
      "step": 723
    },
    {
      "epoch": 0.19887378107402828,
      "grad_norm": 0.3007083535194397,
      "learning_rate": 1.802665933763914e-05,
      "loss": 0.9139,
      "step": 724
    },
    {
      "epoch": 0.1991484686169482,
      "grad_norm": 0.3712599575519562,
      "learning_rate": 1.8023910952315517e-05,
      "loss": 0.9307,
      "step": 725
    },
    {
      "epoch": 0.19942315615986814,
      "grad_norm": 0.3585355281829834,
      "learning_rate": 1.8021162566991894e-05,
      "loss": 1.0693,
      "step": 726
    },
    {
      "epoch": 0.1996978437027881,
      "grad_norm": 0.3540867269039154,
      "learning_rate": 1.801841418166827e-05,
      "loss": 0.9749,
      "step": 727
    },
    {
      "epoch": 0.19997253124570802,
      "grad_norm": 0.24075019359588623,
      "learning_rate": 1.8015665796344648e-05,
      "loss": 0.844,
      "step": 728
    },
    {
      "epoch": 0.20024721878862795,
      "grad_norm": 0.2748813331127167,
      "learning_rate": 1.8012917411021028e-05,
      "loss": 0.6844,
      "step": 729
    },
    {
      "epoch": 0.20052190633154787,
      "grad_norm": 0.37354356050491333,
      "learning_rate": 1.8010169025697405e-05,
      "loss": 0.8496,
      "step": 730
    },
    {
      "epoch": 0.2007965938744678,
      "grad_norm": 0.4196498692035675,
      "learning_rate": 1.8007420640373782e-05,
      "loss": 0.9946,
      "step": 731
    },
    {
      "epoch": 0.20107128141738773,
      "grad_norm": 0.2688349783420563,
      "learning_rate": 1.800467225505016e-05,
      "loss": 0.8751,
      "step": 732
    },
    {
      "epoch": 0.20134596896030765,
      "grad_norm": 0.24008134007453918,
      "learning_rate": 1.8001923869726536e-05,
      "loss": 0.8707,
      "step": 733
    },
    {
      "epoch": 0.20162065650322758,
      "grad_norm": 0.2759401500225067,
      "learning_rate": 1.7999175484402917e-05,
      "loss": 0.8449,
      "step": 734
    },
    {
      "epoch": 0.2018953440461475,
      "grad_norm": 0.27537965774536133,
      "learning_rate": 1.7996427099079294e-05,
      "loss": 1.1672,
      "step": 735
    },
    {
      "epoch": 0.20217003158906743,
      "grad_norm": 0.2753170132637024,
      "learning_rate": 1.799367871375567e-05,
      "loss": 1.1707,
      "step": 736
    },
    {
      "epoch": 0.20244471913198736,
      "grad_norm": 0.23477663099765778,
      "learning_rate": 1.7990930328432048e-05,
      "loss": 0.6934,
      "step": 737
    },
    {
      "epoch": 0.2027194066749073,
      "grad_norm": 0.31253862380981445,
      "learning_rate": 1.7988181943108425e-05,
      "loss": 0.8958,
      "step": 738
    },
    {
      "epoch": 0.20299409421782721,
      "grad_norm": 0.26875433325767517,
      "learning_rate": 1.7985433557784805e-05,
      "loss": 1.0023,
      "step": 739
    },
    {
      "epoch": 0.20326878176074714,
      "grad_norm": 0.265971302986145,
      "learning_rate": 1.7982685172461182e-05,
      "loss": 0.6002,
      "step": 740
    },
    {
      "epoch": 0.20354346930366707,
      "grad_norm": 0.4447736442089081,
      "learning_rate": 1.7979936787137556e-05,
      "loss": 0.8899,
      "step": 741
    },
    {
      "epoch": 0.203818156846587,
      "grad_norm": 0.41261518001556396,
      "learning_rate": 1.7977188401813936e-05,
      "loss": 0.8574,
      "step": 742
    },
    {
      "epoch": 0.20409284438950692,
      "grad_norm": 0.2995651960372925,
      "learning_rate": 1.7974440016490313e-05,
      "loss": 0.9425,
      "step": 743
    },
    {
      "epoch": 0.20436753193242688,
      "grad_norm": 0.3034432530403137,
      "learning_rate": 1.797169163116669e-05,
      "loss": 0.7965,
      "step": 744
    },
    {
      "epoch": 0.2046422194753468,
      "grad_norm": 0.28826403617858887,
      "learning_rate": 1.796894324584307e-05,
      "loss": 0.7708,
      "step": 745
    },
    {
      "epoch": 0.20491690701826673,
      "grad_norm": 0.3444674015045166,
      "learning_rate": 1.7966194860519444e-05,
      "loss": 0.9999,
      "step": 746
    },
    {
      "epoch": 0.20519159456118666,
      "grad_norm": 0.33219632506370544,
      "learning_rate": 1.7963446475195825e-05,
      "loss": 0.8517,
      "step": 747
    },
    {
      "epoch": 0.20546628210410658,
      "grad_norm": 0.2787770628929138,
      "learning_rate": 1.79606980898722e-05,
      "loss": 0.9424,
      "step": 748
    },
    {
      "epoch": 0.2057409696470265,
      "grad_norm": 0.3942496180534363,
      "learning_rate": 1.795794970454858e-05,
      "loss": 1.0898,
      "step": 749
    },
    {
      "epoch": 0.20601565718994644,
      "grad_norm": 0.3176543712615967,
      "learning_rate": 1.795520131922496e-05,
      "loss": 0.76,
      "step": 750
    },
    {
      "epoch": 0.20629034473286637,
      "grad_norm": 0.27828308939933777,
      "learning_rate": 1.7952452933901333e-05,
      "loss": 0.8748,
      "step": 751
    },
    {
      "epoch": 0.2065650322757863,
      "grad_norm": 0.34267207980155945,
      "learning_rate": 1.794970454857771e-05,
      "loss": 0.826,
      "step": 752
    },
    {
      "epoch": 0.20683971981870622,
      "grad_norm": 0.2775469124317169,
      "learning_rate": 1.794695616325409e-05,
      "loss": 0.7767,
      "step": 753
    },
    {
      "epoch": 0.20711440736162615,
      "grad_norm": 0.2909465432167053,
      "learning_rate": 1.7944207777930467e-05,
      "loss": 0.715,
      "step": 754
    },
    {
      "epoch": 0.20738909490454607,
      "grad_norm": 0.35736730694770813,
      "learning_rate": 1.7941459392606844e-05,
      "loss": 1.1322,
      "step": 755
    },
    {
      "epoch": 0.207663782447466,
      "grad_norm": 0.30242788791656494,
      "learning_rate": 1.793871100728322e-05,
      "loss": 0.9148,
      "step": 756
    },
    {
      "epoch": 0.20793846999038593,
      "grad_norm": 0.322777658700943,
      "learning_rate": 1.7935962621959598e-05,
      "loss": 0.8258,
      "step": 757
    },
    {
      "epoch": 0.20821315753330585,
      "grad_norm": 0.26630717515945435,
      "learning_rate": 1.793321423663598e-05,
      "loss": 0.8762,
      "step": 758
    },
    {
      "epoch": 0.20848784507622578,
      "grad_norm": 0.26683974266052246,
      "learning_rate": 1.7930465851312356e-05,
      "loss": 0.8617,
      "step": 759
    },
    {
      "epoch": 0.20876253261914574,
      "grad_norm": 0.39671778678894043,
      "learning_rate": 1.7927717465988733e-05,
      "loss": 0.9137,
      "step": 760
    },
    {
      "epoch": 0.20903722016206566,
      "grad_norm": 0.28011390566825867,
      "learning_rate": 1.792496908066511e-05,
      "loss": 0.7424,
      "step": 761
    },
    {
      "epoch": 0.2093119077049856,
      "grad_norm": 0.32701513171195984,
      "learning_rate": 1.7922220695341487e-05,
      "loss": 0.852,
      "step": 762
    },
    {
      "epoch": 0.20958659524790552,
      "grad_norm": 0.4431382119655609,
      "learning_rate": 1.7919472310017867e-05,
      "loss": 0.9186,
      "step": 763
    },
    {
      "epoch": 0.20986128279082544,
      "grad_norm": 0.26372602581977844,
      "learning_rate": 1.7916723924694244e-05,
      "loss": 0.6734,
      "step": 764
    },
    {
      "epoch": 0.21013597033374537,
      "grad_norm": 0.34279194474220276,
      "learning_rate": 1.791397553937062e-05,
      "loss": 0.8752,
      "step": 765
    },
    {
      "epoch": 0.2104106578766653,
      "grad_norm": 0.31449246406555176,
      "learning_rate": 1.7911227154046998e-05,
      "loss": 0.7587,
      "step": 766
    },
    {
      "epoch": 0.21068534541958522,
      "grad_norm": 0.2922191619873047,
      "learning_rate": 1.7908478768723375e-05,
      "loss": 1.03,
      "step": 767
    },
    {
      "epoch": 0.21096003296250515,
      "grad_norm": 0.3643679916858673,
      "learning_rate": 1.7905730383399752e-05,
      "loss": 0.8372,
      "step": 768
    },
    {
      "epoch": 0.21123472050542508,
      "grad_norm": 0.31123363971710205,
      "learning_rate": 1.7902981998076133e-05,
      "loss": 0.7484,
      "step": 769
    },
    {
      "epoch": 0.211509408048345,
      "grad_norm": 0.2924351692199707,
      "learning_rate": 1.790023361275251e-05,
      "loss": 0.9284,
      "step": 770
    },
    {
      "epoch": 0.21178409559126493,
      "grad_norm": 0.40282925963401794,
      "learning_rate": 1.7897485227428887e-05,
      "loss": 0.9808,
      "step": 771
    },
    {
      "epoch": 0.21205878313418486,
      "grad_norm": 0.35005778074264526,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 0.7843,
      "step": 772
    },
    {
      "epoch": 0.21233347067710479,
      "grad_norm": 0.37175482511520386,
      "learning_rate": 1.789198845678164e-05,
      "loss": 0.8585,
      "step": 773
    },
    {
      "epoch": 0.2126081582200247,
      "grad_norm": 0.29106900095939636,
      "learning_rate": 1.788924007145802e-05,
      "loss": 0.936,
      "step": 774
    },
    {
      "epoch": 0.21288284576294464,
      "grad_norm": 0.3150571882724762,
      "learning_rate": 1.7886491686134398e-05,
      "loss": 0.9359,
      "step": 775
    },
    {
      "epoch": 0.21315753330586457,
      "grad_norm": 0.3887014389038086,
      "learning_rate": 1.7883743300810775e-05,
      "loss": 0.9602,
      "step": 776
    },
    {
      "epoch": 0.21343222084878452,
      "grad_norm": 0.33795976638793945,
      "learning_rate": 1.7880994915487152e-05,
      "loss": 0.9215,
      "step": 777
    },
    {
      "epoch": 0.21370690839170445,
      "grad_norm": 0.2693367302417755,
      "learning_rate": 1.787824653016353e-05,
      "loss": 0.7927,
      "step": 778
    },
    {
      "epoch": 0.21398159593462437,
      "grad_norm": 0.2367921769618988,
      "learning_rate": 1.787549814483991e-05,
      "loss": 0.8273,
      "step": 779
    },
    {
      "epoch": 0.2142562834775443,
      "grad_norm": 0.24470309913158417,
      "learning_rate": 1.7872749759516287e-05,
      "loss": 0.9918,
      "step": 780
    },
    {
      "epoch": 0.21453097102046423,
      "grad_norm": 0.18922662734985352,
      "learning_rate": 1.7870001374192664e-05,
      "loss": 0.7009,
      "step": 781
    },
    {
      "epoch": 0.21480565856338416,
      "grad_norm": 0.29698145389556885,
      "learning_rate": 1.786725298886904e-05,
      "loss": 0.8092,
      "step": 782
    },
    {
      "epoch": 0.21508034610630408,
      "grad_norm": 0.2578592598438263,
      "learning_rate": 1.7864504603545418e-05,
      "loss": 0.8019,
      "step": 783
    },
    {
      "epoch": 0.215355033649224,
      "grad_norm": 0.31181856989860535,
      "learning_rate": 1.7861756218221795e-05,
      "loss": 0.7093,
      "step": 784
    },
    {
      "epoch": 0.21562972119214394,
      "grad_norm": 0.3379547595977783,
      "learning_rate": 1.7859007832898175e-05,
      "loss": 0.9579,
      "step": 785
    },
    {
      "epoch": 0.21590440873506386,
      "grad_norm": 0.3111572861671448,
      "learning_rate": 1.7856259447574552e-05,
      "loss": 0.8085,
      "step": 786
    },
    {
      "epoch": 0.2161790962779838,
      "grad_norm": 0.21899014711380005,
      "learning_rate": 1.785351106225093e-05,
      "loss": 0.7749,
      "step": 787
    },
    {
      "epoch": 0.21645378382090372,
      "grad_norm": 0.23704427480697632,
      "learning_rate": 1.7850762676927306e-05,
      "loss": 0.9369,
      "step": 788
    },
    {
      "epoch": 0.21672847136382364,
      "grad_norm": 0.3838484287261963,
      "learning_rate": 1.7848014291603683e-05,
      "loss": 0.8009,
      "step": 789
    },
    {
      "epoch": 0.21700315890674357,
      "grad_norm": 0.311997652053833,
      "learning_rate": 1.7845265906280064e-05,
      "loss": 0.8274,
      "step": 790
    },
    {
      "epoch": 0.2172778464496635,
      "grad_norm": 0.36650973558425903,
      "learning_rate": 1.784251752095644e-05,
      "loss": 0.8964,
      "step": 791
    },
    {
      "epoch": 0.21755253399258342,
      "grad_norm": 0.24399952590465546,
      "learning_rate": 1.7839769135632818e-05,
      "loss": 0.562,
      "step": 792
    },
    {
      "epoch": 0.21782722153550335,
      "grad_norm": 0.26600927114486694,
      "learning_rate": 1.7837020750309195e-05,
      "loss": 0.847,
      "step": 793
    },
    {
      "epoch": 0.2181019090784233,
      "grad_norm": 0.3237658441066742,
      "learning_rate": 1.7834272364985572e-05,
      "loss": 0.9506,
      "step": 794
    },
    {
      "epoch": 0.21837659662134323,
      "grad_norm": 0.2722218334674835,
      "learning_rate": 1.7831523979661952e-05,
      "loss": 0.8051,
      "step": 795
    },
    {
      "epoch": 0.21865128416426316,
      "grad_norm": 0.27616775035858154,
      "learning_rate": 1.782877559433833e-05,
      "loss": 0.7622,
      "step": 796
    },
    {
      "epoch": 0.2189259717071831,
      "grad_norm": 0.44978851079940796,
      "learning_rate": 1.7826027209014706e-05,
      "loss": 0.916,
      "step": 797
    },
    {
      "epoch": 0.219200659250103,
      "grad_norm": 0.3378292918205261,
      "learning_rate": 1.7823278823691083e-05,
      "loss": 0.8258,
      "step": 798
    },
    {
      "epoch": 0.21947534679302294,
      "grad_norm": 0.29690518975257874,
      "learning_rate": 1.782053043836746e-05,
      "loss": 0.7926,
      "step": 799
    },
    {
      "epoch": 0.21975003433594287,
      "grad_norm": 0.2862563729286194,
      "learning_rate": 1.7817782053043837e-05,
      "loss": 0.774,
      "step": 800
    },
    {
      "epoch": 0.2200247218788628,
      "grad_norm": 0.2631273865699768,
      "learning_rate": 1.7815033667720218e-05,
      "loss": 0.609,
      "step": 801
    },
    {
      "epoch": 0.22029940942178272,
      "grad_norm": 0.347884863615036,
      "learning_rate": 1.7812285282396595e-05,
      "loss": 0.6959,
      "step": 802
    },
    {
      "epoch": 0.22057409696470265,
      "grad_norm": 0.3024451434612274,
      "learning_rate": 1.7809536897072972e-05,
      "loss": 0.9747,
      "step": 803
    },
    {
      "epoch": 0.22084878450762258,
      "grad_norm": 0.323661744594574,
      "learning_rate": 1.780678851174935e-05,
      "loss": 0.9032,
      "step": 804
    },
    {
      "epoch": 0.2211234720505425,
      "grad_norm": 0.3397585451602936,
      "learning_rate": 1.7804040126425726e-05,
      "loss": 1.1094,
      "step": 805
    },
    {
      "epoch": 0.22139815959346243,
      "grad_norm": 0.35923677682876587,
      "learning_rate": 1.7801291741102106e-05,
      "loss": 0.9484,
      "step": 806
    },
    {
      "epoch": 0.22167284713638236,
      "grad_norm": 0.32227712869644165,
      "learning_rate": 1.779854335577848e-05,
      "loss": 1.0813,
      "step": 807
    },
    {
      "epoch": 0.22194753467930228,
      "grad_norm": 0.2829931676387787,
      "learning_rate": 1.7795794970454857e-05,
      "loss": 0.8348,
      "step": 808
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.2721012532711029,
      "learning_rate": 1.7793046585131237e-05,
      "loss": 0.8311,
      "step": 809
    },
    {
      "epoch": 0.22249690976514216,
      "grad_norm": 0.3490399122238159,
      "learning_rate": 1.7790298199807614e-05,
      "loss": 0.9454,
      "step": 810
    },
    {
      "epoch": 0.2227715973080621,
      "grad_norm": 0.4186284840106964,
      "learning_rate": 1.7787549814483995e-05,
      "loss": 0.8524,
      "step": 811
    },
    {
      "epoch": 0.22304628485098202,
      "grad_norm": 0.31781256198883057,
      "learning_rate": 1.7784801429160368e-05,
      "loss": 1.064,
      "step": 812
    },
    {
      "epoch": 0.22332097239390195,
      "grad_norm": 0.3172056972980499,
      "learning_rate": 1.7782053043836745e-05,
      "loss": 0.7554,
      "step": 813
    },
    {
      "epoch": 0.22359565993682187,
      "grad_norm": 0.3403002619743347,
      "learning_rate": 1.7779304658513126e-05,
      "loss": 1.1008,
      "step": 814
    },
    {
      "epoch": 0.2238703474797418,
      "grad_norm": 0.37204933166503906,
      "learning_rate": 1.7776556273189503e-05,
      "loss": 0.9572,
      "step": 815
    },
    {
      "epoch": 0.22414503502266173,
      "grad_norm": 0.24025313556194305,
      "learning_rate": 1.777380788786588e-05,
      "loss": 0.8132,
      "step": 816
    },
    {
      "epoch": 0.22441972256558165,
      "grad_norm": 0.29461055994033813,
      "learning_rate": 1.7771059502542257e-05,
      "loss": 0.5561,
      "step": 817
    },
    {
      "epoch": 0.22469441010850158,
      "grad_norm": 0.2463146448135376,
      "learning_rate": 1.7768311117218634e-05,
      "loss": 0.643,
      "step": 818
    },
    {
      "epoch": 0.2249690976514215,
      "grad_norm": 0.3298480808734894,
      "learning_rate": 1.7765562731895014e-05,
      "loss": 0.9394,
      "step": 819
    },
    {
      "epoch": 0.22524378519434143,
      "grad_norm": 0.2649931013584137,
      "learning_rate": 1.776281434657139e-05,
      "loss": 0.7024,
      "step": 820
    },
    {
      "epoch": 0.22551847273726136,
      "grad_norm": 0.33725422620773315,
      "learning_rate": 1.7760065961247768e-05,
      "loss": 1.1093,
      "step": 821
    },
    {
      "epoch": 0.2257931602801813,
      "grad_norm": 0.31875747442245483,
      "learning_rate": 1.7757317575924145e-05,
      "loss": 0.9258,
      "step": 822
    },
    {
      "epoch": 0.22606784782310121,
      "grad_norm": 0.2702462077140808,
      "learning_rate": 1.7754569190600522e-05,
      "loss": 0.647,
      "step": 823
    },
    {
      "epoch": 0.22634253536602114,
      "grad_norm": 0.32714858651161194,
      "learning_rate": 1.77518208052769e-05,
      "loss": 0.6593,
      "step": 824
    },
    {
      "epoch": 0.22661722290894107,
      "grad_norm": 0.27395063638687134,
      "learning_rate": 1.774907241995328e-05,
      "loss": 0.6897,
      "step": 825
    },
    {
      "epoch": 0.226891910451861,
      "grad_norm": 0.3640950620174408,
      "learning_rate": 1.7746324034629657e-05,
      "loss": 0.8384,
      "step": 826
    },
    {
      "epoch": 0.22716659799478095,
      "grad_norm": 0.27559003233909607,
      "learning_rate": 1.7743575649306034e-05,
      "loss": 0.6076,
      "step": 827
    },
    {
      "epoch": 0.22744128553770088,
      "grad_norm": 0.3430892825126648,
      "learning_rate": 1.774082726398241e-05,
      "loss": 0.8233,
      "step": 828
    },
    {
      "epoch": 0.2277159730806208,
      "grad_norm": 0.40898698568344116,
      "learning_rate": 1.7738078878658788e-05,
      "loss": 0.8675,
      "step": 829
    },
    {
      "epoch": 0.22799066062354073,
      "grad_norm": 0.2666110098361969,
      "learning_rate": 1.7735330493335168e-05,
      "loss": 0.8868,
      "step": 830
    },
    {
      "epoch": 0.22826534816646066,
      "grad_norm": 0.33352920413017273,
      "learning_rate": 1.7732582108011545e-05,
      "loss": 0.9861,
      "step": 831
    },
    {
      "epoch": 0.22854003570938058,
      "grad_norm": 0.2962650954723358,
      "learning_rate": 1.7729833722687922e-05,
      "loss": 0.8963,
      "step": 832
    },
    {
      "epoch": 0.2288147232523005,
      "grad_norm": 0.3395645022392273,
      "learning_rate": 1.77270853373643e-05,
      "loss": 0.7422,
      "step": 833
    },
    {
      "epoch": 0.22908941079522044,
      "grad_norm": 0.32726192474365234,
      "learning_rate": 1.7724336952040676e-05,
      "loss": 0.855,
      "step": 834
    },
    {
      "epoch": 0.22936409833814037,
      "grad_norm": 0.287741094827652,
      "learning_rate": 1.7721588566717057e-05,
      "loss": 0.7209,
      "step": 835
    },
    {
      "epoch": 0.2296387858810603,
      "grad_norm": 0.32963627576828003,
      "learning_rate": 1.7718840181393434e-05,
      "loss": 0.9211,
      "step": 836
    },
    {
      "epoch": 0.22991347342398022,
      "grad_norm": 0.40025079250335693,
      "learning_rate": 1.771609179606981e-05,
      "loss": 0.9315,
      "step": 837
    },
    {
      "epoch": 0.23018816096690015,
      "grad_norm": 0.31655895709991455,
      "learning_rate": 1.7713343410746188e-05,
      "loss": 0.9213,
      "step": 838
    },
    {
      "epoch": 0.23046284850982007,
      "grad_norm": 0.30187520384788513,
      "learning_rate": 1.7710595025422565e-05,
      "loss": 0.8564,
      "step": 839
    },
    {
      "epoch": 0.23073753605274,
      "grad_norm": 0.3384435474872589,
      "learning_rate": 1.7707846640098942e-05,
      "loss": 0.8049,
      "step": 840
    },
    {
      "epoch": 0.23101222359565993,
      "grad_norm": 0.3529997766017914,
      "learning_rate": 1.7705098254775322e-05,
      "loss": 1.0319,
      "step": 841
    },
    {
      "epoch": 0.23128691113857985,
      "grad_norm": 0.31323036551475525,
      "learning_rate": 1.77023498694517e-05,
      "loss": 0.8881,
      "step": 842
    },
    {
      "epoch": 0.23156159868149978,
      "grad_norm": 0.3187207877635956,
      "learning_rate": 1.7699601484128076e-05,
      "loss": 0.8666,
      "step": 843
    },
    {
      "epoch": 0.23183628622441974,
      "grad_norm": 0.34396472573280334,
      "learning_rate": 1.7696853098804453e-05,
      "loss": 0.7882,
      "step": 844
    },
    {
      "epoch": 0.23211097376733966,
      "grad_norm": 0.43988630175590515,
      "learning_rate": 1.769410471348083e-05,
      "loss": 0.8551,
      "step": 845
    },
    {
      "epoch": 0.2323856613102596,
      "grad_norm": 0.30140283703804016,
      "learning_rate": 1.769135632815721e-05,
      "loss": 1.0249,
      "step": 846
    },
    {
      "epoch": 0.23266034885317952,
      "grad_norm": 0.28299033641815186,
      "learning_rate": 1.7688607942833588e-05,
      "loss": 0.7738,
      "step": 847
    },
    {
      "epoch": 0.23293503639609944,
      "grad_norm": 0.23962578177452087,
      "learning_rate": 1.7685859557509965e-05,
      "loss": 1.0218,
      "step": 848
    },
    {
      "epoch": 0.23320972393901937,
      "grad_norm": 0.32310131192207336,
      "learning_rate": 1.7683111172186342e-05,
      "loss": 0.9372,
      "step": 849
    },
    {
      "epoch": 0.2334844114819393,
      "grad_norm": 0.3922351002693176,
      "learning_rate": 1.768036278686272e-05,
      "loss": 0.7479,
      "step": 850
    },
    {
      "epoch": 0.23375909902485922,
      "grad_norm": 0.41298678517341614,
      "learning_rate": 1.76776144015391e-05,
      "loss": 1.0118,
      "step": 851
    },
    {
      "epoch": 0.23403378656777915,
      "grad_norm": 0.3470916450023651,
      "learning_rate": 1.7674866016215476e-05,
      "loss": 0.6401,
      "step": 852
    },
    {
      "epoch": 0.23430847411069908,
      "grad_norm": 0.3532033860683441,
      "learning_rate": 1.7672117630891853e-05,
      "loss": 0.703,
      "step": 853
    },
    {
      "epoch": 0.234583161653619,
      "grad_norm": 0.30398887395858765,
      "learning_rate": 1.766936924556823e-05,
      "loss": 1.1048,
      "step": 854
    },
    {
      "epoch": 0.23485784919653893,
      "grad_norm": 0.3712163269519806,
      "learning_rate": 1.7666620860244607e-05,
      "loss": 0.9842,
      "step": 855
    },
    {
      "epoch": 0.23513253673945886,
      "grad_norm": 0.35790857672691345,
      "learning_rate": 1.7663872474920984e-05,
      "loss": 0.7824,
      "step": 856
    },
    {
      "epoch": 0.23540722428237879,
      "grad_norm": 0.2996363937854767,
      "learning_rate": 1.7661124089597365e-05,
      "loss": 0.9343,
      "step": 857
    },
    {
      "epoch": 0.2356819118252987,
      "grad_norm": 0.33543524146080017,
      "learning_rate": 1.7658375704273742e-05,
      "loss": 0.7111,
      "step": 858
    },
    {
      "epoch": 0.23595659936821864,
      "grad_norm": 0.29422134160995483,
      "learning_rate": 1.765562731895012e-05,
      "loss": 0.9372,
      "step": 859
    },
    {
      "epoch": 0.23623128691113857,
      "grad_norm": 0.29650112986564636,
      "learning_rate": 1.7652878933626496e-05,
      "loss": 0.7292,
      "step": 860
    },
    {
      "epoch": 0.23650597445405852,
      "grad_norm": 0.25177815556526184,
      "learning_rate": 1.7650130548302873e-05,
      "loss": 0.6458,
      "step": 861
    },
    {
      "epoch": 0.23678066199697845,
      "grad_norm": 0.3186068832874298,
      "learning_rate": 1.7647382162979253e-05,
      "loss": 0.5629,
      "step": 862
    },
    {
      "epoch": 0.23705534953989837,
      "grad_norm": 0.3667270839214325,
      "learning_rate": 1.764463377765563e-05,
      "loss": 1.0748,
      "step": 863
    },
    {
      "epoch": 0.2373300370828183,
      "grad_norm": 0.29444172978401184,
      "learning_rate": 1.7641885392332004e-05,
      "loss": 0.877,
      "step": 864
    },
    {
      "epoch": 0.23760472462573823,
      "grad_norm": 0.37133580446243286,
      "learning_rate": 1.7639137007008384e-05,
      "loss": 1.0012,
      "step": 865
    },
    {
      "epoch": 0.23787941216865816,
      "grad_norm": 0.3226100206375122,
      "learning_rate": 1.763638862168476e-05,
      "loss": 0.8489,
      "step": 866
    },
    {
      "epoch": 0.23815409971157808,
      "grad_norm": 0.24510087072849274,
      "learning_rate": 1.763364023636114e-05,
      "loss": 0.7405,
      "step": 867
    },
    {
      "epoch": 0.238428787254498,
      "grad_norm": 0.26519283652305603,
      "learning_rate": 1.763089185103752e-05,
      "loss": 0.8993,
      "step": 868
    },
    {
      "epoch": 0.23870347479741794,
      "grad_norm": 0.38567087054252625,
      "learning_rate": 1.7628143465713892e-05,
      "loss": 0.7048,
      "step": 869
    },
    {
      "epoch": 0.23897816234033786,
      "grad_norm": 0.39935705065727234,
      "learning_rate": 1.7625395080390273e-05,
      "loss": 0.9275,
      "step": 870
    },
    {
      "epoch": 0.2392528498832578,
      "grad_norm": 0.36696237325668335,
      "learning_rate": 1.762264669506665e-05,
      "loss": 0.7874,
      "step": 871
    },
    {
      "epoch": 0.23952753742617772,
      "grad_norm": 0.28319376707077026,
      "learning_rate": 1.7619898309743027e-05,
      "loss": 0.8021,
      "step": 872
    },
    {
      "epoch": 0.23980222496909764,
      "grad_norm": 0.26147565245628357,
      "learning_rate": 1.7617149924419407e-05,
      "loss": 0.9852,
      "step": 873
    },
    {
      "epoch": 0.24007691251201757,
      "grad_norm": 0.3032245934009552,
      "learning_rate": 1.761440153909578e-05,
      "loss": 0.896,
      "step": 874
    },
    {
      "epoch": 0.2403516000549375,
      "grad_norm": 0.24559198319911957,
      "learning_rate": 1.761165315377216e-05,
      "loss": 0.9265,
      "step": 875
    },
    {
      "epoch": 0.24062628759785742,
      "grad_norm": 0.1905454397201538,
      "learning_rate": 1.7608904768448538e-05,
      "loss": 0.5801,
      "step": 876
    },
    {
      "epoch": 0.24090097514077738,
      "grad_norm": 0.32651373744010925,
      "learning_rate": 1.7606156383124915e-05,
      "loss": 0.6949,
      "step": 877
    },
    {
      "epoch": 0.2411756626836973,
      "grad_norm": 0.27530768513679504,
      "learning_rate": 1.7603407997801292e-05,
      "loss": 1.1338,
      "step": 878
    },
    {
      "epoch": 0.24145035022661723,
      "grad_norm": 0.22504949569702148,
      "learning_rate": 1.760065961247767e-05,
      "loss": 0.5567,
      "step": 879
    },
    {
      "epoch": 0.24172503776953716,
      "grad_norm": 0.2908019721508026,
      "learning_rate": 1.7597911227154046e-05,
      "loss": 0.8143,
      "step": 880
    },
    {
      "epoch": 0.2419997253124571,
      "grad_norm": 0.2484211027622223,
      "learning_rate": 1.7595162841830427e-05,
      "loss": 0.8915,
      "step": 881
    },
    {
      "epoch": 0.242274412855377,
      "grad_norm": 0.44817402958869934,
      "learning_rate": 1.7592414456506804e-05,
      "loss": 0.748,
      "step": 882
    },
    {
      "epoch": 0.24254910039829694,
      "grad_norm": 0.27926403284072876,
      "learning_rate": 1.758966607118318e-05,
      "loss": 0.8166,
      "step": 883
    },
    {
      "epoch": 0.24282378794121687,
      "grad_norm": 0.3444117605686188,
      "learning_rate": 1.7586917685859558e-05,
      "loss": 0.9731,
      "step": 884
    },
    {
      "epoch": 0.2430984754841368,
      "grad_norm": 0.32348617911338806,
      "learning_rate": 1.7584169300535935e-05,
      "loss": 0.9985,
      "step": 885
    },
    {
      "epoch": 0.24337316302705672,
      "grad_norm": 0.36972135305404663,
      "learning_rate": 1.7581420915212315e-05,
      "loss": 0.823,
      "step": 886
    },
    {
      "epoch": 0.24364785056997665,
      "grad_norm": 0.33358755707740784,
      "learning_rate": 1.7578672529888692e-05,
      "loss": 0.6668,
      "step": 887
    },
    {
      "epoch": 0.24392253811289658,
      "grad_norm": 0.37444502115249634,
      "learning_rate": 1.757592414456507e-05,
      "loss": 0.8632,
      "step": 888
    },
    {
      "epoch": 0.2441972256558165,
      "grad_norm": 0.336576372385025,
      "learning_rate": 1.7573175759241446e-05,
      "loss": 0.9236,
      "step": 889
    },
    {
      "epoch": 0.24447191319873643,
      "grad_norm": 0.41475629806518555,
      "learning_rate": 1.7570427373917823e-05,
      "loss": 0.9686,
      "step": 890
    },
    {
      "epoch": 0.24474660074165636,
      "grad_norm": 0.3289412260055542,
      "learning_rate": 1.7567678988594204e-05,
      "loss": 0.7844,
      "step": 891
    },
    {
      "epoch": 0.24502128828457628,
      "grad_norm": 0.3279305398464203,
      "learning_rate": 1.756493060327058e-05,
      "loss": 0.6326,
      "step": 892
    },
    {
      "epoch": 0.2452959758274962,
      "grad_norm": 0.4596081078052521,
      "learning_rate": 1.7562182217946958e-05,
      "loss": 0.8869,
      "step": 893
    },
    {
      "epoch": 0.24557066337041616,
      "grad_norm": 0.3494125306606293,
      "learning_rate": 1.7559433832623335e-05,
      "loss": 0.9696,
      "step": 894
    },
    {
      "epoch": 0.2458453509133361,
      "grad_norm": 0.3476659953594208,
      "learning_rate": 1.7556685447299712e-05,
      "loss": 0.8658,
      "step": 895
    },
    {
      "epoch": 0.24612003845625602,
      "grad_norm": 0.333577960729599,
      "learning_rate": 1.755393706197609e-05,
      "loss": 0.7918,
      "step": 896
    },
    {
      "epoch": 0.24639472599917595,
      "grad_norm": 0.3982275426387787,
      "learning_rate": 1.755118867665247e-05,
      "loss": 0.9287,
      "step": 897
    },
    {
      "epoch": 0.24666941354209587,
      "grad_norm": 0.2755137085914612,
      "learning_rate": 1.7548440291328846e-05,
      "loss": 0.7506,
      "step": 898
    },
    {
      "epoch": 0.2469441010850158,
      "grad_norm": 0.2645646035671234,
      "learning_rate": 1.7545691906005223e-05,
      "loss": 0.6884,
      "step": 899
    },
    {
      "epoch": 0.24721878862793573,
      "grad_norm": 0.34356120228767395,
      "learning_rate": 1.75429435206816e-05,
      "loss": 1.1316,
      "step": 900
    },
    {
      "epoch": 0.24749347617085565,
      "grad_norm": 0.26573216915130615,
      "learning_rate": 1.7540195135357977e-05,
      "loss": 0.6931,
      "step": 901
    },
    {
      "epoch": 0.24776816371377558,
      "grad_norm": 0.20259620249271393,
      "learning_rate": 1.7537446750034358e-05,
      "loss": 0.5543,
      "step": 902
    },
    {
      "epoch": 0.2480428512566955,
      "grad_norm": 0.4448135197162628,
      "learning_rate": 1.7534698364710735e-05,
      "loss": 0.7762,
      "step": 903
    },
    {
      "epoch": 0.24831753879961543,
      "grad_norm": 0.4102621078491211,
      "learning_rate": 1.7531949979387112e-05,
      "loss": 0.966,
      "step": 904
    },
    {
      "epoch": 0.24859222634253536,
      "grad_norm": 0.34652993083000183,
      "learning_rate": 1.752920159406349e-05,
      "loss": 0.8984,
      "step": 905
    },
    {
      "epoch": 0.2488669138854553,
      "grad_norm": 0.25910013914108276,
      "learning_rate": 1.7526453208739866e-05,
      "loss": 1.1438,
      "step": 906
    },
    {
      "epoch": 0.24914160142837521,
      "grad_norm": 0.26366716623306274,
      "learning_rate": 1.7523704823416246e-05,
      "loss": 1.0124,
      "step": 907
    },
    {
      "epoch": 0.24941628897129514,
      "grad_norm": 0.32243120670318604,
      "learning_rate": 1.7520956438092623e-05,
      "loss": 0.7665,
      "step": 908
    },
    {
      "epoch": 0.24969097651421507,
      "grad_norm": 0.28684836626052856,
      "learning_rate": 1.7518208052769e-05,
      "loss": 0.8991,
      "step": 909
    },
    {
      "epoch": 0.249965664057135,
      "grad_norm": 0.3130677044391632,
      "learning_rate": 1.7515459667445377e-05,
      "loss": 0.8201,
      "step": 910
    },
    {
      "epoch": 0.2502403516000549,
      "grad_norm": 0.30843693017959595,
      "learning_rate": 1.7512711282121754e-05,
      "loss": 0.9165,
      "step": 911
    },
    {
      "epoch": 0.25051503914297485,
      "grad_norm": 0.34789806604385376,
      "learning_rate": 1.750996289679813e-05,
      "loss": 0.7958,
      "step": 912
    },
    {
      "epoch": 0.2507897266858948,
      "grad_norm": 0.39257335662841797,
      "learning_rate": 1.7507214511474512e-05,
      "loss": 0.8333,
      "step": 913
    },
    {
      "epoch": 0.2510644142288147,
      "grad_norm": 0.32343044877052307,
      "learning_rate": 1.750446612615089e-05,
      "loss": 0.968,
      "step": 914
    },
    {
      "epoch": 0.25133910177173463,
      "grad_norm": 0.33365458250045776,
      "learning_rate": 1.7501717740827266e-05,
      "loss": 0.9101,
      "step": 915
    },
    {
      "epoch": 0.25161378931465456,
      "grad_norm": 0.28654026985168457,
      "learning_rate": 1.7498969355503643e-05,
      "loss": 0.6733,
      "step": 916
    },
    {
      "epoch": 0.2518884768575745,
      "grad_norm": 0.30018648505210876,
      "learning_rate": 1.749622097018002e-05,
      "loss": 0.8613,
      "step": 917
    },
    {
      "epoch": 0.2521631644004944,
      "grad_norm": 0.3280073404312134,
      "learning_rate": 1.74934725848564e-05,
      "loss": 0.5986,
      "step": 918
    },
    {
      "epoch": 0.2524378519434144,
      "grad_norm": 0.31200405955314636,
      "learning_rate": 1.7490724199532777e-05,
      "loss": 1.0331,
      "step": 919
    },
    {
      "epoch": 0.2527125394863343,
      "grad_norm": 0.31301555037498474,
      "learning_rate": 1.7487975814209154e-05,
      "loss": 0.6996,
      "step": 920
    },
    {
      "epoch": 0.25298722702925425,
      "grad_norm": 0.3777190148830414,
      "learning_rate": 1.748522742888553e-05,
      "loss": 0.8997,
      "step": 921
    },
    {
      "epoch": 0.2532619145721742,
      "grad_norm": 0.3202187120914459,
      "learning_rate": 1.748247904356191e-05,
      "loss": 0.7042,
      "step": 922
    },
    {
      "epoch": 0.2535366021150941,
      "grad_norm": 0.33469969034194946,
      "learning_rate": 1.747973065823829e-05,
      "loss": 1.0124,
      "step": 923
    },
    {
      "epoch": 0.253811289658014,
      "grad_norm": 0.2343500852584839,
      "learning_rate": 1.7476982272914666e-05,
      "loss": 0.8614,
      "step": 924
    },
    {
      "epoch": 0.25408597720093395,
      "grad_norm": 0.40839987993240356,
      "learning_rate": 1.7474233887591043e-05,
      "loss": 0.7256,
      "step": 925
    },
    {
      "epoch": 0.2543606647438539,
      "grad_norm": 0.39251065254211426,
      "learning_rate": 1.747148550226742e-05,
      "loss": 0.8582,
      "step": 926
    },
    {
      "epoch": 0.2546353522867738,
      "grad_norm": 0.354196697473526,
      "learning_rate": 1.7468737116943797e-05,
      "loss": 0.8646,
      "step": 927
    },
    {
      "epoch": 0.25491003982969374,
      "grad_norm": 0.32765212655067444,
      "learning_rate": 1.7465988731620174e-05,
      "loss": 0.7509,
      "step": 928
    },
    {
      "epoch": 0.25518472737261366,
      "grad_norm": 0.4955606758594513,
      "learning_rate": 1.7463240346296554e-05,
      "loss": 0.877,
      "step": 929
    },
    {
      "epoch": 0.2554594149155336,
      "grad_norm": 0.3259301781654358,
      "learning_rate": 1.7460491960972928e-05,
      "loss": 0.7924,
      "step": 930
    },
    {
      "epoch": 0.2557341024584535,
      "grad_norm": 0.3045172691345215,
      "learning_rate": 1.7457743575649308e-05,
      "loss": 0.8737,
      "step": 931
    },
    {
      "epoch": 0.25600879000137344,
      "grad_norm": 0.348503053188324,
      "learning_rate": 1.7454995190325685e-05,
      "loss": 0.9822,
      "step": 932
    },
    {
      "epoch": 0.25628347754429337,
      "grad_norm": 0.30051112174987793,
      "learning_rate": 1.7452246805002062e-05,
      "loss": 0.7575,
      "step": 933
    },
    {
      "epoch": 0.2565581650872133,
      "grad_norm": 0.3190141022205353,
      "learning_rate": 1.7449498419678443e-05,
      "loss": 0.714,
      "step": 934
    },
    {
      "epoch": 0.2568328526301332,
      "grad_norm": 0.3754149377346039,
      "learning_rate": 1.7446750034354816e-05,
      "loss": 0.8416,
      "step": 935
    },
    {
      "epoch": 0.25710754017305315,
      "grad_norm": 0.3836405873298645,
      "learning_rate": 1.7444001649031193e-05,
      "loss": 0.8597,
      "step": 936
    },
    {
      "epoch": 0.2573822277159731,
      "grad_norm": 0.392727792263031,
      "learning_rate": 1.7441253263707574e-05,
      "loss": 0.9737,
      "step": 937
    },
    {
      "epoch": 0.257656915258893,
      "grad_norm": 0.3293650448322296,
      "learning_rate": 1.743850487838395e-05,
      "loss": 0.7082,
      "step": 938
    },
    {
      "epoch": 0.25793160280181293,
      "grad_norm": 0.2738098204135895,
      "learning_rate": 1.743575649306033e-05,
      "loss": 0.6653,
      "step": 939
    },
    {
      "epoch": 0.25820629034473286,
      "grad_norm": 0.2856590449810028,
      "learning_rate": 1.7433008107736705e-05,
      "loss": 0.6468,
      "step": 940
    },
    {
      "epoch": 0.2584809778876528,
      "grad_norm": 0.40866196155548096,
      "learning_rate": 1.7430259722413082e-05,
      "loss": 0.7528,
      "step": 941
    },
    {
      "epoch": 0.2587556654305727,
      "grad_norm": 0.33285728096961975,
      "learning_rate": 1.7427511337089462e-05,
      "loss": 0.7857,
      "step": 942
    },
    {
      "epoch": 0.25903035297349264,
      "grad_norm": 0.36194610595703125,
      "learning_rate": 1.742476295176584e-05,
      "loss": 0.8864,
      "step": 943
    },
    {
      "epoch": 0.25930504051641257,
      "grad_norm": 0.38835832476615906,
      "learning_rate": 1.7422014566442216e-05,
      "loss": 1.1141,
      "step": 944
    },
    {
      "epoch": 0.2595797280593325,
      "grad_norm": 0.38530316948890686,
      "learning_rate": 1.7419266181118593e-05,
      "loss": 0.9558,
      "step": 945
    },
    {
      "epoch": 0.2598544156022524,
      "grad_norm": 0.37181195616722107,
      "learning_rate": 1.741651779579497e-05,
      "loss": 1.1235,
      "step": 946
    },
    {
      "epoch": 0.26012910314517235,
      "grad_norm": 0.31136471033096313,
      "learning_rate": 1.741376941047135e-05,
      "loss": 1.0152,
      "step": 947
    },
    {
      "epoch": 0.2604037906880923,
      "grad_norm": 0.20425233244895935,
      "learning_rate": 1.7411021025147728e-05,
      "loss": 0.6005,
      "step": 948
    },
    {
      "epoch": 0.2606784782310122,
      "grad_norm": 0.42588990926742554,
      "learning_rate": 1.7408272639824105e-05,
      "loss": 0.9237,
      "step": 949
    },
    {
      "epoch": 0.2609531657739321,
      "grad_norm": 0.39962732791900635,
      "learning_rate": 1.7405524254500482e-05,
      "loss": 0.9881,
      "step": 950
    },
    {
      "epoch": 0.26122785331685205,
      "grad_norm": 0.35952961444854736,
      "learning_rate": 1.740277586917686e-05,
      "loss": 0.6926,
      "step": 951
    },
    {
      "epoch": 0.26150254085977204,
      "grad_norm": 0.36396533250808716,
      "learning_rate": 1.7400027483853236e-05,
      "loss": 0.8325,
      "step": 952
    },
    {
      "epoch": 0.26177722840269196,
      "grad_norm": 0.4012093245983124,
      "learning_rate": 1.7397279098529616e-05,
      "loss": 0.6744,
      "step": 953
    },
    {
      "epoch": 0.2620519159456119,
      "grad_norm": 0.4181601107120514,
      "learning_rate": 1.7394530713205993e-05,
      "loss": 0.9554,
      "step": 954
    },
    {
      "epoch": 0.2623266034885318,
      "grad_norm": 0.31247442960739136,
      "learning_rate": 1.739178232788237e-05,
      "loss": 1.0749,
      "step": 955
    },
    {
      "epoch": 0.26260129103145174,
      "grad_norm": 0.3588777780532837,
      "learning_rate": 1.7389033942558747e-05,
      "loss": 0.9995,
      "step": 956
    },
    {
      "epoch": 0.26287597857437167,
      "grad_norm": 0.29473501443862915,
      "learning_rate": 1.7386285557235124e-05,
      "loss": 1.088,
      "step": 957
    },
    {
      "epoch": 0.2631506661172916,
      "grad_norm": 0.35107579827308655,
      "learning_rate": 1.7383537171911505e-05,
      "loss": 1.0385,
      "step": 958
    },
    {
      "epoch": 0.2634253536602115,
      "grad_norm": 0.37405309081077576,
      "learning_rate": 1.7380788786587882e-05,
      "loss": 0.8012,
      "step": 959
    },
    {
      "epoch": 0.26370004120313145,
      "grad_norm": 0.39030513167381287,
      "learning_rate": 1.737804040126426e-05,
      "loss": 0.6693,
      "step": 960
    },
    {
      "epoch": 0.2639747287460514,
      "grad_norm": 0.38374897837638855,
      "learning_rate": 1.7375292015940636e-05,
      "loss": 0.8863,
      "step": 961
    },
    {
      "epoch": 0.2642494162889713,
      "grad_norm": 0.25608888268470764,
      "learning_rate": 1.7372543630617013e-05,
      "loss": 0.7745,
      "step": 962
    },
    {
      "epoch": 0.26452410383189123,
      "grad_norm": 0.3108493983745575,
      "learning_rate": 1.736979524529339e-05,
      "loss": 1.1133,
      "step": 963
    },
    {
      "epoch": 0.26479879137481116,
      "grad_norm": 0.44071295857429504,
      "learning_rate": 1.736704685996977e-05,
      "loss": 0.9152,
      "step": 964
    },
    {
      "epoch": 0.2650734789177311,
      "grad_norm": 0.35635676980018616,
      "learning_rate": 1.7364298474646147e-05,
      "loss": 1.0729,
      "step": 965
    },
    {
      "epoch": 0.265348166460651,
      "grad_norm": 0.3941352367401123,
      "learning_rate": 1.7361550089322524e-05,
      "loss": 0.8858,
      "step": 966
    },
    {
      "epoch": 0.26562285400357094,
      "grad_norm": 0.37580519914627075,
      "learning_rate": 1.73588017039989e-05,
      "loss": 0.6607,
      "step": 967
    },
    {
      "epoch": 0.26589754154649087,
      "grad_norm": 0.356142520904541,
      "learning_rate": 1.735605331867528e-05,
      "loss": 0.8361,
      "step": 968
    },
    {
      "epoch": 0.2661722290894108,
      "grad_norm": 0.32789677381515503,
      "learning_rate": 1.735330493335166e-05,
      "loss": 0.6892,
      "step": 969
    },
    {
      "epoch": 0.2664469166323307,
      "grad_norm": 0.28971752524375916,
      "learning_rate": 1.7350556548028036e-05,
      "loss": 0.7005,
      "step": 970
    },
    {
      "epoch": 0.26672160417525065,
      "grad_norm": 0.329448938369751,
      "learning_rate": 1.7347808162704413e-05,
      "loss": 0.8537,
      "step": 971
    },
    {
      "epoch": 0.2669962917181706,
      "grad_norm": 0.29368558526039124,
      "learning_rate": 1.734505977738079e-05,
      "loss": 0.711,
      "step": 972
    },
    {
      "epoch": 0.2672709792610905,
      "grad_norm": 0.4700009524822235,
      "learning_rate": 1.7342311392057167e-05,
      "loss": 0.9326,
      "step": 973
    },
    {
      "epoch": 0.26754566680401043,
      "grad_norm": 0.339903861284256,
      "learning_rate": 1.7339563006733547e-05,
      "loss": 0.7635,
      "step": 974
    },
    {
      "epoch": 0.26782035434693036,
      "grad_norm": 0.2840108871459961,
      "learning_rate": 1.7336814621409924e-05,
      "loss": 0.8658,
      "step": 975
    },
    {
      "epoch": 0.2680950418898503,
      "grad_norm": 0.3524324893951416,
      "learning_rate": 1.73340662360863e-05,
      "loss": 0.7839,
      "step": 976
    },
    {
      "epoch": 0.2683697294327702,
      "grad_norm": 0.33254680037498474,
      "learning_rate": 1.733131785076268e-05,
      "loss": 0.8464,
      "step": 977
    },
    {
      "epoch": 0.26864441697569014,
      "grad_norm": 0.3788672685623169,
      "learning_rate": 1.7328569465439055e-05,
      "loss": 0.8889,
      "step": 978
    },
    {
      "epoch": 0.26891910451861006,
      "grad_norm": 0.2794399857521057,
      "learning_rate": 1.7325821080115432e-05,
      "loss": 0.959,
      "step": 979
    },
    {
      "epoch": 0.26919379206153,
      "grad_norm": 0.3872576951980591,
      "learning_rate": 1.7323072694791813e-05,
      "loss": 0.6717,
      "step": 980
    },
    {
      "epoch": 0.2694684796044499,
      "grad_norm": 0.2687894403934479,
      "learning_rate": 1.732032430946819e-05,
      "loss": 0.9108,
      "step": 981
    },
    {
      "epoch": 0.26974316714736984,
      "grad_norm": 0.2947995960712433,
      "learning_rate": 1.7317575924144567e-05,
      "loss": 0.6932,
      "step": 982
    },
    {
      "epoch": 0.27001785469028977,
      "grad_norm": 0.33253541588783264,
      "learning_rate": 1.7314827538820944e-05,
      "loss": 0.9344,
      "step": 983
    },
    {
      "epoch": 0.2702925422332097,
      "grad_norm": 0.44428956508636475,
      "learning_rate": 1.731207915349732e-05,
      "loss": 0.7604,
      "step": 984
    },
    {
      "epoch": 0.2705672297761296,
      "grad_norm": 0.3325381577014923,
      "learning_rate": 1.73093307681737e-05,
      "loss": 0.8583,
      "step": 985
    },
    {
      "epoch": 0.2708419173190496,
      "grad_norm": 0.2934759557247162,
      "learning_rate": 1.730658238285008e-05,
      "loss": 0.4948,
      "step": 986
    },
    {
      "epoch": 0.27111660486196953,
      "grad_norm": 0.3400583565235138,
      "learning_rate": 1.7303833997526452e-05,
      "loss": 1.1229,
      "step": 987
    },
    {
      "epoch": 0.27139129240488946,
      "grad_norm": 0.20713885128498077,
      "learning_rate": 1.7301085612202832e-05,
      "loss": 0.5953,
      "step": 988
    },
    {
      "epoch": 0.2716659799478094,
      "grad_norm": 0.3441321551799774,
      "learning_rate": 1.729833722687921e-05,
      "loss": 0.7987,
      "step": 989
    },
    {
      "epoch": 0.2719406674907293,
      "grad_norm": 0.34277090430259705,
      "learning_rate": 1.729558884155559e-05,
      "loss": 0.9301,
      "step": 990
    },
    {
      "epoch": 0.27221535503364924,
      "grad_norm": 0.5481919646263123,
      "learning_rate": 1.7292840456231967e-05,
      "loss": 1.0413,
      "step": 991
    },
    {
      "epoch": 0.27249004257656917,
      "grad_norm": 0.2684152126312256,
      "learning_rate": 1.729009207090834e-05,
      "loss": 0.7447,
      "step": 992
    },
    {
      "epoch": 0.2727647301194891,
      "grad_norm": 0.3966268301010132,
      "learning_rate": 1.728734368558472e-05,
      "loss": 0.762,
      "step": 993
    },
    {
      "epoch": 0.273039417662409,
      "grad_norm": 0.3296954035758972,
      "learning_rate": 1.7284595300261098e-05,
      "loss": 1.0323,
      "step": 994
    },
    {
      "epoch": 0.27331410520532895,
      "grad_norm": 0.39776530861854553,
      "learning_rate": 1.7281846914937475e-05,
      "loss": 0.9754,
      "step": 995
    },
    {
      "epoch": 0.2735887927482489,
      "grad_norm": 0.328542023897171,
      "learning_rate": 1.7279098529613852e-05,
      "loss": 0.9589,
      "step": 996
    },
    {
      "epoch": 0.2738634802911688,
      "grad_norm": 0.33081477880477905,
      "learning_rate": 1.727635014429023e-05,
      "loss": 0.7212,
      "step": 997
    },
    {
      "epoch": 0.27413816783408873,
      "grad_norm": 0.3413216769695282,
      "learning_rate": 1.727360175896661e-05,
      "loss": 0.8786,
      "step": 998
    },
    {
      "epoch": 0.27441285537700866,
      "grad_norm": 0.3368081748485565,
      "learning_rate": 1.7270853373642986e-05,
      "loss": 0.8546,
      "step": 999
    },
    {
      "epoch": 0.2746875429199286,
      "grad_norm": 0.35485249757766724,
      "learning_rate": 1.7268104988319363e-05,
      "loss": 0.9538,
      "step": 1000
    },
    {
      "epoch": 0.2749622304628485,
      "grad_norm": 0.34709617495536804,
      "learning_rate": 1.726535660299574e-05,
      "loss": 0.869,
      "step": 1001
    },
    {
      "epoch": 0.27523691800576844,
      "grad_norm": 0.4163241684436798,
      "learning_rate": 1.7262608217672117e-05,
      "loss": 0.7742,
      "step": 1002
    },
    {
      "epoch": 0.27551160554868837,
      "grad_norm": 0.36977237462997437,
      "learning_rate": 1.7259859832348494e-05,
      "loss": 0.7028,
      "step": 1003
    },
    {
      "epoch": 0.2757862930916083,
      "grad_norm": 0.2790045142173767,
      "learning_rate": 1.7257111447024875e-05,
      "loss": 0.5112,
      "step": 1004
    },
    {
      "epoch": 0.2760609806345282,
      "grad_norm": 0.3661898970603943,
      "learning_rate": 1.7254363061701252e-05,
      "loss": 0.7962,
      "step": 1005
    },
    {
      "epoch": 0.27633566817744815,
      "grad_norm": 0.40471935272216797,
      "learning_rate": 1.725161467637763e-05,
      "loss": 0.8273,
      "step": 1006
    },
    {
      "epoch": 0.2766103557203681,
      "grad_norm": 0.49328237771987915,
      "learning_rate": 1.7248866291054006e-05,
      "loss": 0.9133,
      "step": 1007
    },
    {
      "epoch": 0.276885043263288,
      "grad_norm": 0.37307947874069214,
      "learning_rate": 1.7246117905730383e-05,
      "loss": 0.8146,
      "step": 1008
    },
    {
      "epoch": 0.2771597308062079,
      "grad_norm": 0.2919832468032837,
      "learning_rate": 1.7243369520406763e-05,
      "loss": 0.9353,
      "step": 1009
    },
    {
      "epoch": 0.27743441834912785,
      "grad_norm": 0.3953794538974762,
      "learning_rate": 1.724062113508314e-05,
      "loss": 0.9142,
      "step": 1010
    },
    {
      "epoch": 0.2777091058920478,
      "grad_norm": 0.32260119915008545,
      "learning_rate": 1.7237872749759517e-05,
      "loss": 0.9622,
      "step": 1011
    },
    {
      "epoch": 0.2779837934349677,
      "grad_norm": 0.34676721692085266,
      "learning_rate": 1.7235124364435894e-05,
      "loss": 0.8317,
      "step": 1012
    },
    {
      "epoch": 0.27825848097788763,
      "grad_norm": 0.24337591230869293,
      "learning_rate": 1.723237597911227e-05,
      "loss": 0.8861,
      "step": 1013
    },
    {
      "epoch": 0.27853316852080756,
      "grad_norm": 0.26669996976852417,
      "learning_rate": 1.7229627593788652e-05,
      "loss": 0.5373,
      "step": 1014
    },
    {
      "epoch": 0.2788078560637275,
      "grad_norm": 0.3468780517578125,
      "learning_rate": 1.722687920846503e-05,
      "loss": 0.8453,
      "step": 1015
    },
    {
      "epoch": 0.2790825436066474,
      "grad_norm": 0.3765402138233185,
      "learning_rate": 1.7224130823141406e-05,
      "loss": 0.9926,
      "step": 1016
    },
    {
      "epoch": 0.27935723114956734,
      "grad_norm": 0.3099086284637451,
      "learning_rate": 1.7221382437817783e-05,
      "loss": 0.8924,
      "step": 1017
    },
    {
      "epoch": 0.27963191869248727,
      "grad_norm": 0.2934211194515228,
      "learning_rate": 1.721863405249416e-05,
      "loss": 0.8684,
      "step": 1018
    },
    {
      "epoch": 0.27990660623540725,
      "grad_norm": 0.33048585057258606,
      "learning_rate": 1.7215885667170537e-05,
      "loss": 1.0005,
      "step": 1019
    },
    {
      "epoch": 0.2801812937783272,
      "grad_norm": 0.33378198742866516,
      "learning_rate": 1.7213137281846917e-05,
      "loss": 0.6624,
      "step": 1020
    },
    {
      "epoch": 0.2804559813212471,
      "grad_norm": 0.3699290454387665,
      "learning_rate": 1.7210388896523294e-05,
      "loss": 0.9657,
      "step": 1021
    },
    {
      "epoch": 0.28073066886416703,
      "grad_norm": 0.2799227237701416,
      "learning_rate": 1.720764051119967e-05,
      "loss": 0.6925,
      "step": 1022
    },
    {
      "epoch": 0.28100535640708696,
      "grad_norm": 0.32328638434410095,
      "learning_rate": 1.720489212587605e-05,
      "loss": 1.0858,
      "step": 1023
    },
    {
      "epoch": 0.2812800439500069,
      "grad_norm": 0.30896836519241333,
      "learning_rate": 1.7202143740552425e-05,
      "loss": 0.8588,
      "step": 1024
    },
    {
      "epoch": 0.2815547314929268,
      "grad_norm": 0.3190075159072876,
      "learning_rate": 1.7199395355228806e-05,
      "loss": 0.689,
      "step": 1025
    },
    {
      "epoch": 0.28182941903584674,
      "grad_norm": 0.37830671668052673,
      "learning_rate": 1.7196646969905183e-05,
      "loss": 0.7025,
      "step": 1026
    },
    {
      "epoch": 0.28210410657876667,
      "grad_norm": 0.3130069375038147,
      "learning_rate": 1.719389858458156e-05,
      "loss": 0.688,
      "step": 1027
    },
    {
      "epoch": 0.2823787941216866,
      "grad_norm": 0.27711373567581177,
      "learning_rate": 1.7191150199257937e-05,
      "loss": 1.0382,
      "step": 1028
    },
    {
      "epoch": 0.2826534816646065,
      "grad_norm": 0.29817989468574524,
      "learning_rate": 1.7188401813934314e-05,
      "loss": 0.896,
      "step": 1029
    },
    {
      "epoch": 0.28292816920752645,
      "grad_norm": 0.36349472403526306,
      "learning_rate": 1.7185653428610694e-05,
      "loss": 0.8291,
      "step": 1030
    },
    {
      "epoch": 0.2832028567504464,
      "grad_norm": 0.37864160537719727,
      "learning_rate": 1.718290504328707e-05,
      "loss": 0.8551,
      "step": 1031
    },
    {
      "epoch": 0.2834775442933663,
      "grad_norm": 0.29588595032691956,
      "learning_rate": 1.718015665796345e-05,
      "loss": 0.7494,
      "step": 1032
    },
    {
      "epoch": 0.28375223183628623,
      "grad_norm": 0.4194830060005188,
      "learning_rate": 1.7177408272639825e-05,
      "loss": 0.9359,
      "step": 1033
    },
    {
      "epoch": 0.28402691937920616,
      "grad_norm": 0.32878684997558594,
      "learning_rate": 1.7174659887316202e-05,
      "loss": 0.7738,
      "step": 1034
    },
    {
      "epoch": 0.2843016069221261,
      "grad_norm": 0.3597908318042755,
      "learning_rate": 1.717191150199258e-05,
      "loss": 0.7803,
      "step": 1035
    },
    {
      "epoch": 0.284576294465046,
      "grad_norm": 0.3570542335510254,
      "learning_rate": 1.716916311666896e-05,
      "loss": 0.8219,
      "step": 1036
    },
    {
      "epoch": 0.28485098200796594,
      "grad_norm": 0.3172714412212372,
      "learning_rate": 1.7166414731345337e-05,
      "loss": 0.5456,
      "step": 1037
    },
    {
      "epoch": 0.28512566955088586,
      "grad_norm": 0.30903664231300354,
      "learning_rate": 1.7163666346021714e-05,
      "loss": 0.9136,
      "step": 1038
    },
    {
      "epoch": 0.2854003570938058,
      "grad_norm": 0.438429594039917,
      "learning_rate": 1.716091796069809e-05,
      "loss": 0.7673,
      "step": 1039
    },
    {
      "epoch": 0.2856750446367257,
      "grad_norm": 0.3243789076805115,
      "learning_rate": 1.7158169575374468e-05,
      "loss": 0.7078,
      "step": 1040
    },
    {
      "epoch": 0.28594973217964564,
      "grad_norm": 0.2953818738460541,
      "learning_rate": 1.715542119005085e-05,
      "loss": 0.9865,
      "step": 1041
    },
    {
      "epoch": 0.28622441972256557,
      "grad_norm": 0.3322869837284088,
      "learning_rate": 1.7152672804727225e-05,
      "loss": 0.9813,
      "step": 1042
    },
    {
      "epoch": 0.2864991072654855,
      "grad_norm": 0.30504974722862244,
      "learning_rate": 1.7149924419403602e-05,
      "loss": 0.8736,
      "step": 1043
    },
    {
      "epoch": 0.2867737948084054,
      "grad_norm": 0.2992580831050873,
      "learning_rate": 1.714717603407998e-05,
      "loss": 0.8465,
      "step": 1044
    },
    {
      "epoch": 0.28704848235132535,
      "grad_norm": 0.31421101093292236,
      "learning_rate": 1.7144427648756356e-05,
      "loss": 1.0844,
      "step": 1045
    },
    {
      "epoch": 0.2873231698942453,
      "grad_norm": 0.2599547207355499,
      "learning_rate": 1.7141679263432737e-05,
      "loss": 0.7436,
      "step": 1046
    },
    {
      "epoch": 0.2875978574371652,
      "grad_norm": 0.26495805382728577,
      "learning_rate": 1.7138930878109114e-05,
      "loss": 0.6819,
      "step": 1047
    },
    {
      "epoch": 0.28787254498008513,
      "grad_norm": 0.2866683900356293,
      "learning_rate": 1.7136182492785487e-05,
      "loss": 0.8256,
      "step": 1048
    },
    {
      "epoch": 0.28814723252300506,
      "grad_norm": 0.3788217008113861,
      "learning_rate": 1.7133434107461868e-05,
      "loss": 0.7728,
      "step": 1049
    },
    {
      "epoch": 0.288421920065925,
      "grad_norm": 0.28028684854507446,
      "learning_rate": 1.7130685722138245e-05,
      "loss": 0.7829,
      "step": 1050
    },
    {
      "epoch": 0.2886966076088449,
      "grad_norm": 0.352050244808197,
      "learning_rate": 1.7127937336814622e-05,
      "loss": 0.9514,
      "step": 1051
    },
    {
      "epoch": 0.28897129515176484,
      "grad_norm": 0.3421739637851715,
      "learning_rate": 1.7125188951491002e-05,
      "loss": 0.8061,
      "step": 1052
    },
    {
      "epoch": 0.2892459826946848,
      "grad_norm": 0.37665680050849915,
      "learning_rate": 1.7122440566167376e-05,
      "loss": 0.8074,
      "step": 1053
    },
    {
      "epoch": 0.28952067023760475,
      "grad_norm": 0.2837391495704651,
      "learning_rate": 1.7119692180843756e-05,
      "loss": 1.137,
      "step": 1054
    },
    {
      "epoch": 0.2897953577805247,
      "grad_norm": 0.29527807235717773,
      "learning_rate": 1.7116943795520133e-05,
      "loss": 0.96,
      "step": 1055
    },
    {
      "epoch": 0.2900700453234446,
      "grad_norm": 0.3716008961200714,
      "learning_rate": 1.711419541019651e-05,
      "loss": 0.8559,
      "step": 1056
    },
    {
      "epoch": 0.29034473286636453,
      "grad_norm": 0.43157774209976196,
      "learning_rate": 1.711144702487289e-05,
      "loss": 0.828,
      "step": 1057
    },
    {
      "epoch": 0.29061942040928446,
      "grad_norm": 0.3435748815536499,
      "learning_rate": 1.7108698639549264e-05,
      "loss": 0.825,
      "step": 1058
    },
    {
      "epoch": 0.2908941079522044,
      "grad_norm": 0.33553385734558105,
      "learning_rate": 1.710595025422564e-05,
      "loss": 0.5973,
      "step": 1059
    },
    {
      "epoch": 0.2911687954951243,
      "grad_norm": 0.29321742057800293,
      "learning_rate": 1.7103201868902022e-05,
      "loss": 0.8762,
      "step": 1060
    },
    {
      "epoch": 0.29144348303804424,
      "grad_norm": 0.3591887056827545,
      "learning_rate": 1.71004534835784e-05,
      "loss": 0.7605,
      "step": 1061
    },
    {
      "epoch": 0.29171817058096416,
      "grad_norm": 0.3040677011013031,
      "learning_rate": 1.709770509825478e-05,
      "loss": 0.9365,
      "step": 1062
    },
    {
      "epoch": 0.2919928581238841,
      "grad_norm": 0.4054071605205536,
      "learning_rate": 1.7094956712931153e-05,
      "loss": 0.8108,
      "step": 1063
    },
    {
      "epoch": 0.292267545666804,
      "grad_norm": 0.28317344188690186,
      "learning_rate": 1.709220832760753e-05,
      "loss": 0.6609,
      "step": 1064
    },
    {
      "epoch": 0.29254223320972395,
      "grad_norm": 0.3956550359725952,
      "learning_rate": 1.708945994228391e-05,
      "loss": 0.88,
      "step": 1065
    },
    {
      "epoch": 0.29281692075264387,
      "grad_norm": 0.24620455503463745,
      "learning_rate": 1.7086711556960287e-05,
      "loss": 0.7218,
      "step": 1066
    },
    {
      "epoch": 0.2930916082955638,
      "grad_norm": 0.3206275999546051,
      "learning_rate": 1.7083963171636664e-05,
      "loss": 0.8625,
      "step": 1067
    },
    {
      "epoch": 0.2933662958384837,
      "grad_norm": 0.26531147956848145,
      "learning_rate": 1.708121478631304e-05,
      "loss": 0.646,
      "step": 1068
    },
    {
      "epoch": 0.29364098338140365,
      "grad_norm": 0.36370640993118286,
      "learning_rate": 1.707846640098942e-05,
      "loss": 0.75,
      "step": 1069
    },
    {
      "epoch": 0.2939156709243236,
      "grad_norm": 0.35633307695388794,
      "learning_rate": 1.70757180156658e-05,
      "loss": 0.642,
      "step": 1070
    },
    {
      "epoch": 0.2941903584672435,
      "grad_norm": 0.3449651002883911,
      "learning_rate": 1.7072969630342176e-05,
      "loss": 0.8751,
      "step": 1071
    },
    {
      "epoch": 0.29446504601016343,
      "grad_norm": 0.3329438269138336,
      "learning_rate": 1.7070221245018553e-05,
      "loss": 0.795,
      "step": 1072
    },
    {
      "epoch": 0.29473973355308336,
      "grad_norm": 0.4569573700428009,
      "learning_rate": 1.706747285969493e-05,
      "loss": 0.7627,
      "step": 1073
    },
    {
      "epoch": 0.2950144210960033,
      "grad_norm": 0.33673369884490967,
      "learning_rate": 1.7064724474371307e-05,
      "loss": 0.7519,
      "step": 1074
    },
    {
      "epoch": 0.2952891086389232,
      "grad_norm": 0.2703324854373932,
      "learning_rate": 1.7061976089047684e-05,
      "loss": 0.7062,
      "step": 1075
    },
    {
      "epoch": 0.29556379618184314,
      "grad_norm": 0.3275584280490875,
      "learning_rate": 1.7059227703724064e-05,
      "loss": 0.7081,
      "step": 1076
    },
    {
      "epoch": 0.29583848372476307,
      "grad_norm": 0.3792758285999298,
      "learning_rate": 1.705647931840044e-05,
      "loss": 1.0038,
      "step": 1077
    },
    {
      "epoch": 0.296113171267683,
      "grad_norm": 0.33733034133911133,
      "learning_rate": 1.705373093307682e-05,
      "loss": 0.9833,
      "step": 1078
    },
    {
      "epoch": 0.2963878588106029,
      "grad_norm": 0.3268063962459564,
      "learning_rate": 1.7050982547753195e-05,
      "loss": 0.6873,
      "step": 1079
    },
    {
      "epoch": 0.29666254635352285,
      "grad_norm": 0.2686792314052582,
      "learning_rate": 1.7048234162429572e-05,
      "loss": 1.0228,
      "step": 1080
    },
    {
      "epoch": 0.2969372338964428,
      "grad_norm": 0.32374638319015503,
      "learning_rate": 1.7045485777105953e-05,
      "loss": 0.948,
      "step": 1081
    },
    {
      "epoch": 0.2972119214393627,
      "grad_norm": 0.21478934586048126,
      "learning_rate": 1.704273739178233e-05,
      "loss": 0.5441,
      "step": 1082
    },
    {
      "epoch": 0.29748660898228263,
      "grad_norm": 0.34008678793907166,
      "learning_rate": 1.7039989006458707e-05,
      "loss": 0.782,
      "step": 1083
    },
    {
      "epoch": 0.29776129652520256,
      "grad_norm": 0.2902613580226898,
      "learning_rate": 1.7037240621135084e-05,
      "loss": 0.788,
      "step": 1084
    },
    {
      "epoch": 0.2980359840681225,
      "grad_norm": 0.3200162351131439,
      "learning_rate": 1.703449223581146e-05,
      "loss": 0.7458,
      "step": 1085
    },
    {
      "epoch": 0.29831067161104247,
      "grad_norm": 0.35846787691116333,
      "learning_rate": 1.703174385048784e-05,
      "loss": 1.1078,
      "step": 1086
    },
    {
      "epoch": 0.2985853591539624,
      "grad_norm": 0.3536551594734192,
      "learning_rate": 1.702899546516422e-05,
      "loss": 0.9916,
      "step": 1087
    },
    {
      "epoch": 0.2988600466968823,
      "grad_norm": 0.32521992921829224,
      "learning_rate": 1.7026247079840595e-05,
      "loss": 0.9366,
      "step": 1088
    },
    {
      "epoch": 0.29913473423980225,
      "grad_norm": 0.3595747947692871,
      "learning_rate": 1.7023498694516972e-05,
      "loss": 0.8723,
      "step": 1089
    },
    {
      "epoch": 0.2994094217827222,
      "grad_norm": 0.4798397719860077,
      "learning_rate": 1.702075030919335e-05,
      "loss": 1.0749,
      "step": 1090
    },
    {
      "epoch": 0.2996841093256421,
      "grad_norm": 0.2752971947193146,
      "learning_rate": 1.7018001923869726e-05,
      "loss": 0.7531,
      "step": 1091
    },
    {
      "epoch": 0.299958796868562,
      "grad_norm": 0.32776811718940735,
      "learning_rate": 1.7015253538546107e-05,
      "loss": 0.7912,
      "step": 1092
    },
    {
      "epoch": 0.30023348441148195,
      "grad_norm": 0.3777153789997101,
      "learning_rate": 1.7012505153222484e-05,
      "loss": 0.7201,
      "step": 1093
    },
    {
      "epoch": 0.3005081719544019,
      "grad_norm": 0.3460712730884552,
      "learning_rate": 1.700975676789886e-05,
      "loss": 0.6743,
      "step": 1094
    },
    {
      "epoch": 0.3007828594973218,
      "grad_norm": 0.3917783498764038,
      "learning_rate": 1.7007008382575238e-05,
      "loss": 0.7626,
      "step": 1095
    },
    {
      "epoch": 0.30105754704024174,
      "grad_norm": 0.27919304370880127,
      "learning_rate": 1.7004259997251615e-05,
      "loss": 0.8944,
      "step": 1096
    },
    {
      "epoch": 0.30133223458316166,
      "grad_norm": 0.32313886284828186,
      "learning_rate": 1.7001511611927995e-05,
      "loss": 0.7708,
      "step": 1097
    },
    {
      "epoch": 0.3016069221260816,
      "grad_norm": 0.334077924489975,
      "learning_rate": 1.6998763226604372e-05,
      "loss": 0.758,
      "step": 1098
    },
    {
      "epoch": 0.3018816096690015,
      "grad_norm": 0.32968106865882874,
      "learning_rate": 1.699601484128075e-05,
      "loss": 0.7719,
      "step": 1099
    },
    {
      "epoch": 0.30215629721192144,
      "grad_norm": 0.36353713274002075,
      "learning_rate": 1.6993266455957126e-05,
      "loss": 0.8801,
      "step": 1100
    },
    {
      "epoch": 0.30243098475484137,
      "grad_norm": 0.26805394887924194,
      "learning_rate": 1.6990518070633503e-05,
      "loss": 0.6,
      "step": 1101
    },
    {
      "epoch": 0.3027056722977613,
      "grad_norm": 0.32056522369384766,
      "learning_rate": 1.6987769685309884e-05,
      "loss": 0.8818,
      "step": 1102
    },
    {
      "epoch": 0.3029803598406812,
      "grad_norm": 0.3078444004058838,
      "learning_rate": 1.698502129998626e-05,
      "loss": 0.82,
      "step": 1103
    },
    {
      "epoch": 0.30325504738360115,
      "grad_norm": 0.4500022232532501,
      "learning_rate": 1.6982272914662638e-05,
      "loss": 0.8878,
      "step": 1104
    },
    {
      "epoch": 0.3035297349265211,
      "grad_norm": 0.31795036792755127,
      "learning_rate": 1.6979524529339015e-05,
      "loss": 0.7564,
      "step": 1105
    },
    {
      "epoch": 0.303804422469441,
      "grad_norm": 0.3348471522331238,
      "learning_rate": 1.6976776144015392e-05,
      "loss": 0.6751,
      "step": 1106
    },
    {
      "epoch": 0.30407911001236093,
      "grad_norm": 0.3326054513454437,
      "learning_rate": 1.697402775869177e-05,
      "loss": 0.8947,
      "step": 1107
    },
    {
      "epoch": 0.30435379755528086,
      "grad_norm": 0.3906829357147217,
      "learning_rate": 1.697127937336815e-05,
      "loss": 0.733,
      "step": 1108
    },
    {
      "epoch": 0.3046284850982008,
      "grad_norm": 0.44268420338630676,
      "learning_rate": 1.6968530988044526e-05,
      "loss": 0.7676,
      "step": 1109
    },
    {
      "epoch": 0.3049031726411207,
      "grad_norm": 0.4199763834476471,
      "learning_rate": 1.6965782602720903e-05,
      "loss": 0.7479,
      "step": 1110
    },
    {
      "epoch": 0.30517786018404064,
      "grad_norm": 0.42009153962135315,
      "learning_rate": 1.696303421739728e-05,
      "loss": 0.8603,
      "step": 1111
    },
    {
      "epoch": 0.30545254772696057,
      "grad_norm": 0.3375110626220703,
      "learning_rate": 1.6960285832073657e-05,
      "loss": 0.8332,
      "step": 1112
    },
    {
      "epoch": 0.3057272352698805,
      "grad_norm": 0.31942227482795715,
      "learning_rate": 1.6957537446750038e-05,
      "loss": 0.9206,
      "step": 1113
    },
    {
      "epoch": 0.3060019228128004,
      "grad_norm": 0.3359595835208893,
      "learning_rate": 1.6954789061426415e-05,
      "loss": 1.1366,
      "step": 1114
    },
    {
      "epoch": 0.30627661035572035,
      "grad_norm": 0.32085245847702026,
      "learning_rate": 1.695204067610279e-05,
      "loss": 0.7312,
      "step": 1115
    },
    {
      "epoch": 0.3065512978986403,
      "grad_norm": 0.3086831569671631,
      "learning_rate": 1.694929229077917e-05,
      "loss": 0.7215,
      "step": 1116
    },
    {
      "epoch": 0.3068259854415602,
      "grad_norm": 0.38202324509620667,
      "learning_rate": 1.6946543905455546e-05,
      "loss": 0.7148,
      "step": 1117
    },
    {
      "epoch": 0.3071006729844801,
      "grad_norm": 0.2997192144393921,
      "learning_rate": 1.6943795520131926e-05,
      "loss": 0.7907,
      "step": 1118
    },
    {
      "epoch": 0.3073753605274001,
      "grad_norm": 0.3531710207462311,
      "learning_rate": 1.69410471348083e-05,
      "loss": 0.5413,
      "step": 1119
    },
    {
      "epoch": 0.30765004807032004,
      "grad_norm": 0.4477320611476898,
      "learning_rate": 1.6938298749484677e-05,
      "loss": 0.7878,
      "step": 1120
    },
    {
      "epoch": 0.30792473561323996,
      "grad_norm": 0.25792011618614197,
      "learning_rate": 1.6935550364161057e-05,
      "loss": 1.1116,
      "step": 1121
    },
    {
      "epoch": 0.3081994231561599,
      "grad_norm": 0.33830901980400085,
      "learning_rate": 1.6932801978837434e-05,
      "loss": 0.8563,
      "step": 1122
    },
    {
      "epoch": 0.3084741106990798,
      "grad_norm": 0.29077768325805664,
      "learning_rate": 1.693005359351381e-05,
      "loss": 0.8476,
      "step": 1123
    },
    {
      "epoch": 0.30874879824199974,
      "grad_norm": 0.394427627325058,
      "learning_rate": 1.692730520819019e-05,
      "loss": 0.8152,
      "step": 1124
    },
    {
      "epoch": 0.30902348578491967,
      "grad_norm": 0.3278089761734009,
      "learning_rate": 1.6924556822866566e-05,
      "loss": 1.0614,
      "step": 1125
    },
    {
      "epoch": 0.3092981733278396,
      "grad_norm": 0.29241666197776794,
      "learning_rate": 1.6921808437542946e-05,
      "loss": 0.9357,
      "step": 1126
    },
    {
      "epoch": 0.3095728608707595,
      "grad_norm": 0.3260464668273926,
      "learning_rate": 1.6919060052219323e-05,
      "loss": 0.9559,
      "step": 1127
    },
    {
      "epoch": 0.30984754841367945,
      "grad_norm": 0.393545538187027,
      "learning_rate": 1.69163116668957e-05,
      "loss": 0.7874,
      "step": 1128
    },
    {
      "epoch": 0.3101222359565994,
      "grad_norm": 0.3851400315761566,
      "learning_rate": 1.6913563281572077e-05,
      "loss": 0.8577,
      "step": 1129
    },
    {
      "epoch": 0.3103969234995193,
      "grad_norm": 0.3903079330921173,
      "learning_rate": 1.6910814896248454e-05,
      "loss": 1.0338,
      "step": 1130
    },
    {
      "epoch": 0.31067161104243923,
      "grad_norm": 0.3629249036312103,
      "learning_rate": 1.690806651092483e-05,
      "loss": 0.9656,
      "step": 1131
    },
    {
      "epoch": 0.31094629858535916,
      "grad_norm": 0.344447523355484,
      "learning_rate": 1.690531812560121e-05,
      "loss": 0.7703,
      "step": 1132
    },
    {
      "epoch": 0.3112209861282791,
      "grad_norm": 0.27919885516166687,
      "learning_rate": 1.690256974027759e-05,
      "loss": 0.5522,
      "step": 1133
    },
    {
      "epoch": 0.311495673671199,
      "grad_norm": 0.2785206139087677,
      "learning_rate": 1.6899821354953965e-05,
      "loss": 0.5453,
      "step": 1134
    },
    {
      "epoch": 0.31177036121411894,
      "grad_norm": 0.38822874426841736,
      "learning_rate": 1.6897072969630343e-05,
      "loss": 0.7212,
      "step": 1135
    },
    {
      "epoch": 0.31204504875703887,
      "grad_norm": 0.3869542181491852,
      "learning_rate": 1.689432458430672e-05,
      "loss": 0.6632,
      "step": 1136
    },
    {
      "epoch": 0.3123197362999588,
      "grad_norm": 0.23613083362579346,
      "learning_rate": 1.68915761989831e-05,
      "loss": 0.4732,
      "step": 1137
    },
    {
      "epoch": 0.3125944238428787,
      "grad_norm": 0.3181729018688202,
      "learning_rate": 1.6888827813659477e-05,
      "loss": 0.8478,
      "step": 1138
    },
    {
      "epoch": 0.31286911138579865,
      "grad_norm": 0.45161694288253784,
      "learning_rate": 1.6886079428335854e-05,
      "loss": 0.7526,
      "step": 1139
    },
    {
      "epoch": 0.3131437989287186,
      "grad_norm": 0.41017115116119385,
      "learning_rate": 1.688333104301223e-05,
      "loss": 0.8022,
      "step": 1140
    },
    {
      "epoch": 0.3134184864716385,
      "grad_norm": 0.32449468970298767,
      "learning_rate": 1.6880582657688608e-05,
      "loss": 0.7128,
      "step": 1141
    },
    {
      "epoch": 0.31369317401455843,
      "grad_norm": 0.3467469811439514,
      "learning_rate": 1.687783427236499e-05,
      "loss": 0.9192,
      "step": 1142
    },
    {
      "epoch": 0.31396786155747836,
      "grad_norm": 0.2851042151451111,
      "learning_rate": 1.6875085887041365e-05,
      "loss": 0.8342,
      "step": 1143
    },
    {
      "epoch": 0.3142425491003983,
      "grad_norm": 0.3296017348766327,
      "learning_rate": 1.6872337501717742e-05,
      "loss": 1.0602,
      "step": 1144
    },
    {
      "epoch": 0.3145172366433182,
      "grad_norm": 0.39148077368736267,
      "learning_rate": 1.686958911639412e-05,
      "loss": 0.9137,
      "step": 1145
    },
    {
      "epoch": 0.31479192418623814,
      "grad_norm": 0.3021353781223297,
      "learning_rate": 1.6866840731070497e-05,
      "loss": 1.058,
      "step": 1146
    },
    {
      "epoch": 0.31506661172915806,
      "grad_norm": 0.35244086384773254,
      "learning_rate": 1.6864092345746874e-05,
      "loss": 1.0431,
      "step": 1147
    },
    {
      "epoch": 0.315341299272078,
      "grad_norm": 0.4091675579547882,
      "learning_rate": 1.6861343960423254e-05,
      "loss": 0.7207,
      "step": 1148
    },
    {
      "epoch": 0.3156159868149979,
      "grad_norm": 0.2606295347213745,
      "learning_rate": 1.685859557509963e-05,
      "loss": 0.8436,
      "step": 1149
    },
    {
      "epoch": 0.31589067435791784,
      "grad_norm": 0.33511465787887573,
      "learning_rate": 1.6855847189776008e-05,
      "loss": 0.8515,
      "step": 1150
    },
    {
      "epoch": 0.31616536190083777,
      "grad_norm": 0.3676797151565552,
      "learning_rate": 1.6853098804452385e-05,
      "loss": 0.7011,
      "step": 1151
    },
    {
      "epoch": 0.3164400494437577,
      "grad_norm": 0.2767913043498993,
      "learning_rate": 1.6850350419128762e-05,
      "loss": 0.9131,
      "step": 1152
    },
    {
      "epoch": 0.3167147369866777,
      "grad_norm": 0.39779460430145264,
      "learning_rate": 1.6847602033805142e-05,
      "loss": 1.1876,
      "step": 1153
    },
    {
      "epoch": 0.3169894245295976,
      "grad_norm": 0.34843480587005615,
      "learning_rate": 1.684485364848152e-05,
      "loss": 0.6011,
      "step": 1154
    },
    {
      "epoch": 0.31726411207251753,
      "grad_norm": 0.4141124188899994,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 0.8951,
      "step": 1155
    },
    {
      "epoch": 0.31753879961543746,
      "grad_norm": 0.49492746591567993,
      "learning_rate": 1.6839356877834273e-05,
      "loss": 1.025,
      "step": 1156
    },
    {
      "epoch": 0.3178134871583574,
      "grad_norm": 0.4386725127696991,
      "learning_rate": 1.683660849251065e-05,
      "loss": 0.8108,
      "step": 1157
    },
    {
      "epoch": 0.3180881747012773,
      "grad_norm": 0.47087976336479187,
      "learning_rate": 1.683386010718703e-05,
      "loss": 0.9098,
      "step": 1158
    },
    {
      "epoch": 0.31836286224419724,
      "grad_norm": 0.40526247024536133,
      "learning_rate": 1.6831111721863408e-05,
      "loss": 1.1293,
      "step": 1159
    },
    {
      "epoch": 0.31863754978711717,
      "grad_norm": 0.3728295564651489,
      "learning_rate": 1.6828363336539785e-05,
      "loss": 0.9627,
      "step": 1160
    },
    {
      "epoch": 0.3189122373300371,
      "grad_norm": 0.364479660987854,
      "learning_rate": 1.6825614951216162e-05,
      "loss": 0.604,
      "step": 1161
    },
    {
      "epoch": 0.319186924872957,
      "grad_norm": 0.429949551820755,
      "learning_rate": 1.682286656589254e-05,
      "loss": 0.8042,
      "step": 1162
    },
    {
      "epoch": 0.31946161241587695,
      "grad_norm": 0.30086463689804077,
      "learning_rate": 1.6820118180568916e-05,
      "loss": 0.7433,
      "step": 1163
    },
    {
      "epoch": 0.3197362999587969,
      "grad_norm": 0.30742931365966797,
      "learning_rate": 1.6817369795245296e-05,
      "loss": 0.519,
      "step": 1164
    },
    {
      "epoch": 0.3200109875017168,
      "grad_norm": 0.34003183245658875,
      "learning_rate": 1.6814621409921673e-05,
      "loss": 0.8458,
      "step": 1165
    },
    {
      "epoch": 0.32028567504463673,
      "grad_norm": 0.34128716588020325,
      "learning_rate": 1.681187302459805e-05,
      "loss": 0.657,
      "step": 1166
    },
    {
      "epoch": 0.32056036258755666,
      "grad_norm": 0.43373388051986694,
      "learning_rate": 1.6809124639274427e-05,
      "loss": 0.6904,
      "step": 1167
    },
    {
      "epoch": 0.3208350501304766,
      "grad_norm": 0.45134663581848145,
      "learning_rate": 1.6806376253950805e-05,
      "loss": 0.7855,
      "step": 1168
    },
    {
      "epoch": 0.3211097376733965,
      "grad_norm": 0.28571438789367676,
      "learning_rate": 1.6803627868627185e-05,
      "loss": 0.6004,
      "step": 1169
    },
    {
      "epoch": 0.32138442521631644,
      "grad_norm": 0.2459426373243332,
      "learning_rate": 1.6800879483303562e-05,
      "loss": 0.6476,
      "step": 1170
    },
    {
      "epoch": 0.32165911275923637,
      "grad_norm": 0.30403026938438416,
      "learning_rate": 1.6798131097979936e-05,
      "loss": 0.6445,
      "step": 1171
    },
    {
      "epoch": 0.3219338003021563,
      "grad_norm": 0.438427209854126,
      "learning_rate": 1.6795382712656316e-05,
      "loss": 0.8484,
      "step": 1172
    },
    {
      "epoch": 0.3222084878450762,
      "grad_norm": 0.3235408663749695,
      "learning_rate": 1.6792634327332693e-05,
      "loss": 0.9035,
      "step": 1173
    },
    {
      "epoch": 0.32248317538799615,
      "grad_norm": 0.38426846265792847,
      "learning_rate": 1.6789885942009073e-05,
      "loss": 0.839,
      "step": 1174
    },
    {
      "epoch": 0.3227578629309161,
      "grad_norm": 0.2954728603363037,
      "learning_rate": 1.678713755668545e-05,
      "loss": 0.7352,
      "step": 1175
    },
    {
      "epoch": 0.323032550473836,
      "grad_norm": 0.3662184774875641,
      "learning_rate": 1.6784389171361824e-05,
      "loss": 0.8373,
      "step": 1176
    },
    {
      "epoch": 0.3233072380167559,
      "grad_norm": 0.3195166289806366,
      "learning_rate": 1.6781640786038204e-05,
      "loss": 0.7579,
      "step": 1177
    },
    {
      "epoch": 0.32358192555967585,
      "grad_norm": 0.3139706552028656,
      "learning_rate": 1.677889240071458e-05,
      "loss": 0.8545,
      "step": 1178
    },
    {
      "epoch": 0.3238566131025958,
      "grad_norm": 0.3768098056316376,
      "learning_rate": 1.677614401539096e-05,
      "loss": 0.9488,
      "step": 1179
    },
    {
      "epoch": 0.3241313006455157,
      "grad_norm": 0.4428572356700897,
      "learning_rate": 1.677339563006734e-05,
      "loss": 0.8693,
      "step": 1180
    },
    {
      "epoch": 0.32440598818843563,
      "grad_norm": 0.3688257932662964,
      "learning_rate": 1.6770647244743713e-05,
      "loss": 0.5613,
      "step": 1181
    },
    {
      "epoch": 0.32468067573135556,
      "grad_norm": 0.5518360733985901,
      "learning_rate": 1.6767898859420093e-05,
      "loss": 0.8119,
      "step": 1182
    },
    {
      "epoch": 0.3249553632742755,
      "grad_norm": 0.2784770429134369,
      "learning_rate": 1.676515047409647e-05,
      "loss": 0.9703,
      "step": 1183
    },
    {
      "epoch": 0.3252300508171954,
      "grad_norm": 0.3850032389163971,
      "learning_rate": 1.6762402088772847e-05,
      "loss": 0.8544,
      "step": 1184
    },
    {
      "epoch": 0.32550473836011534,
      "grad_norm": 0.36670202016830444,
      "learning_rate": 1.6759653703449224e-05,
      "loss": 1.0589,
      "step": 1185
    },
    {
      "epoch": 0.3257794259030353,
      "grad_norm": 0.24353590607643127,
      "learning_rate": 1.67569053181256e-05,
      "loss": 0.7335,
      "step": 1186
    },
    {
      "epoch": 0.32605411344595525,
      "grad_norm": 0.309049516916275,
      "learning_rate": 1.6754156932801978e-05,
      "loss": 0.6852,
      "step": 1187
    },
    {
      "epoch": 0.3263288009888752,
      "grad_norm": 0.259022057056427,
      "learning_rate": 1.675140854747836e-05,
      "loss": 0.6969,
      "step": 1188
    },
    {
      "epoch": 0.3266034885317951,
      "grad_norm": 0.4289526641368866,
      "learning_rate": 1.6748660162154736e-05,
      "loss": 0.8395,
      "step": 1189
    },
    {
      "epoch": 0.32687817607471503,
      "grad_norm": 0.30379652976989746,
      "learning_rate": 1.6745911776831113e-05,
      "loss": 0.8926,
      "step": 1190
    },
    {
      "epoch": 0.32715286361763496,
      "grad_norm": 0.4871966242790222,
      "learning_rate": 1.674316339150749e-05,
      "loss": 0.9208,
      "step": 1191
    },
    {
      "epoch": 0.3274275511605549,
      "grad_norm": 0.30578678846359253,
      "learning_rate": 1.6740415006183867e-05,
      "loss": 1.0863,
      "step": 1192
    },
    {
      "epoch": 0.3277022387034748,
      "grad_norm": 0.3045530319213867,
      "learning_rate": 1.6737666620860247e-05,
      "loss": 0.6788,
      "step": 1193
    },
    {
      "epoch": 0.32797692624639474,
      "grad_norm": 0.3266822099685669,
      "learning_rate": 1.6734918235536624e-05,
      "loss": 1.036,
      "step": 1194
    },
    {
      "epoch": 0.32825161378931467,
      "grad_norm": 0.3869270384311676,
      "learning_rate": 1.6732169850213e-05,
      "loss": 0.7817,
      "step": 1195
    },
    {
      "epoch": 0.3285263013322346,
      "grad_norm": 0.3550933599472046,
      "learning_rate": 1.6729421464889378e-05,
      "loss": 0.7707,
      "step": 1196
    },
    {
      "epoch": 0.3288009888751545,
      "grad_norm": 0.28778356313705444,
      "learning_rate": 1.6726673079565755e-05,
      "loss": 1.0505,
      "step": 1197
    },
    {
      "epoch": 0.32907567641807445,
      "grad_norm": 0.4342587888240814,
      "learning_rate": 1.6723924694242135e-05,
      "loss": 0.8908,
      "step": 1198
    },
    {
      "epoch": 0.3293503639609944,
      "grad_norm": 0.26576003432273865,
      "learning_rate": 1.6721176308918512e-05,
      "loss": 0.6607,
      "step": 1199
    },
    {
      "epoch": 0.3296250515039143,
      "grad_norm": 0.3413148522377014,
      "learning_rate": 1.671842792359489e-05,
      "loss": 0.7593,
      "step": 1200
    },
    {
      "epoch": 0.32989973904683423,
      "grad_norm": 0.4704531133174896,
      "learning_rate": 1.6715679538271267e-05,
      "loss": 0.8955,
      "step": 1201
    },
    {
      "epoch": 0.33017442658975416,
      "grad_norm": 0.44820237159729004,
      "learning_rate": 1.6712931152947644e-05,
      "loss": 0.86,
      "step": 1202
    },
    {
      "epoch": 0.3304491141326741,
      "grad_norm": 0.40939849615097046,
      "learning_rate": 1.671018276762402e-05,
      "loss": 0.9209,
      "step": 1203
    },
    {
      "epoch": 0.330723801675594,
      "grad_norm": 0.2468651533126831,
      "learning_rate": 1.67074343823004e-05,
      "loss": 0.7344,
      "step": 1204
    },
    {
      "epoch": 0.33099848921851394,
      "grad_norm": 0.633234977722168,
      "learning_rate": 1.6704685996976778e-05,
      "loss": 0.8763,
      "step": 1205
    },
    {
      "epoch": 0.33127317676143386,
      "grad_norm": 0.3937934637069702,
      "learning_rate": 1.6701937611653155e-05,
      "loss": 0.7743,
      "step": 1206
    },
    {
      "epoch": 0.3315478643043538,
      "grad_norm": 0.31630536913871765,
      "learning_rate": 1.6699189226329532e-05,
      "loss": 0.7805,
      "step": 1207
    },
    {
      "epoch": 0.3318225518472737,
      "grad_norm": 0.42655953764915466,
      "learning_rate": 1.669644084100591e-05,
      "loss": 0.8642,
      "step": 1208
    },
    {
      "epoch": 0.33209723939019364,
      "grad_norm": 0.4292066991329193,
      "learning_rate": 1.669369245568229e-05,
      "loss": 0.9852,
      "step": 1209
    },
    {
      "epoch": 0.33237192693311357,
      "grad_norm": 0.3878101408481598,
      "learning_rate": 1.6690944070358666e-05,
      "loss": 1.0403,
      "step": 1210
    },
    {
      "epoch": 0.3326466144760335,
      "grad_norm": 0.37578219175338745,
      "learning_rate": 1.6688195685035044e-05,
      "loss": 0.9049,
      "step": 1211
    },
    {
      "epoch": 0.3329213020189534,
      "grad_norm": 0.35708925127983093,
      "learning_rate": 1.668544729971142e-05,
      "loss": 0.7512,
      "step": 1212
    },
    {
      "epoch": 0.33319598956187335,
      "grad_norm": 0.2525728642940521,
      "learning_rate": 1.6682698914387798e-05,
      "loss": 0.4416,
      "step": 1213
    },
    {
      "epoch": 0.3334706771047933,
      "grad_norm": 0.29642724990844727,
      "learning_rate": 1.6679950529064178e-05,
      "loss": 0.8958,
      "step": 1214
    },
    {
      "epoch": 0.3337453646477132,
      "grad_norm": 0.4493308961391449,
      "learning_rate": 1.6677202143740555e-05,
      "loss": 0.9591,
      "step": 1215
    },
    {
      "epoch": 0.33402005219063313,
      "grad_norm": 0.3432091474533081,
      "learning_rate": 1.6674453758416932e-05,
      "loss": 1.0462,
      "step": 1216
    },
    {
      "epoch": 0.33429473973355306,
      "grad_norm": 0.35243237018585205,
      "learning_rate": 1.667170537309331e-05,
      "loss": 0.9193,
      "step": 1217
    },
    {
      "epoch": 0.334569427276473,
      "grad_norm": 0.2755392789840698,
      "learning_rate": 1.6668956987769686e-05,
      "loss": 0.7076,
      "step": 1218
    },
    {
      "epoch": 0.3348441148193929,
      "grad_norm": 0.34402406215667725,
      "learning_rate": 1.6666208602446063e-05,
      "loss": 0.8099,
      "step": 1219
    },
    {
      "epoch": 0.3351188023623129,
      "grad_norm": 0.27781257033348083,
      "learning_rate": 1.6663460217122443e-05,
      "loss": 0.6888,
      "step": 1220
    },
    {
      "epoch": 0.3353934899052328,
      "grad_norm": 0.4169047772884369,
      "learning_rate": 1.666071183179882e-05,
      "loss": 0.8358,
      "step": 1221
    },
    {
      "epoch": 0.33566817744815275,
      "grad_norm": 0.4277589023113251,
      "learning_rate": 1.6657963446475198e-05,
      "loss": 0.8536,
      "step": 1222
    },
    {
      "epoch": 0.3359428649910727,
      "grad_norm": 0.4160520136356354,
      "learning_rate": 1.6655215061151575e-05,
      "loss": 0.6051,
      "step": 1223
    },
    {
      "epoch": 0.3362175525339926,
      "grad_norm": 0.4532460868358612,
      "learning_rate": 1.665246667582795e-05,
      "loss": 0.9613,
      "step": 1224
    },
    {
      "epoch": 0.33649224007691253,
      "grad_norm": 0.3269125521183014,
      "learning_rate": 1.6649718290504332e-05,
      "loss": 0.6738,
      "step": 1225
    },
    {
      "epoch": 0.33676692761983246,
      "grad_norm": 0.4141317903995514,
      "learning_rate": 1.664696990518071e-05,
      "loss": 0.6811,
      "step": 1226
    },
    {
      "epoch": 0.3370416151627524,
      "grad_norm": 0.32968980073928833,
      "learning_rate": 1.6644221519857086e-05,
      "loss": 0.8715,
      "step": 1227
    },
    {
      "epoch": 0.3373163027056723,
      "grad_norm": 0.32713818550109863,
      "learning_rate": 1.6641473134533463e-05,
      "loss": 0.6585,
      "step": 1228
    },
    {
      "epoch": 0.33759099024859224,
      "grad_norm": 0.39593490958213806,
      "learning_rate": 1.663872474920984e-05,
      "loss": 0.9168,
      "step": 1229
    },
    {
      "epoch": 0.33786567779151216,
      "grad_norm": 0.34461385011672974,
      "learning_rate": 1.663597636388622e-05,
      "loss": 0.9,
      "step": 1230
    },
    {
      "epoch": 0.3381403653344321,
      "grad_norm": 0.3687352240085602,
      "learning_rate": 1.6633227978562597e-05,
      "loss": 0.8552,
      "step": 1231
    },
    {
      "epoch": 0.338415052877352,
      "grad_norm": 0.2653319835662842,
      "learning_rate": 1.6630479593238975e-05,
      "loss": 0.6512,
      "step": 1232
    },
    {
      "epoch": 0.33868974042027195,
      "grad_norm": 0.347817063331604,
      "learning_rate": 1.662773120791535e-05,
      "loss": 0.8206,
      "step": 1233
    },
    {
      "epoch": 0.33896442796319187,
      "grad_norm": 0.4618756175041199,
      "learning_rate": 1.662498282259173e-05,
      "loss": 0.908,
      "step": 1234
    },
    {
      "epoch": 0.3392391155061118,
      "grad_norm": 0.38594162464141846,
      "learning_rate": 1.6622234437268106e-05,
      "loss": 1.0172,
      "step": 1235
    },
    {
      "epoch": 0.3395138030490317,
      "grad_norm": 0.4470430612564087,
      "learning_rate": 1.6619486051944486e-05,
      "loss": 0.7545,
      "step": 1236
    },
    {
      "epoch": 0.33978849059195165,
      "grad_norm": 0.4241577386856079,
      "learning_rate": 1.661673766662086e-05,
      "loss": 1.0297,
      "step": 1237
    },
    {
      "epoch": 0.3400631781348716,
      "grad_norm": 0.5146130323410034,
      "learning_rate": 1.661398928129724e-05,
      "loss": 0.8048,
      "step": 1238
    },
    {
      "epoch": 0.3403378656777915,
      "grad_norm": 0.3316795527935028,
      "learning_rate": 1.6611240895973617e-05,
      "loss": 0.6409,
      "step": 1239
    },
    {
      "epoch": 0.34061255322071143,
      "grad_norm": 0.49625396728515625,
      "learning_rate": 1.6608492510649994e-05,
      "loss": 0.9832,
      "step": 1240
    },
    {
      "epoch": 0.34088724076363136,
      "grad_norm": 0.4560220241546631,
      "learning_rate": 1.6605744125326374e-05,
      "loss": 1.0293,
      "step": 1241
    },
    {
      "epoch": 0.3411619283065513,
      "grad_norm": 0.35307249426841736,
      "learning_rate": 1.6602995740002748e-05,
      "loss": 0.9638,
      "step": 1242
    },
    {
      "epoch": 0.3414366158494712,
      "grad_norm": 0.39645859599113464,
      "learning_rate": 1.6600247354679125e-05,
      "loss": 1.0762,
      "step": 1243
    },
    {
      "epoch": 0.34171130339239114,
      "grad_norm": 0.4066542387008667,
      "learning_rate": 1.6597498969355506e-05,
      "loss": 0.8938,
      "step": 1244
    },
    {
      "epoch": 0.34198599093531107,
      "grad_norm": 0.3553965985774994,
      "learning_rate": 1.6594750584031883e-05,
      "loss": 1.1758,
      "step": 1245
    },
    {
      "epoch": 0.342260678478231,
      "grad_norm": 0.3413005471229553,
      "learning_rate": 1.6592002198708263e-05,
      "loss": 0.7294,
      "step": 1246
    },
    {
      "epoch": 0.3425353660211509,
      "grad_norm": 0.48920294642448425,
      "learning_rate": 1.6589253813384637e-05,
      "loss": 1.1316,
      "step": 1247
    },
    {
      "epoch": 0.34281005356407085,
      "grad_norm": 0.4177711606025696,
      "learning_rate": 1.6586505428061014e-05,
      "loss": 0.7497,
      "step": 1248
    },
    {
      "epoch": 0.3430847411069908,
      "grad_norm": 0.27461886405944824,
      "learning_rate": 1.6583757042737394e-05,
      "loss": 0.681,
      "step": 1249
    },
    {
      "epoch": 0.3433594286499107,
      "grad_norm": 0.3341595232486725,
      "learning_rate": 1.658100865741377e-05,
      "loss": 1.0095,
      "step": 1250
    },
    {
      "epoch": 0.34363411619283063,
      "grad_norm": 0.3425182104110718,
      "learning_rate": 1.6578260272090148e-05,
      "loss": 0.9355,
      "step": 1251
    },
    {
      "epoch": 0.34390880373575056,
      "grad_norm": 0.3662249445915222,
      "learning_rate": 1.6575511886766525e-05,
      "loss": 0.8446,
      "step": 1252
    },
    {
      "epoch": 0.34418349127867054,
      "grad_norm": 0.32262954115867615,
      "learning_rate": 1.6572763501442902e-05,
      "loss": 0.689,
      "step": 1253
    },
    {
      "epoch": 0.34445817882159047,
      "grad_norm": 0.36929795145988464,
      "learning_rate": 1.6570015116119283e-05,
      "loss": 0.8083,
      "step": 1254
    },
    {
      "epoch": 0.3447328663645104,
      "grad_norm": 0.4620105028152466,
      "learning_rate": 1.656726673079566e-05,
      "loss": 0.9714,
      "step": 1255
    },
    {
      "epoch": 0.3450075539074303,
      "grad_norm": 0.47035789489746094,
      "learning_rate": 1.6564518345472037e-05,
      "loss": 0.8694,
      "step": 1256
    },
    {
      "epoch": 0.34528224145035025,
      "grad_norm": 0.3728163242340088,
      "learning_rate": 1.6561769960148414e-05,
      "loss": 0.9627,
      "step": 1257
    },
    {
      "epoch": 0.3455569289932702,
      "grad_norm": 0.33395639061927795,
      "learning_rate": 1.655902157482479e-05,
      "loss": 0.7389,
      "step": 1258
    },
    {
      "epoch": 0.3458316165361901,
      "grad_norm": 0.3573935627937317,
      "learning_rate": 1.6556273189501168e-05,
      "loss": 1.018,
      "step": 1259
    },
    {
      "epoch": 0.34610630407911,
      "grad_norm": 0.33720162510871887,
      "learning_rate": 1.6553524804177548e-05,
      "loss": 0.5921,
      "step": 1260
    },
    {
      "epoch": 0.34638099162202995,
      "grad_norm": 0.3541867434978485,
      "learning_rate": 1.6550776418853925e-05,
      "loss": 0.8844,
      "step": 1261
    },
    {
      "epoch": 0.3466556791649499,
      "grad_norm": 0.3371075987815857,
      "learning_rate": 1.6548028033530302e-05,
      "loss": 0.7556,
      "step": 1262
    },
    {
      "epoch": 0.3469303667078698,
      "grad_norm": 0.24980029463768005,
      "learning_rate": 1.654527964820668e-05,
      "loss": 0.713,
      "step": 1263
    },
    {
      "epoch": 0.34720505425078974,
      "grad_norm": 0.3725094199180603,
      "learning_rate": 1.6542531262883056e-05,
      "loss": 0.9329,
      "step": 1264
    },
    {
      "epoch": 0.34747974179370966,
      "grad_norm": 0.5015193223953247,
      "learning_rate": 1.6539782877559437e-05,
      "loss": 0.8772,
      "step": 1265
    },
    {
      "epoch": 0.3477544293366296,
      "grad_norm": 0.36596518754959106,
      "learning_rate": 1.6537034492235814e-05,
      "loss": 0.8993,
      "step": 1266
    },
    {
      "epoch": 0.3480291168795495,
      "grad_norm": 0.3003566563129425,
      "learning_rate": 1.653428610691219e-05,
      "loss": 0.6793,
      "step": 1267
    },
    {
      "epoch": 0.34830380442246944,
      "grad_norm": 0.43510475754737854,
      "learning_rate": 1.6531537721588568e-05,
      "loss": 0.8482,
      "step": 1268
    },
    {
      "epoch": 0.34857849196538937,
      "grad_norm": 0.5741549134254456,
      "learning_rate": 1.6528789336264945e-05,
      "loss": 0.8051,
      "step": 1269
    },
    {
      "epoch": 0.3488531795083093,
      "grad_norm": 0.28831639885902405,
      "learning_rate": 1.6526040950941325e-05,
      "loss": 0.9313,
      "step": 1270
    },
    {
      "epoch": 0.3491278670512292,
      "grad_norm": 0.32726478576660156,
      "learning_rate": 1.6523292565617702e-05,
      "loss": 1.0264,
      "step": 1271
    },
    {
      "epoch": 0.34940255459414915,
      "grad_norm": 0.3154226541519165,
      "learning_rate": 1.652054418029408e-05,
      "loss": 0.6141,
      "step": 1272
    },
    {
      "epoch": 0.3496772421370691,
      "grad_norm": 0.3713774085044861,
      "learning_rate": 1.6517795794970456e-05,
      "loss": 1.0626,
      "step": 1273
    },
    {
      "epoch": 0.349951929679989,
      "grad_norm": 0.37231898307800293,
      "learning_rate": 1.6515047409646833e-05,
      "loss": 0.7066,
      "step": 1274
    },
    {
      "epoch": 0.35022661722290893,
      "grad_norm": 0.3349222242832184,
      "learning_rate": 1.651229902432321e-05,
      "loss": 0.8452,
      "step": 1275
    },
    {
      "epoch": 0.35050130476582886,
      "grad_norm": 0.31706517934799194,
      "learning_rate": 1.650955063899959e-05,
      "loss": 0.8976,
      "step": 1276
    },
    {
      "epoch": 0.3507759923087488,
      "grad_norm": 0.3562261164188385,
      "learning_rate": 1.6506802253675968e-05,
      "loss": 0.5764,
      "step": 1277
    },
    {
      "epoch": 0.3510506798516687,
      "grad_norm": 0.32737696170806885,
      "learning_rate": 1.6504053868352345e-05,
      "loss": 0.8003,
      "step": 1278
    },
    {
      "epoch": 0.35132536739458864,
      "grad_norm": 0.33902522921562195,
      "learning_rate": 1.650130548302872e-05,
      "loss": 0.6821,
      "step": 1279
    },
    {
      "epoch": 0.35160005493750857,
      "grad_norm": 0.492713063955307,
      "learning_rate": 1.64985570977051e-05,
      "loss": 0.9496,
      "step": 1280
    },
    {
      "epoch": 0.3518747424804285,
      "grad_norm": 0.36341044306755066,
      "learning_rate": 1.649580871238148e-05,
      "loss": 0.7094,
      "step": 1281
    },
    {
      "epoch": 0.3521494300233484,
      "grad_norm": 0.536319375038147,
      "learning_rate": 1.6493060327057856e-05,
      "loss": 0.8643,
      "step": 1282
    },
    {
      "epoch": 0.35242411756626835,
      "grad_norm": 0.3902081847190857,
      "learning_rate": 1.6490311941734233e-05,
      "loss": 0.7581,
      "step": 1283
    },
    {
      "epoch": 0.3526988051091883,
      "grad_norm": 0.3884584605693817,
      "learning_rate": 1.648756355641061e-05,
      "loss": 0.695,
      "step": 1284
    },
    {
      "epoch": 0.3529734926521082,
      "grad_norm": 0.3148411214351654,
      "learning_rate": 1.6484815171086987e-05,
      "loss": 0.5874,
      "step": 1285
    },
    {
      "epoch": 0.3532481801950282,
      "grad_norm": 0.35072454810142517,
      "learning_rate": 1.6482066785763364e-05,
      "loss": 0.9752,
      "step": 1286
    },
    {
      "epoch": 0.3535228677379481,
      "grad_norm": 0.4175049662590027,
      "learning_rate": 1.6479318400439745e-05,
      "loss": 0.8564,
      "step": 1287
    },
    {
      "epoch": 0.35379755528086804,
      "grad_norm": 0.37938231229782104,
      "learning_rate": 1.647657001511612e-05,
      "loss": 0.9746,
      "step": 1288
    },
    {
      "epoch": 0.35407224282378796,
      "grad_norm": 0.3139869272708893,
      "learning_rate": 1.64738216297925e-05,
      "loss": 0.8091,
      "step": 1289
    },
    {
      "epoch": 0.3543469303667079,
      "grad_norm": 0.30755332112312317,
      "learning_rate": 1.6471073244468876e-05,
      "loss": 0.8648,
      "step": 1290
    },
    {
      "epoch": 0.3546216179096278,
      "grad_norm": 0.4068838953971863,
      "learning_rate": 1.6468324859145253e-05,
      "loss": 0.9898,
      "step": 1291
    },
    {
      "epoch": 0.35489630545254774,
      "grad_norm": 0.3245755732059479,
      "learning_rate": 1.6465576473821633e-05,
      "loss": 0.9635,
      "step": 1292
    },
    {
      "epoch": 0.35517099299546767,
      "grad_norm": 0.44564133882522583,
      "learning_rate": 1.646282808849801e-05,
      "loss": 0.85,
      "step": 1293
    },
    {
      "epoch": 0.3554456805383876,
      "grad_norm": 0.2871597707271576,
      "learning_rate": 1.6460079703174384e-05,
      "loss": 0.7478,
      "step": 1294
    },
    {
      "epoch": 0.3557203680813075,
      "grad_norm": 0.3044035732746124,
      "learning_rate": 1.6457331317850764e-05,
      "loss": 0.9005,
      "step": 1295
    },
    {
      "epoch": 0.35599505562422745,
      "grad_norm": 0.35351353883743286,
      "learning_rate": 1.645458293252714e-05,
      "loss": 0.8548,
      "step": 1296
    },
    {
      "epoch": 0.3562697431671474,
      "grad_norm": 0.3598310947418213,
      "learning_rate": 1.645183454720352e-05,
      "loss": 0.5388,
      "step": 1297
    },
    {
      "epoch": 0.3565444307100673,
      "grad_norm": 0.42419594526290894,
      "learning_rate": 1.64490861618799e-05,
      "loss": 0.8384,
      "step": 1298
    },
    {
      "epoch": 0.35681911825298723,
      "grad_norm": 0.3013259768486023,
      "learning_rate": 1.6446337776556272e-05,
      "loss": 0.8162,
      "step": 1299
    },
    {
      "epoch": 0.35709380579590716,
      "grad_norm": 0.45980560779571533,
      "learning_rate": 1.6443589391232653e-05,
      "loss": 0.8064,
      "step": 1300
    },
    {
      "epoch": 0.3573684933388271,
      "grad_norm": 0.30857956409454346,
      "learning_rate": 1.644084100590903e-05,
      "loss": 0.9797,
      "step": 1301
    },
    {
      "epoch": 0.357643180881747,
      "grad_norm": 0.3772880733013153,
      "learning_rate": 1.6438092620585407e-05,
      "loss": 1.0329,
      "step": 1302
    },
    {
      "epoch": 0.35791786842466694,
      "grad_norm": 0.33271050453186035,
      "learning_rate": 1.6435344235261787e-05,
      "loss": 0.7437,
      "step": 1303
    },
    {
      "epoch": 0.35819255596758687,
      "grad_norm": 0.3434244990348816,
      "learning_rate": 1.643259584993816e-05,
      "loss": 0.6971,
      "step": 1304
    },
    {
      "epoch": 0.3584672435105068,
      "grad_norm": 0.3274836242198944,
      "learning_rate": 1.642984746461454e-05,
      "loss": 0.8256,
      "step": 1305
    },
    {
      "epoch": 0.3587419310534267,
      "grad_norm": 0.3942471742630005,
      "learning_rate": 1.6427099079290918e-05,
      "loss": 0.9447,
      "step": 1306
    },
    {
      "epoch": 0.35901661859634665,
      "grad_norm": 0.23653709888458252,
      "learning_rate": 1.6424350693967295e-05,
      "loss": 0.5517,
      "step": 1307
    },
    {
      "epoch": 0.3592913061392666,
      "grad_norm": 0.43571728467941284,
      "learning_rate": 1.6421602308643672e-05,
      "loss": 0.8124,
      "step": 1308
    },
    {
      "epoch": 0.3595659936821865,
      "grad_norm": 0.30755752325057983,
      "learning_rate": 1.641885392332005e-05,
      "loss": 0.5674,
      "step": 1309
    },
    {
      "epoch": 0.35984068122510643,
      "grad_norm": 0.3727714717388153,
      "learning_rate": 1.6416105537996426e-05,
      "loss": 0.6358,
      "step": 1310
    },
    {
      "epoch": 0.36011536876802636,
      "grad_norm": 0.2987203896045685,
      "learning_rate": 1.6413357152672807e-05,
      "loss": 0.8401,
      "step": 1311
    },
    {
      "epoch": 0.3603900563109463,
      "grad_norm": 1.3857454061508179,
      "learning_rate": 1.6410608767349184e-05,
      "loss": 0.7758,
      "step": 1312
    },
    {
      "epoch": 0.3606647438538662,
      "grad_norm": 0.3509828448295593,
      "learning_rate": 1.640786038202556e-05,
      "loss": 0.9084,
      "step": 1313
    },
    {
      "epoch": 0.36093943139678614,
      "grad_norm": 0.4161381721496582,
      "learning_rate": 1.6405111996701938e-05,
      "loss": 0.8135,
      "step": 1314
    },
    {
      "epoch": 0.36121411893970606,
      "grad_norm": 0.32911354303359985,
      "learning_rate": 1.6402363611378315e-05,
      "loss": 1.1211,
      "step": 1315
    },
    {
      "epoch": 0.361488806482626,
      "grad_norm": 0.32821768522262573,
      "learning_rate": 1.6399615226054695e-05,
      "loss": 0.677,
      "step": 1316
    },
    {
      "epoch": 0.3617634940255459,
      "grad_norm": 0.38179028034210205,
      "learning_rate": 1.6396866840731072e-05,
      "loss": 1.0807,
      "step": 1317
    },
    {
      "epoch": 0.36203818156846584,
      "grad_norm": 0.3669012486934662,
      "learning_rate": 1.639411845540745e-05,
      "loss": 0.9053,
      "step": 1318
    },
    {
      "epoch": 0.36231286911138577,
      "grad_norm": 0.2529803514480591,
      "learning_rate": 1.6391370070083826e-05,
      "loss": 0.7509,
      "step": 1319
    },
    {
      "epoch": 0.36258755665430575,
      "grad_norm": 0.35325631499290466,
      "learning_rate": 1.6388621684760203e-05,
      "loss": 1.1046,
      "step": 1320
    },
    {
      "epoch": 0.3628622441972257,
      "grad_norm": 0.30346909165382385,
      "learning_rate": 1.6385873299436584e-05,
      "loss": 0.7557,
      "step": 1321
    },
    {
      "epoch": 0.3631369317401456,
      "grad_norm": 0.25379908084869385,
      "learning_rate": 1.638312491411296e-05,
      "loss": 0.8782,
      "step": 1322
    },
    {
      "epoch": 0.36341161928306553,
      "grad_norm": 0.3699788749217987,
      "learning_rate": 1.6380376528789338e-05,
      "loss": 0.8872,
      "step": 1323
    },
    {
      "epoch": 0.36368630682598546,
      "grad_norm": 0.2910076677799225,
      "learning_rate": 1.6377628143465715e-05,
      "loss": 0.5972,
      "step": 1324
    },
    {
      "epoch": 0.3639609943689054,
      "grad_norm": 0.33011817932128906,
      "learning_rate": 1.637487975814209e-05,
      "loss": 0.6301,
      "step": 1325
    },
    {
      "epoch": 0.3642356819118253,
      "grad_norm": 0.31837645173072815,
      "learning_rate": 1.637213137281847e-05,
      "loss": 0.939,
      "step": 1326
    },
    {
      "epoch": 0.36451036945474524,
      "grad_norm": 0.2299579530954361,
      "learning_rate": 1.636938298749485e-05,
      "loss": 0.7103,
      "step": 1327
    },
    {
      "epoch": 0.36478505699766517,
      "grad_norm": 0.413394957780838,
      "learning_rate": 1.6366634602171226e-05,
      "loss": 0.6857,
      "step": 1328
    },
    {
      "epoch": 0.3650597445405851,
      "grad_norm": 0.4408571124076843,
      "learning_rate": 1.6363886216847603e-05,
      "loss": 0.8828,
      "step": 1329
    },
    {
      "epoch": 0.365334432083505,
      "grad_norm": 0.3481674790382385,
      "learning_rate": 1.636113783152398e-05,
      "loss": 0.9247,
      "step": 1330
    },
    {
      "epoch": 0.36560911962642495,
      "grad_norm": 0.30241096019744873,
      "learning_rate": 1.6358389446200357e-05,
      "loss": 0.7896,
      "step": 1331
    },
    {
      "epoch": 0.3658838071693449,
      "grad_norm": 0.3244688808917999,
      "learning_rate": 1.6355641060876738e-05,
      "loss": 0.5485,
      "step": 1332
    },
    {
      "epoch": 0.3661584947122648,
      "grad_norm": 0.4731455147266388,
      "learning_rate": 1.6352892675553115e-05,
      "loss": 0.8104,
      "step": 1333
    },
    {
      "epoch": 0.36643318225518473,
      "grad_norm": 0.42399853467941284,
      "learning_rate": 1.635014429022949e-05,
      "loss": 0.8769,
      "step": 1334
    },
    {
      "epoch": 0.36670786979810466,
      "grad_norm": 0.4523649513721466,
      "learning_rate": 1.634739590490587e-05,
      "loss": 0.874,
      "step": 1335
    },
    {
      "epoch": 0.3669825573410246,
      "grad_norm": 0.26956915855407715,
      "learning_rate": 1.6344647519582246e-05,
      "loss": 0.5023,
      "step": 1336
    },
    {
      "epoch": 0.3672572448839445,
      "grad_norm": 0.2787874639034271,
      "learning_rate": 1.6341899134258626e-05,
      "loss": 0.7133,
      "step": 1337
    },
    {
      "epoch": 0.36753193242686444,
      "grad_norm": 0.2705073356628418,
      "learning_rate": 1.6339150748935003e-05,
      "loss": 0.5154,
      "step": 1338
    },
    {
      "epoch": 0.36780661996978437,
      "grad_norm": 0.343721866607666,
      "learning_rate": 1.633640236361138e-05,
      "loss": 1.1026,
      "step": 1339
    },
    {
      "epoch": 0.3680813075127043,
      "grad_norm": 0.33782967925071716,
      "learning_rate": 1.6333653978287757e-05,
      "loss": 0.8756,
      "step": 1340
    },
    {
      "epoch": 0.3683559950556242,
      "grad_norm": 0.3685656189918518,
      "learning_rate": 1.6330905592964134e-05,
      "loss": 0.8485,
      "step": 1341
    },
    {
      "epoch": 0.36863068259854415,
      "grad_norm": 0.33008068799972534,
      "learning_rate": 1.632815720764051e-05,
      "loss": 0.6044,
      "step": 1342
    },
    {
      "epoch": 0.3689053701414641,
      "grad_norm": 0.26396816968917847,
      "learning_rate": 1.632540882231689e-05,
      "loss": 0.8258,
      "step": 1343
    },
    {
      "epoch": 0.369180057684384,
      "grad_norm": 0.3352273404598236,
      "learning_rate": 1.632266043699327e-05,
      "loss": 0.6611,
      "step": 1344
    },
    {
      "epoch": 0.3694547452273039,
      "grad_norm": 0.3303733468055725,
      "learning_rate": 1.6319912051669646e-05,
      "loss": 0.7367,
      "step": 1345
    },
    {
      "epoch": 0.36972943277022385,
      "grad_norm": 0.3975073993206024,
      "learning_rate": 1.6317163666346023e-05,
      "loss": 0.9076,
      "step": 1346
    },
    {
      "epoch": 0.3700041203131438,
      "grad_norm": 0.6096312403678894,
      "learning_rate": 1.63144152810224e-05,
      "loss": 1.033,
      "step": 1347
    },
    {
      "epoch": 0.3702788078560637,
      "grad_norm": 0.412743479013443,
      "learning_rate": 1.631166689569878e-05,
      "loss": 0.8726,
      "step": 1348
    },
    {
      "epoch": 0.37055349539898363,
      "grad_norm": 0.328474760055542,
      "learning_rate": 1.6308918510375157e-05,
      "loss": 0.7962,
      "step": 1349
    },
    {
      "epoch": 0.37082818294190356,
      "grad_norm": 0.45432940125465393,
      "learning_rate": 1.6306170125051534e-05,
      "loss": 0.8646,
      "step": 1350
    },
    {
      "epoch": 0.3711028704848235,
      "grad_norm": 0.38049066066741943,
      "learning_rate": 1.630342173972791e-05,
      "loss": 0.7255,
      "step": 1351
    },
    {
      "epoch": 0.3713775580277434,
      "grad_norm": 0.3698313236236572,
      "learning_rate": 1.6300673354404288e-05,
      "loss": 0.9982,
      "step": 1352
    },
    {
      "epoch": 0.3716522455706634,
      "grad_norm": 0.4297863245010376,
      "learning_rate": 1.629792496908067e-05,
      "loss": 0.7853,
      "step": 1353
    },
    {
      "epoch": 0.3719269331135833,
      "grad_norm": 0.3989447057247162,
      "learning_rate": 1.6295176583757046e-05,
      "loss": 1.0129,
      "step": 1354
    },
    {
      "epoch": 0.37220162065650325,
      "grad_norm": 0.3367876708507538,
      "learning_rate": 1.6292428198433423e-05,
      "loss": 0.9388,
      "step": 1355
    },
    {
      "epoch": 0.3724763081994232,
      "grad_norm": 0.4394771456718445,
      "learning_rate": 1.62896798131098e-05,
      "loss": 0.9329,
      "step": 1356
    },
    {
      "epoch": 0.3727509957423431,
      "grad_norm": 0.31999632716178894,
      "learning_rate": 1.6286931427786177e-05,
      "loss": 0.8208,
      "step": 1357
    },
    {
      "epoch": 0.37302568328526303,
      "grad_norm": 0.358635276556015,
      "learning_rate": 1.6284183042462554e-05,
      "loss": 0.9468,
      "step": 1358
    },
    {
      "epoch": 0.37330037082818296,
      "grad_norm": 0.3039748966693878,
      "learning_rate": 1.6281434657138934e-05,
      "loss": 1.0902,
      "step": 1359
    },
    {
      "epoch": 0.3735750583711029,
      "grad_norm": 0.30896836519241333,
      "learning_rate": 1.6278686271815308e-05,
      "loss": 0.6257,
      "step": 1360
    },
    {
      "epoch": 0.3738497459140228,
      "grad_norm": 0.43066996335983276,
      "learning_rate": 1.6275937886491688e-05,
      "loss": 0.7606,
      "step": 1361
    },
    {
      "epoch": 0.37412443345694274,
      "grad_norm": 0.363666832447052,
      "learning_rate": 1.6273189501168065e-05,
      "loss": 1.1204,
      "step": 1362
    },
    {
      "epoch": 0.37439912099986267,
      "grad_norm": 0.4154547154903412,
      "learning_rate": 1.6270441115844442e-05,
      "loss": 0.8329,
      "step": 1363
    },
    {
      "epoch": 0.3746738085427826,
      "grad_norm": 0.49551549553871155,
      "learning_rate": 1.6267692730520823e-05,
      "loss": 0.8868,
      "step": 1364
    },
    {
      "epoch": 0.3749484960857025,
      "grad_norm": 0.6715885400772095,
      "learning_rate": 1.6264944345197196e-05,
      "loss": 0.8975,
      "step": 1365
    },
    {
      "epoch": 0.37522318362862245,
      "grad_norm": 0.3380105197429657,
      "learning_rate": 1.6262195959873573e-05,
      "loss": 0.761,
      "step": 1366
    },
    {
      "epoch": 0.3754978711715424,
      "grad_norm": 0.2724728286266327,
      "learning_rate": 1.6259447574549954e-05,
      "loss": 0.5968,
      "step": 1367
    },
    {
      "epoch": 0.3757725587144623,
      "grad_norm": 0.39799436926841736,
      "learning_rate": 1.625669918922633e-05,
      "loss": 0.9148,
      "step": 1368
    },
    {
      "epoch": 0.37604724625738223,
      "grad_norm": 0.2853657603263855,
      "learning_rate": 1.625395080390271e-05,
      "loss": 0.7917,
      "step": 1369
    },
    {
      "epoch": 0.37632193380030216,
      "grad_norm": 0.38057857751846313,
      "learning_rate": 1.6251202418579085e-05,
      "loss": 0.7636,
      "step": 1370
    },
    {
      "epoch": 0.3765966213432221,
      "grad_norm": 0.42854732275009155,
      "learning_rate": 1.6248454033255462e-05,
      "loss": 0.7782,
      "step": 1371
    },
    {
      "epoch": 0.376871308886142,
      "grad_norm": 0.3923368752002716,
      "learning_rate": 1.6245705647931842e-05,
      "loss": 0.8029,
      "step": 1372
    },
    {
      "epoch": 0.37714599642906194,
      "grad_norm": 0.29061153531074524,
      "learning_rate": 1.624295726260822e-05,
      "loss": 0.7367,
      "step": 1373
    },
    {
      "epoch": 0.37742068397198186,
      "grad_norm": 0.3629797697067261,
      "learning_rate": 1.6240208877284596e-05,
      "loss": 0.7377,
      "step": 1374
    },
    {
      "epoch": 0.3776953715149018,
      "grad_norm": 0.5109005570411682,
      "learning_rate": 1.6237460491960973e-05,
      "loss": 0.9434,
      "step": 1375
    },
    {
      "epoch": 0.3779700590578217,
      "grad_norm": 0.313589483499527,
      "learning_rate": 1.623471210663735e-05,
      "loss": 0.8888,
      "step": 1376
    },
    {
      "epoch": 0.37824474660074164,
      "grad_norm": 0.390830397605896,
      "learning_rate": 1.623196372131373e-05,
      "loss": 0.9483,
      "step": 1377
    },
    {
      "epoch": 0.37851943414366157,
      "grad_norm": 0.38979995250701904,
      "learning_rate": 1.6229215335990108e-05,
      "loss": 0.9085,
      "step": 1378
    },
    {
      "epoch": 0.3787941216865815,
      "grad_norm": 0.37249529361724854,
      "learning_rate": 1.6226466950666485e-05,
      "loss": 0.6882,
      "step": 1379
    },
    {
      "epoch": 0.3790688092295014,
      "grad_norm": 0.45200881361961365,
      "learning_rate": 1.622371856534286e-05,
      "loss": 0.874,
      "step": 1380
    },
    {
      "epoch": 0.37934349677242135,
      "grad_norm": 0.41047778725624084,
      "learning_rate": 1.622097018001924e-05,
      "loss": 0.7136,
      "step": 1381
    },
    {
      "epoch": 0.3796181843153413,
      "grad_norm": 0.37579256296157837,
      "learning_rate": 1.6218221794695616e-05,
      "loss": 1.0301,
      "step": 1382
    },
    {
      "epoch": 0.3798928718582612,
      "grad_norm": 0.3657234311103821,
      "learning_rate": 1.6215473409371996e-05,
      "loss": 0.9839,
      "step": 1383
    },
    {
      "epoch": 0.38016755940118113,
      "grad_norm": 0.4065062403678894,
      "learning_rate": 1.6212725024048373e-05,
      "loss": 0.8379,
      "step": 1384
    },
    {
      "epoch": 0.38044224694410106,
      "grad_norm": 0.4076305329799652,
      "learning_rate": 1.620997663872475e-05,
      "loss": 0.8497,
      "step": 1385
    },
    {
      "epoch": 0.380716934487021,
      "grad_norm": 0.42131277918815613,
      "learning_rate": 1.6207228253401127e-05,
      "loss": 0.7248,
      "step": 1386
    },
    {
      "epoch": 0.38099162202994097,
      "grad_norm": 0.38463717699050903,
      "learning_rate": 1.6204479868077504e-05,
      "loss": 0.7938,
      "step": 1387
    },
    {
      "epoch": 0.3812663095728609,
      "grad_norm": 0.3648097813129425,
      "learning_rate": 1.6201731482753885e-05,
      "loss": 0.6377,
      "step": 1388
    },
    {
      "epoch": 0.3815409971157808,
      "grad_norm": 0.22242610156536102,
      "learning_rate": 1.619898309743026e-05,
      "loss": 0.4829,
      "step": 1389
    },
    {
      "epoch": 0.38181568465870075,
      "grad_norm": 0.47312501072883606,
      "learning_rate": 1.619623471210664e-05,
      "loss": 1.1357,
      "step": 1390
    },
    {
      "epoch": 0.3820903722016207,
      "grad_norm": 0.4447405934333801,
      "learning_rate": 1.6193486326783016e-05,
      "loss": 0.8718,
      "step": 1391
    },
    {
      "epoch": 0.3823650597445406,
      "grad_norm": 0.3940560519695282,
      "learning_rate": 1.6190737941459393e-05,
      "loss": 1.0757,
      "step": 1392
    },
    {
      "epoch": 0.38263974728746053,
      "grad_norm": 0.47080573439598083,
      "learning_rate": 1.6187989556135773e-05,
      "loss": 1.0002,
      "step": 1393
    },
    {
      "epoch": 0.38291443483038046,
      "grad_norm": 0.41301947832107544,
      "learning_rate": 1.618524117081215e-05,
      "loss": 0.7568,
      "step": 1394
    },
    {
      "epoch": 0.3831891223733004,
      "grad_norm": 0.38526085019111633,
      "learning_rate": 1.6182492785488527e-05,
      "loss": 0.8251,
      "step": 1395
    },
    {
      "epoch": 0.3834638099162203,
      "grad_norm": 0.3165360987186432,
      "learning_rate": 1.6179744400164904e-05,
      "loss": 1.1797,
      "step": 1396
    },
    {
      "epoch": 0.38373849745914024,
      "grad_norm": 0.43968138098716736,
      "learning_rate": 1.617699601484128e-05,
      "loss": 0.8924,
      "step": 1397
    },
    {
      "epoch": 0.38401318500206016,
      "grad_norm": 0.3817770183086395,
      "learning_rate": 1.6174247629517658e-05,
      "loss": 0.8818,
      "step": 1398
    },
    {
      "epoch": 0.3842878725449801,
      "grad_norm": 0.38150742650032043,
      "learning_rate": 1.617149924419404e-05,
      "loss": 0.8041,
      "step": 1399
    },
    {
      "epoch": 0.3845625600879,
      "grad_norm": 0.40426144003868103,
      "learning_rate": 1.6168750858870416e-05,
      "loss": 0.8191,
      "step": 1400
    },
    {
      "epoch": 0.38483724763081995,
      "grad_norm": 0.3231002390384674,
      "learning_rate": 1.6166002473546793e-05,
      "loss": 0.8997,
      "step": 1401
    },
    {
      "epoch": 0.38511193517373987,
      "grad_norm": 0.48006296157836914,
      "learning_rate": 1.616325408822317e-05,
      "loss": 0.7882,
      "step": 1402
    },
    {
      "epoch": 0.3853866227166598,
      "grad_norm": 0.3432682156562805,
      "learning_rate": 1.6160505702899547e-05,
      "loss": 0.7579,
      "step": 1403
    },
    {
      "epoch": 0.3856613102595797,
      "grad_norm": 0.4010101854801178,
      "learning_rate": 1.6157757317575927e-05,
      "loss": 0.7806,
      "step": 1404
    },
    {
      "epoch": 0.38593599780249965,
      "grad_norm": 0.41295433044433594,
      "learning_rate": 1.6155008932252304e-05,
      "loss": 1.0446,
      "step": 1405
    },
    {
      "epoch": 0.3862106853454196,
      "grad_norm": 0.45519042015075684,
      "learning_rate": 1.615226054692868e-05,
      "loss": 0.9599,
      "step": 1406
    },
    {
      "epoch": 0.3864853728883395,
      "grad_norm": 0.3579910397529602,
      "learning_rate": 1.6149512161605058e-05,
      "loss": 0.7529,
      "step": 1407
    },
    {
      "epoch": 0.38676006043125943,
      "grad_norm": 0.22718343138694763,
      "learning_rate": 1.6146763776281435e-05,
      "loss": 0.4983,
      "step": 1408
    },
    {
      "epoch": 0.38703474797417936,
      "grad_norm": 0.3595081567764282,
      "learning_rate": 1.6144015390957816e-05,
      "loss": 0.9436,
      "step": 1409
    },
    {
      "epoch": 0.3873094355170993,
      "grad_norm": 0.4646168053150177,
      "learning_rate": 1.6141267005634193e-05,
      "loss": 0.8846,
      "step": 1410
    },
    {
      "epoch": 0.3875841230600192,
      "grad_norm": 0.4007960855960846,
      "learning_rate": 1.613851862031057e-05,
      "loss": 0.8461,
      "step": 1411
    },
    {
      "epoch": 0.38785881060293914,
      "grad_norm": 0.32453206181526184,
      "learning_rate": 1.6135770234986947e-05,
      "loss": 0.9249,
      "step": 1412
    },
    {
      "epoch": 0.38813349814585907,
      "grad_norm": 0.27703478932380676,
      "learning_rate": 1.6133021849663324e-05,
      "loss": 0.6955,
      "step": 1413
    },
    {
      "epoch": 0.388408185688779,
      "grad_norm": 0.3926593065261841,
      "learning_rate": 1.61302734643397e-05,
      "loss": 0.6504,
      "step": 1414
    },
    {
      "epoch": 0.3886828732316989,
      "grad_norm": 0.373972088098526,
      "learning_rate": 1.612752507901608e-05,
      "loss": 0.7065,
      "step": 1415
    },
    {
      "epoch": 0.38895756077461885,
      "grad_norm": 0.30780166387557983,
      "learning_rate": 1.6124776693692458e-05,
      "loss": 0.609,
      "step": 1416
    },
    {
      "epoch": 0.3892322483175388,
      "grad_norm": 0.3795683979988098,
      "learning_rate": 1.6122028308368835e-05,
      "loss": 0.8017,
      "step": 1417
    },
    {
      "epoch": 0.3895069358604587,
      "grad_norm": 0.39143818616867065,
      "learning_rate": 1.6119279923045212e-05,
      "loss": 0.8826,
      "step": 1418
    },
    {
      "epoch": 0.38978162340337863,
      "grad_norm": 0.5490072965621948,
      "learning_rate": 1.611653153772159e-05,
      "loss": 0.9037,
      "step": 1419
    },
    {
      "epoch": 0.3900563109462986,
      "grad_norm": 0.4374969005584717,
      "learning_rate": 1.611378315239797e-05,
      "loss": 0.7938,
      "step": 1420
    },
    {
      "epoch": 0.39033099848921854,
      "grad_norm": 0.36038124561309814,
      "learning_rate": 1.6111034767074347e-05,
      "loss": 0.8098,
      "step": 1421
    },
    {
      "epoch": 0.39060568603213847,
      "grad_norm": 0.30597129464149475,
      "learning_rate": 1.610828638175072e-05,
      "loss": 0.673,
      "step": 1422
    },
    {
      "epoch": 0.3908803735750584,
      "grad_norm": 0.38285449147224426,
      "learning_rate": 1.61055379964271e-05,
      "loss": 0.9164,
      "step": 1423
    },
    {
      "epoch": 0.3911550611179783,
      "grad_norm": 0.3157974183559418,
      "learning_rate": 1.6102789611103478e-05,
      "loss": 0.7232,
      "step": 1424
    },
    {
      "epoch": 0.39142974866089825,
      "grad_norm": 0.44551950693130493,
      "learning_rate": 1.6100041225779858e-05,
      "loss": 0.7847,
      "step": 1425
    },
    {
      "epoch": 0.3917044362038182,
      "grad_norm": 0.31432637572288513,
      "learning_rate": 1.6097292840456232e-05,
      "loss": 0.733,
      "step": 1426
    },
    {
      "epoch": 0.3919791237467381,
      "grad_norm": 0.47513288259506226,
      "learning_rate": 1.609454445513261e-05,
      "loss": 0.6767,
      "step": 1427
    },
    {
      "epoch": 0.392253811289658,
      "grad_norm": 0.39034733176231384,
      "learning_rate": 1.609179606980899e-05,
      "loss": 0.6366,
      "step": 1428
    },
    {
      "epoch": 0.39252849883257795,
      "grad_norm": 0.31440258026123047,
      "learning_rate": 1.6089047684485366e-05,
      "loss": 0.8037,
      "step": 1429
    },
    {
      "epoch": 0.3928031863754979,
      "grad_norm": 0.2810285985469818,
      "learning_rate": 1.6086299299161743e-05,
      "loss": 0.5495,
      "step": 1430
    },
    {
      "epoch": 0.3930778739184178,
      "grad_norm": 0.35470107197761536,
      "learning_rate": 1.608355091383812e-05,
      "loss": 0.7799,
      "step": 1431
    },
    {
      "epoch": 0.39335256146133774,
      "grad_norm": 0.3825291097164154,
      "learning_rate": 1.6080802528514497e-05,
      "loss": 0.8627,
      "step": 1432
    },
    {
      "epoch": 0.39362724900425766,
      "grad_norm": 0.27972668409347534,
      "learning_rate": 1.6078054143190878e-05,
      "loss": 0.8092,
      "step": 1433
    },
    {
      "epoch": 0.3939019365471776,
      "grad_norm": 0.42712923884391785,
      "learning_rate": 1.6075305757867255e-05,
      "loss": 0.8713,
      "step": 1434
    },
    {
      "epoch": 0.3941766240900975,
      "grad_norm": 0.3852687478065491,
      "learning_rate": 1.6072557372543632e-05,
      "loss": 0.9324,
      "step": 1435
    },
    {
      "epoch": 0.39445131163301744,
      "grad_norm": 0.36403271555900574,
      "learning_rate": 1.606980898722001e-05,
      "loss": 0.6603,
      "step": 1436
    },
    {
      "epoch": 0.39472599917593737,
      "grad_norm": 0.33390533924102783,
      "learning_rate": 1.6067060601896386e-05,
      "loss": 0.9848,
      "step": 1437
    },
    {
      "epoch": 0.3950006867188573,
      "grad_norm": 0.3989975154399872,
      "learning_rate": 1.6064312216572763e-05,
      "loss": 0.7129,
      "step": 1438
    },
    {
      "epoch": 0.3952753742617772,
      "grad_norm": 0.3618088662624359,
      "learning_rate": 1.6061563831249143e-05,
      "loss": 0.6833,
      "step": 1439
    },
    {
      "epoch": 0.39555006180469715,
      "grad_norm": 0.2942188084125519,
      "learning_rate": 1.605881544592552e-05,
      "loss": 0.6323,
      "step": 1440
    },
    {
      "epoch": 0.3958247493476171,
      "grad_norm": 0.3917715549468994,
      "learning_rate": 1.6056067060601897e-05,
      "loss": 1.0549,
      "step": 1441
    },
    {
      "epoch": 0.396099436890537,
      "grad_norm": 0.35048046708106995,
      "learning_rate": 1.6053318675278274e-05,
      "loss": 0.7768,
      "step": 1442
    },
    {
      "epoch": 0.39637412443345693,
      "grad_norm": 0.3760538697242737,
      "learning_rate": 1.605057028995465e-05,
      "loss": 0.8691,
      "step": 1443
    },
    {
      "epoch": 0.39664881197637686,
      "grad_norm": 0.33146682381629944,
      "learning_rate": 1.604782190463103e-05,
      "loss": 0.9455,
      "step": 1444
    },
    {
      "epoch": 0.3969234995192968,
      "grad_norm": 0.3907114267349243,
      "learning_rate": 1.604507351930741e-05,
      "loss": 1.1253,
      "step": 1445
    },
    {
      "epoch": 0.3971981870622167,
      "grad_norm": 0.24911193549633026,
      "learning_rate": 1.6042325133983786e-05,
      "loss": 0.6942,
      "step": 1446
    },
    {
      "epoch": 0.39747287460513664,
      "grad_norm": 0.3680749237537384,
      "learning_rate": 1.6039576748660163e-05,
      "loss": 0.8324,
      "step": 1447
    },
    {
      "epoch": 0.39774756214805657,
      "grad_norm": 0.34824177622795105,
      "learning_rate": 1.603682836333654e-05,
      "loss": 0.7129,
      "step": 1448
    },
    {
      "epoch": 0.3980222496909765,
      "grad_norm": 0.3044115900993347,
      "learning_rate": 1.603407997801292e-05,
      "loss": 0.7786,
      "step": 1449
    },
    {
      "epoch": 0.3982969372338964,
      "grad_norm": 0.5445106625556946,
      "learning_rate": 1.6031331592689297e-05,
      "loss": 0.7035,
      "step": 1450
    },
    {
      "epoch": 0.39857162477681635,
      "grad_norm": 0.2739531695842743,
      "learning_rate": 1.6028583207365674e-05,
      "loss": 0.6106,
      "step": 1451
    },
    {
      "epoch": 0.3988463123197363,
      "grad_norm": 0.4184846878051758,
      "learning_rate": 1.602583482204205e-05,
      "loss": 0.7377,
      "step": 1452
    },
    {
      "epoch": 0.39912099986265626,
      "grad_norm": 0.5104706287384033,
      "learning_rate": 1.6023086436718428e-05,
      "loss": 0.6833,
      "step": 1453
    },
    {
      "epoch": 0.3993956874055762,
      "grad_norm": 0.4036400020122528,
      "learning_rate": 1.6020338051394805e-05,
      "loss": 0.9657,
      "step": 1454
    },
    {
      "epoch": 0.3996703749484961,
      "grad_norm": 0.4057066738605499,
      "learning_rate": 1.6017589666071186e-05,
      "loss": 0.6721,
      "step": 1455
    },
    {
      "epoch": 0.39994506249141604,
      "grad_norm": 0.5067551136016846,
      "learning_rate": 1.6014841280747563e-05,
      "loss": 0.9514,
      "step": 1456
    },
    {
      "epoch": 0.40021975003433596,
      "grad_norm": 0.3243044316768646,
      "learning_rate": 1.601209289542394e-05,
      "loss": 1.0603,
      "step": 1457
    },
    {
      "epoch": 0.4004944375772559,
      "grad_norm": 0.3861035406589508,
      "learning_rate": 1.6009344510100317e-05,
      "loss": 0.851,
      "step": 1458
    },
    {
      "epoch": 0.4007691251201758,
      "grad_norm": 0.47284063696861267,
      "learning_rate": 1.6006596124776694e-05,
      "loss": 0.8298,
      "step": 1459
    },
    {
      "epoch": 0.40104381266309574,
      "grad_norm": 0.3901832103729248,
      "learning_rate": 1.6003847739453074e-05,
      "loss": 0.9346,
      "step": 1460
    },
    {
      "epoch": 0.40131850020601567,
      "grad_norm": 0.3175239562988281,
      "learning_rate": 1.600109935412945e-05,
      "loss": 0.8249,
      "step": 1461
    },
    {
      "epoch": 0.4015931877489356,
      "grad_norm": 0.3047941327095032,
      "learning_rate": 1.5998350968805828e-05,
      "loss": 0.6804,
      "step": 1462
    },
    {
      "epoch": 0.4018678752918555,
      "grad_norm": 0.4024829566478729,
      "learning_rate": 1.5995602583482205e-05,
      "loss": 0.8019,
      "step": 1463
    },
    {
      "epoch": 0.40214256283477545,
      "grad_norm": 0.31075918674468994,
      "learning_rate": 1.5992854198158582e-05,
      "loss": 0.6306,
      "step": 1464
    },
    {
      "epoch": 0.4024172503776954,
      "grad_norm": 0.3141782283782959,
      "learning_rate": 1.5990105812834963e-05,
      "loss": 1.1197,
      "step": 1465
    },
    {
      "epoch": 0.4026919379206153,
      "grad_norm": 0.3864113390445709,
      "learning_rate": 1.598735742751134e-05,
      "loss": 0.8029,
      "step": 1466
    },
    {
      "epoch": 0.40296662546353523,
      "grad_norm": 0.48281237483024597,
      "learning_rate": 1.5984609042187717e-05,
      "loss": 0.8378,
      "step": 1467
    },
    {
      "epoch": 0.40324131300645516,
      "grad_norm": 0.45041021704673767,
      "learning_rate": 1.5981860656864094e-05,
      "loss": 0.6176,
      "step": 1468
    },
    {
      "epoch": 0.4035160005493751,
      "grad_norm": 0.41917726397514343,
      "learning_rate": 1.597911227154047e-05,
      "loss": 0.8778,
      "step": 1469
    },
    {
      "epoch": 0.403790688092295,
      "grad_norm": 0.2759922444820404,
      "learning_rate": 1.5976363886216848e-05,
      "loss": 0.5671,
      "step": 1470
    },
    {
      "epoch": 0.40406537563521494,
      "grad_norm": 0.5164101719856262,
      "learning_rate": 1.5973615500893228e-05,
      "loss": 0.8241,
      "step": 1471
    },
    {
      "epoch": 0.40434006317813487,
      "grad_norm": 0.45930951833724976,
      "learning_rate": 1.5970867115569605e-05,
      "loss": 0.9224,
      "step": 1472
    },
    {
      "epoch": 0.4046147507210548,
      "grad_norm": 0.31792503595352173,
      "learning_rate": 1.5968118730245982e-05,
      "loss": 0.5944,
      "step": 1473
    },
    {
      "epoch": 0.4048894382639747,
      "grad_norm": 0.36870500445365906,
      "learning_rate": 1.596537034492236e-05,
      "loss": 0.8427,
      "step": 1474
    },
    {
      "epoch": 0.40516412580689465,
      "grad_norm": 0.40882396697998047,
      "learning_rate": 1.5962621959598736e-05,
      "loss": 0.8965,
      "step": 1475
    },
    {
      "epoch": 0.4054388133498146,
      "grad_norm": 0.31890812516212463,
      "learning_rate": 1.5959873574275117e-05,
      "loss": 0.7216,
      "step": 1476
    },
    {
      "epoch": 0.4057135008927345,
      "grad_norm": 0.46436506509780884,
      "learning_rate": 1.5957125188951494e-05,
      "loss": 1.0114,
      "step": 1477
    },
    {
      "epoch": 0.40598818843565443,
      "grad_norm": 0.2858348786830902,
      "learning_rate": 1.595437680362787e-05,
      "loss": 0.7811,
      "step": 1478
    },
    {
      "epoch": 0.40626287597857436,
      "grad_norm": 0.41206711530685425,
      "learning_rate": 1.5951628418304248e-05,
      "loss": 0.8308,
      "step": 1479
    },
    {
      "epoch": 0.4065375635214943,
      "grad_norm": 0.38554129004478455,
      "learning_rate": 1.5948880032980625e-05,
      "loss": 0.9976,
      "step": 1480
    },
    {
      "epoch": 0.4068122510644142,
      "grad_norm": 0.33337754011154175,
      "learning_rate": 1.5946131647657005e-05,
      "loss": 0.7483,
      "step": 1481
    },
    {
      "epoch": 0.40708693860733414,
      "grad_norm": 0.39667993783950806,
      "learning_rate": 1.5943383262333382e-05,
      "loss": 0.7911,
      "step": 1482
    },
    {
      "epoch": 0.40736162615025406,
      "grad_norm": 0.3029700219631195,
      "learning_rate": 1.5940634877009756e-05,
      "loss": 0.7198,
      "step": 1483
    },
    {
      "epoch": 0.407636313693174,
      "grad_norm": 0.4540368616580963,
      "learning_rate": 1.5937886491686136e-05,
      "loss": 0.9696,
      "step": 1484
    },
    {
      "epoch": 0.4079110012360939,
      "grad_norm": 0.27390041947364807,
      "learning_rate": 1.5935138106362513e-05,
      "loss": 0.6884,
      "step": 1485
    },
    {
      "epoch": 0.40818568877901384,
      "grad_norm": 0.3416951596736908,
      "learning_rate": 1.593238972103889e-05,
      "loss": 0.942,
      "step": 1486
    },
    {
      "epoch": 0.4084603763219338,
      "grad_norm": 0.30048269033432007,
      "learning_rate": 1.592964133571527e-05,
      "loss": 0.8639,
      "step": 1487
    },
    {
      "epoch": 0.40873506386485375,
      "grad_norm": 0.4542458653450012,
      "learning_rate": 1.5926892950391644e-05,
      "loss": 0.8566,
      "step": 1488
    },
    {
      "epoch": 0.4090097514077737,
      "grad_norm": 0.3183557689189911,
      "learning_rate": 1.5924144565068025e-05,
      "loss": 0.8643,
      "step": 1489
    },
    {
      "epoch": 0.4092844389506936,
      "grad_norm": 0.3764091432094574,
      "learning_rate": 1.5921396179744402e-05,
      "loss": 0.6325,
      "step": 1490
    },
    {
      "epoch": 0.40955912649361353,
      "grad_norm": 0.39241743087768555,
      "learning_rate": 1.591864779442078e-05,
      "loss": 1.0403,
      "step": 1491
    },
    {
      "epoch": 0.40983381403653346,
      "grad_norm": 0.4195300340652466,
      "learning_rate": 1.591589940909716e-05,
      "loss": 0.5847,
      "step": 1492
    },
    {
      "epoch": 0.4101085015794534,
      "grad_norm": 0.3713049292564392,
      "learning_rate": 1.5913151023773533e-05,
      "loss": 0.897,
      "step": 1493
    },
    {
      "epoch": 0.4103831891223733,
      "grad_norm": 0.4157247245311737,
      "learning_rate": 1.591040263844991e-05,
      "loss": 0.808,
      "step": 1494
    },
    {
      "epoch": 0.41065787666529324,
      "grad_norm": 0.35031718015670776,
      "learning_rate": 1.590765425312629e-05,
      "loss": 0.7896,
      "step": 1495
    },
    {
      "epoch": 0.41093256420821317,
      "grad_norm": 0.5223699808120728,
      "learning_rate": 1.5904905867802667e-05,
      "loss": 0.7377,
      "step": 1496
    },
    {
      "epoch": 0.4112072517511331,
      "grad_norm": 0.39447689056396484,
      "learning_rate": 1.5902157482479044e-05,
      "loss": 0.8417,
      "step": 1497
    },
    {
      "epoch": 0.411481939294053,
      "grad_norm": 0.318479984998703,
      "learning_rate": 1.589940909715542e-05,
      "loss": 0.881,
      "step": 1498
    },
    {
      "epoch": 0.41175662683697295,
      "grad_norm": 0.3348401188850403,
      "learning_rate": 1.5896660711831798e-05,
      "loss": 0.8753,
      "step": 1499
    },
    {
      "epoch": 0.4120313143798929,
      "grad_norm": 0.39277753233909607,
      "learning_rate": 1.589391232650818e-05,
      "loss": 1.1643,
      "step": 1500
    },
    {
      "epoch": 0.4123060019228128,
      "grad_norm": 0.3406621813774109,
      "learning_rate": 1.5891163941184556e-05,
      "loss": 0.8122,
      "step": 1501
    },
    {
      "epoch": 0.41258068946573273,
      "grad_norm": 0.3583405017852783,
      "learning_rate": 1.5888415555860933e-05,
      "loss": 1.0944,
      "step": 1502
    },
    {
      "epoch": 0.41285537700865266,
      "grad_norm": 0.31168243288993835,
      "learning_rate": 1.588566717053731e-05,
      "loss": 0.686,
      "step": 1503
    },
    {
      "epoch": 0.4131300645515726,
      "grad_norm": 0.4186003506183624,
      "learning_rate": 1.5882918785213687e-05,
      "loss": 0.9378,
      "step": 1504
    },
    {
      "epoch": 0.4134047520944925,
      "grad_norm": 0.4421500265598297,
      "learning_rate": 1.5880170399890067e-05,
      "loss": 0.7465,
      "step": 1505
    },
    {
      "epoch": 0.41367943963741244,
      "grad_norm": 0.3686763346195221,
      "learning_rate": 1.5877422014566444e-05,
      "loss": 0.7954,
      "step": 1506
    },
    {
      "epoch": 0.41395412718033237,
      "grad_norm": 0.3747735619544983,
      "learning_rate": 1.587467362924282e-05,
      "loss": 0.5184,
      "step": 1507
    },
    {
      "epoch": 0.4142288147232523,
      "grad_norm": 0.4699954092502594,
      "learning_rate": 1.5871925243919198e-05,
      "loss": 0.7591,
      "step": 1508
    },
    {
      "epoch": 0.4145035022661722,
      "grad_norm": 0.4746759533882141,
      "learning_rate": 1.5869176858595575e-05,
      "loss": 0.9746,
      "step": 1509
    },
    {
      "epoch": 0.41477818980909215,
      "grad_norm": 0.43386271595954895,
      "learning_rate": 1.5866428473271952e-05,
      "loss": 0.8609,
      "step": 1510
    },
    {
      "epoch": 0.4150528773520121,
      "grad_norm": 0.359747052192688,
      "learning_rate": 1.5863680087948333e-05,
      "loss": 0.6231,
      "step": 1511
    },
    {
      "epoch": 0.415327564894932,
      "grad_norm": 0.38725027441978455,
      "learning_rate": 1.586093170262471e-05,
      "loss": 0.6735,
      "step": 1512
    },
    {
      "epoch": 0.4156022524378519,
      "grad_norm": 0.3824041485786438,
      "learning_rate": 1.5858183317301087e-05,
      "loss": 0.8616,
      "step": 1513
    },
    {
      "epoch": 0.41587693998077185,
      "grad_norm": 0.4277836084365845,
      "learning_rate": 1.5855434931977464e-05,
      "loss": 0.8326,
      "step": 1514
    },
    {
      "epoch": 0.4161516275236918,
      "grad_norm": 0.3438524007797241,
      "learning_rate": 1.585268654665384e-05,
      "loss": 0.5576,
      "step": 1515
    },
    {
      "epoch": 0.4164263150666117,
      "grad_norm": 0.3693961501121521,
      "learning_rate": 1.584993816133022e-05,
      "loss": 0.8275,
      "step": 1516
    },
    {
      "epoch": 0.41670100260953163,
      "grad_norm": 0.2950611710548401,
      "learning_rate": 1.5847189776006598e-05,
      "loss": 0.6247,
      "step": 1517
    },
    {
      "epoch": 0.41697569015245156,
      "grad_norm": 0.4355158805847168,
      "learning_rate": 1.5844441390682975e-05,
      "loss": 0.8289,
      "step": 1518
    },
    {
      "epoch": 0.4172503776953715,
      "grad_norm": 0.27834880352020264,
      "learning_rate": 1.5841693005359352e-05,
      "loss": 0.6327,
      "step": 1519
    },
    {
      "epoch": 0.41752506523829147,
      "grad_norm": 0.3425748944282532,
      "learning_rate": 1.583894462003573e-05,
      "loss": 0.5905,
      "step": 1520
    },
    {
      "epoch": 0.4177997527812114,
      "grad_norm": 0.2854830026626587,
      "learning_rate": 1.583619623471211e-05,
      "loss": 0.6645,
      "step": 1521
    },
    {
      "epoch": 0.4180744403241313,
      "grad_norm": 0.37779802083969116,
      "learning_rate": 1.5833447849388487e-05,
      "loss": 0.8837,
      "step": 1522
    },
    {
      "epoch": 0.41834912786705125,
      "grad_norm": 0.4066874086856842,
      "learning_rate": 1.5830699464064864e-05,
      "loss": 0.7715,
      "step": 1523
    },
    {
      "epoch": 0.4186238154099712,
      "grad_norm": 0.45464426279067993,
      "learning_rate": 1.582795107874124e-05,
      "loss": 0.8585,
      "step": 1524
    },
    {
      "epoch": 0.4188985029528911,
      "grad_norm": 0.3042803406715393,
      "learning_rate": 1.5825202693417618e-05,
      "loss": 0.8502,
      "step": 1525
    },
    {
      "epoch": 0.41917319049581103,
      "grad_norm": 0.43847787380218506,
      "learning_rate": 1.5822454308093995e-05,
      "loss": 0.7405,
      "step": 1526
    },
    {
      "epoch": 0.41944787803873096,
      "grad_norm": 0.39922210574150085,
      "learning_rate": 1.5819705922770375e-05,
      "loss": 0.6369,
      "step": 1527
    },
    {
      "epoch": 0.4197225655816509,
      "grad_norm": 0.385733425617218,
      "learning_rate": 1.5816957537446752e-05,
      "loss": 0.9685,
      "step": 1528
    },
    {
      "epoch": 0.4199972531245708,
      "grad_norm": 0.38224685192108154,
      "learning_rate": 1.581420915212313e-05,
      "loss": 0.9825,
      "step": 1529
    },
    {
      "epoch": 0.42027194066749074,
      "grad_norm": 0.3849998712539673,
      "learning_rate": 1.5811460766799506e-05,
      "loss": 0.8006,
      "step": 1530
    },
    {
      "epoch": 0.42054662821041067,
      "grad_norm": 0.3070632517337799,
      "learning_rate": 1.5808712381475883e-05,
      "loss": 0.8961,
      "step": 1531
    },
    {
      "epoch": 0.4208213157533306,
      "grad_norm": 0.3815346360206604,
      "learning_rate": 1.5805963996152264e-05,
      "loss": 0.7522,
      "step": 1532
    },
    {
      "epoch": 0.4210960032962505,
      "grad_norm": 0.35632675886154175,
      "learning_rate": 1.580321561082864e-05,
      "loss": 0.7811,
      "step": 1533
    },
    {
      "epoch": 0.42137069083917045,
      "grad_norm": 0.45725083351135254,
      "learning_rate": 1.5800467225505018e-05,
      "loss": 0.8321,
      "step": 1534
    },
    {
      "epoch": 0.4216453783820904,
      "grad_norm": 0.2578490972518921,
      "learning_rate": 1.5797718840181395e-05,
      "loss": 0.5428,
      "step": 1535
    },
    {
      "epoch": 0.4219200659250103,
      "grad_norm": 0.3944675028324127,
      "learning_rate": 1.5794970454857772e-05,
      "loss": 0.8436,
      "step": 1536
    },
    {
      "epoch": 0.42219475346793023,
      "grad_norm": 0.3708694279193878,
      "learning_rate": 1.5792222069534152e-05,
      "loss": 0.8362,
      "step": 1537
    },
    {
      "epoch": 0.42246944101085016,
      "grad_norm": 0.39322561025619507,
      "learning_rate": 1.578947368421053e-05,
      "loss": 0.9357,
      "step": 1538
    },
    {
      "epoch": 0.4227441285537701,
      "grad_norm": 0.3231641352176666,
      "learning_rate": 1.5786725298886906e-05,
      "loss": 0.8772,
      "step": 1539
    },
    {
      "epoch": 0.42301881609669,
      "grad_norm": 0.44495677947998047,
      "learning_rate": 1.5783976913563283e-05,
      "loss": 0.976,
      "step": 1540
    },
    {
      "epoch": 0.42329350363960994,
      "grad_norm": 0.29765865206718445,
      "learning_rate": 1.578122852823966e-05,
      "loss": 0.8619,
      "step": 1541
    },
    {
      "epoch": 0.42356819118252986,
      "grad_norm": 0.44307392835617065,
      "learning_rate": 1.5778480142916037e-05,
      "loss": 1.0591,
      "step": 1542
    },
    {
      "epoch": 0.4238428787254498,
      "grad_norm": 0.47513964772224426,
      "learning_rate": 1.5775731757592418e-05,
      "loss": 0.8713,
      "step": 1543
    },
    {
      "epoch": 0.4241175662683697,
      "grad_norm": 0.4401443600654602,
      "learning_rate": 1.5772983372268795e-05,
      "loss": 0.9367,
      "step": 1544
    },
    {
      "epoch": 0.42439225381128964,
      "grad_norm": 0.3519763648509979,
      "learning_rate": 1.5770234986945172e-05,
      "loss": 0.8432,
      "step": 1545
    },
    {
      "epoch": 0.42466694135420957,
      "grad_norm": 0.3580692410469055,
      "learning_rate": 1.576748660162155e-05,
      "loss": 0.8006,
      "step": 1546
    },
    {
      "epoch": 0.4249416288971295,
      "grad_norm": 0.4319553077220917,
      "learning_rate": 1.5764738216297926e-05,
      "loss": 1.083,
      "step": 1547
    },
    {
      "epoch": 0.4252163164400494,
      "grad_norm": 0.3849879503250122,
      "learning_rate": 1.5761989830974306e-05,
      "loss": 0.6127,
      "step": 1548
    },
    {
      "epoch": 0.42549100398296935,
      "grad_norm": 0.2815719246864319,
      "learning_rate": 1.575924144565068e-05,
      "loss": 0.7368,
      "step": 1549
    },
    {
      "epoch": 0.4257656915258893,
      "grad_norm": 0.3633367717266083,
      "learning_rate": 1.5756493060327057e-05,
      "loss": 0.7965,
      "step": 1550
    },
    {
      "epoch": 0.4260403790688092,
      "grad_norm": 0.6244245171546936,
      "learning_rate": 1.5753744675003437e-05,
      "loss": 0.8736,
      "step": 1551
    },
    {
      "epoch": 0.42631506661172913,
      "grad_norm": 0.3201342821121216,
      "learning_rate": 1.5750996289679814e-05,
      "loss": 0.7129,
      "step": 1552
    },
    {
      "epoch": 0.42658975415464906,
      "grad_norm": 0.3190351128578186,
      "learning_rate": 1.5748247904356195e-05,
      "loss": 0.7628,
      "step": 1553
    },
    {
      "epoch": 0.42686444169756904,
      "grad_norm": 0.38973134756088257,
      "learning_rate": 1.574549951903257e-05,
      "loss": 0.9091,
      "step": 1554
    },
    {
      "epoch": 0.42713912924048897,
      "grad_norm": 0.6139158606529236,
      "learning_rate": 1.5742751133708945e-05,
      "loss": 0.8921,
      "step": 1555
    },
    {
      "epoch": 0.4274138167834089,
      "grad_norm": 0.46247732639312744,
      "learning_rate": 1.5740002748385326e-05,
      "loss": 0.7207,
      "step": 1556
    },
    {
      "epoch": 0.4276885043263288,
      "grad_norm": 0.36353495717048645,
      "learning_rate": 1.5737254363061703e-05,
      "loss": 0.6747,
      "step": 1557
    },
    {
      "epoch": 0.42796319186924875,
      "grad_norm": 0.5101955533027649,
      "learning_rate": 1.573450597773808e-05,
      "loss": 0.8391,
      "step": 1558
    },
    {
      "epoch": 0.4282378794121687,
      "grad_norm": 0.34568163752555847,
      "learning_rate": 1.5731757592414457e-05,
      "loss": 0.9088,
      "step": 1559
    },
    {
      "epoch": 0.4285125669550886,
      "grad_norm": 0.5203610062599182,
      "learning_rate": 1.5729009207090834e-05,
      "loss": 0.8532,
      "step": 1560
    },
    {
      "epoch": 0.42878725449800853,
      "grad_norm": 0.3270670771598816,
      "learning_rate": 1.5726260821767214e-05,
      "loss": 0.6883,
      "step": 1561
    },
    {
      "epoch": 0.42906194204092846,
      "grad_norm": 0.36248552799224854,
      "learning_rate": 1.572351243644359e-05,
      "loss": 0.8649,
      "step": 1562
    },
    {
      "epoch": 0.4293366295838484,
      "grad_norm": 0.3505651354789734,
      "learning_rate": 1.5720764051119968e-05,
      "loss": 0.9242,
      "step": 1563
    },
    {
      "epoch": 0.4296113171267683,
      "grad_norm": 0.4237819015979767,
      "learning_rate": 1.5718015665796345e-05,
      "loss": 0.9081,
      "step": 1564
    },
    {
      "epoch": 0.42988600466968824,
      "grad_norm": 0.43519338965415955,
      "learning_rate": 1.5715267280472722e-05,
      "loss": 0.6878,
      "step": 1565
    },
    {
      "epoch": 0.43016069221260816,
      "grad_norm": 0.2525920569896698,
      "learning_rate": 1.57125188951491e-05,
      "loss": 0.5707,
      "step": 1566
    },
    {
      "epoch": 0.4304353797555281,
      "grad_norm": 0.4210473597049713,
      "learning_rate": 1.570977050982548e-05,
      "loss": 1.0023,
      "step": 1567
    },
    {
      "epoch": 0.430710067298448,
      "grad_norm": 0.30767834186553955,
      "learning_rate": 1.5707022124501857e-05,
      "loss": 0.9162,
      "step": 1568
    },
    {
      "epoch": 0.43098475484136795,
      "grad_norm": 0.34758248925209045,
      "learning_rate": 1.5704273739178234e-05,
      "loss": 0.7589,
      "step": 1569
    },
    {
      "epoch": 0.43125944238428787,
      "grad_norm": 0.3588829040527344,
      "learning_rate": 1.570152535385461e-05,
      "loss": 0.8835,
      "step": 1570
    },
    {
      "epoch": 0.4315341299272078,
      "grad_norm": 0.3286401927471161,
      "learning_rate": 1.5698776968530988e-05,
      "loss": 0.6348,
      "step": 1571
    },
    {
      "epoch": 0.4318088174701277,
      "grad_norm": 0.2834451496601105,
      "learning_rate": 1.5696028583207368e-05,
      "loss": 1.0397,
      "step": 1572
    },
    {
      "epoch": 0.43208350501304765,
      "grad_norm": 0.3406881093978882,
      "learning_rate": 1.5693280197883745e-05,
      "loss": 0.6838,
      "step": 1573
    },
    {
      "epoch": 0.4323581925559676,
      "grad_norm": 0.3950667083263397,
      "learning_rate": 1.5690531812560122e-05,
      "loss": 0.7979,
      "step": 1574
    },
    {
      "epoch": 0.4326328800988875,
      "grad_norm": 0.3793924152851105,
      "learning_rate": 1.56877834272365e-05,
      "loss": 1.0681,
      "step": 1575
    },
    {
      "epoch": 0.43290756764180743,
      "grad_norm": 0.45490947365760803,
      "learning_rate": 1.5685035041912876e-05,
      "loss": 0.7183,
      "step": 1576
    },
    {
      "epoch": 0.43318225518472736,
      "grad_norm": 0.3349732756614685,
      "learning_rate": 1.5682286656589257e-05,
      "loss": 1.0707,
      "step": 1577
    },
    {
      "epoch": 0.4334569427276473,
      "grad_norm": 0.3239552080631256,
      "learning_rate": 1.5679538271265634e-05,
      "loss": 0.7646,
      "step": 1578
    },
    {
      "epoch": 0.4337316302705672,
      "grad_norm": 0.5238783359527588,
      "learning_rate": 1.567678988594201e-05,
      "loss": 1.0682,
      "step": 1579
    },
    {
      "epoch": 0.43400631781348714,
      "grad_norm": 0.37779703736305237,
      "learning_rate": 1.5674041500618388e-05,
      "loss": 0.8586,
      "step": 1580
    },
    {
      "epoch": 0.43428100535640707,
      "grad_norm": 0.3776254951953888,
      "learning_rate": 1.5671293115294765e-05,
      "loss": 0.6742,
      "step": 1581
    },
    {
      "epoch": 0.434555692899327,
      "grad_norm": 0.3708952069282532,
      "learning_rate": 1.5668544729971142e-05,
      "loss": 0.9175,
      "step": 1582
    },
    {
      "epoch": 0.4348303804422469,
      "grad_norm": 0.32013803720474243,
      "learning_rate": 1.5665796344647522e-05,
      "loss": 0.8773,
      "step": 1583
    },
    {
      "epoch": 0.43510506798516685,
      "grad_norm": 0.3355189561843872,
      "learning_rate": 1.56630479593239e-05,
      "loss": 0.8745,
      "step": 1584
    },
    {
      "epoch": 0.4353797555280868,
      "grad_norm": 0.3262452483177185,
      "learning_rate": 1.5660299574000276e-05,
      "loss": 0.801,
      "step": 1585
    },
    {
      "epoch": 0.4356544430710067,
      "grad_norm": 0.3358146846294403,
      "learning_rate": 1.5657551188676653e-05,
      "loss": 0.7295,
      "step": 1586
    },
    {
      "epoch": 0.4359291306139267,
      "grad_norm": 0.4679683446884155,
      "learning_rate": 1.565480280335303e-05,
      "loss": 0.9858,
      "step": 1587
    },
    {
      "epoch": 0.4362038181568466,
      "grad_norm": 0.4230237305164337,
      "learning_rate": 1.565205441802941e-05,
      "loss": 0.8178,
      "step": 1588
    },
    {
      "epoch": 0.43647850569976654,
      "grad_norm": 0.3304666578769684,
      "learning_rate": 1.5649306032705788e-05,
      "loss": 0.7874,
      "step": 1589
    },
    {
      "epoch": 0.43675319324268647,
      "grad_norm": 0.27545127272605896,
      "learning_rate": 1.5646557647382165e-05,
      "loss": 0.7472,
      "step": 1590
    },
    {
      "epoch": 0.4370278807856064,
      "grad_norm": 0.380073606967926,
      "learning_rate": 1.5643809262058542e-05,
      "loss": 1.0226,
      "step": 1591
    },
    {
      "epoch": 0.4373025683285263,
      "grad_norm": 0.437730073928833,
      "learning_rate": 1.564106087673492e-05,
      "loss": 0.6927,
      "step": 1592
    },
    {
      "epoch": 0.43757725587144625,
      "grad_norm": 0.3241958022117615,
      "learning_rate": 1.5638312491411296e-05,
      "loss": 0.9546,
      "step": 1593
    },
    {
      "epoch": 0.4378519434143662,
      "grad_norm": 0.29090645909309387,
      "learning_rate": 1.5635564106087676e-05,
      "loss": 0.7289,
      "step": 1594
    },
    {
      "epoch": 0.4381266309572861,
      "grad_norm": 0.4227448105812073,
      "learning_rate": 1.5632815720764053e-05,
      "loss": 0.9625,
      "step": 1595
    },
    {
      "epoch": 0.438401318500206,
      "grad_norm": 0.25714725255966187,
      "learning_rate": 1.563006733544043e-05,
      "loss": 0.795,
      "step": 1596
    },
    {
      "epoch": 0.43867600604312595,
      "grad_norm": 0.327191561460495,
      "learning_rate": 1.5627318950116807e-05,
      "loss": 0.719,
      "step": 1597
    },
    {
      "epoch": 0.4389506935860459,
      "grad_norm": 0.3597950041294098,
      "learning_rate": 1.5624570564793184e-05,
      "loss": 0.6882,
      "step": 1598
    },
    {
      "epoch": 0.4392253811289658,
      "grad_norm": 0.3189670741558075,
      "learning_rate": 1.5621822179469565e-05,
      "loss": 0.7886,
      "step": 1599
    },
    {
      "epoch": 0.43950006867188574,
      "grad_norm": 0.4429094195365906,
      "learning_rate": 1.5619073794145942e-05,
      "loss": 0.8254,
      "step": 1600
    },
    {
      "epoch": 0.43977475621480566,
      "grad_norm": 0.4380422830581665,
      "learning_rate": 1.5616325408822315e-05,
      "loss": 0.6018,
      "step": 1601
    },
    {
      "epoch": 0.4400494437577256,
      "grad_norm": 0.4166739284992218,
      "learning_rate": 1.5613577023498696e-05,
      "loss": 0.9591,
      "step": 1602
    },
    {
      "epoch": 0.4403241313006455,
      "grad_norm": 0.3341524302959442,
      "learning_rate": 1.5610828638175073e-05,
      "loss": 1.0344,
      "step": 1603
    },
    {
      "epoch": 0.44059881884356544,
      "grad_norm": 0.2991352081298828,
      "learning_rate": 1.5608080252851453e-05,
      "loss": 0.869,
      "step": 1604
    },
    {
      "epoch": 0.44087350638648537,
      "grad_norm": 0.49110931158065796,
      "learning_rate": 1.560533186752783e-05,
      "loss": 0.7355,
      "step": 1605
    },
    {
      "epoch": 0.4411481939294053,
      "grad_norm": 0.43560290336608887,
      "learning_rate": 1.5602583482204204e-05,
      "loss": 0.9549,
      "step": 1606
    },
    {
      "epoch": 0.4414228814723252,
      "grad_norm": 0.2542040944099426,
      "learning_rate": 1.5599835096880584e-05,
      "loss": 0.6254,
      "step": 1607
    },
    {
      "epoch": 0.44169756901524515,
      "grad_norm": 0.38314712047576904,
      "learning_rate": 1.559708671155696e-05,
      "loss": 0.766,
      "step": 1608
    },
    {
      "epoch": 0.4419722565581651,
      "grad_norm": 0.35365596413612366,
      "learning_rate": 1.559433832623334e-05,
      "loss": 0.745,
      "step": 1609
    },
    {
      "epoch": 0.442246944101085,
      "grad_norm": 0.34653112292289734,
      "learning_rate": 1.559158994090972e-05,
      "loss": 0.7639,
      "step": 1610
    },
    {
      "epoch": 0.44252163164400493,
      "grad_norm": 0.33676931262016296,
      "learning_rate": 1.5588841555586092e-05,
      "loss": 0.7714,
      "step": 1611
    },
    {
      "epoch": 0.44279631918692486,
      "grad_norm": 0.36860278248786926,
      "learning_rate": 1.5586093170262473e-05,
      "loss": 0.5612,
      "step": 1612
    },
    {
      "epoch": 0.4430710067298448,
      "grad_norm": 0.36205703020095825,
      "learning_rate": 1.558334478493885e-05,
      "loss": 1.0193,
      "step": 1613
    },
    {
      "epoch": 0.4433456942727647,
      "grad_norm": 0.3631129264831543,
      "learning_rate": 1.5580596399615227e-05,
      "loss": 0.7101,
      "step": 1614
    },
    {
      "epoch": 0.44362038181568464,
      "grad_norm": 0.46845105290412903,
      "learning_rate": 1.5577848014291604e-05,
      "loss": 0.8629,
      "step": 1615
    },
    {
      "epoch": 0.44389506935860457,
      "grad_norm": 0.4617152214050293,
      "learning_rate": 1.557509962896798e-05,
      "loss": 0.8505,
      "step": 1616
    },
    {
      "epoch": 0.4441697569015245,
      "grad_norm": 0.3831556737422943,
      "learning_rate": 1.5572351243644358e-05,
      "loss": 0.84,
      "step": 1617
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.40162786841392517,
      "learning_rate": 1.556960285832074e-05,
      "loss": 0.9123,
      "step": 1618
    },
    {
      "epoch": 0.44471913198736435,
      "grad_norm": 0.39805519580841064,
      "learning_rate": 1.5566854472997115e-05,
      "loss": 0.9263,
      "step": 1619
    },
    {
      "epoch": 0.44499381953028433,
      "grad_norm": 0.3763875365257263,
      "learning_rate": 1.5564106087673492e-05,
      "loss": 0.8217,
      "step": 1620
    },
    {
      "epoch": 0.44526850707320426,
      "grad_norm": 0.37483876943588257,
      "learning_rate": 1.556135770234987e-05,
      "loss": 1.0019,
      "step": 1621
    },
    {
      "epoch": 0.4455431946161242,
      "grad_norm": 0.3270060122013092,
      "learning_rate": 1.5558609317026246e-05,
      "loss": 0.6347,
      "step": 1622
    },
    {
      "epoch": 0.4458178821590441,
      "grad_norm": 0.3731410801410675,
      "learning_rate": 1.5555860931702627e-05,
      "loss": 0.9784,
      "step": 1623
    },
    {
      "epoch": 0.44609256970196404,
      "grad_norm": 0.4575214087963104,
      "learning_rate": 1.5553112546379004e-05,
      "loss": 0.9878,
      "step": 1624
    },
    {
      "epoch": 0.44636725724488396,
      "grad_norm": 0.3726681172847748,
      "learning_rate": 1.555036416105538e-05,
      "loss": 0.8091,
      "step": 1625
    },
    {
      "epoch": 0.4466419447878039,
      "grad_norm": 0.4535730481147766,
      "learning_rate": 1.5547615775731758e-05,
      "loss": 0.6457,
      "step": 1626
    },
    {
      "epoch": 0.4469166323307238,
      "grad_norm": 0.4598090648651123,
      "learning_rate": 1.5544867390408135e-05,
      "loss": 0.9812,
      "step": 1627
    },
    {
      "epoch": 0.44719131987364374,
      "grad_norm": 0.5025938749313354,
      "learning_rate": 1.5542119005084515e-05,
      "loss": 0.7336,
      "step": 1628
    },
    {
      "epoch": 0.44746600741656367,
      "grad_norm": 0.33232101798057556,
      "learning_rate": 1.5539370619760892e-05,
      "loss": 0.9985,
      "step": 1629
    },
    {
      "epoch": 0.4477406949594836,
      "grad_norm": 0.35521557927131653,
      "learning_rate": 1.553662223443727e-05,
      "loss": 0.7809,
      "step": 1630
    },
    {
      "epoch": 0.4480153825024035,
      "grad_norm": 0.39871740341186523,
      "learning_rate": 1.5533873849113646e-05,
      "loss": 1.1043,
      "step": 1631
    },
    {
      "epoch": 0.44829007004532345,
      "grad_norm": 0.26813092827796936,
      "learning_rate": 1.5531125463790023e-05,
      "loss": 0.9011,
      "step": 1632
    },
    {
      "epoch": 0.4485647575882434,
      "grad_norm": 0.45360642671585083,
      "learning_rate": 1.55283770784664e-05,
      "loss": 0.8152,
      "step": 1633
    },
    {
      "epoch": 0.4488394451311633,
      "grad_norm": 0.3606756627559662,
      "learning_rate": 1.552562869314278e-05,
      "loss": 0.8347,
      "step": 1634
    },
    {
      "epoch": 0.44911413267408323,
      "grad_norm": 0.5240343809127808,
      "learning_rate": 1.5522880307819158e-05,
      "loss": 0.6572,
      "step": 1635
    },
    {
      "epoch": 0.44938882021700316,
      "grad_norm": 0.30997341871261597,
      "learning_rate": 1.5520131922495535e-05,
      "loss": 0.5773,
      "step": 1636
    },
    {
      "epoch": 0.4496635077599231,
      "grad_norm": 0.433941125869751,
      "learning_rate": 1.5517383537171912e-05,
      "loss": 0.877,
      "step": 1637
    },
    {
      "epoch": 0.449938195302843,
      "grad_norm": 0.3308681547641754,
      "learning_rate": 1.551463515184829e-05,
      "loss": 0.8011,
      "step": 1638
    },
    {
      "epoch": 0.45021288284576294,
      "grad_norm": 0.4714398980140686,
      "learning_rate": 1.551188676652467e-05,
      "loss": 0.9755,
      "step": 1639
    },
    {
      "epoch": 0.45048757038868287,
      "grad_norm": 0.337138295173645,
      "learning_rate": 1.5509138381201046e-05,
      "loss": 0.5997,
      "step": 1640
    },
    {
      "epoch": 0.4507622579316028,
      "grad_norm": 0.4076809287071228,
      "learning_rate": 1.5506389995877423e-05,
      "loss": 0.7887,
      "step": 1641
    },
    {
      "epoch": 0.4510369454745227,
      "grad_norm": 0.4645211398601532,
      "learning_rate": 1.55036416105538e-05,
      "loss": 1.0056,
      "step": 1642
    },
    {
      "epoch": 0.45131163301744265,
      "grad_norm": 0.3819364011287689,
      "learning_rate": 1.5500893225230177e-05,
      "loss": 1.122,
      "step": 1643
    },
    {
      "epoch": 0.4515863205603626,
      "grad_norm": 0.45561280846595764,
      "learning_rate": 1.5498144839906558e-05,
      "loss": 0.8155,
      "step": 1644
    },
    {
      "epoch": 0.4518610081032825,
      "grad_norm": 0.4371189773082733,
      "learning_rate": 1.5495396454582935e-05,
      "loss": 0.878,
      "step": 1645
    },
    {
      "epoch": 0.45213569564620243,
      "grad_norm": 0.4278304874897003,
      "learning_rate": 1.5492648069259312e-05,
      "loss": 0.6994,
      "step": 1646
    },
    {
      "epoch": 0.45241038318912236,
      "grad_norm": 0.3367745578289032,
      "learning_rate": 1.548989968393569e-05,
      "loss": 0.9667,
      "step": 1647
    },
    {
      "epoch": 0.4526850707320423,
      "grad_norm": 0.4719601571559906,
      "learning_rate": 1.5487151298612066e-05,
      "loss": 1.026,
      "step": 1648
    },
    {
      "epoch": 0.4529597582749622,
      "grad_norm": 0.31744658946990967,
      "learning_rate": 1.5484402913288443e-05,
      "loss": 0.9136,
      "step": 1649
    },
    {
      "epoch": 0.45323444581788214,
      "grad_norm": 0.34601184725761414,
      "learning_rate": 1.5481654527964823e-05,
      "loss": 0.9745,
      "step": 1650
    },
    {
      "epoch": 0.45350913336080206,
      "grad_norm": 0.42874273657798767,
      "learning_rate": 1.54789061426412e-05,
      "loss": 0.8979,
      "step": 1651
    },
    {
      "epoch": 0.453783820903722,
      "grad_norm": 0.5118772983551025,
      "learning_rate": 1.5476157757317577e-05,
      "loss": 0.8265,
      "step": 1652
    },
    {
      "epoch": 0.4540585084466419,
      "grad_norm": 0.4313119053840637,
      "learning_rate": 1.5473409371993954e-05,
      "loss": 0.9791,
      "step": 1653
    },
    {
      "epoch": 0.4543331959895619,
      "grad_norm": 0.3173552453517914,
      "learning_rate": 1.547066098667033e-05,
      "loss": 1.045,
      "step": 1654
    },
    {
      "epoch": 0.4546078835324818,
      "grad_norm": 0.5109109282493591,
      "learning_rate": 1.5467912601346712e-05,
      "loss": 0.8392,
      "step": 1655
    },
    {
      "epoch": 0.45488257107540175,
      "grad_norm": 0.3348490297794342,
      "learning_rate": 1.546516421602309e-05,
      "loss": 0.9451,
      "step": 1656
    },
    {
      "epoch": 0.4551572586183217,
      "grad_norm": 0.3850952684879303,
      "learning_rate": 1.5462415830699466e-05,
      "loss": 0.783,
      "step": 1657
    },
    {
      "epoch": 0.4554319461612416,
      "grad_norm": 0.4112633168697357,
      "learning_rate": 1.5459667445375843e-05,
      "loss": 0.9634,
      "step": 1658
    },
    {
      "epoch": 0.45570663370416153,
      "grad_norm": 0.28144633769989014,
      "learning_rate": 1.545691906005222e-05,
      "loss": 0.7854,
      "step": 1659
    },
    {
      "epoch": 0.45598132124708146,
      "grad_norm": 0.5722391605377197,
      "learning_rate": 1.54541706747286e-05,
      "loss": 0.9185,
      "step": 1660
    },
    {
      "epoch": 0.4562560087900014,
      "grad_norm": 0.47092828154563904,
      "learning_rate": 1.5451422289404977e-05,
      "loss": 0.7554,
      "step": 1661
    },
    {
      "epoch": 0.4565306963329213,
      "grad_norm": 0.3501262366771698,
      "learning_rate": 1.5448673904081354e-05,
      "loss": 0.6721,
      "step": 1662
    },
    {
      "epoch": 0.45680538387584124,
      "grad_norm": 0.4469318091869354,
      "learning_rate": 1.544592551875773e-05,
      "loss": 0.7243,
      "step": 1663
    },
    {
      "epoch": 0.45708007141876117,
      "grad_norm": 0.43726876378059387,
      "learning_rate": 1.544317713343411e-05,
      "loss": 0.8671,
      "step": 1664
    },
    {
      "epoch": 0.4573547589616811,
      "grad_norm": 0.3738524317741394,
      "learning_rate": 1.5440428748110485e-05,
      "loss": 0.9106,
      "step": 1665
    },
    {
      "epoch": 0.457629446504601,
      "grad_norm": 0.37829312682151794,
      "learning_rate": 1.5437680362786866e-05,
      "loss": 0.953,
      "step": 1666
    },
    {
      "epoch": 0.45790413404752095,
      "grad_norm": 0.3263157308101654,
      "learning_rate": 1.543493197746324e-05,
      "loss": 0.633,
      "step": 1667
    },
    {
      "epoch": 0.4581788215904409,
      "grad_norm": 0.3709957003593445,
      "learning_rate": 1.543218359213962e-05,
      "loss": 0.6772,
      "step": 1668
    },
    {
      "epoch": 0.4584535091333608,
      "grad_norm": 0.375007688999176,
      "learning_rate": 1.5429435206815997e-05,
      "loss": 0.6578,
      "step": 1669
    },
    {
      "epoch": 0.45872819667628073,
      "grad_norm": 0.36688923835754395,
      "learning_rate": 1.5426686821492374e-05,
      "loss": 0.7834,
      "step": 1670
    },
    {
      "epoch": 0.45900288421920066,
      "grad_norm": 0.431913286447525,
      "learning_rate": 1.5423938436168754e-05,
      "loss": 0.8025,
      "step": 1671
    },
    {
      "epoch": 0.4592775717621206,
      "grad_norm": 0.32965222001075745,
      "learning_rate": 1.5421190050845128e-05,
      "loss": 0.9638,
      "step": 1672
    },
    {
      "epoch": 0.4595522593050405,
      "grad_norm": 0.36552050709724426,
      "learning_rate": 1.5418441665521505e-05,
      "loss": 0.819,
      "step": 1673
    },
    {
      "epoch": 0.45982694684796044,
      "grad_norm": 0.35918837785720825,
      "learning_rate": 1.5415693280197885e-05,
      "loss": 0.6979,
      "step": 1674
    },
    {
      "epoch": 0.46010163439088037,
      "grad_norm": 0.5483434200286865,
      "learning_rate": 1.5412944894874262e-05,
      "loss": 0.8518,
      "step": 1675
    },
    {
      "epoch": 0.4603763219338003,
      "grad_norm": 0.3716599643230438,
      "learning_rate": 1.5410196509550643e-05,
      "loss": 0.8142,
      "step": 1676
    },
    {
      "epoch": 0.4606510094767202,
      "grad_norm": 0.4001762270927429,
      "learning_rate": 1.5407448124227016e-05,
      "loss": 0.9838,
      "step": 1677
    },
    {
      "epoch": 0.46092569701964015,
      "grad_norm": 0.39122360944747925,
      "learning_rate": 1.5404699738903393e-05,
      "loss": 0.5369,
      "step": 1678
    },
    {
      "epoch": 0.4612003845625601,
      "grad_norm": 0.2991906702518463,
      "learning_rate": 1.5401951353579774e-05,
      "loss": 1.048,
      "step": 1679
    },
    {
      "epoch": 0.46147507210548,
      "grad_norm": 0.3818824887275696,
      "learning_rate": 1.539920296825615e-05,
      "loss": 0.9813,
      "step": 1680
    },
    {
      "epoch": 0.4617497596483999,
      "grad_norm": 0.37979021668434143,
      "learning_rate": 1.5396454582932528e-05,
      "loss": 0.6117,
      "step": 1681
    },
    {
      "epoch": 0.46202444719131985,
      "grad_norm": 0.35495316982269287,
      "learning_rate": 1.5393706197608905e-05,
      "loss": 0.8673,
      "step": 1682
    },
    {
      "epoch": 0.4622991347342398,
      "grad_norm": 0.2540515959262848,
      "learning_rate": 1.5390957812285282e-05,
      "loss": 0.8985,
      "step": 1683
    },
    {
      "epoch": 0.4625738222771597,
      "grad_norm": 0.31355369091033936,
      "learning_rate": 1.5388209426961662e-05,
      "loss": 0.9395,
      "step": 1684
    },
    {
      "epoch": 0.46284850982007963,
      "grad_norm": 0.4251198172569275,
      "learning_rate": 1.538546104163804e-05,
      "loss": 0.9344,
      "step": 1685
    },
    {
      "epoch": 0.46312319736299956,
      "grad_norm": 0.3283987045288086,
      "learning_rate": 1.5382712656314416e-05,
      "loss": 0.8563,
      "step": 1686
    },
    {
      "epoch": 0.46339788490591954,
      "grad_norm": 0.4165027141571045,
      "learning_rate": 1.5379964270990793e-05,
      "loss": 0.7369,
      "step": 1687
    },
    {
      "epoch": 0.46367257244883947,
      "grad_norm": 0.453006386756897,
      "learning_rate": 1.537721588566717e-05,
      "loss": 0.8856,
      "step": 1688
    },
    {
      "epoch": 0.4639472599917594,
      "grad_norm": 0.3118075430393219,
      "learning_rate": 1.5374467500343547e-05,
      "loss": 0.9479,
      "step": 1689
    },
    {
      "epoch": 0.4642219475346793,
      "grad_norm": 0.4178313910961151,
      "learning_rate": 1.5371719115019928e-05,
      "loss": 0.7638,
      "step": 1690
    },
    {
      "epoch": 0.46449663507759925,
      "grad_norm": 0.2696802318096161,
      "learning_rate": 1.5368970729696305e-05,
      "loss": 0.7515,
      "step": 1691
    },
    {
      "epoch": 0.4647713226205192,
      "grad_norm": 0.3624226748943329,
      "learning_rate": 1.5366222344372682e-05,
      "loss": 0.9243,
      "step": 1692
    },
    {
      "epoch": 0.4650460101634391,
      "grad_norm": 0.3667970895767212,
      "learning_rate": 1.536347395904906e-05,
      "loss": 0.7955,
      "step": 1693
    },
    {
      "epoch": 0.46532069770635903,
      "grad_norm": 0.408020943403244,
      "learning_rate": 1.5360725573725436e-05,
      "loss": 0.7481,
      "step": 1694
    },
    {
      "epoch": 0.46559538524927896,
      "grad_norm": 0.3460325300693512,
      "learning_rate": 1.5357977188401816e-05,
      "loss": 0.8755,
      "step": 1695
    },
    {
      "epoch": 0.4658700727921989,
      "grad_norm": 0.38280871510505676,
      "learning_rate": 1.5355228803078193e-05,
      "loss": 0.8822,
      "step": 1696
    },
    {
      "epoch": 0.4661447603351188,
      "grad_norm": 0.4130019247531891,
      "learning_rate": 1.535248041775457e-05,
      "loss": 0.9449,
      "step": 1697
    },
    {
      "epoch": 0.46641944787803874,
      "grad_norm": 0.2643417418003082,
      "learning_rate": 1.5349732032430947e-05,
      "loss": 0.6982,
      "step": 1698
    },
    {
      "epoch": 0.46669413542095867,
      "grad_norm": 0.3347115218639374,
      "learning_rate": 1.5346983647107324e-05,
      "loss": 0.9449,
      "step": 1699
    },
    {
      "epoch": 0.4669688229638786,
      "grad_norm": 0.30515795946121216,
      "learning_rate": 1.5344235261783705e-05,
      "loss": 0.9276,
      "step": 1700
    },
    {
      "epoch": 0.4672435105067985,
      "grad_norm": 0.4850423038005829,
      "learning_rate": 1.5341486876460082e-05,
      "loss": 0.7237,
      "step": 1701
    },
    {
      "epoch": 0.46751819804971845,
      "grad_norm": 0.3569662868976593,
      "learning_rate": 1.533873849113646e-05,
      "loss": 0.8857,
      "step": 1702
    },
    {
      "epoch": 0.4677928855926384,
      "grad_norm": 0.39849770069122314,
      "learning_rate": 1.5335990105812836e-05,
      "loss": 0.5712,
      "step": 1703
    },
    {
      "epoch": 0.4680675731355583,
      "grad_norm": 0.36394527554512024,
      "learning_rate": 1.5333241720489213e-05,
      "loss": 0.7588,
      "step": 1704
    },
    {
      "epoch": 0.46834226067847823,
      "grad_norm": 0.461366206407547,
      "learning_rate": 1.533049333516559e-05,
      "loss": 0.9189,
      "step": 1705
    },
    {
      "epoch": 0.46861694822139816,
      "grad_norm": 0.41130372881889343,
      "learning_rate": 1.532774494984197e-05,
      "loss": 0.8524,
      "step": 1706
    },
    {
      "epoch": 0.4688916357643181,
      "grad_norm": 0.4043683111667633,
      "learning_rate": 1.5324996564518347e-05,
      "loss": 0.6081,
      "step": 1707
    },
    {
      "epoch": 0.469166323307238,
      "grad_norm": 0.3834870755672455,
      "learning_rate": 1.5322248179194724e-05,
      "loss": 0.7095,
      "step": 1708
    },
    {
      "epoch": 0.46944101085015794,
      "grad_norm": 0.32102084159851074,
      "learning_rate": 1.53194997938711e-05,
      "loss": 0.8851,
      "step": 1709
    },
    {
      "epoch": 0.46971569839307786,
      "grad_norm": 0.46907201409339905,
      "learning_rate": 1.531675140854748e-05,
      "loss": 0.7389,
      "step": 1710
    },
    {
      "epoch": 0.4699903859359978,
      "grad_norm": 0.5050297379493713,
      "learning_rate": 1.531400302322386e-05,
      "loss": 0.9712,
      "step": 1711
    },
    {
      "epoch": 0.4702650734789177,
      "grad_norm": 0.3419533371925354,
      "learning_rate": 1.5311254637900236e-05,
      "loss": 0.6925,
      "step": 1712
    },
    {
      "epoch": 0.47053976102183764,
      "grad_norm": 0.33604076504707336,
      "learning_rate": 1.5308506252576613e-05,
      "loss": 0.7554,
      "step": 1713
    },
    {
      "epoch": 0.47081444856475757,
      "grad_norm": 0.4876772165298462,
      "learning_rate": 1.530575786725299e-05,
      "loss": 0.8549,
      "step": 1714
    },
    {
      "epoch": 0.4710891361076775,
      "grad_norm": 0.32823631167411804,
      "learning_rate": 1.5303009481929367e-05,
      "loss": 0.5988,
      "step": 1715
    },
    {
      "epoch": 0.4713638236505974,
      "grad_norm": 0.27982428669929504,
      "learning_rate": 1.5300261096605747e-05,
      "loss": 0.7286,
      "step": 1716
    },
    {
      "epoch": 0.47163851119351735,
      "grad_norm": 0.31605079770088196,
      "learning_rate": 1.5297512711282124e-05,
      "loss": 0.8402,
      "step": 1717
    },
    {
      "epoch": 0.4719131987364373,
      "grad_norm": 0.3770042955875397,
      "learning_rate": 1.52947643259585e-05,
      "loss": 0.6654,
      "step": 1718
    },
    {
      "epoch": 0.4721878862793572,
      "grad_norm": 0.4474579989910126,
      "learning_rate": 1.529201594063488e-05,
      "loss": 0.8654,
      "step": 1719
    },
    {
      "epoch": 0.47246257382227713,
      "grad_norm": 0.5032905340194702,
      "learning_rate": 1.5289267555311255e-05,
      "loss": 0.7316,
      "step": 1720
    },
    {
      "epoch": 0.4727372613651971,
      "grad_norm": 0.40043318271636963,
      "learning_rate": 1.5286519169987632e-05,
      "loss": 0.757,
      "step": 1721
    },
    {
      "epoch": 0.47301194890811704,
      "grad_norm": 0.5012389421463013,
      "learning_rate": 1.5283770784664013e-05,
      "loss": 0.9095,
      "step": 1722
    },
    {
      "epoch": 0.47328663645103697,
      "grad_norm": 0.30299112200737,
      "learning_rate": 1.528102239934039e-05,
      "loss": 0.4828,
      "step": 1723
    },
    {
      "epoch": 0.4735613239939569,
      "grad_norm": 0.45837345719337463,
      "learning_rate": 1.5278274014016767e-05,
      "loss": 0.8022,
      "step": 1724
    },
    {
      "epoch": 0.4738360115368768,
      "grad_norm": 0.5122838616371155,
      "learning_rate": 1.5275525628693144e-05,
      "loss": 0.8046,
      "step": 1725
    },
    {
      "epoch": 0.47411069907979675,
      "grad_norm": 0.3718684911727905,
      "learning_rate": 1.527277724336952e-05,
      "loss": 0.8763,
      "step": 1726
    },
    {
      "epoch": 0.4743853866227167,
      "grad_norm": 0.39383429288864136,
      "learning_rate": 1.52700288580459e-05,
      "loss": 0.9603,
      "step": 1727
    },
    {
      "epoch": 0.4746600741656366,
      "grad_norm": 0.3612394332885742,
      "learning_rate": 1.526728047272228e-05,
      "loss": 0.7982,
      "step": 1728
    },
    {
      "epoch": 0.47493476170855653,
      "grad_norm": 0.36803847551345825,
      "learning_rate": 1.5264532087398652e-05,
      "loss": 0.7742,
      "step": 1729
    },
    {
      "epoch": 0.47520944925147646,
      "grad_norm": 0.4272262752056122,
      "learning_rate": 1.5261783702075032e-05,
      "loss": 0.8678,
      "step": 1730
    },
    {
      "epoch": 0.4754841367943964,
      "grad_norm": 0.4963206350803375,
      "learning_rate": 1.525903531675141e-05,
      "loss": 0.7382,
      "step": 1731
    },
    {
      "epoch": 0.4757588243373163,
      "grad_norm": 0.34233883023262024,
      "learning_rate": 1.5256286931427788e-05,
      "loss": 0.9189,
      "step": 1732
    },
    {
      "epoch": 0.47603351188023624,
      "grad_norm": 0.5863878130912781,
      "learning_rate": 1.5253538546104165e-05,
      "loss": 0.795,
      "step": 1733
    },
    {
      "epoch": 0.47630819942315616,
      "grad_norm": 0.6783067584037781,
      "learning_rate": 1.5250790160780542e-05,
      "loss": 0.7435,
      "step": 1734
    },
    {
      "epoch": 0.4765828869660761,
      "grad_norm": 0.5266475081443787,
      "learning_rate": 1.5248041775456921e-05,
      "loss": 0.7849,
      "step": 1735
    },
    {
      "epoch": 0.476857574508996,
      "grad_norm": 0.3700103163719177,
      "learning_rate": 1.5245293390133298e-05,
      "loss": 0.7313,
      "step": 1736
    },
    {
      "epoch": 0.47713226205191595,
      "grad_norm": 0.3726087808609009,
      "learning_rate": 1.5242545004809675e-05,
      "loss": 0.8256,
      "step": 1737
    },
    {
      "epoch": 0.47740694959483587,
      "grad_norm": 0.44527295231819153,
      "learning_rate": 1.5239796619486054e-05,
      "loss": 0.7143,
      "step": 1738
    },
    {
      "epoch": 0.4776816371377558,
      "grad_norm": 0.5019312500953674,
      "learning_rate": 1.523704823416243e-05,
      "loss": 0.886,
      "step": 1739
    },
    {
      "epoch": 0.4779563246806757,
      "grad_norm": 0.6043859720230103,
      "learning_rate": 1.523429984883881e-05,
      "loss": 0.8532,
      "step": 1740
    },
    {
      "epoch": 0.47823101222359565,
      "grad_norm": 0.322735458612442,
      "learning_rate": 1.5231551463515186e-05,
      "loss": 0.8407,
      "step": 1741
    },
    {
      "epoch": 0.4785056997665156,
      "grad_norm": 0.35352256894111633,
      "learning_rate": 1.5228803078191563e-05,
      "loss": 0.7055,
      "step": 1742
    },
    {
      "epoch": 0.4787803873094355,
      "grad_norm": 0.5442696213722229,
      "learning_rate": 1.5226054692867942e-05,
      "loss": 0.8962,
      "step": 1743
    },
    {
      "epoch": 0.47905507485235543,
      "grad_norm": 0.34583428502082825,
      "learning_rate": 1.522330630754432e-05,
      "loss": 0.6888,
      "step": 1744
    },
    {
      "epoch": 0.47932976239527536,
      "grad_norm": 0.39304080605506897,
      "learning_rate": 1.5220557922220696e-05,
      "loss": 0.664,
      "step": 1745
    },
    {
      "epoch": 0.4796044499381953,
      "grad_norm": 0.34155160188674927,
      "learning_rate": 1.5217809536897075e-05,
      "loss": 1.0811,
      "step": 1746
    },
    {
      "epoch": 0.4798791374811152,
      "grad_norm": 0.44109055399894714,
      "learning_rate": 1.5215061151573452e-05,
      "loss": 1.0421,
      "step": 1747
    },
    {
      "epoch": 0.48015382502403514,
      "grad_norm": 0.31140005588531494,
      "learning_rate": 1.521231276624983e-05,
      "loss": 1.0845,
      "step": 1748
    },
    {
      "epoch": 0.48042851256695507,
      "grad_norm": 0.40607210993766785,
      "learning_rate": 1.5209564380926208e-05,
      "loss": 0.8241,
      "step": 1749
    },
    {
      "epoch": 0.480703200109875,
      "grad_norm": 0.3915346562862396,
      "learning_rate": 1.5206815995602585e-05,
      "loss": 0.6977,
      "step": 1750
    },
    {
      "epoch": 0.4809778876527949,
      "grad_norm": 0.30128175020217896,
      "learning_rate": 1.5204067610278963e-05,
      "loss": 0.9958,
      "step": 1751
    },
    {
      "epoch": 0.48125257519571485,
      "grad_norm": 0.40175074338912964,
      "learning_rate": 1.520131922495534e-05,
      "loss": 0.6742,
      "step": 1752
    },
    {
      "epoch": 0.4815272627386348,
      "grad_norm": 0.3423236608505249,
      "learning_rate": 1.5198570839631717e-05,
      "loss": 0.9327,
      "step": 1753
    },
    {
      "epoch": 0.48180195028155476,
      "grad_norm": 0.25899001955986023,
      "learning_rate": 1.5195822454308096e-05,
      "loss": 0.9693,
      "step": 1754
    },
    {
      "epoch": 0.4820766378244747,
      "grad_norm": 0.45510226488113403,
      "learning_rate": 1.5193074068984473e-05,
      "loss": 0.543,
      "step": 1755
    },
    {
      "epoch": 0.4823513253673946,
      "grad_norm": 0.5606169700622559,
      "learning_rate": 1.5190325683660852e-05,
      "loss": 0.7077,
      "step": 1756
    },
    {
      "epoch": 0.48262601291031454,
      "grad_norm": 0.34322530031204224,
      "learning_rate": 1.5187577298337229e-05,
      "loss": 0.7979,
      "step": 1757
    },
    {
      "epoch": 0.48290070045323447,
      "grad_norm": 0.35664859414100647,
      "learning_rate": 1.5184828913013604e-05,
      "loss": 0.9294,
      "step": 1758
    },
    {
      "epoch": 0.4831753879961544,
      "grad_norm": 1.4071060419082642,
      "learning_rate": 1.5182080527689985e-05,
      "loss": 0.9374,
      "step": 1759
    },
    {
      "epoch": 0.4834500755390743,
      "grad_norm": 0.47719112038612366,
      "learning_rate": 1.5179332142366362e-05,
      "loss": 0.9281,
      "step": 1760
    },
    {
      "epoch": 0.48372476308199425,
      "grad_norm": 0.35354065895080566,
      "learning_rate": 1.5176583757042737e-05,
      "loss": 0.7407,
      "step": 1761
    },
    {
      "epoch": 0.4839994506249142,
      "grad_norm": 0.41742199659347534,
      "learning_rate": 1.5173835371719117e-05,
      "loss": 0.9341,
      "step": 1762
    },
    {
      "epoch": 0.4842741381678341,
      "grad_norm": 0.3764139413833618,
      "learning_rate": 1.5171086986395493e-05,
      "loss": 0.809,
      "step": 1763
    },
    {
      "epoch": 0.484548825710754,
      "grad_norm": 0.3707970678806305,
      "learning_rate": 1.5168338601071873e-05,
      "loss": 0.7251,
      "step": 1764
    },
    {
      "epoch": 0.48482351325367395,
      "grad_norm": 0.42317095398902893,
      "learning_rate": 1.5165590215748248e-05,
      "loss": 0.6753,
      "step": 1765
    },
    {
      "epoch": 0.4850982007965939,
      "grad_norm": 0.3993869125843048,
      "learning_rate": 1.5162841830424625e-05,
      "loss": 0.7012,
      "step": 1766
    },
    {
      "epoch": 0.4853728883395138,
      "grad_norm": 0.3359992802143097,
      "learning_rate": 1.5160093445101006e-05,
      "loss": 0.9162,
      "step": 1767
    },
    {
      "epoch": 0.48564757588243374,
      "grad_norm": 0.4025302827358246,
      "learning_rate": 1.5157345059777381e-05,
      "loss": 0.9382,
      "step": 1768
    },
    {
      "epoch": 0.48592226342535366,
      "grad_norm": 0.39680391550064087,
      "learning_rate": 1.5154596674453758e-05,
      "loss": 0.7477,
      "step": 1769
    },
    {
      "epoch": 0.4861969509682736,
      "grad_norm": 0.35963091254234314,
      "learning_rate": 1.5151848289130137e-05,
      "loss": 0.7185,
      "step": 1770
    },
    {
      "epoch": 0.4864716385111935,
      "grad_norm": 0.41155266761779785,
      "learning_rate": 1.5149099903806514e-05,
      "loss": 0.8516,
      "step": 1771
    },
    {
      "epoch": 0.48674632605411344,
      "grad_norm": 0.3983778953552246,
      "learning_rate": 1.5146351518482894e-05,
      "loss": 0.8194,
      "step": 1772
    },
    {
      "epoch": 0.48702101359703337,
      "grad_norm": 0.37782275676727295,
      "learning_rate": 1.514360313315927e-05,
      "loss": 0.9312,
      "step": 1773
    },
    {
      "epoch": 0.4872957011399533,
      "grad_norm": 0.41353747248649597,
      "learning_rate": 1.5140854747835647e-05,
      "loss": 0.7767,
      "step": 1774
    },
    {
      "epoch": 0.4875703886828732,
      "grad_norm": 0.43969136476516724,
      "learning_rate": 1.5138106362512025e-05,
      "loss": 0.8721,
      "step": 1775
    },
    {
      "epoch": 0.48784507622579315,
      "grad_norm": 0.42617136240005493,
      "learning_rate": 1.5135357977188402e-05,
      "loss": 0.8443,
      "step": 1776
    },
    {
      "epoch": 0.4881197637687131,
      "grad_norm": 0.34805455803871155,
      "learning_rate": 1.513260959186478e-05,
      "loss": 0.7015,
      "step": 1777
    },
    {
      "epoch": 0.488394451311633,
      "grad_norm": 0.3783893883228302,
      "learning_rate": 1.5129861206541158e-05,
      "loss": 0.8839,
      "step": 1778
    },
    {
      "epoch": 0.48866913885455293,
      "grad_norm": 0.3429986834526062,
      "learning_rate": 1.5127112821217535e-05,
      "loss": 0.8536,
      "step": 1779
    },
    {
      "epoch": 0.48894382639747286,
      "grad_norm": 0.42853808403015137,
      "learning_rate": 1.5124364435893914e-05,
      "loss": 0.7787,
      "step": 1780
    },
    {
      "epoch": 0.4892185139403928,
      "grad_norm": 0.49906325340270996,
      "learning_rate": 1.5121616050570291e-05,
      "loss": 0.8149,
      "step": 1781
    },
    {
      "epoch": 0.4894932014833127,
      "grad_norm": 0.32787907123565674,
      "learning_rate": 1.5118867665246668e-05,
      "loss": 0.7422,
      "step": 1782
    },
    {
      "epoch": 0.48976788902623264,
      "grad_norm": 0.4772665500640869,
      "learning_rate": 1.5116119279923047e-05,
      "loss": 0.5603,
      "step": 1783
    },
    {
      "epoch": 0.49004257656915257,
      "grad_norm": 0.4489178955554962,
      "learning_rate": 1.5113370894599424e-05,
      "loss": 0.9427,
      "step": 1784
    },
    {
      "epoch": 0.4903172641120725,
      "grad_norm": 0.2983334958553314,
      "learning_rate": 1.51106225092758e-05,
      "loss": 0.7962,
      "step": 1785
    },
    {
      "epoch": 0.4905919516549924,
      "grad_norm": 0.5449787378311157,
      "learning_rate": 1.510787412395218e-05,
      "loss": 0.8826,
      "step": 1786
    },
    {
      "epoch": 0.4908666391979124,
      "grad_norm": 0.4114105999469757,
      "learning_rate": 1.5105125738628556e-05,
      "loss": 0.7644,
      "step": 1787
    },
    {
      "epoch": 0.49114132674083233,
      "grad_norm": 0.3533714711666107,
      "learning_rate": 1.5102377353304935e-05,
      "loss": 0.7878,
      "step": 1788
    },
    {
      "epoch": 0.49141601428375226,
      "grad_norm": 0.3111441433429718,
      "learning_rate": 1.5099628967981312e-05,
      "loss": 0.7891,
      "step": 1789
    },
    {
      "epoch": 0.4916907018266722,
      "grad_norm": 0.4337693750858307,
      "learning_rate": 1.509688058265769e-05,
      "loss": 0.6363,
      "step": 1790
    },
    {
      "epoch": 0.4919653893695921,
      "grad_norm": 0.4334784150123596,
      "learning_rate": 1.5094132197334068e-05,
      "loss": 0.9311,
      "step": 1791
    },
    {
      "epoch": 0.49224007691251204,
      "grad_norm": 0.4551042318344116,
      "learning_rate": 1.5091383812010445e-05,
      "loss": 0.8057,
      "step": 1792
    },
    {
      "epoch": 0.49251476445543196,
      "grad_norm": 0.376151442527771,
      "learning_rate": 1.5088635426686822e-05,
      "loss": 0.6458,
      "step": 1793
    },
    {
      "epoch": 0.4927894519983519,
      "grad_norm": 0.26825475692749023,
      "learning_rate": 1.50858870413632e-05,
      "loss": 0.5729,
      "step": 1794
    },
    {
      "epoch": 0.4930641395412718,
      "grad_norm": 0.3467924892902374,
      "learning_rate": 1.5083138656039578e-05,
      "loss": 0.8207,
      "step": 1795
    },
    {
      "epoch": 0.49333882708419174,
      "grad_norm": 0.4016483724117279,
      "learning_rate": 1.5080390270715956e-05,
      "loss": 0.9425,
      "step": 1796
    },
    {
      "epoch": 0.49361351462711167,
      "grad_norm": 0.4291705787181854,
      "learning_rate": 1.5077641885392333e-05,
      "loss": 0.6695,
      "step": 1797
    },
    {
      "epoch": 0.4938882021700316,
      "grad_norm": 0.18411192297935486,
      "learning_rate": 1.507489350006871e-05,
      "loss": 0.4158,
      "step": 1798
    },
    {
      "epoch": 0.4941628897129515,
      "grad_norm": 0.4342506527900696,
      "learning_rate": 1.507214511474509e-05,
      "loss": 0.6203,
      "step": 1799
    },
    {
      "epoch": 0.49443757725587145,
      "grad_norm": 0.6155367493629456,
      "learning_rate": 1.5069396729421466e-05,
      "loss": 0.8501,
      "step": 1800
    },
    {
      "epoch": 0.4947122647987914,
      "grad_norm": 0.4223400950431824,
      "learning_rate": 1.5066648344097843e-05,
      "loss": 0.8642,
      "step": 1801
    },
    {
      "epoch": 0.4949869523417113,
      "grad_norm": 0.3707761764526367,
      "learning_rate": 1.5063899958774222e-05,
      "loss": 1.0463,
      "step": 1802
    },
    {
      "epoch": 0.49526163988463123,
      "grad_norm": 0.31697607040405273,
      "learning_rate": 1.5061151573450599e-05,
      "loss": 0.623,
      "step": 1803
    },
    {
      "epoch": 0.49553632742755116,
      "grad_norm": 0.41592636704444885,
      "learning_rate": 1.5058403188126978e-05,
      "loss": 0.7083,
      "step": 1804
    },
    {
      "epoch": 0.4958110149704711,
      "grad_norm": 0.4073021113872528,
      "learning_rate": 1.5055654802803355e-05,
      "loss": 0.8194,
      "step": 1805
    },
    {
      "epoch": 0.496085702513391,
      "grad_norm": 0.3764280378818512,
      "learning_rate": 1.5052906417479732e-05,
      "loss": 0.7678,
      "step": 1806
    },
    {
      "epoch": 0.49636039005631094,
      "grad_norm": 0.4980545938014984,
      "learning_rate": 1.505015803215611e-05,
      "loss": 0.7599,
      "step": 1807
    },
    {
      "epoch": 0.49663507759923087,
      "grad_norm": 0.44343098998069763,
      "learning_rate": 1.5047409646832487e-05,
      "loss": 1.0757,
      "step": 1808
    },
    {
      "epoch": 0.4969097651421508,
      "grad_norm": 0.2981955111026764,
      "learning_rate": 1.5044661261508864e-05,
      "loss": 0.8469,
      "step": 1809
    },
    {
      "epoch": 0.4971844526850707,
      "grad_norm": 0.4096852242946625,
      "learning_rate": 1.5041912876185243e-05,
      "loss": 0.8498,
      "step": 1810
    },
    {
      "epoch": 0.49745914022799065,
      "grad_norm": 0.35476431250572205,
      "learning_rate": 1.503916449086162e-05,
      "loss": 0.8131,
      "step": 1811
    },
    {
      "epoch": 0.4977338277709106,
      "grad_norm": 0.43178582191467285,
      "learning_rate": 1.5036416105537999e-05,
      "loss": 0.9281,
      "step": 1812
    },
    {
      "epoch": 0.4980085153138305,
      "grad_norm": 0.34955891966819763,
      "learning_rate": 1.5033667720214376e-05,
      "loss": 0.9202,
      "step": 1813
    },
    {
      "epoch": 0.49828320285675043,
      "grad_norm": 0.37984898686408997,
      "learning_rate": 1.5030919334890753e-05,
      "loss": 0.8141,
      "step": 1814
    },
    {
      "epoch": 0.49855789039967036,
      "grad_norm": 0.4321152865886688,
      "learning_rate": 1.5028170949567132e-05,
      "loss": 0.585,
      "step": 1815
    },
    {
      "epoch": 0.4988325779425903,
      "grad_norm": 0.3585307002067566,
      "learning_rate": 1.5025422564243509e-05,
      "loss": 0.8581,
      "step": 1816
    },
    {
      "epoch": 0.4991072654855102,
      "grad_norm": 0.32126280665397644,
      "learning_rate": 1.5022674178919886e-05,
      "loss": 0.8807,
      "step": 1817
    },
    {
      "epoch": 0.49938195302843014,
      "grad_norm": 0.3777679204940796,
      "learning_rate": 1.5019925793596264e-05,
      "loss": 0.768,
      "step": 1818
    },
    {
      "epoch": 0.49965664057135006,
      "grad_norm": 0.4447338283061981,
      "learning_rate": 1.5017177408272641e-05,
      "loss": 1.0333,
      "step": 1819
    },
    {
      "epoch": 0.49993132811427,
      "grad_norm": 0.27767300605773926,
      "learning_rate": 1.501442902294902e-05,
      "loss": 0.9069,
      "step": 1820
    },
    {
      "epoch": 0.5002060156571899,
      "grad_norm": 0.3866852819919586,
      "learning_rate": 1.5011680637625397e-05,
      "loss": 0.7422,
      "step": 1821
    },
    {
      "epoch": 0.5004807032001098,
      "grad_norm": 0.38326260447502136,
      "learning_rate": 1.5008932252301773e-05,
      "loss": 0.9446,
      "step": 1822
    },
    {
      "epoch": 0.5007553907430298,
      "grad_norm": 0.5053102970123291,
      "learning_rate": 1.5006183866978153e-05,
      "loss": 0.8152,
      "step": 1823
    },
    {
      "epoch": 0.5010300782859497,
      "grad_norm": 0.48157528042793274,
      "learning_rate": 1.500343548165453e-05,
      "loss": 0.8689,
      "step": 1824
    },
    {
      "epoch": 0.5013047658288696,
      "grad_norm": 0.37919414043426514,
      "learning_rate": 1.5000687096330905e-05,
      "loss": 0.6693,
      "step": 1825
    },
    {
      "epoch": 0.5015794533717896,
      "grad_norm": 0.5460142493247986,
      "learning_rate": 1.4997938711007286e-05,
      "loss": 0.8689,
      "step": 1826
    },
    {
      "epoch": 0.5018541409147095,
      "grad_norm": 0.3446902632713318,
      "learning_rate": 1.4995190325683661e-05,
      "loss": 1.1336,
      "step": 1827
    },
    {
      "epoch": 0.5021288284576294,
      "grad_norm": 0.48599550127983093,
      "learning_rate": 1.4992441940360041e-05,
      "loss": 0.9625,
      "step": 1828
    },
    {
      "epoch": 0.5024035160005493,
      "grad_norm": 0.4371981918811798,
      "learning_rate": 1.4989693555036417e-05,
      "loss": 0.895,
      "step": 1829
    },
    {
      "epoch": 0.5026782035434693,
      "grad_norm": 0.34314578771591187,
      "learning_rate": 1.4986945169712794e-05,
      "loss": 0.6702,
      "step": 1830
    },
    {
      "epoch": 0.5029528910863892,
      "grad_norm": 0.27589061856269836,
      "learning_rate": 1.4984196784389174e-05,
      "loss": 0.8012,
      "step": 1831
    },
    {
      "epoch": 0.5032275786293091,
      "grad_norm": 0.43598106503486633,
      "learning_rate": 1.498144839906555e-05,
      "loss": 0.6742,
      "step": 1832
    },
    {
      "epoch": 0.503502266172229,
      "grad_norm": 0.49213793873786926,
      "learning_rate": 1.4978700013741927e-05,
      "loss": 0.858,
      "step": 1833
    },
    {
      "epoch": 0.503776953715149,
      "grad_norm": 0.4597683250904083,
      "learning_rate": 1.4975951628418305e-05,
      "loss": 0.8637,
      "step": 1834
    },
    {
      "epoch": 0.5040516412580689,
      "grad_norm": 0.48858878016471863,
      "learning_rate": 1.4973203243094682e-05,
      "loss": 0.9239,
      "step": 1835
    },
    {
      "epoch": 0.5043263288009888,
      "grad_norm": 0.5328519344329834,
      "learning_rate": 1.4970454857771061e-05,
      "loss": 0.8329,
      "step": 1836
    },
    {
      "epoch": 0.5046010163439089,
      "grad_norm": 0.3910696804523468,
      "learning_rate": 1.4967706472447438e-05,
      "loss": 1.0252,
      "step": 1837
    },
    {
      "epoch": 0.5048757038868288,
      "grad_norm": 0.3972834646701813,
      "learning_rate": 1.4964958087123815e-05,
      "loss": 0.6155,
      "step": 1838
    },
    {
      "epoch": 0.5051503914297487,
      "grad_norm": 0.423946738243103,
      "learning_rate": 1.4962209701800194e-05,
      "loss": 1.1287,
      "step": 1839
    },
    {
      "epoch": 0.5054250789726686,
      "grad_norm": 0.3472439646720886,
      "learning_rate": 1.495946131647657e-05,
      "loss": 0.9535,
      "step": 1840
    },
    {
      "epoch": 0.5056997665155886,
      "grad_norm": 0.3463735282421112,
      "learning_rate": 1.4956712931152948e-05,
      "loss": 1.1258,
      "step": 1841
    },
    {
      "epoch": 0.5059744540585085,
      "grad_norm": 0.3997192084789276,
      "learning_rate": 1.4953964545829326e-05,
      "loss": 0.8627,
      "step": 1842
    },
    {
      "epoch": 0.5062491416014284,
      "grad_norm": 0.32668259739875793,
      "learning_rate": 1.4951216160505704e-05,
      "loss": 0.8938,
      "step": 1843
    },
    {
      "epoch": 0.5065238291443483,
      "grad_norm": 0.39249947667121887,
      "learning_rate": 1.4948467775182082e-05,
      "loss": 0.6849,
      "step": 1844
    },
    {
      "epoch": 0.5067985166872683,
      "grad_norm": 0.4158893823623657,
      "learning_rate": 1.494571938985846e-05,
      "loss": 0.7985,
      "step": 1845
    },
    {
      "epoch": 0.5070732042301882,
      "grad_norm": 0.3468565046787262,
      "learning_rate": 1.4942971004534836e-05,
      "loss": 0.7483,
      "step": 1846
    },
    {
      "epoch": 0.5073478917731081,
      "grad_norm": 0.355884850025177,
      "learning_rate": 1.4940222619211215e-05,
      "loss": 0.7723,
      "step": 1847
    },
    {
      "epoch": 0.507622579316028,
      "grad_norm": 0.34024879336357117,
      "learning_rate": 1.4937474233887592e-05,
      "loss": 0.5786,
      "step": 1848
    },
    {
      "epoch": 0.507897266858948,
      "grad_norm": 0.49022313952445984,
      "learning_rate": 1.4934725848563969e-05,
      "loss": 0.7724,
      "step": 1849
    },
    {
      "epoch": 0.5081719544018679,
      "grad_norm": 0.4358989894390106,
      "learning_rate": 1.4931977463240348e-05,
      "loss": 0.8931,
      "step": 1850
    },
    {
      "epoch": 0.5084466419447878,
      "grad_norm": 0.41119417548179626,
      "learning_rate": 1.4929229077916725e-05,
      "loss": 0.7253,
      "step": 1851
    },
    {
      "epoch": 0.5087213294877078,
      "grad_norm": 0.4251299500465393,
      "learning_rate": 1.4926480692593103e-05,
      "loss": 0.8703,
      "step": 1852
    },
    {
      "epoch": 0.5089960170306277,
      "grad_norm": 0.39267709851264954,
      "learning_rate": 1.492373230726948e-05,
      "loss": 0.6749,
      "step": 1853
    },
    {
      "epoch": 0.5092707045735476,
      "grad_norm": 0.39821162819862366,
      "learning_rate": 1.4920983921945858e-05,
      "loss": 0.9939,
      "step": 1854
    },
    {
      "epoch": 0.5095453921164675,
      "grad_norm": 0.35704025626182556,
      "learning_rate": 1.4918235536622236e-05,
      "loss": 0.8783,
      "step": 1855
    },
    {
      "epoch": 0.5098200796593875,
      "grad_norm": 0.4849885106086731,
      "learning_rate": 1.4915487151298613e-05,
      "loss": 0.7775,
      "step": 1856
    },
    {
      "epoch": 0.5100947672023074,
      "grad_norm": 0.42794865369796753,
      "learning_rate": 1.491273876597499e-05,
      "loss": 0.8686,
      "step": 1857
    },
    {
      "epoch": 0.5103694547452273,
      "grad_norm": 0.45820242166519165,
      "learning_rate": 1.4909990380651369e-05,
      "loss": 0.6247,
      "step": 1858
    },
    {
      "epoch": 0.5106441422881473,
      "grad_norm": 0.33635449409484863,
      "learning_rate": 1.4907241995327746e-05,
      "loss": 0.8097,
      "step": 1859
    },
    {
      "epoch": 0.5109188298310672,
      "grad_norm": 0.4166935384273529,
      "learning_rate": 1.4904493610004125e-05,
      "loss": 0.9022,
      "step": 1860
    },
    {
      "epoch": 0.5111935173739871,
      "grad_norm": 0.3700962960720062,
      "learning_rate": 1.4901745224680502e-05,
      "loss": 0.8433,
      "step": 1861
    },
    {
      "epoch": 0.511468204916907,
      "grad_norm": 0.48065146803855896,
      "learning_rate": 1.4898996839356879e-05,
      "loss": 0.9458,
      "step": 1862
    },
    {
      "epoch": 0.511742892459827,
      "grad_norm": 0.4087192714214325,
      "learning_rate": 1.4896248454033257e-05,
      "loss": 0.918,
      "step": 1863
    },
    {
      "epoch": 0.5120175800027469,
      "grad_norm": 0.4517883062362671,
      "learning_rate": 1.4893500068709634e-05,
      "loss": 0.9151,
      "step": 1864
    },
    {
      "epoch": 0.5122922675456668,
      "grad_norm": 0.4098823666572571,
      "learning_rate": 1.4890751683386012e-05,
      "loss": 0.8722,
      "step": 1865
    },
    {
      "epoch": 0.5125669550885867,
      "grad_norm": 0.5189796090126038,
      "learning_rate": 1.488800329806239e-05,
      "loss": 0.8665,
      "step": 1866
    },
    {
      "epoch": 0.5128416426315067,
      "grad_norm": 0.4055899679660797,
      "learning_rate": 1.4885254912738767e-05,
      "loss": 0.9705,
      "step": 1867
    },
    {
      "epoch": 0.5131163301744266,
      "grad_norm": 0.39640066027641296,
      "learning_rate": 1.4882506527415146e-05,
      "loss": 0.7063,
      "step": 1868
    },
    {
      "epoch": 0.5133910177173465,
      "grad_norm": 0.26832911372184753,
      "learning_rate": 1.4879758142091523e-05,
      "loss": 0.4893,
      "step": 1869
    },
    {
      "epoch": 0.5136657052602664,
      "grad_norm": 0.4029400944709778,
      "learning_rate": 1.48770097567679e-05,
      "loss": 0.8198,
      "step": 1870
    },
    {
      "epoch": 0.5139403928031864,
      "grad_norm": 0.4172610640525818,
      "learning_rate": 1.4874261371444279e-05,
      "loss": 0.9807,
      "step": 1871
    },
    {
      "epoch": 0.5142150803461063,
      "grad_norm": 0.45050862431526184,
      "learning_rate": 1.4871512986120656e-05,
      "loss": 0.8388,
      "step": 1872
    },
    {
      "epoch": 0.5144897678890262,
      "grad_norm": 0.42162761092185974,
      "learning_rate": 1.4868764600797033e-05,
      "loss": 0.8888,
      "step": 1873
    },
    {
      "epoch": 0.5147644554319462,
      "grad_norm": 0.3001844882965088,
      "learning_rate": 1.4866016215473411e-05,
      "loss": 0.8034,
      "step": 1874
    },
    {
      "epoch": 0.5150391429748661,
      "grad_norm": 0.5313382744789124,
      "learning_rate": 1.4863267830149789e-05,
      "loss": 0.8131,
      "step": 1875
    },
    {
      "epoch": 0.515313830517786,
      "grad_norm": 0.3644242286682129,
      "learning_rate": 1.4860519444826167e-05,
      "loss": 0.5431,
      "step": 1876
    },
    {
      "epoch": 0.5155885180607059,
      "grad_norm": 0.4579552114009857,
      "learning_rate": 1.4857771059502544e-05,
      "loss": 0.8044,
      "step": 1877
    },
    {
      "epoch": 0.5158632056036259,
      "grad_norm": 0.44566693902015686,
      "learning_rate": 1.4855022674178921e-05,
      "loss": 0.8459,
      "step": 1878
    },
    {
      "epoch": 0.5161378931465458,
      "grad_norm": 0.5060917735099792,
      "learning_rate": 1.48522742888553e-05,
      "loss": 0.9385,
      "step": 1879
    },
    {
      "epoch": 0.5164125806894657,
      "grad_norm": 0.40121081471443176,
      "learning_rate": 1.4849525903531677e-05,
      "loss": 0.6789,
      "step": 1880
    },
    {
      "epoch": 0.5166872682323856,
      "grad_norm": 0.30303794145584106,
      "learning_rate": 1.4846777518208052e-05,
      "loss": 0.5044,
      "step": 1881
    },
    {
      "epoch": 0.5169619557753056,
      "grad_norm": 0.3317880630493164,
      "learning_rate": 1.4844029132884433e-05,
      "loss": 0.8246,
      "step": 1882
    },
    {
      "epoch": 0.5172366433182255,
      "grad_norm": 0.4291694164276123,
      "learning_rate": 1.484128074756081e-05,
      "loss": 0.7148,
      "step": 1883
    },
    {
      "epoch": 0.5175113308611454,
      "grad_norm": 0.3847351670265198,
      "learning_rate": 1.4838532362237188e-05,
      "loss": 0.8389,
      "step": 1884
    },
    {
      "epoch": 0.5177860184040654,
      "grad_norm": 0.46282958984375,
      "learning_rate": 1.4835783976913565e-05,
      "loss": 0.9377,
      "step": 1885
    },
    {
      "epoch": 0.5180607059469853,
      "grad_norm": 0.3724949061870575,
      "learning_rate": 1.483303559158994e-05,
      "loss": 0.9639,
      "step": 1886
    },
    {
      "epoch": 0.5183353934899052,
      "grad_norm": 0.569284200668335,
      "learning_rate": 1.4830287206266321e-05,
      "loss": 1.0062,
      "step": 1887
    },
    {
      "epoch": 0.5186100810328251,
      "grad_norm": 0.4784832298755646,
      "learning_rate": 1.4827538820942697e-05,
      "loss": 0.8918,
      "step": 1888
    },
    {
      "epoch": 0.5188847685757451,
      "grad_norm": 0.45698845386505127,
      "learning_rate": 1.4824790435619074e-05,
      "loss": 0.695,
      "step": 1889
    },
    {
      "epoch": 0.519159456118665,
      "grad_norm": 0.437092661857605,
      "learning_rate": 1.4822042050295454e-05,
      "loss": 0.8483,
      "step": 1890
    },
    {
      "epoch": 0.5194341436615849,
      "grad_norm": 0.4388923943042755,
      "learning_rate": 1.481929366497183e-05,
      "loss": 1.0129,
      "step": 1891
    },
    {
      "epoch": 0.5197088312045048,
      "grad_norm": 0.39617031812667847,
      "learning_rate": 1.481654527964821e-05,
      "loss": 0.8727,
      "step": 1892
    },
    {
      "epoch": 0.5199835187474248,
      "grad_norm": 0.33274412155151367,
      "learning_rate": 1.4813796894324585e-05,
      "loss": 0.8987,
      "step": 1893
    },
    {
      "epoch": 0.5202582062903447,
      "grad_norm": 0.4987473487854004,
      "learning_rate": 1.4811048509000962e-05,
      "loss": 0.9094,
      "step": 1894
    },
    {
      "epoch": 0.5205328938332646,
      "grad_norm": 0.5353153944015503,
      "learning_rate": 1.480830012367734e-05,
      "loss": 0.8969,
      "step": 1895
    },
    {
      "epoch": 0.5208075813761845,
      "grad_norm": 0.37733331322669983,
      "learning_rate": 1.4805551738353718e-05,
      "loss": 0.7499,
      "step": 1896
    },
    {
      "epoch": 0.5210822689191045,
      "grad_norm": 0.38766318559646606,
      "learning_rate": 1.4802803353030095e-05,
      "loss": 0.8555,
      "step": 1897
    },
    {
      "epoch": 0.5213569564620244,
      "grad_norm": 0.48471599817276,
      "learning_rate": 1.4800054967706474e-05,
      "loss": 0.8485,
      "step": 1898
    },
    {
      "epoch": 0.5216316440049443,
      "grad_norm": 0.31719446182250977,
      "learning_rate": 1.479730658238285e-05,
      "loss": 0.7398,
      "step": 1899
    },
    {
      "epoch": 0.5219063315478643,
      "grad_norm": 0.4500076472759247,
      "learning_rate": 1.479455819705923e-05,
      "loss": 0.8543,
      "step": 1900
    },
    {
      "epoch": 0.5221810190907842,
      "grad_norm": 0.574049174785614,
      "learning_rate": 1.4791809811735606e-05,
      "loss": 0.7126,
      "step": 1901
    },
    {
      "epoch": 0.5224557066337041,
      "grad_norm": 0.3799774944782257,
      "learning_rate": 1.4789061426411983e-05,
      "loss": 0.8451,
      "step": 1902
    },
    {
      "epoch": 0.522730394176624,
      "grad_norm": 0.37605515122413635,
      "learning_rate": 1.4786313041088362e-05,
      "loss": 0.7755,
      "step": 1903
    },
    {
      "epoch": 0.5230050817195441,
      "grad_norm": 0.4302169978618622,
      "learning_rate": 1.4783564655764739e-05,
      "loss": 0.7622,
      "step": 1904
    },
    {
      "epoch": 0.523279769262464,
      "grad_norm": 0.3914578855037689,
      "learning_rate": 1.4780816270441116e-05,
      "loss": 0.8904,
      "step": 1905
    },
    {
      "epoch": 0.5235544568053839,
      "grad_norm": 0.25931891798973083,
      "learning_rate": 1.4778067885117495e-05,
      "loss": 0.6151,
      "step": 1906
    },
    {
      "epoch": 0.5238291443483039,
      "grad_norm": 0.39938756823539734,
      "learning_rate": 1.4775319499793872e-05,
      "loss": 0.9247,
      "step": 1907
    },
    {
      "epoch": 0.5241038318912238,
      "grad_norm": 0.3384731113910675,
      "learning_rate": 1.4772571114470249e-05,
      "loss": 0.723,
      "step": 1908
    },
    {
      "epoch": 0.5243785194341437,
      "grad_norm": 0.4336825907230377,
      "learning_rate": 1.4769822729146628e-05,
      "loss": 0.8848,
      "step": 1909
    },
    {
      "epoch": 0.5246532069770636,
      "grad_norm": 0.34666910767555237,
      "learning_rate": 1.4767074343823005e-05,
      "loss": 0.8896,
      "step": 1910
    },
    {
      "epoch": 0.5249278945199836,
      "grad_norm": 0.3660959303379059,
      "learning_rate": 1.4764325958499383e-05,
      "loss": 0.7752,
      "step": 1911
    },
    {
      "epoch": 0.5252025820629035,
      "grad_norm": 0.41329535841941833,
      "learning_rate": 1.476157757317576e-05,
      "loss": 0.8173,
      "step": 1912
    },
    {
      "epoch": 0.5254772696058234,
      "grad_norm": 0.3945826590061188,
      "learning_rate": 1.4758829187852137e-05,
      "loss": 0.7271,
      "step": 1913
    },
    {
      "epoch": 0.5257519571487433,
      "grad_norm": 0.3442266583442688,
      "learning_rate": 1.4756080802528516e-05,
      "loss": 0.7636,
      "step": 1914
    },
    {
      "epoch": 0.5260266446916633,
      "grad_norm": 0.4068133533000946,
      "learning_rate": 1.4753332417204893e-05,
      "loss": 0.7835,
      "step": 1915
    },
    {
      "epoch": 0.5263013322345832,
      "grad_norm": 0.3379283547401428,
      "learning_rate": 1.475058403188127e-05,
      "loss": 0.6127,
      "step": 1916
    },
    {
      "epoch": 0.5265760197775031,
      "grad_norm": 0.36721745133399963,
      "learning_rate": 1.4747835646557649e-05,
      "loss": 0.9082,
      "step": 1917
    },
    {
      "epoch": 0.526850707320423,
      "grad_norm": 0.3483150899410248,
      "learning_rate": 1.4745087261234026e-05,
      "loss": 0.6711,
      "step": 1918
    },
    {
      "epoch": 0.527125394863343,
      "grad_norm": 0.5012164115905762,
      "learning_rate": 1.4742338875910405e-05,
      "loss": 0.7972,
      "step": 1919
    },
    {
      "epoch": 0.5274000824062629,
      "grad_norm": 0.4287840723991394,
      "learning_rate": 1.4739590490586782e-05,
      "loss": 0.9187,
      "step": 1920
    },
    {
      "epoch": 0.5276747699491828,
      "grad_norm": 0.34943264722824097,
      "learning_rate": 1.4736842105263159e-05,
      "loss": 0.7612,
      "step": 1921
    },
    {
      "epoch": 0.5279494574921028,
      "grad_norm": 0.3780249059200287,
      "learning_rate": 1.4734093719939537e-05,
      "loss": 0.8901,
      "step": 1922
    },
    {
      "epoch": 0.5282241450350227,
      "grad_norm": 0.4172881841659546,
      "learning_rate": 1.4731345334615914e-05,
      "loss": 0.7343,
      "step": 1923
    },
    {
      "epoch": 0.5284988325779426,
      "grad_norm": 0.32258209586143494,
      "learning_rate": 1.4728596949292291e-05,
      "loss": 0.6878,
      "step": 1924
    },
    {
      "epoch": 0.5287735201208625,
      "grad_norm": 0.2996840476989746,
      "learning_rate": 1.472584856396867e-05,
      "loss": 0.9719,
      "step": 1925
    },
    {
      "epoch": 0.5290482076637825,
      "grad_norm": 0.4877471327781677,
      "learning_rate": 1.4723100178645047e-05,
      "loss": 1.0697,
      "step": 1926
    },
    {
      "epoch": 0.5293228952067024,
      "grad_norm": 0.4135098159313202,
      "learning_rate": 1.4720351793321426e-05,
      "loss": 0.9194,
      "step": 1927
    },
    {
      "epoch": 0.5295975827496223,
      "grad_norm": 0.44737622141838074,
      "learning_rate": 1.4717603407997803e-05,
      "loss": 0.6931,
      "step": 1928
    },
    {
      "epoch": 0.5298722702925422,
      "grad_norm": 0.4255816638469696,
      "learning_rate": 1.471485502267418e-05,
      "loss": 0.8632,
      "step": 1929
    },
    {
      "epoch": 0.5301469578354622,
      "grad_norm": 0.4599296748638153,
      "learning_rate": 1.4712106637350559e-05,
      "loss": 0.8507,
      "step": 1930
    },
    {
      "epoch": 0.5304216453783821,
      "grad_norm": 0.3927897810935974,
      "learning_rate": 1.4709358252026936e-05,
      "loss": 0.8786,
      "step": 1931
    },
    {
      "epoch": 0.530696332921302,
      "grad_norm": 0.34990447759628296,
      "learning_rate": 1.4706609866703313e-05,
      "loss": 0.6354,
      "step": 1932
    },
    {
      "epoch": 0.530971020464222,
      "grad_norm": 0.32567963004112244,
      "learning_rate": 1.4703861481379691e-05,
      "loss": 0.7737,
      "step": 1933
    },
    {
      "epoch": 0.5312457080071419,
      "grad_norm": 0.5134514570236206,
      "learning_rate": 1.4701113096056068e-05,
      "loss": 0.6978,
      "step": 1934
    },
    {
      "epoch": 0.5315203955500618,
      "grad_norm": 0.33618539571762085,
      "learning_rate": 1.4698364710732447e-05,
      "loss": 0.9467,
      "step": 1935
    },
    {
      "epoch": 0.5317950830929817,
      "grad_norm": 0.4595257341861725,
      "learning_rate": 1.4695616325408824e-05,
      "loss": 0.9661,
      "step": 1936
    },
    {
      "epoch": 0.5320697706359017,
      "grad_norm": 0.29370027780532837,
      "learning_rate": 1.4692867940085201e-05,
      "loss": 0.8938,
      "step": 1937
    },
    {
      "epoch": 0.5323444581788216,
      "grad_norm": 0.4053916037082672,
      "learning_rate": 1.469011955476158e-05,
      "loss": 0.9675,
      "step": 1938
    },
    {
      "epoch": 0.5326191457217415,
      "grad_norm": 0.37841615080833435,
      "learning_rate": 1.4687371169437957e-05,
      "loss": 0.7495,
      "step": 1939
    },
    {
      "epoch": 0.5328938332646614,
      "grad_norm": 0.2530868649482727,
      "learning_rate": 1.4684622784114332e-05,
      "loss": 0.8086,
      "step": 1940
    },
    {
      "epoch": 0.5331685208075814,
      "grad_norm": 0.3922579288482666,
      "learning_rate": 1.4681874398790713e-05,
      "loss": 0.8572,
      "step": 1941
    },
    {
      "epoch": 0.5334432083505013,
      "grad_norm": 0.5174602270126343,
      "learning_rate": 1.467912601346709e-05,
      "loss": 0.9021,
      "step": 1942
    },
    {
      "epoch": 0.5337178958934212,
      "grad_norm": 0.3721178472042084,
      "learning_rate": 1.4676377628143468e-05,
      "loss": 0.6795,
      "step": 1943
    },
    {
      "epoch": 0.5339925834363412,
      "grad_norm": 0.3532138764858246,
      "learning_rate": 1.4673629242819845e-05,
      "loss": 0.9538,
      "step": 1944
    },
    {
      "epoch": 0.5342672709792611,
      "grad_norm": 0.3366652727127075,
      "learning_rate": 1.467088085749622e-05,
      "loss": 0.5866,
      "step": 1945
    },
    {
      "epoch": 0.534541958522181,
      "grad_norm": 0.4550861418247223,
      "learning_rate": 1.4668132472172601e-05,
      "loss": 0.8068,
      "step": 1946
    },
    {
      "epoch": 0.5348166460651009,
      "grad_norm": 0.2992922365665436,
      "learning_rate": 1.4665384086848976e-05,
      "loss": 0.7526,
      "step": 1947
    },
    {
      "epoch": 0.5350913336080209,
      "grad_norm": 0.3986852765083313,
      "learning_rate": 1.4662635701525353e-05,
      "loss": 0.8491,
      "step": 1948
    },
    {
      "epoch": 0.5353660211509408,
      "grad_norm": 0.4550762176513672,
      "learning_rate": 1.4659887316201734e-05,
      "loss": 0.8274,
      "step": 1949
    },
    {
      "epoch": 0.5356407086938607,
      "grad_norm": 0.4841098189353943,
      "learning_rate": 1.4657138930878109e-05,
      "loss": 0.8027,
      "step": 1950
    },
    {
      "epoch": 0.5359153962367806,
      "grad_norm": 0.3223857581615448,
      "learning_rate": 1.465439054555449e-05,
      "loss": 0.5865,
      "step": 1951
    },
    {
      "epoch": 0.5361900837797006,
      "grad_norm": 0.3902454078197479,
      "learning_rate": 1.4651642160230865e-05,
      "loss": 0.832,
      "step": 1952
    },
    {
      "epoch": 0.5364647713226205,
      "grad_norm": 0.44486376643180847,
      "learning_rate": 1.4648893774907242e-05,
      "loss": 0.869,
      "step": 1953
    },
    {
      "epoch": 0.5367394588655404,
      "grad_norm": 0.34311404824256897,
      "learning_rate": 1.464614538958362e-05,
      "loss": 0.9286,
      "step": 1954
    },
    {
      "epoch": 0.5370141464084603,
      "grad_norm": 0.46717435121536255,
      "learning_rate": 1.4643397004259998e-05,
      "loss": 0.8053,
      "step": 1955
    },
    {
      "epoch": 0.5372888339513803,
      "grad_norm": 0.3327638506889343,
      "learning_rate": 1.4640648618936375e-05,
      "loss": 0.6176,
      "step": 1956
    },
    {
      "epoch": 0.5375635214943002,
      "grad_norm": 0.375698447227478,
      "learning_rate": 1.4637900233612753e-05,
      "loss": 0.8513,
      "step": 1957
    },
    {
      "epoch": 0.5378382090372201,
      "grad_norm": 0.26798272132873535,
      "learning_rate": 1.463515184828913e-05,
      "loss": 0.502,
      "step": 1958
    },
    {
      "epoch": 0.53811289658014,
      "grad_norm": 0.42146435379981995,
      "learning_rate": 1.4632403462965509e-05,
      "loss": 0.7759,
      "step": 1959
    },
    {
      "epoch": 0.53838758412306,
      "grad_norm": 0.42748475074768066,
      "learning_rate": 1.4629655077641886e-05,
      "loss": 0.6564,
      "step": 1960
    },
    {
      "epoch": 0.5386622716659799,
      "grad_norm": 0.33597663044929504,
      "learning_rate": 1.4626906692318263e-05,
      "loss": 0.6174,
      "step": 1961
    },
    {
      "epoch": 0.5389369592088998,
      "grad_norm": 0.4991922378540039,
      "learning_rate": 1.4624158306994642e-05,
      "loss": 0.8584,
      "step": 1962
    },
    {
      "epoch": 0.5392116467518198,
      "grad_norm": 0.35667479038238525,
      "learning_rate": 1.4621409921671019e-05,
      "loss": 0.8972,
      "step": 1963
    },
    {
      "epoch": 0.5394863342947397,
      "grad_norm": 0.4231280982494354,
      "learning_rate": 1.4618661536347396e-05,
      "loss": 0.8753,
      "step": 1964
    },
    {
      "epoch": 0.5397610218376596,
      "grad_norm": 0.4833252727985382,
      "learning_rate": 1.4615913151023775e-05,
      "loss": 0.7999,
      "step": 1965
    },
    {
      "epoch": 0.5400357093805795,
      "grad_norm": 0.46652811765670776,
      "learning_rate": 1.4613164765700152e-05,
      "loss": 0.72,
      "step": 1966
    },
    {
      "epoch": 0.5403103969234995,
      "grad_norm": 0.3631334900856018,
      "learning_rate": 1.461041638037653e-05,
      "loss": 0.8048,
      "step": 1967
    },
    {
      "epoch": 0.5405850844664194,
      "grad_norm": 0.3676178753376007,
      "learning_rate": 1.4607667995052907e-05,
      "loss": 0.5832,
      "step": 1968
    },
    {
      "epoch": 0.5408597720093393,
      "grad_norm": 0.3768589496612549,
      "learning_rate": 1.4604919609729284e-05,
      "loss": 0.833,
      "step": 1969
    },
    {
      "epoch": 0.5411344595522593,
      "grad_norm": 0.5356301665306091,
      "learning_rate": 1.4602171224405663e-05,
      "loss": 0.7977,
      "step": 1970
    },
    {
      "epoch": 0.5414091470951793,
      "grad_norm": 0.46990638971328735,
      "learning_rate": 1.459942283908204e-05,
      "loss": 0.576,
      "step": 1971
    },
    {
      "epoch": 0.5416838346380992,
      "grad_norm": 0.4229279160499573,
      "learning_rate": 1.4596674453758417e-05,
      "loss": 0.7027,
      "step": 1972
    },
    {
      "epoch": 0.5419585221810191,
      "grad_norm": 0.42901259660720825,
      "learning_rate": 1.4593926068434796e-05,
      "loss": 0.5695,
      "step": 1973
    },
    {
      "epoch": 0.5422332097239391,
      "grad_norm": 0.42031392455101013,
      "learning_rate": 1.4591177683111173e-05,
      "loss": 0.875,
      "step": 1974
    },
    {
      "epoch": 0.542507897266859,
      "grad_norm": 0.41739779710769653,
      "learning_rate": 1.4588429297787552e-05,
      "loss": 0.8219,
      "step": 1975
    },
    {
      "epoch": 0.5427825848097789,
      "grad_norm": 0.462126225233078,
      "learning_rate": 1.4585680912463929e-05,
      "loss": 1.0706,
      "step": 1976
    },
    {
      "epoch": 0.5430572723526988,
      "grad_norm": 0.35250604152679443,
      "learning_rate": 1.4582932527140306e-05,
      "loss": 0.6753,
      "step": 1977
    },
    {
      "epoch": 0.5433319598956188,
      "grad_norm": 0.36060020327568054,
      "learning_rate": 1.4580184141816684e-05,
      "loss": 0.7076,
      "step": 1978
    },
    {
      "epoch": 0.5436066474385387,
      "grad_norm": 0.3256727159023285,
      "learning_rate": 1.4577435756493061e-05,
      "loss": 0.4368,
      "step": 1979
    },
    {
      "epoch": 0.5438813349814586,
      "grad_norm": 0.404375284910202,
      "learning_rate": 1.4574687371169438e-05,
      "loss": 0.7966,
      "step": 1980
    },
    {
      "epoch": 0.5441560225243786,
      "grad_norm": 0.36992907524108887,
      "learning_rate": 1.4571938985845817e-05,
      "loss": 0.7406,
      "step": 1981
    },
    {
      "epoch": 0.5444307100672985,
      "grad_norm": 0.37437841296195984,
      "learning_rate": 1.4569190600522194e-05,
      "loss": 0.6867,
      "step": 1982
    },
    {
      "epoch": 0.5447053976102184,
      "grad_norm": 0.32635825872421265,
      "learning_rate": 1.4566442215198573e-05,
      "loss": 0.7482,
      "step": 1983
    },
    {
      "epoch": 0.5449800851531383,
      "grad_norm": 0.5968158841133118,
      "learning_rate": 1.456369382987495e-05,
      "loss": 0.9001,
      "step": 1984
    },
    {
      "epoch": 0.5452547726960583,
      "grad_norm": 0.3787171244621277,
      "learning_rate": 1.4560945444551327e-05,
      "loss": 1.0111,
      "step": 1985
    },
    {
      "epoch": 0.5455294602389782,
      "grad_norm": 0.3830743134021759,
      "learning_rate": 1.4558197059227706e-05,
      "loss": 0.8314,
      "step": 1986
    },
    {
      "epoch": 0.5458041477818981,
      "grad_norm": 0.38278353214263916,
      "learning_rate": 1.4555448673904083e-05,
      "loss": 0.9797,
      "step": 1987
    },
    {
      "epoch": 0.546078835324818,
      "grad_norm": 0.46235841512680054,
      "learning_rate": 1.455270028858046e-05,
      "loss": 0.9429,
      "step": 1988
    },
    {
      "epoch": 0.546353522867738,
      "grad_norm": 0.41826751828193665,
      "learning_rate": 1.4549951903256838e-05,
      "loss": 0.7375,
      "step": 1989
    },
    {
      "epoch": 0.5466282104106579,
      "grad_norm": 0.34495848417282104,
      "learning_rate": 1.4547203517933215e-05,
      "loss": 0.9917,
      "step": 1990
    },
    {
      "epoch": 0.5469028979535778,
      "grad_norm": 0.4414818584918976,
      "learning_rate": 1.4544455132609594e-05,
      "loss": 0.744,
      "step": 1991
    },
    {
      "epoch": 0.5471775854964978,
      "grad_norm": 0.4475363790988922,
      "learning_rate": 1.4541706747285971e-05,
      "loss": 0.8988,
      "step": 1992
    },
    {
      "epoch": 0.5474522730394177,
      "grad_norm": 0.4193134009838104,
      "learning_rate": 1.4538958361962348e-05,
      "loss": 0.8817,
      "step": 1993
    },
    {
      "epoch": 0.5477269605823376,
      "grad_norm": 0.4059814512729645,
      "learning_rate": 1.4536209976638727e-05,
      "loss": 0.862,
      "step": 1994
    },
    {
      "epoch": 0.5480016481252575,
      "grad_norm": 0.44238221645355225,
      "learning_rate": 1.4533461591315104e-05,
      "loss": 0.7049,
      "step": 1995
    },
    {
      "epoch": 0.5482763356681775,
      "grad_norm": 0.37093666195869446,
      "learning_rate": 1.4530713205991481e-05,
      "loss": 0.8226,
      "step": 1996
    },
    {
      "epoch": 0.5485510232110974,
      "grad_norm": 0.37669000029563904,
      "learning_rate": 1.452796482066786e-05,
      "loss": 0.8,
      "step": 1997
    },
    {
      "epoch": 0.5488257107540173,
      "grad_norm": 0.32667282223701477,
      "learning_rate": 1.4525216435344237e-05,
      "loss": 0.8448,
      "step": 1998
    },
    {
      "epoch": 0.5491003982969372,
      "grad_norm": 0.2887444794178009,
      "learning_rate": 1.4522468050020615e-05,
      "loss": 0.5222,
      "step": 1999
    },
    {
      "epoch": 0.5493750858398572,
      "grad_norm": 0.5591833591461182,
      "learning_rate": 1.4519719664696992e-05,
      "loss": 1.0962,
      "step": 2000
    },
    {
      "epoch": 0.5496497733827771,
      "grad_norm": 0.3125847578048706,
      "learning_rate": 1.451697127937337e-05,
      "loss": 0.4734,
      "step": 2001
    },
    {
      "epoch": 0.549924460925697,
      "grad_norm": 0.48768049478530884,
      "learning_rate": 1.4514222894049748e-05,
      "loss": 1.023,
      "step": 2002
    },
    {
      "epoch": 0.550199148468617,
      "grad_norm": 0.4843409061431885,
      "learning_rate": 1.4511474508726125e-05,
      "loss": 0.7309,
      "step": 2003
    },
    {
      "epoch": 0.5504738360115369,
      "grad_norm": 0.37693047523498535,
      "learning_rate": 1.45087261234025e-05,
      "loss": 0.8791,
      "step": 2004
    },
    {
      "epoch": 0.5507485235544568,
      "grad_norm": 0.4105978012084961,
      "learning_rate": 1.450597773807888e-05,
      "loss": 0.5973,
      "step": 2005
    },
    {
      "epoch": 0.5510232110973767,
      "grad_norm": 0.4829584062099457,
      "learning_rate": 1.4503229352755258e-05,
      "loss": 0.9181,
      "step": 2006
    },
    {
      "epoch": 0.5512978986402967,
      "grad_norm": 0.3222130537033081,
      "learning_rate": 1.4500480967431637e-05,
      "loss": 0.6609,
      "step": 2007
    },
    {
      "epoch": 0.5515725861832166,
      "grad_norm": 0.3657008707523346,
      "learning_rate": 1.4497732582108014e-05,
      "loss": 0.7715,
      "step": 2008
    },
    {
      "epoch": 0.5518472737261365,
      "grad_norm": 0.49810972809791565,
      "learning_rate": 1.4494984196784389e-05,
      "loss": 0.7514,
      "step": 2009
    },
    {
      "epoch": 0.5521219612690564,
      "grad_norm": 0.4560610055923462,
      "learning_rate": 1.449223581146077e-05,
      "loss": 0.6883,
      "step": 2010
    },
    {
      "epoch": 0.5523966488119764,
      "grad_norm": 0.4109652042388916,
      "learning_rate": 1.4489487426137145e-05,
      "loss": 1.1211,
      "step": 2011
    },
    {
      "epoch": 0.5526713363548963,
      "grad_norm": 0.4757671058177948,
      "learning_rate": 1.4486739040813522e-05,
      "loss": 0.9611,
      "step": 2012
    },
    {
      "epoch": 0.5529460238978162,
      "grad_norm": 0.30716225504875183,
      "learning_rate": 1.4483990655489902e-05,
      "loss": 0.5463,
      "step": 2013
    },
    {
      "epoch": 0.5532207114407361,
      "grad_norm": 0.30037519335746765,
      "learning_rate": 1.4481242270166277e-05,
      "loss": 0.6827,
      "step": 2014
    },
    {
      "epoch": 0.5534953989836561,
      "grad_norm": 0.36217522621154785,
      "learning_rate": 1.4478493884842658e-05,
      "loss": 0.9174,
      "step": 2015
    },
    {
      "epoch": 0.553770086526576,
      "grad_norm": 0.42983195185661316,
      "learning_rate": 1.4475745499519033e-05,
      "loss": 1.0061,
      "step": 2016
    },
    {
      "epoch": 0.5540447740694959,
      "grad_norm": 0.3296232521533966,
      "learning_rate": 1.447299711419541e-05,
      "loss": 0.7641,
      "step": 2017
    },
    {
      "epoch": 0.5543194616124159,
      "grad_norm": 0.2294737696647644,
      "learning_rate": 1.4470248728871789e-05,
      "loss": 0.4821,
      "step": 2018
    },
    {
      "epoch": 0.5545941491553358,
      "grad_norm": 0.4463423490524292,
      "learning_rate": 1.4467500343548166e-05,
      "loss": 0.8233,
      "step": 2019
    },
    {
      "epoch": 0.5548688366982557,
      "grad_norm": 0.30809515714645386,
      "learning_rate": 1.4464751958224543e-05,
      "loss": 0.7024,
      "step": 2020
    },
    {
      "epoch": 0.5551435242411756,
      "grad_norm": 0.39970147609710693,
      "learning_rate": 1.4462003572900922e-05,
      "loss": 0.9065,
      "step": 2021
    },
    {
      "epoch": 0.5554182117840956,
      "grad_norm": 0.424552321434021,
      "learning_rate": 1.4459255187577299e-05,
      "loss": 0.8263,
      "step": 2022
    },
    {
      "epoch": 0.5556928993270155,
      "grad_norm": 0.5085424780845642,
      "learning_rate": 1.4456506802253677e-05,
      "loss": 0.89,
      "step": 2023
    },
    {
      "epoch": 0.5559675868699354,
      "grad_norm": 0.4064384996891022,
      "learning_rate": 1.4453758416930054e-05,
      "loss": 0.8705,
      "step": 2024
    },
    {
      "epoch": 0.5562422744128553,
      "grad_norm": 0.45239222049713135,
      "learning_rate": 1.4451010031606431e-05,
      "loss": 0.7425,
      "step": 2025
    },
    {
      "epoch": 0.5565169619557753,
      "grad_norm": 0.5172197818756104,
      "learning_rate": 1.444826164628281e-05,
      "loss": 0.7222,
      "step": 2026
    },
    {
      "epoch": 0.5567916494986952,
      "grad_norm": 0.33925551176071167,
      "learning_rate": 1.4445513260959187e-05,
      "loss": 0.7109,
      "step": 2027
    },
    {
      "epoch": 0.5570663370416151,
      "grad_norm": 0.2995378375053406,
      "learning_rate": 1.4442764875635564e-05,
      "loss": 0.7442,
      "step": 2028
    },
    {
      "epoch": 0.557341024584535,
      "grad_norm": 0.3829374611377716,
      "learning_rate": 1.4440016490311943e-05,
      "loss": 0.8523,
      "step": 2029
    },
    {
      "epoch": 0.557615712127455,
      "grad_norm": 0.4981348216533661,
      "learning_rate": 1.443726810498832e-05,
      "loss": 0.7799,
      "step": 2030
    },
    {
      "epoch": 0.5578903996703749,
      "grad_norm": 0.4656449258327484,
      "learning_rate": 1.4434519719664699e-05,
      "loss": 0.9049,
      "step": 2031
    },
    {
      "epoch": 0.5581650872132948,
      "grad_norm": 0.40494269132614136,
      "learning_rate": 1.4431771334341076e-05,
      "loss": 1.0489,
      "step": 2032
    },
    {
      "epoch": 0.5584397747562148,
      "grad_norm": 0.37615731358528137,
      "learning_rate": 1.4429022949017453e-05,
      "loss": 0.795,
      "step": 2033
    },
    {
      "epoch": 0.5587144622991347,
      "grad_norm": 0.3447233736515045,
      "learning_rate": 1.4426274563693831e-05,
      "loss": 0.9811,
      "step": 2034
    },
    {
      "epoch": 0.5589891498420546,
      "grad_norm": 0.4607302248477936,
      "learning_rate": 1.4423526178370208e-05,
      "loss": 0.8168,
      "step": 2035
    },
    {
      "epoch": 0.5592638373849745,
      "grad_norm": 0.36946335434913635,
      "learning_rate": 1.4420777793046585e-05,
      "loss": 0.8766,
      "step": 2036
    },
    {
      "epoch": 0.5595385249278945,
      "grad_norm": 0.3762160837650299,
      "learning_rate": 1.4418029407722964e-05,
      "loss": 0.7601,
      "step": 2037
    },
    {
      "epoch": 0.5598132124708145,
      "grad_norm": 0.41896116733551025,
      "learning_rate": 1.4415281022399341e-05,
      "loss": 0.6798,
      "step": 2038
    },
    {
      "epoch": 0.5600879000137344,
      "grad_norm": 0.4779491722583771,
      "learning_rate": 1.441253263707572e-05,
      "loss": 0.7775,
      "step": 2039
    },
    {
      "epoch": 0.5603625875566544,
      "grad_norm": 0.46326562762260437,
      "learning_rate": 1.4409784251752097e-05,
      "loss": 0.929,
      "step": 2040
    },
    {
      "epoch": 0.5606372750995743,
      "grad_norm": 0.42503100633621216,
      "learning_rate": 1.4407035866428474e-05,
      "loss": 0.89,
      "step": 2041
    },
    {
      "epoch": 0.5609119626424942,
      "grad_norm": 0.4689219295978546,
      "learning_rate": 1.4404287481104853e-05,
      "loss": 1.0766,
      "step": 2042
    },
    {
      "epoch": 0.5611866501854141,
      "grad_norm": 0.43797430396080017,
      "learning_rate": 1.440153909578123e-05,
      "loss": 0.8551,
      "step": 2043
    },
    {
      "epoch": 0.5614613377283341,
      "grad_norm": 0.3566569983959198,
      "learning_rate": 1.4398790710457607e-05,
      "loss": 0.7234,
      "step": 2044
    },
    {
      "epoch": 0.561736025271254,
      "grad_norm": 0.3423462510108948,
      "learning_rate": 1.4396042325133985e-05,
      "loss": 0.6467,
      "step": 2045
    },
    {
      "epoch": 0.5620107128141739,
      "grad_norm": 0.37698450684547424,
      "learning_rate": 1.4393293939810362e-05,
      "loss": 0.9082,
      "step": 2046
    },
    {
      "epoch": 0.5622854003570938,
      "grad_norm": 0.522363543510437,
      "learning_rate": 1.4390545554486741e-05,
      "loss": 0.948,
      "step": 2047
    },
    {
      "epoch": 0.5625600879000138,
      "grad_norm": 0.3546285629272461,
      "learning_rate": 1.4387797169163118e-05,
      "loss": 0.9672,
      "step": 2048
    },
    {
      "epoch": 0.5628347754429337,
      "grad_norm": 0.34626102447509766,
      "learning_rate": 1.4385048783839495e-05,
      "loss": 0.8205,
      "step": 2049
    },
    {
      "epoch": 0.5631094629858536,
      "grad_norm": 0.4418241083621979,
      "learning_rate": 1.4382300398515874e-05,
      "loss": 0.9691,
      "step": 2050
    },
    {
      "epoch": 0.5633841505287736,
      "grad_norm": 0.46060940623283386,
      "learning_rate": 1.4379552013192251e-05,
      "loss": 0.9324,
      "step": 2051
    },
    {
      "epoch": 0.5636588380716935,
      "grad_norm": 0.47403499484062195,
      "learning_rate": 1.4376803627868628e-05,
      "loss": 0.8402,
      "step": 2052
    },
    {
      "epoch": 0.5639335256146134,
      "grad_norm": 0.42271625995635986,
      "learning_rate": 1.4374055242545007e-05,
      "loss": 0.6376,
      "step": 2053
    },
    {
      "epoch": 0.5642082131575333,
      "grad_norm": 0.43698742985725403,
      "learning_rate": 1.4371306857221384e-05,
      "loss": 0.9993,
      "step": 2054
    },
    {
      "epoch": 0.5644829007004533,
      "grad_norm": 0.3474348187446594,
      "learning_rate": 1.4368558471897762e-05,
      "loss": 0.8787,
      "step": 2055
    },
    {
      "epoch": 0.5647575882433732,
      "grad_norm": 0.33036133646965027,
      "learning_rate": 1.436581008657414e-05,
      "loss": 0.7251,
      "step": 2056
    },
    {
      "epoch": 0.5650322757862931,
      "grad_norm": 0.4494204819202423,
      "learning_rate": 1.4363061701250516e-05,
      "loss": 0.717,
      "step": 2057
    },
    {
      "epoch": 0.565306963329213,
      "grad_norm": 0.5029399394989014,
      "learning_rate": 1.4360313315926895e-05,
      "loss": 0.8446,
      "step": 2058
    },
    {
      "epoch": 0.565581650872133,
      "grad_norm": 0.3977581560611725,
      "learning_rate": 1.4357564930603272e-05,
      "loss": 0.8746,
      "step": 2059
    },
    {
      "epoch": 0.5658563384150529,
      "grad_norm": 0.6550084948539734,
      "learning_rate": 1.435481654527965e-05,
      "loss": 0.706,
      "step": 2060
    },
    {
      "epoch": 0.5661310259579728,
      "grad_norm": 0.4677705764770508,
      "learning_rate": 1.4352068159956028e-05,
      "loss": 0.7728,
      "step": 2061
    },
    {
      "epoch": 0.5664057135008927,
      "grad_norm": 0.4332510232925415,
      "learning_rate": 1.4349319774632405e-05,
      "loss": 0.7248,
      "step": 2062
    },
    {
      "epoch": 0.5666804010438127,
      "grad_norm": 0.4319644570350647,
      "learning_rate": 1.4346571389308784e-05,
      "loss": 0.5841,
      "step": 2063
    },
    {
      "epoch": 0.5669550885867326,
      "grad_norm": 0.5591549873352051,
      "learning_rate": 1.434382300398516e-05,
      "loss": 0.8062,
      "step": 2064
    },
    {
      "epoch": 0.5672297761296525,
      "grad_norm": 0.4446990191936493,
      "learning_rate": 1.4341074618661538e-05,
      "loss": 0.6153,
      "step": 2065
    },
    {
      "epoch": 0.5675044636725725,
      "grad_norm": 0.2825070917606354,
      "learning_rate": 1.4338326233337916e-05,
      "loss": 0.6573,
      "step": 2066
    },
    {
      "epoch": 0.5677791512154924,
      "grad_norm": 0.4860974848270416,
      "learning_rate": 1.4335577848014293e-05,
      "loss": 0.8895,
      "step": 2067
    },
    {
      "epoch": 0.5680538387584123,
      "grad_norm": 0.36241135001182556,
      "learning_rate": 1.4332829462690669e-05,
      "loss": 0.7383,
      "step": 2068
    },
    {
      "epoch": 0.5683285263013322,
      "grad_norm": 0.4190361201763153,
      "learning_rate": 1.4330081077367049e-05,
      "loss": 0.8321,
      "step": 2069
    },
    {
      "epoch": 0.5686032138442522,
      "grad_norm": 0.34443703293800354,
      "learning_rate": 1.4327332692043424e-05,
      "loss": 0.865,
      "step": 2070
    },
    {
      "epoch": 0.5688779013871721,
      "grad_norm": 0.4927169978618622,
      "learning_rate": 1.4324584306719805e-05,
      "loss": 0.7817,
      "step": 2071
    },
    {
      "epoch": 0.569152588930092,
      "grad_norm": 0.4324863851070404,
      "learning_rate": 1.4321835921396182e-05,
      "loss": 0.9055,
      "step": 2072
    },
    {
      "epoch": 0.569427276473012,
      "grad_norm": 0.35150542855262756,
      "learning_rate": 1.4319087536072557e-05,
      "loss": 0.7944,
      "step": 2073
    },
    {
      "epoch": 0.5697019640159319,
      "grad_norm": 0.3808491826057434,
      "learning_rate": 1.4316339150748938e-05,
      "loss": 0.835,
      "step": 2074
    },
    {
      "epoch": 0.5699766515588518,
      "grad_norm": 0.33833688497543335,
      "learning_rate": 1.4313590765425313e-05,
      "loss": 0.8067,
      "step": 2075
    },
    {
      "epoch": 0.5702513391017717,
      "grad_norm": 0.561021625995636,
      "learning_rate": 1.431084238010169e-05,
      "loss": 1.0768,
      "step": 2076
    },
    {
      "epoch": 0.5705260266446917,
      "grad_norm": 0.47327113151550293,
      "learning_rate": 1.4308093994778069e-05,
      "loss": 0.9145,
      "step": 2077
    },
    {
      "epoch": 0.5708007141876116,
      "grad_norm": 0.3764543831348419,
      "learning_rate": 1.4305345609454446e-05,
      "loss": 0.8178,
      "step": 2078
    },
    {
      "epoch": 0.5710754017305315,
      "grad_norm": 0.5259787440299988,
      "learning_rate": 1.4302597224130826e-05,
      "loss": 0.8075,
      "step": 2079
    },
    {
      "epoch": 0.5713500892734514,
      "grad_norm": 0.37366345524787903,
      "learning_rate": 1.4299848838807201e-05,
      "loss": 0.7629,
      "step": 2080
    },
    {
      "epoch": 0.5716247768163714,
      "grad_norm": 0.49388161301612854,
      "learning_rate": 1.4297100453483578e-05,
      "loss": 0.9716,
      "step": 2081
    },
    {
      "epoch": 0.5718994643592913,
      "grad_norm": 0.43368151783943176,
      "learning_rate": 1.4294352068159957e-05,
      "loss": 0.9632,
      "step": 2082
    },
    {
      "epoch": 0.5721741519022112,
      "grad_norm": 0.38490408658981323,
      "learning_rate": 1.4291603682836334e-05,
      "loss": 0.7795,
      "step": 2083
    },
    {
      "epoch": 0.5724488394451311,
      "grad_norm": 0.4210168719291687,
      "learning_rate": 1.4288855297512711e-05,
      "loss": 0.8228,
      "step": 2084
    },
    {
      "epoch": 0.5727235269880511,
      "grad_norm": 0.36043933033943176,
      "learning_rate": 1.428610691218909e-05,
      "loss": 0.7442,
      "step": 2085
    },
    {
      "epoch": 0.572998214530971,
      "grad_norm": 0.49955853819847107,
      "learning_rate": 1.4283358526865467e-05,
      "loss": 0.7122,
      "step": 2086
    },
    {
      "epoch": 0.5732729020738909,
      "grad_norm": 0.3624882102012634,
      "learning_rate": 1.4280610141541846e-05,
      "loss": 0.5096,
      "step": 2087
    },
    {
      "epoch": 0.5735475896168108,
      "grad_norm": 0.5074044466018677,
      "learning_rate": 1.4277861756218223e-05,
      "loss": 0.8118,
      "step": 2088
    },
    {
      "epoch": 0.5738222771597308,
      "grad_norm": 0.5081160068511963,
      "learning_rate": 1.42751133708946e-05,
      "loss": 0.7722,
      "step": 2089
    },
    {
      "epoch": 0.5740969647026507,
      "grad_norm": 0.4836336076259613,
      "learning_rate": 1.4272364985570978e-05,
      "loss": 0.6922,
      "step": 2090
    },
    {
      "epoch": 0.5743716522455706,
      "grad_norm": 0.318734735250473,
      "learning_rate": 1.4269616600247355e-05,
      "loss": 0.9408,
      "step": 2091
    },
    {
      "epoch": 0.5746463397884906,
      "grad_norm": 0.46481823921203613,
      "learning_rate": 1.4266868214923732e-05,
      "loss": 0.8567,
      "step": 2092
    },
    {
      "epoch": 0.5749210273314105,
      "grad_norm": 0.4520277678966522,
      "learning_rate": 1.4264119829600111e-05,
      "loss": 0.9379,
      "step": 2093
    },
    {
      "epoch": 0.5751957148743304,
      "grad_norm": 0.3401288390159607,
      "learning_rate": 1.4261371444276488e-05,
      "loss": 0.9235,
      "step": 2094
    },
    {
      "epoch": 0.5754704024172503,
      "grad_norm": 0.42379462718963623,
      "learning_rate": 1.4258623058952867e-05,
      "loss": 0.7742,
      "step": 2095
    },
    {
      "epoch": 0.5757450899601703,
      "grad_norm": 0.6149909496307373,
      "learning_rate": 1.4255874673629244e-05,
      "loss": 0.8892,
      "step": 2096
    },
    {
      "epoch": 0.5760197775030902,
      "grad_norm": 0.4090857207775116,
      "learning_rate": 1.4253126288305621e-05,
      "loss": 0.7277,
      "step": 2097
    },
    {
      "epoch": 0.5762944650460101,
      "grad_norm": 0.37905585765838623,
      "learning_rate": 1.4250377902982e-05,
      "loss": 0.8717,
      "step": 2098
    },
    {
      "epoch": 0.57656915258893,
      "grad_norm": 0.39358314871788025,
      "learning_rate": 1.4247629517658377e-05,
      "loss": 0.7126,
      "step": 2099
    },
    {
      "epoch": 0.57684384013185,
      "grad_norm": 0.40072834491729736,
      "learning_rate": 1.4244881132334754e-05,
      "loss": 0.7617,
      "step": 2100
    },
    {
      "epoch": 0.5771185276747699,
      "grad_norm": 0.3354143798351288,
      "learning_rate": 1.4242132747011132e-05,
      "loss": 0.4261,
      "step": 2101
    },
    {
      "epoch": 0.5773932152176898,
      "grad_norm": 0.4375704228878021,
      "learning_rate": 1.423938436168751e-05,
      "loss": 0.9053,
      "step": 2102
    },
    {
      "epoch": 0.5776679027606098,
      "grad_norm": 0.45481374859809875,
      "learning_rate": 1.4236635976363888e-05,
      "loss": 0.8556,
      "step": 2103
    },
    {
      "epoch": 0.5779425903035297,
      "grad_norm": 0.4237217307090759,
      "learning_rate": 1.4233887591040265e-05,
      "loss": 1.0823,
      "step": 2104
    },
    {
      "epoch": 0.5782172778464497,
      "grad_norm": 0.48465025424957275,
      "learning_rate": 1.4231139205716642e-05,
      "loss": 0.7564,
      "step": 2105
    },
    {
      "epoch": 0.5784919653893696,
      "grad_norm": 0.3485870957374573,
      "learning_rate": 1.4228390820393021e-05,
      "loss": 0.6875,
      "step": 2106
    },
    {
      "epoch": 0.5787666529322896,
      "grad_norm": 0.3916664123535156,
      "learning_rate": 1.4225642435069398e-05,
      "loss": 0.8888,
      "step": 2107
    },
    {
      "epoch": 0.5790413404752095,
      "grad_norm": 0.48496323823928833,
      "learning_rate": 1.4222894049745775e-05,
      "loss": 0.7967,
      "step": 2108
    },
    {
      "epoch": 0.5793160280181294,
      "grad_norm": 0.3647613823413849,
      "learning_rate": 1.4220145664422154e-05,
      "loss": 0.8786,
      "step": 2109
    },
    {
      "epoch": 0.5795907155610494,
      "grad_norm": 0.4361792206764221,
      "learning_rate": 1.421739727909853e-05,
      "loss": 0.7905,
      "step": 2110
    },
    {
      "epoch": 0.5798654031039693,
      "grad_norm": 0.3605886697769165,
      "learning_rate": 1.421464889377491e-05,
      "loss": 0.7812,
      "step": 2111
    },
    {
      "epoch": 0.5801400906468892,
      "grad_norm": 0.46065816283226013,
      "learning_rate": 1.4211900508451286e-05,
      "loss": 0.8427,
      "step": 2112
    },
    {
      "epoch": 0.5804147781898091,
      "grad_norm": 0.456674188375473,
      "learning_rate": 1.4209152123127663e-05,
      "loss": 0.8429,
      "step": 2113
    },
    {
      "epoch": 0.5806894657327291,
      "grad_norm": 0.34307631850242615,
      "learning_rate": 1.4206403737804042e-05,
      "loss": 0.7924,
      "step": 2114
    },
    {
      "epoch": 0.580964153275649,
      "grad_norm": 0.3616185486316681,
      "learning_rate": 1.420365535248042e-05,
      "loss": 0.6384,
      "step": 2115
    },
    {
      "epoch": 0.5812388408185689,
      "grad_norm": 0.561634361743927,
      "learning_rate": 1.4200906967156796e-05,
      "loss": 0.7751,
      "step": 2116
    },
    {
      "epoch": 0.5815135283614888,
      "grad_norm": 0.4239600896835327,
      "learning_rate": 1.4198158581833175e-05,
      "loss": 0.6391,
      "step": 2117
    },
    {
      "epoch": 0.5817882159044088,
      "grad_norm": 0.3875822126865387,
      "learning_rate": 1.4195410196509552e-05,
      "loss": 0.6161,
      "step": 2118
    },
    {
      "epoch": 0.5820629034473287,
      "grad_norm": 0.32066917419433594,
      "learning_rate": 1.419266181118593e-05,
      "loss": 1.0607,
      "step": 2119
    },
    {
      "epoch": 0.5823375909902486,
      "grad_norm": 0.3816903531551361,
      "learning_rate": 1.4189913425862308e-05,
      "loss": 0.6893,
      "step": 2120
    },
    {
      "epoch": 0.5826122785331685,
      "grad_norm": 0.3573743999004364,
      "learning_rate": 1.4187165040538685e-05,
      "loss": 0.7865,
      "step": 2121
    },
    {
      "epoch": 0.5828869660760885,
      "grad_norm": 0.3579890727996826,
      "learning_rate": 1.4184416655215063e-05,
      "loss": 0.808,
      "step": 2122
    },
    {
      "epoch": 0.5831616536190084,
      "grad_norm": 0.3771617114543915,
      "learning_rate": 1.418166826989144e-05,
      "loss": 0.797,
      "step": 2123
    },
    {
      "epoch": 0.5834363411619283,
      "grad_norm": 0.40594515204429626,
      "learning_rate": 1.4178919884567817e-05,
      "loss": 0.7468,
      "step": 2124
    },
    {
      "epoch": 0.5837110287048483,
      "grad_norm": 0.5013450980186462,
      "learning_rate": 1.4176171499244196e-05,
      "loss": 0.9289,
      "step": 2125
    },
    {
      "epoch": 0.5839857162477682,
      "grad_norm": 0.35900038480758667,
      "learning_rate": 1.4173423113920573e-05,
      "loss": 0.8254,
      "step": 2126
    },
    {
      "epoch": 0.5842604037906881,
      "grad_norm": 0.32886648178100586,
      "learning_rate": 1.4170674728596952e-05,
      "loss": 0.8483,
      "step": 2127
    },
    {
      "epoch": 0.584535091333608,
      "grad_norm": 0.40815770626068115,
      "learning_rate": 1.4167926343273329e-05,
      "loss": 0.721,
      "step": 2128
    },
    {
      "epoch": 0.584809778876528,
      "grad_norm": 0.44482213258743286,
      "learning_rate": 1.4165177957949704e-05,
      "loss": 0.9206,
      "step": 2129
    },
    {
      "epoch": 0.5850844664194479,
      "grad_norm": 0.38966333866119385,
      "learning_rate": 1.4162429572626085e-05,
      "loss": 0.7279,
      "step": 2130
    },
    {
      "epoch": 0.5853591539623678,
      "grad_norm": 0.35033607482910156,
      "learning_rate": 1.4159681187302462e-05,
      "loss": 0.8451,
      "step": 2131
    },
    {
      "epoch": 0.5856338415052877,
      "grad_norm": 0.3533806800842285,
      "learning_rate": 1.4156932801978837e-05,
      "loss": 1.0441,
      "step": 2132
    },
    {
      "epoch": 0.5859085290482077,
      "grad_norm": 0.8065562844276428,
      "learning_rate": 1.4154184416655217e-05,
      "loss": 0.9268,
      "step": 2133
    },
    {
      "epoch": 0.5861832165911276,
      "grad_norm": 0.44246309995651245,
      "learning_rate": 1.4151436031331593e-05,
      "loss": 0.9528,
      "step": 2134
    },
    {
      "epoch": 0.5864579041340475,
      "grad_norm": 0.3435840606689453,
      "learning_rate": 1.4148687646007973e-05,
      "loss": 0.6921,
      "step": 2135
    },
    {
      "epoch": 0.5867325916769675,
      "grad_norm": 0.36764952540397644,
      "learning_rate": 1.4145939260684348e-05,
      "loss": 0.7719,
      "step": 2136
    },
    {
      "epoch": 0.5870072792198874,
      "grad_norm": 0.37460625171661377,
      "learning_rate": 1.4143190875360725e-05,
      "loss": 0.7381,
      "step": 2137
    },
    {
      "epoch": 0.5872819667628073,
      "grad_norm": 0.2999773919582367,
      "learning_rate": 1.4140442490037106e-05,
      "loss": 0.7433,
      "step": 2138
    },
    {
      "epoch": 0.5875566543057272,
      "grad_norm": 0.36898571252822876,
      "learning_rate": 1.4137694104713481e-05,
      "loss": 0.6287,
      "step": 2139
    },
    {
      "epoch": 0.5878313418486472,
      "grad_norm": 0.32348689436912537,
      "learning_rate": 1.4134945719389858e-05,
      "loss": 0.7842,
      "step": 2140
    },
    {
      "epoch": 0.5881060293915671,
      "grad_norm": 0.42699843645095825,
      "learning_rate": 1.4132197334066237e-05,
      "loss": 0.9072,
      "step": 2141
    },
    {
      "epoch": 0.588380716934487,
      "grad_norm": 0.36722034215927124,
      "learning_rate": 1.4129448948742614e-05,
      "loss": 1.0248,
      "step": 2142
    },
    {
      "epoch": 0.5886554044774069,
      "grad_norm": 0.5946866869926453,
      "learning_rate": 1.4126700563418993e-05,
      "loss": 0.8162,
      "step": 2143
    },
    {
      "epoch": 0.5889300920203269,
      "grad_norm": 0.36136388778686523,
      "learning_rate": 1.412395217809537e-05,
      "loss": 0.9899,
      "step": 2144
    },
    {
      "epoch": 0.5892047795632468,
      "grad_norm": 0.35914087295532227,
      "learning_rate": 1.4121203792771747e-05,
      "loss": 0.7111,
      "step": 2145
    },
    {
      "epoch": 0.5894794671061667,
      "grad_norm": 0.28923115134239197,
      "learning_rate": 1.4118455407448125e-05,
      "loss": 0.905,
      "step": 2146
    },
    {
      "epoch": 0.5897541546490866,
      "grad_norm": 0.4052952826023102,
      "learning_rate": 1.4115707022124502e-05,
      "loss": 0.9797,
      "step": 2147
    },
    {
      "epoch": 0.5900288421920066,
      "grad_norm": 0.42772677540779114,
      "learning_rate": 1.411295863680088e-05,
      "loss": 0.893,
      "step": 2148
    },
    {
      "epoch": 0.5903035297349265,
      "grad_norm": 0.4425853192806244,
      "learning_rate": 1.4110210251477258e-05,
      "loss": 0.9436,
      "step": 2149
    },
    {
      "epoch": 0.5905782172778464,
      "grad_norm": 0.44531765580177307,
      "learning_rate": 1.4107461866153635e-05,
      "loss": 0.8947,
      "step": 2150
    },
    {
      "epoch": 0.5908529048207664,
      "grad_norm": 0.4252610206604004,
      "learning_rate": 1.4104713480830014e-05,
      "loss": 0.6284,
      "step": 2151
    },
    {
      "epoch": 0.5911275923636863,
      "grad_norm": 0.364403635263443,
      "learning_rate": 1.4101965095506391e-05,
      "loss": 0.6788,
      "step": 2152
    },
    {
      "epoch": 0.5914022799066062,
      "grad_norm": 0.469208687543869,
      "learning_rate": 1.4099216710182768e-05,
      "loss": 1.0254,
      "step": 2153
    },
    {
      "epoch": 0.5916769674495261,
      "grad_norm": 0.6087056398391724,
      "learning_rate": 1.4096468324859147e-05,
      "loss": 0.8734,
      "step": 2154
    },
    {
      "epoch": 0.5919516549924461,
      "grad_norm": 0.4503013491630554,
      "learning_rate": 1.4093719939535524e-05,
      "loss": 0.8807,
      "step": 2155
    },
    {
      "epoch": 0.592226342535366,
      "grad_norm": 0.4258202314376831,
      "learning_rate": 1.40909715542119e-05,
      "loss": 0.9592,
      "step": 2156
    },
    {
      "epoch": 0.5925010300782859,
      "grad_norm": 0.36056211590766907,
      "learning_rate": 1.408822316888828e-05,
      "loss": 0.9481,
      "step": 2157
    },
    {
      "epoch": 0.5927757176212058,
      "grad_norm": 0.3823760151863098,
      "learning_rate": 1.4085474783564656e-05,
      "loss": 0.7086,
      "step": 2158
    },
    {
      "epoch": 0.5930504051641258,
      "grad_norm": 0.30323925614356995,
      "learning_rate": 1.4082726398241035e-05,
      "loss": 0.963,
      "step": 2159
    },
    {
      "epoch": 0.5933250927070457,
      "grad_norm": 0.436206579208374,
      "learning_rate": 1.4079978012917412e-05,
      "loss": 0.8186,
      "step": 2160
    },
    {
      "epoch": 0.5935997802499656,
      "grad_norm": 0.3810781240463257,
      "learning_rate": 1.407722962759379e-05,
      "loss": 0.7962,
      "step": 2161
    },
    {
      "epoch": 0.5938744677928856,
      "grad_norm": 0.33725935220718384,
      "learning_rate": 1.4074481242270168e-05,
      "loss": 0.8522,
      "step": 2162
    },
    {
      "epoch": 0.5941491553358055,
      "grad_norm": 0.3637582063674927,
      "learning_rate": 1.4071732856946545e-05,
      "loss": 0.7379,
      "step": 2163
    },
    {
      "epoch": 0.5944238428787254,
      "grad_norm": 0.33679524064064026,
      "learning_rate": 1.4068984471622922e-05,
      "loss": 0.9322,
      "step": 2164
    },
    {
      "epoch": 0.5946985304216453,
      "grad_norm": 0.4768944978713989,
      "learning_rate": 1.40662360862993e-05,
      "loss": 0.9874,
      "step": 2165
    },
    {
      "epoch": 0.5949732179645653,
      "grad_norm": 0.44240590929985046,
      "learning_rate": 1.4063487700975678e-05,
      "loss": 0.9616,
      "step": 2166
    },
    {
      "epoch": 0.5952479055074852,
      "grad_norm": 0.39741677045822144,
      "learning_rate": 1.4060739315652056e-05,
      "loss": 0.7351,
      "step": 2167
    },
    {
      "epoch": 0.5955225930504051,
      "grad_norm": 0.40630993247032166,
      "learning_rate": 1.4057990930328433e-05,
      "loss": 0.9938,
      "step": 2168
    },
    {
      "epoch": 0.595797280593325,
      "grad_norm": 0.432026743888855,
      "learning_rate": 1.405524254500481e-05,
      "loss": 0.7667,
      "step": 2169
    },
    {
      "epoch": 0.596071968136245,
      "grad_norm": 0.3272799849510193,
      "learning_rate": 1.405249415968119e-05,
      "loss": 0.8256,
      "step": 2170
    },
    {
      "epoch": 0.596346655679165,
      "grad_norm": 0.34998437762260437,
      "learning_rate": 1.4049745774357566e-05,
      "loss": 0.6719,
      "step": 2171
    },
    {
      "epoch": 0.5966213432220849,
      "grad_norm": 0.3943890333175659,
      "learning_rate": 1.4046997389033943e-05,
      "loss": 0.8027,
      "step": 2172
    },
    {
      "epoch": 0.5968960307650049,
      "grad_norm": 0.3647398352622986,
      "learning_rate": 1.4044249003710322e-05,
      "loss": 0.8755,
      "step": 2173
    },
    {
      "epoch": 0.5971707183079248,
      "grad_norm": 0.38467690348625183,
      "learning_rate": 1.4041500618386699e-05,
      "loss": 0.8117,
      "step": 2174
    },
    {
      "epoch": 0.5974454058508447,
      "grad_norm": 0.38361576199531555,
      "learning_rate": 1.4038752233063078e-05,
      "loss": 0.9221,
      "step": 2175
    },
    {
      "epoch": 0.5977200933937646,
      "grad_norm": 0.29895636439323425,
      "learning_rate": 1.4036003847739455e-05,
      "loss": 0.9148,
      "step": 2176
    },
    {
      "epoch": 0.5979947809366846,
      "grad_norm": 0.4352515637874603,
      "learning_rate": 1.4033255462415832e-05,
      "loss": 0.7175,
      "step": 2177
    },
    {
      "epoch": 0.5982694684796045,
      "grad_norm": 0.35748156905174255,
      "learning_rate": 1.403050707709221e-05,
      "loss": 0.8211,
      "step": 2178
    },
    {
      "epoch": 0.5985441560225244,
      "grad_norm": 0.33292150497436523,
      "learning_rate": 1.4027758691768587e-05,
      "loss": 0.6419,
      "step": 2179
    },
    {
      "epoch": 0.5988188435654443,
      "grad_norm": 0.36785373091697693,
      "learning_rate": 1.4025010306444964e-05,
      "loss": 0.8578,
      "step": 2180
    },
    {
      "epoch": 0.5990935311083643,
      "grad_norm": 0.42598143219947815,
      "learning_rate": 1.4022261921121343e-05,
      "loss": 0.8199,
      "step": 2181
    },
    {
      "epoch": 0.5993682186512842,
      "grad_norm": 0.4431101977825165,
      "learning_rate": 1.401951353579772e-05,
      "loss": 0.7937,
      "step": 2182
    },
    {
      "epoch": 0.5996429061942041,
      "grad_norm": 0.4075913727283478,
      "learning_rate": 1.4016765150474099e-05,
      "loss": 0.7269,
      "step": 2183
    },
    {
      "epoch": 0.599917593737124,
      "grad_norm": 0.4396545886993408,
      "learning_rate": 1.4014016765150476e-05,
      "loss": 0.6824,
      "step": 2184
    },
    {
      "epoch": 0.600192281280044,
      "grad_norm": 0.3920479416847229,
      "learning_rate": 1.4011268379826853e-05,
      "loss": 1.0596,
      "step": 2185
    },
    {
      "epoch": 0.6004669688229639,
      "grad_norm": 0.28966236114501953,
      "learning_rate": 1.4008519994503232e-05,
      "loss": 0.5472,
      "step": 2186
    },
    {
      "epoch": 0.6007416563658838,
      "grad_norm": 0.38918545842170715,
      "learning_rate": 1.4005771609179609e-05,
      "loss": 0.5106,
      "step": 2187
    },
    {
      "epoch": 0.6010163439088038,
      "grad_norm": 0.3176487982273102,
      "learning_rate": 1.4003023223855984e-05,
      "loss": 0.56,
      "step": 2188
    },
    {
      "epoch": 0.6012910314517237,
      "grad_norm": 0.5583871603012085,
      "learning_rate": 1.4000274838532364e-05,
      "loss": 1.068,
      "step": 2189
    },
    {
      "epoch": 0.6015657189946436,
      "grad_norm": 0.3696845769882202,
      "learning_rate": 1.3997526453208741e-05,
      "loss": 0.8233,
      "step": 2190
    },
    {
      "epoch": 0.6018404065375635,
      "grad_norm": 0.3991202414035797,
      "learning_rate": 1.399477806788512e-05,
      "loss": 0.8109,
      "step": 2191
    },
    {
      "epoch": 0.6021150940804835,
      "grad_norm": 0.4499439001083374,
      "learning_rate": 1.3992029682561497e-05,
      "loss": 0.7141,
      "step": 2192
    },
    {
      "epoch": 0.6023897816234034,
      "grad_norm": 0.4221056401729584,
      "learning_rate": 1.3989281297237873e-05,
      "loss": 0.6821,
      "step": 2193
    },
    {
      "epoch": 0.6026644691663233,
      "grad_norm": 0.42494437098503113,
      "learning_rate": 1.3986532911914253e-05,
      "loss": 0.8198,
      "step": 2194
    },
    {
      "epoch": 0.6029391567092433,
      "grad_norm": 0.4378654360771179,
      "learning_rate": 1.398378452659063e-05,
      "loss": 0.6465,
      "step": 2195
    },
    {
      "epoch": 0.6032138442521632,
      "grad_norm": 0.24139918386936188,
      "learning_rate": 1.3981036141267005e-05,
      "loss": 0.612,
      "step": 2196
    },
    {
      "epoch": 0.6034885317950831,
      "grad_norm": 0.4207821786403656,
      "learning_rate": 1.3978287755943386e-05,
      "loss": 0.776,
      "step": 2197
    },
    {
      "epoch": 0.603763219338003,
      "grad_norm": 0.382057249546051,
      "learning_rate": 1.3975539370619761e-05,
      "loss": 0.6983,
      "step": 2198
    },
    {
      "epoch": 0.604037906880923,
      "grad_norm": 0.47806239128112793,
      "learning_rate": 1.3972790985296141e-05,
      "loss": 0.8657,
      "step": 2199
    },
    {
      "epoch": 0.6043125944238429,
      "grad_norm": 0.38379839062690735,
      "learning_rate": 1.3970042599972517e-05,
      "loss": 0.7476,
      "step": 2200
    },
    {
      "epoch": 0.6045872819667628,
      "grad_norm": 0.570395290851593,
      "learning_rate": 1.3967294214648894e-05,
      "loss": 0.9714,
      "step": 2201
    },
    {
      "epoch": 0.6048619695096827,
      "grad_norm": 0.3847440779209137,
      "learning_rate": 1.3964545829325274e-05,
      "loss": 0.5534,
      "step": 2202
    },
    {
      "epoch": 0.6051366570526027,
      "grad_norm": 0.2819223403930664,
      "learning_rate": 1.396179744400165e-05,
      "loss": 0.8271,
      "step": 2203
    },
    {
      "epoch": 0.6054113445955226,
      "grad_norm": 0.4246405065059662,
      "learning_rate": 1.3959049058678027e-05,
      "loss": 1.0053,
      "step": 2204
    },
    {
      "epoch": 0.6056860321384425,
      "grad_norm": 0.31506940722465515,
      "learning_rate": 1.3956300673354405e-05,
      "loss": 0.7314,
      "step": 2205
    },
    {
      "epoch": 0.6059607196813624,
      "grad_norm": 0.28221186995506287,
      "learning_rate": 1.3953552288030782e-05,
      "loss": 0.6465,
      "step": 2206
    },
    {
      "epoch": 0.6062354072242824,
      "grad_norm": 0.43218573927879333,
      "learning_rate": 1.3950803902707161e-05,
      "loss": 1.1975,
      "step": 2207
    },
    {
      "epoch": 0.6065100947672023,
      "grad_norm": 0.4113632142543793,
      "learning_rate": 1.3948055517383538e-05,
      "loss": 0.9621,
      "step": 2208
    },
    {
      "epoch": 0.6067847823101222,
      "grad_norm": 0.44851237535476685,
      "learning_rate": 1.3945307132059915e-05,
      "loss": 0.8626,
      "step": 2209
    },
    {
      "epoch": 0.6070594698530422,
      "grad_norm": 0.29599544405937195,
      "learning_rate": 1.3942558746736294e-05,
      "loss": 0.8506,
      "step": 2210
    },
    {
      "epoch": 0.6073341573959621,
      "grad_norm": 0.3665797710418701,
      "learning_rate": 1.393981036141267e-05,
      "loss": 0.8909,
      "step": 2211
    },
    {
      "epoch": 0.607608844938882,
      "grad_norm": 0.2992560863494873,
      "learning_rate": 1.3937061976089048e-05,
      "loss": 0.855,
      "step": 2212
    },
    {
      "epoch": 0.6078835324818019,
      "grad_norm": 0.3414834439754486,
      "learning_rate": 1.3934313590765427e-05,
      "loss": 0.5937,
      "step": 2213
    },
    {
      "epoch": 0.6081582200247219,
      "grad_norm": 0.39167219400405884,
      "learning_rate": 1.3931565205441804e-05,
      "loss": 0.8694,
      "step": 2214
    },
    {
      "epoch": 0.6084329075676418,
      "grad_norm": 0.38991400599479675,
      "learning_rate": 1.3928816820118182e-05,
      "loss": 0.9507,
      "step": 2215
    },
    {
      "epoch": 0.6087075951105617,
      "grad_norm": 0.4310379922389984,
      "learning_rate": 1.392606843479456e-05,
      "loss": 0.7102,
      "step": 2216
    },
    {
      "epoch": 0.6089822826534816,
      "grad_norm": 0.54993736743927,
      "learning_rate": 1.3923320049470936e-05,
      "loss": 0.952,
      "step": 2217
    },
    {
      "epoch": 0.6092569701964016,
      "grad_norm": 0.4249913990497589,
      "learning_rate": 1.3920571664147315e-05,
      "loss": 1.0169,
      "step": 2218
    },
    {
      "epoch": 0.6095316577393215,
      "grad_norm": 0.3306918144226074,
      "learning_rate": 1.3917823278823692e-05,
      "loss": 0.9337,
      "step": 2219
    },
    {
      "epoch": 0.6098063452822414,
      "grad_norm": 0.4128495156764984,
      "learning_rate": 1.3915074893500069e-05,
      "loss": 0.6008,
      "step": 2220
    },
    {
      "epoch": 0.6100810328251614,
      "grad_norm": 0.36853575706481934,
      "learning_rate": 1.3912326508176448e-05,
      "loss": 0.8161,
      "step": 2221
    },
    {
      "epoch": 0.6103557203680813,
      "grad_norm": 0.3862062096595764,
      "learning_rate": 1.3909578122852825e-05,
      "loss": 0.721,
      "step": 2222
    },
    {
      "epoch": 0.6106304079110012,
      "grad_norm": 0.48273271322250366,
      "learning_rate": 1.3906829737529202e-05,
      "loss": 0.6705,
      "step": 2223
    },
    {
      "epoch": 0.6109050954539211,
      "grad_norm": 0.6325643658638,
      "learning_rate": 1.390408135220558e-05,
      "loss": 0.745,
      "step": 2224
    },
    {
      "epoch": 0.6111797829968411,
      "grad_norm": 0.3720119297504425,
      "learning_rate": 1.3901332966881958e-05,
      "loss": 1.0371,
      "step": 2225
    },
    {
      "epoch": 0.611454470539761,
      "grad_norm": 0.350078284740448,
      "learning_rate": 1.3898584581558336e-05,
      "loss": 0.8306,
      "step": 2226
    },
    {
      "epoch": 0.6117291580826809,
      "grad_norm": 0.3963256776332855,
      "learning_rate": 1.3895836196234713e-05,
      "loss": 0.8172,
      "step": 2227
    },
    {
      "epoch": 0.6120038456256008,
      "grad_norm": 0.47065746784210205,
      "learning_rate": 1.389308781091109e-05,
      "loss": 0.7694,
      "step": 2228
    },
    {
      "epoch": 0.6122785331685208,
      "grad_norm": 0.4083212614059448,
      "learning_rate": 1.3890339425587469e-05,
      "loss": 0.7459,
      "step": 2229
    },
    {
      "epoch": 0.6125532207114407,
      "grad_norm": 0.2760429084300995,
      "learning_rate": 1.3887591040263846e-05,
      "loss": 0.6444,
      "step": 2230
    },
    {
      "epoch": 0.6128279082543606,
      "grad_norm": 0.3110139071941376,
      "learning_rate": 1.3884842654940223e-05,
      "loss": 0.8856,
      "step": 2231
    },
    {
      "epoch": 0.6131025957972805,
      "grad_norm": 0.4459973871707916,
      "learning_rate": 1.3882094269616602e-05,
      "loss": 0.9182,
      "step": 2232
    },
    {
      "epoch": 0.6133772833402005,
      "grad_norm": 0.6236221194267273,
      "learning_rate": 1.3879345884292979e-05,
      "loss": 0.757,
      "step": 2233
    },
    {
      "epoch": 0.6136519708831204,
      "grad_norm": 0.4012565016746521,
      "learning_rate": 1.3876597498969357e-05,
      "loss": 0.8648,
      "step": 2234
    },
    {
      "epoch": 0.6139266584260403,
      "grad_norm": 0.48627063632011414,
      "learning_rate": 1.3873849113645735e-05,
      "loss": 0.8117,
      "step": 2235
    },
    {
      "epoch": 0.6142013459689603,
      "grad_norm": 0.39827895164489746,
      "learning_rate": 1.3871100728322112e-05,
      "loss": 1.1419,
      "step": 2236
    },
    {
      "epoch": 0.6144760335118802,
      "grad_norm": 0.4371322989463806,
      "learning_rate": 1.386835234299849e-05,
      "loss": 0.7731,
      "step": 2237
    },
    {
      "epoch": 0.6147507210548002,
      "grad_norm": 0.44775092601776123,
      "learning_rate": 1.3865603957674867e-05,
      "loss": 0.8301,
      "step": 2238
    },
    {
      "epoch": 0.6150254085977201,
      "grad_norm": 0.35511699318885803,
      "learning_rate": 1.3862855572351244e-05,
      "loss": 0.8961,
      "step": 2239
    },
    {
      "epoch": 0.6153000961406401,
      "grad_norm": 0.337271511554718,
      "learning_rate": 1.3860107187027623e-05,
      "loss": 0.7222,
      "step": 2240
    },
    {
      "epoch": 0.61557478368356,
      "grad_norm": 0.3742680847644806,
      "learning_rate": 1.3857358801704e-05,
      "loss": 0.6412,
      "step": 2241
    },
    {
      "epoch": 0.6158494712264799,
      "grad_norm": 0.5092702507972717,
      "learning_rate": 1.3854610416380379e-05,
      "loss": 0.831,
      "step": 2242
    },
    {
      "epoch": 0.6161241587693999,
      "grad_norm": 0.47734442353248596,
      "learning_rate": 1.3851862031056756e-05,
      "loss": 0.8522,
      "step": 2243
    },
    {
      "epoch": 0.6163988463123198,
      "grad_norm": 0.4336296021938324,
      "learning_rate": 1.3849113645733133e-05,
      "loss": 0.6721,
      "step": 2244
    },
    {
      "epoch": 0.6166735338552397,
      "grad_norm": 0.37327468395233154,
      "learning_rate": 1.3846365260409512e-05,
      "loss": 0.6185,
      "step": 2245
    },
    {
      "epoch": 0.6169482213981596,
      "grad_norm": 0.409164696931839,
      "learning_rate": 1.3843616875085889e-05,
      "loss": 0.7484,
      "step": 2246
    },
    {
      "epoch": 0.6172229089410796,
      "grad_norm": 0.36110544204711914,
      "learning_rate": 1.3840868489762266e-05,
      "loss": 0.4986,
      "step": 2247
    },
    {
      "epoch": 0.6174975964839995,
      "grad_norm": 0.5063424110412598,
      "learning_rate": 1.3838120104438644e-05,
      "loss": 0.8026,
      "step": 2248
    },
    {
      "epoch": 0.6177722840269194,
      "grad_norm": 0.49934881925582886,
      "learning_rate": 1.3835371719115021e-05,
      "loss": 0.6865,
      "step": 2249
    },
    {
      "epoch": 0.6180469715698393,
      "grad_norm": 0.4297367334365845,
      "learning_rate": 1.38326233337914e-05,
      "loss": 0.8025,
      "step": 2250
    },
    {
      "epoch": 0.6183216591127593,
      "grad_norm": 0.29868102073669434,
      "learning_rate": 1.3829874948467777e-05,
      "loss": 0.651,
      "step": 2251
    },
    {
      "epoch": 0.6185963466556792,
      "grad_norm": 0.453419953584671,
      "learning_rate": 1.3827126563144152e-05,
      "loss": 0.7661,
      "step": 2252
    },
    {
      "epoch": 0.6188710341985991,
      "grad_norm": 0.4056752026081085,
      "learning_rate": 1.3824378177820533e-05,
      "loss": 0.7224,
      "step": 2253
    },
    {
      "epoch": 0.619145721741519,
      "grad_norm": 0.36368775367736816,
      "learning_rate": 1.382162979249691e-05,
      "loss": 0.7624,
      "step": 2254
    },
    {
      "epoch": 0.619420409284439,
      "grad_norm": 0.33160287141799927,
      "learning_rate": 1.3818881407173285e-05,
      "loss": 0.8753,
      "step": 2255
    },
    {
      "epoch": 0.6196950968273589,
      "grad_norm": 0.44637808203697205,
      "learning_rate": 1.3816133021849666e-05,
      "loss": 0.9,
      "step": 2256
    },
    {
      "epoch": 0.6199697843702788,
      "grad_norm": 0.39974772930145264,
      "learning_rate": 1.381338463652604e-05,
      "loss": 0.737,
      "step": 2257
    },
    {
      "epoch": 0.6202444719131988,
      "grad_norm": 0.3253493905067444,
      "learning_rate": 1.3810636251202421e-05,
      "loss": 0.7201,
      "step": 2258
    },
    {
      "epoch": 0.6205191594561187,
      "grad_norm": 0.5724297761917114,
      "learning_rate": 1.3807887865878797e-05,
      "loss": 0.8595,
      "step": 2259
    },
    {
      "epoch": 0.6207938469990386,
      "grad_norm": 0.43648335337638855,
      "learning_rate": 1.3805139480555174e-05,
      "loss": 0.7673,
      "step": 2260
    },
    {
      "epoch": 0.6210685345419585,
      "grad_norm": 0.4484272301197052,
      "learning_rate": 1.3802391095231554e-05,
      "loss": 0.7837,
      "step": 2261
    },
    {
      "epoch": 0.6213432220848785,
      "grad_norm": 0.3884648382663727,
      "learning_rate": 1.379964270990793e-05,
      "loss": 0.7985,
      "step": 2262
    },
    {
      "epoch": 0.6216179096277984,
      "grad_norm": 0.5114205479621887,
      "learning_rate": 1.3796894324584306e-05,
      "loss": 0.8146,
      "step": 2263
    },
    {
      "epoch": 0.6218925971707183,
      "grad_norm": 0.3830573260784149,
      "learning_rate": 1.3794145939260685e-05,
      "loss": 0.8655,
      "step": 2264
    },
    {
      "epoch": 0.6221672847136382,
      "grad_norm": 0.38257214426994324,
      "learning_rate": 1.3791397553937062e-05,
      "loss": 0.9346,
      "step": 2265
    },
    {
      "epoch": 0.6224419722565582,
      "grad_norm": 0.6161715984344482,
      "learning_rate": 1.378864916861344e-05,
      "loss": 0.873,
      "step": 2266
    },
    {
      "epoch": 0.6227166597994781,
      "grad_norm": 0.3538625240325928,
      "learning_rate": 1.3785900783289818e-05,
      "loss": 0.4874,
      "step": 2267
    },
    {
      "epoch": 0.622991347342398,
      "grad_norm": 0.3451964259147644,
      "learning_rate": 1.3783152397966195e-05,
      "loss": 1.1433,
      "step": 2268
    },
    {
      "epoch": 0.623266034885318,
      "grad_norm": 0.34519919753074646,
      "learning_rate": 1.3780404012642574e-05,
      "loss": 0.8024,
      "step": 2269
    },
    {
      "epoch": 0.6235407224282379,
      "grad_norm": 0.37479686737060547,
      "learning_rate": 1.377765562731895e-05,
      "loss": 1.0057,
      "step": 2270
    },
    {
      "epoch": 0.6238154099711578,
      "grad_norm": 0.3238420784473419,
      "learning_rate": 1.3774907241995328e-05,
      "loss": 0.6149,
      "step": 2271
    },
    {
      "epoch": 0.6240900975140777,
      "grad_norm": 0.3647848069667816,
      "learning_rate": 1.3772158856671706e-05,
      "loss": 0.8993,
      "step": 2272
    },
    {
      "epoch": 0.6243647850569977,
      "grad_norm": 0.5940234661102295,
      "learning_rate": 1.3769410471348083e-05,
      "loss": 0.679,
      "step": 2273
    },
    {
      "epoch": 0.6246394725999176,
      "grad_norm": 0.524986743927002,
      "learning_rate": 1.3766662086024462e-05,
      "loss": 0.8789,
      "step": 2274
    },
    {
      "epoch": 0.6249141601428375,
      "grad_norm": 0.45124056935310364,
      "learning_rate": 1.3763913700700839e-05,
      "loss": 0.8139,
      "step": 2275
    },
    {
      "epoch": 0.6251888476857574,
      "grad_norm": 0.5034171938896179,
      "learning_rate": 1.3761165315377216e-05,
      "loss": 0.9029,
      "step": 2276
    },
    {
      "epoch": 0.6254635352286774,
      "grad_norm": 0.5189480185508728,
      "learning_rate": 1.3758416930053595e-05,
      "loss": 0.8809,
      "step": 2277
    },
    {
      "epoch": 0.6257382227715973,
      "grad_norm": 0.415792852640152,
      "learning_rate": 1.3755668544729972e-05,
      "loss": 0.7027,
      "step": 2278
    },
    {
      "epoch": 0.6260129103145172,
      "grad_norm": 0.3805093467235565,
      "learning_rate": 1.3752920159406349e-05,
      "loss": 1.1592,
      "step": 2279
    },
    {
      "epoch": 0.6262875978574372,
      "grad_norm": 0.4954253137111664,
      "learning_rate": 1.3750171774082728e-05,
      "loss": 0.7689,
      "step": 2280
    },
    {
      "epoch": 0.6265622854003571,
      "grad_norm": 0.4684828817844391,
      "learning_rate": 1.3747423388759105e-05,
      "loss": 0.9652,
      "step": 2281
    },
    {
      "epoch": 0.626836972943277,
      "grad_norm": 0.5203549265861511,
      "learning_rate": 1.3744675003435483e-05,
      "loss": 0.7689,
      "step": 2282
    },
    {
      "epoch": 0.6271116604861969,
      "grad_norm": 0.4571121037006378,
      "learning_rate": 1.374192661811186e-05,
      "loss": 0.8916,
      "step": 2283
    },
    {
      "epoch": 0.6273863480291169,
      "grad_norm": 0.3533395230770111,
      "learning_rate": 1.3739178232788237e-05,
      "loss": 0.5334,
      "step": 2284
    },
    {
      "epoch": 0.6276610355720368,
      "grad_norm": 0.4418266713619232,
      "learning_rate": 1.3736429847464616e-05,
      "loss": 0.7319,
      "step": 2285
    },
    {
      "epoch": 0.6279357231149567,
      "grad_norm": 0.4129978120326996,
      "learning_rate": 1.3733681462140993e-05,
      "loss": 0.7487,
      "step": 2286
    },
    {
      "epoch": 0.6282104106578766,
      "grad_norm": 0.4199274480342865,
      "learning_rate": 1.373093307681737e-05,
      "loss": 0.9079,
      "step": 2287
    },
    {
      "epoch": 0.6284850982007966,
      "grad_norm": 0.2950742840766907,
      "learning_rate": 1.3728184691493749e-05,
      "loss": 0.6509,
      "step": 2288
    },
    {
      "epoch": 0.6287597857437165,
      "grad_norm": 0.3860020041465759,
      "learning_rate": 1.3725436306170126e-05,
      "loss": 0.764,
      "step": 2289
    },
    {
      "epoch": 0.6290344732866364,
      "grad_norm": 0.5009716749191284,
      "learning_rate": 1.3722687920846505e-05,
      "loss": 0.8852,
      "step": 2290
    },
    {
      "epoch": 0.6293091608295563,
      "grad_norm": 0.4828962981700897,
      "learning_rate": 1.3719939535522882e-05,
      "loss": 0.9201,
      "step": 2291
    },
    {
      "epoch": 0.6295838483724763,
      "grad_norm": 0.3240280747413635,
      "learning_rate": 1.3717191150199259e-05,
      "loss": 0.8004,
      "step": 2292
    },
    {
      "epoch": 0.6298585359153962,
      "grad_norm": 0.3075036406517029,
      "learning_rate": 1.3714442764875637e-05,
      "loss": 0.4901,
      "step": 2293
    },
    {
      "epoch": 0.6301332234583161,
      "grad_norm": 0.4263114631175995,
      "learning_rate": 1.3711694379552014e-05,
      "loss": 0.8376,
      "step": 2294
    },
    {
      "epoch": 0.630407911001236,
      "grad_norm": 0.37726637721061707,
      "learning_rate": 1.3708945994228391e-05,
      "loss": 0.7498,
      "step": 2295
    },
    {
      "epoch": 0.630682598544156,
      "grad_norm": 0.37326154112815857,
      "learning_rate": 1.370619760890477e-05,
      "loss": 0.8213,
      "step": 2296
    },
    {
      "epoch": 0.6309572860870759,
      "grad_norm": 0.3755851089954376,
      "learning_rate": 1.3703449223581147e-05,
      "loss": 0.5593,
      "step": 2297
    },
    {
      "epoch": 0.6312319736299958,
      "grad_norm": 0.36328887939453125,
      "learning_rate": 1.3700700838257526e-05,
      "loss": 0.8223,
      "step": 2298
    },
    {
      "epoch": 0.6315066611729158,
      "grad_norm": 0.4204294979572296,
      "learning_rate": 1.3697952452933903e-05,
      "loss": 0.6723,
      "step": 2299
    },
    {
      "epoch": 0.6317813487158357,
      "grad_norm": 0.4815131723880768,
      "learning_rate": 1.369520406761028e-05,
      "loss": 0.8125,
      "step": 2300
    },
    {
      "epoch": 0.6320560362587556,
      "grad_norm": 0.4824061095714569,
      "learning_rate": 1.3692455682286659e-05,
      "loss": 0.8566,
      "step": 2301
    },
    {
      "epoch": 0.6323307238016755,
      "grad_norm": 0.338011771440506,
      "learning_rate": 1.3689707296963036e-05,
      "loss": 0.7624,
      "step": 2302
    },
    {
      "epoch": 0.6326054113445955,
      "grad_norm": 0.387234628200531,
      "learning_rate": 1.3686958911639413e-05,
      "loss": 0.7041,
      "step": 2303
    },
    {
      "epoch": 0.6328800988875154,
      "grad_norm": 0.40714478492736816,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 1.057,
      "step": 2304
    },
    {
      "epoch": 0.6331547864304354,
      "grad_norm": 0.34651824831962585,
      "learning_rate": 1.3681462140992168e-05,
      "loss": 1.0996,
      "step": 2305
    },
    {
      "epoch": 0.6334294739733554,
      "grad_norm": 0.4064626097679138,
      "learning_rate": 1.3678713755668547e-05,
      "loss": 0.814,
      "step": 2306
    },
    {
      "epoch": 0.6337041615162753,
      "grad_norm": 0.4752848446369171,
      "learning_rate": 1.3675965370344924e-05,
      "loss": 0.8994,
      "step": 2307
    },
    {
      "epoch": 0.6339788490591952,
      "grad_norm": 0.34376758337020874,
      "learning_rate": 1.3673216985021301e-05,
      "loss": 0.8255,
      "step": 2308
    },
    {
      "epoch": 0.6342535366021151,
      "grad_norm": 0.3091365396976471,
      "learning_rate": 1.367046859969768e-05,
      "loss": 0.5592,
      "step": 2309
    },
    {
      "epoch": 0.6345282241450351,
      "grad_norm": 0.39238566160202026,
      "learning_rate": 1.3667720214374057e-05,
      "loss": 0.9537,
      "step": 2310
    },
    {
      "epoch": 0.634802911687955,
      "grad_norm": 0.40737366676330566,
      "learning_rate": 1.3664971829050432e-05,
      "loss": 0.9172,
      "step": 2311
    },
    {
      "epoch": 0.6350775992308749,
      "grad_norm": 0.5081478953361511,
      "learning_rate": 1.3662223443726813e-05,
      "loss": 0.7578,
      "step": 2312
    },
    {
      "epoch": 0.6353522867737948,
      "grad_norm": 0.3957054913043976,
      "learning_rate": 1.365947505840319e-05,
      "loss": 0.4963,
      "step": 2313
    },
    {
      "epoch": 0.6356269743167148,
      "grad_norm": 0.5439357757568359,
      "learning_rate": 1.3656726673079568e-05,
      "loss": 0.9641,
      "step": 2314
    },
    {
      "epoch": 0.6359016618596347,
      "grad_norm": 0.36711081862449646,
      "learning_rate": 1.3653978287755945e-05,
      "loss": 0.6268,
      "step": 2315
    },
    {
      "epoch": 0.6361763494025546,
      "grad_norm": 0.5085931420326233,
      "learning_rate": 1.365122990243232e-05,
      "loss": 0.8599,
      "step": 2316
    },
    {
      "epoch": 0.6364510369454746,
      "grad_norm": 0.44270187616348267,
      "learning_rate": 1.3648481517108701e-05,
      "loss": 0.6956,
      "step": 2317
    },
    {
      "epoch": 0.6367257244883945,
      "grad_norm": 0.4419419765472412,
      "learning_rate": 1.3645733131785076e-05,
      "loss": 0.8308,
      "step": 2318
    },
    {
      "epoch": 0.6370004120313144,
      "grad_norm": 0.42054322361946106,
      "learning_rate": 1.3642984746461453e-05,
      "loss": 0.7269,
      "step": 2319
    },
    {
      "epoch": 0.6372750995742343,
      "grad_norm": 0.5442548394203186,
      "learning_rate": 1.3640236361137834e-05,
      "loss": 0.8352,
      "step": 2320
    },
    {
      "epoch": 0.6375497871171543,
      "grad_norm": 0.37819376587867737,
      "learning_rate": 1.3637487975814209e-05,
      "loss": 0.962,
      "step": 2321
    },
    {
      "epoch": 0.6378244746600742,
      "grad_norm": 0.37310898303985596,
      "learning_rate": 1.363473959049059e-05,
      "loss": 0.6592,
      "step": 2322
    },
    {
      "epoch": 0.6380991622029941,
      "grad_norm": 0.40460261702537537,
      "learning_rate": 1.3631991205166965e-05,
      "loss": 0.6991,
      "step": 2323
    },
    {
      "epoch": 0.638373849745914,
      "grad_norm": 0.4447297155857086,
      "learning_rate": 1.3629242819843342e-05,
      "loss": 0.9883,
      "step": 2324
    },
    {
      "epoch": 0.638648537288834,
      "grad_norm": 0.35064756870269775,
      "learning_rate": 1.362649443451972e-05,
      "loss": 1.0539,
      "step": 2325
    },
    {
      "epoch": 0.6389232248317539,
      "grad_norm": 0.36741676926612854,
      "learning_rate": 1.3623746049196098e-05,
      "loss": 0.6554,
      "step": 2326
    },
    {
      "epoch": 0.6391979123746738,
      "grad_norm": 0.34420129656791687,
      "learning_rate": 1.3620997663872475e-05,
      "loss": 0.9907,
      "step": 2327
    },
    {
      "epoch": 0.6394725999175938,
      "grad_norm": 0.36849308013916016,
      "learning_rate": 1.3618249278548853e-05,
      "loss": 0.6216,
      "step": 2328
    },
    {
      "epoch": 0.6397472874605137,
      "grad_norm": 0.505434513092041,
      "learning_rate": 1.361550089322523e-05,
      "loss": 0.743,
      "step": 2329
    },
    {
      "epoch": 0.6400219750034336,
      "grad_norm": 0.3957449197769165,
      "learning_rate": 1.3612752507901609e-05,
      "loss": 0.7037,
      "step": 2330
    },
    {
      "epoch": 0.6402966625463535,
      "grad_norm": 0.4282202422618866,
      "learning_rate": 1.3610004122577986e-05,
      "loss": 0.9809,
      "step": 2331
    },
    {
      "epoch": 0.6405713500892735,
      "grad_norm": 0.39539432525634766,
      "learning_rate": 1.3607255737254363e-05,
      "loss": 0.6102,
      "step": 2332
    },
    {
      "epoch": 0.6408460376321934,
      "grad_norm": 0.428027480840683,
      "learning_rate": 1.3604507351930742e-05,
      "loss": 0.7598,
      "step": 2333
    },
    {
      "epoch": 0.6411207251751133,
      "grad_norm": 0.5774027705192566,
      "learning_rate": 1.3601758966607119e-05,
      "loss": 0.7645,
      "step": 2334
    },
    {
      "epoch": 0.6413954127180332,
      "grad_norm": 0.41361406445503235,
      "learning_rate": 1.3599010581283496e-05,
      "loss": 1.0251,
      "step": 2335
    },
    {
      "epoch": 0.6416701002609532,
      "grad_norm": 0.5250378847122192,
      "learning_rate": 1.3596262195959875e-05,
      "loss": 0.5227,
      "step": 2336
    },
    {
      "epoch": 0.6419447878038731,
      "grad_norm": 0.4537384510040283,
      "learning_rate": 1.3593513810636252e-05,
      "loss": 0.6929,
      "step": 2337
    },
    {
      "epoch": 0.642219475346793,
      "grad_norm": 0.40106964111328125,
      "learning_rate": 1.359076542531263e-05,
      "loss": 0.8252,
      "step": 2338
    },
    {
      "epoch": 0.642494162889713,
      "grad_norm": 0.3440161347389221,
      "learning_rate": 1.3588017039989007e-05,
      "loss": 0.8399,
      "step": 2339
    },
    {
      "epoch": 0.6427688504326329,
      "grad_norm": 0.36020800471305847,
      "learning_rate": 1.3585268654665384e-05,
      "loss": 0.9125,
      "step": 2340
    },
    {
      "epoch": 0.6430435379755528,
      "grad_norm": 0.3849390745162964,
      "learning_rate": 1.3582520269341763e-05,
      "loss": 0.7285,
      "step": 2341
    },
    {
      "epoch": 0.6433182255184727,
      "grad_norm": 0.36177510023117065,
      "learning_rate": 1.357977188401814e-05,
      "loss": 0.9313,
      "step": 2342
    },
    {
      "epoch": 0.6435929130613927,
      "grad_norm": 0.44168806076049805,
      "learning_rate": 1.3577023498694517e-05,
      "loss": 0.7355,
      "step": 2343
    },
    {
      "epoch": 0.6438676006043126,
      "grad_norm": 0.4890860319137573,
      "learning_rate": 1.3574275113370896e-05,
      "loss": 0.7264,
      "step": 2344
    },
    {
      "epoch": 0.6441422881472325,
      "grad_norm": 0.46768996119499207,
      "learning_rate": 1.3571526728047273e-05,
      "loss": 0.8319,
      "step": 2345
    },
    {
      "epoch": 0.6444169756901524,
      "grad_norm": 0.4128832221031189,
      "learning_rate": 1.3568778342723652e-05,
      "loss": 0.9242,
      "step": 2346
    },
    {
      "epoch": 0.6446916632330724,
      "grad_norm": 0.44847244024276733,
      "learning_rate": 1.3566029957400029e-05,
      "loss": 0.8719,
      "step": 2347
    },
    {
      "epoch": 0.6449663507759923,
      "grad_norm": 0.3058076500892639,
      "learning_rate": 1.3563281572076406e-05,
      "loss": 0.9357,
      "step": 2348
    },
    {
      "epoch": 0.6452410383189122,
      "grad_norm": 0.3974904716014862,
      "learning_rate": 1.3560533186752784e-05,
      "loss": 0.8108,
      "step": 2349
    },
    {
      "epoch": 0.6455157258618321,
      "grad_norm": 0.49994638562202454,
      "learning_rate": 1.3557784801429161e-05,
      "loss": 0.7764,
      "step": 2350
    },
    {
      "epoch": 0.6457904134047521,
      "grad_norm": 0.3114233911037445,
      "learning_rate": 1.3555036416105538e-05,
      "loss": 0.6319,
      "step": 2351
    },
    {
      "epoch": 0.646065100947672,
      "grad_norm": 0.5249606966972351,
      "learning_rate": 1.3552288030781917e-05,
      "loss": 0.7977,
      "step": 2352
    },
    {
      "epoch": 0.6463397884905919,
      "grad_norm": 0.5561675429344177,
      "learning_rate": 1.3549539645458294e-05,
      "loss": 0.7691,
      "step": 2353
    },
    {
      "epoch": 0.6466144760335119,
      "grad_norm": 0.3384697735309601,
      "learning_rate": 1.3546791260134673e-05,
      "loss": 0.6627,
      "step": 2354
    },
    {
      "epoch": 0.6468891635764318,
      "grad_norm": 0.43513116240501404,
      "learning_rate": 1.354404287481105e-05,
      "loss": 0.7085,
      "step": 2355
    },
    {
      "epoch": 0.6471638511193517,
      "grad_norm": 0.5474324822425842,
      "learning_rate": 1.3541294489487427e-05,
      "loss": 0.7825,
      "step": 2356
    },
    {
      "epoch": 0.6474385386622716,
      "grad_norm": 0.6291898488998413,
      "learning_rate": 1.3538546104163806e-05,
      "loss": 0.8113,
      "step": 2357
    },
    {
      "epoch": 0.6477132262051916,
      "grad_norm": 0.45073139667510986,
      "learning_rate": 1.3535797718840183e-05,
      "loss": 0.8469,
      "step": 2358
    },
    {
      "epoch": 0.6479879137481115,
      "grad_norm": 0.43603816628456116,
      "learning_rate": 1.353304933351656e-05,
      "loss": 0.9508,
      "step": 2359
    },
    {
      "epoch": 0.6482626012910314,
      "grad_norm": 0.4450383186340332,
      "learning_rate": 1.3530300948192938e-05,
      "loss": 0.7209,
      "step": 2360
    },
    {
      "epoch": 0.6485372888339513,
      "grad_norm": 0.25476881861686707,
      "learning_rate": 1.3527552562869315e-05,
      "loss": 0.5458,
      "step": 2361
    },
    {
      "epoch": 0.6488119763768713,
      "grad_norm": 0.38038942217826843,
      "learning_rate": 1.3524804177545694e-05,
      "loss": 1.0486,
      "step": 2362
    },
    {
      "epoch": 0.6490866639197912,
      "grad_norm": 0.3837694525718689,
      "learning_rate": 1.3522055792222071e-05,
      "loss": 0.7924,
      "step": 2363
    },
    {
      "epoch": 0.6493613514627111,
      "grad_norm": 0.5338654518127441,
      "learning_rate": 1.3519307406898448e-05,
      "loss": 0.911,
      "step": 2364
    },
    {
      "epoch": 0.649636039005631,
      "grad_norm": 0.5153733491897583,
      "learning_rate": 1.3516559021574827e-05,
      "loss": 0.9288,
      "step": 2365
    },
    {
      "epoch": 0.649910726548551,
      "grad_norm": 0.30804675817489624,
      "learning_rate": 1.3513810636251204e-05,
      "loss": 0.7552,
      "step": 2366
    },
    {
      "epoch": 0.6501854140914709,
      "grad_norm": 0.4805564880371094,
      "learning_rate": 1.3511062250927581e-05,
      "loss": 0.8605,
      "step": 2367
    },
    {
      "epoch": 0.6504601016343908,
      "grad_norm": 0.4430308938026428,
      "learning_rate": 1.350831386560396e-05,
      "loss": 0.7643,
      "step": 2368
    },
    {
      "epoch": 0.6507347891773108,
      "grad_norm": 0.41115009784698486,
      "learning_rate": 1.3505565480280337e-05,
      "loss": 0.6483,
      "step": 2369
    },
    {
      "epoch": 0.6510094767202307,
      "grad_norm": 0.4013855755329132,
      "learning_rate": 1.3502817094956715e-05,
      "loss": 1.0813,
      "step": 2370
    },
    {
      "epoch": 0.6512841642631506,
      "grad_norm": 0.5120947360992432,
      "learning_rate": 1.3500068709633092e-05,
      "loss": 0.6503,
      "step": 2371
    },
    {
      "epoch": 0.6515588518060706,
      "grad_norm": 0.4197784960269928,
      "learning_rate": 1.349732032430947e-05,
      "loss": 0.9167,
      "step": 2372
    },
    {
      "epoch": 0.6518335393489906,
      "grad_norm": 0.3481389582157135,
      "learning_rate": 1.3494571938985848e-05,
      "loss": 0.6268,
      "step": 2373
    },
    {
      "epoch": 0.6521082268919105,
      "grad_norm": 0.384831041097641,
      "learning_rate": 1.3491823553662225e-05,
      "loss": 0.9106,
      "step": 2374
    },
    {
      "epoch": 0.6523829144348304,
      "grad_norm": 0.32769128680229187,
      "learning_rate": 1.34890751683386e-05,
      "loss": 0.449,
      "step": 2375
    },
    {
      "epoch": 0.6526576019777504,
      "grad_norm": 0.5006611347198486,
      "learning_rate": 1.3486326783014981e-05,
      "loss": 0.7005,
      "step": 2376
    },
    {
      "epoch": 0.6529322895206703,
      "grad_norm": 0.41752004623413086,
      "learning_rate": 1.3483578397691356e-05,
      "loss": 0.887,
      "step": 2377
    },
    {
      "epoch": 0.6532069770635902,
      "grad_norm": 0.3809186518192291,
      "learning_rate": 1.3480830012367737e-05,
      "loss": 1.0321,
      "step": 2378
    },
    {
      "epoch": 0.6534816646065101,
      "grad_norm": 0.5124394297599792,
      "learning_rate": 1.3478081627044114e-05,
      "loss": 0.7343,
      "step": 2379
    },
    {
      "epoch": 0.6537563521494301,
      "grad_norm": 0.3689250349998474,
      "learning_rate": 1.3475333241720489e-05,
      "loss": 0.5819,
      "step": 2380
    },
    {
      "epoch": 0.65403103969235,
      "grad_norm": 0.39527514576911926,
      "learning_rate": 1.347258485639687e-05,
      "loss": 0.8487,
      "step": 2381
    },
    {
      "epoch": 0.6543057272352699,
      "grad_norm": 0.41065335273742676,
      "learning_rate": 1.3469836471073245e-05,
      "loss": 0.9179,
      "step": 2382
    },
    {
      "epoch": 0.6545804147781898,
      "grad_norm": 0.3599492013454437,
      "learning_rate": 1.3467088085749622e-05,
      "loss": 1.0236,
      "step": 2383
    },
    {
      "epoch": 0.6548551023211098,
      "grad_norm": 0.4119132161140442,
      "learning_rate": 1.3464339700426e-05,
      "loss": 0.9187,
      "step": 2384
    },
    {
      "epoch": 0.6551297898640297,
      "grad_norm": 0.4809182286262512,
      "learning_rate": 1.3461591315102377e-05,
      "loss": 0.8876,
      "step": 2385
    },
    {
      "epoch": 0.6554044774069496,
      "grad_norm": 0.4022350311279297,
      "learning_rate": 1.3458842929778758e-05,
      "loss": 0.8559,
      "step": 2386
    },
    {
      "epoch": 0.6556791649498696,
      "grad_norm": 0.43583348393440247,
      "learning_rate": 1.3456094544455133e-05,
      "loss": 0.7606,
      "step": 2387
    },
    {
      "epoch": 0.6559538524927895,
      "grad_norm": 0.5429250001907349,
      "learning_rate": 1.345334615913151e-05,
      "loss": 0.8148,
      "step": 2388
    },
    {
      "epoch": 0.6562285400357094,
      "grad_norm": 0.48866623640060425,
      "learning_rate": 1.3450597773807889e-05,
      "loss": 0.7373,
      "step": 2389
    },
    {
      "epoch": 0.6565032275786293,
      "grad_norm": 0.4396297037601471,
      "learning_rate": 1.3447849388484266e-05,
      "loss": 0.8765,
      "step": 2390
    },
    {
      "epoch": 0.6567779151215493,
      "grad_norm": 0.4093182682991028,
      "learning_rate": 1.3445101003160643e-05,
      "loss": 0.8943,
      "step": 2391
    },
    {
      "epoch": 0.6570526026644692,
      "grad_norm": 0.500918984413147,
      "learning_rate": 1.3442352617837022e-05,
      "loss": 0.9697,
      "step": 2392
    },
    {
      "epoch": 0.6573272902073891,
      "grad_norm": 0.6044387817382812,
      "learning_rate": 1.3439604232513399e-05,
      "loss": 0.5608,
      "step": 2393
    },
    {
      "epoch": 0.657601977750309,
      "grad_norm": 0.4923231899738312,
      "learning_rate": 1.3436855847189777e-05,
      "loss": 0.8541,
      "step": 2394
    },
    {
      "epoch": 0.657876665293229,
      "grad_norm": 0.3079429566860199,
      "learning_rate": 1.3434107461866154e-05,
      "loss": 0.7491,
      "step": 2395
    },
    {
      "epoch": 0.6581513528361489,
      "grad_norm": 0.3777804374694824,
      "learning_rate": 1.3431359076542531e-05,
      "loss": 0.6698,
      "step": 2396
    },
    {
      "epoch": 0.6584260403790688,
      "grad_norm": 0.4391746520996094,
      "learning_rate": 1.342861069121891e-05,
      "loss": 0.7827,
      "step": 2397
    },
    {
      "epoch": 0.6587007279219887,
      "grad_norm": 0.6223986744880676,
      "learning_rate": 1.3425862305895287e-05,
      "loss": 0.8837,
      "step": 2398
    },
    {
      "epoch": 0.6589754154649087,
      "grad_norm": 0.49607428908348083,
      "learning_rate": 1.3423113920571664e-05,
      "loss": 0.7911,
      "step": 2399
    },
    {
      "epoch": 0.6592501030078286,
      "grad_norm": 0.35844582319259644,
      "learning_rate": 1.3420365535248043e-05,
      "loss": 0.8361,
      "step": 2400
    },
    {
      "epoch": 0.6595247905507485,
      "grad_norm": 0.4243994653224945,
      "learning_rate": 1.341761714992442e-05,
      "loss": 0.4981,
      "step": 2401
    },
    {
      "epoch": 0.6597994780936685,
      "grad_norm": 0.3533773124217987,
      "learning_rate": 1.3414868764600799e-05,
      "loss": 0.6772,
      "step": 2402
    },
    {
      "epoch": 0.6600741656365884,
      "grad_norm": 0.3925134539604187,
      "learning_rate": 1.3412120379277176e-05,
      "loss": 1.0029,
      "step": 2403
    },
    {
      "epoch": 0.6603488531795083,
      "grad_norm": 0.40739625692367554,
      "learning_rate": 1.3409371993953553e-05,
      "loss": 0.8742,
      "step": 2404
    },
    {
      "epoch": 0.6606235407224282,
      "grad_norm": 0.4334717392921448,
      "learning_rate": 1.3406623608629931e-05,
      "loss": 0.6891,
      "step": 2405
    },
    {
      "epoch": 0.6608982282653482,
      "grad_norm": 0.46060827374458313,
      "learning_rate": 1.3403875223306308e-05,
      "loss": 0.9674,
      "step": 2406
    },
    {
      "epoch": 0.6611729158082681,
      "grad_norm": 0.3604790270328522,
      "learning_rate": 1.3401126837982685e-05,
      "loss": 0.6349,
      "step": 2407
    },
    {
      "epoch": 0.661447603351188,
      "grad_norm": 0.4108806252479553,
      "learning_rate": 1.3398378452659064e-05,
      "loss": 0.7354,
      "step": 2408
    },
    {
      "epoch": 0.661722290894108,
      "grad_norm": 0.37303629517555237,
      "learning_rate": 1.3395630067335441e-05,
      "loss": 0.9208,
      "step": 2409
    },
    {
      "epoch": 0.6619969784370279,
      "grad_norm": 0.5090392827987671,
      "learning_rate": 1.339288168201182e-05,
      "loss": 0.8946,
      "step": 2410
    },
    {
      "epoch": 0.6622716659799478,
      "grad_norm": 0.4597971439361572,
      "learning_rate": 1.3390133296688197e-05,
      "loss": 0.894,
      "step": 2411
    },
    {
      "epoch": 0.6625463535228677,
      "grad_norm": 0.4562961161136627,
      "learning_rate": 1.3387384911364574e-05,
      "loss": 0.8169,
      "step": 2412
    },
    {
      "epoch": 0.6628210410657877,
      "grad_norm": 0.506190836429596,
      "learning_rate": 1.3384636526040953e-05,
      "loss": 0.7071,
      "step": 2413
    },
    {
      "epoch": 0.6630957286087076,
      "grad_norm": 0.32461845874786377,
      "learning_rate": 1.338188814071733e-05,
      "loss": 1.0478,
      "step": 2414
    },
    {
      "epoch": 0.6633704161516275,
      "grad_norm": 0.3828307092189789,
      "learning_rate": 1.3379139755393707e-05,
      "loss": 0.8934,
      "step": 2415
    },
    {
      "epoch": 0.6636451036945474,
      "grad_norm": 0.5203040838241577,
      "learning_rate": 1.3376391370070085e-05,
      "loss": 0.6976,
      "step": 2416
    },
    {
      "epoch": 0.6639197912374674,
      "grad_norm": 0.36671510338783264,
      "learning_rate": 1.3373642984746462e-05,
      "loss": 0.5319,
      "step": 2417
    },
    {
      "epoch": 0.6641944787803873,
      "grad_norm": 0.47190582752227783,
      "learning_rate": 1.3370894599422841e-05,
      "loss": 0.9081,
      "step": 2418
    },
    {
      "epoch": 0.6644691663233072,
      "grad_norm": 0.2965547442436218,
      "learning_rate": 1.3368146214099218e-05,
      "loss": 0.9304,
      "step": 2419
    },
    {
      "epoch": 0.6647438538662271,
      "grad_norm": 0.42530620098114014,
      "learning_rate": 1.3365397828775595e-05,
      "loss": 1.0349,
      "step": 2420
    },
    {
      "epoch": 0.6650185414091471,
      "grad_norm": 0.43432918190956116,
      "learning_rate": 1.3362649443451974e-05,
      "loss": 0.7721,
      "step": 2421
    },
    {
      "epoch": 0.665293228952067,
      "grad_norm": 0.4949570596218109,
      "learning_rate": 1.3359901058128351e-05,
      "loss": 0.8655,
      "step": 2422
    },
    {
      "epoch": 0.6655679164949869,
      "grad_norm": 0.4468547999858856,
      "learning_rate": 1.3357152672804728e-05,
      "loss": 0.8295,
      "step": 2423
    },
    {
      "epoch": 0.6658426040379068,
      "grad_norm": 0.3562890887260437,
      "learning_rate": 1.3354404287481107e-05,
      "loss": 0.9854,
      "step": 2424
    },
    {
      "epoch": 0.6661172915808268,
      "grad_norm": 0.4297083616256714,
      "learning_rate": 1.3351655902157484e-05,
      "loss": 0.8656,
      "step": 2425
    },
    {
      "epoch": 0.6663919791237467,
      "grad_norm": 0.39351534843444824,
      "learning_rate": 1.3348907516833862e-05,
      "loss": 0.6604,
      "step": 2426
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.4184670150279999,
      "learning_rate": 1.334615913151024e-05,
      "loss": 1.0356,
      "step": 2427
    },
    {
      "epoch": 0.6669413542095866,
      "grad_norm": 0.4230524003505707,
      "learning_rate": 1.3343410746186616e-05,
      "loss": 0.9577,
      "step": 2428
    },
    {
      "epoch": 0.6672160417525065,
      "grad_norm": 0.46622228622436523,
      "learning_rate": 1.3340662360862995e-05,
      "loss": 0.8602,
      "step": 2429
    },
    {
      "epoch": 0.6674907292954264,
      "grad_norm": 0.40768077969551086,
      "learning_rate": 1.3337913975539372e-05,
      "loss": 0.9657,
      "step": 2430
    },
    {
      "epoch": 0.6677654168383463,
      "grad_norm": 0.4251699447631836,
      "learning_rate": 1.333516559021575e-05,
      "loss": 0.7787,
      "step": 2431
    },
    {
      "epoch": 0.6680401043812663,
      "grad_norm": 0.4231991469860077,
      "learning_rate": 1.3332417204892128e-05,
      "loss": 0.6217,
      "step": 2432
    },
    {
      "epoch": 0.6683147919241862,
      "grad_norm": 0.5107226371765137,
      "learning_rate": 1.3329668819568505e-05,
      "loss": 0.9015,
      "step": 2433
    },
    {
      "epoch": 0.6685894794671061,
      "grad_norm": 0.5022569298744202,
      "learning_rate": 1.3326920434244884e-05,
      "loss": 0.6124,
      "step": 2434
    },
    {
      "epoch": 0.668864167010026,
      "grad_norm": 0.40583640336990356,
      "learning_rate": 1.332417204892126e-05,
      "loss": 0.7845,
      "step": 2435
    },
    {
      "epoch": 0.669138854552946,
      "grad_norm": 0.43253880739212036,
      "learning_rate": 1.3321423663597638e-05,
      "loss": 0.8442,
      "step": 2436
    },
    {
      "epoch": 0.6694135420958659,
      "grad_norm": 0.40084096789360046,
      "learning_rate": 1.3318675278274016e-05,
      "loss": 0.7027,
      "step": 2437
    },
    {
      "epoch": 0.6696882296387858,
      "grad_norm": 0.41996264457702637,
      "learning_rate": 1.3315926892950393e-05,
      "loss": 0.8508,
      "step": 2438
    },
    {
      "epoch": 0.6699629171817059,
      "grad_norm": 0.4295591115951538,
      "learning_rate": 1.3313178507626769e-05,
      "loss": 0.9094,
      "step": 2439
    },
    {
      "epoch": 0.6702376047246258,
      "grad_norm": 0.6567744612693787,
      "learning_rate": 1.3310430122303149e-05,
      "loss": 0.8204,
      "step": 2440
    },
    {
      "epoch": 0.6705122922675457,
      "grad_norm": 0.38912293314933777,
      "learning_rate": 1.3307681736979524e-05,
      "loss": 1.0007,
      "step": 2441
    },
    {
      "epoch": 0.6707869798104656,
      "grad_norm": 0.4013628363609314,
      "learning_rate": 1.3304933351655905e-05,
      "loss": 0.9219,
      "step": 2442
    },
    {
      "epoch": 0.6710616673533856,
      "grad_norm": 0.38805294036865234,
      "learning_rate": 1.3302184966332282e-05,
      "loss": 0.7969,
      "step": 2443
    },
    {
      "epoch": 0.6713363548963055,
      "grad_norm": 0.37638819217681885,
      "learning_rate": 1.3299436581008657e-05,
      "loss": 0.7394,
      "step": 2444
    },
    {
      "epoch": 0.6716110424392254,
      "grad_norm": 0.30139660835266113,
      "learning_rate": 1.3296688195685038e-05,
      "loss": 1.1148,
      "step": 2445
    },
    {
      "epoch": 0.6718857299821454,
      "grad_norm": 0.4495241343975067,
      "learning_rate": 1.3293939810361413e-05,
      "loss": 1.0641,
      "step": 2446
    },
    {
      "epoch": 0.6721604175250653,
      "grad_norm": 0.4117393493652344,
      "learning_rate": 1.329119142503779e-05,
      "loss": 0.7452,
      "step": 2447
    },
    {
      "epoch": 0.6724351050679852,
      "grad_norm": 0.42835739254951477,
      "learning_rate": 1.3288443039714169e-05,
      "loss": 0.7658,
      "step": 2448
    },
    {
      "epoch": 0.6727097926109051,
      "grad_norm": 0.28486406803131104,
      "learning_rate": 1.3285694654390546e-05,
      "loss": 0.5044,
      "step": 2449
    },
    {
      "epoch": 0.6729844801538251,
      "grad_norm": 0.5996279120445251,
      "learning_rate": 1.3282946269066926e-05,
      "loss": 0.8916,
      "step": 2450
    },
    {
      "epoch": 0.673259167696745,
      "grad_norm": 0.5393305420875549,
      "learning_rate": 1.3280197883743301e-05,
      "loss": 0.7705,
      "step": 2451
    },
    {
      "epoch": 0.6735338552396649,
      "grad_norm": 0.4392099976539612,
      "learning_rate": 1.3277449498419678e-05,
      "loss": 0.8262,
      "step": 2452
    },
    {
      "epoch": 0.6738085427825848,
      "grad_norm": 0.46219876408576965,
      "learning_rate": 1.3274701113096057e-05,
      "loss": 0.8659,
      "step": 2453
    },
    {
      "epoch": 0.6740832303255048,
      "grad_norm": 0.39464783668518066,
      "learning_rate": 1.3271952727772434e-05,
      "loss": 0.5921,
      "step": 2454
    },
    {
      "epoch": 0.6743579178684247,
      "grad_norm": 0.48897257447242737,
      "learning_rate": 1.3269204342448811e-05,
      "loss": 0.9107,
      "step": 2455
    },
    {
      "epoch": 0.6746326054113446,
      "grad_norm": 0.489510715007782,
      "learning_rate": 1.326645595712519e-05,
      "loss": 0.9033,
      "step": 2456
    },
    {
      "epoch": 0.6749072929542645,
      "grad_norm": 0.41860470175743103,
      "learning_rate": 1.3263707571801567e-05,
      "loss": 0.9332,
      "step": 2457
    },
    {
      "epoch": 0.6751819804971845,
      "grad_norm": 0.341756135225296,
      "learning_rate": 1.3260959186477946e-05,
      "loss": 0.9554,
      "step": 2458
    },
    {
      "epoch": 0.6754566680401044,
      "grad_norm": 0.5766615271568298,
      "learning_rate": 1.3258210801154323e-05,
      "loss": 0.9448,
      "step": 2459
    },
    {
      "epoch": 0.6757313555830243,
      "grad_norm": 0.4076853394508362,
      "learning_rate": 1.32554624158307e-05,
      "loss": 0.993,
      "step": 2460
    },
    {
      "epoch": 0.6760060431259443,
      "grad_norm": 0.3664436638355255,
      "learning_rate": 1.3252714030507078e-05,
      "loss": 0.7568,
      "step": 2461
    },
    {
      "epoch": 0.6762807306688642,
      "grad_norm": 0.40126875042915344,
      "learning_rate": 1.3249965645183455e-05,
      "loss": 0.7947,
      "step": 2462
    },
    {
      "epoch": 0.6765554182117841,
      "grad_norm": 0.5068717002868652,
      "learning_rate": 1.3247217259859832e-05,
      "loss": 0.9974,
      "step": 2463
    },
    {
      "epoch": 0.676830105754704,
      "grad_norm": 0.3692241609096527,
      "learning_rate": 1.3244468874536211e-05,
      "loss": 0.7331,
      "step": 2464
    },
    {
      "epoch": 0.677104793297624,
      "grad_norm": 2.906212329864502,
      "learning_rate": 1.3241720489212588e-05,
      "loss": 1.0407,
      "step": 2465
    },
    {
      "epoch": 0.6773794808405439,
      "grad_norm": 0.7538865804672241,
      "learning_rate": 1.3238972103888967e-05,
      "loss": 0.8144,
      "step": 2466
    },
    {
      "epoch": 0.6776541683834638,
      "grad_norm": 0.36285486817359924,
      "learning_rate": 1.3236223718565344e-05,
      "loss": 0.7335,
      "step": 2467
    },
    {
      "epoch": 0.6779288559263837,
      "grad_norm": 0.4182887375354767,
      "learning_rate": 1.3233475333241721e-05,
      "loss": 0.6384,
      "step": 2468
    },
    {
      "epoch": 0.6782035434693037,
      "grad_norm": 0.40643030405044556,
      "learning_rate": 1.32307269479181e-05,
      "loss": 0.958,
      "step": 2469
    },
    {
      "epoch": 0.6784782310122236,
      "grad_norm": 0.32545945048332214,
      "learning_rate": 1.3227978562594477e-05,
      "loss": 0.8651,
      "step": 2470
    },
    {
      "epoch": 0.6787529185551435,
      "grad_norm": 0.3444758355617523,
      "learning_rate": 1.3225230177270854e-05,
      "loss": 0.9555,
      "step": 2471
    },
    {
      "epoch": 0.6790276060980635,
      "grad_norm": 0.4164741635322571,
      "learning_rate": 1.3222481791947232e-05,
      "loss": 0.6378,
      "step": 2472
    },
    {
      "epoch": 0.6793022936409834,
      "grad_norm": 0.5032501220703125,
      "learning_rate": 1.321973340662361e-05,
      "loss": 0.7652,
      "step": 2473
    },
    {
      "epoch": 0.6795769811839033,
      "grad_norm": 0.3933190703392029,
      "learning_rate": 1.3216985021299988e-05,
      "loss": 0.9882,
      "step": 2474
    },
    {
      "epoch": 0.6798516687268232,
      "grad_norm": 0.3553745448589325,
      "learning_rate": 1.3214236635976365e-05,
      "loss": 0.9058,
      "step": 2475
    },
    {
      "epoch": 0.6801263562697432,
      "grad_norm": 0.46272826194763184,
      "learning_rate": 1.3211488250652742e-05,
      "loss": 0.8905,
      "step": 2476
    },
    {
      "epoch": 0.6804010438126631,
      "grad_norm": 0.39290666580200195,
      "learning_rate": 1.3208739865329121e-05,
      "loss": 0.7458,
      "step": 2477
    },
    {
      "epoch": 0.680675731355583,
      "grad_norm": 0.49539539217948914,
      "learning_rate": 1.3205991480005498e-05,
      "loss": 0.7566,
      "step": 2478
    },
    {
      "epoch": 0.6809504188985029,
      "grad_norm": 0.4711631238460541,
      "learning_rate": 1.3203243094681875e-05,
      "loss": 0.7361,
      "step": 2479
    },
    {
      "epoch": 0.6812251064414229,
      "grad_norm": 0.4841075539588928,
      "learning_rate": 1.3200494709358254e-05,
      "loss": 0.8209,
      "step": 2480
    },
    {
      "epoch": 0.6814997939843428,
      "grad_norm": 0.4113127291202545,
      "learning_rate": 1.319774632403463e-05,
      "loss": 0.6299,
      "step": 2481
    },
    {
      "epoch": 0.6817744815272627,
      "grad_norm": 0.2932221591472626,
      "learning_rate": 1.319499793871101e-05,
      "loss": 0.9503,
      "step": 2482
    },
    {
      "epoch": 0.6820491690701826,
      "grad_norm": 0.4917018413543701,
      "learning_rate": 1.3192249553387386e-05,
      "loss": 0.9771,
      "step": 2483
    },
    {
      "epoch": 0.6823238566131026,
      "grad_norm": 0.48003146052360535,
      "learning_rate": 1.3189501168063763e-05,
      "loss": 0.5909,
      "step": 2484
    },
    {
      "epoch": 0.6825985441560225,
      "grad_norm": 0.4643349349498749,
      "learning_rate": 1.3186752782740142e-05,
      "loss": 0.7432,
      "step": 2485
    },
    {
      "epoch": 0.6828732316989424,
      "grad_norm": 0.4077408015727997,
      "learning_rate": 1.318400439741652e-05,
      "loss": 0.8101,
      "step": 2486
    },
    {
      "epoch": 0.6831479192418624,
      "grad_norm": 0.5649300813674927,
      "learning_rate": 1.3181256012092896e-05,
      "loss": 0.8287,
      "step": 2487
    },
    {
      "epoch": 0.6834226067847823,
      "grad_norm": 0.41024187207221985,
      "learning_rate": 1.3178507626769275e-05,
      "loss": 1.0516,
      "step": 2488
    },
    {
      "epoch": 0.6836972943277022,
      "grad_norm": 0.43574437499046326,
      "learning_rate": 1.3175759241445652e-05,
      "loss": 1.0881,
      "step": 2489
    },
    {
      "epoch": 0.6839719818706221,
      "grad_norm": 0.30976012349128723,
      "learning_rate": 1.317301085612203e-05,
      "loss": 0.5466,
      "step": 2490
    },
    {
      "epoch": 0.6842466694135421,
      "grad_norm": 0.43249672651290894,
      "learning_rate": 1.3170262470798408e-05,
      "loss": 0.8208,
      "step": 2491
    },
    {
      "epoch": 0.684521356956462,
      "grad_norm": 0.3732079863548279,
      "learning_rate": 1.3167514085474785e-05,
      "loss": 0.8146,
      "step": 2492
    },
    {
      "epoch": 0.6847960444993819,
      "grad_norm": 0.4268312454223633,
      "learning_rate": 1.3164765700151163e-05,
      "loss": 0.6655,
      "step": 2493
    },
    {
      "epoch": 0.6850707320423018,
      "grad_norm": 0.3623804450035095,
      "learning_rate": 1.316201731482754e-05,
      "loss": 0.9184,
      "step": 2494
    },
    {
      "epoch": 0.6853454195852218,
      "grad_norm": 0.46376362442970276,
      "learning_rate": 1.3159268929503917e-05,
      "loss": 0.5305,
      "step": 2495
    },
    {
      "epoch": 0.6856201071281417,
      "grad_norm": 0.5163015127182007,
      "learning_rate": 1.3156520544180296e-05,
      "loss": 1.0265,
      "step": 2496
    },
    {
      "epoch": 0.6858947946710616,
      "grad_norm": 0.4940735697746277,
      "learning_rate": 1.3153772158856673e-05,
      "loss": 0.6833,
      "step": 2497
    },
    {
      "epoch": 0.6861694822139816,
      "grad_norm": 0.3722177743911743,
      "learning_rate": 1.3151023773533052e-05,
      "loss": 0.6693,
      "step": 2498
    },
    {
      "epoch": 0.6864441697569015,
      "grad_norm": 0.3981100618839264,
      "learning_rate": 1.3148275388209429e-05,
      "loss": 0.7223,
      "step": 2499
    },
    {
      "epoch": 0.6867188572998214,
      "grad_norm": 0.40905219316482544,
      "learning_rate": 1.3145527002885804e-05,
      "loss": 0.664,
      "step": 2500
    },
    {
      "epoch": 0.6869935448427413,
      "grad_norm": 0.4297630190849304,
      "learning_rate": 1.3142778617562185e-05,
      "loss": 0.8879,
      "step": 2501
    },
    {
      "epoch": 0.6872682323856613,
      "grad_norm": 0.43739378452301025,
      "learning_rate": 1.3140030232238562e-05,
      "loss": 0.9767,
      "step": 2502
    },
    {
      "epoch": 0.6875429199285812,
      "grad_norm": 0.29156166315078735,
      "learning_rate": 1.3137281846914937e-05,
      "loss": 0.8025,
      "step": 2503
    },
    {
      "epoch": 0.6878176074715011,
      "grad_norm": 0.4391503632068634,
      "learning_rate": 1.3134533461591317e-05,
      "loss": 0.9125,
      "step": 2504
    },
    {
      "epoch": 0.6880922950144212,
      "grad_norm": 0.5905311107635498,
      "learning_rate": 1.3131785076267693e-05,
      "loss": 0.7763,
      "step": 2505
    },
    {
      "epoch": 0.6883669825573411,
      "grad_norm": 0.6847760677337646,
      "learning_rate": 1.3129036690944073e-05,
      "loss": 0.9257,
      "step": 2506
    },
    {
      "epoch": 0.688641670100261,
      "grad_norm": 0.5752922892570496,
      "learning_rate": 1.3126288305620448e-05,
      "loss": 0.7945,
      "step": 2507
    },
    {
      "epoch": 0.6889163576431809,
      "grad_norm": 0.5122411847114563,
      "learning_rate": 1.3123539920296826e-05,
      "loss": 0.8452,
      "step": 2508
    },
    {
      "epoch": 0.6891910451861009,
      "grad_norm": 0.3471599519252777,
      "learning_rate": 1.3120791534973206e-05,
      "loss": 0.5596,
      "step": 2509
    },
    {
      "epoch": 0.6894657327290208,
      "grad_norm": 0.41370877623558044,
      "learning_rate": 1.3118043149649581e-05,
      "loss": 0.5963,
      "step": 2510
    },
    {
      "epoch": 0.6897404202719407,
      "grad_norm": 0.43347224593162537,
      "learning_rate": 1.3115294764325958e-05,
      "loss": 0.753,
      "step": 2511
    },
    {
      "epoch": 0.6900151078148606,
      "grad_norm": 0.3470153510570526,
      "learning_rate": 1.3112546379002337e-05,
      "loss": 0.9106,
      "step": 2512
    },
    {
      "epoch": 0.6902897953577806,
      "grad_norm": 0.4391767382621765,
      "learning_rate": 1.3109797993678714e-05,
      "loss": 0.9441,
      "step": 2513
    },
    {
      "epoch": 0.6905644829007005,
      "grad_norm": 0.4279528260231018,
      "learning_rate": 1.3107049608355093e-05,
      "loss": 0.6295,
      "step": 2514
    },
    {
      "epoch": 0.6908391704436204,
      "grad_norm": 0.4235847294330597,
      "learning_rate": 1.310430122303147e-05,
      "loss": 0.9579,
      "step": 2515
    },
    {
      "epoch": 0.6911138579865403,
      "grad_norm": 0.3999024033546448,
      "learning_rate": 1.3101552837707847e-05,
      "loss": 0.8104,
      "step": 2516
    },
    {
      "epoch": 0.6913885455294603,
      "grad_norm": 0.42281657457351685,
      "learning_rate": 1.3098804452384225e-05,
      "loss": 0.7414,
      "step": 2517
    },
    {
      "epoch": 0.6916632330723802,
      "grad_norm": 0.4335510730743408,
      "learning_rate": 1.3096056067060603e-05,
      "loss": 0.5212,
      "step": 2518
    },
    {
      "epoch": 0.6919379206153001,
      "grad_norm": 0.4730035066604614,
      "learning_rate": 1.309330768173698e-05,
      "loss": 0.9721,
      "step": 2519
    },
    {
      "epoch": 0.69221260815822,
      "grad_norm": 0.4455912709236145,
      "learning_rate": 1.3090559296413358e-05,
      "loss": 0.7537,
      "step": 2520
    },
    {
      "epoch": 0.69248729570114,
      "grad_norm": 0.3873048722743988,
      "learning_rate": 1.3087810911089735e-05,
      "loss": 0.665,
      "step": 2521
    },
    {
      "epoch": 0.6927619832440599,
      "grad_norm": 0.5281676650047302,
      "learning_rate": 1.3085062525766114e-05,
      "loss": 0.6287,
      "step": 2522
    },
    {
      "epoch": 0.6930366707869798,
      "grad_norm": 0.3268296718597412,
      "learning_rate": 1.3082314140442491e-05,
      "loss": 0.4737,
      "step": 2523
    },
    {
      "epoch": 0.6933113583298998,
      "grad_norm": 0.3779178261756897,
      "learning_rate": 1.3079565755118868e-05,
      "loss": 0.9269,
      "step": 2524
    },
    {
      "epoch": 0.6935860458728197,
      "grad_norm": 0.355900377035141,
      "learning_rate": 1.3076817369795247e-05,
      "loss": 1.0051,
      "step": 2525
    },
    {
      "epoch": 0.6938607334157396,
      "grad_norm": 0.44494393467903137,
      "learning_rate": 1.3074068984471624e-05,
      "loss": 0.7684,
      "step": 2526
    },
    {
      "epoch": 0.6941354209586595,
      "grad_norm": 0.4277920126914978,
      "learning_rate": 1.3071320599148e-05,
      "loss": 0.6743,
      "step": 2527
    },
    {
      "epoch": 0.6944101085015795,
      "grad_norm": 0.48165300488471985,
      "learning_rate": 1.306857221382438e-05,
      "loss": 0.8865,
      "step": 2528
    },
    {
      "epoch": 0.6946847960444994,
      "grad_norm": 0.36710917949676514,
      "learning_rate": 1.3065823828500757e-05,
      "loss": 0.6028,
      "step": 2529
    },
    {
      "epoch": 0.6949594835874193,
      "grad_norm": 0.42719176411628723,
      "learning_rate": 1.3063075443177135e-05,
      "loss": 0.8835,
      "step": 2530
    },
    {
      "epoch": 0.6952341711303393,
      "grad_norm": 0.43599027395248413,
      "learning_rate": 1.3060327057853512e-05,
      "loss": 0.6275,
      "step": 2531
    },
    {
      "epoch": 0.6955088586732592,
      "grad_norm": 0.3605963885784149,
      "learning_rate": 1.305757867252989e-05,
      "loss": 0.7719,
      "step": 2532
    },
    {
      "epoch": 0.6957835462161791,
      "grad_norm": 0.31869590282440186,
      "learning_rate": 1.3054830287206268e-05,
      "loss": 1.0039,
      "step": 2533
    },
    {
      "epoch": 0.696058233759099,
      "grad_norm": 0.5453352332115173,
      "learning_rate": 1.3052081901882645e-05,
      "loss": 0.7997,
      "step": 2534
    },
    {
      "epoch": 0.696332921302019,
      "grad_norm": 0.45424187183380127,
      "learning_rate": 1.3049333516559022e-05,
      "loss": 0.7858,
      "step": 2535
    },
    {
      "epoch": 0.6966076088449389,
      "grad_norm": 0.44945260882377625,
      "learning_rate": 1.30465851312354e-05,
      "loss": 0.8525,
      "step": 2536
    },
    {
      "epoch": 0.6968822963878588,
      "grad_norm": 0.3765419125556946,
      "learning_rate": 1.3043836745911778e-05,
      "loss": 0.8205,
      "step": 2537
    },
    {
      "epoch": 0.6971569839307787,
      "grad_norm": 0.37883126735687256,
      "learning_rate": 1.3041088360588155e-05,
      "loss": 0.6572,
      "step": 2538
    },
    {
      "epoch": 0.6974316714736987,
      "grad_norm": 0.4723029136657715,
      "learning_rate": 1.3038339975264533e-05,
      "loss": 0.7354,
      "step": 2539
    },
    {
      "epoch": 0.6977063590166186,
      "grad_norm": 0.3703017830848694,
      "learning_rate": 1.303559158994091e-05,
      "loss": 0.8276,
      "step": 2540
    },
    {
      "epoch": 0.6979810465595385,
      "grad_norm": 0.45571646094322205,
      "learning_rate": 1.303284320461729e-05,
      "loss": 0.8886,
      "step": 2541
    },
    {
      "epoch": 0.6982557341024584,
      "grad_norm": 0.4841305911540985,
      "learning_rate": 1.3030094819293666e-05,
      "loss": 0.8805,
      "step": 2542
    },
    {
      "epoch": 0.6985304216453784,
      "grad_norm": 0.43448135256767273,
      "learning_rate": 1.3027346433970043e-05,
      "loss": 0.8653,
      "step": 2543
    },
    {
      "epoch": 0.6988051091882983,
      "grad_norm": 0.40910738706588745,
      "learning_rate": 1.3024598048646422e-05,
      "loss": 0.7497,
      "step": 2544
    },
    {
      "epoch": 0.6990797967312182,
      "grad_norm": 0.3577800393104553,
      "learning_rate": 1.3021849663322799e-05,
      "loss": 0.9854,
      "step": 2545
    },
    {
      "epoch": 0.6993544842741382,
      "grad_norm": 0.46194887161254883,
      "learning_rate": 1.3019101277999176e-05,
      "loss": 0.7247,
      "step": 2546
    },
    {
      "epoch": 0.6996291718170581,
      "grad_norm": 0.4110136926174164,
      "learning_rate": 1.3016352892675555e-05,
      "loss": 0.8512,
      "step": 2547
    },
    {
      "epoch": 0.699903859359978,
      "grad_norm": 0.4741031229496002,
      "learning_rate": 1.3013604507351932e-05,
      "loss": 0.8446,
      "step": 2548
    },
    {
      "epoch": 0.7001785469028979,
      "grad_norm": 0.4223114550113678,
      "learning_rate": 1.301085612202831e-05,
      "loss": 0.954,
      "step": 2549
    },
    {
      "epoch": 0.7004532344458179,
      "grad_norm": 0.44700562953948975,
      "learning_rate": 1.3008107736704687e-05,
      "loss": 0.5769,
      "step": 2550
    },
    {
      "epoch": 0.7007279219887378,
      "grad_norm": 0.6098933219909668,
      "learning_rate": 1.3005359351381065e-05,
      "loss": 0.753,
      "step": 2551
    },
    {
      "epoch": 0.7010026095316577,
      "grad_norm": 0.28204071521759033,
      "learning_rate": 1.3002610966057443e-05,
      "loss": 0.6723,
      "step": 2552
    },
    {
      "epoch": 0.7012772970745776,
      "grad_norm": 0.327251136302948,
      "learning_rate": 1.299986258073382e-05,
      "loss": 0.5782,
      "step": 2553
    },
    {
      "epoch": 0.7015519846174976,
      "grad_norm": 0.5971869230270386,
      "learning_rate": 1.2997114195410197e-05,
      "loss": 0.8378,
      "step": 2554
    },
    {
      "epoch": 0.7018266721604175,
      "grad_norm": 0.3665277659893036,
      "learning_rate": 1.2994365810086576e-05,
      "loss": 1.0483,
      "step": 2555
    },
    {
      "epoch": 0.7021013597033374,
      "grad_norm": 0.45382818579673767,
      "learning_rate": 1.2991617424762953e-05,
      "loss": 0.6652,
      "step": 2556
    },
    {
      "epoch": 0.7023760472462574,
      "grad_norm": 0.4437173008918762,
      "learning_rate": 1.2988869039439332e-05,
      "loss": 0.771,
      "step": 2557
    },
    {
      "epoch": 0.7026507347891773,
      "grad_norm": 0.4334750175476074,
      "learning_rate": 1.2986120654115709e-05,
      "loss": 0.8705,
      "step": 2558
    },
    {
      "epoch": 0.7029254223320972,
      "grad_norm": 0.4235488176345825,
      "learning_rate": 1.2983372268792084e-05,
      "loss": 0.8334,
      "step": 2559
    },
    {
      "epoch": 0.7032001098750171,
      "grad_norm": 0.47960159182548523,
      "learning_rate": 1.2980623883468464e-05,
      "loss": 0.7354,
      "step": 2560
    },
    {
      "epoch": 0.7034747974179371,
      "grad_norm": 0.4613870680332184,
      "learning_rate": 1.2977875498144842e-05,
      "loss": 0.7781,
      "step": 2561
    },
    {
      "epoch": 0.703749484960857,
      "grad_norm": 0.42740172147750854,
      "learning_rate": 1.2975127112821217e-05,
      "loss": 0.8644,
      "step": 2562
    },
    {
      "epoch": 0.7040241725037769,
      "grad_norm": 0.5442057251930237,
      "learning_rate": 1.2972378727497597e-05,
      "loss": 0.8888,
      "step": 2563
    },
    {
      "epoch": 0.7042988600466968,
      "grad_norm": 0.3874342441558838,
      "learning_rate": 1.2969630342173973e-05,
      "loss": 0.6562,
      "step": 2564
    },
    {
      "epoch": 0.7045735475896168,
      "grad_norm": 0.43796640634536743,
      "learning_rate": 1.2966881956850353e-05,
      "loss": 0.7591,
      "step": 2565
    },
    {
      "epoch": 0.7048482351325367,
      "grad_norm": 0.5469735860824585,
      "learning_rate": 1.2964133571526728e-05,
      "loss": 0.9445,
      "step": 2566
    },
    {
      "epoch": 0.7051229226754566,
      "grad_norm": 0.40011274814605713,
      "learning_rate": 1.2961385186203105e-05,
      "loss": 0.8659,
      "step": 2567
    },
    {
      "epoch": 0.7053976102183765,
      "grad_norm": 0.3112790286540985,
      "learning_rate": 1.2958636800879486e-05,
      "loss": 0.7775,
      "step": 2568
    },
    {
      "epoch": 0.7056722977612965,
      "grad_norm": 0.519625723361969,
      "learning_rate": 1.2955888415555861e-05,
      "loss": 0.9616,
      "step": 2569
    },
    {
      "epoch": 0.7059469853042164,
      "grad_norm": 0.5869515538215637,
      "learning_rate": 1.2953140030232238e-05,
      "loss": 0.9127,
      "step": 2570
    },
    {
      "epoch": 0.7062216728471363,
      "grad_norm": 0.38749751448631287,
      "learning_rate": 1.2950391644908617e-05,
      "loss": 1.0198,
      "step": 2571
    },
    {
      "epoch": 0.7064963603900564,
      "grad_norm": 0.46599212288856506,
      "learning_rate": 1.2947643259584994e-05,
      "loss": 1.0208,
      "step": 2572
    },
    {
      "epoch": 0.7067710479329763,
      "grad_norm": 0.32162952423095703,
      "learning_rate": 1.2944894874261373e-05,
      "loss": 0.5738,
      "step": 2573
    },
    {
      "epoch": 0.7070457354758962,
      "grad_norm": 0.374640554189682,
      "learning_rate": 1.294214648893775e-05,
      "loss": 0.5691,
      "step": 2574
    },
    {
      "epoch": 0.7073204230188161,
      "grad_norm": 0.5253236293792725,
      "learning_rate": 1.2939398103614127e-05,
      "loss": 0.7155,
      "step": 2575
    },
    {
      "epoch": 0.7075951105617361,
      "grad_norm": 0.5020178556442261,
      "learning_rate": 1.2936649718290505e-05,
      "loss": 0.8794,
      "step": 2576
    },
    {
      "epoch": 0.707869798104656,
      "grad_norm": 0.33916693925857544,
      "learning_rate": 1.2933901332966882e-05,
      "loss": 0.4387,
      "step": 2577
    },
    {
      "epoch": 0.7081444856475759,
      "grad_norm": 0.4010704457759857,
      "learning_rate": 1.293115294764326e-05,
      "loss": 0.7833,
      "step": 2578
    },
    {
      "epoch": 0.7084191731904959,
      "grad_norm": 0.43975725769996643,
      "learning_rate": 1.2928404562319638e-05,
      "loss": 0.9405,
      "step": 2579
    },
    {
      "epoch": 0.7086938607334158,
      "grad_norm": 0.2482573688030243,
      "learning_rate": 1.2925656176996015e-05,
      "loss": 0.4484,
      "step": 2580
    },
    {
      "epoch": 0.7089685482763357,
      "grad_norm": 0.3891310691833496,
      "learning_rate": 1.2922907791672394e-05,
      "loss": 0.663,
      "step": 2581
    },
    {
      "epoch": 0.7092432358192556,
      "grad_norm": 0.49695122241973877,
      "learning_rate": 1.292015940634877e-05,
      "loss": 0.7385,
      "step": 2582
    },
    {
      "epoch": 0.7095179233621756,
      "grad_norm": 0.33610647916793823,
      "learning_rate": 1.2917411021025148e-05,
      "loss": 0.9952,
      "step": 2583
    },
    {
      "epoch": 0.7097926109050955,
      "grad_norm": 0.3785381019115448,
      "learning_rate": 1.2914662635701527e-05,
      "loss": 0.8761,
      "step": 2584
    },
    {
      "epoch": 0.7100672984480154,
      "grad_norm": 0.3717294931411743,
      "learning_rate": 1.2911914250377904e-05,
      "loss": 0.8209,
      "step": 2585
    },
    {
      "epoch": 0.7103419859909353,
      "grad_norm": 0.5123809576034546,
      "learning_rate": 1.290916586505428e-05,
      "loss": 1.0634,
      "step": 2586
    },
    {
      "epoch": 0.7106166735338553,
      "grad_norm": 0.34137460589408875,
      "learning_rate": 1.290641747973066e-05,
      "loss": 0.5841,
      "step": 2587
    },
    {
      "epoch": 0.7108913610767752,
      "grad_norm": 0.3912794589996338,
      "learning_rate": 1.2903669094407036e-05,
      "loss": 0.5522,
      "step": 2588
    },
    {
      "epoch": 0.7111660486196951,
      "grad_norm": 0.4955684542655945,
      "learning_rate": 1.2900920709083415e-05,
      "loss": 0.7053,
      "step": 2589
    },
    {
      "epoch": 0.711440736162615,
      "grad_norm": 0.3681042790412903,
      "learning_rate": 1.2898172323759792e-05,
      "loss": 0.8767,
      "step": 2590
    },
    {
      "epoch": 0.711715423705535,
      "grad_norm": 0.3753027021884918,
      "learning_rate": 1.2895423938436169e-05,
      "loss": 0.804,
      "step": 2591
    },
    {
      "epoch": 0.7119901112484549,
      "grad_norm": 0.3904483914375305,
      "learning_rate": 1.2892675553112548e-05,
      "loss": 0.892,
      "step": 2592
    },
    {
      "epoch": 0.7122647987913748,
      "grad_norm": 0.5284411907196045,
      "learning_rate": 1.2889927167788925e-05,
      "loss": 1.1069,
      "step": 2593
    },
    {
      "epoch": 0.7125394863342948,
      "grad_norm": 0.389761358499527,
      "learning_rate": 1.2887178782465302e-05,
      "loss": 0.7252,
      "step": 2594
    },
    {
      "epoch": 0.7128141738772147,
      "grad_norm": 0.361841082572937,
      "learning_rate": 1.288443039714168e-05,
      "loss": 0.5223,
      "step": 2595
    },
    {
      "epoch": 0.7130888614201346,
      "grad_norm": 0.5237256288528442,
      "learning_rate": 1.2881682011818058e-05,
      "loss": 0.8461,
      "step": 2596
    },
    {
      "epoch": 0.7133635489630545,
      "grad_norm": 0.33471670746803284,
      "learning_rate": 1.2878933626494436e-05,
      "loss": 0.6314,
      "step": 2597
    },
    {
      "epoch": 0.7136382365059745,
      "grad_norm": 0.38082727789878845,
      "learning_rate": 1.2876185241170813e-05,
      "loss": 0.6581,
      "step": 2598
    },
    {
      "epoch": 0.7139129240488944,
      "grad_norm": 0.46476441621780396,
      "learning_rate": 1.287343685584719e-05,
      "loss": 0.7351,
      "step": 2599
    },
    {
      "epoch": 0.7141876115918143,
      "grad_norm": 0.40470507740974426,
      "learning_rate": 1.2870688470523569e-05,
      "loss": 0.7544,
      "step": 2600
    },
    {
      "epoch": 0.7144622991347342,
      "grad_norm": 0.4632035195827484,
      "learning_rate": 1.2867940085199946e-05,
      "loss": 0.5701,
      "step": 2601
    },
    {
      "epoch": 0.7147369866776542,
      "grad_norm": 0.42518576979637146,
      "learning_rate": 1.2865191699876323e-05,
      "loss": 0.9546,
      "step": 2602
    },
    {
      "epoch": 0.7150116742205741,
      "grad_norm": 0.3912157118320465,
      "learning_rate": 1.2862443314552702e-05,
      "loss": 0.7295,
      "step": 2603
    },
    {
      "epoch": 0.715286361763494,
      "grad_norm": 0.32500001788139343,
      "learning_rate": 1.2859694929229079e-05,
      "loss": 0.6392,
      "step": 2604
    },
    {
      "epoch": 0.715561049306414,
      "grad_norm": 0.42134955525398254,
      "learning_rate": 1.2856946543905458e-05,
      "loss": 0.5027,
      "step": 2605
    },
    {
      "epoch": 0.7158357368493339,
      "grad_norm": 0.32763397693634033,
      "learning_rate": 1.2854198158581835e-05,
      "loss": 0.5842,
      "step": 2606
    },
    {
      "epoch": 0.7161104243922538,
      "grad_norm": 0.5489423274993896,
      "learning_rate": 1.2851449773258212e-05,
      "loss": 0.7816,
      "step": 2607
    },
    {
      "epoch": 0.7163851119351737,
      "grad_norm": 0.488581120967865,
      "learning_rate": 1.284870138793459e-05,
      "loss": 0.8728,
      "step": 2608
    },
    {
      "epoch": 0.7166597994780937,
      "grad_norm": 0.38944026827812195,
      "learning_rate": 1.2845953002610967e-05,
      "loss": 0.7332,
      "step": 2609
    },
    {
      "epoch": 0.7169344870210136,
      "grad_norm": 0.42928871512413025,
      "learning_rate": 1.2843204617287344e-05,
      "loss": 0.7796,
      "step": 2610
    },
    {
      "epoch": 0.7172091745639335,
      "grad_norm": 0.4415881335735321,
      "learning_rate": 1.2840456231963723e-05,
      "loss": 0.5668,
      "step": 2611
    },
    {
      "epoch": 0.7174838621068534,
      "grad_norm": 0.33534786105155945,
      "learning_rate": 1.28377078466401e-05,
      "loss": 0.949,
      "step": 2612
    },
    {
      "epoch": 0.7177585496497734,
      "grad_norm": 0.4195427894592285,
      "learning_rate": 1.2834959461316479e-05,
      "loss": 0.9385,
      "step": 2613
    },
    {
      "epoch": 0.7180332371926933,
      "grad_norm": 0.5358110070228577,
      "learning_rate": 1.2832211075992856e-05,
      "loss": 0.6526,
      "step": 2614
    },
    {
      "epoch": 0.7183079247356132,
      "grad_norm": 0.35317903757095337,
      "learning_rate": 1.2829462690669233e-05,
      "loss": 0.6334,
      "step": 2615
    },
    {
      "epoch": 0.7185826122785332,
      "grad_norm": 0.30869513750076294,
      "learning_rate": 1.2826714305345612e-05,
      "loss": 0.5587,
      "step": 2616
    },
    {
      "epoch": 0.7188572998214531,
      "grad_norm": 0.452627032995224,
      "learning_rate": 1.2823965920021989e-05,
      "loss": 0.8379,
      "step": 2617
    },
    {
      "epoch": 0.719131987364373,
      "grad_norm": 0.4689362347126007,
      "learning_rate": 1.2821217534698364e-05,
      "loss": 0.8144,
      "step": 2618
    },
    {
      "epoch": 0.7194066749072929,
      "grad_norm": 0.3419845700263977,
      "learning_rate": 1.2818469149374744e-05,
      "loss": 0.5747,
      "step": 2619
    },
    {
      "epoch": 0.7196813624502129,
      "grad_norm": 0.43731752038002014,
      "learning_rate": 1.2815720764051121e-05,
      "loss": 0.7884,
      "step": 2620
    },
    {
      "epoch": 0.7199560499931328,
      "grad_norm": 0.3218039274215698,
      "learning_rate": 1.28129723787275e-05,
      "loss": 0.8027,
      "step": 2621
    },
    {
      "epoch": 0.7202307375360527,
      "grad_norm": 0.304639995098114,
      "learning_rate": 1.2810223993403877e-05,
      "loss": 0.5732,
      "step": 2622
    },
    {
      "epoch": 0.7205054250789726,
      "grad_norm": 0.33327195048332214,
      "learning_rate": 1.2807475608080252e-05,
      "loss": 0.7363,
      "step": 2623
    },
    {
      "epoch": 0.7207801126218926,
      "grad_norm": 0.26708731055259705,
      "learning_rate": 1.2804727222756633e-05,
      "loss": 0.5877,
      "step": 2624
    },
    {
      "epoch": 0.7210548001648125,
      "grad_norm": 0.42177706956863403,
      "learning_rate": 1.280197883743301e-05,
      "loss": 0.5867,
      "step": 2625
    },
    {
      "epoch": 0.7213294877077324,
      "grad_norm": 0.3864443898200989,
      "learning_rate": 1.2799230452109385e-05,
      "loss": 1.0244,
      "step": 2626
    },
    {
      "epoch": 0.7216041752506523,
      "grad_norm": 0.38053980469703674,
      "learning_rate": 1.2796482066785766e-05,
      "loss": 0.8757,
      "step": 2627
    },
    {
      "epoch": 0.7218788627935723,
      "grad_norm": 0.37073343992233276,
      "learning_rate": 1.2793733681462141e-05,
      "loss": 0.8586,
      "step": 2628
    },
    {
      "epoch": 0.7221535503364922,
      "grad_norm": 0.39966726303100586,
      "learning_rate": 1.2790985296138521e-05,
      "loss": 0.9038,
      "step": 2629
    },
    {
      "epoch": 0.7224282378794121,
      "grad_norm": 0.3199721872806549,
      "learning_rate": 1.2788236910814897e-05,
      "loss": 0.7806,
      "step": 2630
    },
    {
      "epoch": 0.722702925422332,
      "grad_norm": 0.3459722101688385,
      "learning_rate": 1.2785488525491274e-05,
      "loss": 0.9141,
      "step": 2631
    },
    {
      "epoch": 0.722977612965252,
      "grad_norm": 0.3430350720882416,
      "learning_rate": 1.2782740140167654e-05,
      "loss": 0.706,
      "step": 2632
    },
    {
      "epoch": 0.7232523005081719,
      "grad_norm": 0.3631364703178406,
      "learning_rate": 1.277999175484403e-05,
      "loss": 0.9804,
      "step": 2633
    },
    {
      "epoch": 0.7235269880510918,
      "grad_norm": 0.5737611651420593,
      "learning_rate": 1.2777243369520406e-05,
      "loss": 0.767,
      "step": 2634
    },
    {
      "epoch": 0.7238016755940118,
      "grad_norm": 0.45769378542900085,
      "learning_rate": 1.2774494984196785e-05,
      "loss": 0.8744,
      "step": 2635
    },
    {
      "epoch": 0.7240763631369317,
      "grad_norm": 0.3383202850818634,
      "learning_rate": 1.2771746598873162e-05,
      "loss": 0.6743,
      "step": 2636
    },
    {
      "epoch": 0.7243510506798516,
      "grad_norm": 0.4197041690349579,
      "learning_rate": 1.276899821354954e-05,
      "loss": 1.0766,
      "step": 2637
    },
    {
      "epoch": 0.7246257382227715,
      "grad_norm": 0.36645662784576416,
      "learning_rate": 1.2766249828225918e-05,
      "loss": 0.6932,
      "step": 2638
    },
    {
      "epoch": 0.7249004257656916,
      "grad_norm": 0.4998141825199127,
      "learning_rate": 1.2763501442902295e-05,
      "loss": 0.6036,
      "step": 2639
    },
    {
      "epoch": 0.7251751133086115,
      "grad_norm": 0.4875708222389221,
      "learning_rate": 1.2760753057578674e-05,
      "loss": 0.7347,
      "step": 2640
    },
    {
      "epoch": 0.7254498008515314,
      "grad_norm": 0.41486626863479614,
      "learning_rate": 1.275800467225505e-05,
      "loss": 0.6821,
      "step": 2641
    },
    {
      "epoch": 0.7257244883944514,
      "grad_norm": 0.4197487533092499,
      "learning_rate": 1.2755256286931428e-05,
      "loss": 0.8723,
      "step": 2642
    },
    {
      "epoch": 0.7259991759373713,
      "grad_norm": 0.39878642559051514,
      "learning_rate": 1.2752507901607806e-05,
      "loss": 0.7714,
      "step": 2643
    },
    {
      "epoch": 0.7262738634802912,
      "grad_norm": 0.45609110593795776,
      "learning_rate": 1.2749759516284183e-05,
      "loss": 0.7379,
      "step": 2644
    },
    {
      "epoch": 0.7265485510232111,
      "grad_norm": 0.3231832683086395,
      "learning_rate": 1.2747011130960562e-05,
      "loss": 0.8502,
      "step": 2645
    },
    {
      "epoch": 0.7268232385661311,
      "grad_norm": 0.40170589089393616,
      "learning_rate": 1.2744262745636939e-05,
      "loss": 0.8331,
      "step": 2646
    },
    {
      "epoch": 0.727097926109051,
      "grad_norm": 0.4380098283290863,
      "learning_rate": 1.2741514360313316e-05,
      "loss": 0.8608,
      "step": 2647
    },
    {
      "epoch": 0.7273726136519709,
      "grad_norm": 0.34390056133270264,
      "learning_rate": 1.2738765974989695e-05,
      "loss": 0.9193,
      "step": 2648
    },
    {
      "epoch": 0.7276473011948908,
      "grad_norm": 0.4012860357761383,
      "learning_rate": 1.2736017589666072e-05,
      "loss": 0.9664,
      "step": 2649
    },
    {
      "epoch": 0.7279219887378108,
      "grad_norm": 0.35220104455947876,
      "learning_rate": 1.2733269204342449e-05,
      "loss": 0.9635,
      "step": 2650
    },
    {
      "epoch": 0.7281966762807307,
      "grad_norm": 0.4053080677986145,
      "learning_rate": 1.2730520819018828e-05,
      "loss": 0.8873,
      "step": 2651
    },
    {
      "epoch": 0.7284713638236506,
      "grad_norm": 0.31364622712135315,
      "learning_rate": 1.2727772433695205e-05,
      "loss": 0.4544,
      "step": 2652
    },
    {
      "epoch": 0.7287460513665706,
      "grad_norm": 0.3024512231349945,
      "learning_rate": 1.2725024048371583e-05,
      "loss": 0.6749,
      "step": 2653
    },
    {
      "epoch": 0.7290207389094905,
      "grad_norm": 0.4257940948009491,
      "learning_rate": 1.272227566304796e-05,
      "loss": 0.7726,
      "step": 2654
    },
    {
      "epoch": 0.7292954264524104,
      "grad_norm": 0.41107177734375,
      "learning_rate": 1.2719527277724337e-05,
      "loss": 0.962,
      "step": 2655
    },
    {
      "epoch": 0.7295701139953303,
      "grad_norm": 0.5166407227516174,
      "learning_rate": 1.2716778892400716e-05,
      "loss": 0.6795,
      "step": 2656
    },
    {
      "epoch": 0.7298448015382503,
      "grad_norm": 0.3221108317375183,
      "learning_rate": 1.2714030507077093e-05,
      "loss": 0.725,
      "step": 2657
    },
    {
      "epoch": 0.7301194890811702,
      "grad_norm": 0.4840562641620636,
      "learning_rate": 1.271128212175347e-05,
      "loss": 0.7534,
      "step": 2658
    },
    {
      "epoch": 0.7303941766240901,
      "grad_norm": 0.3618643879890442,
      "learning_rate": 1.2708533736429849e-05,
      "loss": 0.7731,
      "step": 2659
    },
    {
      "epoch": 0.73066886416701,
      "grad_norm": 0.41029301285743713,
      "learning_rate": 1.2705785351106226e-05,
      "loss": 0.8749,
      "step": 2660
    },
    {
      "epoch": 0.73094355170993,
      "grad_norm": 0.3459523618221283,
      "learning_rate": 1.2703036965782605e-05,
      "loss": 0.6692,
      "step": 2661
    },
    {
      "epoch": 0.7312182392528499,
      "grad_norm": 0.5006167888641357,
      "learning_rate": 1.2700288580458982e-05,
      "loss": 0.7273,
      "step": 2662
    },
    {
      "epoch": 0.7314929267957698,
      "grad_norm": 0.3192788362503052,
      "learning_rate": 1.2697540195135359e-05,
      "loss": 1.036,
      "step": 2663
    },
    {
      "epoch": 0.7317676143386898,
      "grad_norm": 0.4546571373939514,
      "learning_rate": 1.2694791809811737e-05,
      "loss": 0.6187,
      "step": 2664
    },
    {
      "epoch": 0.7320423018816097,
      "grad_norm": 0.48162686824798584,
      "learning_rate": 1.2692043424488114e-05,
      "loss": 0.82,
      "step": 2665
    },
    {
      "epoch": 0.7323169894245296,
      "grad_norm": 0.49704667925834656,
      "learning_rate": 1.2689295039164491e-05,
      "loss": 0.7795,
      "step": 2666
    },
    {
      "epoch": 0.7325916769674495,
      "grad_norm": 0.3876737952232361,
      "learning_rate": 1.268654665384087e-05,
      "loss": 0.9278,
      "step": 2667
    },
    {
      "epoch": 0.7328663645103695,
      "grad_norm": 0.36634212732315063,
      "learning_rate": 1.2683798268517247e-05,
      "loss": 0.6368,
      "step": 2668
    },
    {
      "epoch": 0.7331410520532894,
      "grad_norm": 0.36456647515296936,
      "learning_rate": 1.2681049883193626e-05,
      "loss": 0.6168,
      "step": 2669
    },
    {
      "epoch": 0.7334157395962093,
      "grad_norm": 0.45884788036346436,
      "learning_rate": 1.2678301497870003e-05,
      "loss": 0.6106,
      "step": 2670
    },
    {
      "epoch": 0.7336904271391292,
      "grad_norm": 0.334961473941803,
      "learning_rate": 1.267555311254638e-05,
      "loss": 0.6139,
      "step": 2671
    },
    {
      "epoch": 0.7339651146820492,
      "grad_norm": 0.4308086931705475,
      "learning_rate": 1.2672804727222759e-05,
      "loss": 0.8558,
      "step": 2672
    },
    {
      "epoch": 0.7342398022249691,
      "grad_norm": 0.45520108938217163,
      "learning_rate": 1.2670056341899136e-05,
      "loss": 0.861,
      "step": 2673
    },
    {
      "epoch": 0.734514489767889,
      "grad_norm": 0.5594604015350342,
      "learning_rate": 1.2667307956575513e-05,
      "loss": 0.855,
      "step": 2674
    },
    {
      "epoch": 0.734789177310809,
      "grad_norm": 0.3736240565776825,
      "learning_rate": 1.2664559571251891e-05,
      "loss": 0.9285,
      "step": 2675
    },
    {
      "epoch": 0.7350638648537289,
      "grad_norm": 0.41455721855163574,
      "learning_rate": 1.2661811185928268e-05,
      "loss": 1.1272,
      "step": 2676
    },
    {
      "epoch": 0.7353385523966488,
      "grad_norm": 0.4263284206390381,
      "learning_rate": 1.2659062800604647e-05,
      "loss": 0.9222,
      "step": 2677
    },
    {
      "epoch": 0.7356132399395687,
      "grad_norm": 0.4594009220600128,
      "learning_rate": 1.2656314415281024e-05,
      "loss": 0.7273,
      "step": 2678
    },
    {
      "epoch": 0.7358879274824887,
      "grad_norm": 0.26561567187309265,
      "learning_rate": 1.2653566029957401e-05,
      "loss": 0.5828,
      "step": 2679
    },
    {
      "epoch": 0.7361626150254086,
      "grad_norm": 0.4782300889492035,
      "learning_rate": 1.265081764463378e-05,
      "loss": 1.056,
      "step": 2680
    },
    {
      "epoch": 0.7364373025683285,
      "grad_norm": 0.45629435777664185,
      "learning_rate": 1.2648069259310157e-05,
      "loss": 0.8729,
      "step": 2681
    },
    {
      "epoch": 0.7367119901112484,
      "grad_norm": 0.42822304368019104,
      "learning_rate": 1.2645320873986532e-05,
      "loss": 0.9187,
      "step": 2682
    },
    {
      "epoch": 0.7369866776541684,
      "grad_norm": 0.4970606863498688,
      "learning_rate": 1.2642572488662913e-05,
      "loss": 0.9148,
      "step": 2683
    },
    {
      "epoch": 0.7372613651970883,
      "grad_norm": 0.575200080871582,
      "learning_rate": 1.263982410333929e-05,
      "loss": 0.785,
      "step": 2684
    },
    {
      "epoch": 0.7375360527400082,
      "grad_norm": 0.3137892782688141,
      "learning_rate": 1.2637075718015668e-05,
      "loss": 0.6582,
      "step": 2685
    },
    {
      "epoch": 0.7378107402829281,
      "grad_norm": 0.5692229866981506,
      "learning_rate": 1.2634327332692045e-05,
      "loss": 0.7939,
      "step": 2686
    },
    {
      "epoch": 0.7380854278258481,
      "grad_norm": 0.5534932613372803,
      "learning_rate": 1.263157894736842e-05,
      "loss": 0.7833,
      "step": 2687
    },
    {
      "epoch": 0.738360115368768,
      "grad_norm": 0.5322375297546387,
      "learning_rate": 1.2628830562044801e-05,
      "loss": 0.8498,
      "step": 2688
    },
    {
      "epoch": 0.7386348029116879,
      "grad_norm": 0.4174712300300598,
      "learning_rate": 1.2626082176721176e-05,
      "loss": 0.6643,
      "step": 2689
    },
    {
      "epoch": 0.7389094904546079,
      "grad_norm": 0.3051123023033142,
      "learning_rate": 1.2623333791397553e-05,
      "loss": 0.6756,
      "step": 2690
    },
    {
      "epoch": 0.7391841779975278,
      "grad_norm": 0.5492472648620605,
      "learning_rate": 1.2620585406073934e-05,
      "loss": 0.7913,
      "step": 2691
    },
    {
      "epoch": 0.7394588655404477,
      "grad_norm": 0.508948802947998,
      "learning_rate": 1.2617837020750309e-05,
      "loss": 0.6468,
      "step": 2692
    },
    {
      "epoch": 0.7397335530833676,
      "grad_norm": 0.4019610583782196,
      "learning_rate": 1.261508863542669e-05,
      "loss": 0.7979,
      "step": 2693
    },
    {
      "epoch": 0.7400082406262876,
      "grad_norm": 0.2937764823436737,
      "learning_rate": 1.2612340250103065e-05,
      "loss": 0.772,
      "step": 2694
    },
    {
      "epoch": 0.7402829281692075,
      "grad_norm": 0.43755391240119934,
      "learning_rate": 1.2609591864779442e-05,
      "loss": 0.752,
      "step": 2695
    },
    {
      "epoch": 0.7405576157121274,
      "grad_norm": 0.4251433312892914,
      "learning_rate": 1.260684347945582e-05,
      "loss": 0.7893,
      "step": 2696
    },
    {
      "epoch": 0.7408323032550473,
      "grad_norm": 0.5110555291175842,
      "learning_rate": 1.2604095094132198e-05,
      "loss": 0.8759,
      "step": 2697
    },
    {
      "epoch": 0.7411069907979673,
      "grad_norm": 0.32156845927238464,
      "learning_rate": 1.2601346708808575e-05,
      "loss": 0.7584,
      "step": 2698
    },
    {
      "epoch": 0.7413816783408872,
      "grad_norm": 0.5449675917625427,
      "learning_rate": 1.2598598323484953e-05,
      "loss": 0.6204,
      "step": 2699
    },
    {
      "epoch": 0.7416563658838071,
      "grad_norm": 0.47504767775535583,
      "learning_rate": 1.259584993816133e-05,
      "loss": 0.8022,
      "step": 2700
    },
    {
      "epoch": 0.741931053426727,
      "grad_norm": 0.3736690282821655,
      "learning_rate": 1.2593101552837709e-05,
      "loss": 0.8386,
      "step": 2701
    },
    {
      "epoch": 0.742205740969647,
      "grad_norm": 0.4156479835510254,
      "learning_rate": 1.2590353167514086e-05,
      "loss": 0.9753,
      "step": 2702
    },
    {
      "epoch": 0.7424804285125669,
      "grad_norm": 0.4008459150791168,
      "learning_rate": 1.2587604782190463e-05,
      "loss": 0.8061,
      "step": 2703
    },
    {
      "epoch": 0.7427551160554868,
      "grad_norm": 0.27138614654541016,
      "learning_rate": 1.2584856396866842e-05,
      "loss": 0.7856,
      "step": 2704
    },
    {
      "epoch": 0.7430298035984068,
      "grad_norm": 0.4496285915374756,
      "learning_rate": 1.2582108011543219e-05,
      "loss": 0.7918,
      "step": 2705
    },
    {
      "epoch": 0.7433044911413268,
      "grad_norm": 0.466139554977417,
      "learning_rate": 1.2579359626219596e-05,
      "loss": 0.6942,
      "step": 2706
    },
    {
      "epoch": 0.7435791786842467,
      "grad_norm": 0.44289010763168335,
      "learning_rate": 1.2576611240895975e-05,
      "loss": 0.8692,
      "step": 2707
    },
    {
      "epoch": 0.7438538662271666,
      "grad_norm": 0.4149937033653259,
      "learning_rate": 1.2573862855572352e-05,
      "loss": 0.8391,
      "step": 2708
    },
    {
      "epoch": 0.7441285537700866,
      "grad_norm": 0.4624582827091217,
      "learning_rate": 1.257111447024873e-05,
      "loss": 0.8514,
      "step": 2709
    },
    {
      "epoch": 0.7444032413130065,
      "grad_norm": 0.4108308255672455,
      "learning_rate": 1.2568366084925107e-05,
      "loss": 0.6886,
      "step": 2710
    },
    {
      "epoch": 0.7446779288559264,
      "grad_norm": 0.45699119567871094,
      "learning_rate": 1.2565617699601484e-05,
      "loss": 0.791,
      "step": 2711
    },
    {
      "epoch": 0.7449526163988464,
      "grad_norm": 0.4129078686237335,
      "learning_rate": 1.2562869314277863e-05,
      "loss": 0.7195,
      "step": 2712
    },
    {
      "epoch": 0.7452273039417663,
      "grad_norm": 0.4573785066604614,
      "learning_rate": 1.256012092895424e-05,
      "loss": 0.6837,
      "step": 2713
    },
    {
      "epoch": 0.7455019914846862,
      "grad_norm": 0.28073686361312866,
      "learning_rate": 1.2557372543630617e-05,
      "loss": 0.5314,
      "step": 2714
    },
    {
      "epoch": 0.7457766790276061,
      "grad_norm": 0.42188987135887146,
      "learning_rate": 1.2554624158306996e-05,
      "loss": 0.8792,
      "step": 2715
    },
    {
      "epoch": 0.7460513665705261,
      "grad_norm": 0.3263362944126129,
      "learning_rate": 1.2551875772983373e-05,
      "loss": 0.9144,
      "step": 2716
    },
    {
      "epoch": 0.746326054113446,
      "grad_norm": 0.4027324318885803,
      "learning_rate": 1.2549127387659752e-05,
      "loss": 0.6171,
      "step": 2717
    },
    {
      "epoch": 0.7466007416563659,
      "grad_norm": 0.48830586671829224,
      "learning_rate": 1.2546379002336129e-05,
      "loss": 0.7491,
      "step": 2718
    },
    {
      "epoch": 0.7468754291992858,
      "grad_norm": 0.32444772124290466,
      "learning_rate": 1.2543630617012506e-05,
      "loss": 0.583,
      "step": 2719
    },
    {
      "epoch": 0.7471501167422058,
      "grad_norm": 0.3379845917224884,
      "learning_rate": 1.2540882231688884e-05,
      "loss": 0.6968,
      "step": 2720
    },
    {
      "epoch": 0.7474248042851257,
      "grad_norm": 0.3869752883911133,
      "learning_rate": 1.2538133846365261e-05,
      "loss": 0.8248,
      "step": 2721
    },
    {
      "epoch": 0.7476994918280456,
      "grad_norm": 0.4435710608959198,
      "learning_rate": 1.2535385461041638e-05,
      "loss": 0.9483,
      "step": 2722
    },
    {
      "epoch": 0.7479741793709656,
      "grad_norm": 0.40100860595703125,
      "learning_rate": 1.2532637075718017e-05,
      "loss": 0.9414,
      "step": 2723
    },
    {
      "epoch": 0.7482488669138855,
      "grad_norm": 0.46914586424827576,
      "learning_rate": 1.2529888690394394e-05,
      "loss": 0.6127,
      "step": 2724
    },
    {
      "epoch": 0.7485235544568054,
      "grad_norm": 0.4456745386123657,
      "learning_rate": 1.2527140305070773e-05,
      "loss": 0.7925,
      "step": 2725
    },
    {
      "epoch": 0.7487982419997253,
      "grad_norm": 0.3865029811859131,
      "learning_rate": 1.252439191974715e-05,
      "loss": 0.6782,
      "step": 2726
    },
    {
      "epoch": 0.7490729295426453,
      "grad_norm": 0.40850940346717834,
      "learning_rate": 1.2521643534423527e-05,
      "loss": 0.8634,
      "step": 2727
    },
    {
      "epoch": 0.7493476170855652,
      "grad_norm": 0.3572840690612793,
      "learning_rate": 1.2518895149099906e-05,
      "loss": 0.7085,
      "step": 2728
    },
    {
      "epoch": 0.7496223046284851,
      "grad_norm": 0.43457648158073425,
      "learning_rate": 1.2516146763776283e-05,
      "loss": 0.6117,
      "step": 2729
    },
    {
      "epoch": 0.749896992171405,
      "grad_norm": 0.40651974081993103,
      "learning_rate": 1.251339837845266e-05,
      "loss": 0.6467,
      "step": 2730
    },
    {
      "epoch": 0.750171679714325,
      "grad_norm": 0.4454154372215271,
      "learning_rate": 1.2510649993129038e-05,
      "loss": 0.7813,
      "step": 2731
    },
    {
      "epoch": 0.7504463672572449,
      "grad_norm": 0.4495835304260254,
      "learning_rate": 1.2507901607805415e-05,
      "loss": 0.6368,
      "step": 2732
    },
    {
      "epoch": 0.7507210548001648,
      "grad_norm": 0.3770275115966797,
      "learning_rate": 1.2505153222481794e-05,
      "loss": 0.9945,
      "step": 2733
    },
    {
      "epoch": 0.7509957423430847,
      "grad_norm": 0.47982028126716614,
      "learning_rate": 1.2502404837158171e-05,
      "loss": 0.7614,
      "step": 2734
    },
    {
      "epoch": 0.7512704298860047,
      "grad_norm": 0.44457748532295227,
      "learning_rate": 1.2499656451834548e-05,
      "loss": 0.5942,
      "step": 2735
    },
    {
      "epoch": 0.7515451174289246,
      "grad_norm": 0.3679920732975006,
      "learning_rate": 1.2496908066510927e-05,
      "loss": 0.7311,
      "step": 2736
    },
    {
      "epoch": 0.7518198049718445,
      "grad_norm": 0.639054536819458,
      "learning_rate": 1.2494159681187304e-05,
      "loss": 0.6532,
      "step": 2737
    },
    {
      "epoch": 0.7520944925147645,
      "grad_norm": 0.5089520215988159,
      "learning_rate": 1.2491411295863681e-05,
      "loss": 0.7437,
      "step": 2738
    },
    {
      "epoch": 0.7523691800576844,
      "grad_norm": 0.40228453278541565,
      "learning_rate": 1.248866291054006e-05,
      "loss": 0.7679,
      "step": 2739
    },
    {
      "epoch": 0.7526438676006043,
      "grad_norm": 0.4628741443157196,
      "learning_rate": 1.2485914525216437e-05,
      "loss": 0.6789,
      "step": 2740
    },
    {
      "epoch": 0.7529185551435242,
      "grad_norm": 0.5101096630096436,
      "learning_rate": 1.2483166139892815e-05,
      "loss": 0.8112,
      "step": 2741
    },
    {
      "epoch": 0.7531932426864442,
      "grad_norm": 0.4485037922859192,
      "learning_rate": 1.2480417754569192e-05,
      "loss": 0.7608,
      "step": 2742
    },
    {
      "epoch": 0.7534679302293641,
      "grad_norm": 0.30424538254737854,
      "learning_rate": 1.247766936924557e-05,
      "loss": 0.8364,
      "step": 2743
    },
    {
      "epoch": 0.753742617772284,
      "grad_norm": 0.3430722951889038,
      "learning_rate": 1.2474920983921948e-05,
      "loss": 0.9639,
      "step": 2744
    },
    {
      "epoch": 0.754017305315204,
      "grad_norm": 0.4376615583896637,
      "learning_rate": 1.2472172598598325e-05,
      "loss": 0.9993,
      "step": 2745
    },
    {
      "epoch": 0.7542919928581239,
      "grad_norm": 0.5028585195541382,
      "learning_rate": 1.24694242132747e-05,
      "loss": 0.6567,
      "step": 2746
    },
    {
      "epoch": 0.7545666804010438,
      "grad_norm": 0.36697596311569214,
      "learning_rate": 1.2466675827951081e-05,
      "loss": 0.513,
      "step": 2747
    },
    {
      "epoch": 0.7548413679439637,
      "grad_norm": 0.335035502910614,
      "learning_rate": 1.2463927442627456e-05,
      "loss": 0.9686,
      "step": 2748
    },
    {
      "epoch": 0.7551160554868837,
      "grad_norm": 0.38358280062675476,
      "learning_rate": 1.2461179057303837e-05,
      "loss": 0.7924,
      "step": 2749
    },
    {
      "epoch": 0.7553907430298036,
      "grad_norm": 0.40973955392837524,
      "learning_rate": 1.2458430671980214e-05,
      "loss": 1.0657,
      "step": 2750
    },
    {
      "epoch": 0.7556654305727235,
      "grad_norm": 0.42449089884757996,
      "learning_rate": 1.2455682286656589e-05,
      "loss": 0.7532,
      "step": 2751
    },
    {
      "epoch": 0.7559401181156434,
      "grad_norm": 0.5221022367477417,
      "learning_rate": 1.245293390133297e-05,
      "loss": 0.8303,
      "step": 2752
    },
    {
      "epoch": 0.7562148056585634,
      "grad_norm": 0.45302173495292664,
      "learning_rate": 1.2450185516009345e-05,
      "loss": 0.8859,
      "step": 2753
    },
    {
      "epoch": 0.7564894932014833,
      "grad_norm": 0.32886338233947754,
      "learning_rate": 1.2447437130685722e-05,
      "loss": 0.5391,
      "step": 2754
    },
    {
      "epoch": 0.7567641807444032,
      "grad_norm": 0.4265406131744385,
      "learning_rate": 1.24446887453621e-05,
      "loss": 1.0174,
      "step": 2755
    },
    {
      "epoch": 0.7570388682873231,
      "grad_norm": 0.3724417984485626,
      "learning_rate": 1.2441940360038477e-05,
      "loss": 0.7305,
      "step": 2756
    },
    {
      "epoch": 0.7573135558302431,
      "grad_norm": 0.5888676643371582,
      "learning_rate": 1.2439191974714858e-05,
      "loss": 0.7708,
      "step": 2757
    },
    {
      "epoch": 0.757588243373163,
      "grad_norm": 0.34608393907546997,
      "learning_rate": 1.2436443589391233e-05,
      "loss": 0.8035,
      "step": 2758
    },
    {
      "epoch": 0.7578629309160829,
      "grad_norm": 0.3046647310256958,
      "learning_rate": 1.243369520406761e-05,
      "loss": 0.6179,
      "step": 2759
    },
    {
      "epoch": 0.7581376184590028,
      "grad_norm": 0.34475910663604736,
      "learning_rate": 1.2430946818743989e-05,
      "loss": 0.8456,
      "step": 2760
    },
    {
      "epoch": 0.7584123060019228,
      "grad_norm": 0.3253839910030365,
      "learning_rate": 1.2428198433420366e-05,
      "loss": 1.1358,
      "step": 2761
    },
    {
      "epoch": 0.7586869935448427,
      "grad_norm": 0.39693474769592285,
      "learning_rate": 1.2425450048096743e-05,
      "loss": 0.9157,
      "step": 2762
    },
    {
      "epoch": 0.7589616810877626,
      "grad_norm": 0.39607444405555725,
      "learning_rate": 1.2422701662773122e-05,
      "loss": 0.8221,
      "step": 2763
    },
    {
      "epoch": 0.7592363686306826,
      "grad_norm": 0.4735741913318634,
      "learning_rate": 1.2419953277449499e-05,
      "loss": 0.6338,
      "step": 2764
    },
    {
      "epoch": 0.7595110561736025,
      "grad_norm": 0.4223717451095581,
      "learning_rate": 1.2417204892125877e-05,
      "loss": 0.8895,
      "step": 2765
    },
    {
      "epoch": 0.7597857437165224,
      "grad_norm": 0.2933275103569031,
      "learning_rate": 1.2414456506802254e-05,
      "loss": 0.7614,
      "step": 2766
    },
    {
      "epoch": 0.7600604312594423,
      "grad_norm": 0.41127467155456543,
      "learning_rate": 1.2411708121478631e-05,
      "loss": 0.9284,
      "step": 2767
    },
    {
      "epoch": 0.7603351188023623,
      "grad_norm": 0.45078879594802856,
      "learning_rate": 1.240895973615501e-05,
      "loss": 0.9342,
      "step": 2768
    },
    {
      "epoch": 0.7606098063452822,
      "grad_norm": 0.40736398100852966,
      "learning_rate": 1.2406211350831387e-05,
      "loss": 0.8272,
      "step": 2769
    },
    {
      "epoch": 0.7608844938882021,
      "grad_norm": 0.395685613155365,
      "learning_rate": 1.2403462965507764e-05,
      "loss": 0.7192,
      "step": 2770
    },
    {
      "epoch": 0.761159181431122,
      "grad_norm": 0.2817753553390503,
      "learning_rate": 1.2400714580184143e-05,
      "loss": 0.763,
      "step": 2771
    },
    {
      "epoch": 0.761433868974042,
      "grad_norm": 0.36590275168418884,
      "learning_rate": 1.239796619486052e-05,
      "loss": 0.9386,
      "step": 2772
    },
    {
      "epoch": 0.761708556516962,
      "grad_norm": 0.380031943321228,
      "learning_rate": 1.2395217809536899e-05,
      "loss": 0.8297,
      "step": 2773
    },
    {
      "epoch": 0.7619832440598819,
      "grad_norm": 0.3725648820400238,
      "learning_rate": 1.2392469424213276e-05,
      "loss": 0.7566,
      "step": 2774
    },
    {
      "epoch": 0.7622579316028019,
      "grad_norm": 0.3393590450286865,
      "learning_rate": 1.2389721038889653e-05,
      "loss": 0.5699,
      "step": 2775
    },
    {
      "epoch": 0.7625326191457218,
      "grad_norm": 0.4672906696796417,
      "learning_rate": 1.2386972653566031e-05,
      "loss": 0.9248,
      "step": 2776
    },
    {
      "epoch": 0.7628073066886417,
      "grad_norm": 0.43857428431510925,
      "learning_rate": 1.2384224268242408e-05,
      "loss": 1.0102,
      "step": 2777
    },
    {
      "epoch": 0.7630819942315616,
      "grad_norm": 0.3563160002231598,
      "learning_rate": 1.2381475882918785e-05,
      "loss": 0.6362,
      "step": 2778
    },
    {
      "epoch": 0.7633566817744816,
      "grad_norm": 0.2839386761188507,
      "learning_rate": 1.2378727497595164e-05,
      "loss": 0.736,
      "step": 2779
    },
    {
      "epoch": 0.7636313693174015,
      "grad_norm": 0.4900609254837036,
      "learning_rate": 1.2375979112271541e-05,
      "loss": 0.7982,
      "step": 2780
    },
    {
      "epoch": 0.7639060568603214,
      "grad_norm": 0.41263309121131897,
      "learning_rate": 1.237323072694792e-05,
      "loss": 0.6825,
      "step": 2781
    },
    {
      "epoch": 0.7641807444032414,
      "grad_norm": 0.36929851770401,
      "learning_rate": 1.2370482341624297e-05,
      "loss": 0.8491,
      "step": 2782
    },
    {
      "epoch": 0.7644554319461613,
      "grad_norm": 0.2897096574306488,
      "learning_rate": 1.2367733956300674e-05,
      "loss": 0.6285,
      "step": 2783
    },
    {
      "epoch": 0.7647301194890812,
      "grad_norm": 0.44509270787239075,
      "learning_rate": 1.2364985570977053e-05,
      "loss": 0.872,
      "step": 2784
    },
    {
      "epoch": 0.7650048070320011,
      "grad_norm": 0.5956674814224243,
      "learning_rate": 1.236223718565343e-05,
      "loss": 0.8328,
      "step": 2785
    },
    {
      "epoch": 0.7652794945749211,
      "grad_norm": 0.3945281505584717,
      "learning_rate": 1.2359488800329807e-05,
      "loss": 0.5439,
      "step": 2786
    },
    {
      "epoch": 0.765554182117841,
      "grad_norm": 0.34722402691841125,
      "learning_rate": 1.2356740415006185e-05,
      "loss": 0.8007,
      "step": 2787
    },
    {
      "epoch": 0.7658288696607609,
      "grad_norm": 0.36594900488853455,
      "learning_rate": 1.2353992029682562e-05,
      "loss": 0.6262,
      "step": 2788
    },
    {
      "epoch": 0.7661035572036808,
      "grad_norm": 0.3798649311065674,
      "learning_rate": 1.2351243644358941e-05,
      "loss": 0.7623,
      "step": 2789
    },
    {
      "epoch": 0.7663782447466008,
      "grad_norm": 0.6004505157470703,
      "learning_rate": 1.2348495259035318e-05,
      "loss": 0.7139,
      "step": 2790
    },
    {
      "epoch": 0.7666529322895207,
      "grad_norm": 0.4017181694507599,
      "learning_rate": 1.2345746873711695e-05,
      "loss": 0.9588,
      "step": 2791
    },
    {
      "epoch": 0.7669276198324406,
      "grad_norm": 0.35286837816238403,
      "learning_rate": 1.2342998488388074e-05,
      "loss": 0.7289,
      "step": 2792
    },
    {
      "epoch": 0.7672023073753605,
      "grad_norm": 0.4075562357902527,
      "learning_rate": 1.2340250103064451e-05,
      "loss": 0.969,
      "step": 2793
    },
    {
      "epoch": 0.7674769949182805,
      "grad_norm": 0.4694030284881592,
      "learning_rate": 1.2337501717740828e-05,
      "loss": 0.8701,
      "step": 2794
    },
    {
      "epoch": 0.7677516824612004,
      "grad_norm": 0.45876190066337585,
      "learning_rate": 1.2334753332417207e-05,
      "loss": 0.8955,
      "step": 2795
    },
    {
      "epoch": 0.7680263700041203,
      "grad_norm": 0.37815529108047485,
      "learning_rate": 1.2332004947093584e-05,
      "loss": 0.839,
      "step": 2796
    },
    {
      "epoch": 0.7683010575470403,
      "grad_norm": 0.46743249893188477,
      "learning_rate": 1.2329256561769962e-05,
      "loss": 0.6828,
      "step": 2797
    },
    {
      "epoch": 0.7685757450899602,
      "grad_norm": 0.3494011461734772,
      "learning_rate": 1.232650817644634e-05,
      "loss": 0.8664,
      "step": 2798
    },
    {
      "epoch": 0.7688504326328801,
      "grad_norm": 0.3969676196575165,
      "learning_rate": 1.2323759791122716e-05,
      "loss": 0.8592,
      "step": 2799
    },
    {
      "epoch": 0.7691251201758,
      "grad_norm": 0.5172176957130432,
      "learning_rate": 1.2321011405799095e-05,
      "loss": 0.6434,
      "step": 2800
    },
    {
      "epoch": 0.76939980771872,
      "grad_norm": 0.4073917865753174,
      "learning_rate": 1.2318263020475472e-05,
      "loss": 0.6855,
      "step": 2801
    },
    {
      "epoch": 0.7696744952616399,
      "grad_norm": 0.46564266085624695,
      "learning_rate": 1.231551463515185e-05,
      "loss": 0.7074,
      "step": 2802
    },
    {
      "epoch": 0.7699491828045598,
      "grad_norm": 0.41759058833122253,
      "learning_rate": 1.2312766249828228e-05,
      "loss": 0.891,
      "step": 2803
    },
    {
      "epoch": 0.7702238703474797,
      "grad_norm": 0.4072630703449249,
      "learning_rate": 1.2310017864504605e-05,
      "loss": 0.8629,
      "step": 2804
    },
    {
      "epoch": 0.7704985578903997,
      "grad_norm": 0.41321003437042236,
      "learning_rate": 1.2307269479180984e-05,
      "loss": 1.0163,
      "step": 2805
    },
    {
      "epoch": 0.7707732454333196,
      "grad_norm": 0.37025725841522217,
      "learning_rate": 1.230452109385736e-05,
      "loss": 0.8497,
      "step": 2806
    },
    {
      "epoch": 0.7710479329762395,
      "grad_norm": 0.38870295882225037,
      "learning_rate": 1.2301772708533736e-05,
      "loss": 1.1092,
      "step": 2807
    },
    {
      "epoch": 0.7713226205191595,
      "grad_norm": 0.2967287003993988,
      "learning_rate": 1.2299024323210116e-05,
      "loss": 0.9071,
      "step": 2808
    },
    {
      "epoch": 0.7715973080620794,
      "grad_norm": 0.5159684419631958,
      "learning_rate": 1.2296275937886493e-05,
      "loss": 0.985,
      "step": 2809
    },
    {
      "epoch": 0.7718719956049993,
      "grad_norm": 0.45214012265205383,
      "learning_rate": 1.2293527552562869e-05,
      "loss": 1.0278,
      "step": 2810
    },
    {
      "epoch": 0.7721466831479192,
      "grad_norm": 0.3313060998916626,
      "learning_rate": 1.229077916723925e-05,
      "loss": 0.8901,
      "step": 2811
    },
    {
      "epoch": 0.7724213706908392,
      "grad_norm": 0.6387304663658142,
      "learning_rate": 1.2288030781915624e-05,
      "loss": 0.7545,
      "step": 2812
    },
    {
      "epoch": 0.7726960582337591,
      "grad_norm": 0.5369746088981628,
      "learning_rate": 1.2285282396592005e-05,
      "loss": 0.9751,
      "step": 2813
    },
    {
      "epoch": 0.772970745776679,
      "grad_norm": 0.509718656539917,
      "learning_rate": 1.2282534011268382e-05,
      "loss": 0.7975,
      "step": 2814
    },
    {
      "epoch": 0.7732454333195989,
      "grad_norm": 0.4240679442882538,
      "learning_rate": 1.2279785625944757e-05,
      "loss": 0.79,
      "step": 2815
    },
    {
      "epoch": 0.7735201208625189,
      "grad_norm": 0.39171671867370605,
      "learning_rate": 1.2277037240621138e-05,
      "loss": 0.9897,
      "step": 2816
    },
    {
      "epoch": 0.7737948084054388,
      "grad_norm": 0.3799555003643036,
      "learning_rate": 1.2274288855297513e-05,
      "loss": 0.9886,
      "step": 2817
    },
    {
      "epoch": 0.7740694959483587,
      "grad_norm": 0.35818812251091003,
      "learning_rate": 1.227154046997389e-05,
      "loss": 0.791,
      "step": 2818
    },
    {
      "epoch": 0.7743441834912786,
      "grad_norm": 0.44445690512657166,
      "learning_rate": 1.2268792084650269e-05,
      "loss": 0.5182,
      "step": 2819
    },
    {
      "epoch": 0.7746188710341986,
      "grad_norm": 0.4444340169429779,
      "learning_rate": 1.2266043699326646e-05,
      "loss": 0.9453,
      "step": 2820
    },
    {
      "epoch": 0.7748935585771185,
      "grad_norm": 0.4177187383174896,
      "learning_rate": 1.2263295314003026e-05,
      "loss": 0.6854,
      "step": 2821
    },
    {
      "epoch": 0.7751682461200384,
      "grad_norm": 0.5563009977340698,
      "learning_rate": 1.2260546928679401e-05,
      "loss": 0.6996,
      "step": 2822
    },
    {
      "epoch": 0.7754429336629584,
      "grad_norm": 0.3756076991558075,
      "learning_rate": 1.2257798543355778e-05,
      "loss": 0.7923,
      "step": 2823
    },
    {
      "epoch": 0.7757176212058783,
      "grad_norm": 0.3705824017524719,
      "learning_rate": 1.2255050158032157e-05,
      "loss": 1.0417,
      "step": 2824
    },
    {
      "epoch": 0.7759923087487982,
      "grad_norm": 0.43830275535583496,
      "learning_rate": 1.2252301772708534e-05,
      "loss": 0.6939,
      "step": 2825
    },
    {
      "epoch": 0.7762669962917181,
      "grad_norm": 0.46757063269615173,
      "learning_rate": 1.2249553387384911e-05,
      "loss": 0.9611,
      "step": 2826
    },
    {
      "epoch": 0.7765416838346381,
      "grad_norm": 0.4635539948940277,
      "learning_rate": 1.224680500206129e-05,
      "loss": 0.8564,
      "step": 2827
    },
    {
      "epoch": 0.776816371377558,
      "grad_norm": 0.544963538646698,
      "learning_rate": 1.2244056616737667e-05,
      "loss": 0.9619,
      "step": 2828
    },
    {
      "epoch": 0.7770910589204779,
      "grad_norm": 0.32570815086364746,
      "learning_rate": 1.2241308231414046e-05,
      "loss": 0.6945,
      "step": 2829
    },
    {
      "epoch": 0.7773657464633978,
      "grad_norm": 0.4102344214916229,
      "learning_rate": 1.2238559846090423e-05,
      "loss": 0.6504,
      "step": 2830
    },
    {
      "epoch": 0.7776404340063178,
      "grad_norm": 0.4070529341697693,
      "learning_rate": 1.22358114607668e-05,
      "loss": 0.7495,
      "step": 2831
    },
    {
      "epoch": 0.7779151215492377,
      "grad_norm": 0.4209791421890259,
      "learning_rate": 1.2233063075443178e-05,
      "loss": 0.7244,
      "step": 2832
    },
    {
      "epoch": 0.7781898090921576,
      "grad_norm": 0.39029428362846375,
      "learning_rate": 1.2230314690119555e-05,
      "loss": 0.688,
      "step": 2833
    },
    {
      "epoch": 0.7784644966350776,
      "grad_norm": 0.43187856674194336,
      "learning_rate": 1.2227566304795932e-05,
      "loss": 0.636,
      "step": 2834
    },
    {
      "epoch": 0.7787391841779975,
      "grad_norm": 0.38435399532318115,
      "learning_rate": 1.2224817919472311e-05,
      "loss": 0.6372,
      "step": 2835
    },
    {
      "epoch": 0.7790138717209174,
      "grad_norm": 0.5166619420051575,
      "learning_rate": 1.2222069534148688e-05,
      "loss": 0.8307,
      "step": 2836
    },
    {
      "epoch": 0.7792885592638373,
      "grad_norm": 0.4158237874507904,
      "learning_rate": 1.2219321148825067e-05,
      "loss": 0.7943,
      "step": 2837
    },
    {
      "epoch": 0.7795632468067573,
      "grad_norm": 0.4088239371776581,
      "learning_rate": 1.2216572763501444e-05,
      "loss": 0.9366,
      "step": 2838
    },
    {
      "epoch": 0.7798379343496773,
      "grad_norm": 0.6008925437927246,
      "learning_rate": 1.2213824378177821e-05,
      "loss": 0.5878,
      "step": 2839
    },
    {
      "epoch": 0.7801126218925972,
      "grad_norm": 0.5085877776145935,
      "learning_rate": 1.22110759928542e-05,
      "loss": 0.7359,
      "step": 2840
    },
    {
      "epoch": 0.7803873094355172,
      "grad_norm": 0.41613632440567017,
      "learning_rate": 1.2208327607530577e-05,
      "loss": 0.6311,
      "step": 2841
    },
    {
      "epoch": 0.7806619969784371,
      "grad_norm": 0.39659443497657776,
      "learning_rate": 1.2205579222206954e-05,
      "loss": 0.8962,
      "step": 2842
    },
    {
      "epoch": 0.780936684521357,
      "grad_norm": 0.6131373643875122,
      "learning_rate": 1.2202830836883332e-05,
      "loss": 0.7788,
      "step": 2843
    },
    {
      "epoch": 0.7812113720642769,
      "grad_norm": 0.5657122731208801,
      "learning_rate": 1.220008245155971e-05,
      "loss": 0.9172,
      "step": 2844
    },
    {
      "epoch": 0.7814860596071969,
      "grad_norm": 0.5362435579299927,
      "learning_rate": 1.2197334066236088e-05,
      "loss": 0.7196,
      "step": 2845
    },
    {
      "epoch": 0.7817607471501168,
      "grad_norm": 0.5777600407600403,
      "learning_rate": 1.2194585680912465e-05,
      "loss": 0.7742,
      "step": 2846
    },
    {
      "epoch": 0.7820354346930367,
      "grad_norm": 0.36044904589653015,
      "learning_rate": 1.2191837295588842e-05,
      "loss": 1.0511,
      "step": 2847
    },
    {
      "epoch": 0.7823101222359566,
      "grad_norm": 0.5003953576087952,
      "learning_rate": 1.2189088910265221e-05,
      "loss": 0.7998,
      "step": 2848
    },
    {
      "epoch": 0.7825848097788766,
      "grad_norm": 0.422465980052948,
      "learning_rate": 1.2186340524941598e-05,
      "loss": 0.8204,
      "step": 2849
    },
    {
      "epoch": 0.7828594973217965,
      "grad_norm": 0.3659236431121826,
      "learning_rate": 1.2183592139617975e-05,
      "loss": 0.5549,
      "step": 2850
    },
    {
      "epoch": 0.7831341848647164,
      "grad_norm": 0.5041361451148987,
      "learning_rate": 1.2180843754294354e-05,
      "loss": 0.8933,
      "step": 2851
    },
    {
      "epoch": 0.7834088724076363,
      "grad_norm": 0.47476115822792053,
      "learning_rate": 1.217809536897073e-05,
      "loss": 0.638,
      "step": 2852
    },
    {
      "epoch": 0.7836835599505563,
      "grad_norm": 0.4850045442581177,
      "learning_rate": 1.217534698364711e-05,
      "loss": 0.7333,
      "step": 2853
    },
    {
      "epoch": 0.7839582474934762,
      "grad_norm": 0.6448707580566406,
      "learning_rate": 1.2172598598323486e-05,
      "loss": 0.9188,
      "step": 2854
    },
    {
      "epoch": 0.7842329350363961,
      "grad_norm": 0.5617122054100037,
      "learning_rate": 1.2169850212999863e-05,
      "loss": 0.6247,
      "step": 2855
    },
    {
      "epoch": 0.784507622579316,
      "grad_norm": 0.4472850263118744,
      "learning_rate": 1.2167101827676242e-05,
      "loss": 0.718,
      "step": 2856
    },
    {
      "epoch": 0.784782310122236,
      "grad_norm": 0.25152671337127686,
      "learning_rate": 1.216435344235262e-05,
      "loss": 0.5829,
      "step": 2857
    },
    {
      "epoch": 0.7850569976651559,
      "grad_norm": 0.5382258892059326,
      "learning_rate": 1.2161605057028996e-05,
      "loss": 0.5763,
      "step": 2858
    },
    {
      "epoch": 0.7853316852080758,
      "grad_norm": 0.42956727743148804,
      "learning_rate": 1.2158856671705375e-05,
      "loss": 0.8538,
      "step": 2859
    },
    {
      "epoch": 0.7856063727509958,
      "grad_norm": 0.30421045422554016,
      "learning_rate": 1.2156108286381752e-05,
      "loss": 0.6846,
      "step": 2860
    },
    {
      "epoch": 0.7858810602939157,
      "grad_norm": 0.38191866874694824,
      "learning_rate": 1.2153359901058129e-05,
      "loss": 0.5422,
      "step": 2861
    },
    {
      "epoch": 0.7861557478368356,
      "grad_norm": 0.44091886281967163,
      "learning_rate": 1.2150611515734508e-05,
      "loss": 0.8085,
      "step": 2862
    },
    {
      "epoch": 0.7864304353797555,
      "grad_norm": 0.4677603542804718,
      "learning_rate": 1.2147863130410885e-05,
      "loss": 0.6564,
      "step": 2863
    },
    {
      "epoch": 0.7867051229226755,
      "grad_norm": 0.4926683008670807,
      "learning_rate": 1.2145114745087263e-05,
      "loss": 0.7409,
      "step": 2864
    },
    {
      "epoch": 0.7869798104655954,
      "grad_norm": 0.3388977348804474,
      "learning_rate": 1.214236635976364e-05,
      "loss": 0.7922,
      "step": 2865
    },
    {
      "epoch": 0.7872544980085153,
      "grad_norm": 0.4609636664390564,
      "learning_rate": 1.2139617974440017e-05,
      "loss": 0.8606,
      "step": 2866
    },
    {
      "epoch": 0.7875291855514353,
      "grad_norm": 0.5832532644271851,
      "learning_rate": 1.2136869589116396e-05,
      "loss": 0.9186,
      "step": 2867
    },
    {
      "epoch": 0.7878038730943552,
      "grad_norm": 0.49118879437446594,
      "learning_rate": 1.2134121203792773e-05,
      "loss": 0.638,
      "step": 2868
    },
    {
      "epoch": 0.7880785606372751,
      "grad_norm": 0.4461996555328369,
      "learning_rate": 1.2131372818469149e-05,
      "loss": 0.9568,
      "step": 2869
    },
    {
      "epoch": 0.788353248180195,
      "grad_norm": 0.4763333797454834,
      "learning_rate": 1.2128624433145529e-05,
      "loss": 0.8265,
      "step": 2870
    },
    {
      "epoch": 0.788627935723115,
      "grad_norm": 0.3654263913631439,
      "learning_rate": 1.2125876047821904e-05,
      "loss": 0.8044,
      "step": 2871
    },
    {
      "epoch": 0.7889026232660349,
      "grad_norm": 0.3531612753868103,
      "learning_rate": 1.2123127662498285e-05,
      "loss": 0.7935,
      "step": 2872
    },
    {
      "epoch": 0.7891773108089548,
      "grad_norm": 0.5414294004440308,
      "learning_rate": 1.2120379277174662e-05,
      "loss": 0.7133,
      "step": 2873
    },
    {
      "epoch": 0.7894519983518747,
      "grad_norm": 0.3803199529647827,
      "learning_rate": 1.2117630891851037e-05,
      "loss": 0.95,
      "step": 2874
    },
    {
      "epoch": 0.7897266858947947,
      "grad_norm": 0.46596759557724,
      "learning_rate": 1.2114882506527417e-05,
      "loss": 0.7336,
      "step": 2875
    },
    {
      "epoch": 0.7900013734377146,
      "grad_norm": 0.39684563875198364,
      "learning_rate": 1.2112134121203793e-05,
      "loss": 0.874,
      "step": 2876
    },
    {
      "epoch": 0.7902760609806345,
      "grad_norm": 0.41522282361984253,
      "learning_rate": 1.210938573588017e-05,
      "loss": 0.7311,
      "step": 2877
    },
    {
      "epoch": 0.7905507485235544,
      "grad_norm": 0.4401734173297882,
      "learning_rate": 1.2106637350556549e-05,
      "loss": 0.9215,
      "step": 2878
    },
    {
      "epoch": 0.7908254360664744,
      "grad_norm": 0.34901008009910583,
      "learning_rate": 1.2103888965232926e-05,
      "loss": 0.9571,
      "step": 2879
    },
    {
      "epoch": 0.7911001236093943,
      "grad_norm": 0.40214142203330994,
      "learning_rate": 1.2101140579909306e-05,
      "loss": 1.014,
      "step": 2880
    },
    {
      "epoch": 0.7913748111523142,
      "grad_norm": 0.47265928983688354,
      "learning_rate": 1.2098392194585681e-05,
      "loss": 1.028,
      "step": 2881
    },
    {
      "epoch": 0.7916494986952342,
      "grad_norm": 0.35457611083984375,
      "learning_rate": 1.2095643809262058e-05,
      "loss": 0.7248,
      "step": 2882
    },
    {
      "epoch": 0.7919241862381541,
      "grad_norm": 0.4931648075580597,
      "learning_rate": 1.2092895423938437e-05,
      "loss": 0.955,
      "step": 2883
    },
    {
      "epoch": 0.792198873781074,
      "grad_norm": 0.49492549896240234,
      "learning_rate": 1.2090147038614814e-05,
      "loss": 0.8108,
      "step": 2884
    },
    {
      "epoch": 0.7924735613239939,
      "grad_norm": 0.3361770510673523,
      "learning_rate": 1.2087398653291191e-05,
      "loss": 0.8545,
      "step": 2885
    },
    {
      "epoch": 0.7927482488669139,
      "grad_norm": 0.48249298334121704,
      "learning_rate": 1.208465026796757e-05,
      "loss": 0.744,
      "step": 2886
    },
    {
      "epoch": 0.7930229364098338,
      "grad_norm": 0.415426641702652,
      "learning_rate": 1.2081901882643947e-05,
      "loss": 0.7765,
      "step": 2887
    },
    {
      "epoch": 0.7932976239527537,
      "grad_norm": 0.4058968424797058,
      "learning_rate": 1.2079153497320326e-05,
      "loss": 0.9529,
      "step": 2888
    },
    {
      "epoch": 0.7935723114956736,
      "grad_norm": 0.43363896012306213,
      "learning_rate": 1.2076405111996703e-05,
      "loss": 0.7839,
      "step": 2889
    },
    {
      "epoch": 0.7938469990385936,
      "grad_norm": 0.40511682629585266,
      "learning_rate": 1.207365672667308e-05,
      "loss": 0.7203,
      "step": 2890
    },
    {
      "epoch": 0.7941216865815135,
      "grad_norm": 0.4767276346683502,
      "learning_rate": 1.2070908341349458e-05,
      "loss": 0.8749,
      "step": 2891
    },
    {
      "epoch": 0.7943963741244334,
      "grad_norm": 0.36034122109413147,
      "learning_rate": 1.2068159956025835e-05,
      "loss": 0.4979,
      "step": 2892
    },
    {
      "epoch": 0.7946710616673534,
      "grad_norm": 0.34237194061279297,
      "learning_rate": 1.2065411570702212e-05,
      "loss": 0.7947,
      "step": 2893
    },
    {
      "epoch": 0.7949457492102733,
      "grad_norm": 0.3820399343967438,
      "learning_rate": 1.2062663185378591e-05,
      "loss": 0.7653,
      "step": 2894
    },
    {
      "epoch": 0.7952204367531932,
      "grad_norm": 0.4631752371788025,
      "learning_rate": 1.2059914800054968e-05,
      "loss": 0.7407,
      "step": 2895
    },
    {
      "epoch": 0.7954951242961131,
      "grad_norm": 0.35353323817253113,
      "learning_rate": 1.2057166414731347e-05,
      "loss": 0.6235,
      "step": 2896
    },
    {
      "epoch": 0.7957698118390331,
      "grad_norm": 0.35757946968078613,
      "learning_rate": 1.2054418029407724e-05,
      "loss": 0.751,
      "step": 2897
    },
    {
      "epoch": 0.796044499381953,
      "grad_norm": 0.41555097699165344,
      "learning_rate": 1.20516696440841e-05,
      "loss": 0.8477,
      "step": 2898
    },
    {
      "epoch": 0.7963191869248729,
      "grad_norm": 0.4046950340270996,
      "learning_rate": 1.204892125876048e-05,
      "loss": 0.9032,
      "step": 2899
    },
    {
      "epoch": 0.7965938744677928,
      "grad_norm": 0.49568188190460205,
      "learning_rate": 1.2046172873436857e-05,
      "loss": 0.9214,
      "step": 2900
    },
    {
      "epoch": 0.7968685620107128,
      "grad_norm": 0.47973552346229553,
      "learning_rate": 1.2043424488113234e-05,
      "loss": 0.7199,
      "step": 2901
    },
    {
      "epoch": 0.7971432495536327,
      "grad_norm": 0.49843722581863403,
      "learning_rate": 1.2040676102789612e-05,
      "loss": 0.8264,
      "step": 2902
    },
    {
      "epoch": 0.7974179370965526,
      "grad_norm": 0.5633012652397156,
      "learning_rate": 1.203792771746599e-05,
      "loss": 1.0019,
      "step": 2903
    },
    {
      "epoch": 0.7976926246394725,
      "grad_norm": 0.34044748544692993,
      "learning_rate": 1.2035179332142368e-05,
      "loss": 0.9255,
      "step": 2904
    },
    {
      "epoch": 0.7979673121823925,
      "grad_norm": 0.4005977213382721,
      "learning_rate": 1.2032430946818745e-05,
      "loss": 0.8462,
      "step": 2905
    },
    {
      "epoch": 0.7982419997253125,
      "grad_norm": 0.4951169788837433,
      "learning_rate": 1.2029682561495122e-05,
      "loss": 0.7251,
      "step": 2906
    },
    {
      "epoch": 0.7985166872682324,
      "grad_norm": 0.5321128964424133,
      "learning_rate": 1.20269341761715e-05,
      "loss": 0.7849,
      "step": 2907
    },
    {
      "epoch": 0.7987913748111524,
      "grad_norm": 0.37196049094200134,
      "learning_rate": 1.2024185790847878e-05,
      "loss": 0.8867,
      "step": 2908
    },
    {
      "epoch": 0.7990660623540723,
      "grad_norm": 0.40556469559669495,
      "learning_rate": 1.2021437405524255e-05,
      "loss": 0.9448,
      "step": 2909
    },
    {
      "epoch": 0.7993407498969922,
      "grad_norm": 0.3291759192943573,
      "learning_rate": 1.2018689020200634e-05,
      "loss": 0.9829,
      "step": 2910
    },
    {
      "epoch": 0.7996154374399121,
      "grad_norm": 0.5107349157333374,
      "learning_rate": 1.201594063487701e-05,
      "loss": 0.7874,
      "step": 2911
    },
    {
      "epoch": 0.7998901249828321,
      "grad_norm": 0.4955521523952484,
      "learning_rate": 1.201319224955339e-05,
      "loss": 0.7407,
      "step": 2912
    },
    {
      "epoch": 0.800164812525752,
      "grad_norm": 0.4039120078086853,
      "learning_rate": 1.2010443864229766e-05,
      "loss": 0.6909,
      "step": 2913
    },
    {
      "epoch": 0.8004395000686719,
      "grad_norm": 0.3856528103351593,
      "learning_rate": 1.2007695478906143e-05,
      "loss": 0.6862,
      "step": 2914
    },
    {
      "epoch": 0.8007141876115919,
      "grad_norm": 0.4010237753391266,
      "learning_rate": 1.2004947093582522e-05,
      "loss": 0.792,
      "step": 2915
    },
    {
      "epoch": 0.8009888751545118,
      "grad_norm": 0.32858023047447205,
      "learning_rate": 1.2002198708258899e-05,
      "loss": 0.5907,
      "step": 2916
    },
    {
      "epoch": 0.8012635626974317,
      "grad_norm": 0.3746885061264038,
      "learning_rate": 1.1999450322935276e-05,
      "loss": 0.9345,
      "step": 2917
    },
    {
      "epoch": 0.8015382502403516,
      "grad_norm": 0.3815154433250427,
      "learning_rate": 1.1996701937611655e-05,
      "loss": 1.005,
      "step": 2918
    },
    {
      "epoch": 0.8018129377832716,
      "grad_norm": 0.5492666959762573,
      "learning_rate": 1.1993953552288032e-05,
      "loss": 0.7166,
      "step": 2919
    },
    {
      "epoch": 0.8020876253261915,
      "grad_norm": 0.405931681394577,
      "learning_rate": 1.199120516696441e-05,
      "loss": 1.0878,
      "step": 2920
    },
    {
      "epoch": 0.8023623128691114,
      "grad_norm": 0.40201571583747864,
      "learning_rate": 1.1988456781640788e-05,
      "loss": 0.7054,
      "step": 2921
    },
    {
      "epoch": 0.8026370004120313,
      "grad_norm": 0.5080817341804504,
      "learning_rate": 1.1985708396317165e-05,
      "loss": 0.6143,
      "step": 2922
    },
    {
      "epoch": 0.8029116879549513,
      "grad_norm": 0.438705712556839,
      "learning_rate": 1.1982960010993543e-05,
      "loss": 0.8209,
      "step": 2923
    },
    {
      "epoch": 0.8031863754978712,
      "grad_norm": 0.5048743486404419,
      "learning_rate": 1.198021162566992e-05,
      "loss": 0.6773,
      "step": 2924
    },
    {
      "epoch": 0.8034610630407911,
      "grad_norm": 0.36271870136260986,
      "learning_rate": 1.1977463240346297e-05,
      "loss": 0.8386,
      "step": 2925
    },
    {
      "epoch": 0.803735750583711,
      "grad_norm": 0.4847554862499237,
      "learning_rate": 1.1974714855022676e-05,
      "loss": 0.8664,
      "step": 2926
    },
    {
      "epoch": 0.804010438126631,
      "grad_norm": 0.4133825898170471,
      "learning_rate": 1.1971966469699053e-05,
      "loss": 0.863,
      "step": 2927
    },
    {
      "epoch": 0.8042851256695509,
      "grad_norm": 0.3796490728855133,
      "learning_rate": 1.1969218084375432e-05,
      "loss": 0.9494,
      "step": 2928
    },
    {
      "epoch": 0.8045598132124708,
      "grad_norm": 0.4316178262233734,
      "learning_rate": 1.1966469699051809e-05,
      "loss": 0.7844,
      "step": 2929
    },
    {
      "epoch": 0.8048345007553908,
      "grad_norm": 0.4414903223514557,
      "learning_rate": 1.1963721313728184e-05,
      "loss": 0.8317,
      "step": 2930
    },
    {
      "epoch": 0.8051091882983107,
      "grad_norm": 0.3410714566707611,
      "learning_rate": 1.1960972928404565e-05,
      "loss": 0.8234,
      "step": 2931
    },
    {
      "epoch": 0.8053838758412306,
      "grad_norm": 0.31910577416419983,
      "learning_rate": 1.1958224543080942e-05,
      "loss": 0.6433,
      "step": 2932
    },
    {
      "epoch": 0.8056585633841505,
      "grad_norm": 0.42890414595603943,
      "learning_rate": 1.1955476157757317e-05,
      "loss": 0.5968,
      "step": 2933
    },
    {
      "epoch": 0.8059332509270705,
      "grad_norm": 0.5803917050361633,
      "learning_rate": 1.1952727772433697e-05,
      "loss": 0.7347,
      "step": 2934
    },
    {
      "epoch": 0.8062079384699904,
      "grad_norm": 0.4389863908290863,
      "learning_rate": 1.1949979387110073e-05,
      "loss": 0.7372,
      "step": 2935
    },
    {
      "epoch": 0.8064826260129103,
      "grad_norm": 0.405086874961853,
      "learning_rate": 1.1947231001786453e-05,
      "loss": 0.4884,
      "step": 2936
    },
    {
      "epoch": 0.8067573135558302,
      "grad_norm": 0.3392961323261261,
      "learning_rate": 1.1944482616462828e-05,
      "loss": 0.6886,
      "step": 2937
    },
    {
      "epoch": 0.8070320010987502,
      "grad_norm": 0.4596550464630127,
      "learning_rate": 1.1941734231139205e-05,
      "loss": 0.8147,
      "step": 2938
    },
    {
      "epoch": 0.8073066886416701,
      "grad_norm": 0.509949803352356,
      "learning_rate": 1.1938985845815586e-05,
      "loss": 0.6255,
      "step": 2939
    },
    {
      "epoch": 0.80758137618459,
      "grad_norm": 0.4427347779273987,
      "learning_rate": 1.1936237460491961e-05,
      "loss": 0.6555,
      "step": 2940
    },
    {
      "epoch": 0.80785606372751,
      "grad_norm": 0.4757686257362366,
      "learning_rate": 1.1933489075168338e-05,
      "loss": 0.7348,
      "step": 2941
    },
    {
      "epoch": 0.8081307512704299,
      "grad_norm": 0.4560914635658264,
      "learning_rate": 1.1930740689844717e-05,
      "loss": 0.9902,
      "step": 2942
    },
    {
      "epoch": 0.8084054388133498,
      "grad_norm": 0.4319315552711487,
      "learning_rate": 1.1927992304521094e-05,
      "loss": 0.7026,
      "step": 2943
    },
    {
      "epoch": 0.8086801263562697,
      "grad_norm": 0.35273709893226624,
      "learning_rate": 1.1925243919197473e-05,
      "loss": 0.7673,
      "step": 2944
    },
    {
      "epoch": 0.8089548138991897,
      "grad_norm": 0.403579980134964,
      "learning_rate": 1.192249553387385e-05,
      "loss": 0.7673,
      "step": 2945
    },
    {
      "epoch": 0.8092295014421096,
      "grad_norm": 0.3337711691856384,
      "learning_rate": 1.1919747148550227e-05,
      "loss": 0.6204,
      "step": 2946
    },
    {
      "epoch": 0.8095041889850295,
      "grad_norm": 0.48716408014297485,
      "learning_rate": 1.1916998763226605e-05,
      "loss": 0.8657,
      "step": 2947
    },
    {
      "epoch": 0.8097788765279494,
      "grad_norm": 0.3411198556423187,
      "learning_rate": 1.1914250377902982e-05,
      "loss": 0.7776,
      "step": 2948
    },
    {
      "epoch": 0.8100535640708694,
      "grad_norm": 0.5106936693191528,
      "learning_rate": 1.191150199257936e-05,
      "loss": 0.9082,
      "step": 2949
    },
    {
      "epoch": 0.8103282516137893,
      "grad_norm": 0.42391616106033325,
      "learning_rate": 1.1908753607255738e-05,
      "loss": 0.5818,
      "step": 2950
    },
    {
      "epoch": 0.8106029391567092,
      "grad_norm": 0.4005684554576874,
      "learning_rate": 1.1906005221932115e-05,
      "loss": 0.9625,
      "step": 2951
    },
    {
      "epoch": 0.8108776266996292,
      "grad_norm": 0.39401325583457947,
      "learning_rate": 1.1903256836608494e-05,
      "loss": 0.7192,
      "step": 2952
    },
    {
      "epoch": 0.8111523142425491,
      "grad_norm": 0.4275623857975006,
      "learning_rate": 1.190050845128487e-05,
      "loss": 0.7092,
      "step": 2953
    },
    {
      "epoch": 0.811427001785469,
      "grad_norm": 0.47644898295402527,
      "learning_rate": 1.1897760065961248e-05,
      "loss": 0.6974,
      "step": 2954
    },
    {
      "epoch": 0.8117016893283889,
      "grad_norm": 0.4449723958969116,
      "learning_rate": 1.1895011680637627e-05,
      "loss": 0.7557,
      "step": 2955
    },
    {
      "epoch": 0.8119763768713089,
      "grad_norm": 0.46816927194595337,
      "learning_rate": 1.1892263295314004e-05,
      "loss": 0.8267,
      "step": 2956
    },
    {
      "epoch": 0.8122510644142288,
      "grad_norm": 0.3641548454761505,
      "learning_rate": 1.188951490999038e-05,
      "loss": 0.702,
      "step": 2957
    },
    {
      "epoch": 0.8125257519571487,
      "grad_norm": 0.36956191062927246,
      "learning_rate": 1.188676652466676e-05,
      "loss": 0.9572,
      "step": 2958
    },
    {
      "epoch": 0.8128004395000686,
      "grad_norm": 0.5848672389984131,
      "learning_rate": 1.1884018139343136e-05,
      "loss": 0.9081,
      "step": 2959
    },
    {
      "epoch": 0.8130751270429886,
      "grad_norm": 0.3303700387477875,
      "learning_rate": 1.1881269754019515e-05,
      "loss": 0.9424,
      "step": 2960
    },
    {
      "epoch": 0.8133498145859085,
      "grad_norm": 0.4375174343585968,
      "learning_rate": 1.1878521368695892e-05,
      "loss": 0.7791,
      "step": 2961
    },
    {
      "epoch": 0.8136245021288284,
      "grad_norm": 0.3216687738895416,
      "learning_rate": 1.1875772983372269e-05,
      "loss": 0.7555,
      "step": 2962
    },
    {
      "epoch": 0.8138991896717483,
      "grad_norm": 0.46480774879455566,
      "learning_rate": 1.1873024598048648e-05,
      "loss": 0.7734,
      "step": 2963
    },
    {
      "epoch": 0.8141738772146683,
      "grad_norm": 0.47249168157577515,
      "learning_rate": 1.1870276212725025e-05,
      "loss": 0.5836,
      "step": 2964
    },
    {
      "epoch": 0.8144485647575882,
      "grad_norm": 0.4837609529495239,
      "learning_rate": 1.1867527827401402e-05,
      "loss": 0.8981,
      "step": 2965
    },
    {
      "epoch": 0.8147232523005081,
      "grad_norm": 0.28863024711608887,
      "learning_rate": 1.186477944207778e-05,
      "loss": 0.5756,
      "step": 2966
    },
    {
      "epoch": 0.814997939843428,
      "grad_norm": 0.5195584297180176,
      "learning_rate": 1.1862031056754158e-05,
      "loss": 0.6045,
      "step": 2967
    },
    {
      "epoch": 0.815272627386348,
      "grad_norm": 0.4132479131221771,
      "learning_rate": 1.1859282671430536e-05,
      "loss": 0.9685,
      "step": 2968
    },
    {
      "epoch": 0.8155473149292679,
      "grad_norm": 0.45716366171836853,
      "learning_rate": 1.1856534286106913e-05,
      "loss": 0.9127,
      "step": 2969
    },
    {
      "epoch": 0.8158220024721878,
      "grad_norm": 0.5252676606178284,
      "learning_rate": 1.185378590078329e-05,
      "loss": 0.7715,
      "step": 2970
    },
    {
      "epoch": 0.8160966900151078,
      "grad_norm": 0.5384502410888672,
      "learning_rate": 1.1851037515459669e-05,
      "loss": 0.7149,
      "step": 2971
    },
    {
      "epoch": 0.8163713775580277,
      "grad_norm": 0.2932322025299072,
      "learning_rate": 1.1848289130136046e-05,
      "loss": 0.7191,
      "step": 2972
    },
    {
      "epoch": 0.8166460651009477,
      "grad_norm": 0.312090128660202,
      "learning_rate": 1.1845540744812423e-05,
      "loss": 0.9851,
      "step": 2973
    },
    {
      "epoch": 0.8169207526438677,
      "grad_norm": 0.37411949038505554,
      "learning_rate": 1.1842792359488802e-05,
      "loss": 0.6983,
      "step": 2974
    },
    {
      "epoch": 0.8171954401867876,
      "grad_norm": 0.42768168449401855,
      "learning_rate": 1.1840043974165179e-05,
      "loss": 0.5914,
      "step": 2975
    },
    {
      "epoch": 0.8174701277297075,
      "grad_norm": 0.4848911762237549,
      "learning_rate": 1.1837295588841558e-05,
      "loss": 0.8943,
      "step": 2976
    },
    {
      "epoch": 0.8177448152726274,
      "grad_norm": 0.47556495666503906,
      "learning_rate": 1.1834547203517935e-05,
      "loss": 0.6355,
      "step": 2977
    },
    {
      "epoch": 0.8180195028155474,
      "grad_norm": 0.3392391800880432,
      "learning_rate": 1.1831798818194312e-05,
      "loss": 1.0316,
      "step": 2978
    },
    {
      "epoch": 0.8182941903584673,
      "grad_norm": 0.3510320484638214,
      "learning_rate": 1.182905043287069e-05,
      "loss": 0.6088,
      "step": 2979
    },
    {
      "epoch": 0.8185688779013872,
      "grad_norm": 0.3817397654056549,
      "learning_rate": 1.1826302047547067e-05,
      "loss": 0.7799,
      "step": 2980
    },
    {
      "epoch": 0.8188435654443071,
      "grad_norm": 0.3366994261741638,
      "learning_rate": 1.1823553662223444e-05,
      "loss": 0.8519,
      "step": 2981
    },
    {
      "epoch": 0.8191182529872271,
      "grad_norm": 0.39042040705680847,
      "learning_rate": 1.1820805276899823e-05,
      "loss": 0.5656,
      "step": 2982
    },
    {
      "epoch": 0.819392940530147,
      "grad_norm": 0.39717748761177063,
      "learning_rate": 1.18180568915762e-05,
      "loss": 0.9798,
      "step": 2983
    },
    {
      "epoch": 0.8196676280730669,
      "grad_norm": 0.32317256927490234,
      "learning_rate": 1.1815308506252579e-05,
      "loss": 0.5513,
      "step": 2984
    },
    {
      "epoch": 0.8199423156159868,
      "grad_norm": 0.42637982964515686,
      "learning_rate": 1.1812560120928956e-05,
      "loss": 0.9405,
      "step": 2985
    },
    {
      "epoch": 0.8202170031589068,
      "grad_norm": 0.34767454862594604,
      "learning_rate": 1.1809811735605333e-05,
      "loss": 0.5318,
      "step": 2986
    },
    {
      "epoch": 0.8204916907018267,
      "grad_norm": 0.5396771430969238,
      "learning_rate": 1.1807063350281712e-05,
      "loss": 0.6692,
      "step": 2987
    },
    {
      "epoch": 0.8207663782447466,
      "grad_norm": 0.557276725769043,
      "learning_rate": 1.1804314964958089e-05,
      "loss": 0.7886,
      "step": 2988
    },
    {
      "epoch": 0.8210410657876666,
      "grad_norm": 0.38026368618011475,
      "learning_rate": 1.1801566579634464e-05,
      "loss": 0.6405,
      "step": 2989
    },
    {
      "epoch": 0.8213157533305865,
      "grad_norm": 0.4298712909221649,
      "learning_rate": 1.1798818194310844e-05,
      "loss": 1.0706,
      "step": 2990
    },
    {
      "epoch": 0.8215904408735064,
      "grad_norm": 0.4503450393676758,
      "learning_rate": 1.1796069808987221e-05,
      "loss": 0.9605,
      "step": 2991
    },
    {
      "epoch": 0.8218651284164263,
      "grad_norm": 0.288457453250885,
      "learning_rate": 1.17933214236636e-05,
      "loss": 0.5316,
      "step": 2992
    },
    {
      "epoch": 0.8221398159593463,
      "grad_norm": 0.39268627762794495,
      "learning_rate": 1.1790573038339977e-05,
      "loss": 0.6798,
      "step": 2993
    },
    {
      "epoch": 0.8224145035022662,
      "grad_norm": 0.42928826808929443,
      "learning_rate": 1.1787824653016352e-05,
      "loss": 0.774,
      "step": 2994
    },
    {
      "epoch": 0.8226891910451861,
      "grad_norm": 0.4681522250175476,
      "learning_rate": 1.1785076267692733e-05,
      "loss": 0.8104,
      "step": 2995
    },
    {
      "epoch": 0.822963878588106,
      "grad_norm": 0.47222596406936646,
      "learning_rate": 1.1782327882369108e-05,
      "loss": 0.655,
      "step": 2996
    },
    {
      "epoch": 0.823238566131026,
      "grad_norm": 0.4950777590274811,
      "learning_rate": 1.1779579497045485e-05,
      "loss": 0.8087,
      "step": 2997
    },
    {
      "epoch": 0.8235132536739459,
      "grad_norm": 0.3705980181694031,
      "learning_rate": 1.1776831111721866e-05,
      "loss": 0.712,
      "step": 2998
    },
    {
      "epoch": 0.8237879412168658,
      "grad_norm": 0.3644186556339264,
      "learning_rate": 1.1774082726398241e-05,
      "loss": 0.7307,
      "step": 2999
    },
    {
      "epoch": 0.8240626287597858,
      "grad_norm": 0.45935943722724915,
      "learning_rate": 1.1771334341074621e-05,
      "loss": 0.9541,
      "step": 3000
    },
    {
      "epoch": 0.8243373163027057,
      "grad_norm": 0.36076226830482483,
      "learning_rate": 1.1768585955750997e-05,
      "loss": 1.0309,
      "step": 3001
    },
    {
      "epoch": 0.8246120038456256,
      "grad_norm": 0.3176887631416321,
      "learning_rate": 1.1765837570427374e-05,
      "loss": 0.6295,
      "step": 3002
    },
    {
      "epoch": 0.8248866913885455,
      "grad_norm": 0.4514355957508087,
      "learning_rate": 1.1763089185103752e-05,
      "loss": 0.5932,
      "step": 3003
    },
    {
      "epoch": 0.8251613789314655,
      "grad_norm": 0.5228477120399475,
      "learning_rate": 1.176034079978013e-05,
      "loss": 0.8905,
      "step": 3004
    },
    {
      "epoch": 0.8254360664743854,
      "grad_norm": 0.5135506391525269,
      "learning_rate": 1.1757592414456506e-05,
      "loss": 0.6967,
      "step": 3005
    },
    {
      "epoch": 0.8257107540173053,
      "grad_norm": 0.4420495629310608,
      "learning_rate": 1.1754844029132885e-05,
      "loss": 0.8415,
      "step": 3006
    },
    {
      "epoch": 0.8259854415602252,
      "grad_norm": 0.3707277774810791,
      "learning_rate": 1.1752095643809262e-05,
      "loss": 0.9119,
      "step": 3007
    },
    {
      "epoch": 0.8262601291031452,
      "grad_norm": 0.38196709752082825,
      "learning_rate": 1.174934725848564e-05,
      "loss": 0.7266,
      "step": 3008
    },
    {
      "epoch": 0.8265348166460651,
      "grad_norm": 0.5202895998954773,
      "learning_rate": 1.1746598873162018e-05,
      "loss": 0.6227,
      "step": 3009
    },
    {
      "epoch": 0.826809504188985,
      "grad_norm": 0.4711443781852722,
      "learning_rate": 1.1743850487838395e-05,
      "loss": 0.6238,
      "step": 3010
    },
    {
      "epoch": 0.827084191731905,
      "grad_norm": 0.5103205442428589,
      "learning_rate": 1.1741102102514774e-05,
      "loss": 0.8677,
      "step": 3011
    },
    {
      "epoch": 0.8273588792748249,
      "grad_norm": 0.3960525095462799,
      "learning_rate": 1.173835371719115e-05,
      "loss": 0.7515,
      "step": 3012
    },
    {
      "epoch": 0.8276335668177448,
      "grad_norm": 0.4292113780975342,
      "learning_rate": 1.1735605331867528e-05,
      "loss": 0.5874,
      "step": 3013
    },
    {
      "epoch": 0.8279082543606647,
      "grad_norm": 0.4081830084323883,
      "learning_rate": 1.1732856946543906e-05,
      "loss": 0.6767,
      "step": 3014
    },
    {
      "epoch": 0.8281829419035847,
      "grad_norm": 0.45018303394317627,
      "learning_rate": 1.1730108561220283e-05,
      "loss": 0.6351,
      "step": 3015
    },
    {
      "epoch": 0.8284576294465046,
      "grad_norm": 0.45074641704559326,
      "learning_rate": 1.1727360175896662e-05,
      "loss": 0.8528,
      "step": 3016
    },
    {
      "epoch": 0.8287323169894245,
      "grad_norm": 0.35897448658943176,
      "learning_rate": 1.1724611790573039e-05,
      "loss": 0.5312,
      "step": 3017
    },
    {
      "epoch": 0.8290070045323444,
      "grad_norm": 0.3957553803920746,
      "learning_rate": 1.1721863405249416e-05,
      "loss": 0.6667,
      "step": 3018
    },
    {
      "epoch": 0.8292816920752644,
      "grad_norm": 0.3573541045188904,
      "learning_rate": 1.1719115019925795e-05,
      "loss": 0.823,
      "step": 3019
    },
    {
      "epoch": 0.8295563796181843,
      "grad_norm": 0.2817539870738983,
      "learning_rate": 1.1716366634602172e-05,
      "loss": 0.5731,
      "step": 3020
    },
    {
      "epoch": 0.8298310671611042,
      "grad_norm": 0.4718209505081177,
      "learning_rate": 1.1713618249278549e-05,
      "loss": 0.9488,
      "step": 3021
    },
    {
      "epoch": 0.8301057547040241,
      "grad_norm": 0.4151826798915863,
      "learning_rate": 1.1710869863954928e-05,
      "loss": 0.8236,
      "step": 3022
    },
    {
      "epoch": 0.8303804422469441,
      "grad_norm": 0.5668027997016907,
      "learning_rate": 1.1708121478631305e-05,
      "loss": 0.8329,
      "step": 3023
    },
    {
      "epoch": 0.830655129789864,
      "grad_norm": 0.4221627116203308,
      "learning_rate": 1.1705373093307683e-05,
      "loss": 0.6349,
      "step": 3024
    },
    {
      "epoch": 0.8309298173327839,
      "grad_norm": 0.3915141820907593,
      "learning_rate": 1.170262470798406e-05,
      "loss": 0.8084,
      "step": 3025
    },
    {
      "epoch": 0.8312045048757039,
      "grad_norm": 0.4621562659740448,
      "learning_rate": 1.1699876322660437e-05,
      "loss": 0.8187,
      "step": 3026
    },
    {
      "epoch": 0.8314791924186238,
      "grad_norm": 0.38151320815086365,
      "learning_rate": 1.1697127937336816e-05,
      "loss": 0.7855,
      "step": 3027
    },
    {
      "epoch": 0.8317538799615437,
      "grad_norm": 0.6550028324127197,
      "learning_rate": 1.1694379552013193e-05,
      "loss": 0.853,
      "step": 3028
    },
    {
      "epoch": 0.8320285675044636,
      "grad_norm": 0.40612098574638367,
      "learning_rate": 1.169163116668957e-05,
      "loss": 0.9226,
      "step": 3029
    },
    {
      "epoch": 0.8323032550473836,
      "grad_norm": 0.512313187122345,
      "learning_rate": 1.1688882781365949e-05,
      "loss": 0.9825,
      "step": 3030
    },
    {
      "epoch": 0.8325779425903035,
      "grad_norm": 0.4918607175350189,
      "learning_rate": 1.1686134396042326e-05,
      "loss": 0.8249,
      "step": 3031
    },
    {
      "epoch": 0.8328526301332234,
      "grad_norm": 0.48703718185424805,
      "learning_rate": 1.1683386010718705e-05,
      "loss": 0.7112,
      "step": 3032
    },
    {
      "epoch": 0.8331273176761433,
      "grad_norm": 0.49653732776641846,
      "learning_rate": 1.1680637625395082e-05,
      "loss": 0.8253,
      "step": 3033
    },
    {
      "epoch": 0.8334020052190633,
      "grad_norm": 0.38720932602882385,
      "learning_rate": 1.1677889240071459e-05,
      "loss": 0.6794,
      "step": 3034
    },
    {
      "epoch": 0.8336766927619832,
      "grad_norm": 0.410971999168396,
      "learning_rate": 1.1675140854747837e-05,
      "loss": 0.8084,
      "step": 3035
    },
    {
      "epoch": 0.8339513803049031,
      "grad_norm": 0.3583734631538391,
      "learning_rate": 1.1672392469424214e-05,
      "loss": 0.8911,
      "step": 3036
    },
    {
      "epoch": 0.834226067847823,
      "grad_norm": 0.3453867435455322,
      "learning_rate": 1.1669644084100591e-05,
      "loss": 1.0725,
      "step": 3037
    },
    {
      "epoch": 0.834500755390743,
      "grad_norm": 0.4904463291168213,
      "learning_rate": 1.166689569877697e-05,
      "loss": 0.7214,
      "step": 3038
    },
    {
      "epoch": 0.8347754429336629,
      "grad_norm": 0.34127047657966614,
      "learning_rate": 1.1664147313453347e-05,
      "loss": 0.7794,
      "step": 3039
    },
    {
      "epoch": 0.8350501304765829,
      "grad_norm": 0.45382198691368103,
      "learning_rate": 1.1661398928129726e-05,
      "loss": 0.76,
      "step": 3040
    },
    {
      "epoch": 0.8353248180195029,
      "grad_norm": 0.48800602555274963,
      "learning_rate": 1.1658650542806103e-05,
      "loss": 0.9206,
      "step": 3041
    },
    {
      "epoch": 0.8355995055624228,
      "grad_norm": 0.4633256793022156,
      "learning_rate": 1.165590215748248e-05,
      "loss": 0.7242,
      "step": 3042
    },
    {
      "epoch": 0.8358741931053427,
      "grad_norm": 0.4002763628959656,
      "learning_rate": 1.1653153772158859e-05,
      "loss": 0.5396,
      "step": 3043
    },
    {
      "epoch": 0.8361488806482626,
      "grad_norm": 0.539190411567688,
      "learning_rate": 1.1650405386835236e-05,
      "loss": 0.9132,
      "step": 3044
    },
    {
      "epoch": 0.8364235681911826,
      "grad_norm": 0.48738858103752136,
      "learning_rate": 1.1647657001511613e-05,
      "loss": 0.7538,
      "step": 3045
    },
    {
      "epoch": 0.8366982557341025,
      "grad_norm": 0.32156580686569214,
      "learning_rate": 1.1644908616187991e-05,
      "loss": 0.5259,
      "step": 3046
    },
    {
      "epoch": 0.8369729432770224,
      "grad_norm": 0.37729722261428833,
      "learning_rate": 1.1642160230864368e-05,
      "loss": 0.8313,
      "step": 3047
    },
    {
      "epoch": 0.8372476308199424,
      "grad_norm": 0.4075693190097809,
      "learning_rate": 1.1639411845540747e-05,
      "loss": 0.7792,
      "step": 3048
    },
    {
      "epoch": 0.8375223183628623,
      "grad_norm": 0.48419389128685,
      "learning_rate": 1.1636663460217124e-05,
      "loss": 0.8717,
      "step": 3049
    },
    {
      "epoch": 0.8377970059057822,
      "grad_norm": 0.3837912678718567,
      "learning_rate": 1.1633915074893501e-05,
      "loss": 1.015,
      "step": 3050
    },
    {
      "epoch": 0.8380716934487021,
      "grad_norm": 0.43308067321777344,
      "learning_rate": 1.163116668956988e-05,
      "loss": 0.8907,
      "step": 3051
    },
    {
      "epoch": 0.8383463809916221,
      "grad_norm": 0.44432708621025085,
      "learning_rate": 1.1628418304246257e-05,
      "loss": 0.9795,
      "step": 3052
    },
    {
      "epoch": 0.838621068534542,
      "grad_norm": 0.4011573791503906,
      "learning_rate": 1.1625669918922632e-05,
      "loss": 0.586,
      "step": 3053
    },
    {
      "epoch": 0.8388957560774619,
      "grad_norm": 0.3917035162448883,
      "learning_rate": 1.1622921533599013e-05,
      "loss": 0.528,
      "step": 3054
    },
    {
      "epoch": 0.8391704436203818,
      "grad_norm": 0.38670772314071655,
      "learning_rate": 1.162017314827539e-05,
      "loss": 0.5891,
      "step": 3055
    },
    {
      "epoch": 0.8394451311633018,
      "grad_norm": 0.34910187125205994,
      "learning_rate": 1.1617424762951768e-05,
      "loss": 0.6405,
      "step": 3056
    },
    {
      "epoch": 0.8397198187062217,
      "grad_norm": 0.3530575931072235,
      "learning_rate": 1.1614676377628145e-05,
      "loss": 0.5465,
      "step": 3057
    },
    {
      "epoch": 0.8399945062491416,
      "grad_norm": 0.38631486892700195,
      "learning_rate": 1.161192799230452e-05,
      "loss": 0.7224,
      "step": 3058
    },
    {
      "epoch": 0.8402691937920616,
      "grad_norm": 0.43046852946281433,
      "learning_rate": 1.1609179606980901e-05,
      "loss": 0.864,
      "step": 3059
    },
    {
      "epoch": 0.8405438813349815,
      "grad_norm": 0.40826985239982605,
      "learning_rate": 1.1606431221657276e-05,
      "loss": 0.8048,
      "step": 3060
    },
    {
      "epoch": 0.8408185688779014,
      "grad_norm": 0.4431944489479065,
      "learning_rate": 1.1603682836333653e-05,
      "loss": 0.8404,
      "step": 3061
    },
    {
      "epoch": 0.8410932564208213,
      "grad_norm": 0.4200015962123871,
      "learning_rate": 1.1600934451010034e-05,
      "loss": 0.9435,
      "step": 3062
    },
    {
      "epoch": 0.8413679439637413,
      "grad_norm": 0.42036595940589905,
      "learning_rate": 1.159818606568641e-05,
      "loss": 0.8189,
      "step": 3063
    },
    {
      "epoch": 0.8416426315066612,
      "grad_norm": 0.4014994204044342,
      "learning_rate": 1.159543768036279e-05,
      "loss": 0.7949,
      "step": 3064
    },
    {
      "epoch": 0.8419173190495811,
      "grad_norm": 0.5010548830032349,
      "learning_rate": 1.1592689295039165e-05,
      "loss": 0.7399,
      "step": 3065
    },
    {
      "epoch": 0.842192006592501,
      "grad_norm": 0.4467915892601013,
      "learning_rate": 1.1589940909715542e-05,
      "loss": 0.7892,
      "step": 3066
    },
    {
      "epoch": 0.842466694135421,
      "grad_norm": 0.3889675736427307,
      "learning_rate": 1.158719252439192e-05,
      "loss": 0.5921,
      "step": 3067
    },
    {
      "epoch": 0.8427413816783409,
      "grad_norm": 0.39464884996414185,
      "learning_rate": 1.1584444139068298e-05,
      "loss": 0.5767,
      "step": 3068
    },
    {
      "epoch": 0.8430160692212608,
      "grad_norm": 0.5658791661262512,
      "learning_rate": 1.1581695753744675e-05,
      "loss": 0.6725,
      "step": 3069
    },
    {
      "epoch": 0.8432907567641807,
      "grad_norm": 0.32179468870162964,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 0.6087,
      "step": 3070
    },
    {
      "epoch": 0.8435654443071007,
      "grad_norm": 0.37258270382881165,
      "learning_rate": 1.157619898309743e-05,
      "loss": 0.7279,
      "step": 3071
    },
    {
      "epoch": 0.8438401318500206,
      "grad_norm": 0.4470842480659485,
      "learning_rate": 1.1573450597773809e-05,
      "loss": 0.7464,
      "step": 3072
    },
    {
      "epoch": 0.8441148193929405,
      "grad_norm": 0.4590134620666504,
      "learning_rate": 1.1570702212450186e-05,
      "loss": 0.6697,
      "step": 3073
    },
    {
      "epoch": 0.8443895069358605,
      "grad_norm": 0.4662724435329437,
      "learning_rate": 1.1567953827126563e-05,
      "loss": 0.7669,
      "step": 3074
    },
    {
      "epoch": 0.8446641944787804,
      "grad_norm": 0.5616170167922974,
      "learning_rate": 1.1565205441802942e-05,
      "loss": 0.6854,
      "step": 3075
    },
    {
      "epoch": 0.8449388820217003,
      "grad_norm": 0.34969329833984375,
      "learning_rate": 1.1562457056479319e-05,
      "loss": 0.7537,
      "step": 3076
    },
    {
      "epoch": 0.8452135695646202,
      "grad_norm": 0.2865254282951355,
      "learning_rate": 1.1559708671155696e-05,
      "loss": 0.8278,
      "step": 3077
    },
    {
      "epoch": 0.8454882571075402,
      "grad_norm": 0.368308961391449,
      "learning_rate": 1.1556960285832075e-05,
      "loss": 0.6419,
      "step": 3078
    },
    {
      "epoch": 0.8457629446504601,
      "grad_norm": 0.4578734040260315,
      "learning_rate": 1.1554211900508452e-05,
      "loss": 0.8278,
      "step": 3079
    },
    {
      "epoch": 0.84603763219338,
      "grad_norm": 0.4127757251262665,
      "learning_rate": 1.155146351518483e-05,
      "loss": 0.8419,
      "step": 3080
    },
    {
      "epoch": 0.8463123197363,
      "grad_norm": 0.3131892681121826,
      "learning_rate": 1.1548715129861207e-05,
      "loss": 0.4094,
      "step": 3081
    },
    {
      "epoch": 0.8465870072792199,
      "grad_norm": 0.4497797191143036,
      "learning_rate": 1.1545966744537584e-05,
      "loss": 0.7082,
      "step": 3082
    },
    {
      "epoch": 0.8468616948221398,
      "grad_norm": 0.3344283103942871,
      "learning_rate": 1.1543218359213963e-05,
      "loss": 0.7115,
      "step": 3083
    },
    {
      "epoch": 0.8471363823650597,
      "grad_norm": 0.43287602066993713,
      "learning_rate": 1.154046997389034e-05,
      "loss": 0.6753,
      "step": 3084
    },
    {
      "epoch": 0.8474110699079797,
      "grad_norm": 0.5588279962539673,
      "learning_rate": 1.1537721588566717e-05,
      "loss": 0.7619,
      "step": 3085
    },
    {
      "epoch": 0.8476857574508996,
      "grad_norm": 0.4427691400051117,
      "learning_rate": 1.1534973203243096e-05,
      "loss": 0.9094,
      "step": 3086
    },
    {
      "epoch": 0.8479604449938195,
      "grad_norm": 0.4623594582080841,
      "learning_rate": 1.1532224817919473e-05,
      "loss": 0.6962,
      "step": 3087
    },
    {
      "epoch": 0.8482351325367394,
      "grad_norm": 0.39457133412361145,
      "learning_rate": 1.1529476432595852e-05,
      "loss": 0.9836,
      "step": 3088
    },
    {
      "epoch": 0.8485098200796594,
      "grad_norm": 0.4549407660961151,
      "learning_rate": 1.1526728047272229e-05,
      "loss": 0.8795,
      "step": 3089
    },
    {
      "epoch": 0.8487845076225793,
      "grad_norm": 0.38868260383605957,
      "learning_rate": 1.1523979661948606e-05,
      "loss": 0.6544,
      "step": 3090
    },
    {
      "epoch": 0.8490591951654992,
      "grad_norm": 0.3944456875324249,
      "learning_rate": 1.1521231276624984e-05,
      "loss": 0.6778,
      "step": 3091
    },
    {
      "epoch": 0.8493338827084191,
      "grad_norm": 0.4517955482006073,
      "learning_rate": 1.1518482891301361e-05,
      "loss": 0.6359,
      "step": 3092
    },
    {
      "epoch": 0.8496085702513391,
      "grad_norm": 0.47844675183296204,
      "learning_rate": 1.1515734505977738e-05,
      "loss": 1.108,
      "step": 3093
    },
    {
      "epoch": 0.849883257794259,
      "grad_norm": 0.30169638991355896,
      "learning_rate": 1.1512986120654117e-05,
      "loss": 0.7355,
      "step": 3094
    },
    {
      "epoch": 0.8501579453371789,
      "grad_norm": 0.4344729483127594,
      "learning_rate": 1.1510237735330494e-05,
      "loss": 0.8797,
      "step": 3095
    },
    {
      "epoch": 0.8504326328800988,
      "grad_norm": 0.38660094141960144,
      "learning_rate": 1.1507489350006873e-05,
      "loss": 0.9966,
      "step": 3096
    },
    {
      "epoch": 0.8507073204230188,
      "grad_norm": 0.45192375779151917,
      "learning_rate": 1.150474096468325e-05,
      "loss": 0.9715,
      "step": 3097
    },
    {
      "epoch": 0.8509820079659387,
      "grad_norm": 0.3869803249835968,
      "learning_rate": 1.1501992579359627e-05,
      "loss": 0.6676,
      "step": 3098
    },
    {
      "epoch": 0.8512566955088586,
      "grad_norm": 0.3616040050983429,
      "learning_rate": 1.1499244194036006e-05,
      "loss": 0.5633,
      "step": 3099
    },
    {
      "epoch": 0.8515313830517786,
      "grad_norm": 0.49830499291419983,
      "learning_rate": 1.1496495808712383e-05,
      "loss": 0.8037,
      "step": 3100
    },
    {
      "epoch": 0.8518060705946985,
      "grad_norm": 0.433014452457428,
      "learning_rate": 1.149374742338876e-05,
      "loss": 1.0075,
      "step": 3101
    },
    {
      "epoch": 0.8520807581376184,
      "grad_norm": 0.3248509168624878,
      "learning_rate": 1.1490999038065138e-05,
      "loss": 0.9225,
      "step": 3102
    },
    {
      "epoch": 0.8523554456805383,
      "grad_norm": 0.43194642663002014,
      "learning_rate": 1.1488250652741515e-05,
      "loss": 0.6523,
      "step": 3103
    },
    {
      "epoch": 0.8526301332234583,
      "grad_norm": 0.44691506028175354,
      "learning_rate": 1.1485502267417894e-05,
      "loss": 0.706,
      "step": 3104
    },
    {
      "epoch": 0.8529048207663782,
      "grad_norm": 0.3730623722076416,
      "learning_rate": 1.1482753882094271e-05,
      "loss": 0.7765,
      "step": 3105
    },
    {
      "epoch": 0.8531795083092981,
      "grad_norm": 0.5383691787719727,
      "learning_rate": 1.1480005496770648e-05,
      "loss": 0.757,
      "step": 3106
    },
    {
      "epoch": 0.8534541958522182,
      "grad_norm": 0.4475453197956085,
      "learning_rate": 1.1477257111447027e-05,
      "loss": 0.8215,
      "step": 3107
    },
    {
      "epoch": 0.8537288833951381,
      "grad_norm": 0.3862798511981964,
      "learning_rate": 1.1474508726123404e-05,
      "loss": 0.8661,
      "step": 3108
    },
    {
      "epoch": 0.854003570938058,
      "grad_norm": 0.34773823618888855,
      "learning_rate": 1.1471760340799781e-05,
      "loss": 0.7313,
      "step": 3109
    },
    {
      "epoch": 0.8542782584809779,
      "grad_norm": 0.4752775728702545,
      "learning_rate": 1.146901195547616e-05,
      "loss": 0.8801,
      "step": 3110
    },
    {
      "epoch": 0.8545529460238979,
      "grad_norm": 0.47360706329345703,
      "learning_rate": 1.1466263570152537e-05,
      "loss": 0.7076,
      "step": 3111
    },
    {
      "epoch": 0.8548276335668178,
      "grad_norm": 0.5632167458534241,
      "learning_rate": 1.1463515184828915e-05,
      "loss": 0.7351,
      "step": 3112
    },
    {
      "epoch": 0.8551023211097377,
      "grad_norm": 0.4351915419101715,
      "learning_rate": 1.1460766799505292e-05,
      "loss": 0.7077,
      "step": 3113
    },
    {
      "epoch": 0.8553770086526576,
      "grad_norm": 0.4137261211872101,
      "learning_rate": 1.145801841418167e-05,
      "loss": 0.7813,
      "step": 3114
    },
    {
      "epoch": 0.8556516961955776,
      "grad_norm": 0.5697649121284485,
      "learning_rate": 1.1455270028858048e-05,
      "loss": 0.7517,
      "step": 3115
    },
    {
      "epoch": 0.8559263837384975,
      "grad_norm": 0.3948134481906891,
      "learning_rate": 1.1452521643534425e-05,
      "loss": 0.7552,
      "step": 3116
    },
    {
      "epoch": 0.8562010712814174,
      "grad_norm": 0.42917710542678833,
      "learning_rate": 1.14497732582108e-05,
      "loss": 0.9166,
      "step": 3117
    },
    {
      "epoch": 0.8564757588243374,
      "grad_norm": 0.5345103144645691,
      "learning_rate": 1.1447024872887181e-05,
      "loss": 0.7329,
      "step": 3118
    },
    {
      "epoch": 0.8567504463672573,
      "grad_norm": 0.45051273703575134,
      "learning_rate": 1.1444276487563556e-05,
      "loss": 0.8184,
      "step": 3119
    },
    {
      "epoch": 0.8570251339101772,
      "grad_norm": 0.4472975730895996,
      "learning_rate": 1.1441528102239937e-05,
      "loss": 0.6922,
      "step": 3120
    },
    {
      "epoch": 0.8572998214530971,
      "grad_norm": 0.45391255617141724,
      "learning_rate": 1.1438779716916314e-05,
      "loss": 0.9978,
      "step": 3121
    },
    {
      "epoch": 0.8575745089960171,
      "grad_norm": 0.4828939735889435,
      "learning_rate": 1.1436031331592689e-05,
      "loss": 0.938,
      "step": 3122
    },
    {
      "epoch": 0.857849196538937,
      "grad_norm": 0.4893663823604584,
      "learning_rate": 1.143328294626907e-05,
      "loss": 0.6684,
      "step": 3123
    },
    {
      "epoch": 0.8581238840818569,
      "grad_norm": 0.5519631505012512,
      "learning_rate": 1.1430534560945445e-05,
      "loss": 1.0098,
      "step": 3124
    },
    {
      "epoch": 0.8583985716247768,
      "grad_norm": 0.41245782375335693,
      "learning_rate": 1.1427786175621822e-05,
      "loss": 0.7705,
      "step": 3125
    },
    {
      "epoch": 0.8586732591676968,
      "grad_norm": 0.4342738091945648,
      "learning_rate": 1.14250377902982e-05,
      "loss": 0.8948,
      "step": 3126
    },
    {
      "epoch": 0.8589479467106167,
      "grad_norm": 0.41424617171287537,
      "learning_rate": 1.1422289404974577e-05,
      "loss": 0.6996,
      "step": 3127
    },
    {
      "epoch": 0.8592226342535366,
      "grad_norm": 0.4482669234275818,
      "learning_rate": 1.1419541019650958e-05,
      "loss": 0.6248,
      "step": 3128
    },
    {
      "epoch": 0.8594973217964565,
      "grad_norm": 0.36194169521331787,
      "learning_rate": 1.1416792634327333e-05,
      "loss": 0.5607,
      "step": 3129
    },
    {
      "epoch": 0.8597720093393765,
      "grad_norm": 0.464934378862381,
      "learning_rate": 1.141404424900371e-05,
      "loss": 0.8371,
      "step": 3130
    },
    {
      "epoch": 0.8600466968822964,
      "grad_norm": 0.4617985785007477,
      "learning_rate": 1.1411295863680089e-05,
      "loss": 0.894,
      "step": 3131
    },
    {
      "epoch": 0.8603213844252163,
      "grad_norm": 0.4209194481372833,
      "learning_rate": 1.1408547478356466e-05,
      "loss": 0.7605,
      "step": 3132
    },
    {
      "epoch": 0.8605960719681363,
      "grad_norm": 0.4490082859992981,
      "learning_rate": 1.1405799093032843e-05,
      "loss": 0.7428,
      "step": 3133
    },
    {
      "epoch": 0.8608707595110562,
      "grad_norm": 0.435513973236084,
      "learning_rate": 1.1403050707709222e-05,
      "loss": 0.8588,
      "step": 3134
    },
    {
      "epoch": 0.8611454470539761,
      "grad_norm": 0.4324148893356323,
      "learning_rate": 1.1400302322385599e-05,
      "loss": 0.9334,
      "step": 3135
    },
    {
      "epoch": 0.861420134596896,
      "grad_norm": 0.4446139335632324,
      "learning_rate": 1.1397553937061977e-05,
      "loss": 0.9098,
      "step": 3136
    },
    {
      "epoch": 0.861694822139816,
      "grad_norm": 0.4419865608215332,
      "learning_rate": 1.1394805551738354e-05,
      "loss": 0.6808,
      "step": 3137
    },
    {
      "epoch": 0.8619695096827359,
      "grad_norm": 0.4547271430492401,
      "learning_rate": 1.1392057166414731e-05,
      "loss": 0.7878,
      "step": 3138
    },
    {
      "epoch": 0.8622441972256558,
      "grad_norm": 0.6273484230041504,
      "learning_rate": 1.138930878109111e-05,
      "loss": 0.7154,
      "step": 3139
    },
    {
      "epoch": 0.8625188847685757,
      "grad_norm": 0.3340033292770386,
      "learning_rate": 1.1386560395767487e-05,
      "loss": 1.075,
      "step": 3140
    },
    {
      "epoch": 0.8627935723114957,
      "grad_norm": 0.3238706588745117,
      "learning_rate": 1.1383812010443864e-05,
      "loss": 0.9486,
      "step": 3141
    },
    {
      "epoch": 0.8630682598544156,
      "grad_norm": 0.46107158064842224,
      "learning_rate": 1.1381063625120243e-05,
      "loss": 0.703,
      "step": 3142
    },
    {
      "epoch": 0.8633429473973355,
      "grad_norm": 0.4031849801540375,
      "learning_rate": 1.137831523979662e-05,
      "loss": 0.692,
      "step": 3143
    },
    {
      "epoch": 0.8636176349402555,
      "grad_norm": 0.4054923355579376,
      "learning_rate": 1.1375566854472999e-05,
      "loss": 0.8145,
      "step": 3144
    },
    {
      "epoch": 0.8638923224831754,
      "grad_norm": 0.4974101185798645,
      "learning_rate": 1.1372818469149376e-05,
      "loss": 0.7097,
      "step": 3145
    },
    {
      "epoch": 0.8641670100260953,
      "grad_norm": 0.345196932554245,
      "learning_rate": 1.1370070083825753e-05,
      "loss": 0.7171,
      "step": 3146
    },
    {
      "epoch": 0.8644416975690152,
      "grad_norm": 0.6699345707893372,
      "learning_rate": 1.1367321698502131e-05,
      "loss": 0.7351,
      "step": 3147
    },
    {
      "epoch": 0.8647163851119352,
      "grad_norm": 0.3587840795516968,
      "learning_rate": 1.1364573313178508e-05,
      "loss": 0.8819,
      "step": 3148
    },
    {
      "epoch": 0.8649910726548551,
      "grad_norm": 0.34946155548095703,
      "learning_rate": 1.1361824927854885e-05,
      "loss": 0.7794,
      "step": 3149
    },
    {
      "epoch": 0.865265760197775,
      "grad_norm": 0.4659494459629059,
      "learning_rate": 1.1359076542531264e-05,
      "loss": 0.6956,
      "step": 3150
    },
    {
      "epoch": 0.8655404477406949,
      "grad_norm": 0.4975051283836365,
      "learning_rate": 1.1356328157207641e-05,
      "loss": 0.6451,
      "step": 3151
    },
    {
      "epoch": 0.8658151352836149,
      "grad_norm": 0.3203812837600708,
      "learning_rate": 1.135357977188402e-05,
      "loss": 0.5559,
      "step": 3152
    },
    {
      "epoch": 0.8660898228265348,
      "grad_norm": 0.48671433329582214,
      "learning_rate": 1.1350831386560397e-05,
      "loss": 0.7804,
      "step": 3153
    },
    {
      "epoch": 0.8663645103694547,
      "grad_norm": 0.42902350425720215,
      "learning_rate": 1.1348083001236774e-05,
      "loss": 0.6184,
      "step": 3154
    },
    {
      "epoch": 0.8666391979123746,
      "grad_norm": 0.34849390387535095,
      "learning_rate": 1.1345334615913153e-05,
      "loss": 0.7807,
      "step": 3155
    },
    {
      "epoch": 0.8669138854552946,
      "grad_norm": 0.4959639012813568,
      "learning_rate": 1.134258623058953e-05,
      "loss": 0.8157,
      "step": 3156
    },
    {
      "epoch": 0.8671885729982145,
      "grad_norm": 0.44072654843330383,
      "learning_rate": 1.1339837845265907e-05,
      "loss": 0.6543,
      "step": 3157
    },
    {
      "epoch": 0.8674632605411344,
      "grad_norm": 0.448758602142334,
      "learning_rate": 1.1337089459942285e-05,
      "loss": 0.7737,
      "step": 3158
    },
    {
      "epoch": 0.8677379480840544,
      "grad_norm": 0.30234986543655396,
      "learning_rate": 1.1334341074618662e-05,
      "loss": 0.8977,
      "step": 3159
    },
    {
      "epoch": 0.8680126356269743,
      "grad_norm": 0.47651103138923645,
      "learning_rate": 1.1331592689295041e-05,
      "loss": 0.6342,
      "step": 3160
    },
    {
      "epoch": 0.8682873231698942,
      "grad_norm": 0.3961711823940277,
      "learning_rate": 1.1328844303971418e-05,
      "loss": 0.789,
      "step": 3161
    },
    {
      "epoch": 0.8685620107128141,
      "grad_norm": 0.4648817181587219,
      "learning_rate": 1.1326095918647795e-05,
      "loss": 0.6614,
      "step": 3162
    },
    {
      "epoch": 0.8688366982557341,
      "grad_norm": 0.33974143862724304,
      "learning_rate": 1.1323347533324174e-05,
      "loss": 0.6418,
      "step": 3163
    },
    {
      "epoch": 0.869111385798654,
      "grad_norm": 0.6454595923423767,
      "learning_rate": 1.1320599148000551e-05,
      "loss": 0.7775,
      "step": 3164
    },
    {
      "epoch": 0.8693860733415739,
      "grad_norm": 0.4367125928401947,
      "learning_rate": 1.1317850762676928e-05,
      "loss": 0.864,
      "step": 3165
    },
    {
      "epoch": 0.8696607608844938,
      "grad_norm": 0.35395655035972595,
      "learning_rate": 1.1315102377353307e-05,
      "loss": 0.5281,
      "step": 3166
    },
    {
      "epoch": 0.8699354484274138,
      "grad_norm": 0.3497166633605957,
      "learning_rate": 1.1312353992029684e-05,
      "loss": 0.6489,
      "step": 3167
    },
    {
      "epoch": 0.8702101359703337,
      "grad_norm": 0.49256059527397156,
      "learning_rate": 1.1309605606706062e-05,
      "loss": 0.6713,
      "step": 3168
    },
    {
      "epoch": 0.8704848235132536,
      "grad_norm": 0.5567436218261719,
      "learning_rate": 1.130685722138244e-05,
      "loss": 0.6379,
      "step": 3169
    },
    {
      "epoch": 0.8707595110561736,
      "grad_norm": 0.549852728843689,
      "learning_rate": 1.1304108836058816e-05,
      "loss": 0.7162,
      "step": 3170
    },
    {
      "epoch": 0.8710341985990935,
      "grad_norm": 0.3690211772918701,
      "learning_rate": 1.1301360450735195e-05,
      "loss": 0.9699,
      "step": 3171
    },
    {
      "epoch": 0.8713088861420134,
      "grad_norm": 0.45453381538391113,
      "learning_rate": 1.1298612065411572e-05,
      "loss": 0.6089,
      "step": 3172
    },
    {
      "epoch": 0.8715835736849334,
      "grad_norm": 0.35067418217658997,
      "learning_rate": 1.129586368008795e-05,
      "loss": 1.1753,
      "step": 3173
    },
    {
      "epoch": 0.8718582612278534,
      "grad_norm": 0.5738176107406616,
      "learning_rate": 1.1293115294764328e-05,
      "loss": 0.6905,
      "step": 3174
    },
    {
      "epoch": 0.8721329487707733,
      "grad_norm": 0.44897744059562683,
      "learning_rate": 1.1290366909440705e-05,
      "loss": 0.8861,
      "step": 3175
    },
    {
      "epoch": 0.8724076363136932,
      "grad_norm": 0.46319594979286194,
      "learning_rate": 1.128761852411708e-05,
      "loss": 0.7102,
      "step": 3176
    },
    {
      "epoch": 0.8726823238566132,
      "grad_norm": 0.4412339925765991,
      "learning_rate": 1.128487013879346e-05,
      "loss": 0.5886,
      "step": 3177
    },
    {
      "epoch": 0.8729570113995331,
      "grad_norm": 0.5197470784187317,
      "learning_rate": 1.1282121753469836e-05,
      "loss": 0.623,
      "step": 3178
    },
    {
      "epoch": 0.873231698942453,
      "grad_norm": 0.5692662596702576,
      "learning_rate": 1.1279373368146216e-05,
      "loss": 0.7134,
      "step": 3179
    },
    {
      "epoch": 0.8735063864853729,
      "grad_norm": 0.4462834596633911,
      "learning_rate": 1.1276624982822593e-05,
      "loss": 0.6449,
      "step": 3180
    },
    {
      "epoch": 0.8737810740282929,
      "grad_norm": 0.47803816199302673,
      "learning_rate": 1.1273876597498969e-05,
      "loss": 0.6469,
      "step": 3181
    },
    {
      "epoch": 0.8740557615712128,
      "grad_norm": 0.4256519079208374,
      "learning_rate": 1.127112821217535e-05,
      "loss": 0.7527,
      "step": 3182
    },
    {
      "epoch": 0.8743304491141327,
      "grad_norm": 0.40546828508377075,
      "learning_rate": 1.1268379826851725e-05,
      "loss": 0.6617,
      "step": 3183
    },
    {
      "epoch": 0.8746051366570526,
      "grad_norm": 0.5381118059158325,
      "learning_rate": 1.1265631441528102e-05,
      "loss": 0.9893,
      "step": 3184
    },
    {
      "epoch": 0.8748798241999726,
      "grad_norm": 0.4884776175022125,
      "learning_rate": 1.126288305620448e-05,
      "loss": 0.8564,
      "step": 3185
    },
    {
      "epoch": 0.8751545117428925,
      "grad_norm": 0.4887281060218811,
      "learning_rate": 1.1260134670880857e-05,
      "loss": 0.771,
      "step": 3186
    },
    {
      "epoch": 0.8754291992858124,
      "grad_norm": 0.41346776485443115,
      "learning_rate": 1.1257386285557238e-05,
      "loss": 1.0855,
      "step": 3187
    },
    {
      "epoch": 0.8757038868287323,
      "grad_norm": 0.3243221938610077,
      "learning_rate": 1.1254637900233613e-05,
      "loss": 0.6071,
      "step": 3188
    },
    {
      "epoch": 0.8759785743716523,
      "grad_norm": 0.4154301881790161,
      "learning_rate": 1.125188951490999e-05,
      "loss": 0.7193,
      "step": 3189
    },
    {
      "epoch": 0.8762532619145722,
      "grad_norm": 0.3014715313911438,
      "learning_rate": 1.1249141129586369e-05,
      "loss": 1.0663,
      "step": 3190
    },
    {
      "epoch": 0.8765279494574921,
      "grad_norm": 0.33377358317375183,
      "learning_rate": 1.1246392744262746e-05,
      "loss": 0.8275,
      "step": 3191
    },
    {
      "epoch": 0.876802637000412,
      "grad_norm": 0.4143403172492981,
      "learning_rate": 1.1243644358939123e-05,
      "loss": 0.9394,
      "step": 3192
    },
    {
      "epoch": 0.877077324543332,
      "grad_norm": 0.4747360348701477,
      "learning_rate": 1.1240895973615501e-05,
      "loss": 0.8146,
      "step": 3193
    },
    {
      "epoch": 0.8773520120862519,
      "grad_norm": 0.40260738134384155,
      "learning_rate": 1.1238147588291879e-05,
      "loss": 0.5856,
      "step": 3194
    },
    {
      "epoch": 0.8776266996291718,
      "grad_norm": 0.38692769408226013,
      "learning_rate": 1.1235399202968257e-05,
      "loss": 0.7187,
      "step": 3195
    },
    {
      "epoch": 0.8779013871720918,
      "grad_norm": 0.4515823423862457,
      "learning_rate": 1.1232650817644634e-05,
      "loss": 0.8074,
      "step": 3196
    },
    {
      "epoch": 0.8781760747150117,
      "grad_norm": 0.3793233036994934,
      "learning_rate": 1.1229902432321011e-05,
      "loss": 0.6669,
      "step": 3197
    },
    {
      "epoch": 0.8784507622579316,
      "grad_norm": 0.43010273575782776,
      "learning_rate": 1.122715404699739e-05,
      "loss": 0.6561,
      "step": 3198
    },
    {
      "epoch": 0.8787254498008515,
      "grad_norm": 0.34302711486816406,
      "learning_rate": 1.1224405661673767e-05,
      "loss": 0.5563,
      "step": 3199
    },
    {
      "epoch": 0.8790001373437715,
      "grad_norm": 0.3966692388057709,
      "learning_rate": 1.1221657276350144e-05,
      "loss": 0.6349,
      "step": 3200
    },
    {
      "epoch": 0.8792748248866914,
      "grad_norm": 0.3678834140300751,
      "learning_rate": 1.1218908891026523e-05,
      "loss": 0.6374,
      "step": 3201
    },
    {
      "epoch": 0.8795495124296113,
      "grad_norm": 0.5533472299575806,
      "learning_rate": 1.12161605057029e-05,
      "loss": 0.7424,
      "step": 3202
    },
    {
      "epoch": 0.8798241999725313,
      "grad_norm": 0.4416320025920868,
      "learning_rate": 1.1213412120379278e-05,
      "loss": 1.0405,
      "step": 3203
    },
    {
      "epoch": 0.8800988875154512,
      "grad_norm": 0.38877177238464355,
      "learning_rate": 1.1210663735055656e-05,
      "loss": 0.8761,
      "step": 3204
    },
    {
      "epoch": 0.8803735750583711,
      "grad_norm": 0.48078417778015137,
      "learning_rate": 1.1207915349732033e-05,
      "loss": 0.735,
      "step": 3205
    },
    {
      "epoch": 0.880648262601291,
      "grad_norm": 0.3501462936401367,
      "learning_rate": 1.1205166964408411e-05,
      "loss": 0.8742,
      "step": 3206
    },
    {
      "epoch": 0.880922950144211,
      "grad_norm": 0.4186016619205475,
      "learning_rate": 1.1202418579084788e-05,
      "loss": 0.7252,
      "step": 3207
    },
    {
      "epoch": 0.8811976376871309,
      "grad_norm": 0.25918641686439514,
      "learning_rate": 1.1199670193761165e-05,
      "loss": 0.7087,
      "step": 3208
    },
    {
      "epoch": 0.8814723252300508,
      "grad_norm": 0.5158475637435913,
      "learning_rate": 1.1196921808437544e-05,
      "loss": 0.6717,
      "step": 3209
    },
    {
      "epoch": 0.8817470127729707,
      "grad_norm": 0.35248512029647827,
      "learning_rate": 1.1194173423113921e-05,
      "loss": 0.7827,
      "step": 3210
    },
    {
      "epoch": 0.8820217003158907,
      "grad_norm": 0.42646124958992004,
      "learning_rate": 1.11914250377903e-05,
      "loss": 0.9708,
      "step": 3211
    },
    {
      "epoch": 0.8822963878588106,
      "grad_norm": 0.4517992436885834,
      "learning_rate": 1.1188676652466677e-05,
      "loss": 0.7005,
      "step": 3212
    },
    {
      "epoch": 0.8825710754017305,
      "grad_norm": 0.40593621134757996,
      "learning_rate": 1.1185928267143054e-05,
      "loss": 0.6186,
      "step": 3213
    },
    {
      "epoch": 0.8828457629446504,
      "grad_norm": 0.4042162299156189,
      "learning_rate": 1.1183179881819432e-05,
      "loss": 0.6827,
      "step": 3214
    },
    {
      "epoch": 0.8831204504875704,
      "grad_norm": 0.460520476102829,
      "learning_rate": 1.118043149649581e-05,
      "loss": 0.764,
      "step": 3215
    },
    {
      "epoch": 0.8833951380304903,
      "grad_norm": 0.41877779364585876,
      "learning_rate": 1.1177683111172187e-05,
      "loss": 0.7792,
      "step": 3216
    },
    {
      "epoch": 0.8836698255734102,
      "grad_norm": 0.3513909578323364,
      "learning_rate": 1.1174934725848565e-05,
      "loss": 0.9574,
      "step": 3217
    },
    {
      "epoch": 0.8839445131163302,
      "grad_norm": 0.5707758665084839,
      "learning_rate": 1.1172186340524942e-05,
      "loss": 0.908,
      "step": 3218
    },
    {
      "epoch": 0.8842192006592501,
      "grad_norm": 0.43094611167907715,
      "learning_rate": 1.1169437955201321e-05,
      "loss": 0.6815,
      "step": 3219
    },
    {
      "epoch": 0.88449388820217,
      "grad_norm": 0.36677369475364685,
      "learning_rate": 1.1166689569877698e-05,
      "loss": 0.7146,
      "step": 3220
    },
    {
      "epoch": 0.8847685757450899,
      "grad_norm": 0.3439498841762543,
      "learning_rate": 1.1163941184554075e-05,
      "loss": 0.7188,
      "step": 3221
    },
    {
      "epoch": 0.8850432632880099,
      "grad_norm": 0.5029342174530029,
      "learning_rate": 1.1161192799230454e-05,
      "loss": 0.8115,
      "step": 3222
    },
    {
      "epoch": 0.8853179508309298,
      "grad_norm": 0.5307128429412842,
      "learning_rate": 1.115844441390683e-05,
      "loss": 0.8168,
      "step": 3223
    },
    {
      "epoch": 0.8855926383738497,
      "grad_norm": 0.5077596306800842,
      "learning_rate": 1.1155696028583208e-05,
      "loss": 0.6862,
      "step": 3224
    },
    {
      "epoch": 0.8858673259167696,
      "grad_norm": 0.4017813205718994,
      "learning_rate": 1.1152947643259586e-05,
      "loss": 0.992,
      "step": 3225
    },
    {
      "epoch": 0.8861420134596896,
      "grad_norm": 0.5834123492240906,
      "learning_rate": 1.1150199257935964e-05,
      "loss": 0.9066,
      "step": 3226
    },
    {
      "epoch": 0.8864167010026095,
      "grad_norm": 0.6411367654800415,
      "learning_rate": 1.1147450872612342e-05,
      "loss": 0.919,
      "step": 3227
    },
    {
      "epoch": 0.8866913885455294,
      "grad_norm": 0.3863753080368042,
      "learning_rate": 1.114470248728872e-05,
      "loss": 0.954,
      "step": 3228
    },
    {
      "epoch": 0.8869660760884494,
      "grad_norm": 0.3479141294956207,
      "learning_rate": 1.1141954101965096e-05,
      "loss": 0.7322,
      "step": 3229
    },
    {
      "epoch": 0.8872407636313693,
      "grad_norm": 0.44803979992866516,
      "learning_rate": 1.1139205716641475e-05,
      "loss": 0.7503,
      "step": 3230
    },
    {
      "epoch": 0.8875154511742892,
      "grad_norm": 0.46756017208099365,
      "learning_rate": 1.1136457331317852e-05,
      "loss": 0.8803,
      "step": 3231
    },
    {
      "epoch": 0.8877901387172091,
      "grad_norm": 0.38953468203544617,
      "learning_rate": 1.1133708945994229e-05,
      "loss": 0.7531,
      "step": 3232
    },
    {
      "epoch": 0.8880648262601291,
      "grad_norm": 0.5302019715309143,
      "learning_rate": 1.1130960560670608e-05,
      "loss": 0.8151,
      "step": 3233
    },
    {
      "epoch": 0.888339513803049,
      "grad_norm": 0.45141729712486267,
      "learning_rate": 1.1128212175346985e-05,
      "loss": 0.6976,
      "step": 3234
    },
    {
      "epoch": 0.8886142013459689,
      "grad_norm": 0.3565881550312042,
      "learning_rate": 1.1125463790023363e-05,
      "loss": 0.9138,
      "step": 3235
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.38121017813682556,
      "learning_rate": 1.112271540469974e-05,
      "loss": 0.8209,
      "step": 3236
    },
    {
      "epoch": 0.8891635764318088,
      "grad_norm": 0.5290696620941162,
      "learning_rate": 1.1119967019376116e-05,
      "loss": 0.8078,
      "step": 3237
    },
    {
      "epoch": 0.8894382639747287,
      "grad_norm": 0.5946629643440247,
      "learning_rate": 1.1117218634052496e-05,
      "loss": 0.8139,
      "step": 3238
    },
    {
      "epoch": 0.8897129515176486,
      "grad_norm": 0.4996741712093353,
      "learning_rate": 1.1114470248728873e-05,
      "loss": 0.6866,
      "step": 3239
    },
    {
      "epoch": 0.8899876390605687,
      "grad_norm": 0.3569605052471161,
      "learning_rate": 1.1111721863405249e-05,
      "loss": 0.8076,
      "step": 3240
    },
    {
      "epoch": 0.8902623266034886,
      "grad_norm": 0.5425332188606262,
      "learning_rate": 1.1108973478081629e-05,
      "loss": 0.8398,
      "step": 3241
    },
    {
      "epoch": 0.8905370141464085,
      "grad_norm": 0.48060545325279236,
      "learning_rate": 1.1106225092758004e-05,
      "loss": 0.7247,
      "step": 3242
    },
    {
      "epoch": 0.8908117016893284,
      "grad_norm": 0.44545429944992065,
      "learning_rate": 1.1103476707434385e-05,
      "loss": 0.9932,
      "step": 3243
    },
    {
      "epoch": 0.8910863892322484,
      "grad_norm": 0.35129210352897644,
      "learning_rate": 1.1100728322110762e-05,
      "loss": 0.9089,
      "step": 3244
    },
    {
      "epoch": 0.8913610767751683,
      "grad_norm": 0.4271351993083954,
      "learning_rate": 1.1097979936787137e-05,
      "loss": 0.7938,
      "step": 3245
    },
    {
      "epoch": 0.8916357643180882,
      "grad_norm": 0.44677162170410156,
      "learning_rate": 1.1095231551463517e-05,
      "loss": 0.7106,
      "step": 3246
    },
    {
      "epoch": 0.8919104518610081,
      "grad_norm": 0.4299044907093048,
      "learning_rate": 1.1092483166139893e-05,
      "loss": 0.9255,
      "step": 3247
    },
    {
      "epoch": 0.8921851394039281,
      "grad_norm": 0.4119156002998352,
      "learning_rate": 1.108973478081627e-05,
      "loss": 0.6862,
      "step": 3248
    },
    {
      "epoch": 0.892459826946848,
      "grad_norm": 0.39982709288597107,
      "learning_rate": 1.1086986395492649e-05,
      "loss": 0.6118,
      "step": 3249
    },
    {
      "epoch": 0.8927345144897679,
      "grad_norm": 0.5436339378356934,
      "learning_rate": 1.1084238010169026e-05,
      "loss": 0.5397,
      "step": 3250
    },
    {
      "epoch": 0.8930092020326879,
      "grad_norm": 0.580186128616333,
      "learning_rate": 1.1081489624845406e-05,
      "loss": 0.9195,
      "step": 3251
    },
    {
      "epoch": 0.8932838895756078,
      "grad_norm": 0.3730548918247223,
      "learning_rate": 1.1078741239521781e-05,
      "loss": 0.9894,
      "step": 3252
    },
    {
      "epoch": 0.8935585771185277,
      "grad_norm": 0.3532906770706177,
      "learning_rate": 1.1075992854198158e-05,
      "loss": 0.5352,
      "step": 3253
    },
    {
      "epoch": 0.8938332646614476,
      "grad_norm": 0.5609377026557922,
      "learning_rate": 1.1073244468874537e-05,
      "loss": 0.9176,
      "step": 3254
    },
    {
      "epoch": 0.8941079522043676,
      "grad_norm": 0.6163805723190308,
      "learning_rate": 1.1070496083550914e-05,
      "loss": 0.7812,
      "step": 3255
    },
    {
      "epoch": 0.8943826397472875,
      "grad_norm": 0.35918089747428894,
      "learning_rate": 1.1067747698227291e-05,
      "loss": 1.0417,
      "step": 3256
    },
    {
      "epoch": 0.8946573272902074,
      "grad_norm": 0.4066706895828247,
      "learning_rate": 1.106499931290367e-05,
      "loss": 0.8916,
      "step": 3257
    },
    {
      "epoch": 0.8949320148331273,
      "grad_norm": 0.4682488739490509,
      "learning_rate": 1.1062250927580047e-05,
      "loss": 0.5625,
      "step": 3258
    },
    {
      "epoch": 0.8952067023760473,
      "grad_norm": 0.4743712544441223,
      "learning_rate": 1.1059502542256426e-05,
      "loss": 0.8272,
      "step": 3259
    },
    {
      "epoch": 0.8954813899189672,
      "grad_norm": 0.47509533166885376,
      "learning_rate": 1.1056754156932803e-05,
      "loss": 0.7805,
      "step": 3260
    },
    {
      "epoch": 0.8957560774618871,
      "grad_norm": 0.3193657100200653,
      "learning_rate": 1.105400577160918e-05,
      "loss": 0.4793,
      "step": 3261
    },
    {
      "epoch": 0.896030765004807,
      "grad_norm": 0.5371000170707703,
      "learning_rate": 1.1051257386285558e-05,
      "loss": 0.8719,
      "step": 3262
    },
    {
      "epoch": 0.896305452547727,
      "grad_norm": 0.7095023989677429,
      "learning_rate": 1.1048509000961935e-05,
      "loss": 0.7522,
      "step": 3263
    },
    {
      "epoch": 0.8965801400906469,
      "grad_norm": 0.48195046186447144,
      "learning_rate": 1.1045760615638312e-05,
      "loss": 0.7948,
      "step": 3264
    },
    {
      "epoch": 0.8968548276335668,
      "grad_norm": 0.4237311780452728,
      "learning_rate": 1.1043012230314691e-05,
      "loss": 0.6116,
      "step": 3265
    },
    {
      "epoch": 0.8971295151764868,
      "grad_norm": 0.460176020860672,
      "learning_rate": 1.1040263844991068e-05,
      "loss": 0.6295,
      "step": 3266
    },
    {
      "epoch": 0.8974042027194067,
      "grad_norm": 0.4480782151222229,
      "learning_rate": 1.1037515459667447e-05,
      "loss": 0.6865,
      "step": 3267
    },
    {
      "epoch": 0.8976788902623266,
      "grad_norm": 0.5415688753128052,
      "learning_rate": 1.1034767074343824e-05,
      "loss": 0.8887,
      "step": 3268
    },
    {
      "epoch": 0.8979535778052465,
      "grad_norm": 0.45753082633018494,
      "learning_rate": 1.10320186890202e-05,
      "loss": 0.799,
      "step": 3269
    },
    {
      "epoch": 0.8982282653481665,
      "grad_norm": 0.44746094942092896,
      "learning_rate": 1.102927030369658e-05,
      "loss": 0.6682,
      "step": 3270
    },
    {
      "epoch": 0.8985029528910864,
      "grad_norm": 0.43462294340133667,
      "learning_rate": 1.1026521918372957e-05,
      "loss": 0.6685,
      "step": 3271
    },
    {
      "epoch": 0.8987776404340063,
      "grad_norm": 0.43827179074287415,
      "learning_rate": 1.1023773533049334e-05,
      "loss": 0.6493,
      "step": 3272
    },
    {
      "epoch": 0.8990523279769262,
      "grad_norm": 0.41883593797683716,
      "learning_rate": 1.1021025147725712e-05,
      "loss": 0.6134,
      "step": 3273
    },
    {
      "epoch": 0.8993270155198462,
      "grad_norm": 0.30487337708473206,
      "learning_rate": 1.101827676240209e-05,
      "loss": 0.7054,
      "step": 3274
    },
    {
      "epoch": 0.8996017030627661,
      "grad_norm": 0.32610929012298584,
      "learning_rate": 1.1015528377078468e-05,
      "loss": 0.7651,
      "step": 3275
    },
    {
      "epoch": 0.899876390605686,
      "grad_norm": 0.4300110638141632,
      "learning_rate": 1.1012779991754845e-05,
      "loss": 0.8342,
      "step": 3276
    },
    {
      "epoch": 0.900151078148606,
      "grad_norm": 0.4587155878543854,
      "learning_rate": 1.1010031606431222e-05,
      "loss": 0.59,
      "step": 3277
    },
    {
      "epoch": 0.9004257656915259,
      "grad_norm": 0.48943886160850525,
      "learning_rate": 1.10072832211076e-05,
      "loss": 0.8691,
      "step": 3278
    },
    {
      "epoch": 0.9007004532344458,
      "grad_norm": 0.3597392737865448,
      "learning_rate": 1.1004534835783978e-05,
      "loss": 0.6895,
      "step": 3279
    },
    {
      "epoch": 0.9009751407773657,
      "grad_norm": 0.43375417590141296,
      "learning_rate": 1.1001786450460355e-05,
      "loss": 0.7101,
      "step": 3280
    },
    {
      "epoch": 0.9012498283202857,
      "grad_norm": 0.4659120440483093,
      "learning_rate": 1.0999038065136734e-05,
      "loss": 0.838,
      "step": 3281
    },
    {
      "epoch": 0.9015245158632056,
      "grad_norm": 0.45352140069007874,
      "learning_rate": 1.099628967981311e-05,
      "loss": 0.7778,
      "step": 3282
    },
    {
      "epoch": 0.9017992034061255,
      "grad_norm": 0.41904976963996887,
      "learning_rate": 1.099354129448949e-05,
      "loss": 0.9363,
      "step": 3283
    },
    {
      "epoch": 0.9020738909490454,
      "grad_norm": 0.3697563409805298,
      "learning_rate": 1.0990792909165866e-05,
      "loss": 0.4938,
      "step": 3284
    },
    {
      "epoch": 0.9023485784919654,
      "grad_norm": 0.35856762528419495,
      "learning_rate": 1.0988044523842243e-05,
      "loss": 0.6572,
      "step": 3285
    },
    {
      "epoch": 0.9026232660348853,
      "grad_norm": 0.46948206424713135,
      "learning_rate": 1.0985296138518622e-05,
      "loss": 0.7328,
      "step": 3286
    },
    {
      "epoch": 0.9028979535778052,
      "grad_norm": 0.6157584190368652,
      "learning_rate": 1.0982547753194999e-05,
      "loss": 0.8207,
      "step": 3287
    },
    {
      "epoch": 0.9031726411207252,
      "grad_norm": 0.47197529673576355,
      "learning_rate": 1.0979799367871376e-05,
      "loss": 0.9237,
      "step": 3288
    },
    {
      "epoch": 0.9034473286636451,
      "grad_norm": 0.485146164894104,
      "learning_rate": 1.0977050982547755e-05,
      "loss": 1.0073,
      "step": 3289
    },
    {
      "epoch": 0.903722016206565,
      "grad_norm": 0.42907199263572693,
      "learning_rate": 1.0974302597224132e-05,
      "loss": 0.8284,
      "step": 3290
    },
    {
      "epoch": 0.9039967037494849,
      "grad_norm": 0.4710270166397095,
      "learning_rate": 1.097155421190051e-05,
      "loss": 0.804,
      "step": 3291
    },
    {
      "epoch": 0.9042713912924049,
      "grad_norm": 0.4655184745788574,
      "learning_rate": 1.0968805826576888e-05,
      "loss": 0.8785,
      "step": 3292
    },
    {
      "epoch": 0.9045460788353248,
      "grad_norm": 0.4888700246810913,
      "learning_rate": 1.0966057441253265e-05,
      "loss": 0.8339,
      "step": 3293
    },
    {
      "epoch": 0.9048207663782447,
      "grad_norm": 0.385762482881546,
      "learning_rate": 1.0963309055929643e-05,
      "loss": 0.9014,
      "step": 3294
    },
    {
      "epoch": 0.9050954539211646,
      "grad_norm": 0.4227830171585083,
      "learning_rate": 1.096056067060602e-05,
      "loss": 0.6323,
      "step": 3295
    },
    {
      "epoch": 0.9053701414640846,
      "grad_norm": 0.5389531850814819,
      "learning_rate": 1.0957812285282397e-05,
      "loss": 0.8462,
      "step": 3296
    },
    {
      "epoch": 0.9056448290070045,
      "grad_norm": 0.45470044016838074,
      "learning_rate": 1.0955063899958776e-05,
      "loss": 0.5567,
      "step": 3297
    },
    {
      "epoch": 0.9059195165499244,
      "grad_norm": 0.3707863390445709,
      "learning_rate": 1.0952315514635153e-05,
      "loss": 0.8692,
      "step": 3298
    },
    {
      "epoch": 0.9061942040928443,
      "grad_norm": 0.4534496068954468,
      "learning_rate": 1.0949567129311532e-05,
      "loss": 0.8149,
      "step": 3299
    },
    {
      "epoch": 0.9064688916357643,
      "grad_norm": 0.46771663427352905,
      "learning_rate": 1.0946818743987909e-05,
      "loss": 0.8005,
      "step": 3300
    },
    {
      "epoch": 0.9067435791786842,
      "grad_norm": 0.2992889881134033,
      "learning_rate": 1.0944070358664284e-05,
      "loss": 0.7862,
      "step": 3301
    },
    {
      "epoch": 0.9070182667216041,
      "grad_norm": 0.38543182611465454,
      "learning_rate": 1.0941321973340665e-05,
      "loss": 0.8305,
      "step": 3302
    },
    {
      "epoch": 0.907292954264524,
      "grad_norm": 0.5244500637054443,
      "learning_rate": 1.0938573588017042e-05,
      "loss": 0.7889,
      "step": 3303
    },
    {
      "epoch": 0.907567641807444,
      "grad_norm": 0.3556429445743561,
      "learning_rate": 1.0935825202693417e-05,
      "loss": 0.7987,
      "step": 3304
    },
    {
      "epoch": 0.9078423293503639,
      "grad_norm": 0.5324926972389221,
      "learning_rate": 1.0933076817369797e-05,
      "loss": 0.6161,
      "step": 3305
    },
    {
      "epoch": 0.9081170168932838,
      "grad_norm": 0.4921017289161682,
      "learning_rate": 1.0930328432046173e-05,
      "loss": 0.7808,
      "step": 3306
    },
    {
      "epoch": 0.9083917044362039,
      "grad_norm": 0.5175229907035828,
      "learning_rate": 1.0927580046722553e-05,
      "loss": 0.6761,
      "step": 3307
    },
    {
      "epoch": 0.9086663919791238,
      "grad_norm": 0.5082927346229553,
      "learning_rate": 1.0924831661398928e-05,
      "loss": 0.6192,
      "step": 3308
    },
    {
      "epoch": 0.9089410795220437,
      "grad_norm": 0.49705174565315247,
      "learning_rate": 1.0922083276075305e-05,
      "loss": 0.9037,
      "step": 3309
    },
    {
      "epoch": 0.9092157670649637,
      "grad_norm": 0.4760642647743225,
      "learning_rate": 1.0919334890751686e-05,
      "loss": 0.9619,
      "step": 3310
    },
    {
      "epoch": 0.9094904546078836,
      "grad_norm": 0.5807300806045532,
      "learning_rate": 1.0916586505428061e-05,
      "loss": 0.8743,
      "step": 3311
    },
    {
      "epoch": 0.9097651421508035,
      "grad_norm": 0.4725587069988251,
      "learning_rate": 1.0913838120104438e-05,
      "loss": 0.6274,
      "step": 3312
    },
    {
      "epoch": 0.9100398296937234,
      "grad_norm": 0.5274493098258972,
      "learning_rate": 1.0911089734780817e-05,
      "loss": 0.8033,
      "step": 3313
    },
    {
      "epoch": 0.9103145172366434,
      "grad_norm": 0.47205621004104614,
      "learning_rate": 1.0908341349457194e-05,
      "loss": 0.8303,
      "step": 3314
    },
    {
      "epoch": 0.9105892047795633,
      "grad_norm": 0.4193306863307953,
      "learning_rate": 1.0905592964133573e-05,
      "loss": 0.6076,
      "step": 3315
    },
    {
      "epoch": 0.9108638923224832,
      "grad_norm": 0.4968576729297638,
      "learning_rate": 1.090284457880995e-05,
      "loss": 0.9151,
      "step": 3316
    },
    {
      "epoch": 0.9111385798654031,
      "grad_norm": 0.3048945367336273,
      "learning_rate": 1.0900096193486327e-05,
      "loss": 0.6126,
      "step": 3317
    },
    {
      "epoch": 0.9114132674083231,
      "grad_norm": 0.46058306097984314,
      "learning_rate": 1.0897347808162705e-05,
      "loss": 0.7443,
      "step": 3318
    },
    {
      "epoch": 0.911687954951243,
      "grad_norm": 0.5704224109649658,
      "learning_rate": 1.0894599422839082e-05,
      "loss": 0.7417,
      "step": 3319
    },
    {
      "epoch": 0.9119626424941629,
      "grad_norm": 0.5011089444160461,
      "learning_rate": 1.089185103751546e-05,
      "loss": 0.825,
      "step": 3320
    },
    {
      "epoch": 0.9122373300370828,
      "grad_norm": 0.46560609340667725,
      "learning_rate": 1.0889102652191838e-05,
      "loss": 0.7716,
      "step": 3321
    },
    {
      "epoch": 0.9125120175800028,
      "grad_norm": 0.5513193607330322,
      "learning_rate": 1.0886354266868215e-05,
      "loss": 0.8649,
      "step": 3322
    },
    {
      "epoch": 0.9127867051229227,
      "grad_norm": 0.42002683877944946,
      "learning_rate": 1.0883605881544594e-05,
      "loss": 0.858,
      "step": 3323
    },
    {
      "epoch": 0.9130613926658426,
      "grad_norm": 0.5121695399284363,
      "learning_rate": 1.088085749622097e-05,
      "loss": 0.7177,
      "step": 3324
    },
    {
      "epoch": 0.9133360802087626,
      "grad_norm": 0.33616697788238525,
      "learning_rate": 1.0878109110897348e-05,
      "loss": 0.8196,
      "step": 3325
    },
    {
      "epoch": 0.9136107677516825,
      "grad_norm": 0.3807526230812073,
      "learning_rate": 1.0875360725573727e-05,
      "loss": 0.8451,
      "step": 3326
    },
    {
      "epoch": 0.9138854552946024,
      "grad_norm": 0.41291528940200806,
      "learning_rate": 1.0872612340250104e-05,
      "loss": 0.674,
      "step": 3327
    },
    {
      "epoch": 0.9141601428375223,
      "grad_norm": 0.330866277217865,
      "learning_rate": 1.086986395492648e-05,
      "loss": 0.8111,
      "step": 3328
    },
    {
      "epoch": 0.9144348303804423,
      "grad_norm": 0.47553718090057373,
      "learning_rate": 1.086711556960286e-05,
      "loss": 0.7551,
      "step": 3329
    },
    {
      "epoch": 0.9147095179233622,
      "grad_norm": 0.3627781569957733,
      "learning_rate": 1.0864367184279236e-05,
      "loss": 0.5799,
      "step": 3330
    },
    {
      "epoch": 0.9149842054662821,
      "grad_norm": 0.48784366250038147,
      "learning_rate": 1.0861618798955615e-05,
      "loss": 0.6223,
      "step": 3331
    },
    {
      "epoch": 0.915258893009202,
      "grad_norm": 0.40558990836143494,
      "learning_rate": 1.0858870413631992e-05,
      "loss": 0.8096,
      "step": 3332
    },
    {
      "epoch": 0.915533580552122,
      "grad_norm": 0.41259947419166565,
      "learning_rate": 1.0856122028308369e-05,
      "loss": 0.7411,
      "step": 3333
    },
    {
      "epoch": 0.9158082680950419,
      "grad_norm": 0.3352629840373993,
      "learning_rate": 1.0853373642984748e-05,
      "loss": 0.5473,
      "step": 3334
    },
    {
      "epoch": 0.9160829556379618,
      "grad_norm": 0.38502928614616394,
      "learning_rate": 1.0850625257661125e-05,
      "loss": 0.4934,
      "step": 3335
    },
    {
      "epoch": 0.9163576431808818,
      "grad_norm": 0.5271713733673096,
      "learning_rate": 1.0847876872337502e-05,
      "loss": 0.9837,
      "step": 3336
    },
    {
      "epoch": 0.9166323307238017,
      "grad_norm": 0.4747161269187927,
      "learning_rate": 1.084512848701388e-05,
      "loss": 0.8576,
      "step": 3337
    },
    {
      "epoch": 0.9169070182667216,
      "grad_norm": 0.5046903491020203,
      "learning_rate": 1.0842380101690258e-05,
      "loss": 0.8153,
      "step": 3338
    },
    {
      "epoch": 0.9171817058096415,
      "grad_norm": 0.3188874125480652,
      "learning_rate": 1.0839631716366636e-05,
      "loss": 0.748,
      "step": 3339
    },
    {
      "epoch": 0.9174563933525615,
      "grad_norm": 0.462249219417572,
      "learning_rate": 1.0836883331043013e-05,
      "loss": 0.7489,
      "step": 3340
    },
    {
      "epoch": 0.9177310808954814,
      "grad_norm": 0.4011991322040558,
      "learning_rate": 1.083413494571939e-05,
      "loss": 0.816,
      "step": 3341
    },
    {
      "epoch": 0.9180057684384013,
      "grad_norm": 0.37654486298561096,
      "learning_rate": 1.0831386560395769e-05,
      "loss": 0.9894,
      "step": 3342
    },
    {
      "epoch": 0.9182804559813212,
      "grad_norm": 0.5051063299179077,
      "learning_rate": 1.0828638175072146e-05,
      "loss": 1.069,
      "step": 3343
    },
    {
      "epoch": 0.9185551435242412,
      "grad_norm": 0.3990648090839386,
      "learning_rate": 1.0825889789748523e-05,
      "loss": 0.498,
      "step": 3344
    },
    {
      "epoch": 0.9188298310671611,
      "grad_norm": 0.3879188001155853,
      "learning_rate": 1.0823141404424902e-05,
      "loss": 0.9367,
      "step": 3345
    },
    {
      "epoch": 0.919104518610081,
      "grad_norm": 0.367260605096817,
      "learning_rate": 1.0820393019101279e-05,
      "loss": 0.8625,
      "step": 3346
    },
    {
      "epoch": 0.919379206153001,
      "grad_norm": 0.45506730675697327,
      "learning_rate": 1.0817644633777658e-05,
      "loss": 0.7674,
      "step": 3347
    },
    {
      "epoch": 0.9196538936959209,
      "grad_norm": 0.4580680727958679,
      "learning_rate": 1.0814896248454035e-05,
      "loss": 0.6343,
      "step": 3348
    },
    {
      "epoch": 0.9199285812388408,
      "grad_norm": 0.3797946870326996,
      "learning_rate": 1.0812147863130412e-05,
      "loss": 0.7894,
      "step": 3349
    },
    {
      "epoch": 0.9202032687817607,
      "grad_norm": 0.35966381430625916,
      "learning_rate": 1.080939947780679e-05,
      "loss": 0.7237,
      "step": 3350
    },
    {
      "epoch": 0.9204779563246807,
      "grad_norm": 0.4575766623020172,
      "learning_rate": 1.0806651092483167e-05,
      "loss": 0.7494,
      "step": 3351
    },
    {
      "epoch": 0.9207526438676006,
      "grad_norm": 0.5403180122375488,
      "learning_rate": 1.0803902707159544e-05,
      "loss": 0.6993,
      "step": 3352
    },
    {
      "epoch": 0.9210273314105205,
      "grad_norm": 0.5685340166091919,
      "learning_rate": 1.0801154321835923e-05,
      "loss": 0.8825,
      "step": 3353
    },
    {
      "epoch": 0.9213020189534404,
      "grad_norm": 0.4294430911540985,
      "learning_rate": 1.07984059365123e-05,
      "loss": 0.9662,
      "step": 3354
    },
    {
      "epoch": 0.9215767064963604,
      "grad_norm": 0.4639984965324402,
      "learning_rate": 1.0795657551188679e-05,
      "loss": 0.6183,
      "step": 3355
    },
    {
      "epoch": 0.9218513940392803,
      "grad_norm": 0.3808353543281555,
      "learning_rate": 1.0792909165865056e-05,
      "loss": 1.1675,
      "step": 3356
    },
    {
      "epoch": 0.9221260815822002,
      "grad_norm": 0.5085517764091492,
      "learning_rate": 1.0790160780541433e-05,
      "loss": 0.8188,
      "step": 3357
    },
    {
      "epoch": 0.9224007691251201,
      "grad_norm": 0.3996214270591736,
      "learning_rate": 1.0787412395217812e-05,
      "loss": 0.8288,
      "step": 3358
    },
    {
      "epoch": 0.9226754566680401,
      "grad_norm": 0.4848426878452301,
      "learning_rate": 1.0784664009894189e-05,
      "loss": 0.9078,
      "step": 3359
    },
    {
      "epoch": 0.92295014421096,
      "grad_norm": 0.5383610129356384,
      "learning_rate": 1.0781915624570564e-05,
      "loss": 0.7674,
      "step": 3360
    },
    {
      "epoch": 0.9232248317538799,
      "grad_norm": 0.4107929468154907,
      "learning_rate": 1.0779167239246944e-05,
      "loss": 0.5941,
      "step": 3361
    },
    {
      "epoch": 0.9234995192967999,
      "grad_norm": 0.5598790049552917,
      "learning_rate": 1.0776418853923321e-05,
      "loss": 0.7036,
      "step": 3362
    },
    {
      "epoch": 0.9237742068397198,
      "grad_norm": 0.46257728338241577,
      "learning_rate": 1.07736704685997e-05,
      "loss": 0.6684,
      "step": 3363
    },
    {
      "epoch": 0.9240488943826397,
      "grad_norm": 0.42782360315322876,
      "learning_rate": 1.0770922083276077e-05,
      "loss": 0.6799,
      "step": 3364
    },
    {
      "epoch": 0.9243235819255596,
      "grad_norm": 0.5527724623680115,
      "learning_rate": 1.0768173697952452e-05,
      "loss": 0.834,
      "step": 3365
    },
    {
      "epoch": 0.9245982694684796,
      "grad_norm": 0.6076619625091553,
      "learning_rate": 1.0765425312628833e-05,
      "loss": 0.7446,
      "step": 3366
    },
    {
      "epoch": 0.9248729570113995,
      "grad_norm": 0.49025875329971313,
      "learning_rate": 1.0762676927305208e-05,
      "loss": 0.5668,
      "step": 3367
    },
    {
      "epoch": 0.9251476445543194,
      "grad_norm": 0.3233032524585724,
      "learning_rate": 1.0759928541981585e-05,
      "loss": 0.6123,
      "step": 3368
    },
    {
      "epoch": 0.9254223320972393,
      "grad_norm": 0.5357702374458313,
      "learning_rate": 1.0757180156657966e-05,
      "loss": 0.6101,
      "step": 3369
    },
    {
      "epoch": 0.9256970196401593,
      "grad_norm": 0.5599467158317566,
      "learning_rate": 1.0754431771334341e-05,
      "loss": 0.8838,
      "step": 3370
    },
    {
      "epoch": 0.9259717071830792,
      "grad_norm": 0.4113526940345764,
      "learning_rate": 1.0751683386010721e-05,
      "loss": 0.6212,
      "step": 3371
    },
    {
      "epoch": 0.9262463947259991,
      "grad_norm": 0.4146808087825775,
      "learning_rate": 1.0748935000687097e-05,
      "loss": 0.8647,
      "step": 3372
    },
    {
      "epoch": 0.926521082268919,
      "grad_norm": 0.5776706337928772,
      "learning_rate": 1.0746186615363474e-05,
      "loss": 0.9499,
      "step": 3373
    },
    {
      "epoch": 0.9267957698118391,
      "grad_norm": 0.36336904764175415,
      "learning_rate": 1.0743438230039852e-05,
      "loss": 0.725,
      "step": 3374
    },
    {
      "epoch": 0.927070457354759,
      "grad_norm": 0.4598216414451599,
      "learning_rate": 1.074068984471623e-05,
      "loss": 0.7633,
      "step": 3375
    },
    {
      "epoch": 0.9273451448976789,
      "grad_norm": 0.4982246458530426,
      "learning_rate": 1.0737941459392606e-05,
      "loss": 0.6859,
      "step": 3376
    },
    {
      "epoch": 0.9276198324405989,
      "grad_norm": 0.3768463730812073,
      "learning_rate": 1.0735193074068985e-05,
      "loss": 0.6927,
      "step": 3377
    },
    {
      "epoch": 0.9278945199835188,
      "grad_norm": 0.4912184178829193,
      "learning_rate": 1.0732444688745362e-05,
      "loss": 0.9568,
      "step": 3378
    },
    {
      "epoch": 0.9281692075264387,
      "grad_norm": 0.3806002140045166,
      "learning_rate": 1.0729696303421741e-05,
      "loss": 1.0254,
      "step": 3379
    },
    {
      "epoch": 0.9284438950693586,
      "grad_norm": 0.3158327639102936,
      "learning_rate": 1.0726947918098118e-05,
      "loss": 0.629,
      "step": 3380
    },
    {
      "epoch": 0.9287185826122786,
      "grad_norm": 0.4823381304740906,
      "learning_rate": 1.0724199532774495e-05,
      "loss": 0.5763,
      "step": 3381
    },
    {
      "epoch": 0.9289932701551985,
      "grad_norm": 0.38425716757774353,
      "learning_rate": 1.0721451147450874e-05,
      "loss": 1.0193,
      "step": 3382
    },
    {
      "epoch": 0.9292679576981184,
      "grad_norm": 0.46833550930023193,
      "learning_rate": 1.071870276212725e-05,
      "loss": 0.79,
      "step": 3383
    },
    {
      "epoch": 0.9295426452410384,
      "grad_norm": 0.5206587910652161,
      "learning_rate": 1.0715954376803628e-05,
      "loss": 0.7837,
      "step": 3384
    },
    {
      "epoch": 0.9298173327839583,
      "grad_norm": 0.39904266595840454,
      "learning_rate": 1.0713205991480006e-05,
      "loss": 0.7305,
      "step": 3385
    },
    {
      "epoch": 0.9300920203268782,
      "grad_norm": 0.4615829288959503,
      "learning_rate": 1.0710457606156383e-05,
      "loss": 0.9043,
      "step": 3386
    },
    {
      "epoch": 0.9303667078697981,
      "grad_norm": 0.580036461353302,
      "learning_rate": 1.0707709220832762e-05,
      "loss": 0.9538,
      "step": 3387
    },
    {
      "epoch": 0.9306413954127181,
      "grad_norm": 0.45450952649116516,
      "learning_rate": 1.0704960835509139e-05,
      "loss": 0.747,
      "step": 3388
    },
    {
      "epoch": 0.930916082955638,
      "grad_norm": 0.37642085552215576,
      "learning_rate": 1.0702212450185516e-05,
      "loss": 0.9437,
      "step": 3389
    },
    {
      "epoch": 0.9311907704985579,
      "grad_norm": 0.5487443208694458,
      "learning_rate": 1.0699464064861895e-05,
      "loss": 0.8187,
      "step": 3390
    },
    {
      "epoch": 0.9314654580414778,
      "grad_norm": 0.5175371766090393,
      "learning_rate": 1.0696715679538272e-05,
      "loss": 0.7385,
      "step": 3391
    },
    {
      "epoch": 0.9317401455843978,
      "grad_norm": 0.6628782153129578,
      "learning_rate": 1.0693967294214649e-05,
      "loss": 0.6834,
      "step": 3392
    },
    {
      "epoch": 0.9320148331273177,
      "grad_norm": 0.5078772902488708,
      "learning_rate": 1.0691218908891028e-05,
      "loss": 0.7821,
      "step": 3393
    },
    {
      "epoch": 0.9322895206702376,
      "grad_norm": 0.4375612735748291,
      "learning_rate": 1.0688470523567405e-05,
      "loss": 1.1275,
      "step": 3394
    },
    {
      "epoch": 0.9325642082131576,
      "grad_norm": 0.4290148615837097,
      "learning_rate": 1.0685722138243783e-05,
      "loss": 0.7499,
      "step": 3395
    },
    {
      "epoch": 0.9328388957560775,
      "grad_norm": 0.31619903445243835,
      "learning_rate": 1.068297375292016e-05,
      "loss": 0.6209,
      "step": 3396
    },
    {
      "epoch": 0.9331135832989974,
      "grad_norm": 0.454396516084671,
      "learning_rate": 1.0680225367596537e-05,
      "loss": 0.7344,
      "step": 3397
    },
    {
      "epoch": 0.9333882708419173,
      "grad_norm": 0.43646422028541565,
      "learning_rate": 1.0677476982272916e-05,
      "loss": 0.5402,
      "step": 3398
    },
    {
      "epoch": 0.9336629583848373,
      "grad_norm": 0.5473254919052124,
      "learning_rate": 1.0674728596949293e-05,
      "loss": 0.8963,
      "step": 3399
    },
    {
      "epoch": 0.9339376459277572,
      "grad_norm": 0.4628221392631531,
      "learning_rate": 1.067198021162567e-05,
      "loss": 0.7855,
      "step": 3400
    },
    {
      "epoch": 0.9342123334706771,
      "grad_norm": 0.4006953537464142,
      "learning_rate": 1.0669231826302049e-05,
      "loss": 0.8672,
      "step": 3401
    },
    {
      "epoch": 0.934487021013597,
      "grad_norm": 0.4137876033782959,
      "learning_rate": 1.0666483440978426e-05,
      "loss": 0.5913,
      "step": 3402
    },
    {
      "epoch": 0.934761708556517,
      "grad_norm": 0.397334486246109,
      "learning_rate": 1.0663735055654805e-05,
      "loss": 0.9124,
      "step": 3403
    },
    {
      "epoch": 0.9350363960994369,
      "grad_norm": 0.40748119354248047,
      "learning_rate": 1.0660986670331182e-05,
      "loss": 0.7234,
      "step": 3404
    },
    {
      "epoch": 0.9353110836423568,
      "grad_norm": 0.5301898121833801,
      "learning_rate": 1.0658238285007559e-05,
      "loss": 0.7613,
      "step": 3405
    },
    {
      "epoch": 0.9355857711852767,
      "grad_norm": 0.5261777639389038,
      "learning_rate": 1.0655489899683937e-05,
      "loss": 0.6812,
      "step": 3406
    },
    {
      "epoch": 0.9358604587281967,
      "grad_norm": 0.41836225986480713,
      "learning_rate": 1.0652741514360314e-05,
      "loss": 0.7919,
      "step": 3407
    },
    {
      "epoch": 0.9361351462711166,
      "grad_norm": 0.4521649479866028,
      "learning_rate": 1.0649993129036691e-05,
      "loss": 0.9464,
      "step": 3408
    },
    {
      "epoch": 0.9364098338140365,
      "grad_norm": 0.4270244836807251,
      "learning_rate": 1.064724474371307e-05,
      "loss": 0.7145,
      "step": 3409
    },
    {
      "epoch": 0.9366845213569565,
      "grad_norm": 0.5600605607032776,
      "learning_rate": 1.0644496358389447e-05,
      "loss": 0.7775,
      "step": 3410
    },
    {
      "epoch": 0.9369592088998764,
      "grad_norm": 0.33577436208724976,
      "learning_rate": 1.0641747973065826e-05,
      "loss": 0.6052,
      "step": 3411
    },
    {
      "epoch": 0.9372338964427963,
      "grad_norm": 0.43283456563949585,
      "learning_rate": 1.0638999587742203e-05,
      "loss": 0.7935,
      "step": 3412
    },
    {
      "epoch": 0.9375085839857162,
      "grad_norm": 0.4030177891254425,
      "learning_rate": 1.063625120241858e-05,
      "loss": 0.5204,
      "step": 3413
    },
    {
      "epoch": 0.9377832715286362,
      "grad_norm": 0.32486090064048767,
      "learning_rate": 1.0633502817094959e-05,
      "loss": 0.5921,
      "step": 3414
    },
    {
      "epoch": 0.9380579590715561,
      "grad_norm": 0.4790309965610504,
      "learning_rate": 1.0630754431771336e-05,
      "loss": 0.6841,
      "step": 3415
    },
    {
      "epoch": 0.938332646614476,
      "grad_norm": 0.459753155708313,
      "learning_rate": 1.0628006046447713e-05,
      "loss": 0.7685,
      "step": 3416
    },
    {
      "epoch": 0.938607334157396,
      "grad_norm": 0.40922099351882935,
      "learning_rate": 1.0625257661124091e-05,
      "loss": 0.8649,
      "step": 3417
    },
    {
      "epoch": 0.9388820217003159,
      "grad_norm": 0.3832552134990692,
      "learning_rate": 1.0622509275800468e-05,
      "loss": 0.9737,
      "step": 3418
    },
    {
      "epoch": 0.9391567092432358,
      "grad_norm": 0.5063005089759827,
      "learning_rate": 1.0619760890476847e-05,
      "loss": 0.9059,
      "step": 3419
    },
    {
      "epoch": 0.9394313967861557,
      "grad_norm": 0.45283815264701843,
      "learning_rate": 1.0617012505153224e-05,
      "loss": 0.9346,
      "step": 3420
    },
    {
      "epoch": 0.9397060843290757,
      "grad_norm": 0.4152452051639557,
      "learning_rate": 1.0614264119829601e-05,
      "loss": 1.0226,
      "step": 3421
    },
    {
      "epoch": 0.9399807718719956,
      "grad_norm": 0.6699652671813965,
      "learning_rate": 1.061151573450598e-05,
      "loss": 0.7186,
      "step": 3422
    },
    {
      "epoch": 0.9402554594149155,
      "grad_norm": 0.3912934362888336,
      "learning_rate": 1.0608767349182357e-05,
      "loss": 0.6726,
      "step": 3423
    },
    {
      "epoch": 0.9405301469578354,
      "grad_norm": 0.5550257563591003,
      "learning_rate": 1.0606018963858732e-05,
      "loss": 0.6031,
      "step": 3424
    },
    {
      "epoch": 0.9408048345007554,
      "grad_norm": 0.45351648330688477,
      "learning_rate": 1.0603270578535113e-05,
      "loss": 0.9449,
      "step": 3425
    },
    {
      "epoch": 0.9410795220436753,
      "grad_norm": 0.5015707015991211,
      "learning_rate": 1.0600522193211488e-05,
      "loss": 0.6771,
      "step": 3426
    },
    {
      "epoch": 0.9413542095865952,
      "grad_norm": 0.41755613684654236,
      "learning_rate": 1.0597773807887868e-05,
      "loss": 0.7425,
      "step": 3427
    },
    {
      "epoch": 0.9416288971295151,
      "grad_norm": 0.4781467616558075,
      "learning_rate": 1.0595025422564245e-05,
      "loss": 0.5561,
      "step": 3428
    },
    {
      "epoch": 0.9419035846724351,
      "grad_norm": 0.4753360152244568,
      "learning_rate": 1.059227703724062e-05,
      "loss": 0.7974,
      "step": 3429
    },
    {
      "epoch": 0.942178272215355,
      "grad_norm": 0.35272273421287537,
      "learning_rate": 1.0589528651917001e-05,
      "loss": 0.5744,
      "step": 3430
    },
    {
      "epoch": 0.9424529597582749,
      "grad_norm": 0.33814942836761475,
      "learning_rate": 1.0586780266593376e-05,
      "loss": 0.915,
      "step": 3431
    },
    {
      "epoch": 0.9427276473011948,
      "grad_norm": 0.4248135983943939,
      "learning_rate": 1.0584031881269753e-05,
      "loss": 0.9867,
      "step": 3432
    },
    {
      "epoch": 0.9430023348441148,
      "grad_norm": 0.38870468735694885,
      "learning_rate": 1.0581283495946134e-05,
      "loss": 0.6044,
      "step": 3433
    },
    {
      "epoch": 0.9432770223870347,
      "grad_norm": 0.4382290840148926,
      "learning_rate": 1.057853511062251e-05,
      "loss": 1.0196,
      "step": 3434
    },
    {
      "epoch": 0.9435517099299546,
      "grad_norm": 0.29979637265205383,
      "learning_rate": 1.057578672529889e-05,
      "loss": 0.6454,
      "step": 3435
    },
    {
      "epoch": 0.9438263974728746,
      "grad_norm": 0.362723171710968,
      "learning_rate": 1.0573038339975265e-05,
      "loss": 0.8699,
      "step": 3436
    },
    {
      "epoch": 0.9441010850157945,
      "grad_norm": 0.6236101388931274,
      "learning_rate": 1.0570289954651642e-05,
      "loss": 0.7172,
      "step": 3437
    },
    {
      "epoch": 0.9443757725587144,
      "grad_norm": 0.39172685146331787,
      "learning_rate": 1.056754156932802e-05,
      "loss": 0.9662,
      "step": 3438
    },
    {
      "epoch": 0.9446504601016343,
      "grad_norm": 0.34201085567474365,
      "learning_rate": 1.0564793184004398e-05,
      "loss": 0.7965,
      "step": 3439
    },
    {
      "epoch": 0.9449251476445543,
      "grad_norm": 0.3824758529663086,
      "learning_rate": 1.0562044798680775e-05,
      "loss": 0.9889,
      "step": 3440
    },
    {
      "epoch": 0.9451998351874743,
      "grad_norm": 0.4421567916870117,
      "learning_rate": 1.0559296413357153e-05,
      "loss": 0.9906,
      "step": 3441
    },
    {
      "epoch": 0.9454745227303942,
      "grad_norm": 0.41488853096961975,
      "learning_rate": 1.055654802803353e-05,
      "loss": 0.7389,
      "step": 3442
    },
    {
      "epoch": 0.9457492102733142,
      "grad_norm": 0.47174394130706787,
      "learning_rate": 1.055379964270991e-05,
      "loss": 0.5571,
      "step": 3443
    },
    {
      "epoch": 0.9460238978162341,
      "grad_norm": 0.45016223192214966,
      "learning_rate": 1.0551051257386286e-05,
      "loss": 0.8452,
      "step": 3444
    },
    {
      "epoch": 0.946298585359154,
      "grad_norm": 0.48184508085250854,
      "learning_rate": 1.0548302872062663e-05,
      "loss": 0.817,
      "step": 3445
    },
    {
      "epoch": 0.9465732729020739,
      "grad_norm": 0.5387427806854248,
      "learning_rate": 1.0545554486739042e-05,
      "loss": 0.6384,
      "step": 3446
    },
    {
      "epoch": 0.9468479604449939,
      "grad_norm": 0.29767361283302307,
      "learning_rate": 1.0542806101415419e-05,
      "loss": 0.6753,
      "step": 3447
    },
    {
      "epoch": 0.9471226479879138,
      "grad_norm": 0.6502910256385803,
      "learning_rate": 1.0540057716091796e-05,
      "loss": 1.0243,
      "step": 3448
    },
    {
      "epoch": 0.9473973355308337,
      "grad_norm": 0.2949610948562622,
      "learning_rate": 1.0537309330768175e-05,
      "loss": 0.6807,
      "step": 3449
    },
    {
      "epoch": 0.9476720230737536,
      "grad_norm": 0.4021822512149811,
      "learning_rate": 1.0534560945444552e-05,
      "loss": 0.8286,
      "step": 3450
    },
    {
      "epoch": 0.9479467106166736,
      "grad_norm": 0.5276903510093689,
      "learning_rate": 1.053181256012093e-05,
      "loss": 0.7979,
      "step": 3451
    },
    {
      "epoch": 0.9482213981595935,
      "grad_norm": 0.37115687131881714,
      "learning_rate": 1.0529064174797307e-05,
      "loss": 0.7409,
      "step": 3452
    },
    {
      "epoch": 0.9484960857025134,
      "grad_norm": 0.40323197841644287,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 0.7736,
      "step": 3453
    },
    {
      "epoch": 0.9487707732454334,
      "grad_norm": 0.5238466858863831,
      "learning_rate": 1.0523567404150063e-05,
      "loss": 0.6987,
      "step": 3454
    },
    {
      "epoch": 0.9490454607883533,
      "grad_norm": 0.3032049238681793,
      "learning_rate": 1.052081901882644e-05,
      "loss": 0.7544,
      "step": 3455
    },
    {
      "epoch": 0.9493201483312732,
      "grad_norm": 0.45495739579200745,
      "learning_rate": 1.0518070633502817e-05,
      "loss": 0.9093,
      "step": 3456
    },
    {
      "epoch": 0.9495948358741931,
      "grad_norm": 0.47376498579978943,
      "learning_rate": 1.0515322248179196e-05,
      "loss": 0.7361,
      "step": 3457
    },
    {
      "epoch": 0.9498695234171131,
      "grad_norm": 0.44213151931762695,
      "learning_rate": 1.0512573862855573e-05,
      "loss": 0.8824,
      "step": 3458
    },
    {
      "epoch": 0.950144210960033,
      "grad_norm": 0.4305657148361206,
      "learning_rate": 1.0509825477531952e-05,
      "loss": 0.4606,
      "step": 3459
    },
    {
      "epoch": 0.9504188985029529,
      "grad_norm": 0.51978999376297,
      "learning_rate": 1.0507077092208329e-05,
      "loss": 0.6934,
      "step": 3460
    },
    {
      "epoch": 0.9506935860458728,
      "grad_norm": 0.5031760931015015,
      "learning_rate": 1.0504328706884706e-05,
      "loss": 0.842,
      "step": 3461
    },
    {
      "epoch": 0.9509682735887928,
      "grad_norm": 0.6276075839996338,
      "learning_rate": 1.0501580321561084e-05,
      "loss": 0.7074,
      "step": 3462
    },
    {
      "epoch": 0.9512429611317127,
      "grad_norm": 0.47487232089042664,
      "learning_rate": 1.0498831936237461e-05,
      "loss": 0.6801,
      "step": 3463
    },
    {
      "epoch": 0.9515176486746326,
      "grad_norm": 0.38165783882141113,
      "learning_rate": 1.0496083550913838e-05,
      "loss": 0.756,
      "step": 3464
    },
    {
      "epoch": 0.9517923362175525,
      "grad_norm": 0.38636425137519836,
      "learning_rate": 1.0493335165590217e-05,
      "loss": 0.9309,
      "step": 3465
    },
    {
      "epoch": 0.9520670237604725,
      "grad_norm": 0.4340437054634094,
      "learning_rate": 1.0490586780266594e-05,
      "loss": 0.6672,
      "step": 3466
    },
    {
      "epoch": 0.9523417113033924,
      "grad_norm": 0.3973347544670105,
      "learning_rate": 1.0487838394942973e-05,
      "loss": 0.7688,
      "step": 3467
    },
    {
      "epoch": 0.9526163988463123,
      "grad_norm": 0.3525451421737671,
      "learning_rate": 1.048509000961935e-05,
      "loss": 0.7494,
      "step": 3468
    },
    {
      "epoch": 0.9528910863892323,
      "grad_norm": 0.42509981989860535,
      "learning_rate": 1.0482341624295727e-05,
      "loss": 0.9495,
      "step": 3469
    },
    {
      "epoch": 0.9531657739321522,
      "grad_norm": 0.48979511857032776,
      "learning_rate": 1.0479593238972106e-05,
      "loss": 0.7126,
      "step": 3470
    },
    {
      "epoch": 0.9534404614750721,
      "grad_norm": 0.3853795826435089,
      "learning_rate": 1.0476844853648483e-05,
      "loss": 0.7834,
      "step": 3471
    },
    {
      "epoch": 0.953715149017992,
      "grad_norm": 0.49972298741340637,
      "learning_rate": 1.047409646832486e-05,
      "loss": 0.8772,
      "step": 3472
    },
    {
      "epoch": 0.953989836560912,
      "grad_norm": 0.4070413410663605,
      "learning_rate": 1.0471348083001238e-05,
      "loss": 0.7892,
      "step": 3473
    },
    {
      "epoch": 0.9542645241038319,
      "grad_norm": 0.4023621678352356,
      "learning_rate": 1.0468599697677615e-05,
      "loss": 0.7221,
      "step": 3474
    },
    {
      "epoch": 0.9545392116467518,
      "grad_norm": 0.4372531473636627,
      "learning_rate": 1.0465851312353994e-05,
      "loss": 0.7629,
      "step": 3475
    },
    {
      "epoch": 0.9548138991896717,
      "grad_norm": 0.4789023697376251,
      "learning_rate": 1.0463102927030371e-05,
      "loss": 0.7669,
      "step": 3476
    },
    {
      "epoch": 0.9550885867325917,
      "grad_norm": 0.33191755414009094,
      "learning_rate": 1.0460354541706748e-05,
      "loss": 0.9325,
      "step": 3477
    },
    {
      "epoch": 0.9553632742755116,
      "grad_norm": 0.41308414936065674,
      "learning_rate": 1.0457606156383127e-05,
      "loss": 0.6629,
      "step": 3478
    },
    {
      "epoch": 0.9556379618184315,
      "grad_norm": 0.4417121410369873,
      "learning_rate": 1.0454857771059504e-05,
      "loss": 0.7717,
      "step": 3479
    },
    {
      "epoch": 0.9559126493613515,
      "grad_norm": 0.33432692289352417,
      "learning_rate": 1.0452109385735881e-05,
      "loss": 0.4183,
      "step": 3480
    },
    {
      "epoch": 0.9561873369042714,
      "grad_norm": 0.4539678394794464,
      "learning_rate": 1.044936100041226e-05,
      "loss": 0.8324,
      "step": 3481
    },
    {
      "epoch": 0.9564620244471913,
      "grad_norm": 0.37105610966682434,
      "learning_rate": 1.0446612615088637e-05,
      "loss": 0.7375,
      "step": 3482
    },
    {
      "epoch": 0.9567367119901112,
      "grad_norm": 0.46459051966667175,
      "learning_rate": 1.0443864229765015e-05,
      "loss": 0.5757,
      "step": 3483
    },
    {
      "epoch": 0.9570113995330312,
      "grad_norm": 0.4490734338760376,
      "learning_rate": 1.0441115844441392e-05,
      "loss": 0.6831,
      "step": 3484
    },
    {
      "epoch": 0.9572860870759511,
      "grad_norm": 0.41708293557167053,
      "learning_rate": 1.043836745911777e-05,
      "loss": 0.8437,
      "step": 3485
    },
    {
      "epoch": 0.957560774618871,
      "grad_norm": 0.5814176797866821,
      "learning_rate": 1.0435619073794148e-05,
      "loss": 0.7736,
      "step": 3486
    },
    {
      "epoch": 0.9578354621617909,
      "grad_norm": 0.4687255322933197,
      "learning_rate": 1.0432870688470525e-05,
      "loss": 0.7098,
      "step": 3487
    },
    {
      "epoch": 0.9581101497047109,
      "grad_norm": 0.35038402676582336,
      "learning_rate": 1.04301223031469e-05,
      "loss": 0.6803,
      "step": 3488
    },
    {
      "epoch": 0.9583848372476308,
      "grad_norm": 0.5776459574699402,
      "learning_rate": 1.0427373917823281e-05,
      "loss": 0.7295,
      "step": 3489
    },
    {
      "epoch": 0.9586595247905507,
      "grad_norm": 0.40931206941604614,
      "learning_rate": 1.0424625532499656e-05,
      "loss": 0.7894,
      "step": 3490
    },
    {
      "epoch": 0.9589342123334706,
      "grad_norm": 0.4615342915058136,
      "learning_rate": 1.0421877147176033e-05,
      "loss": 0.6244,
      "step": 3491
    },
    {
      "epoch": 0.9592088998763906,
      "grad_norm": 0.42673611640930176,
      "learning_rate": 1.0419128761852414e-05,
      "loss": 0.5246,
      "step": 3492
    },
    {
      "epoch": 0.9594835874193105,
      "grad_norm": 0.5128232836723328,
      "learning_rate": 1.0416380376528789e-05,
      "loss": 0.8696,
      "step": 3493
    },
    {
      "epoch": 0.9597582749622304,
      "grad_norm": 0.49929043650627136,
      "learning_rate": 1.041363199120517e-05,
      "loss": 0.8065,
      "step": 3494
    },
    {
      "epoch": 0.9600329625051504,
      "grad_norm": 0.5832487940788269,
      "learning_rate": 1.0410883605881545e-05,
      "loss": 0.8261,
      "step": 3495
    },
    {
      "epoch": 0.9603076500480703,
      "grad_norm": 0.37833720445632935,
      "learning_rate": 1.0408135220557922e-05,
      "loss": 0.8637,
      "step": 3496
    },
    {
      "epoch": 0.9605823375909902,
      "grad_norm": 0.4356192648410797,
      "learning_rate": 1.04053868352343e-05,
      "loss": 0.7455,
      "step": 3497
    },
    {
      "epoch": 0.9608570251339101,
      "grad_norm": 0.4866577684879303,
      "learning_rate": 1.0402638449910677e-05,
      "loss": 0.803,
      "step": 3498
    },
    {
      "epoch": 0.9611317126768301,
      "grad_norm": 0.41742339730262756,
      "learning_rate": 1.0399890064587055e-05,
      "loss": 0.735,
      "step": 3499
    },
    {
      "epoch": 0.96140640021975,
      "grad_norm": 0.4263286888599396,
      "learning_rate": 1.0397141679263433e-05,
      "loss": 1.0721,
      "step": 3500
    },
    {
      "epoch": 0.9616810877626699,
      "grad_norm": 0.3995981216430664,
      "learning_rate": 1.039439329393981e-05,
      "loss": 0.5403,
      "step": 3501
    },
    {
      "epoch": 0.9619557753055898,
      "grad_norm": 0.2971172630786896,
      "learning_rate": 1.0391644908616189e-05,
      "loss": 0.8254,
      "step": 3502
    },
    {
      "epoch": 0.9622304628485098,
      "grad_norm": 0.4629015624523163,
      "learning_rate": 1.0388896523292566e-05,
      "loss": 0.8797,
      "step": 3503
    },
    {
      "epoch": 0.9625051503914297,
      "grad_norm": 0.32548418641090393,
      "learning_rate": 1.0386148137968943e-05,
      "loss": 0.6567,
      "step": 3504
    },
    {
      "epoch": 0.9627798379343496,
      "grad_norm": 0.45039960741996765,
      "learning_rate": 1.0383399752645322e-05,
      "loss": 0.6875,
      "step": 3505
    },
    {
      "epoch": 0.9630545254772696,
      "grad_norm": 0.5396274328231812,
      "learning_rate": 1.0380651367321699e-05,
      "loss": 0.6739,
      "step": 3506
    },
    {
      "epoch": 0.9633292130201896,
      "grad_norm": 0.3866516351699829,
      "learning_rate": 1.0377902981998076e-05,
      "loss": 0.8969,
      "step": 3507
    },
    {
      "epoch": 0.9636039005631095,
      "grad_norm": 0.4911908805370331,
      "learning_rate": 1.0375154596674454e-05,
      "loss": 0.8556,
      "step": 3508
    },
    {
      "epoch": 0.9638785881060294,
      "grad_norm": 0.3845323622226715,
      "learning_rate": 1.0372406211350831e-05,
      "loss": 0.7565,
      "step": 3509
    },
    {
      "epoch": 0.9641532756489494,
      "grad_norm": 0.38927289843559265,
      "learning_rate": 1.036965782602721e-05,
      "loss": 0.8826,
      "step": 3510
    },
    {
      "epoch": 0.9644279631918693,
      "grad_norm": 0.5762121677398682,
      "learning_rate": 1.0366909440703587e-05,
      "loss": 0.7922,
      "step": 3511
    },
    {
      "epoch": 0.9647026507347892,
      "grad_norm": 0.4726410508155823,
      "learning_rate": 1.0364161055379964e-05,
      "loss": 0.6534,
      "step": 3512
    },
    {
      "epoch": 0.9649773382777092,
      "grad_norm": 0.34655457735061646,
      "learning_rate": 1.0361412670056343e-05,
      "loss": 0.6525,
      "step": 3513
    },
    {
      "epoch": 0.9652520258206291,
      "grad_norm": 0.41961249709129333,
      "learning_rate": 1.035866428473272e-05,
      "loss": 0.7815,
      "step": 3514
    },
    {
      "epoch": 0.965526713363549,
      "grad_norm": 0.40892985463142395,
      "learning_rate": 1.0355915899409097e-05,
      "loss": 0.6287,
      "step": 3515
    },
    {
      "epoch": 0.9658014009064689,
      "grad_norm": 0.39567291736602783,
      "learning_rate": 1.0353167514085476e-05,
      "loss": 0.853,
      "step": 3516
    },
    {
      "epoch": 0.9660760884493889,
      "grad_norm": 0.47972142696380615,
      "learning_rate": 1.0350419128761853e-05,
      "loss": 0.6888,
      "step": 3517
    },
    {
      "epoch": 0.9663507759923088,
      "grad_norm": 0.4460701644420624,
      "learning_rate": 1.0347670743438231e-05,
      "loss": 0.7618,
      "step": 3518
    },
    {
      "epoch": 0.9666254635352287,
      "grad_norm": 0.49363526701927185,
      "learning_rate": 1.0344922358114608e-05,
      "loss": 0.956,
      "step": 3519
    },
    {
      "epoch": 0.9669001510781486,
      "grad_norm": 0.41439035534858704,
      "learning_rate": 1.0342173972790985e-05,
      "loss": 0.9509,
      "step": 3520
    },
    {
      "epoch": 0.9671748386210686,
      "grad_norm": 0.48853710293769836,
      "learning_rate": 1.0339425587467364e-05,
      "loss": 0.7339,
      "step": 3521
    },
    {
      "epoch": 0.9674495261639885,
      "grad_norm": 0.5988669991493225,
      "learning_rate": 1.0336677202143741e-05,
      "loss": 0.8438,
      "step": 3522
    },
    {
      "epoch": 0.9677242137069084,
      "grad_norm": 0.640031635761261,
      "learning_rate": 1.0333928816820118e-05,
      "loss": 0.7849,
      "step": 3523
    },
    {
      "epoch": 0.9679989012498283,
      "grad_norm": 0.4746118485927582,
      "learning_rate": 1.0331180431496497e-05,
      "loss": 0.9775,
      "step": 3524
    },
    {
      "epoch": 0.9682735887927483,
      "grad_norm": 0.4792206287384033,
      "learning_rate": 1.0328432046172874e-05,
      "loss": 0.7391,
      "step": 3525
    },
    {
      "epoch": 0.9685482763356682,
      "grad_norm": 0.48237931728363037,
      "learning_rate": 1.0325683660849253e-05,
      "loss": 0.6251,
      "step": 3526
    },
    {
      "epoch": 0.9688229638785881,
      "grad_norm": 0.46685928106307983,
      "learning_rate": 1.032293527552563e-05,
      "loss": 0.5909,
      "step": 3527
    },
    {
      "epoch": 0.969097651421508,
      "grad_norm": 0.463850200176239,
      "learning_rate": 1.0320186890202007e-05,
      "loss": 0.5929,
      "step": 3528
    },
    {
      "epoch": 0.969372338964428,
      "grad_norm": 0.4325132369995117,
      "learning_rate": 1.0317438504878385e-05,
      "loss": 0.7148,
      "step": 3529
    },
    {
      "epoch": 0.9696470265073479,
      "grad_norm": 0.46691611409187317,
      "learning_rate": 1.0314690119554762e-05,
      "loss": 0.8906,
      "step": 3530
    },
    {
      "epoch": 0.9699217140502678,
      "grad_norm": 0.42744210362434387,
      "learning_rate": 1.031194173423114e-05,
      "loss": 0.6428,
      "step": 3531
    },
    {
      "epoch": 0.9701964015931878,
      "grad_norm": 0.3958432376384735,
      "learning_rate": 1.0309193348907518e-05,
      "loss": 0.6547,
      "step": 3532
    },
    {
      "epoch": 0.9704710891361077,
      "grad_norm": 0.5026724934577942,
      "learning_rate": 1.0306444963583895e-05,
      "loss": 0.8583,
      "step": 3533
    },
    {
      "epoch": 0.9707457766790276,
      "grad_norm": 0.33033519983291626,
      "learning_rate": 1.0303696578260274e-05,
      "loss": 0.465,
      "step": 3534
    },
    {
      "epoch": 0.9710204642219475,
      "grad_norm": 0.4438226819038391,
      "learning_rate": 1.0300948192936651e-05,
      "loss": 0.7357,
      "step": 3535
    },
    {
      "epoch": 0.9712951517648675,
      "grad_norm": 0.42194581031799316,
      "learning_rate": 1.0298199807613028e-05,
      "loss": 0.8681,
      "step": 3536
    },
    {
      "epoch": 0.9715698393077874,
      "grad_norm": 0.4394244849681854,
      "learning_rate": 1.0295451422289407e-05,
      "loss": 0.5086,
      "step": 3537
    },
    {
      "epoch": 0.9718445268507073,
      "grad_norm": 0.5173270106315613,
      "learning_rate": 1.0292703036965784e-05,
      "loss": 0.7226,
      "step": 3538
    },
    {
      "epoch": 0.9721192143936273,
      "grad_norm": 0.3122175931930542,
      "learning_rate": 1.028995465164216e-05,
      "loss": 0.4992,
      "step": 3539
    },
    {
      "epoch": 0.9723939019365472,
      "grad_norm": 0.41186216473579407,
      "learning_rate": 1.028720626631854e-05,
      "loss": 0.8097,
      "step": 3540
    },
    {
      "epoch": 0.9726685894794671,
      "grad_norm": 0.43987950682640076,
      "learning_rate": 1.0284457880994916e-05,
      "loss": 0.9756,
      "step": 3541
    },
    {
      "epoch": 0.972943277022387,
      "grad_norm": 0.3639916777610779,
      "learning_rate": 1.0281709495671295e-05,
      "loss": 0.9152,
      "step": 3542
    },
    {
      "epoch": 0.973217964565307,
      "grad_norm": 0.345971018075943,
      "learning_rate": 1.0278961110347672e-05,
      "loss": 0.5695,
      "step": 3543
    },
    {
      "epoch": 0.9734926521082269,
      "grad_norm": 0.36689701676368713,
      "learning_rate": 1.027621272502405e-05,
      "loss": 0.7335,
      "step": 3544
    },
    {
      "epoch": 0.9737673396511468,
      "grad_norm": 0.39604488015174866,
      "learning_rate": 1.0273464339700428e-05,
      "loss": 0.5371,
      "step": 3545
    },
    {
      "epoch": 0.9740420271940667,
      "grad_norm": 0.348307341337204,
      "learning_rate": 1.0270715954376805e-05,
      "loss": 0.6456,
      "step": 3546
    },
    {
      "epoch": 0.9743167147369867,
      "grad_norm": 0.33245041966438293,
      "learning_rate": 1.026796756905318e-05,
      "loss": 0.8192,
      "step": 3547
    },
    {
      "epoch": 0.9745914022799066,
      "grad_norm": 0.3890763521194458,
      "learning_rate": 1.026521918372956e-05,
      "loss": 0.7531,
      "step": 3548
    },
    {
      "epoch": 0.9748660898228265,
      "grad_norm": 0.42552876472473145,
      "learning_rate": 1.0262470798405936e-05,
      "loss": 0.8486,
      "step": 3549
    },
    {
      "epoch": 0.9751407773657464,
      "grad_norm": 0.30184468626976013,
      "learning_rate": 1.0259722413082316e-05,
      "loss": 0.4457,
      "step": 3550
    },
    {
      "epoch": 0.9754154649086664,
      "grad_norm": 0.42904379963874817,
      "learning_rate": 1.0256974027758693e-05,
      "loss": 0.7142,
      "step": 3551
    },
    {
      "epoch": 0.9756901524515863,
      "grad_norm": 0.42363569140434265,
      "learning_rate": 1.0254225642435069e-05,
      "loss": 0.9434,
      "step": 3552
    },
    {
      "epoch": 0.9759648399945062,
      "grad_norm": 0.3336479365825653,
      "learning_rate": 1.025147725711145e-05,
      "loss": 0.8,
      "step": 3553
    },
    {
      "epoch": 0.9762395275374262,
      "grad_norm": 0.4110511839389801,
      "learning_rate": 1.0248728871787825e-05,
      "loss": 0.8223,
      "step": 3554
    },
    {
      "epoch": 0.9765142150803461,
      "grad_norm": 0.425628662109375,
      "learning_rate": 1.0245980486464202e-05,
      "loss": 0.6472,
      "step": 3555
    },
    {
      "epoch": 0.976788902623266,
      "grad_norm": 0.352995902299881,
      "learning_rate": 1.024323210114058e-05,
      "loss": 0.7555,
      "step": 3556
    },
    {
      "epoch": 0.9770635901661859,
      "grad_norm": 0.5201056003570557,
      "learning_rate": 1.0240483715816957e-05,
      "loss": 0.8158,
      "step": 3557
    },
    {
      "epoch": 0.9773382777091059,
      "grad_norm": 0.5116705894470215,
      "learning_rate": 1.0237735330493338e-05,
      "loss": 0.8715,
      "step": 3558
    },
    {
      "epoch": 0.9776129652520258,
      "grad_norm": 0.4445383548736572,
      "learning_rate": 1.0234986945169713e-05,
      "loss": 0.7972,
      "step": 3559
    },
    {
      "epoch": 0.9778876527949457,
      "grad_norm": 0.4756864011287689,
      "learning_rate": 1.023223855984609e-05,
      "loss": 0.9944,
      "step": 3560
    },
    {
      "epoch": 0.9781623403378656,
      "grad_norm": 0.48496192693710327,
      "learning_rate": 1.0229490174522469e-05,
      "loss": 0.8837,
      "step": 3561
    },
    {
      "epoch": 0.9784370278807856,
      "grad_norm": 0.6488956809043884,
      "learning_rate": 1.0226741789198846e-05,
      "loss": 0.8631,
      "step": 3562
    },
    {
      "epoch": 0.9787117154237055,
      "grad_norm": 0.48001301288604736,
      "learning_rate": 1.0223993403875223e-05,
      "loss": 0.9403,
      "step": 3563
    },
    {
      "epoch": 0.9789864029666254,
      "grad_norm": 0.3998594284057617,
      "learning_rate": 1.0221245018551602e-05,
      "loss": 0.8107,
      "step": 3564
    },
    {
      "epoch": 0.9792610905095454,
      "grad_norm": 0.3702256381511688,
      "learning_rate": 1.0218496633227979e-05,
      "loss": 0.6651,
      "step": 3565
    },
    {
      "epoch": 0.9795357780524653,
      "grad_norm": 0.4133528470993042,
      "learning_rate": 1.0215748247904357e-05,
      "loss": 0.7646,
      "step": 3566
    },
    {
      "epoch": 0.9798104655953852,
      "grad_norm": 0.3897297978401184,
      "learning_rate": 1.0212999862580734e-05,
      "loss": 0.7511,
      "step": 3567
    },
    {
      "epoch": 0.9800851531383051,
      "grad_norm": 0.44021204113960266,
      "learning_rate": 1.0210251477257111e-05,
      "loss": 0.8604,
      "step": 3568
    },
    {
      "epoch": 0.9803598406812251,
      "grad_norm": 0.381754606962204,
      "learning_rate": 1.020750309193349e-05,
      "loss": 0.9103,
      "step": 3569
    },
    {
      "epoch": 0.980634528224145,
      "grad_norm": 0.3415520489215851,
      "learning_rate": 1.0204754706609867e-05,
      "loss": 0.6239,
      "step": 3570
    },
    {
      "epoch": 0.9809092157670649,
      "grad_norm": 0.47775998711586,
      "learning_rate": 1.0202006321286244e-05,
      "loss": 0.7059,
      "step": 3571
    },
    {
      "epoch": 0.9811839033099848,
      "grad_norm": 0.48240458965301514,
      "learning_rate": 1.0199257935962623e-05,
      "loss": 0.7552,
      "step": 3572
    },
    {
      "epoch": 0.9814585908529048,
      "grad_norm": 0.38294318318367004,
      "learning_rate": 1.0196509550639e-05,
      "loss": 0.6998,
      "step": 3573
    },
    {
      "epoch": 0.9817332783958248,
      "grad_norm": 0.3943803012371063,
      "learning_rate": 1.0193761165315379e-05,
      "loss": 0.7679,
      "step": 3574
    },
    {
      "epoch": 0.9820079659387447,
      "grad_norm": 0.4505015015602112,
      "learning_rate": 1.0191012779991756e-05,
      "loss": 0.6978,
      "step": 3575
    },
    {
      "epoch": 0.9822826534816647,
      "grad_norm": 0.38551026582717896,
      "learning_rate": 1.0188264394668133e-05,
      "loss": 0.8891,
      "step": 3576
    },
    {
      "epoch": 0.9825573410245846,
      "grad_norm": 0.41836118698120117,
      "learning_rate": 1.0185516009344511e-05,
      "loss": 0.6672,
      "step": 3577
    },
    {
      "epoch": 0.9828320285675045,
      "grad_norm": 0.5301141738891602,
      "learning_rate": 1.0182767624020888e-05,
      "loss": 0.9471,
      "step": 3578
    },
    {
      "epoch": 0.9831067161104244,
      "grad_norm": 0.47068366408348083,
      "learning_rate": 1.0180019238697265e-05,
      "loss": 0.7151,
      "step": 3579
    },
    {
      "epoch": 0.9833814036533444,
      "grad_norm": 0.41128605604171753,
      "learning_rate": 1.0177270853373644e-05,
      "loss": 0.7052,
      "step": 3580
    },
    {
      "epoch": 0.9836560911962643,
      "grad_norm": 0.4177156984806061,
      "learning_rate": 1.0174522468050021e-05,
      "loss": 0.7087,
      "step": 3581
    },
    {
      "epoch": 0.9839307787391842,
      "grad_norm": 0.34703949093818665,
      "learning_rate": 1.01717740827264e-05,
      "loss": 0.9101,
      "step": 3582
    },
    {
      "epoch": 0.9842054662821041,
      "grad_norm": 0.4179423451423645,
      "learning_rate": 1.0169025697402777e-05,
      "loss": 0.8485,
      "step": 3583
    },
    {
      "epoch": 0.9844801538250241,
      "grad_norm": 0.6361936926841736,
      "learning_rate": 1.0166277312079154e-05,
      "loss": 0.8534,
      "step": 3584
    },
    {
      "epoch": 0.984754841367944,
      "grad_norm": 0.43126776814460754,
      "learning_rate": 1.0163528926755533e-05,
      "loss": 0.681,
      "step": 3585
    },
    {
      "epoch": 0.9850295289108639,
      "grad_norm": 0.37301862239837646,
      "learning_rate": 1.016078054143191e-05,
      "loss": 0.8683,
      "step": 3586
    },
    {
      "epoch": 0.9853042164537839,
      "grad_norm": 0.4738551676273346,
      "learning_rate": 1.0158032156108287e-05,
      "loss": 0.9137,
      "step": 3587
    },
    {
      "epoch": 0.9855789039967038,
      "grad_norm": 0.34116634726524353,
      "learning_rate": 1.0155283770784665e-05,
      "loss": 0.6779,
      "step": 3588
    },
    {
      "epoch": 0.9858535915396237,
      "grad_norm": 0.5737759470939636,
      "learning_rate": 1.0152535385461042e-05,
      "loss": 0.8474,
      "step": 3589
    },
    {
      "epoch": 0.9861282790825436,
      "grad_norm": 0.4838612675666809,
      "learning_rate": 1.0149787000137421e-05,
      "loss": 0.651,
      "step": 3590
    },
    {
      "epoch": 0.9864029666254636,
      "grad_norm": 0.39447590708732605,
      "learning_rate": 1.0147038614813798e-05,
      "loss": 0.9079,
      "step": 3591
    },
    {
      "epoch": 0.9866776541683835,
      "grad_norm": 0.392793744802475,
      "learning_rate": 1.0144290229490175e-05,
      "loss": 0.657,
      "step": 3592
    },
    {
      "epoch": 0.9869523417113034,
      "grad_norm": 0.7363808751106262,
      "learning_rate": 1.0141541844166554e-05,
      "loss": 0.7448,
      "step": 3593
    },
    {
      "epoch": 0.9872270292542233,
      "grad_norm": 0.4069053530693054,
      "learning_rate": 1.013879345884293e-05,
      "loss": 0.7651,
      "step": 3594
    },
    {
      "epoch": 0.9875017167971433,
      "grad_norm": 0.4729968011379242,
      "learning_rate": 1.0136045073519308e-05,
      "loss": 0.8954,
      "step": 3595
    },
    {
      "epoch": 0.9877764043400632,
      "grad_norm": 0.41176795959472656,
      "learning_rate": 1.0133296688195687e-05,
      "loss": 0.8816,
      "step": 3596
    },
    {
      "epoch": 0.9880510918829831,
      "grad_norm": 0.37685057520866394,
      "learning_rate": 1.0130548302872064e-05,
      "loss": 0.9022,
      "step": 3597
    },
    {
      "epoch": 0.988325779425903,
      "grad_norm": 0.4606911838054657,
      "learning_rate": 1.0127799917548442e-05,
      "loss": 0.7042,
      "step": 3598
    },
    {
      "epoch": 0.988600466968823,
      "grad_norm": 0.38871660828590393,
      "learning_rate": 1.012505153222482e-05,
      "loss": 0.8981,
      "step": 3599
    },
    {
      "epoch": 0.9888751545117429,
      "grad_norm": 0.42629286646842957,
      "learning_rate": 1.0122303146901196e-05,
      "loss": 1.0908,
      "step": 3600
    },
    {
      "epoch": 0.9891498420546628,
      "grad_norm": 0.4843009412288666,
      "learning_rate": 1.0119554761577575e-05,
      "loss": 0.6689,
      "step": 3601
    },
    {
      "epoch": 0.9894245295975828,
      "grad_norm": 0.5581785440444946,
      "learning_rate": 1.0116806376253952e-05,
      "loss": 0.7475,
      "step": 3602
    },
    {
      "epoch": 0.9896992171405027,
      "grad_norm": 0.4357980489730835,
      "learning_rate": 1.0114057990930329e-05,
      "loss": 0.7076,
      "step": 3603
    },
    {
      "epoch": 0.9899739046834226,
      "grad_norm": 0.39406296610832214,
      "learning_rate": 1.0111309605606708e-05,
      "loss": 0.7185,
      "step": 3604
    },
    {
      "epoch": 0.9902485922263425,
      "grad_norm": 0.3198079466819763,
      "learning_rate": 1.0108561220283085e-05,
      "loss": 0.8025,
      "step": 3605
    },
    {
      "epoch": 0.9905232797692625,
      "grad_norm": 0.4846961796283722,
      "learning_rate": 1.0105812834959463e-05,
      "loss": 0.8431,
      "step": 3606
    },
    {
      "epoch": 0.9907979673121824,
      "grad_norm": 0.5061333179473877,
      "learning_rate": 1.010306444963584e-05,
      "loss": 0.8636,
      "step": 3607
    },
    {
      "epoch": 0.9910726548551023,
      "grad_norm": 0.44804924726486206,
      "learning_rate": 1.0100316064312216e-05,
      "loss": 0.8046,
      "step": 3608
    },
    {
      "epoch": 0.9913473423980222,
      "grad_norm": 0.40287601947784424,
      "learning_rate": 1.0097567678988596e-05,
      "loss": 0.5812,
      "step": 3609
    },
    {
      "epoch": 0.9916220299409422,
      "grad_norm": 0.4023846387863159,
      "learning_rate": 1.0094819293664973e-05,
      "loss": 1.0396,
      "step": 3610
    },
    {
      "epoch": 0.9918967174838621,
      "grad_norm": 0.3921183943748474,
      "learning_rate": 1.0092070908341349e-05,
      "loss": 0.6646,
      "step": 3611
    },
    {
      "epoch": 0.992171405026782,
      "grad_norm": 0.43720686435699463,
      "learning_rate": 1.0089322523017729e-05,
      "loss": 0.9016,
      "step": 3612
    },
    {
      "epoch": 0.992446092569702,
      "grad_norm": 0.42564478516578674,
      "learning_rate": 1.0086574137694104e-05,
      "loss": 0.834,
      "step": 3613
    },
    {
      "epoch": 0.9927207801126219,
      "grad_norm": 0.37655043601989746,
      "learning_rate": 1.0083825752370485e-05,
      "loss": 0.6916,
      "step": 3614
    },
    {
      "epoch": 0.9929954676555418,
      "grad_norm": 0.5453474521636963,
      "learning_rate": 1.008107736704686e-05,
      "loss": 0.8109,
      "step": 3615
    },
    {
      "epoch": 0.9932701551984617,
      "grad_norm": 0.4201655089855194,
      "learning_rate": 1.0078328981723237e-05,
      "loss": 0.792,
      "step": 3616
    },
    {
      "epoch": 0.9935448427413817,
      "grad_norm": 0.49262502789497375,
      "learning_rate": 1.0075580596399617e-05,
      "loss": 0.8219,
      "step": 3617
    },
    {
      "epoch": 0.9938195302843016,
      "grad_norm": 0.5329756140708923,
      "learning_rate": 1.0072832211075993e-05,
      "loss": 0.9214,
      "step": 3618
    },
    {
      "epoch": 0.9940942178272215,
      "grad_norm": 0.6736170649528503,
      "learning_rate": 1.007008382575237e-05,
      "loss": 0.8043,
      "step": 3619
    },
    {
      "epoch": 0.9943689053701414,
      "grad_norm": 0.5995938181877136,
      "learning_rate": 1.0067335440428749e-05,
      "loss": 0.8901,
      "step": 3620
    },
    {
      "epoch": 0.9946435929130614,
      "grad_norm": 0.33236902952194214,
      "learning_rate": 1.0064587055105126e-05,
      "loss": 0.6162,
      "step": 3621
    },
    {
      "epoch": 0.9949182804559813,
      "grad_norm": 0.5294310450553894,
      "learning_rate": 1.0061838669781504e-05,
      "loss": 0.7727,
      "step": 3622
    },
    {
      "epoch": 0.9951929679989012,
      "grad_norm": 0.511700451374054,
      "learning_rate": 1.0059090284457881e-05,
      "loss": 0.9332,
      "step": 3623
    },
    {
      "epoch": 0.9954676555418212,
      "grad_norm": 0.4210036098957062,
      "learning_rate": 1.0056341899134258e-05,
      "loss": 0.8943,
      "step": 3624
    },
    {
      "epoch": 0.9957423430847411,
      "grad_norm": 0.425214558839798,
      "learning_rate": 1.0053593513810637e-05,
      "loss": 0.6603,
      "step": 3625
    },
    {
      "epoch": 0.996017030627661,
      "grad_norm": 0.49906355142593384,
      "learning_rate": 1.0050845128487014e-05,
      "loss": 0.8353,
      "step": 3626
    },
    {
      "epoch": 0.9962917181705809,
      "grad_norm": 0.4537266194820404,
      "learning_rate": 1.0048096743163391e-05,
      "loss": 0.7387,
      "step": 3627
    },
    {
      "epoch": 0.9965664057135009,
      "grad_norm": 0.35782766342163086,
      "learning_rate": 1.004534835783977e-05,
      "loss": 0.9607,
      "step": 3628
    },
    {
      "epoch": 0.9968410932564208,
      "grad_norm": 0.6286492347717285,
      "learning_rate": 1.0042599972516147e-05,
      "loss": 0.8009,
      "step": 3629
    },
    {
      "epoch": 0.9971157807993407,
      "grad_norm": 0.4656655192375183,
      "learning_rate": 1.0039851587192526e-05,
      "loss": 0.878,
      "step": 3630
    },
    {
      "epoch": 0.9973904683422606,
      "grad_norm": 0.44942307472229004,
      "learning_rate": 1.0037103201868903e-05,
      "loss": 0.6019,
      "step": 3631
    },
    {
      "epoch": 0.9976651558851806,
      "grad_norm": 0.36015942692756653,
      "learning_rate": 1.003435481654528e-05,
      "loss": 0.6703,
      "step": 3632
    },
    {
      "epoch": 0.9979398434281005,
      "grad_norm": 0.4545639455318451,
      "learning_rate": 1.0031606431221658e-05,
      "loss": 0.7832,
      "step": 3633
    },
    {
      "epoch": 0.9982145309710204,
      "grad_norm": 0.44608673453330994,
      "learning_rate": 1.0028858045898035e-05,
      "loss": 1.0657,
      "step": 3634
    },
    {
      "epoch": 0.9984892185139403,
      "grad_norm": 0.4870340824127197,
      "learning_rate": 1.0026109660574412e-05,
      "loss": 0.6458,
      "step": 3635
    },
    {
      "epoch": 0.9987639060568603,
      "grad_norm": 0.5649285912513733,
      "learning_rate": 1.0023361275250791e-05,
      "loss": 0.6397,
      "step": 3636
    },
    {
      "epoch": 0.9990385935997802,
      "grad_norm": 0.3944137394428253,
      "learning_rate": 1.0020612889927168e-05,
      "loss": 0.7631,
      "step": 3637
    },
    {
      "epoch": 0.9993132811427001,
      "grad_norm": 0.3731857240200043,
      "learning_rate": 1.0017864504603547e-05,
      "loss": 0.9083,
      "step": 3638
    },
    {
      "epoch": 0.99958796868562,
      "grad_norm": 0.5318137407302856,
      "learning_rate": 1.0015116119279924e-05,
      "loss": 0.859,
      "step": 3639
    },
    {
      "epoch": 0.99986265622854,
      "grad_norm": 0.43524986505508423,
      "learning_rate": 1.00123677339563e-05,
      "loss": 0.8087,
      "step": 3640
    },
    {
      "epoch": 1.00027468754292,
      "grad_norm": 0.7773877382278442,
      "learning_rate": 1.000961934863268e-05,
      "loss": 1.5258,
      "step": 3641
    },
    {
      "epoch": 1.0005493750858399,
      "grad_norm": 0.4738582968711853,
      "learning_rate": 1.0006870963309057e-05,
      "loss": 0.8489,
      "step": 3642
    },
    {
      "epoch": 1.00082406262876,
      "grad_norm": 0.2943694293498993,
      "learning_rate": 1.0004122577985434e-05,
      "loss": 0.7276,
      "step": 3643
    },
    {
      "epoch": 1.0010987501716797,
      "grad_norm": 0.45934486389160156,
      "learning_rate": 1.0001374192661812e-05,
      "loss": 0.7561,
      "step": 3644
    },
    {
      "epoch": 1.0013734377145997,
      "grad_norm": 0.524002730846405,
      "learning_rate": 9.99862580733819e-06,
      "loss": 0.6999,
      "step": 3645
    },
    {
      "epoch": 1.0016481252575196,
      "grad_norm": 0.5366028547286987,
      "learning_rate": 9.995877422014566e-06,
      "loss": 0.6127,
      "step": 3646
    },
    {
      "epoch": 1.0019228128004396,
      "grad_norm": 0.44266971945762634,
      "learning_rate": 9.993129036690945e-06,
      "loss": 0.4697,
      "step": 3647
    },
    {
      "epoch": 1.0021975003433594,
      "grad_norm": 0.4188106060028076,
      "learning_rate": 9.990380651367324e-06,
      "loss": 0.732,
      "step": 3648
    },
    {
      "epoch": 1.0024721878862795,
      "grad_norm": 0.6381589770317078,
      "learning_rate": 9.987632266043699e-06,
      "loss": 0.7324,
      "step": 3649
    },
    {
      "epoch": 1.0027468754291993,
      "grad_norm": 0.47882843017578125,
      "learning_rate": 9.984883880720078e-06,
      "loss": 0.6988,
      "step": 3650
    },
    {
      "epoch": 1.0030215629721193,
      "grad_norm": 0.5032527446746826,
      "learning_rate": 9.982135495396455e-06,
      "loss": 0.884,
      "step": 3651
    },
    {
      "epoch": 1.0032962505150391,
      "grad_norm": 0.3544904589653015,
      "learning_rate": 9.979387110072834e-06,
      "loss": 0.7542,
      "step": 3652
    },
    {
      "epoch": 1.0035709380579592,
      "grad_norm": 0.30218732357025146,
      "learning_rate": 9.97663872474921e-06,
      "loss": 0.8407,
      "step": 3653
    },
    {
      "epoch": 1.003845625600879,
      "grad_norm": 0.4430505931377411,
      "learning_rate": 9.973890339425588e-06,
      "loss": 0.8529,
      "step": 3654
    },
    {
      "epoch": 1.004120313143799,
      "grad_norm": 0.42019811272621155,
      "learning_rate": 9.971141954101966e-06,
      "loss": 0.5633,
      "step": 3655
    },
    {
      "epoch": 1.0043950006867188,
      "grad_norm": 0.4418288767337799,
      "learning_rate": 9.968393568778343e-06,
      "loss": 0.7213,
      "step": 3656
    },
    {
      "epoch": 1.0046696882296389,
      "grad_norm": 0.4904439449310303,
      "learning_rate": 9.96564518345472e-06,
      "loss": 0.8332,
      "step": 3657
    },
    {
      "epoch": 1.0049443757725587,
      "grad_norm": 0.33900830149650574,
      "learning_rate": 9.962896798131099e-06,
      "loss": 0.6871,
      "step": 3658
    },
    {
      "epoch": 1.0052190633154787,
      "grad_norm": 0.38686972856521606,
      "learning_rate": 9.960148412807476e-06,
      "loss": 0.9893,
      "step": 3659
    },
    {
      "epoch": 1.0054937508583985,
      "grad_norm": 0.4352690875530243,
      "learning_rate": 9.957400027483855e-06,
      "loss": 0.7185,
      "step": 3660
    },
    {
      "epoch": 1.0057684384013186,
      "grad_norm": 0.4430035948753357,
      "learning_rate": 9.954651642160232e-06,
      "loss": 0.706,
      "step": 3661
    },
    {
      "epoch": 1.0060431259442384,
      "grad_norm": 0.4669344127178192,
      "learning_rate": 9.951903256836609e-06,
      "loss": 0.6101,
      "step": 3662
    },
    {
      "epoch": 1.0063178134871584,
      "grad_norm": 0.3861996531486511,
      "learning_rate": 9.949154871512988e-06,
      "loss": 0.9065,
      "step": 3663
    },
    {
      "epoch": 1.0065925010300782,
      "grad_norm": 0.43282386660575867,
      "learning_rate": 9.946406486189365e-06,
      "loss": 0.7204,
      "step": 3664
    },
    {
      "epoch": 1.0068671885729983,
      "grad_norm": 0.517202615737915,
      "learning_rate": 9.943658100865742e-06,
      "loss": 0.8704,
      "step": 3665
    },
    {
      "epoch": 1.007141876115918,
      "grad_norm": 0.4609034061431885,
      "learning_rate": 9.94090971554212e-06,
      "loss": 0.8331,
      "step": 3666
    },
    {
      "epoch": 1.0074165636588381,
      "grad_norm": 0.49621063470840454,
      "learning_rate": 9.938161330218497e-06,
      "loss": 0.8428,
      "step": 3667
    },
    {
      "epoch": 1.007691251201758,
      "grad_norm": 0.6222859621047974,
      "learning_rate": 9.935412944894876e-06,
      "loss": 0.7575,
      "step": 3668
    },
    {
      "epoch": 1.007965938744678,
      "grad_norm": 0.558480441570282,
      "learning_rate": 9.932664559571253e-06,
      "loss": 0.582,
      "step": 3669
    },
    {
      "epoch": 1.0082406262875978,
      "grad_norm": 0.48372167348861694,
      "learning_rate": 9.92991617424763e-06,
      "loss": 0.7067,
      "step": 3670
    },
    {
      "epoch": 1.0085153138305178,
      "grad_norm": 0.4184461236000061,
      "learning_rate": 9.927167788924009e-06,
      "loss": 0.7765,
      "step": 3671
    },
    {
      "epoch": 1.0087900013734377,
      "grad_norm": 0.42103704810142517,
      "learning_rate": 9.924419403600386e-06,
      "loss": 0.9257,
      "step": 3672
    },
    {
      "epoch": 1.0090646889163577,
      "grad_norm": 0.5034600496292114,
      "learning_rate": 9.921671018276763e-06,
      "loss": 0.8385,
      "step": 3673
    },
    {
      "epoch": 1.0093393764592775,
      "grad_norm": 0.5352590084075928,
      "learning_rate": 9.918922632953142e-06,
      "loss": 0.8825,
      "step": 3674
    },
    {
      "epoch": 1.0096140640021976,
      "grad_norm": 0.34530752897262573,
      "learning_rate": 9.916174247629519e-06,
      "loss": 0.8897,
      "step": 3675
    },
    {
      "epoch": 1.0098887515451174,
      "grad_norm": 0.40129411220550537,
      "learning_rate": 9.913425862305897e-06,
      "loss": 0.825,
      "step": 3676
    },
    {
      "epoch": 1.0101634390880374,
      "grad_norm": 0.43488526344299316,
      "learning_rate": 9.910677476982273e-06,
      "loss": 1.0215,
      "step": 3677
    },
    {
      "epoch": 1.0104381266309572,
      "grad_norm": 0.42354637384414673,
      "learning_rate": 9.907929091658651e-06,
      "loss": 0.7264,
      "step": 3678
    },
    {
      "epoch": 1.0107128141738773,
      "grad_norm": 0.5124500393867493,
      "learning_rate": 9.905180706335028e-06,
      "loss": 0.7169,
      "step": 3679
    },
    {
      "epoch": 1.010987501716797,
      "grad_norm": 0.4788925349712372,
      "learning_rate": 9.902432321011407e-06,
      "loss": 0.6775,
      "step": 3680
    },
    {
      "epoch": 1.0112621892597171,
      "grad_norm": 0.3680455684661865,
      "learning_rate": 9.899683935687784e-06,
      "loss": 0.576,
      "step": 3681
    },
    {
      "epoch": 1.011536876802637,
      "grad_norm": 0.5382637977600098,
      "learning_rate": 9.896935550364161e-06,
      "loss": 0.7924,
      "step": 3682
    },
    {
      "epoch": 1.011811564345557,
      "grad_norm": 0.44476431608200073,
      "learning_rate": 9.89418716504054e-06,
      "loss": 0.6013,
      "step": 3683
    },
    {
      "epoch": 1.0120862518884768,
      "grad_norm": 0.4605240225791931,
      "learning_rate": 9.891438779716917e-06,
      "loss": 0.6884,
      "step": 3684
    },
    {
      "epoch": 1.0123609394313968,
      "grad_norm": 0.4362699091434479,
      "learning_rate": 9.888690394393294e-06,
      "loss": 0.8024,
      "step": 3685
    },
    {
      "epoch": 1.0126356269743166,
      "grad_norm": 0.329354852437973,
      "learning_rate": 9.885942009069673e-06,
      "loss": 0.7326,
      "step": 3686
    },
    {
      "epoch": 1.0129103145172367,
      "grad_norm": 0.36932891607284546,
      "learning_rate": 9.88319362374605e-06,
      "loss": 0.7457,
      "step": 3687
    },
    {
      "epoch": 1.0131850020601565,
      "grad_norm": 0.4112457036972046,
      "learning_rate": 9.880445238422428e-06,
      "loss": 0.8106,
      "step": 3688
    },
    {
      "epoch": 1.0134596896030765,
      "grad_norm": 0.39122530817985535,
      "learning_rate": 9.877696853098805e-06,
      "loss": 0.9087,
      "step": 3689
    },
    {
      "epoch": 1.0137343771459963,
      "grad_norm": 0.42583325505256653,
      "learning_rate": 9.874948467775182e-06,
      "loss": 0.5268,
      "step": 3690
    },
    {
      "epoch": 1.0140090646889164,
      "grad_norm": 0.4983881711959839,
      "learning_rate": 9.872200082451561e-06,
      "loss": 0.8073,
      "step": 3691
    },
    {
      "epoch": 1.0142837522318362,
      "grad_norm": 0.407642662525177,
      "learning_rate": 9.869451697127938e-06,
      "loss": 0.8327,
      "step": 3692
    },
    {
      "epoch": 1.0145584397747562,
      "grad_norm": 0.5237122178077698,
      "learning_rate": 9.866703311804315e-06,
      "loss": 1.0737,
      "step": 3693
    },
    {
      "epoch": 1.014833127317676,
      "grad_norm": 0.4127925932407379,
      "learning_rate": 9.863954926480694e-06,
      "loss": 0.7378,
      "step": 3694
    },
    {
      "epoch": 1.015107814860596,
      "grad_norm": 0.5114954710006714,
      "learning_rate": 9.861206541157071e-06,
      "loss": 0.5753,
      "step": 3695
    },
    {
      "epoch": 1.015382502403516,
      "grad_norm": 0.45269083976745605,
      "learning_rate": 9.85845815583345e-06,
      "loss": 0.9212,
      "step": 3696
    },
    {
      "epoch": 1.015657189946436,
      "grad_norm": 0.38372519612312317,
      "learning_rate": 9.855709770509827e-06,
      "loss": 0.6607,
      "step": 3697
    },
    {
      "epoch": 1.0159318774893558,
      "grad_norm": 0.4477578103542328,
      "learning_rate": 9.852961385186204e-06,
      "loss": 0.8123,
      "step": 3698
    },
    {
      "epoch": 1.0162065650322758,
      "grad_norm": 0.4897669851779938,
      "learning_rate": 9.850212999862582e-06,
      "loss": 0.7638,
      "step": 3699
    },
    {
      "epoch": 1.0164812525751956,
      "grad_norm": 0.36383435130119324,
      "learning_rate": 9.84746461453896e-06,
      "loss": 0.781,
      "step": 3700
    },
    {
      "epoch": 1.0167559401181157,
      "grad_norm": 0.5327789783477783,
      "learning_rate": 9.844716229215336e-06,
      "loss": 0.6215,
      "step": 3701
    },
    {
      "epoch": 1.0170306276610355,
      "grad_norm": 0.3720916509628296,
      "learning_rate": 9.841967843891715e-06,
      "loss": 0.8422,
      "step": 3702
    },
    {
      "epoch": 1.0173053152039555,
      "grad_norm": 0.39736562967300415,
      "learning_rate": 9.839219458568092e-06,
      "loss": 0.7697,
      "step": 3703
    },
    {
      "epoch": 1.0175800027468753,
      "grad_norm": 0.3466986119747162,
      "learning_rate": 9.83647107324447e-06,
      "loss": 0.8368,
      "step": 3704
    },
    {
      "epoch": 1.0178546902897954,
      "grad_norm": 0.4343137741088867,
      "learning_rate": 9.833722687920846e-06,
      "loss": 0.8107,
      "step": 3705
    },
    {
      "epoch": 1.0181293778327154,
      "grad_norm": 0.5024486780166626,
      "learning_rate": 9.830974302597225e-06,
      "loss": 0.9314,
      "step": 3706
    },
    {
      "epoch": 1.0184040653756352,
      "grad_norm": 0.3919193148612976,
      "learning_rate": 9.828225917273604e-06,
      "loss": 0.754,
      "step": 3707
    },
    {
      "epoch": 1.0186787529185553,
      "grad_norm": 0.4763863980770111,
      "learning_rate": 9.82547753194998e-06,
      "loss": 0.6322,
      "step": 3708
    },
    {
      "epoch": 1.018953440461475,
      "grad_norm": 0.385661244392395,
      "learning_rate": 9.822729146626358e-06,
      "loss": 0.9654,
      "step": 3709
    },
    {
      "epoch": 1.019228128004395,
      "grad_norm": 0.40983518958091736,
      "learning_rate": 9.819980761302735e-06,
      "loss": 0.8301,
      "step": 3710
    },
    {
      "epoch": 1.019502815547315,
      "grad_norm": 0.42838796973228455,
      "learning_rate": 9.817232375979113e-06,
      "loss": 0.6741,
      "step": 3711
    },
    {
      "epoch": 1.019777503090235,
      "grad_norm": 0.3760108947753906,
      "learning_rate": 9.81448399065549e-06,
      "loss": 0.9891,
      "step": 3712
    },
    {
      "epoch": 1.0200521906331548,
      "grad_norm": 0.3883536756038666,
      "learning_rate": 9.811735605331867e-06,
      "loss": 0.7652,
      "step": 3713
    },
    {
      "epoch": 1.0203268781760748,
      "grad_norm": 0.4803607165813446,
      "learning_rate": 9.808987220008246e-06,
      "loss": 0.8857,
      "step": 3714
    },
    {
      "epoch": 1.0206015657189946,
      "grad_norm": 0.45138663053512573,
      "learning_rate": 9.806238834684623e-06,
      "loss": 0.6473,
      "step": 3715
    },
    {
      "epoch": 1.0208762532619147,
      "grad_norm": 0.4883204996585846,
      "learning_rate": 9.803490449361002e-06,
      "loss": 0.7648,
      "step": 3716
    },
    {
      "epoch": 1.0211509408048345,
      "grad_norm": 0.46915435791015625,
      "learning_rate": 9.800742064037379e-06,
      "loss": 0.7016,
      "step": 3717
    },
    {
      "epoch": 1.0214256283477545,
      "grad_norm": 0.4924217164516449,
      "learning_rate": 9.797993678713756e-06,
      "loss": 0.7957,
      "step": 3718
    },
    {
      "epoch": 1.0217003158906743,
      "grad_norm": 0.37309733033180237,
      "learning_rate": 9.795245293390135e-06,
      "loss": 0.8689,
      "step": 3719
    },
    {
      "epoch": 1.0219750034335944,
      "grad_norm": 0.38333600759506226,
      "learning_rate": 9.792496908066512e-06,
      "loss": 0.682,
      "step": 3720
    },
    {
      "epoch": 1.0222496909765142,
      "grad_norm": 0.38862136006355286,
      "learning_rate": 9.789748522742889e-06,
      "loss": 0.9098,
      "step": 3721
    },
    {
      "epoch": 1.0225243785194342,
      "grad_norm": 0.34912094473838806,
      "learning_rate": 9.787000137419267e-06,
      "loss": 0.7564,
      "step": 3722
    },
    {
      "epoch": 1.022799066062354,
      "grad_norm": 0.3914547562599182,
      "learning_rate": 9.784251752095644e-06,
      "loss": 0.9159,
      "step": 3723
    },
    {
      "epoch": 1.023073753605274,
      "grad_norm": 0.3530879616737366,
      "learning_rate": 9.781503366772023e-06,
      "loss": 0.6182,
      "step": 3724
    },
    {
      "epoch": 1.023348441148194,
      "grad_norm": 0.42260029911994934,
      "learning_rate": 9.7787549814484e-06,
      "loss": 0.9701,
      "step": 3725
    },
    {
      "epoch": 1.023623128691114,
      "grad_norm": 0.3770493268966675,
      "learning_rate": 9.776006596124777e-06,
      "loss": 0.9881,
      "step": 3726
    },
    {
      "epoch": 1.0238978162340338,
      "grad_norm": 0.39916136860847473,
      "learning_rate": 9.773258210801156e-06,
      "loss": 0.5739,
      "step": 3727
    },
    {
      "epoch": 1.0241725037769538,
      "grad_norm": 0.3869968056678772,
      "learning_rate": 9.770509825477533e-06,
      "loss": 0.8233,
      "step": 3728
    },
    {
      "epoch": 1.0244471913198736,
      "grad_norm": 0.49805930256843567,
      "learning_rate": 9.76776144015391e-06,
      "loss": 0.9969,
      "step": 3729
    },
    {
      "epoch": 1.0247218788627936,
      "grad_norm": 0.3335803151130676,
      "learning_rate": 9.765013054830289e-06,
      "loss": 0.9997,
      "step": 3730
    },
    {
      "epoch": 1.0249965664057135,
      "grad_norm": 0.4761928617954254,
      "learning_rate": 9.762264669506666e-06,
      "loss": 0.5185,
      "step": 3731
    },
    {
      "epoch": 1.0252712539486335,
      "grad_norm": 0.49729686975479126,
      "learning_rate": 9.759516284183044e-06,
      "loss": 0.6137,
      "step": 3732
    },
    {
      "epoch": 1.0255459414915533,
      "grad_norm": 0.432540625333786,
      "learning_rate": 9.756767898859421e-06,
      "loss": 0.8978,
      "step": 3733
    },
    {
      "epoch": 1.0258206290344734,
      "grad_norm": 0.41703054308891296,
      "learning_rate": 9.754019513535798e-06,
      "loss": 0.6561,
      "step": 3734
    },
    {
      "epoch": 1.0260953165773932,
      "grad_norm": 0.3910958766937256,
      "learning_rate": 9.751271128212177e-06,
      "loss": 0.8419,
      "step": 3735
    },
    {
      "epoch": 1.0263700041203132,
      "grad_norm": 0.4142461121082306,
      "learning_rate": 9.748522742888554e-06,
      "loss": 0.7193,
      "step": 3736
    },
    {
      "epoch": 1.026644691663233,
      "grad_norm": 0.6277196407318115,
      "learning_rate": 9.745774357564931e-06,
      "loss": 0.6738,
      "step": 3737
    },
    {
      "epoch": 1.026919379206153,
      "grad_norm": 0.3802291452884674,
      "learning_rate": 9.743025972241308e-06,
      "loss": 0.6558,
      "step": 3738
    },
    {
      "epoch": 1.0271940667490729,
      "grad_norm": 0.4131438732147217,
      "learning_rate": 9.740277586917687e-06,
      "loss": 0.8042,
      "step": 3739
    },
    {
      "epoch": 1.027468754291993,
      "grad_norm": 0.40727466344833374,
      "learning_rate": 9.737529201594066e-06,
      "loss": 0.8067,
      "step": 3740
    },
    {
      "epoch": 1.0277434418349127,
      "grad_norm": 0.3359214663505554,
      "learning_rate": 9.734780816270441e-06,
      "loss": 0.562,
      "step": 3741
    },
    {
      "epoch": 1.0280181293778328,
      "grad_norm": 0.4676520526409149,
      "learning_rate": 9.73203243094682e-06,
      "loss": 0.7548,
      "step": 3742
    },
    {
      "epoch": 1.0282928169207526,
      "grad_norm": 0.5624427795410156,
      "learning_rate": 9.729284045623197e-06,
      "loss": 0.8531,
      "step": 3743
    },
    {
      "epoch": 1.0285675044636726,
      "grad_norm": 0.41295960545539856,
      "learning_rate": 9.726535660299575e-06,
      "loss": 0.743,
      "step": 3744
    },
    {
      "epoch": 1.0288421920065924,
      "grad_norm": 0.383123517036438,
      "learning_rate": 9.723787274975952e-06,
      "loss": 0.5847,
      "step": 3745
    },
    {
      "epoch": 1.0291168795495125,
      "grad_norm": 0.5229427814483643,
      "learning_rate": 9.72103888965233e-06,
      "loss": 0.8339,
      "step": 3746
    },
    {
      "epoch": 1.0293915670924323,
      "grad_norm": 0.4228321313858032,
      "learning_rate": 9.718290504328708e-06,
      "loss": 0.7192,
      "step": 3747
    },
    {
      "epoch": 1.0296662546353523,
      "grad_norm": 0.45991310477256775,
      "learning_rate": 9.715542119005085e-06,
      "loss": 0.6939,
      "step": 3748
    },
    {
      "epoch": 1.0299409421782721,
      "grad_norm": 0.45587682723999023,
      "learning_rate": 9.712793733681462e-06,
      "loss": 0.7492,
      "step": 3749
    },
    {
      "epoch": 1.0302156297211922,
      "grad_norm": 0.3838711977005005,
      "learning_rate": 9.710045348357841e-06,
      "loss": 0.8692,
      "step": 3750
    },
    {
      "epoch": 1.030490317264112,
      "grad_norm": 0.5099546909332275,
      "learning_rate": 9.707296963034218e-06,
      "loss": 0.7888,
      "step": 3751
    },
    {
      "epoch": 1.030765004807032,
      "grad_norm": 0.4424188435077667,
      "learning_rate": 9.704548577710597e-06,
      "loss": 0.7841,
      "step": 3752
    },
    {
      "epoch": 1.0310396923499519,
      "grad_norm": 0.45174941420555115,
      "learning_rate": 9.701800192386974e-06,
      "loss": 0.7622,
      "step": 3753
    },
    {
      "epoch": 1.031314379892872,
      "grad_norm": 0.43213436007499695,
      "learning_rate": 9.69905180706335e-06,
      "loss": 0.7299,
      "step": 3754
    },
    {
      "epoch": 1.0315890674357917,
      "grad_norm": 0.4502948522567749,
      "learning_rate": 9.69630342173973e-06,
      "loss": 0.8609,
      "step": 3755
    },
    {
      "epoch": 1.0318637549787117,
      "grad_norm": 0.4987502098083496,
      "learning_rate": 9.693555036416106e-06,
      "loss": 0.8054,
      "step": 3756
    },
    {
      "epoch": 1.0321384425216316,
      "grad_norm": 0.4415982961654663,
      "learning_rate": 9.690806651092483e-06,
      "loss": 0.5806,
      "step": 3757
    },
    {
      "epoch": 1.0324131300645516,
      "grad_norm": 0.3565589487552643,
      "learning_rate": 9.688058265768862e-06,
      "loss": 0.9379,
      "step": 3758
    },
    {
      "epoch": 1.0326878176074714,
      "grad_norm": 0.39268720149993896,
      "learning_rate": 9.68530988044524e-06,
      "loss": 0.9042,
      "step": 3759
    },
    {
      "epoch": 1.0329625051503915,
      "grad_norm": 0.5059788823127747,
      "learning_rate": 9.682561495121618e-06,
      "loss": 0.5853,
      "step": 3760
    },
    {
      "epoch": 1.0332371926933113,
      "grad_norm": 0.3801974952220917,
      "learning_rate": 9.679813109797995e-06,
      "loss": 0.5471,
      "step": 3761
    },
    {
      "epoch": 1.0335118802362313,
      "grad_norm": 0.4039478302001953,
      "learning_rate": 9.677064724474372e-06,
      "loss": 0.5915,
      "step": 3762
    },
    {
      "epoch": 1.0337865677791511,
      "grad_norm": 0.47645336389541626,
      "learning_rate": 9.67431633915075e-06,
      "loss": 0.879,
      "step": 3763
    },
    {
      "epoch": 1.0340612553220712,
      "grad_norm": 0.43104225397109985,
      "learning_rate": 9.671567953827128e-06,
      "loss": 0.6349,
      "step": 3764
    },
    {
      "epoch": 1.034335942864991,
      "grad_norm": 0.6356921195983887,
      "learning_rate": 9.668819568503505e-06,
      "loss": 0.8331,
      "step": 3765
    },
    {
      "epoch": 1.034610630407911,
      "grad_norm": 0.5529979467391968,
      "learning_rate": 9.666071183179883e-06,
      "loss": 0.6731,
      "step": 3766
    },
    {
      "epoch": 1.0348853179508308,
      "grad_norm": 0.5060014724731445,
      "learning_rate": 9.66332279785626e-06,
      "loss": 0.7584,
      "step": 3767
    },
    {
      "epoch": 1.0351600054937509,
      "grad_norm": 0.5256634950637817,
      "learning_rate": 9.660574412532639e-06,
      "loss": 0.9438,
      "step": 3768
    },
    {
      "epoch": 1.0354346930366707,
      "grad_norm": 0.41200733184814453,
      "learning_rate": 9.657826027209014e-06,
      "loss": 0.9027,
      "step": 3769
    },
    {
      "epoch": 1.0357093805795907,
      "grad_norm": 0.6332233548164368,
      "learning_rate": 9.655077641885393e-06,
      "loss": 0.8618,
      "step": 3770
    },
    {
      "epoch": 1.0359840681225105,
      "grad_norm": 0.46623682975769043,
      "learning_rate": 9.65232925656177e-06,
      "loss": 0.7085,
      "step": 3771
    },
    {
      "epoch": 1.0362587556654306,
      "grad_norm": 0.49040457606315613,
      "learning_rate": 9.649580871238149e-06,
      "loss": 0.8325,
      "step": 3772
    },
    {
      "epoch": 1.0365334432083504,
      "grad_norm": 0.3598412573337555,
      "learning_rate": 9.646832485914526e-06,
      "loss": 0.6338,
      "step": 3773
    },
    {
      "epoch": 1.0368081307512704,
      "grad_norm": 0.3790982663631439,
      "learning_rate": 9.644084100590903e-06,
      "loss": 0.7244,
      "step": 3774
    },
    {
      "epoch": 1.0370828182941905,
      "grad_norm": 0.4014667868614197,
      "learning_rate": 9.641335715267282e-06,
      "loss": 1.0023,
      "step": 3775
    },
    {
      "epoch": 1.0373575058371103,
      "grad_norm": 0.41882768273353577,
      "learning_rate": 9.638587329943659e-06,
      "loss": 0.8914,
      "step": 3776
    },
    {
      "epoch": 1.0376321933800303,
      "grad_norm": 0.6040247678756714,
      "learning_rate": 9.635838944620036e-06,
      "loss": 0.8396,
      "step": 3777
    },
    {
      "epoch": 1.0379068809229501,
      "grad_norm": 0.36010733246803284,
      "learning_rate": 9.633090559296414e-06,
      "loss": 0.8862,
      "step": 3778
    },
    {
      "epoch": 1.0381815684658702,
      "grad_norm": 0.46953675150871277,
      "learning_rate": 9.630342173972791e-06,
      "loss": 0.7555,
      "step": 3779
    },
    {
      "epoch": 1.03845625600879,
      "grad_norm": 0.5645854473114014,
      "learning_rate": 9.62759378864917e-06,
      "loss": 0.9575,
      "step": 3780
    },
    {
      "epoch": 1.03873094355171,
      "grad_norm": 0.5214871168136597,
      "learning_rate": 9.624845403325547e-06,
      "loss": 0.7532,
      "step": 3781
    },
    {
      "epoch": 1.0390056310946298,
      "grad_norm": 0.5418582558631897,
      "learning_rate": 9.622097018001924e-06,
      "loss": 0.759,
      "step": 3782
    },
    {
      "epoch": 1.0392803186375499,
      "grad_norm": 0.39873388409614563,
      "learning_rate": 9.619348632678303e-06,
      "loss": 0.7728,
      "step": 3783
    },
    {
      "epoch": 1.0395550061804697,
      "grad_norm": 0.526309609413147,
      "learning_rate": 9.61660024735468e-06,
      "loss": 0.8116,
      "step": 3784
    },
    {
      "epoch": 1.0398296937233897,
      "grad_norm": 0.42385271191596985,
      "learning_rate": 9.613851862031057e-06,
      "loss": 0.5688,
      "step": 3785
    },
    {
      "epoch": 1.0401043812663096,
      "grad_norm": 0.5238037109375,
      "learning_rate": 9.611103476707436e-06,
      "loss": 0.6598,
      "step": 3786
    },
    {
      "epoch": 1.0403790688092296,
      "grad_norm": 0.37199461460113525,
      "learning_rate": 9.608355091383813e-06,
      "loss": 1.0844,
      "step": 3787
    },
    {
      "epoch": 1.0406537563521494,
      "grad_norm": 0.33015504479408264,
      "learning_rate": 9.605606706060191e-06,
      "loss": 1.0297,
      "step": 3788
    },
    {
      "epoch": 1.0409284438950694,
      "grad_norm": 0.4896422028541565,
      "learning_rate": 9.602858320736568e-06,
      "loss": 0.7455,
      "step": 3789
    },
    {
      "epoch": 1.0412031314379893,
      "grad_norm": 0.4750913977622986,
      "learning_rate": 9.600109935412945e-06,
      "loss": 0.6207,
      "step": 3790
    },
    {
      "epoch": 1.0414778189809093,
      "grad_norm": 0.482089638710022,
      "learning_rate": 9.597361550089324e-06,
      "loss": 0.9225,
      "step": 3791
    },
    {
      "epoch": 1.0417525065238291,
      "grad_norm": 0.41680315136909485,
      "learning_rate": 9.594613164765701e-06,
      "loss": 0.77,
      "step": 3792
    },
    {
      "epoch": 1.0420271940667492,
      "grad_norm": 0.3964472711086273,
      "learning_rate": 9.591864779442078e-06,
      "loss": 0.5694,
      "step": 3793
    },
    {
      "epoch": 1.042301881609669,
      "grad_norm": 0.5011022090911865,
      "learning_rate": 9.589116394118457e-06,
      "loss": 0.8703,
      "step": 3794
    },
    {
      "epoch": 1.042576569152589,
      "grad_norm": 0.465829074382782,
      "learning_rate": 9.586368008794834e-06,
      "loss": 0.7086,
      "step": 3795
    },
    {
      "epoch": 1.0428512566955088,
      "grad_norm": 0.5819718837738037,
      "learning_rate": 9.583619623471213e-06,
      "loss": 0.8667,
      "step": 3796
    },
    {
      "epoch": 1.0431259442384289,
      "grad_norm": 0.45685943961143494,
      "learning_rate": 9.580871238147588e-06,
      "loss": 0.9353,
      "step": 3797
    },
    {
      "epoch": 1.0434006317813487,
      "grad_norm": 0.5066664218902588,
      "learning_rate": 9.578122852823967e-06,
      "loss": 0.9075,
      "step": 3798
    },
    {
      "epoch": 1.0436753193242687,
      "grad_norm": 0.43311235308647156,
      "learning_rate": 9.575374467500345e-06,
      "loss": 0.939,
      "step": 3799
    },
    {
      "epoch": 1.0439500068671885,
      "grad_norm": 0.46577611565589905,
      "learning_rate": 9.572626082176722e-06,
      "loss": 0.8313,
      "step": 3800
    },
    {
      "epoch": 1.0442246944101086,
      "grad_norm": 0.5620462894439697,
      "learning_rate": 9.5698776968531e-06,
      "loss": 0.7435,
      "step": 3801
    },
    {
      "epoch": 1.0444993819530284,
      "grad_norm": 0.45614010095596313,
      "learning_rate": 9.567129311529476e-06,
      "loss": 0.8377,
      "step": 3802
    },
    {
      "epoch": 1.0447740694959484,
      "grad_norm": 0.4001118838787079,
      "learning_rate": 9.564380926205855e-06,
      "loss": 0.4657,
      "step": 3803
    },
    {
      "epoch": 1.0450487570388682,
      "grad_norm": 0.4199645221233368,
      "learning_rate": 9.561632540882232e-06,
      "loss": 0.8674,
      "step": 3804
    },
    {
      "epoch": 1.0453234445817883,
      "grad_norm": 0.4385525584220886,
      "learning_rate": 9.55888415555861e-06,
      "loss": 1.051,
      "step": 3805
    },
    {
      "epoch": 1.045598132124708,
      "grad_norm": 0.5265312194824219,
      "learning_rate": 9.556135770234988e-06,
      "loss": 0.7342,
      "step": 3806
    },
    {
      "epoch": 1.0458728196676281,
      "grad_norm": 0.5397762060165405,
      "learning_rate": 9.553387384911365e-06,
      "loss": 0.8661,
      "step": 3807
    },
    {
      "epoch": 1.046147507210548,
      "grad_norm": 0.3705483376979828,
      "learning_rate": 9.550638999587742e-06,
      "loss": 0.618,
      "step": 3808
    },
    {
      "epoch": 1.046422194753468,
      "grad_norm": 0.5165073871612549,
      "learning_rate": 9.54789061426412e-06,
      "loss": 0.8134,
      "step": 3809
    },
    {
      "epoch": 1.0466968822963878,
      "grad_norm": 0.49816378951072693,
      "learning_rate": 9.545142228940498e-06,
      "loss": 0.9744,
      "step": 3810
    },
    {
      "epoch": 1.0469715698393078,
      "grad_norm": 0.488060861825943,
      "learning_rate": 9.542393843616876e-06,
      "loss": 0.8881,
      "step": 3811
    },
    {
      "epoch": 1.0472462573822277,
      "grad_norm": 0.3446560800075531,
      "learning_rate": 9.539645458293253e-06,
      "loss": 1.0657,
      "step": 3812
    },
    {
      "epoch": 1.0475209449251477,
      "grad_norm": 0.3348295986652374,
      "learning_rate": 9.53689707296963e-06,
      "loss": 0.6212,
      "step": 3813
    },
    {
      "epoch": 1.0477956324680675,
      "grad_norm": 0.4515681564807892,
      "learning_rate": 9.53414868764601e-06,
      "loss": 0.7462,
      "step": 3814
    },
    {
      "epoch": 1.0480703200109875,
      "grad_norm": 0.4114763140678406,
      "learning_rate": 9.531400302322386e-06,
      "loss": 0.7735,
      "step": 3815
    },
    {
      "epoch": 1.0483450075539074,
      "grad_norm": 0.57194983959198,
      "learning_rate": 9.528651916998763e-06,
      "loss": 0.5583,
      "step": 3816
    },
    {
      "epoch": 1.0486196950968274,
      "grad_norm": 0.5350444316864014,
      "learning_rate": 9.525903531675142e-06,
      "loss": 0.9448,
      "step": 3817
    },
    {
      "epoch": 1.0488943826397472,
      "grad_norm": 0.5222076177597046,
      "learning_rate": 9.523155146351519e-06,
      "loss": 0.6341,
      "step": 3818
    },
    {
      "epoch": 1.0491690701826673,
      "grad_norm": 0.3862020671367645,
      "learning_rate": 9.520406761027898e-06,
      "loss": 0.692,
      "step": 3819
    },
    {
      "epoch": 1.049443757725587,
      "grad_norm": 0.4266696870326996,
      "learning_rate": 9.517658375704275e-06,
      "loss": 0.8279,
      "step": 3820
    },
    {
      "epoch": 1.049718445268507,
      "grad_norm": 0.5575244426727295,
      "learning_rate": 9.514909990380652e-06,
      "loss": 0.7364,
      "step": 3821
    },
    {
      "epoch": 1.049993132811427,
      "grad_norm": 0.4935869872570038,
      "learning_rate": 9.51216160505703e-06,
      "loss": 0.8012,
      "step": 3822
    },
    {
      "epoch": 1.050267820354347,
      "grad_norm": 0.5516129732131958,
      "learning_rate": 9.509413219733407e-06,
      "loss": 0.8292,
      "step": 3823
    },
    {
      "epoch": 1.0505425078972668,
      "grad_norm": 0.4351520538330078,
      "learning_rate": 9.506664834409784e-06,
      "loss": 0.9839,
      "step": 3824
    },
    {
      "epoch": 1.0508171954401868,
      "grad_norm": 0.5171826481819153,
      "learning_rate": 9.503916449086163e-06,
      "loss": 0.7984,
      "step": 3825
    },
    {
      "epoch": 1.0510918829831066,
      "grad_norm": 0.2759772837162018,
      "learning_rate": 9.50116806376254e-06,
      "loss": 0.6038,
      "step": 3826
    },
    {
      "epoch": 1.0513665705260267,
      "grad_norm": 0.34276247024536133,
      "learning_rate": 9.498419678438919e-06,
      "loss": 0.8191,
      "step": 3827
    },
    {
      "epoch": 1.0516412580689465,
      "grad_norm": 0.5324795246124268,
      "learning_rate": 9.495671293115294e-06,
      "loss": 0.8142,
      "step": 3828
    },
    {
      "epoch": 1.0519159456118665,
      "grad_norm": 0.3891792893409729,
      "learning_rate": 9.492922907791673e-06,
      "loss": 0.7342,
      "step": 3829
    },
    {
      "epoch": 1.0521906331547863,
      "grad_norm": 0.45104917883872986,
      "learning_rate": 9.49017452246805e-06,
      "loss": 0.7965,
      "step": 3830
    },
    {
      "epoch": 1.0524653206977064,
      "grad_norm": 0.4331902265548706,
      "learning_rate": 9.487426137144429e-06,
      "loss": 0.7432,
      "step": 3831
    },
    {
      "epoch": 1.0527400082406262,
      "grad_norm": 0.4006405770778656,
      "learning_rate": 9.484677751820806e-06,
      "loss": 0.7642,
      "step": 3832
    },
    {
      "epoch": 1.0530146957835462,
      "grad_norm": 0.42431747913360596,
      "learning_rate": 9.481929366497183e-06,
      "loss": 0.775,
      "step": 3833
    },
    {
      "epoch": 1.053289383326466,
      "grad_norm": 0.40049347281455994,
      "learning_rate": 9.479180981173561e-06,
      "loss": 0.7887,
      "step": 3834
    },
    {
      "epoch": 1.053564070869386,
      "grad_norm": 0.5158541798591614,
      "learning_rate": 9.476432595849938e-06,
      "loss": 0.7694,
      "step": 3835
    },
    {
      "epoch": 1.053838758412306,
      "grad_norm": 0.5547837615013123,
      "learning_rate": 9.473684210526315e-06,
      "loss": 0.7375,
      "step": 3836
    },
    {
      "epoch": 1.054113445955226,
      "grad_norm": 0.3836875557899475,
      "learning_rate": 9.470935825202694e-06,
      "loss": 0.7631,
      "step": 3837
    },
    {
      "epoch": 1.0543881334981458,
      "grad_norm": 0.41924071311950684,
      "learning_rate": 9.468187439879071e-06,
      "loss": 0.646,
      "step": 3838
    },
    {
      "epoch": 1.0546628210410658,
      "grad_norm": 0.4701293408870697,
      "learning_rate": 9.46543905455545e-06,
      "loss": 0.7162,
      "step": 3839
    },
    {
      "epoch": 1.0549375085839858,
      "grad_norm": 0.49539491534233093,
      "learning_rate": 9.462690669231827e-06,
      "loss": 0.8767,
      "step": 3840
    },
    {
      "epoch": 1.0552121961269056,
      "grad_norm": 0.350653737783432,
      "learning_rate": 9.459942283908204e-06,
      "loss": 0.7823,
      "step": 3841
    },
    {
      "epoch": 1.0554868836698255,
      "grad_norm": 0.4119814336299896,
      "learning_rate": 9.457193898584583e-06,
      "loss": 0.97,
      "step": 3842
    },
    {
      "epoch": 1.0557615712127455,
      "grad_norm": 0.5355620980262756,
      "learning_rate": 9.45444551326096e-06,
      "loss": 0.7987,
      "step": 3843
    },
    {
      "epoch": 1.0560362587556655,
      "grad_norm": 0.33243682980537415,
      "learning_rate": 9.451697127937337e-06,
      "loss": 0.562,
      "step": 3844
    },
    {
      "epoch": 1.0563109462985854,
      "grad_norm": 0.30657869577407837,
      "learning_rate": 9.448948742613715e-06,
      "loss": 0.5573,
      "step": 3845
    },
    {
      "epoch": 1.0565856338415054,
      "grad_norm": 0.4759327471256256,
      "learning_rate": 9.446200357290092e-06,
      "loss": 0.7925,
      "step": 3846
    },
    {
      "epoch": 1.0568603213844252,
      "grad_norm": 0.29805395007133484,
      "learning_rate": 9.443451971966471e-06,
      "loss": 0.7945,
      "step": 3847
    },
    {
      "epoch": 1.0571350089273452,
      "grad_norm": 0.394307017326355,
      "learning_rate": 9.440703586642848e-06,
      "loss": 0.7969,
      "step": 3848
    },
    {
      "epoch": 1.057409696470265,
      "grad_norm": 0.6042557954788208,
      "learning_rate": 9.437955201319225e-06,
      "loss": 0.7591,
      "step": 3849
    },
    {
      "epoch": 1.057684384013185,
      "grad_norm": 0.40051236748695374,
      "learning_rate": 9.435206815995604e-06,
      "loss": 1.0288,
      "step": 3850
    },
    {
      "epoch": 1.057959071556105,
      "grad_norm": 0.35650497674942017,
      "learning_rate": 9.432458430671981e-06,
      "loss": 0.6922,
      "step": 3851
    },
    {
      "epoch": 1.058233759099025,
      "grad_norm": 0.38561326265335083,
      "learning_rate": 9.429710045348358e-06,
      "loss": 0.6311,
      "step": 3852
    },
    {
      "epoch": 1.0585084466419448,
      "grad_norm": 0.3680665194988251,
      "learning_rate": 9.426961660024737e-06,
      "loss": 0.7029,
      "step": 3853
    },
    {
      "epoch": 1.0587831341848648,
      "grad_norm": 0.4408940076828003,
      "learning_rate": 9.424213274701114e-06,
      "loss": 1.0307,
      "step": 3854
    },
    {
      "epoch": 1.0590578217277846,
      "grad_norm": 0.3844451606273651,
      "learning_rate": 9.421464889377492e-06,
      "loss": 0.7636,
      "step": 3855
    },
    {
      "epoch": 1.0593325092707047,
      "grad_norm": 0.40213048458099365,
      "learning_rate": 9.418716504053868e-06,
      "loss": 0.6909,
      "step": 3856
    },
    {
      "epoch": 1.0596071968136245,
      "grad_norm": 0.3583410084247589,
      "learning_rate": 9.415968118730246e-06,
      "loss": 0.6442,
      "step": 3857
    },
    {
      "epoch": 1.0598818843565445,
      "grad_norm": 0.43081435561180115,
      "learning_rate": 9.413219733406625e-06,
      "loss": 0.7806,
      "step": 3858
    },
    {
      "epoch": 1.0601565718994643,
      "grad_norm": 0.40399858355522156,
      "learning_rate": 9.410471348083002e-06,
      "loss": 0.8849,
      "step": 3859
    },
    {
      "epoch": 1.0604312594423844,
      "grad_norm": 0.4172636568546295,
      "learning_rate": 9.40772296275938e-06,
      "loss": 0.8349,
      "step": 3860
    },
    {
      "epoch": 1.0607059469853042,
      "grad_norm": 0.5793783664703369,
      "learning_rate": 9.404974577435756e-06,
      "loss": 0.8408,
      "step": 3861
    },
    {
      "epoch": 1.0609806345282242,
      "grad_norm": 0.528430163860321,
      "learning_rate": 9.402226192112135e-06,
      "loss": 0.9924,
      "step": 3862
    },
    {
      "epoch": 1.061255322071144,
      "grad_norm": 0.3953479528427124,
      "learning_rate": 9.399477806788514e-06,
      "loss": 1.114,
      "step": 3863
    },
    {
      "epoch": 1.061530009614064,
      "grad_norm": 0.39338478446006775,
      "learning_rate": 9.396729421464889e-06,
      "loss": 1.0847,
      "step": 3864
    },
    {
      "epoch": 1.061804697156984,
      "grad_norm": 0.4123649001121521,
      "learning_rate": 9.393981036141268e-06,
      "loss": 0.7293,
      "step": 3865
    },
    {
      "epoch": 1.062079384699904,
      "grad_norm": 0.3572738468647003,
      "learning_rate": 9.391232650817645e-06,
      "loss": 0.582,
      "step": 3866
    },
    {
      "epoch": 1.0623540722428237,
      "grad_norm": 0.37108123302459717,
      "learning_rate": 9.388484265494023e-06,
      "loss": 0.9125,
      "step": 3867
    },
    {
      "epoch": 1.0626287597857438,
      "grad_norm": 0.5442267656326294,
      "learning_rate": 9.3857358801704e-06,
      "loss": 0.7475,
      "step": 3868
    },
    {
      "epoch": 1.0629034473286636,
      "grad_norm": 0.563513994216919,
      "learning_rate": 9.382987494846778e-06,
      "loss": 0.8259,
      "step": 3869
    },
    {
      "epoch": 1.0631781348715836,
      "grad_norm": 0.4019607603549957,
      "learning_rate": 9.380239109523156e-06,
      "loss": 0.805,
      "step": 3870
    },
    {
      "epoch": 1.0634528224145035,
      "grad_norm": 0.6065915822982788,
      "learning_rate": 9.377490724199533e-06,
      "loss": 0.7312,
      "step": 3871
    },
    {
      "epoch": 1.0637275099574235,
      "grad_norm": 0.4089483320713043,
      "learning_rate": 9.37474233887591e-06,
      "loss": 0.8025,
      "step": 3872
    },
    {
      "epoch": 1.0640021975003433,
      "grad_norm": 0.5689536333084106,
      "learning_rate": 9.371993953552289e-06,
      "loss": 0.6904,
      "step": 3873
    },
    {
      "epoch": 1.0642768850432633,
      "grad_norm": 0.46814361214637756,
      "learning_rate": 9.369245568228666e-06,
      "loss": 0.5405,
      "step": 3874
    },
    {
      "epoch": 1.0645515725861832,
      "grad_norm": 0.5040010213851929,
      "learning_rate": 9.366497182905045e-06,
      "loss": 0.8606,
      "step": 3875
    },
    {
      "epoch": 1.0648262601291032,
      "grad_norm": 0.44161659479141235,
      "learning_rate": 9.363748797581422e-06,
      "loss": 0.702,
      "step": 3876
    },
    {
      "epoch": 1.065100947672023,
      "grad_norm": 0.36072301864624023,
      "learning_rate": 9.361000412257799e-06,
      "loss": 0.9315,
      "step": 3877
    },
    {
      "epoch": 1.065375635214943,
      "grad_norm": 0.3731571435928345,
      "learning_rate": 9.358252026934177e-06,
      "loss": 0.751,
      "step": 3878
    },
    {
      "epoch": 1.0656503227578629,
      "grad_norm": 0.4086410105228424,
      "learning_rate": 9.355503641610554e-06,
      "loss": 0.9323,
      "step": 3879
    },
    {
      "epoch": 1.065925010300783,
      "grad_norm": 0.47410017251968384,
      "learning_rate": 9.352755256286932e-06,
      "loss": 0.7525,
      "step": 3880
    },
    {
      "epoch": 1.0661996978437027,
      "grad_norm": 0.43121299147605896,
      "learning_rate": 9.35000687096331e-06,
      "loss": 0.639,
      "step": 3881
    },
    {
      "epoch": 1.0664743853866228,
      "grad_norm": 0.4133700430393219,
      "learning_rate": 9.347258485639687e-06,
      "loss": 0.8491,
      "step": 3882
    },
    {
      "epoch": 1.0667490729295426,
      "grad_norm": 0.44997724890708923,
      "learning_rate": 9.344510100316066e-06,
      "loss": 0.9032,
      "step": 3883
    },
    {
      "epoch": 1.0670237604724626,
      "grad_norm": 0.5262262225151062,
      "learning_rate": 9.341761714992443e-06,
      "loss": 0.6525,
      "step": 3884
    },
    {
      "epoch": 1.0672984480153824,
      "grad_norm": 0.4147252142429352,
      "learning_rate": 9.33901332966882e-06,
      "loss": 0.6937,
      "step": 3885
    },
    {
      "epoch": 1.0675731355583025,
      "grad_norm": 0.45425182580947876,
      "learning_rate": 9.336264944345199e-06,
      "loss": 0.9112,
      "step": 3886
    },
    {
      "epoch": 1.0678478231012223,
      "grad_norm": 0.3866460919380188,
      "learning_rate": 9.333516559021576e-06,
      "loss": 0.805,
      "step": 3887
    },
    {
      "epoch": 1.0681225106441423,
      "grad_norm": 0.5174725651741028,
      "learning_rate": 9.330768173697953e-06,
      "loss": 0.8508,
      "step": 3888
    },
    {
      "epoch": 1.0683971981870621,
      "grad_norm": 0.42662087082862854,
      "learning_rate": 9.328019788374331e-06,
      "loss": 0.695,
      "step": 3889
    },
    {
      "epoch": 1.0686718857299822,
      "grad_norm": 0.6277583241462708,
      "learning_rate": 9.325271403050708e-06,
      "loss": 0.8522,
      "step": 3890
    },
    {
      "epoch": 1.068946573272902,
      "grad_norm": 0.38994720578193665,
      "learning_rate": 9.322523017727087e-06,
      "loss": 0.7245,
      "step": 3891
    },
    {
      "epoch": 1.069221260815822,
      "grad_norm": 0.32074546813964844,
      "learning_rate": 9.319774632403463e-06,
      "loss": 0.8756,
      "step": 3892
    },
    {
      "epoch": 1.0694959483587418,
      "grad_norm": 0.5628466606140137,
      "learning_rate": 9.317026247079841e-06,
      "loss": 0.6205,
      "step": 3893
    },
    {
      "epoch": 1.0697706359016619,
      "grad_norm": 0.35141006112098694,
      "learning_rate": 9.314277861756218e-06,
      "loss": 1.0834,
      "step": 3894
    },
    {
      "epoch": 1.0700453234445817,
      "grad_norm": 0.5165079236030579,
      "learning_rate": 9.311529476432597e-06,
      "loss": 0.7107,
      "step": 3895
    },
    {
      "epoch": 1.0703200109875017,
      "grad_norm": 0.4212515354156494,
      "learning_rate": 9.308781091108974e-06,
      "loss": 0.5756,
      "step": 3896
    },
    {
      "epoch": 1.0705946985304216,
      "grad_norm": 0.5738928318023682,
      "learning_rate": 9.306032705785351e-06,
      "loss": 0.7892,
      "step": 3897
    },
    {
      "epoch": 1.0708693860733416,
      "grad_norm": 0.5221078395843506,
      "learning_rate": 9.30328432046173e-06,
      "loss": 0.69,
      "step": 3898
    },
    {
      "epoch": 1.0711440736162614,
      "grad_norm": 0.46513721346855164,
      "learning_rate": 9.300535935138107e-06,
      "loss": 0.7624,
      "step": 3899
    },
    {
      "epoch": 1.0714187611591814,
      "grad_norm": 0.35684844851493835,
      "learning_rate": 9.297787549814484e-06,
      "loss": 0.6057,
      "step": 3900
    },
    {
      "epoch": 1.0716934487021013,
      "grad_norm": 0.48240524530410767,
      "learning_rate": 9.295039164490863e-06,
      "loss": 0.7076,
      "step": 3901
    },
    {
      "epoch": 1.0719681362450213,
      "grad_norm": 0.4640994369983673,
      "learning_rate": 9.29229077916724e-06,
      "loss": 0.8603,
      "step": 3902
    },
    {
      "epoch": 1.0722428237879411,
      "grad_norm": 0.41045865416526794,
      "learning_rate": 9.289542393843618e-06,
      "loss": 0.7216,
      "step": 3903
    },
    {
      "epoch": 1.0725175113308612,
      "grad_norm": 0.5481833815574646,
      "learning_rate": 9.286794008519995e-06,
      "loss": 0.8988,
      "step": 3904
    },
    {
      "epoch": 1.0727921988737812,
      "grad_norm": 0.4722384810447693,
      "learning_rate": 9.284045623196372e-06,
      "loss": 0.8402,
      "step": 3905
    },
    {
      "epoch": 1.073066886416701,
      "grad_norm": 0.3828066885471344,
      "learning_rate": 9.281297237872751e-06,
      "loss": 0.791,
      "step": 3906
    },
    {
      "epoch": 1.0733415739596208,
      "grad_norm": 0.7280402183532715,
      "learning_rate": 9.278548852549128e-06,
      "loss": 0.7181,
      "step": 3907
    },
    {
      "epoch": 1.0736162615025409,
      "grad_norm": 0.5290054678916931,
      "learning_rate": 9.275800467225505e-06,
      "loss": 0.7159,
      "step": 3908
    },
    {
      "epoch": 1.073890949045461,
      "grad_norm": 0.5486999154090881,
      "learning_rate": 9.273052081901884e-06,
      "loss": 0.7788,
      "step": 3909
    },
    {
      "epoch": 1.0741656365883807,
      "grad_norm": 0.45973169803619385,
      "learning_rate": 9.27030369657826e-06,
      "loss": 0.6906,
      "step": 3910
    },
    {
      "epoch": 1.0744403241313005,
      "grad_norm": 0.38840559124946594,
      "learning_rate": 9.26755531125464e-06,
      "loss": 0.8999,
      "step": 3911
    },
    {
      "epoch": 1.0747150116742206,
      "grad_norm": 0.4187402129173279,
      "learning_rate": 9.264806925931017e-06,
      "loss": 0.8195,
      "step": 3912
    },
    {
      "epoch": 1.0749896992171406,
      "grad_norm": 0.5123761296272278,
      "learning_rate": 9.262058540607394e-06,
      "loss": 0.7282,
      "step": 3913
    },
    {
      "epoch": 1.0752643867600604,
      "grad_norm": 0.3415219187736511,
      "learning_rate": 9.259310155283772e-06,
      "loss": 1.0011,
      "step": 3914
    },
    {
      "epoch": 1.0755390743029805,
      "grad_norm": 0.5059516429901123,
      "learning_rate": 9.25656176996015e-06,
      "loss": 0.6768,
      "step": 3915
    },
    {
      "epoch": 1.0758137618459003,
      "grad_norm": 0.478687584400177,
      "learning_rate": 9.253813384636526e-06,
      "loss": 0.5582,
      "step": 3916
    },
    {
      "epoch": 1.0760884493888203,
      "grad_norm": 0.4861970543861389,
      "learning_rate": 9.251064999312905e-06,
      "loss": 1.0376,
      "step": 3917
    },
    {
      "epoch": 1.0763631369317401,
      "grad_norm": 0.40869832038879395,
      "learning_rate": 9.248316613989282e-06,
      "loss": 0.7447,
      "step": 3918
    },
    {
      "epoch": 1.0766378244746602,
      "grad_norm": 0.6007975935935974,
      "learning_rate": 9.24556822866566e-06,
      "loss": 0.8275,
      "step": 3919
    },
    {
      "epoch": 1.07691251201758,
      "grad_norm": 0.5144460201263428,
      "learning_rate": 9.242819843342036e-06,
      "loss": 0.8566,
      "step": 3920
    },
    {
      "epoch": 1.0771871995605,
      "grad_norm": 0.4891836643218994,
      "learning_rate": 9.240071458018415e-06,
      "loss": 0.7505,
      "step": 3921
    },
    {
      "epoch": 1.0774618871034198,
      "grad_norm": 0.41579577326774597,
      "learning_rate": 9.237323072694793e-06,
      "loss": 0.8866,
      "step": 3922
    },
    {
      "epoch": 1.0777365746463399,
      "grad_norm": 0.42554184794425964,
      "learning_rate": 9.23457468737117e-06,
      "loss": 0.5281,
      "step": 3923
    },
    {
      "epoch": 1.0780112621892597,
      "grad_norm": 0.33732444047927856,
      "learning_rate": 9.231826302047548e-06,
      "loss": 0.8962,
      "step": 3924
    },
    {
      "epoch": 1.0782859497321797,
      "grad_norm": 0.5208308100700378,
      "learning_rate": 9.229077916723925e-06,
      "loss": 0.7343,
      "step": 3925
    },
    {
      "epoch": 1.0785606372750995,
      "grad_norm": 0.3728395402431488,
      "learning_rate": 9.226329531400303e-06,
      "loss": 0.5165,
      "step": 3926
    },
    {
      "epoch": 1.0788353248180196,
      "grad_norm": 0.33870676159858704,
      "learning_rate": 9.22358114607668e-06,
      "loss": 0.7409,
      "step": 3927
    },
    {
      "epoch": 1.0791100123609394,
      "grad_norm": 0.43530020117759705,
      "learning_rate": 9.220832760753057e-06,
      "loss": 0.6567,
      "step": 3928
    },
    {
      "epoch": 1.0793846999038594,
      "grad_norm": 0.4128011465072632,
      "learning_rate": 9.218084375429436e-06,
      "loss": 0.5357,
      "step": 3929
    },
    {
      "epoch": 1.0796593874467793,
      "grad_norm": 0.4924473166465759,
      "learning_rate": 9.215335990105813e-06,
      "loss": 0.9022,
      "step": 3930
    },
    {
      "epoch": 1.0799340749896993,
      "grad_norm": 0.5248915553092957,
      "learning_rate": 9.212587604782192e-06,
      "loss": 0.6942,
      "step": 3931
    },
    {
      "epoch": 1.080208762532619,
      "grad_norm": 0.4647020995616913,
      "learning_rate": 9.209839219458569e-06,
      "loss": 0.8844,
      "step": 3932
    },
    {
      "epoch": 1.0804834500755391,
      "grad_norm": 0.4515751898288727,
      "learning_rate": 9.207090834134946e-06,
      "loss": 0.9258,
      "step": 3933
    },
    {
      "epoch": 1.080758137618459,
      "grad_norm": 0.4110581576824188,
      "learning_rate": 9.204342448811325e-06,
      "loss": 1.1311,
      "step": 3934
    },
    {
      "epoch": 1.081032825161379,
      "grad_norm": 0.3482224643230438,
      "learning_rate": 9.201594063487702e-06,
      "loss": 1.0041,
      "step": 3935
    },
    {
      "epoch": 1.0813075127042988,
      "grad_norm": 0.4788133502006531,
      "learning_rate": 9.198845678164079e-06,
      "loss": 0.5994,
      "step": 3936
    },
    {
      "epoch": 1.0815822002472189,
      "grad_norm": 0.49346816539764404,
      "learning_rate": 9.196097292840457e-06,
      "loss": 0.6308,
      "step": 3937
    },
    {
      "epoch": 1.0818568877901387,
      "grad_norm": 0.38373371958732605,
      "learning_rate": 9.193348907516834e-06,
      "loss": 0.6206,
      "step": 3938
    },
    {
      "epoch": 1.0821315753330587,
      "grad_norm": 0.4328956604003906,
      "learning_rate": 9.190600522193213e-06,
      "loss": 0.6333,
      "step": 3939
    },
    {
      "epoch": 1.0824062628759785,
      "grad_norm": 0.5228323936462402,
      "learning_rate": 9.18785213686959e-06,
      "loss": 0.6287,
      "step": 3940
    },
    {
      "epoch": 1.0826809504188986,
      "grad_norm": 0.4025191068649292,
      "learning_rate": 9.185103751545967e-06,
      "loss": 0.7913,
      "step": 3941
    },
    {
      "epoch": 1.0829556379618184,
      "grad_norm": 0.5484251976013184,
      "learning_rate": 9.182355366222346e-06,
      "loss": 0.6598,
      "step": 3942
    },
    {
      "epoch": 1.0832303255047384,
      "grad_norm": 0.3304372727870941,
      "learning_rate": 9.179606980898723e-06,
      "loss": 0.9943,
      "step": 3943
    },
    {
      "epoch": 1.0835050130476582,
      "grad_norm": 0.4944940507411957,
      "learning_rate": 9.1768585955751e-06,
      "loss": 0.7407,
      "step": 3944
    },
    {
      "epoch": 1.0837797005905783,
      "grad_norm": 0.3881138265132904,
      "learning_rate": 9.174110210251479e-06,
      "loss": 0.9202,
      "step": 3945
    },
    {
      "epoch": 1.084054388133498,
      "grad_norm": 0.4028731882572174,
      "learning_rate": 9.171361824927856e-06,
      "loss": 0.626,
      "step": 3946
    },
    {
      "epoch": 1.0843290756764181,
      "grad_norm": 0.31609317660331726,
      "learning_rate": 9.168613439604234e-06,
      "loss": 0.6204,
      "step": 3947
    },
    {
      "epoch": 1.084603763219338,
      "grad_norm": 0.4085569977760315,
      "learning_rate": 9.165865054280611e-06,
      "loss": 0.8737,
      "step": 3948
    },
    {
      "epoch": 1.084878450762258,
      "grad_norm": 0.5174620151519775,
      "learning_rate": 9.163116668956988e-06,
      "loss": 0.6727,
      "step": 3949
    },
    {
      "epoch": 1.0851531383051778,
      "grad_norm": 0.44680505990982056,
      "learning_rate": 9.160368283633367e-06,
      "loss": 0.9652,
      "step": 3950
    },
    {
      "epoch": 1.0854278258480978,
      "grad_norm": 0.4425711929798126,
      "learning_rate": 9.157619898309744e-06,
      "loss": 1.0604,
      "step": 3951
    },
    {
      "epoch": 1.0857025133910176,
      "grad_norm": 0.5342880487442017,
      "learning_rate": 9.154871512986121e-06,
      "loss": 0.8378,
      "step": 3952
    },
    {
      "epoch": 1.0859772009339377,
      "grad_norm": 0.5431009531021118,
      "learning_rate": 9.152123127662498e-06,
      "loss": 0.8178,
      "step": 3953
    },
    {
      "epoch": 1.0862518884768575,
      "grad_norm": 0.5073680281639099,
      "learning_rate": 9.149374742338877e-06,
      "loss": 0.711,
      "step": 3954
    },
    {
      "epoch": 1.0865265760197775,
      "grad_norm": 0.5825316309928894,
      "learning_rate": 9.146626357015256e-06,
      "loss": 0.852,
      "step": 3955
    },
    {
      "epoch": 1.0868012635626974,
      "grad_norm": 0.3022661507129669,
      "learning_rate": 9.14387797169163e-06,
      "loss": 0.5468,
      "step": 3956
    },
    {
      "epoch": 1.0870759511056174,
      "grad_norm": 0.40758973360061646,
      "learning_rate": 9.14112958636801e-06,
      "loss": 0.6429,
      "step": 3957
    },
    {
      "epoch": 1.0873506386485372,
      "grad_norm": 0.4258359670639038,
      "learning_rate": 9.138381201044387e-06,
      "loss": 0.7644,
      "step": 3958
    },
    {
      "epoch": 1.0876253261914572,
      "grad_norm": 0.45637282729148865,
      "learning_rate": 9.135632815720765e-06,
      "loss": 0.5368,
      "step": 3959
    },
    {
      "epoch": 1.087900013734377,
      "grad_norm": 0.2899613380432129,
      "learning_rate": 9.132884430397142e-06,
      "loss": 0.6024,
      "step": 3960
    },
    {
      "epoch": 1.088174701277297,
      "grad_norm": 0.4144079089164734,
      "learning_rate": 9.13013604507352e-06,
      "loss": 0.7707,
      "step": 3961
    },
    {
      "epoch": 1.088449388820217,
      "grad_norm": 0.3918832838535309,
      "learning_rate": 9.127387659749898e-06,
      "loss": 1.023,
      "step": 3962
    },
    {
      "epoch": 1.088724076363137,
      "grad_norm": 0.3436899483203888,
      "learning_rate": 9.124639274426275e-06,
      "loss": 0.4596,
      "step": 3963
    },
    {
      "epoch": 1.0889987639060568,
      "grad_norm": 0.411487340927124,
      "learning_rate": 9.121890889102652e-06,
      "loss": 0.6324,
      "step": 3964
    },
    {
      "epoch": 1.0892734514489768,
      "grad_norm": 0.4711022675037384,
      "learning_rate": 9.11914250377903e-06,
      "loss": 0.7569,
      "step": 3965
    },
    {
      "epoch": 1.0895481389918966,
      "grad_norm": 0.4481630027294159,
      "learning_rate": 9.116394118455408e-06,
      "loss": 0.8065,
      "step": 3966
    },
    {
      "epoch": 1.0898228265348167,
      "grad_norm": 0.5379236936569214,
      "learning_rate": 9.113645733131787e-06,
      "loss": 0.7386,
      "step": 3967
    },
    {
      "epoch": 1.0900975140777365,
      "grad_norm": 0.4471421241760254,
      "learning_rate": 9.110897347808164e-06,
      "loss": 0.8875,
      "step": 3968
    },
    {
      "epoch": 1.0903722016206565,
      "grad_norm": 0.37672746181488037,
      "learning_rate": 9.10814896248454e-06,
      "loss": 0.5676,
      "step": 3969
    },
    {
      "epoch": 1.0906468891635765,
      "grad_norm": 0.41942650079727173,
      "learning_rate": 9.10540057716092e-06,
      "loss": 0.7298,
      "step": 3970
    },
    {
      "epoch": 1.0909215767064964,
      "grad_norm": 0.4688628315925598,
      "learning_rate": 9.102652191837296e-06,
      "loss": 0.5535,
      "step": 3971
    },
    {
      "epoch": 1.0911962642494162,
      "grad_norm": 0.4375166893005371,
      "learning_rate": 9.099903806513673e-06,
      "loss": 0.8557,
      "step": 3972
    },
    {
      "epoch": 1.0914709517923362,
      "grad_norm": 0.35658594965934753,
      "learning_rate": 9.097155421190052e-06,
      "loss": 0.7711,
      "step": 3973
    },
    {
      "epoch": 1.0917456393352563,
      "grad_norm": 0.4641742706298828,
      "learning_rate": 9.094407035866429e-06,
      "loss": 0.8989,
      "step": 3974
    },
    {
      "epoch": 1.092020326878176,
      "grad_norm": 0.48032066226005554,
      "learning_rate": 9.091658650542808e-06,
      "loss": 0.7413,
      "step": 3975
    },
    {
      "epoch": 1.0922950144210959,
      "grad_norm": 0.6156169772148132,
      "learning_rate": 9.088910265219185e-06,
      "loss": 0.9069,
      "step": 3976
    },
    {
      "epoch": 1.092569701964016,
      "grad_norm": 0.5904116630554199,
      "learning_rate": 9.086161879895562e-06,
      "loss": 0.6518,
      "step": 3977
    },
    {
      "epoch": 1.092844389506936,
      "grad_norm": 0.4362366497516632,
      "learning_rate": 9.08341349457194e-06,
      "loss": 1.022,
      "step": 3978
    },
    {
      "epoch": 1.0931190770498558,
      "grad_norm": 0.5405244827270508,
      "learning_rate": 9.080665109248318e-06,
      "loss": 0.741,
      "step": 3979
    },
    {
      "epoch": 1.0933937645927758,
      "grad_norm": 0.6547083258628845,
      "learning_rate": 9.077916723924695e-06,
      "loss": 0.8547,
      "step": 3980
    },
    {
      "epoch": 1.0936684521356956,
      "grad_norm": 0.5257118344306946,
      "learning_rate": 9.075168338601073e-06,
      "loss": 0.8792,
      "step": 3981
    },
    {
      "epoch": 1.0939431396786157,
      "grad_norm": 0.42347249388694763,
      "learning_rate": 9.07241995327745e-06,
      "loss": 0.7248,
      "step": 3982
    },
    {
      "epoch": 1.0942178272215355,
      "grad_norm": 0.43988117575645447,
      "learning_rate": 9.069671567953829e-06,
      "loss": 0.6663,
      "step": 3983
    },
    {
      "epoch": 1.0944925147644555,
      "grad_norm": 0.505987823009491,
      "learning_rate": 9.066923182630204e-06,
      "loss": 0.7074,
      "step": 3984
    },
    {
      "epoch": 1.0947672023073753,
      "grad_norm": 0.4593750536441803,
      "learning_rate": 9.064174797306583e-06,
      "loss": 0.9039,
      "step": 3985
    },
    {
      "epoch": 1.0950418898502954,
      "grad_norm": 0.36029139161109924,
      "learning_rate": 9.06142641198296e-06,
      "loss": 0.6207,
      "step": 3986
    },
    {
      "epoch": 1.0953165773932152,
      "grad_norm": 0.4307313859462738,
      "learning_rate": 9.058678026659339e-06,
      "loss": 0.9112,
      "step": 3987
    },
    {
      "epoch": 1.0955912649361352,
      "grad_norm": 0.497462660074234,
      "learning_rate": 9.055929641335716e-06,
      "loss": 1.061,
      "step": 3988
    },
    {
      "epoch": 1.095865952479055,
      "grad_norm": 0.3974326252937317,
      "learning_rate": 9.053181256012093e-06,
      "loss": 0.8492,
      "step": 3989
    },
    {
      "epoch": 1.096140640021975,
      "grad_norm": 0.4576719403266907,
      "learning_rate": 9.050432870688472e-06,
      "loss": 0.7462,
      "step": 3990
    },
    {
      "epoch": 1.096415327564895,
      "grad_norm": 0.4604439437389374,
      "learning_rate": 9.047684485364849e-06,
      "loss": 0.8482,
      "step": 3991
    },
    {
      "epoch": 1.096690015107815,
      "grad_norm": 0.4701593518257141,
      "learning_rate": 9.044936100041226e-06,
      "loss": 0.865,
      "step": 3992
    },
    {
      "epoch": 1.0969647026507348,
      "grad_norm": 0.5293518900871277,
      "learning_rate": 9.042187714717604e-06,
      "loss": 0.8782,
      "step": 3993
    },
    {
      "epoch": 1.0972393901936548,
      "grad_norm": 0.4645938575267792,
      "learning_rate": 9.039439329393981e-06,
      "loss": 0.8152,
      "step": 3994
    },
    {
      "epoch": 1.0975140777365746,
      "grad_norm": 0.37127894163131714,
      "learning_rate": 9.03669094407036e-06,
      "loss": 0.7501,
      "step": 3995
    },
    {
      "epoch": 1.0977887652794946,
      "grad_norm": 0.4184896945953369,
      "learning_rate": 9.033942558746737e-06,
      "loss": 0.9207,
      "step": 3996
    },
    {
      "epoch": 1.0980634528224145,
      "grad_norm": 0.2673567533493042,
      "learning_rate": 9.031194173423114e-06,
      "loss": 0.5114,
      "step": 3997
    },
    {
      "epoch": 1.0983381403653345,
      "grad_norm": 0.24638080596923828,
      "learning_rate": 9.028445788099493e-06,
      "loss": 0.5601,
      "step": 3998
    },
    {
      "epoch": 1.0986128279082543,
      "grad_norm": 0.41832271218299866,
      "learning_rate": 9.02569740277587e-06,
      "loss": 0.7619,
      "step": 3999
    },
    {
      "epoch": 1.0988875154511744,
      "grad_norm": 0.4098522663116455,
      "learning_rate": 9.022949017452247e-06,
      "loss": 0.5919,
      "step": 4000
    },
    {
      "epoch": 1.0991622029940942,
      "grad_norm": 0.44471025466918945,
      "learning_rate": 9.020200632128626e-06,
      "loss": 0.6567,
      "step": 4001
    },
    {
      "epoch": 1.0994368905370142,
      "grad_norm": 0.47432610392570496,
      "learning_rate": 9.017452246805003e-06,
      "loss": 0.6229,
      "step": 4002
    },
    {
      "epoch": 1.099711578079934,
      "grad_norm": 0.44337669014930725,
      "learning_rate": 9.014703861481381e-06,
      "loss": 0.7036,
      "step": 4003
    },
    {
      "epoch": 1.099986265622854,
      "grad_norm": 0.43888646364212036,
      "learning_rate": 9.011955476157758e-06,
      "loss": 0.6283,
      "step": 4004
    },
    {
      "epoch": 1.1002609531657739,
      "grad_norm": 0.33118999004364014,
      "learning_rate": 9.009207090834135e-06,
      "loss": 0.5924,
      "step": 4005
    },
    {
      "epoch": 1.100535640708694,
      "grad_norm": 0.47267478704452515,
      "learning_rate": 9.006458705510514e-06,
      "loss": 0.9016,
      "step": 4006
    },
    {
      "epoch": 1.1008103282516137,
      "grad_norm": 0.4156203866004944,
      "learning_rate": 9.003710320186891e-06,
      "loss": 0.8268,
      "step": 4007
    },
    {
      "epoch": 1.1010850157945338,
      "grad_norm": 0.5013849139213562,
      "learning_rate": 9.000961934863268e-06,
      "loss": 0.7348,
      "step": 4008
    },
    {
      "epoch": 1.1013597033374536,
      "grad_norm": 0.5900996327400208,
      "learning_rate": 8.998213549539647e-06,
      "loss": 0.7781,
      "step": 4009
    },
    {
      "epoch": 1.1016343908803736,
      "grad_norm": 0.5411255359649658,
      "learning_rate": 8.995465164216024e-06,
      "loss": 0.7771,
      "step": 4010
    },
    {
      "epoch": 1.1019090784232934,
      "grad_norm": 0.44909462332725525,
      "learning_rate": 8.992716778892403e-06,
      "loss": 0.67,
      "step": 4011
    },
    {
      "epoch": 1.1021837659662135,
      "grad_norm": 0.5196014046669006,
      "learning_rate": 8.989968393568778e-06,
      "loss": 0.632,
      "step": 4012
    },
    {
      "epoch": 1.1024584535091333,
      "grad_norm": 0.446262001991272,
      "learning_rate": 8.987220008245157e-06,
      "loss": 0.9137,
      "step": 4013
    },
    {
      "epoch": 1.1027331410520533,
      "grad_norm": 0.41301730275154114,
      "learning_rate": 8.984471622921535e-06,
      "loss": 0.9952,
      "step": 4014
    },
    {
      "epoch": 1.1030078285949731,
      "grad_norm": 0.44374382495880127,
      "learning_rate": 8.981723237597912e-06,
      "loss": 0.8125,
      "step": 4015
    },
    {
      "epoch": 1.1032825161378932,
      "grad_norm": 0.5140025019645691,
      "learning_rate": 8.97897485227429e-06,
      "loss": 0.8304,
      "step": 4016
    },
    {
      "epoch": 1.103557203680813,
      "grad_norm": 0.3812014162540436,
      "learning_rate": 8.976226466950666e-06,
      "loss": 0.7576,
      "step": 4017
    },
    {
      "epoch": 1.103831891223733,
      "grad_norm": 0.46502217650413513,
      "learning_rate": 8.973478081627045e-06,
      "loss": 0.7153,
      "step": 4018
    },
    {
      "epoch": 1.1041065787666529,
      "grad_norm": 0.4364437460899353,
      "learning_rate": 8.970729696303422e-06,
      "loss": 0.5459,
      "step": 4019
    },
    {
      "epoch": 1.104381266309573,
      "grad_norm": 0.5787715315818787,
      "learning_rate": 8.967981310979799e-06,
      "loss": 0.6887,
      "step": 4020
    },
    {
      "epoch": 1.1046559538524927,
      "grad_norm": 0.3823210895061493,
      "learning_rate": 8.965232925656178e-06,
      "loss": 0.9234,
      "step": 4021
    },
    {
      "epoch": 1.1049306413954127,
      "grad_norm": 0.46403393149375916,
      "learning_rate": 8.962484540332555e-06,
      "loss": 0.7591,
      "step": 4022
    },
    {
      "epoch": 1.1052053289383326,
      "grad_norm": 0.44278034567832947,
      "learning_rate": 8.959736155008934e-06,
      "loss": 0.8075,
      "step": 4023
    },
    {
      "epoch": 1.1054800164812526,
      "grad_norm": 0.4740288555622101,
      "learning_rate": 8.95698776968531e-06,
      "loss": 0.7313,
      "step": 4024
    },
    {
      "epoch": 1.1057547040241724,
      "grad_norm": 0.38568297028541565,
      "learning_rate": 8.954239384361688e-06,
      "loss": 0.7337,
      "step": 4025
    },
    {
      "epoch": 1.1060293915670925,
      "grad_norm": 0.46776971220970154,
      "learning_rate": 8.951490999038066e-06,
      "loss": 0.5179,
      "step": 4026
    },
    {
      "epoch": 1.1063040791100123,
      "grad_norm": 0.37515562772750854,
      "learning_rate": 8.948742613714443e-06,
      "loss": 0.5733,
      "step": 4027
    },
    {
      "epoch": 1.1065787666529323,
      "grad_norm": 0.5345409512519836,
      "learning_rate": 8.94599422839082e-06,
      "loss": 0.7139,
      "step": 4028
    },
    {
      "epoch": 1.1068534541958521,
      "grad_norm": 0.32215937972068787,
      "learning_rate": 8.943245843067199e-06,
      "loss": 0.8772,
      "step": 4029
    },
    {
      "epoch": 1.1071281417387722,
      "grad_norm": 0.4501116871833801,
      "learning_rate": 8.940497457743576e-06,
      "loss": 0.497,
      "step": 4030
    },
    {
      "epoch": 1.107402829281692,
      "grad_norm": 0.53240966796875,
      "learning_rate": 8.937749072419955e-06,
      "loss": 0.5957,
      "step": 4031
    },
    {
      "epoch": 1.107677516824612,
      "grad_norm": 0.5287309885025024,
      "learning_rate": 8.935000687096332e-06,
      "loss": 0.7986,
      "step": 4032
    },
    {
      "epoch": 1.1079522043675318,
      "grad_norm": 0.5374777317047119,
      "learning_rate": 8.932252301772709e-06,
      "loss": 0.7815,
      "step": 4033
    },
    {
      "epoch": 1.1082268919104519,
      "grad_norm": 0.49788913130760193,
      "learning_rate": 8.929503916449088e-06,
      "loss": 0.7475,
      "step": 4034
    },
    {
      "epoch": 1.1085015794533717,
      "grad_norm": 0.43924012780189514,
      "learning_rate": 8.926755531125465e-06,
      "loss": 0.961,
      "step": 4035
    },
    {
      "epoch": 1.1087762669962917,
      "grad_norm": 0.5222353339195251,
      "learning_rate": 8.924007145801842e-06,
      "loss": 0.7984,
      "step": 4036
    },
    {
      "epoch": 1.1090509545392115,
      "grad_norm": 0.4737566411495209,
      "learning_rate": 8.92125876047822e-06,
      "loss": 0.7174,
      "step": 4037
    },
    {
      "epoch": 1.1093256420821316,
      "grad_norm": 0.48419710993766785,
      "learning_rate": 8.918510375154597e-06,
      "loss": 0.6479,
      "step": 4038
    },
    {
      "epoch": 1.1096003296250516,
      "grad_norm": 0.3896210193634033,
      "learning_rate": 8.915761989830976e-06,
      "loss": 0.6118,
      "step": 4039
    },
    {
      "epoch": 1.1098750171679714,
      "grad_norm": 0.47781315445899963,
      "learning_rate": 8.913013604507353e-06,
      "loss": 0.9114,
      "step": 4040
    },
    {
      "epoch": 1.1101497047108912,
      "grad_norm": 0.4294223189353943,
      "learning_rate": 8.91026521918373e-06,
      "loss": 0.6886,
      "step": 4041
    },
    {
      "epoch": 1.1104243922538113,
      "grad_norm": 0.4344167411327362,
      "learning_rate": 8.907516833860109e-06,
      "loss": 0.702,
      "step": 4042
    },
    {
      "epoch": 1.1106990797967313,
      "grad_norm": 0.37454596161842346,
      "learning_rate": 8.904768448536486e-06,
      "loss": 0.7412,
      "step": 4043
    },
    {
      "epoch": 1.1109737673396511,
      "grad_norm": 0.45513293147087097,
      "learning_rate": 8.902020063212863e-06,
      "loss": 0.5764,
      "step": 4044
    },
    {
      "epoch": 1.1112484548825712,
      "grad_norm": 0.5619507431983948,
      "learning_rate": 8.89927167788924e-06,
      "loss": 0.7645,
      "step": 4045
    },
    {
      "epoch": 1.111523142425491,
      "grad_norm": 0.5967808961868286,
      "learning_rate": 8.896523292565619e-06,
      "loss": 0.671,
      "step": 4046
    },
    {
      "epoch": 1.111797829968411,
      "grad_norm": 0.31456294655799866,
      "learning_rate": 8.893774907241997e-06,
      "loss": 0.7383,
      "step": 4047
    },
    {
      "epoch": 1.1120725175113308,
      "grad_norm": 0.41461730003356934,
      "learning_rate": 8.891026521918373e-06,
      "loss": 0.8728,
      "step": 4048
    },
    {
      "epoch": 1.1123472050542509,
      "grad_norm": 0.49544647336006165,
      "learning_rate": 8.888278136594751e-06,
      "loss": 0.689,
      "step": 4049
    },
    {
      "epoch": 1.1126218925971707,
      "grad_norm": 0.5439872741699219,
      "learning_rate": 8.885529751271128e-06,
      "loss": 0.7924,
      "step": 4050
    },
    {
      "epoch": 1.1128965801400907,
      "grad_norm": 0.4226680397987366,
      "learning_rate": 8.882781365947507e-06,
      "loss": 0.7183,
      "step": 4051
    },
    {
      "epoch": 1.1131712676830106,
      "grad_norm": 0.38719990849494934,
      "learning_rate": 8.880032980623884e-06,
      "loss": 0.7559,
      "step": 4052
    },
    {
      "epoch": 1.1134459552259306,
      "grad_norm": 0.44871455430984497,
      "learning_rate": 8.877284595300261e-06,
      "loss": 0.7214,
      "step": 4053
    },
    {
      "epoch": 1.1137206427688504,
      "grad_norm": 0.35783684253692627,
      "learning_rate": 8.87453620997664e-06,
      "loss": 0.5321,
      "step": 4054
    },
    {
      "epoch": 1.1139953303117704,
      "grad_norm": 0.5230251550674438,
      "learning_rate": 8.871787824653017e-06,
      "loss": 0.7656,
      "step": 4055
    },
    {
      "epoch": 1.1142700178546903,
      "grad_norm": 0.4236011505126953,
      "learning_rate": 8.869039439329394e-06,
      "loss": 0.8291,
      "step": 4056
    },
    {
      "epoch": 1.1145447053976103,
      "grad_norm": 0.3731664717197418,
      "learning_rate": 8.866291054005773e-06,
      "loss": 0.861,
      "step": 4057
    },
    {
      "epoch": 1.1148193929405301,
      "grad_norm": 0.5977442860603333,
      "learning_rate": 8.86354266868215e-06,
      "loss": 0.8405,
      "step": 4058
    },
    {
      "epoch": 1.1150940804834502,
      "grad_norm": 0.4762936234474182,
      "learning_rate": 8.860794283358528e-06,
      "loss": 0.8188,
      "step": 4059
    },
    {
      "epoch": 1.11536876802637,
      "grad_norm": 0.43810322880744934,
      "learning_rate": 8.858045898034905e-06,
      "loss": 0.6461,
      "step": 4060
    },
    {
      "epoch": 1.11564345556929,
      "grad_norm": 0.44681572914123535,
      "learning_rate": 8.855297512711282e-06,
      "loss": 0.7009,
      "step": 4061
    },
    {
      "epoch": 1.1159181431122098,
      "grad_norm": 0.48020172119140625,
      "learning_rate": 8.852549127387661e-06,
      "loss": 0.899,
      "step": 4062
    },
    {
      "epoch": 1.1161928306551299,
      "grad_norm": 0.47030481696128845,
      "learning_rate": 8.849800742064038e-06,
      "loss": 0.6839,
      "step": 4063
    },
    {
      "epoch": 1.1164675181980497,
      "grad_norm": 0.4137589633464813,
      "learning_rate": 8.847052356740415e-06,
      "loss": 0.5506,
      "step": 4064
    },
    {
      "epoch": 1.1167422057409697,
      "grad_norm": 0.5272451639175415,
      "learning_rate": 8.844303971416794e-06,
      "loss": 0.7124,
      "step": 4065
    },
    {
      "epoch": 1.1170168932838895,
      "grad_norm": 0.3467077314853668,
      "learning_rate": 8.841555586093171e-06,
      "loss": 0.605,
      "step": 4066
    },
    {
      "epoch": 1.1172915808268096,
      "grad_norm": 0.46630144119262695,
      "learning_rate": 8.83880720076955e-06,
      "loss": 0.7847,
      "step": 4067
    },
    {
      "epoch": 1.1175662683697294,
      "grad_norm": 0.3375801146030426,
      "learning_rate": 8.836058815445927e-06,
      "loss": 0.9232,
      "step": 4068
    },
    {
      "epoch": 1.1178409559126494,
      "grad_norm": 0.5996243357658386,
      "learning_rate": 8.833310430122304e-06,
      "loss": 0.7673,
      "step": 4069
    },
    {
      "epoch": 1.1181156434555692,
      "grad_norm": 0.5322815775871277,
      "learning_rate": 8.830562044798682e-06,
      "loss": 0.7699,
      "step": 4070
    },
    {
      "epoch": 1.1183903309984893,
      "grad_norm": 0.4210908114910126,
      "learning_rate": 8.82781365947506e-06,
      "loss": 0.7708,
      "step": 4071
    },
    {
      "epoch": 1.118665018541409,
      "grad_norm": 0.3564922511577606,
      "learning_rate": 8.825065274151436e-06,
      "loss": 0.7185,
      "step": 4072
    },
    {
      "epoch": 1.1189397060843291,
      "grad_norm": 0.5791550278663635,
      "learning_rate": 8.822316888827815e-06,
      "loss": 0.8486,
      "step": 4073
    },
    {
      "epoch": 1.119214393627249,
      "grad_norm": 0.5051422715187073,
      "learning_rate": 8.819568503504192e-06,
      "loss": 0.8282,
      "step": 4074
    },
    {
      "epoch": 1.119489081170169,
      "grad_norm": 0.3493426442146301,
      "learning_rate": 8.81682011818057e-06,
      "loss": 0.6533,
      "step": 4075
    },
    {
      "epoch": 1.1197637687130888,
      "grad_norm": 0.4652740955352783,
      "learning_rate": 8.814071732856946e-06,
      "loss": 0.8394,
      "step": 4076
    },
    {
      "epoch": 1.1200384562560088,
      "grad_norm": 0.3672747015953064,
      "learning_rate": 8.811323347533325e-06,
      "loss": 0.9991,
      "step": 4077
    },
    {
      "epoch": 1.1203131437989287,
      "grad_norm": 0.5134392976760864,
      "learning_rate": 8.808574962209704e-06,
      "loss": 0.7955,
      "step": 4078
    },
    {
      "epoch": 1.1205878313418487,
      "grad_norm": 0.466653048992157,
      "learning_rate": 8.80582657688608e-06,
      "loss": 0.9857,
      "step": 4079
    },
    {
      "epoch": 1.1208625188847685,
      "grad_norm": 0.42358270287513733,
      "learning_rate": 8.803078191562458e-06,
      "loss": 0.9855,
      "step": 4080
    },
    {
      "epoch": 1.1211372064276885,
      "grad_norm": 0.4358750283718109,
      "learning_rate": 8.800329806238835e-06,
      "loss": 0.6829,
      "step": 4081
    },
    {
      "epoch": 1.1214118939706084,
      "grad_norm": 0.4293590784072876,
      "learning_rate": 8.797581420915213e-06,
      "loss": 0.9065,
      "step": 4082
    },
    {
      "epoch": 1.1216865815135284,
      "grad_norm": 0.47900253534317017,
      "learning_rate": 8.79483303559159e-06,
      "loss": 0.863,
      "step": 4083
    },
    {
      "epoch": 1.1219612690564482,
      "grad_norm": 0.46837642788887024,
      "learning_rate": 8.792084650267967e-06,
      "loss": 0.8318,
      "step": 4084
    },
    {
      "epoch": 1.1222359565993683,
      "grad_norm": 0.3425785303115845,
      "learning_rate": 8.789336264944346e-06,
      "loss": 0.7295,
      "step": 4085
    },
    {
      "epoch": 1.122510644142288,
      "grad_norm": 0.4713115990161896,
      "learning_rate": 8.786587879620723e-06,
      "loss": 0.7698,
      "step": 4086
    },
    {
      "epoch": 1.122785331685208,
      "grad_norm": 0.3788217008113861,
      "learning_rate": 8.783839494297102e-06,
      "loss": 0.9393,
      "step": 4087
    },
    {
      "epoch": 1.123060019228128,
      "grad_norm": 0.4206651449203491,
      "learning_rate": 8.781091108973479e-06,
      "loss": 0.7851,
      "step": 4088
    },
    {
      "epoch": 1.123334706771048,
      "grad_norm": 0.37786027789115906,
      "learning_rate": 8.778342723649856e-06,
      "loss": 0.7258,
      "step": 4089
    },
    {
      "epoch": 1.1236093943139678,
      "grad_norm": 0.4185289442539215,
      "learning_rate": 8.775594338326235e-06,
      "loss": 0.6505,
      "step": 4090
    },
    {
      "epoch": 1.1238840818568878,
      "grad_norm": 0.4316391050815582,
      "learning_rate": 8.772845953002612e-06,
      "loss": 1.0369,
      "step": 4091
    },
    {
      "epoch": 1.1241587693998076,
      "grad_norm": 0.4293273389339447,
      "learning_rate": 8.770097567678989e-06,
      "loss": 0.6416,
      "step": 4092
    },
    {
      "epoch": 1.1244334569427277,
      "grad_norm": 0.4762743413448334,
      "learning_rate": 8.767349182355367e-06,
      "loss": 0.9984,
      "step": 4093
    },
    {
      "epoch": 1.1247081444856475,
      "grad_norm": 0.3469218313694,
      "learning_rate": 8.764600797031744e-06,
      "loss": 0.5942,
      "step": 4094
    },
    {
      "epoch": 1.1249828320285675,
      "grad_norm": 0.5656812787055969,
      "learning_rate": 8.761852411708123e-06,
      "loss": 0.6527,
      "step": 4095
    },
    {
      "epoch": 1.1252575195714873,
      "grad_norm": 0.48361095786094666,
      "learning_rate": 8.7591040263845e-06,
      "loss": 0.5411,
      "step": 4096
    },
    {
      "epoch": 1.1255322071144074,
      "grad_norm": 0.43314671516418457,
      "learning_rate": 8.756355641060877e-06,
      "loss": 0.7989,
      "step": 4097
    },
    {
      "epoch": 1.1258068946573272,
      "grad_norm": 0.4274429678916931,
      "learning_rate": 8.753607255737256e-06,
      "loss": 0.9082,
      "step": 4098
    },
    {
      "epoch": 1.1260815822002472,
      "grad_norm": 0.5110147595405579,
      "learning_rate": 8.750858870413633e-06,
      "loss": 0.6961,
      "step": 4099
    },
    {
      "epoch": 1.1263562697431673,
      "grad_norm": 0.5823311805725098,
      "learning_rate": 8.74811048509001e-06,
      "loss": 0.6128,
      "step": 4100
    },
    {
      "epoch": 1.126630957286087,
      "grad_norm": 0.3589326739311218,
      "learning_rate": 8.745362099766389e-06,
      "loss": 0.9144,
      "step": 4101
    },
    {
      "epoch": 1.126905644829007,
      "grad_norm": 0.4325469136238098,
      "learning_rate": 8.742613714442766e-06,
      "loss": 0.7892,
      "step": 4102
    },
    {
      "epoch": 1.127180332371927,
      "grad_norm": 0.33765995502471924,
      "learning_rate": 8.739865329119144e-06,
      "loss": 0.4827,
      "step": 4103
    },
    {
      "epoch": 1.127455019914847,
      "grad_norm": 0.4293377697467804,
      "learning_rate": 8.737116943795521e-06,
      "loss": 0.7985,
      "step": 4104
    },
    {
      "epoch": 1.1277297074577668,
      "grad_norm": 0.7667198777198792,
      "learning_rate": 8.734368558471898e-06,
      "loss": 0.8288,
      "step": 4105
    },
    {
      "epoch": 1.1280043950006866,
      "grad_norm": 0.40668758749961853,
      "learning_rate": 8.731620173148277e-06,
      "loss": 0.7298,
      "step": 4106
    },
    {
      "epoch": 1.1282790825436066,
      "grad_norm": 0.369150847196579,
      "learning_rate": 8.728871787824654e-06,
      "loss": 0.6213,
      "step": 4107
    },
    {
      "epoch": 1.1285537700865267,
      "grad_norm": 0.3973424732685089,
      "learning_rate": 8.726123402501031e-06,
      "loss": 1.022,
      "step": 4108
    },
    {
      "epoch": 1.1288284576294465,
      "grad_norm": 0.5955417156219482,
      "learning_rate": 8.723375017177408e-06,
      "loss": 0.8848,
      "step": 4109
    },
    {
      "epoch": 1.1291031451723663,
      "grad_norm": 0.5159307718276978,
      "learning_rate": 8.720626631853787e-06,
      "loss": 0.8326,
      "step": 4110
    },
    {
      "epoch": 1.1293778327152864,
      "grad_norm": 0.5654878616333008,
      "learning_rate": 8.717878246530166e-06,
      "loss": 0.8089,
      "step": 4111
    },
    {
      "epoch": 1.1296525202582064,
      "grad_norm": 0.5845798850059509,
      "learning_rate": 8.715129861206541e-06,
      "loss": 0.7923,
      "step": 4112
    },
    {
      "epoch": 1.1299272078011262,
      "grad_norm": 0.4587545692920685,
      "learning_rate": 8.71238147588292e-06,
      "loss": 0.7087,
      "step": 4113
    },
    {
      "epoch": 1.130201895344046,
      "grad_norm": 0.37858638167381287,
      "learning_rate": 8.709633090559297e-06,
      "loss": 0.8192,
      "step": 4114
    },
    {
      "epoch": 1.130476582886966,
      "grad_norm": 0.4894179403781891,
      "learning_rate": 8.706884705235675e-06,
      "loss": 0.9622,
      "step": 4115
    },
    {
      "epoch": 1.130751270429886,
      "grad_norm": 0.590070903301239,
      "learning_rate": 8.704136319912052e-06,
      "loss": 0.7014,
      "step": 4116
    },
    {
      "epoch": 1.131025957972806,
      "grad_norm": 0.3722393214702606,
      "learning_rate": 8.70138793458843e-06,
      "loss": 0.5622,
      "step": 4117
    },
    {
      "epoch": 1.131300645515726,
      "grad_norm": 0.6144607663154602,
      "learning_rate": 8.698639549264808e-06,
      "loss": 0.8439,
      "step": 4118
    },
    {
      "epoch": 1.1315753330586458,
      "grad_norm": 0.500579297542572,
      "learning_rate": 8.695891163941185e-06,
      "loss": 0.771,
      "step": 4119
    },
    {
      "epoch": 1.1318500206015658,
      "grad_norm": 0.44815099239349365,
      "learning_rate": 8.693142778617562e-06,
      "loss": 0.9765,
      "step": 4120
    },
    {
      "epoch": 1.1321247081444856,
      "grad_norm": 0.4063316881656647,
      "learning_rate": 8.690394393293941e-06,
      "loss": 0.6789,
      "step": 4121
    },
    {
      "epoch": 1.1323993956874057,
      "grad_norm": 0.495960533618927,
      "learning_rate": 8.687646007970318e-06,
      "loss": 0.7519,
      "step": 4122
    },
    {
      "epoch": 1.1326740832303255,
      "grad_norm": 0.4690568745136261,
      "learning_rate": 8.684897622646695e-06,
      "loss": 0.9975,
      "step": 4123
    },
    {
      "epoch": 1.1329487707732455,
      "grad_norm": 0.5898585915565491,
      "learning_rate": 8.682149237323074e-06,
      "loss": 0.8013,
      "step": 4124
    },
    {
      "epoch": 1.1332234583161653,
      "grad_norm": 0.31086671352386475,
      "learning_rate": 8.67940085199945e-06,
      "loss": 0.7552,
      "step": 4125
    },
    {
      "epoch": 1.1334981458590854,
      "grad_norm": 0.5109190940856934,
      "learning_rate": 8.67665246667583e-06,
      "loss": 0.7939,
      "step": 4126
    },
    {
      "epoch": 1.1337728334020052,
      "grad_norm": 0.5964676141738892,
      "learning_rate": 8.673904081352206e-06,
      "loss": 0.9854,
      "step": 4127
    },
    {
      "epoch": 1.1340475209449252,
      "grad_norm": 0.5946398973464966,
      "learning_rate": 8.671155696028583e-06,
      "loss": 0.8267,
      "step": 4128
    },
    {
      "epoch": 1.134322208487845,
      "grad_norm": 0.36500731110572815,
      "learning_rate": 8.668407310704962e-06,
      "loss": 0.4898,
      "step": 4129
    },
    {
      "epoch": 1.134596896030765,
      "grad_norm": 0.4332418739795685,
      "learning_rate": 8.66565892538134e-06,
      "loss": 0.7174,
      "step": 4130
    },
    {
      "epoch": 1.134871583573685,
      "grad_norm": 0.4319138824939728,
      "learning_rate": 8.662910540057716e-06,
      "loss": 0.6438,
      "step": 4131
    },
    {
      "epoch": 1.135146271116605,
      "grad_norm": 0.6818703413009644,
      "learning_rate": 8.660162154734095e-06,
      "loss": 0.742,
      "step": 4132
    },
    {
      "epoch": 1.1354209586595247,
      "grad_norm": 0.4537076950073242,
      "learning_rate": 8.657413769410472e-06,
      "loss": 0.8702,
      "step": 4133
    },
    {
      "epoch": 1.1356956462024448,
      "grad_norm": 0.490191251039505,
      "learning_rate": 8.65466538408685e-06,
      "loss": 0.6131,
      "step": 4134
    },
    {
      "epoch": 1.1359703337453646,
      "grad_norm": 0.4037730395793915,
      "learning_rate": 8.651916998763226e-06,
      "loss": 0.8436,
      "step": 4135
    },
    {
      "epoch": 1.1362450212882846,
      "grad_norm": 0.4030740559101105,
      "learning_rate": 8.649168613439605e-06,
      "loss": 0.8534,
      "step": 4136
    },
    {
      "epoch": 1.1365197088312045,
      "grad_norm": 0.3850644528865814,
      "learning_rate": 8.646420228115983e-06,
      "loss": 0.6938,
      "step": 4137
    },
    {
      "epoch": 1.1367943963741245,
      "grad_norm": 0.31368333101272583,
      "learning_rate": 8.64367184279236e-06,
      "loss": 0.8463,
      "step": 4138
    },
    {
      "epoch": 1.1370690839170443,
      "grad_norm": 0.4207240641117096,
      "learning_rate": 8.640923457468737e-06,
      "loss": 0.6783,
      "step": 4139
    },
    {
      "epoch": 1.1373437714599643,
      "grad_norm": 0.4655756950378418,
      "learning_rate": 8.638175072145114e-06,
      "loss": 0.8901,
      "step": 4140
    },
    {
      "epoch": 1.1376184590028842,
      "grad_norm": 0.4481129050254822,
      "learning_rate": 8.635426686821493e-06,
      "loss": 0.9329,
      "step": 4141
    },
    {
      "epoch": 1.1378931465458042,
      "grad_norm": 0.44569942355155945,
      "learning_rate": 8.63267830149787e-06,
      "loss": 0.7526,
      "step": 4142
    },
    {
      "epoch": 1.138167834088724,
      "grad_norm": 0.5258368849754333,
      "learning_rate": 8.629929916174247e-06,
      "loss": 0.9294,
      "step": 4143
    },
    {
      "epoch": 1.138442521631644,
      "grad_norm": 0.39695966243743896,
      "learning_rate": 8.627181530850626e-06,
      "loss": 0.7061,
      "step": 4144
    },
    {
      "epoch": 1.1387172091745639,
      "grad_norm": 0.3839893639087677,
      "learning_rate": 8.624433145527003e-06,
      "loss": 0.8983,
      "step": 4145
    },
    {
      "epoch": 1.138991896717484,
      "grad_norm": 0.5174344778060913,
      "learning_rate": 8.621684760203382e-06,
      "loss": 0.8656,
      "step": 4146
    },
    {
      "epoch": 1.1392665842604037,
      "grad_norm": 0.47410884499549866,
      "learning_rate": 8.618936374879759e-06,
      "loss": 0.7954,
      "step": 4147
    },
    {
      "epoch": 1.1395412718033238,
      "grad_norm": 0.438782274723053,
      "learning_rate": 8.616187989556136e-06,
      "loss": 0.9252,
      "step": 4148
    },
    {
      "epoch": 1.1398159593462436,
      "grad_norm": 0.3836534321308136,
      "learning_rate": 8.613439604232514e-06,
      "loss": 0.6789,
      "step": 4149
    },
    {
      "epoch": 1.1400906468891636,
      "grad_norm": 0.364042729139328,
      "learning_rate": 8.610691218908891e-06,
      "loss": 1.0456,
      "step": 4150
    },
    {
      "epoch": 1.1403653344320834,
      "grad_norm": 0.6514015197753906,
      "learning_rate": 8.607942833585268e-06,
      "loss": 0.7138,
      "step": 4151
    },
    {
      "epoch": 1.1406400219750035,
      "grad_norm": 0.534455418586731,
      "learning_rate": 8.605194448261647e-06,
      "loss": 0.6645,
      "step": 4152
    },
    {
      "epoch": 1.1409147095179233,
      "grad_norm": 0.5718715190887451,
      "learning_rate": 8.602446062938024e-06,
      "loss": 0.906,
      "step": 4153
    },
    {
      "epoch": 1.1411893970608433,
      "grad_norm": 0.46590906381607056,
      "learning_rate": 8.599697677614403e-06,
      "loss": 0.4736,
      "step": 4154
    },
    {
      "epoch": 1.1414640846037631,
      "grad_norm": 0.4736765921115875,
      "learning_rate": 8.59694929229078e-06,
      "loss": 0.7177,
      "step": 4155
    },
    {
      "epoch": 1.1417387721466832,
      "grad_norm": 0.4049414396286011,
      "learning_rate": 8.594200906967157e-06,
      "loss": 0.8211,
      "step": 4156
    },
    {
      "epoch": 1.142013459689603,
      "grad_norm": 0.5716152191162109,
      "learning_rate": 8.591452521643536e-06,
      "loss": 0.7355,
      "step": 4157
    },
    {
      "epoch": 1.142288147232523,
      "grad_norm": 0.3840794563293457,
      "learning_rate": 8.588704136319913e-06,
      "loss": 0.7859,
      "step": 4158
    },
    {
      "epoch": 1.1425628347754428,
      "grad_norm": 0.5418487787246704,
      "learning_rate": 8.58595575099629e-06,
      "loss": 0.7905,
      "step": 4159
    },
    {
      "epoch": 1.1428375223183629,
      "grad_norm": 0.5029850602149963,
      "learning_rate": 8.583207365672668e-06,
      "loss": 0.8353,
      "step": 4160
    },
    {
      "epoch": 1.1431122098612827,
      "grad_norm": 0.5120168924331665,
      "learning_rate": 8.580458980349045e-06,
      "loss": 0.8821,
      "step": 4161
    },
    {
      "epoch": 1.1433868974042027,
      "grad_norm": 0.5306359529495239,
      "learning_rate": 8.577710595025424e-06,
      "loss": 0.7621,
      "step": 4162
    },
    {
      "epoch": 1.1436615849471226,
      "grad_norm": 0.4419049024581909,
      "learning_rate": 8.574962209701801e-06,
      "loss": 0.5961,
      "step": 4163
    },
    {
      "epoch": 1.1439362724900426,
      "grad_norm": 0.5014894008636475,
      "learning_rate": 8.572213824378178e-06,
      "loss": 0.8282,
      "step": 4164
    },
    {
      "epoch": 1.1442109600329624,
      "grad_norm": 0.44452500343322754,
      "learning_rate": 8.569465439054557e-06,
      "loss": 0.5735,
      "step": 4165
    },
    {
      "epoch": 1.1444856475758824,
      "grad_norm": 0.509362518787384,
      "learning_rate": 8.566717053730934e-06,
      "loss": 0.6961,
      "step": 4166
    },
    {
      "epoch": 1.1447603351188023,
      "grad_norm": 0.4679470956325531,
      "learning_rate": 8.563968668407311e-06,
      "loss": 0.9625,
      "step": 4167
    },
    {
      "epoch": 1.1450350226617223,
      "grad_norm": 0.521384596824646,
      "learning_rate": 8.561220283083688e-06,
      "loss": 0.6917,
      "step": 4168
    },
    {
      "epoch": 1.1453097102046423,
      "grad_norm": 0.39251771569252014,
      "learning_rate": 8.558471897760067e-06,
      "loss": 0.7401,
      "step": 4169
    },
    {
      "epoch": 1.1455843977475622,
      "grad_norm": 0.35048457980155945,
      "learning_rate": 8.555723512436445e-06,
      "loss": 0.7707,
      "step": 4170
    },
    {
      "epoch": 1.145859085290482,
      "grad_norm": 0.4542633295059204,
      "learning_rate": 8.55297512711282e-06,
      "loss": 0.6612,
      "step": 4171
    },
    {
      "epoch": 1.146133772833402,
      "grad_norm": 0.423666775226593,
      "learning_rate": 8.5502267417892e-06,
      "loss": 0.81,
      "step": 4172
    },
    {
      "epoch": 1.146408460376322,
      "grad_norm": 0.3712381422519684,
      "learning_rate": 8.547478356465576e-06,
      "loss": 0.6495,
      "step": 4173
    },
    {
      "epoch": 1.1466831479192419,
      "grad_norm": 0.3878575265407562,
      "learning_rate": 8.544729971141955e-06,
      "loss": 0.832,
      "step": 4174
    },
    {
      "epoch": 1.1469578354621617,
      "grad_norm": 0.3261931836605072,
      "learning_rate": 8.541981585818332e-06,
      "loss": 0.7442,
      "step": 4175
    },
    {
      "epoch": 1.1472325230050817,
      "grad_norm": 0.4378376007080078,
      "learning_rate": 8.53923320049471e-06,
      "loss": 0.7596,
      "step": 4176
    },
    {
      "epoch": 1.1475072105480018,
      "grad_norm": 0.39890074729919434,
      "learning_rate": 8.536484815171088e-06,
      "loss": 0.5872,
      "step": 4177
    },
    {
      "epoch": 1.1477818980909216,
      "grad_norm": 0.5436393022537231,
      "learning_rate": 8.533736429847465e-06,
      "loss": 0.6971,
      "step": 4178
    },
    {
      "epoch": 1.1480565856338414,
      "grad_norm": 0.4045860767364502,
      "learning_rate": 8.530988044523842e-06,
      "loss": 0.6356,
      "step": 4179
    },
    {
      "epoch": 1.1483312731767614,
      "grad_norm": 0.47622641921043396,
      "learning_rate": 8.52823965920022e-06,
      "loss": 0.7527,
      "step": 4180
    },
    {
      "epoch": 1.1486059607196815,
      "grad_norm": 0.514976978302002,
      "learning_rate": 8.525491273876598e-06,
      "loss": 0.7445,
      "step": 4181
    },
    {
      "epoch": 1.1488806482626013,
      "grad_norm": 0.4783588647842407,
      "learning_rate": 8.522742888552976e-06,
      "loss": 0.691,
      "step": 4182
    },
    {
      "epoch": 1.1491553358055213,
      "grad_norm": 0.4455702602863312,
      "learning_rate": 8.519994503229353e-06,
      "loss": 0.7782,
      "step": 4183
    },
    {
      "epoch": 1.1494300233484411,
      "grad_norm": 0.40012040734291077,
      "learning_rate": 8.51724611790573e-06,
      "loss": 0.8385,
      "step": 4184
    },
    {
      "epoch": 1.1497047108913612,
      "grad_norm": 0.41451773047447205,
      "learning_rate": 8.51449773258211e-06,
      "loss": 0.7485,
      "step": 4185
    },
    {
      "epoch": 1.149979398434281,
      "grad_norm": 0.4120958745479584,
      "learning_rate": 8.511749347258486e-06,
      "loss": 0.8129,
      "step": 4186
    },
    {
      "epoch": 1.150254085977201,
      "grad_norm": 0.48473691940307617,
      "learning_rate": 8.509000961934863e-06,
      "loss": 0.7286,
      "step": 4187
    },
    {
      "epoch": 1.1505287735201208,
      "grad_norm": 0.4832138121128082,
      "learning_rate": 8.506252576611242e-06,
      "loss": 0.8139,
      "step": 4188
    },
    {
      "epoch": 1.1508034610630409,
      "grad_norm": 0.5176620483398438,
      "learning_rate": 8.503504191287619e-06,
      "loss": 0.8722,
      "step": 4189
    },
    {
      "epoch": 1.1510781486059607,
      "grad_norm": 0.5381171107292175,
      "learning_rate": 8.500755805963998e-06,
      "loss": 0.8722,
      "step": 4190
    },
    {
      "epoch": 1.1513528361488807,
      "grad_norm": 0.5323605537414551,
      "learning_rate": 8.498007420640375e-06,
      "loss": 0.7265,
      "step": 4191
    },
    {
      "epoch": 1.1516275236918005,
      "grad_norm": 0.5309060215950012,
      "learning_rate": 8.495259035316752e-06,
      "loss": 0.8182,
      "step": 4192
    },
    {
      "epoch": 1.1519022112347206,
      "grad_norm": 0.475058913230896,
      "learning_rate": 8.49251064999313e-06,
      "loss": 0.6203,
      "step": 4193
    },
    {
      "epoch": 1.1521768987776404,
      "grad_norm": 0.4085457921028137,
      "learning_rate": 8.489762264669507e-06,
      "loss": 0.5595,
      "step": 4194
    },
    {
      "epoch": 1.1524515863205604,
      "grad_norm": 0.44733282923698425,
      "learning_rate": 8.487013879345884e-06,
      "loss": 0.9035,
      "step": 4195
    },
    {
      "epoch": 1.1527262738634803,
      "grad_norm": 0.4571681320667267,
      "learning_rate": 8.484265494022263e-06,
      "loss": 0.5866,
      "step": 4196
    },
    {
      "epoch": 1.1530009614064003,
      "grad_norm": 0.3704262673854828,
      "learning_rate": 8.48151710869864e-06,
      "loss": 0.7555,
      "step": 4197
    },
    {
      "epoch": 1.15327564894932,
      "grad_norm": 0.4282141625881195,
      "learning_rate": 8.478768723375019e-06,
      "loss": 0.7369,
      "step": 4198
    },
    {
      "epoch": 1.1535503364922401,
      "grad_norm": 0.4151206612586975,
      "learning_rate": 8.476020338051394e-06,
      "loss": 0.7039,
      "step": 4199
    },
    {
      "epoch": 1.15382502403516,
      "grad_norm": 0.44901686906814575,
      "learning_rate": 8.473271952727773e-06,
      "loss": 0.8509,
      "step": 4200
    },
    {
      "epoch": 1.15409971157808,
      "grad_norm": 0.4795321226119995,
      "learning_rate": 8.47052356740415e-06,
      "loss": 0.9235,
      "step": 4201
    },
    {
      "epoch": 1.1543743991209998,
      "grad_norm": 0.3587228059768677,
      "learning_rate": 8.467775182080529e-06,
      "loss": 0.6995,
      "step": 4202
    },
    {
      "epoch": 1.1546490866639199,
      "grad_norm": 0.3889414668083191,
      "learning_rate": 8.465026796756906e-06,
      "loss": 0.7171,
      "step": 4203
    },
    {
      "epoch": 1.1549237742068397,
      "grad_norm": 0.43348976969718933,
      "learning_rate": 8.462278411433283e-06,
      "loss": 0.8531,
      "step": 4204
    },
    {
      "epoch": 1.1551984617497597,
      "grad_norm": 0.437216192483902,
      "learning_rate": 8.459530026109661e-06,
      "loss": 0.6924,
      "step": 4205
    },
    {
      "epoch": 1.1554731492926795,
      "grad_norm": 0.5269249081611633,
      "learning_rate": 8.456781640786038e-06,
      "loss": 0.9137,
      "step": 4206
    },
    {
      "epoch": 1.1557478368355996,
      "grad_norm": 0.4781726896762848,
      "learning_rate": 8.454033255462416e-06,
      "loss": 0.785,
      "step": 4207
    },
    {
      "epoch": 1.1560225243785194,
      "grad_norm": 0.49377837777137756,
      "learning_rate": 8.451284870138794e-06,
      "loss": 0.7525,
      "step": 4208
    },
    {
      "epoch": 1.1562972119214394,
      "grad_norm": 0.4330034554004669,
      "learning_rate": 8.448536484815171e-06,
      "loss": 0.8744,
      "step": 4209
    },
    {
      "epoch": 1.1565718994643592,
      "grad_norm": 0.4719419777393341,
      "learning_rate": 8.44578809949155e-06,
      "loss": 0.8572,
      "step": 4210
    },
    {
      "epoch": 1.1568465870072793,
      "grad_norm": 0.3586428761482239,
      "learning_rate": 8.443039714167927e-06,
      "loss": 0.7229,
      "step": 4211
    },
    {
      "epoch": 1.157121274550199,
      "grad_norm": 0.4599624574184418,
      "learning_rate": 8.440291328844304e-06,
      "loss": 1.1008,
      "step": 4212
    },
    {
      "epoch": 1.1573959620931191,
      "grad_norm": 0.49391117691993713,
      "learning_rate": 8.437542943520683e-06,
      "loss": 0.6089,
      "step": 4213
    },
    {
      "epoch": 1.157670649636039,
      "grad_norm": 0.480849951505661,
      "learning_rate": 8.43479455819706e-06,
      "loss": 0.6471,
      "step": 4214
    },
    {
      "epoch": 1.157945337178959,
      "grad_norm": 0.6061499714851379,
      "learning_rate": 8.432046172873437e-06,
      "loss": 0.9606,
      "step": 4215
    },
    {
      "epoch": 1.1582200247218788,
      "grad_norm": 0.4555921256542206,
      "learning_rate": 8.429297787549815e-06,
      "loss": 1.0847,
      "step": 4216
    },
    {
      "epoch": 1.1584947122647988,
      "grad_norm": 0.5059974789619446,
      "learning_rate": 8.426549402226192e-06,
      "loss": 0.682,
      "step": 4217
    },
    {
      "epoch": 1.1587693998077186,
      "grad_norm": 0.5199582576751709,
      "learning_rate": 8.423801016902571e-06,
      "loss": 0.7417,
      "step": 4218
    },
    {
      "epoch": 1.1590440873506387,
      "grad_norm": 0.3829125463962555,
      "learning_rate": 8.421052631578948e-06,
      "loss": 0.4998,
      "step": 4219
    },
    {
      "epoch": 1.1593187748935585,
      "grad_norm": 0.437107115983963,
      "learning_rate": 8.418304246255325e-06,
      "loss": 0.8072,
      "step": 4220
    },
    {
      "epoch": 1.1595934624364785,
      "grad_norm": 0.3704370856285095,
      "learning_rate": 8.415555860931704e-06,
      "loss": 0.726,
      "step": 4221
    },
    {
      "epoch": 1.1598681499793984,
      "grad_norm": 0.4829767346382141,
      "learning_rate": 8.412807475608081e-06,
      "loss": 0.838,
      "step": 4222
    },
    {
      "epoch": 1.1601428375223184,
      "grad_norm": 0.49250560998916626,
      "learning_rate": 8.410059090284458e-06,
      "loss": 0.8878,
      "step": 4223
    },
    {
      "epoch": 1.1604175250652382,
      "grad_norm": 0.519269585609436,
      "learning_rate": 8.407310704960837e-06,
      "loss": 0.7297,
      "step": 4224
    },
    {
      "epoch": 1.1606922126081582,
      "grad_norm": 0.41984933614730835,
      "learning_rate": 8.404562319637214e-06,
      "loss": 0.8589,
      "step": 4225
    },
    {
      "epoch": 1.160966900151078,
      "grad_norm": 0.4385926127433777,
      "learning_rate": 8.401813934313592e-06,
      "loss": 0.5082,
      "step": 4226
    },
    {
      "epoch": 1.161241587693998,
      "grad_norm": 0.3732292354106903,
      "learning_rate": 8.399065548989968e-06,
      "loss": 1.1174,
      "step": 4227
    },
    {
      "epoch": 1.161516275236918,
      "grad_norm": 0.43764978647232056,
      "learning_rate": 8.396317163666347e-06,
      "loss": 0.6891,
      "step": 4228
    },
    {
      "epoch": 1.161790962779838,
      "grad_norm": 0.4533981680870056,
      "learning_rate": 8.393568778342725e-06,
      "loss": 0.7207,
      "step": 4229
    },
    {
      "epoch": 1.1620656503227578,
      "grad_norm": 0.511467456817627,
      "learning_rate": 8.390820393019102e-06,
      "loss": 0.6976,
      "step": 4230
    },
    {
      "epoch": 1.1623403378656778,
      "grad_norm": 0.46760568022727966,
      "learning_rate": 8.38807200769548e-06,
      "loss": 0.7692,
      "step": 4231
    },
    {
      "epoch": 1.1626150254085976,
      "grad_norm": 0.4725264608860016,
      "learning_rate": 8.385323622371856e-06,
      "loss": 0.748,
      "step": 4232
    },
    {
      "epoch": 1.1628897129515177,
      "grad_norm": 0.46940186619758606,
      "learning_rate": 8.382575237048235e-06,
      "loss": 0.5968,
      "step": 4233
    },
    {
      "epoch": 1.1631644004944377,
      "grad_norm": 0.40717753767967224,
      "learning_rate": 8.379826851724612e-06,
      "loss": 0.5176,
      "step": 4234
    },
    {
      "epoch": 1.1634390880373575,
      "grad_norm": 0.5079889297485352,
      "learning_rate": 8.377078466400989e-06,
      "loss": 0.9764,
      "step": 4235
    },
    {
      "epoch": 1.1637137755802773,
      "grad_norm": 0.35354337096214294,
      "learning_rate": 8.374330081077368e-06,
      "loss": 0.6917,
      "step": 4236
    },
    {
      "epoch": 1.1639884631231974,
      "grad_norm": 0.3739694654941559,
      "learning_rate": 8.371581695753745e-06,
      "loss": 0.7185,
      "step": 4237
    },
    {
      "epoch": 1.1642631506661174,
      "grad_norm": 0.4746476709842682,
      "learning_rate": 8.368833310430123e-06,
      "loss": 0.7874,
      "step": 4238
    },
    {
      "epoch": 1.1645378382090372,
      "grad_norm": 0.503607988357544,
      "learning_rate": 8.3660849251065e-06,
      "loss": 0.6459,
      "step": 4239
    },
    {
      "epoch": 1.164812525751957,
      "grad_norm": 0.34228184819221497,
      "learning_rate": 8.363336539782878e-06,
      "loss": 0.6226,
      "step": 4240
    },
    {
      "epoch": 1.165087213294877,
      "grad_norm": 0.47393131256103516,
      "learning_rate": 8.360588154459256e-06,
      "loss": 0.6884,
      "step": 4241
    },
    {
      "epoch": 1.1653619008377971,
      "grad_norm": 0.4002237915992737,
      "learning_rate": 8.357839769135633e-06,
      "loss": 0.9706,
      "step": 4242
    },
    {
      "epoch": 1.165636588380717,
      "grad_norm": 0.29627183079719543,
      "learning_rate": 8.35509138381201e-06,
      "loss": 0.8344,
      "step": 4243
    },
    {
      "epoch": 1.1659112759236367,
      "grad_norm": 0.45142364501953125,
      "learning_rate": 8.352342998488389e-06,
      "loss": 0.9511,
      "step": 4244
    },
    {
      "epoch": 1.1661859634665568,
      "grad_norm": 0.41057321429252625,
      "learning_rate": 8.349594613164766e-06,
      "loss": 0.4099,
      "step": 4245
    },
    {
      "epoch": 1.1664606510094768,
      "grad_norm": 0.4363393187522888,
      "learning_rate": 8.346846227841145e-06,
      "loss": 0.9409,
      "step": 4246
    },
    {
      "epoch": 1.1667353385523966,
      "grad_norm": 0.5647244453430176,
      "learning_rate": 8.344097842517522e-06,
      "loss": 0.724,
      "step": 4247
    },
    {
      "epoch": 1.1670100260953165,
      "grad_norm": 0.4183550477027893,
      "learning_rate": 8.341349457193899e-06,
      "loss": 0.6476,
      "step": 4248
    },
    {
      "epoch": 1.1672847136382365,
      "grad_norm": 0.49137353897094727,
      "learning_rate": 8.338601071870277e-06,
      "loss": 0.7133,
      "step": 4249
    },
    {
      "epoch": 1.1675594011811565,
      "grad_norm": 0.5185673832893372,
      "learning_rate": 8.335852686546655e-06,
      "loss": 0.8532,
      "step": 4250
    },
    {
      "epoch": 1.1678340887240763,
      "grad_norm": 0.44973328709602356,
      "learning_rate": 8.333104301223032e-06,
      "loss": 0.8372,
      "step": 4251
    },
    {
      "epoch": 1.1681087762669964,
      "grad_norm": 0.5826719403266907,
      "learning_rate": 8.33035591589941e-06,
      "loss": 0.841,
      "step": 4252
    },
    {
      "epoch": 1.1683834638099162,
      "grad_norm": 0.41452422738075256,
      "learning_rate": 8.327607530575787e-06,
      "loss": 0.7597,
      "step": 4253
    },
    {
      "epoch": 1.1686581513528362,
      "grad_norm": 0.334652841091156,
      "learning_rate": 8.324859145252166e-06,
      "loss": 0.799,
      "step": 4254
    },
    {
      "epoch": 1.168932838895756,
      "grad_norm": 0.4484504461288452,
      "learning_rate": 8.322110759928543e-06,
      "loss": 0.8802,
      "step": 4255
    },
    {
      "epoch": 1.169207526438676,
      "grad_norm": 0.5343471169471741,
      "learning_rate": 8.31936237460492e-06,
      "loss": 0.9445,
      "step": 4256
    },
    {
      "epoch": 1.169482213981596,
      "grad_norm": 0.4095478355884552,
      "learning_rate": 8.316613989281299e-06,
      "loss": 0.7939,
      "step": 4257
    },
    {
      "epoch": 1.169756901524516,
      "grad_norm": 0.3358507752418518,
      "learning_rate": 8.313865603957676e-06,
      "loss": 0.9405,
      "step": 4258
    },
    {
      "epoch": 1.1700315890674358,
      "grad_norm": 0.5054279565811157,
      "learning_rate": 8.311117218634053e-06,
      "loss": 0.8381,
      "step": 4259
    },
    {
      "epoch": 1.1703062766103558,
      "grad_norm": 0.4462062418460846,
      "learning_rate": 8.30836883331043e-06,
      "loss": 0.9411,
      "step": 4260
    },
    {
      "epoch": 1.1705809641532756,
      "grad_norm": 0.5909618139266968,
      "learning_rate": 8.305620447986809e-06,
      "loss": 0.6356,
      "step": 4261
    },
    {
      "epoch": 1.1708556516961957,
      "grad_norm": 0.37309733033180237,
      "learning_rate": 8.302872062663187e-06,
      "loss": 0.915,
      "step": 4262
    },
    {
      "epoch": 1.1711303392391155,
      "grad_norm": 0.37568753957748413,
      "learning_rate": 8.300123677339563e-06,
      "loss": 1.0314,
      "step": 4263
    },
    {
      "epoch": 1.1714050267820355,
      "grad_norm": 0.3805834650993347,
      "learning_rate": 8.297375292015941e-06,
      "loss": 0.8659,
      "step": 4264
    },
    {
      "epoch": 1.1716797143249553,
      "grad_norm": 0.3949870765209198,
      "learning_rate": 8.294626906692318e-06,
      "loss": 0.5771,
      "step": 4265
    },
    {
      "epoch": 1.1719544018678754,
      "grad_norm": 0.5058557987213135,
      "learning_rate": 8.291878521368697e-06,
      "loss": 0.8937,
      "step": 4266
    },
    {
      "epoch": 1.1722290894107952,
      "grad_norm": 0.3685765564441681,
      "learning_rate": 8.289130136045074e-06,
      "loss": 0.8516,
      "step": 4267
    },
    {
      "epoch": 1.1725037769537152,
      "grad_norm": 0.49919626116752625,
      "learning_rate": 8.286381750721451e-06,
      "loss": 0.8372,
      "step": 4268
    },
    {
      "epoch": 1.172778464496635,
      "grad_norm": 0.5063122510910034,
      "learning_rate": 8.28363336539783e-06,
      "loss": 0.793,
      "step": 4269
    },
    {
      "epoch": 1.173053152039555,
      "grad_norm": 0.5544183850288391,
      "learning_rate": 8.280884980074207e-06,
      "loss": 0.7741,
      "step": 4270
    },
    {
      "epoch": 1.1733278395824749,
      "grad_norm": 0.39944228529930115,
      "learning_rate": 8.278136594750584e-06,
      "loss": 0.8579,
      "step": 4271
    },
    {
      "epoch": 1.173602527125395,
      "grad_norm": 0.3251881003379822,
      "learning_rate": 8.275388209426963e-06,
      "loss": 0.821,
      "step": 4272
    },
    {
      "epoch": 1.1738772146683147,
      "grad_norm": 0.42006009817123413,
      "learning_rate": 8.27263982410334e-06,
      "loss": 0.6804,
      "step": 4273
    },
    {
      "epoch": 1.1741519022112348,
      "grad_norm": 0.4530197083950043,
      "learning_rate": 8.269891438779718e-06,
      "loss": 0.5296,
      "step": 4274
    },
    {
      "epoch": 1.1744265897541546,
      "grad_norm": 0.38821882009506226,
      "learning_rate": 8.267143053456095e-06,
      "loss": 0.7702,
      "step": 4275
    },
    {
      "epoch": 1.1747012772970746,
      "grad_norm": 0.44201359152793884,
      "learning_rate": 8.264394668132472e-06,
      "loss": 0.877,
      "step": 4276
    },
    {
      "epoch": 1.1749759648399944,
      "grad_norm": 0.3535976707935333,
      "learning_rate": 8.261646282808851e-06,
      "loss": 1.0252,
      "step": 4277
    },
    {
      "epoch": 1.1752506523829145,
      "grad_norm": 0.6487314105033875,
      "learning_rate": 8.258897897485228e-06,
      "loss": 0.8731,
      "step": 4278
    },
    {
      "epoch": 1.1755253399258343,
      "grad_norm": 0.2986038625240326,
      "learning_rate": 8.256149512161605e-06,
      "loss": 0.7535,
      "step": 4279
    },
    {
      "epoch": 1.1758000274687543,
      "grad_norm": 0.40870171785354614,
      "learning_rate": 8.253401126837984e-06,
      "loss": 0.7968,
      "step": 4280
    },
    {
      "epoch": 1.1760747150116742,
      "grad_norm": 0.4534000754356384,
      "learning_rate": 8.25065274151436e-06,
      "loss": 0.8891,
      "step": 4281
    },
    {
      "epoch": 1.1763494025545942,
      "grad_norm": 0.44943320751190186,
      "learning_rate": 8.24790435619074e-06,
      "loss": 0.7675,
      "step": 4282
    },
    {
      "epoch": 1.176624090097514,
      "grad_norm": 0.2971000373363495,
      "learning_rate": 8.245155970867117e-06,
      "loss": 0.5833,
      "step": 4283
    },
    {
      "epoch": 1.176898777640434,
      "grad_norm": 0.35908347368240356,
      "learning_rate": 8.242407585543494e-06,
      "loss": 0.4973,
      "step": 4284
    },
    {
      "epoch": 1.1771734651833539,
      "grad_norm": 0.42064768075942993,
      "learning_rate": 8.239659200219872e-06,
      "loss": 0.9133,
      "step": 4285
    },
    {
      "epoch": 1.177448152726274,
      "grad_norm": 0.3748287260532379,
      "learning_rate": 8.23691081489625e-06,
      "loss": 0.7141,
      "step": 4286
    },
    {
      "epoch": 1.1777228402691937,
      "grad_norm": 0.40590861439704895,
      "learning_rate": 8.234162429572626e-06,
      "loss": 0.8371,
      "step": 4287
    },
    {
      "epoch": 1.1779975278121138,
      "grad_norm": 0.4245665669441223,
      "learning_rate": 8.231414044249005e-06,
      "loss": 0.6838,
      "step": 4288
    },
    {
      "epoch": 1.1782722153550336,
      "grad_norm": 0.7051292657852173,
      "learning_rate": 8.228665658925382e-06,
      "loss": 0.8549,
      "step": 4289
    },
    {
      "epoch": 1.1785469028979536,
      "grad_norm": 0.5030488967895508,
      "learning_rate": 8.22591727360176e-06,
      "loss": 0.8661,
      "step": 4290
    },
    {
      "epoch": 1.1788215904408734,
      "grad_norm": 0.4307345747947693,
      "learning_rate": 8.223168888278136e-06,
      "loss": 0.7343,
      "step": 4291
    },
    {
      "epoch": 1.1790962779837935,
      "grad_norm": 0.47651055455207825,
      "learning_rate": 8.220420502954515e-06,
      "loss": 0.9166,
      "step": 4292
    },
    {
      "epoch": 1.1793709655267133,
      "grad_norm": 0.4743233621120453,
      "learning_rate": 8.217672117630894e-06,
      "loss": 0.7167,
      "step": 4293
    },
    {
      "epoch": 1.1796456530696333,
      "grad_norm": 0.34158453345298767,
      "learning_rate": 8.21492373230727e-06,
      "loss": 0.8716,
      "step": 4294
    },
    {
      "epoch": 1.1799203406125531,
      "grad_norm": 0.2995149791240692,
      "learning_rate": 8.212175346983648e-06,
      "loss": 0.4941,
      "step": 4295
    },
    {
      "epoch": 1.1801950281554732,
      "grad_norm": 0.4862208366394043,
      "learning_rate": 8.209426961660025e-06,
      "loss": 1.067,
      "step": 4296
    },
    {
      "epoch": 1.180469715698393,
      "grad_norm": 0.7367683053016663,
      "learning_rate": 8.206678576336403e-06,
      "loss": 0.7078,
      "step": 4297
    },
    {
      "epoch": 1.180744403241313,
      "grad_norm": 0.3946194350719452,
      "learning_rate": 8.20393019101278e-06,
      "loss": 0.6231,
      "step": 4298
    },
    {
      "epoch": 1.1810190907842328,
      "grad_norm": 0.3716247081756592,
      "learning_rate": 8.201181805689157e-06,
      "loss": 0.9436,
      "step": 4299
    },
    {
      "epoch": 1.1812937783271529,
      "grad_norm": 0.3981447219848633,
      "learning_rate": 8.198433420365536e-06,
      "loss": 0.7868,
      "step": 4300
    },
    {
      "epoch": 1.1815684658700727,
      "grad_norm": 0.4226513206958771,
      "learning_rate": 8.195685035041913e-06,
      "loss": 1.0068,
      "step": 4301
    },
    {
      "epoch": 1.1818431534129927,
      "grad_norm": 0.37626636028289795,
      "learning_rate": 8.192936649718292e-06,
      "loss": 1.1486,
      "step": 4302
    },
    {
      "epoch": 1.1821178409559128,
      "grad_norm": 0.5271945595741272,
      "learning_rate": 8.190188264394669e-06,
      "loss": 0.7626,
      "step": 4303
    },
    {
      "epoch": 1.1823925284988326,
      "grad_norm": 0.6033921241760254,
      "learning_rate": 8.187439879071046e-06,
      "loss": 0.7225,
      "step": 4304
    },
    {
      "epoch": 1.1826672160417524,
      "grad_norm": 0.4664328396320343,
      "learning_rate": 8.184691493747425e-06,
      "loss": 0.7182,
      "step": 4305
    },
    {
      "epoch": 1.1829419035846724,
      "grad_norm": 0.4649301767349243,
      "learning_rate": 8.181943108423802e-06,
      "loss": 0.8111,
      "step": 4306
    },
    {
      "epoch": 1.1832165911275925,
      "grad_norm": 0.3548402786254883,
      "learning_rate": 8.179194723100179e-06,
      "loss": 0.7594,
      "step": 4307
    },
    {
      "epoch": 1.1834912786705123,
      "grad_norm": 0.5062845945358276,
      "learning_rate": 8.176446337776557e-06,
      "loss": 0.9106,
      "step": 4308
    },
    {
      "epoch": 1.183765966213432,
      "grad_norm": 0.44972488284111023,
      "learning_rate": 8.173697952452934e-06,
      "loss": 1.0119,
      "step": 4309
    },
    {
      "epoch": 1.1840406537563521,
      "grad_norm": 0.414818674325943,
      "learning_rate": 8.170949567129313e-06,
      "loss": 0.6782,
      "step": 4310
    },
    {
      "epoch": 1.1843153412992722,
      "grad_norm": 0.40353479981422424,
      "learning_rate": 8.16820118180569e-06,
      "loss": 0.6952,
      "step": 4311
    },
    {
      "epoch": 1.184590028842192,
      "grad_norm": 0.43238329887390137,
      "learning_rate": 8.165452796482067e-06,
      "loss": 0.6345,
      "step": 4312
    },
    {
      "epoch": 1.1848647163851118,
      "grad_norm": 0.4679376780986786,
      "learning_rate": 8.162704411158446e-06,
      "loss": 0.778,
      "step": 4313
    },
    {
      "epoch": 1.1851394039280319,
      "grad_norm": 0.473751038312912,
      "learning_rate": 8.159956025834823e-06,
      "loss": 0.7523,
      "step": 4314
    },
    {
      "epoch": 1.185414091470952,
      "grad_norm": 0.4960268437862396,
      "learning_rate": 8.1572076405112e-06,
      "loss": 0.9353,
      "step": 4315
    },
    {
      "epoch": 1.1856887790138717,
      "grad_norm": 0.5040628910064697,
      "learning_rate": 8.154459255187579e-06,
      "loss": 0.7132,
      "step": 4316
    },
    {
      "epoch": 1.1859634665567917,
      "grad_norm": 0.4014180600643158,
      "learning_rate": 8.151710869863956e-06,
      "loss": 0.855,
      "step": 4317
    },
    {
      "epoch": 1.1862381540997116,
      "grad_norm": 0.39820343255996704,
      "learning_rate": 8.148962484540334e-06,
      "loss": 0.8258,
      "step": 4318
    },
    {
      "epoch": 1.1865128416426316,
      "grad_norm": 0.5544390082359314,
      "learning_rate": 8.146214099216711e-06,
      "loss": 0.7418,
      "step": 4319
    },
    {
      "epoch": 1.1867875291855514,
      "grad_norm": 0.37048688530921936,
      "learning_rate": 8.143465713893088e-06,
      "loss": 0.7156,
      "step": 4320
    },
    {
      "epoch": 1.1870622167284715,
      "grad_norm": 0.4213361442089081,
      "learning_rate": 8.140717328569467e-06,
      "loss": 0.7591,
      "step": 4321
    },
    {
      "epoch": 1.1873369042713913,
      "grad_norm": 0.42606526613235474,
      "learning_rate": 8.137968943245844e-06,
      "loss": 0.686,
      "step": 4322
    },
    {
      "epoch": 1.1876115918143113,
      "grad_norm": 0.4521978497505188,
      "learning_rate": 8.135220557922221e-06,
      "loss": 0.6757,
      "step": 4323
    },
    {
      "epoch": 1.1878862793572311,
      "grad_norm": 0.3285727798938751,
      "learning_rate": 8.132472172598598e-06,
      "loss": 0.6826,
      "step": 4324
    },
    {
      "epoch": 1.1881609669001512,
      "grad_norm": 0.44205641746520996,
      "learning_rate": 8.129723787274977e-06,
      "loss": 0.7203,
      "step": 4325
    },
    {
      "epoch": 1.188435654443071,
      "grad_norm": 0.4377013146877289,
      "learning_rate": 8.126975401951356e-06,
      "loss": 0.6251,
      "step": 4326
    },
    {
      "epoch": 1.188710341985991,
      "grad_norm": 0.42547518014907837,
      "learning_rate": 8.124227016627731e-06,
      "loss": 0.4702,
      "step": 4327
    },
    {
      "epoch": 1.1889850295289108,
      "grad_norm": 0.5759099721908569,
      "learning_rate": 8.12147863130411e-06,
      "loss": 0.7853,
      "step": 4328
    },
    {
      "epoch": 1.1892597170718309,
      "grad_norm": 0.5426904559135437,
      "learning_rate": 8.118730245980487e-06,
      "loss": 0.6101,
      "step": 4329
    },
    {
      "epoch": 1.1895344046147507,
      "grad_norm": 0.3460353910923004,
      "learning_rate": 8.115981860656865e-06,
      "loss": 0.6921,
      "step": 4330
    },
    {
      "epoch": 1.1898090921576707,
      "grad_norm": 0.6189436316490173,
      "learning_rate": 8.113233475333242e-06,
      "loss": 0.8038,
      "step": 4331
    },
    {
      "epoch": 1.1900837797005905,
      "grad_norm": 0.5984348654747009,
      "learning_rate": 8.11048509000962e-06,
      "loss": 0.7486,
      "step": 4332
    },
    {
      "epoch": 1.1903584672435106,
      "grad_norm": 0.5099565386772156,
      "learning_rate": 8.107736704685998e-06,
      "loss": 1.0435,
      "step": 4333
    },
    {
      "epoch": 1.1906331547864304,
      "grad_norm": 0.5005665421485901,
      "learning_rate": 8.104988319362375e-06,
      "loss": 0.894,
      "step": 4334
    },
    {
      "epoch": 1.1909078423293504,
      "grad_norm": 0.5075063109397888,
      "learning_rate": 8.102239934038752e-06,
      "loss": 0.8121,
      "step": 4335
    },
    {
      "epoch": 1.1911825298722702,
      "grad_norm": 0.6751214265823364,
      "learning_rate": 8.09949154871513e-06,
      "loss": 0.75,
      "step": 4336
    },
    {
      "epoch": 1.1914572174151903,
      "grad_norm": 0.42071086168289185,
      "learning_rate": 8.096743163391508e-06,
      "loss": 1.0474,
      "step": 4337
    },
    {
      "epoch": 1.19173190495811,
      "grad_norm": 0.5125982165336609,
      "learning_rate": 8.093994778067887e-06,
      "loss": 0.6044,
      "step": 4338
    },
    {
      "epoch": 1.1920065925010301,
      "grad_norm": 0.40918222069740295,
      "learning_rate": 8.091246392744264e-06,
      "loss": 0.5,
      "step": 4339
    },
    {
      "epoch": 1.19228128004395,
      "grad_norm": 0.44986462593078613,
      "learning_rate": 8.08849800742064e-06,
      "loss": 0.7403,
      "step": 4340
    },
    {
      "epoch": 1.19255596758687,
      "grad_norm": 0.3820019066333771,
      "learning_rate": 8.08574962209702e-06,
      "loss": 0.6477,
      "step": 4341
    },
    {
      "epoch": 1.1928306551297898,
      "grad_norm": 0.43447864055633545,
      "learning_rate": 8.083001236773396e-06,
      "loss": 0.7212,
      "step": 4342
    },
    {
      "epoch": 1.1931053426727098,
      "grad_norm": 0.43180587887763977,
      "learning_rate": 8.080252851449773e-06,
      "loss": 0.6407,
      "step": 4343
    },
    {
      "epoch": 1.1933800302156297,
      "grad_norm": 0.4465253949165344,
      "learning_rate": 8.077504466126152e-06,
      "loss": 0.8855,
      "step": 4344
    },
    {
      "epoch": 1.1936547177585497,
      "grad_norm": 0.4253655970096588,
      "learning_rate": 8.074756080802529e-06,
      "loss": 1.0166,
      "step": 4345
    },
    {
      "epoch": 1.1939294053014695,
      "grad_norm": 0.42755064368247986,
      "learning_rate": 8.072007695478908e-06,
      "loss": 0.5911,
      "step": 4346
    },
    {
      "epoch": 1.1942040928443896,
      "grad_norm": 0.4090331494808197,
      "learning_rate": 8.069259310155285e-06,
      "loss": 0.7522,
      "step": 4347
    },
    {
      "epoch": 1.1944787803873094,
      "grad_norm": 0.5452840328216553,
      "learning_rate": 8.066510924831662e-06,
      "loss": 0.9348,
      "step": 4348
    },
    {
      "epoch": 1.1947534679302294,
      "grad_norm": 0.42100846767425537,
      "learning_rate": 8.06376253950804e-06,
      "loss": 0.6196,
      "step": 4349
    },
    {
      "epoch": 1.1950281554731492,
      "grad_norm": 0.42283642292022705,
      "learning_rate": 8.061014154184418e-06,
      "loss": 0.7486,
      "step": 4350
    },
    {
      "epoch": 1.1953028430160693,
      "grad_norm": 0.4347936511039734,
      "learning_rate": 8.058265768860795e-06,
      "loss": 0.5703,
      "step": 4351
    },
    {
      "epoch": 1.195577530558989,
      "grad_norm": 0.3848274052143097,
      "learning_rate": 8.055517383537173e-06,
      "loss": 0.7216,
      "step": 4352
    },
    {
      "epoch": 1.1958522181019091,
      "grad_norm": 0.41134074330329895,
      "learning_rate": 8.05276899821355e-06,
      "loss": 0.956,
      "step": 4353
    },
    {
      "epoch": 1.196126905644829,
      "grad_norm": 0.3844347596168518,
      "learning_rate": 8.050020612889929e-06,
      "loss": 0.4879,
      "step": 4354
    },
    {
      "epoch": 1.196401593187749,
      "grad_norm": 0.4829365015029907,
      "learning_rate": 8.047272227566304e-06,
      "loss": 0.7753,
      "step": 4355
    },
    {
      "epoch": 1.1966762807306688,
      "grad_norm": 0.5213337540626526,
      "learning_rate": 8.044523842242683e-06,
      "loss": 0.7455,
      "step": 4356
    },
    {
      "epoch": 1.1969509682735888,
      "grad_norm": 0.6088531613349915,
      "learning_rate": 8.04177545691906e-06,
      "loss": 0.6693,
      "step": 4357
    },
    {
      "epoch": 1.1972256558165086,
      "grad_norm": 0.4402381479740143,
      "learning_rate": 8.039027071595439e-06,
      "loss": 0.7874,
      "step": 4358
    },
    {
      "epoch": 1.1975003433594287,
      "grad_norm": 0.3602839708328247,
      "learning_rate": 8.036278686271816e-06,
      "loss": 0.7659,
      "step": 4359
    },
    {
      "epoch": 1.1977750309023485,
      "grad_norm": 0.6082988381385803,
      "learning_rate": 8.033530300948193e-06,
      "loss": 0.7809,
      "step": 4360
    },
    {
      "epoch": 1.1980497184452685,
      "grad_norm": 0.4504801034927368,
      "learning_rate": 8.030781915624572e-06,
      "loss": 0.7178,
      "step": 4361
    },
    {
      "epoch": 1.1983244059881883,
      "grad_norm": 0.5000601410865784,
      "learning_rate": 8.028033530300949e-06,
      "loss": 0.6992,
      "step": 4362
    },
    {
      "epoch": 1.1985990935311084,
      "grad_norm": 0.5544174909591675,
      "learning_rate": 8.025285144977326e-06,
      "loss": 0.7771,
      "step": 4363
    },
    {
      "epoch": 1.1988737810740282,
      "grad_norm": 0.29268619418144226,
      "learning_rate": 8.022536759653704e-06,
      "loss": 0.7734,
      "step": 4364
    },
    {
      "epoch": 1.1991484686169482,
      "grad_norm": 0.5712262988090515,
      "learning_rate": 8.019788374330081e-06,
      "loss": 0.9218,
      "step": 4365
    },
    {
      "epoch": 1.199423156159868,
      "grad_norm": 0.5192736387252808,
      "learning_rate": 8.01703998900646e-06,
      "loss": 0.7693,
      "step": 4366
    },
    {
      "epoch": 1.199697843702788,
      "grad_norm": 0.5461355447769165,
      "learning_rate": 8.014291603682837e-06,
      "loss": 0.5505,
      "step": 4367
    },
    {
      "epoch": 1.1999725312457081,
      "grad_norm": 0.49984169006347656,
      "learning_rate": 8.011543218359214e-06,
      "loss": 0.7152,
      "step": 4368
    },
    {
      "epoch": 1.200247218788628,
      "grad_norm": 0.3327881395816803,
      "learning_rate": 8.008794833035593e-06,
      "loss": 0.5174,
      "step": 4369
    },
    {
      "epoch": 1.2005219063315478,
      "grad_norm": 0.5232974290847778,
      "learning_rate": 8.00604644771197e-06,
      "loss": 0.6087,
      "step": 4370
    },
    {
      "epoch": 1.2007965938744678,
      "grad_norm": 0.47899916768074036,
      "learning_rate": 8.003298062388347e-06,
      "loss": 0.8901,
      "step": 4371
    },
    {
      "epoch": 1.2010712814173878,
      "grad_norm": 0.5159834027290344,
      "learning_rate": 8.000549677064726e-06,
      "loss": 0.8761,
      "step": 4372
    },
    {
      "epoch": 1.2013459689603077,
      "grad_norm": 0.5041170120239258,
      "learning_rate": 7.997801291741103e-06,
      "loss": 0.77,
      "step": 4373
    },
    {
      "epoch": 1.2016206565032275,
      "grad_norm": 0.39919888973236084,
      "learning_rate": 7.995052906417481e-06,
      "loss": 0.8086,
      "step": 4374
    },
    {
      "epoch": 1.2018953440461475,
      "grad_norm": 0.4533628821372986,
      "learning_rate": 7.992304521093858e-06,
      "loss": 0.6352,
      "step": 4375
    },
    {
      "epoch": 1.2021700315890675,
      "grad_norm": 0.5679082870483398,
      "learning_rate": 7.989556135770235e-06,
      "loss": 0.6955,
      "step": 4376
    },
    {
      "epoch": 1.2024447191319874,
      "grad_norm": 0.4349135756492615,
      "learning_rate": 7.986807750446614e-06,
      "loss": 0.872,
      "step": 4377
    },
    {
      "epoch": 1.2027194066749072,
      "grad_norm": 0.40241554379463196,
      "learning_rate": 7.984059365122991e-06,
      "loss": 0.8835,
      "step": 4378
    },
    {
      "epoch": 1.2029940942178272,
      "grad_norm": 0.4479895234107971,
      "learning_rate": 7.981310979799368e-06,
      "loss": 0.9695,
      "step": 4379
    },
    {
      "epoch": 1.2032687817607473,
      "grad_norm": 0.350294291973114,
      "learning_rate": 7.978562594475747e-06,
      "loss": 0.4606,
      "step": 4380
    },
    {
      "epoch": 1.203543469303667,
      "grad_norm": 0.5521461367607117,
      "learning_rate": 7.975814209152124e-06,
      "loss": 0.8274,
      "step": 4381
    },
    {
      "epoch": 1.2038181568465869,
      "grad_norm": 0.42674246430397034,
      "learning_rate": 7.973065823828503e-06,
      "loss": 0.6583,
      "step": 4382
    },
    {
      "epoch": 1.204092844389507,
      "grad_norm": 0.44709470868110657,
      "learning_rate": 7.970317438504878e-06,
      "loss": 0.551,
      "step": 4383
    },
    {
      "epoch": 1.204367531932427,
      "grad_norm": 0.33063092827796936,
      "learning_rate": 7.967569053181257e-06,
      "loss": 0.7394,
      "step": 4384
    },
    {
      "epoch": 1.2046422194753468,
      "grad_norm": 0.5624905824661255,
      "learning_rate": 7.964820667857635e-06,
      "loss": 0.9421,
      "step": 4385
    },
    {
      "epoch": 1.2049169070182668,
      "grad_norm": 0.3813599944114685,
      "learning_rate": 7.962072282534012e-06,
      "loss": 0.8559,
      "step": 4386
    },
    {
      "epoch": 1.2051915945611866,
      "grad_norm": 0.3825848400592804,
      "learning_rate": 7.95932389721039e-06,
      "loss": 0.7545,
      "step": 4387
    },
    {
      "epoch": 1.2054662821041067,
      "grad_norm": 0.314808189868927,
      "learning_rate": 7.956575511886766e-06,
      "loss": 0.5586,
      "step": 4388
    },
    {
      "epoch": 1.2057409696470265,
      "grad_norm": 0.4418025612831116,
      "learning_rate": 7.953827126563145e-06,
      "loss": 0.7515,
      "step": 4389
    },
    {
      "epoch": 1.2060156571899465,
      "grad_norm": 0.5469188094139099,
      "learning_rate": 7.951078741239522e-06,
      "loss": 0.6801,
      "step": 4390
    },
    {
      "epoch": 1.2062903447328663,
      "grad_norm": 0.5065515637397766,
      "learning_rate": 7.948330355915899e-06,
      "loss": 0.5833,
      "step": 4391
    },
    {
      "epoch": 1.2065650322757864,
      "grad_norm": 0.4539637267589569,
      "learning_rate": 7.945581970592278e-06,
      "loss": 0.582,
      "step": 4392
    },
    {
      "epoch": 1.2068397198187062,
      "grad_norm": 0.4382917582988739,
      "learning_rate": 7.942833585268655e-06,
      "loss": 0.6019,
      "step": 4393
    },
    {
      "epoch": 1.2071144073616262,
      "grad_norm": 0.48640215396881104,
      "learning_rate": 7.940085199945034e-06,
      "loss": 0.6368,
      "step": 4394
    },
    {
      "epoch": 1.207389094904546,
      "grad_norm": 0.4552904963493347,
      "learning_rate": 7.93733681462141e-06,
      "loss": 0.7083,
      "step": 4395
    },
    {
      "epoch": 1.207663782447466,
      "grad_norm": 0.42167750000953674,
      "learning_rate": 7.934588429297788e-06,
      "loss": 0.6596,
      "step": 4396
    },
    {
      "epoch": 1.207938469990386,
      "grad_norm": 0.4950363337993622,
      "learning_rate": 7.931840043974166e-06,
      "loss": 0.6938,
      "step": 4397
    },
    {
      "epoch": 1.208213157533306,
      "grad_norm": 0.31878337264060974,
      "learning_rate": 7.929091658650543e-06,
      "loss": 0.8751,
      "step": 4398
    },
    {
      "epoch": 1.2084878450762258,
      "grad_norm": 0.4305116832256317,
      "learning_rate": 7.92634327332692e-06,
      "loss": 0.8547,
      "step": 4399
    },
    {
      "epoch": 1.2087625326191458,
      "grad_norm": 0.45864731073379517,
      "learning_rate": 7.923594888003299e-06,
      "loss": 0.7155,
      "step": 4400
    },
    {
      "epoch": 1.2090372201620656,
      "grad_norm": 0.3830079138278961,
      "learning_rate": 7.920846502679676e-06,
      "loss": 0.6761,
      "step": 4401
    },
    {
      "epoch": 1.2093119077049856,
      "grad_norm": 0.38334399461746216,
      "learning_rate": 7.918098117356055e-06,
      "loss": 0.5962,
      "step": 4402
    },
    {
      "epoch": 1.2095865952479055,
      "grad_norm": 0.44774019718170166,
      "learning_rate": 7.915349732032432e-06,
      "loss": 0.9012,
      "step": 4403
    },
    {
      "epoch": 1.2098612827908255,
      "grad_norm": 0.4665972888469696,
      "learning_rate": 7.912601346708809e-06,
      "loss": 0.6875,
      "step": 4404
    },
    {
      "epoch": 1.2101359703337453,
      "grad_norm": 0.43563681840896606,
      "learning_rate": 7.909852961385188e-06,
      "loss": 0.8383,
      "step": 4405
    },
    {
      "epoch": 1.2104106578766654,
      "grad_norm": 0.42455416917800903,
      "learning_rate": 7.907104576061565e-06,
      "loss": 0.9802,
      "step": 4406
    },
    {
      "epoch": 1.2106853454195852,
      "grad_norm": 0.34304726123809814,
      "learning_rate": 7.904356190737942e-06,
      "loss": 0.6797,
      "step": 4407
    },
    {
      "epoch": 1.2109600329625052,
      "grad_norm": 0.45593181252479553,
      "learning_rate": 7.90160780541432e-06,
      "loss": 0.6079,
      "step": 4408
    },
    {
      "epoch": 1.211234720505425,
      "grad_norm": 0.42851072549819946,
      "learning_rate": 7.898859420090697e-06,
      "loss": 0.7302,
      "step": 4409
    },
    {
      "epoch": 1.211509408048345,
      "grad_norm": 0.45757320523262024,
      "learning_rate": 7.896111034767076e-06,
      "loss": 0.7243,
      "step": 4410
    },
    {
      "epoch": 1.2117840955912649,
      "grad_norm": 0.6601826548576355,
      "learning_rate": 7.893362649443453e-06,
      "loss": 0.9178,
      "step": 4411
    },
    {
      "epoch": 1.212058783134185,
      "grad_norm": 0.3049335777759552,
      "learning_rate": 7.89061426411983e-06,
      "loss": 0.666,
      "step": 4412
    },
    {
      "epoch": 1.2123334706771047,
      "grad_norm": 0.4776659607887268,
      "learning_rate": 7.887865878796209e-06,
      "loss": 0.9076,
      "step": 4413
    },
    {
      "epoch": 1.2126081582200248,
      "grad_norm": 0.40953344106674194,
      "learning_rate": 7.885117493472586e-06,
      "loss": 0.8122,
      "step": 4414
    },
    {
      "epoch": 1.2128828457629446,
      "grad_norm": 0.3426366150379181,
      "learning_rate": 7.882369108148963e-06,
      "loss": 0.5844,
      "step": 4415
    },
    {
      "epoch": 1.2131575333058646,
      "grad_norm": 0.5465901494026184,
      "learning_rate": 7.87962072282534e-06,
      "loss": 0.8513,
      "step": 4416
    },
    {
      "epoch": 1.2134322208487844,
      "grad_norm": 0.3909267485141754,
      "learning_rate": 7.876872337501719e-06,
      "loss": 0.889,
      "step": 4417
    },
    {
      "epoch": 1.2137069083917045,
      "grad_norm": 0.5134691596031189,
      "learning_rate": 7.874123952178097e-06,
      "loss": 0.5203,
      "step": 4418
    },
    {
      "epoch": 1.2139815959346243,
      "grad_norm": 0.6086059212684631,
      "learning_rate": 7.871375566854473e-06,
      "loss": 0.7317,
      "step": 4419
    },
    {
      "epoch": 1.2142562834775443,
      "grad_norm": 0.3894067406654358,
      "learning_rate": 7.868627181530851e-06,
      "loss": 0.6233,
      "step": 4420
    },
    {
      "epoch": 1.2145309710204641,
      "grad_norm": 0.3771394193172455,
      "learning_rate": 7.865878796207228e-06,
      "loss": 0.8641,
      "step": 4421
    },
    {
      "epoch": 1.2148056585633842,
      "grad_norm": 0.38084936141967773,
      "learning_rate": 7.863130410883607e-06,
      "loss": 0.7397,
      "step": 4422
    },
    {
      "epoch": 1.215080346106304,
      "grad_norm": 0.6611708402633667,
      "learning_rate": 7.860382025559984e-06,
      "loss": 1.0365,
      "step": 4423
    },
    {
      "epoch": 1.215355033649224,
      "grad_norm": 0.4667147994041443,
      "learning_rate": 7.857633640236361e-06,
      "loss": 0.7581,
      "step": 4424
    },
    {
      "epoch": 1.2156297211921439,
      "grad_norm": 0.4315234422683716,
      "learning_rate": 7.85488525491274e-06,
      "loss": 0.5599,
      "step": 4425
    },
    {
      "epoch": 1.215904408735064,
      "grad_norm": 0.48773884773254395,
      "learning_rate": 7.852136869589117e-06,
      "loss": 0.7097,
      "step": 4426
    },
    {
      "epoch": 1.2161790962779837,
      "grad_norm": 0.36107516288757324,
      "learning_rate": 7.849388484265494e-06,
      "loss": 0.8193,
      "step": 4427
    },
    {
      "epoch": 1.2164537838209037,
      "grad_norm": 0.43460601568222046,
      "learning_rate": 7.846640098941873e-06,
      "loss": 0.6915,
      "step": 4428
    },
    {
      "epoch": 1.2167284713638236,
      "grad_norm": 0.433943510055542,
      "learning_rate": 7.84389171361825e-06,
      "loss": 0.7219,
      "step": 4429
    },
    {
      "epoch": 1.2170031589067436,
      "grad_norm": 0.6023010015487671,
      "learning_rate": 7.841143328294628e-06,
      "loss": 0.8635,
      "step": 4430
    },
    {
      "epoch": 1.2172778464496634,
      "grad_norm": 0.4957656264305115,
      "learning_rate": 7.838394942971005e-06,
      "loss": 0.6996,
      "step": 4431
    },
    {
      "epoch": 1.2175525339925835,
      "grad_norm": 0.4816265106201172,
      "learning_rate": 7.835646557647382e-06,
      "loss": 0.8179,
      "step": 4432
    },
    {
      "epoch": 1.2178272215355033,
      "grad_norm": 0.4076264500617981,
      "learning_rate": 7.832898172323761e-06,
      "loss": 0.868,
      "step": 4433
    },
    {
      "epoch": 1.2181019090784233,
      "grad_norm": 0.33592063188552856,
      "learning_rate": 7.830149787000138e-06,
      "loss": 0.8263,
      "step": 4434
    },
    {
      "epoch": 1.2183765966213431,
      "grad_norm": 0.4221339225769043,
      "learning_rate": 7.827401401676515e-06,
      "loss": 0.7383,
      "step": 4435
    },
    {
      "epoch": 1.2186512841642632,
      "grad_norm": 0.5709002614021301,
      "learning_rate": 7.824653016352894e-06,
      "loss": 0.623,
      "step": 4436
    },
    {
      "epoch": 1.2189259717071832,
      "grad_norm": 0.4883991777896881,
      "learning_rate": 7.821904631029271e-06,
      "loss": 0.7163,
      "step": 4437
    },
    {
      "epoch": 1.219200659250103,
      "grad_norm": 0.4021482765674591,
      "learning_rate": 7.819156245705648e-06,
      "loss": 0.9748,
      "step": 4438
    },
    {
      "epoch": 1.2194753467930228,
      "grad_norm": 0.5068091750144958,
      "learning_rate": 7.816407860382027e-06,
      "loss": 0.883,
      "step": 4439
    },
    {
      "epoch": 1.2197500343359429,
      "grad_norm": 0.6168105006217957,
      "learning_rate": 7.813659475058404e-06,
      "loss": 0.9452,
      "step": 4440
    },
    {
      "epoch": 1.220024721878863,
      "grad_norm": 0.5632764101028442,
      "learning_rate": 7.810911089734782e-06,
      "loss": 0.6916,
      "step": 4441
    },
    {
      "epoch": 1.2202994094217827,
      "grad_norm": 0.32559913396835327,
      "learning_rate": 7.808162704411158e-06,
      "loss": 0.762,
      "step": 4442
    },
    {
      "epoch": 1.2205740969647025,
      "grad_norm": 0.37739744782447815,
      "learning_rate": 7.805414319087536e-06,
      "loss": 0.5891,
      "step": 4443
    },
    {
      "epoch": 1.2208487845076226,
      "grad_norm": 0.6183237433433533,
      "learning_rate": 7.802665933763915e-06,
      "loss": 0.6079,
      "step": 4444
    },
    {
      "epoch": 1.2211234720505426,
      "grad_norm": 0.4653232991695404,
      "learning_rate": 7.799917548440292e-06,
      "loss": 0.7254,
      "step": 4445
    },
    {
      "epoch": 1.2213981595934624,
      "grad_norm": 0.3400241434574127,
      "learning_rate": 7.79716916311667e-06,
      "loss": 0.5675,
      "step": 4446
    },
    {
      "epoch": 1.2216728471363822,
      "grad_norm": 0.40220633149147034,
      "learning_rate": 7.794420777793046e-06,
      "loss": 0.8484,
      "step": 4447
    },
    {
      "epoch": 1.2219475346793023,
      "grad_norm": 0.37519314885139465,
      "learning_rate": 7.791672392469425e-06,
      "loss": 0.9969,
      "step": 4448
    },
    {
      "epoch": 1.2222222222222223,
      "grad_norm": 0.36526116728782654,
      "learning_rate": 7.788924007145802e-06,
      "loss": 0.7448,
      "step": 4449
    },
    {
      "epoch": 1.2224969097651421,
      "grad_norm": 0.4168228507041931,
      "learning_rate": 7.786175621822179e-06,
      "loss": 0.9387,
      "step": 4450
    },
    {
      "epoch": 1.2227715973080622,
      "grad_norm": 0.33881962299346924,
      "learning_rate": 7.783427236498558e-06,
      "loss": 0.7099,
      "step": 4451
    },
    {
      "epoch": 1.223046284850982,
      "grad_norm": 0.5817686915397644,
      "learning_rate": 7.780678851174935e-06,
      "loss": 0.6918,
      "step": 4452
    },
    {
      "epoch": 1.223320972393902,
      "grad_norm": 0.375512033700943,
      "learning_rate": 7.777930465851313e-06,
      "loss": 0.9107,
      "step": 4453
    },
    {
      "epoch": 1.2235956599368218,
      "grad_norm": 0.47743937373161316,
      "learning_rate": 7.77518208052769e-06,
      "loss": 0.7646,
      "step": 4454
    },
    {
      "epoch": 1.2238703474797419,
      "grad_norm": 0.4286329746246338,
      "learning_rate": 7.772433695204067e-06,
      "loss": 0.7891,
      "step": 4455
    },
    {
      "epoch": 1.2241450350226617,
      "grad_norm": 0.5087169408798218,
      "learning_rate": 7.769685309880446e-06,
      "loss": 0.7273,
      "step": 4456
    },
    {
      "epoch": 1.2244197225655817,
      "grad_norm": 0.38509631156921387,
      "learning_rate": 7.766936924556823e-06,
      "loss": 0.8675,
      "step": 4457
    },
    {
      "epoch": 1.2246944101085016,
      "grad_norm": 0.40424373745918274,
      "learning_rate": 7.7641885392332e-06,
      "loss": 0.6415,
      "step": 4458
    },
    {
      "epoch": 1.2249690976514216,
      "grad_norm": 0.47004565596580505,
      "learning_rate": 7.761440153909579e-06,
      "loss": 0.6339,
      "step": 4459
    },
    {
      "epoch": 1.2252437851943414,
      "grad_norm": 0.46175718307495117,
      "learning_rate": 7.758691768585956e-06,
      "loss": 0.7722,
      "step": 4460
    },
    {
      "epoch": 1.2255184727372614,
      "grad_norm": 0.5548505783081055,
      "learning_rate": 7.755943383262335e-06,
      "loss": 0.8666,
      "step": 4461
    },
    {
      "epoch": 1.2257931602801813,
      "grad_norm": 0.441882461309433,
      "learning_rate": 7.753194997938712e-06,
      "loss": 0.9,
      "step": 4462
    },
    {
      "epoch": 1.2260678478231013,
      "grad_norm": 0.5441725850105286,
      "learning_rate": 7.750446612615089e-06,
      "loss": 0.7878,
      "step": 4463
    },
    {
      "epoch": 1.2263425353660211,
      "grad_norm": 0.3717654049396515,
      "learning_rate": 7.747698227291467e-06,
      "loss": 0.9715,
      "step": 4464
    },
    {
      "epoch": 1.2266172229089412,
      "grad_norm": 0.429291307926178,
      "learning_rate": 7.744949841967844e-06,
      "loss": 0.7608,
      "step": 4465
    },
    {
      "epoch": 1.226891910451861,
      "grad_norm": 0.4545179009437561,
      "learning_rate": 7.742201456644221e-06,
      "loss": 0.8528,
      "step": 4466
    },
    {
      "epoch": 1.227166597994781,
      "grad_norm": 0.3419816792011261,
      "learning_rate": 7.7394530713206e-06,
      "loss": 1.0021,
      "step": 4467
    },
    {
      "epoch": 1.2274412855377008,
      "grad_norm": 0.4695431888103485,
      "learning_rate": 7.736704685996977e-06,
      "loss": 0.8162,
      "step": 4468
    },
    {
      "epoch": 1.2277159730806209,
      "grad_norm": 0.42033812403678894,
      "learning_rate": 7.733956300673356e-06,
      "loss": 0.9427,
      "step": 4469
    },
    {
      "epoch": 1.2279906606235407,
      "grad_norm": 0.3829508125782013,
      "learning_rate": 7.731207915349733e-06,
      "loss": 1.0332,
      "step": 4470
    },
    {
      "epoch": 1.2282653481664607,
      "grad_norm": 0.5302209258079529,
      "learning_rate": 7.72845953002611e-06,
      "loss": 0.9659,
      "step": 4471
    },
    {
      "epoch": 1.2285400357093805,
      "grad_norm": 0.35888805985450745,
      "learning_rate": 7.725711144702489e-06,
      "loss": 0.6919,
      "step": 4472
    },
    {
      "epoch": 1.2288147232523006,
      "grad_norm": 0.5640972852706909,
      "learning_rate": 7.722962759378866e-06,
      "loss": 0.8272,
      "step": 4473
    },
    {
      "epoch": 1.2290894107952204,
      "grad_norm": 0.46543797850608826,
      "learning_rate": 7.720214374055243e-06,
      "loss": 0.5712,
      "step": 4474
    },
    {
      "epoch": 1.2293640983381404,
      "grad_norm": 0.6132771372795105,
      "learning_rate": 7.71746598873162e-06,
      "loss": 0.78,
      "step": 4475
    },
    {
      "epoch": 1.2296387858810602,
      "grad_norm": 0.4273255169391632,
      "learning_rate": 7.714717603407998e-06,
      "loss": 0.7612,
      "step": 4476
    },
    {
      "epoch": 1.2299134734239803,
      "grad_norm": 0.5622982382774353,
      "learning_rate": 7.711969218084377e-06,
      "loss": 0.8966,
      "step": 4477
    },
    {
      "epoch": 1.2301881609669,
      "grad_norm": 0.5771257877349854,
      "learning_rate": 7.709220832760752e-06,
      "loss": 0.7739,
      "step": 4478
    },
    {
      "epoch": 1.2304628485098201,
      "grad_norm": 0.3980194628238678,
      "learning_rate": 7.706472447437131e-06,
      "loss": 0.5271,
      "step": 4479
    },
    {
      "epoch": 1.23073753605274,
      "grad_norm": 0.4557589292526245,
      "learning_rate": 7.703724062113508e-06,
      "loss": 0.661,
      "step": 4480
    },
    {
      "epoch": 1.23101222359566,
      "grad_norm": 0.690471887588501,
      "learning_rate": 7.700975676789887e-06,
      "loss": 0.7291,
      "step": 4481
    },
    {
      "epoch": 1.2312869111385798,
      "grad_norm": 0.45587611198425293,
      "learning_rate": 7.698227291466264e-06,
      "loss": 0.7331,
      "step": 4482
    },
    {
      "epoch": 1.2315615986814998,
      "grad_norm": 0.5862184166908264,
      "learning_rate": 7.695478906142641e-06,
      "loss": 0.879,
      "step": 4483
    },
    {
      "epoch": 1.2318362862244197,
      "grad_norm": 0.40048497915267944,
      "learning_rate": 7.69273052081902e-06,
      "loss": 0.7183,
      "step": 4484
    },
    {
      "epoch": 1.2321109737673397,
      "grad_norm": 0.420218825340271,
      "learning_rate": 7.689982135495397e-06,
      "loss": 0.7076,
      "step": 4485
    },
    {
      "epoch": 1.2323856613102595,
      "grad_norm": 0.40106019377708435,
      "learning_rate": 7.687233750171774e-06,
      "loss": 1.0025,
      "step": 4486
    },
    {
      "epoch": 1.2326603488531795,
      "grad_norm": 0.5470634698867798,
      "learning_rate": 7.684485364848152e-06,
      "loss": 0.6844,
      "step": 4487
    },
    {
      "epoch": 1.2329350363960994,
      "grad_norm": 0.400137722492218,
      "learning_rate": 7.68173697952453e-06,
      "loss": 0.7186,
      "step": 4488
    },
    {
      "epoch": 1.2332097239390194,
      "grad_norm": 0.515718936920166,
      "learning_rate": 7.678988594200908e-06,
      "loss": 0.9405,
      "step": 4489
    },
    {
      "epoch": 1.2334844114819392,
      "grad_norm": 0.3478454351425171,
      "learning_rate": 7.676240208877285e-06,
      "loss": 0.7998,
      "step": 4490
    },
    {
      "epoch": 1.2337590990248593,
      "grad_norm": 0.39497464895248413,
      "learning_rate": 7.673491823553662e-06,
      "loss": 0.9293,
      "step": 4491
    },
    {
      "epoch": 1.234033786567779,
      "grad_norm": 0.47236168384552,
      "learning_rate": 7.670743438230041e-06,
      "loss": 0.8085,
      "step": 4492
    },
    {
      "epoch": 1.234308474110699,
      "grad_norm": 0.4436910152435303,
      "learning_rate": 7.667995052906418e-06,
      "loss": 0.7003,
      "step": 4493
    },
    {
      "epoch": 1.234583161653619,
      "grad_norm": 0.34086260199546814,
      "learning_rate": 7.665246667582795e-06,
      "loss": 0.5957,
      "step": 4494
    },
    {
      "epoch": 1.234857849196539,
      "grad_norm": 0.5625371932983398,
      "learning_rate": 7.662498282259174e-06,
      "loss": 1.0249,
      "step": 4495
    },
    {
      "epoch": 1.2351325367394588,
      "grad_norm": 0.487893283367157,
      "learning_rate": 7.65974989693555e-06,
      "loss": 0.7683,
      "step": 4496
    },
    {
      "epoch": 1.2354072242823788,
      "grad_norm": 0.5933549404144287,
      "learning_rate": 7.65700151161193e-06,
      "loss": 0.7696,
      "step": 4497
    },
    {
      "epoch": 1.2356819118252986,
      "grad_norm": 0.3512812554836273,
      "learning_rate": 7.654253126288306e-06,
      "loss": 0.7029,
      "step": 4498
    },
    {
      "epoch": 1.2359565993682187,
      "grad_norm": 0.5662689208984375,
      "learning_rate": 7.651504740964683e-06,
      "loss": 0.8999,
      "step": 4499
    },
    {
      "epoch": 1.2362312869111385,
      "grad_norm": 0.3580576777458191,
      "learning_rate": 7.648756355641062e-06,
      "loss": 0.5682,
      "step": 4500
    },
    {
      "epoch": 1.2365059744540585,
      "grad_norm": 0.4707838296890259,
      "learning_rate": 7.64600797031744e-06,
      "loss": 0.7936,
      "step": 4501
    },
    {
      "epoch": 1.2367806619969786,
      "grad_norm": 0.34420347213745117,
      "learning_rate": 7.643259584993816e-06,
      "loss": 0.8506,
      "step": 4502
    },
    {
      "epoch": 1.2370553495398984,
      "grad_norm": 0.48294371366500854,
      "learning_rate": 7.640511199670195e-06,
      "loss": 0.8366,
      "step": 4503
    },
    {
      "epoch": 1.2373300370828182,
      "grad_norm": 0.4135560691356659,
      "learning_rate": 7.637762814346572e-06,
      "loss": 0.8714,
      "step": 4504
    },
    {
      "epoch": 1.2376047246257382,
      "grad_norm": 0.39106059074401855,
      "learning_rate": 7.63501442902295e-06,
      "loss": 0.6617,
      "step": 4505
    },
    {
      "epoch": 1.2378794121686583,
      "grad_norm": 0.38744550943374634,
      "learning_rate": 7.632266043699326e-06,
      "loss": 0.8827,
      "step": 4506
    },
    {
      "epoch": 1.238154099711578,
      "grad_norm": 0.31478193402290344,
      "learning_rate": 7.629517658375705e-06,
      "loss": 0.8371,
      "step": 4507
    },
    {
      "epoch": 1.238428787254498,
      "grad_norm": 0.5821689367294312,
      "learning_rate": 7.626769273052083e-06,
      "loss": 0.7724,
      "step": 4508
    },
    {
      "epoch": 1.238703474797418,
      "grad_norm": 0.4225611984729767,
      "learning_rate": 7.6240208877284604e-06,
      "loss": 0.9789,
      "step": 4509
    },
    {
      "epoch": 1.238978162340338,
      "grad_norm": 0.6864148378372192,
      "learning_rate": 7.6212725024048375e-06,
      "loss": 0.7871,
      "step": 4510
    },
    {
      "epoch": 1.2392528498832578,
      "grad_norm": 0.5559916496276855,
      "learning_rate": 7.618524117081215e-06,
      "loss": 0.7298,
      "step": 4511
    },
    {
      "epoch": 1.2395275374261776,
      "grad_norm": 0.44815897941589355,
      "learning_rate": 7.615775731757593e-06,
      "loss": 0.7601,
      "step": 4512
    },
    {
      "epoch": 1.2398022249690976,
      "grad_norm": 0.3349950611591339,
      "learning_rate": 7.613027346433971e-06,
      "loss": 0.6693,
      "step": 4513
    },
    {
      "epoch": 1.2400769125120177,
      "grad_norm": 0.34492114186286926,
      "learning_rate": 7.610278961110348e-06,
      "loss": 0.6969,
      "step": 4514
    },
    {
      "epoch": 1.2403516000549375,
      "grad_norm": 0.40952402353286743,
      "learning_rate": 7.607530575786726e-06,
      "loss": 0.7327,
      "step": 4515
    },
    {
      "epoch": 1.2406262875978573,
      "grad_norm": 0.3547569811344147,
      "learning_rate": 7.604782190463104e-06,
      "loss": 0.853,
      "step": 4516
    },
    {
      "epoch": 1.2409009751407774,
      "grad_norm": 0.41641899943351746,
      "learning_rate": 7.602033805139482e-06,
      "loss": 0.7871,
      "step": 4517
    },
    {
      "epoch": 1.2411756626836974,
      "grad_norm": 0.46452394127845764,
      "learning_rate": 7.599285419815859e-06,
      "loss": 0.7563,
      "step": 4518
    },
    {
      "epoch": 1.2414503502266172,
      "grad_norm": 0.4704127013683319,
      "learning_rate": 7.596537034492237e-06,
      "loss": 0.8444,
      "step": 4519
    },
    {
      "epoch": 1.2417250377695372,
      "grad_norm": 0.5887008905410767,
      "learning_rate": 7.5937886491686145e-06,
      "loss": 0.7282,
      "step": 4520
    },
    {
      "epoch": 1.241999725312457,
      "grad_norm": 0.4017654359340668,
      "learning_rate": 7.591040263844992e-06,
      "loss": 0.6515,
      "step": 4521
    },
    {
      "epoch": 1.242274412855377,
      "grad_norm": 0.4587480425834656,
      "learning_rate": 7.5882918785213685e-06,
      "loss": 0.9273,
      "step": 4522
    },
    {
      "epoch": 1.242549100398297,
      "grad_norm": 0.4142567813396454,
      "learning_rate": 7.585543493197746e-06,
      "loss": 0.6743,
      "step": 4523
    },
    {
      "epoch": 1.242823787941217,
      "grad_norm": 0.4128339886665344,
      "learning_rate": 7.582795107874124e-06,
      "loss": 0.9471,
      "step": 4524
    },
    {
      "epoch": 1.2430984754841368,
      "grad_norm": 0.34725716710090637,
      "learning_rate": 7.580046722550503e-06,
      "loss": 0.6953,
      "step": 4525
    },
    {
      "epoch": 1.2433731630270568,
      "grad_norm": 1.4469400644302368,
      "learning_rate": 7.577298337226879e-06,
      "loss": 1.0532,
      "step": 4526
    },
    {
      "epoch": 1.2436478505699766,
      "grad_norm": 0.3826884329319,
      "learning_rate": 7.574549951903257e-06,
      "loss": 0.8263,
      "step": 4527
    },
    {
      "epoch": 1.2439225381128967,
      "grad_norm": 0.3978562653064728,
      "learning_rate": 7.571801566579635e-06,
      "loss": 0.6582,
      "step": 4528
    },
    {
      "epoch": 1.2441972256558165,
      "grad_norm": 0.4641948938369751,
      "learning_rate": 7.569053181256013e-06,
      "loss": 0.7252,
      "step": 4529
    },
    {
      "epoch": 1.2444719131987365,
      "grad_norm": 0.3392391800880432,
      "learning_rate": 7.56630479593239e-06,
      "loss": 0.7968,
      "step": 4530
    },
    {
      "epoch": 1.2447466007416563,
      "grad_norm": 0.3889738619327545,
      "learning_rate": 7.563556410608768e-06,
      "loss": 0.8296,
      "step": 4531
    },
    {
      "epoch": 1.2450212882845764,
      "grad_norm": 0.3996177911758423,
      "learning_rate": 7.5608080252851455e-06,
      "loss": 0.866,
      "step": 4532
    },
    {
      "epoch": 1.2452959758274962,
      "grad_norm": 0.4664579927921295,
      "learning_rate": 7.558059639961523e-06,
      "loss": 0.7179,
      "step": 4533
    },
    {
      "epoch": 1.2455706633704162,
      "grad_norm": 0.41135188937187195,
      "learning_rate": 7.5553112546379e-06,
      "loss": 0.688,
      "step": 4534
    },
    {
      "epoch": 1.245845350913336,
      "grad_norm": 0.4892445504665375,
      "learning_rate": 7.552562869314278e-06,
      "loss": 0.8912,
      "step": 4535
    },
    {
      "epoch": 1.246120038456256,
      "grad_norm": 0.41815367341041565,
      "learning_rate": 7.549814483990656e-06,
      "loss": 0.6775,
      "step": 4536
    },
    {
      "epoch": 1.246394725999176,
      "grad_norm": 0.46289366483688354,
      "learning_rate": 7.547066098667034e-06,
      "loss": 0.5992,
      "step": 4537
    },
    {
      "epoch": 1.246669413542096,
      "grad_norm": 0.5388553738594055,
      "learning_rate": 7.544317713343411e-06,
      "loss": 0.617,
      "step": 4538
    },
    {
      "epoch": 1.2469441010850157,
      "grad_norm": 0.4903644621372223,
      "learning_rate": 7.541569328019789e-06,
      "loss": 0.6216,
      "step": 4539
    },
    {
      "epoch": 1.2472187886279358,
      "grad_norm": 0.33209434151649475,
      "learning_rate": 7.538820942696167e-06,
      "loss": 0.6387,
      "step": 4540
    },
    {
      "epoch": 1.2474934761708556,
      "grad_norm": 0.3650531470775604,
      "learning_rate": 7.536072557372545e-06,
      "loss": 1.0081,
      "step": 4541
    },
    {
      "epoch": 1.2477681637137756,
      "grad_norm": 0.4558880925178528,
      "learning_rate": 7.533324172048922e-06,
      "loss": 0.7119,
      "step": 4542
    },
    {
      "epoch": 1.2480428512566955,
      "grad_norm": 0.48822838068008423,
      "learning_rate": 7.5305757867252995e-06,
      "loss": 0.6935,
      "step": 4543
    },
    {
      "epoch": 1.2483175387996155,
      "grad_norm": 0.30679503083229065,
      "learning_rate": 7.527827401401677e-06,
      "loss": 0.3971,
      "step": 4544
    },
    {
      "epoch": 1.2485922263425353,
      "grad_norm": 0.5207659602165222,
      "learning_rate": 7.525079016078055e-06,
      "loss": 0.9058,
      "step": 4545
    },
    {
      "epoch": 1.2488669138854553,
      "grad_norm": 0.4541795551776886,
      "learning_rate": 7.522330630754432e-06,
      "loss": 0.9222,
      "step": 4546
    },
    {
      "epoch": 1.2491416014283752,
      "grad_norm": 0.3855384886264801,
      "learning_rate": 7.51958224543081e-06,
      "loss": 0.7236,
      "step": 4547
    },
    {
      "epoch": 1.2494162889712952,
      "grad_norm": 0.3742593824863434,
      "learning_rate": 7.516833860107188e-06,
      "loss": 0.7798,
      "step": 4548
    },
    {
      "epoch": 1.249690976514215,
      "grad_norm": 0.43392252922058105,
      "learning_rate": 7.514085474783566e-06,
      "loss": 0.6774,
      "step": 4549
    },
    {
      "epoch": 1.249965664057135,
      "grad_norm": 0.5120749473571777,
      "learning_rate": 7.511337089459943e-06,
      "loss": 0.7175,
      "step": 4550
    },
    {
      "epoch": 1.2502403516000549,
      "grad_norm": 0.412675142288208,
      "learning_rate": 7.508588704136321e-06,
      "loss": 0.9203,
      "step": 4551
    },
    {
      "epoch": 1.250515039142975,
      "grad_norm": 0.44011470675468445,
      "learning_rate": 7.505840318812699e-06,
      "loss": 0.764,
      "step": 4552
    },
    {
      "epoch": 1.2507897266858947,
      "grad_norm": 0.407272607088089,
      "learning_rate": 7.5030919334890765e-06,
      "loss": 0.932,
      "step": 4553
    },
    {
      "epoch": 1.2510644142288148,
      "grad_norm": 0.3997441828250885,
      "learning_rate": 7.500343548165453e-06,
      "loss": 0.9667,
      "step": 4554
    },
    {
      "epoch": 1.2513391017717346,
      "grad_norm": 0.3597617447376251,
      "learning_rate": 7.4975951628418305e-06,
      "loss": 0.95,
      "step": 4555
    },
    {
      "epoch": 1.2516137893146546,
      "grad_norm": 0.33781278133392334,
      "learning_rate": 7.494846777518208e-06,
      "loss": 0.7209,
      "step": 4556
    },
    {
      "epoch": 1.2518884768575744,
      "grad_norm": 0.4565044641494751,
      "learning_rate": 7.492098392194587e-06,
      "loss": 0.6248,
      "step": 4557
    },
    {
      "epoch": 1.2521631644004945,
      "grad_norm": 0.5202114582061768,
      "learning_rate": 7.489350006870963e-06,
      "loss": 0.7484,
      "step": 4558
    },
    {
      "epoch": 1.2524378519434145,
      "grad_norm": 0.38143521547317505,
      "learning_rate": 7.486601621547341e-06,
      "loss": 0.5947,
      "step": 4559
    },
    {
      "epoch": 1.2527125394863343,
      "grad_norm": 0.340680330991745,
      "learning_rate": 7.483853236223719e-06,
      "loss": 0.8842,
      "step": 4560
    },
    {
      "epoch": 1.2529872270292541,
      "grad_norm": 0.46039512753486633,
      "learning_rate": 7.481104850900097e-06,
      "loss": 0.7852,
      "step": 4561
    },
    {
      "epoch": 1.2532619145721742,
      "grad_norm": 0.40246376395225525,
      "learning_rate": 7.478356465576474e-06,
      "loss": 0.6746,
      "step": 4562
    },
    {
      "epoch": 1.2535366021150942,
      "grad_norm": 0.40966957807540894,
      "learning_rate": 7.475608080252852e-06,
      "loss": 0.8888,
      "step": 4563
    },
    {
      "epoch": 1.253811289658014,
      "grad_norm": 0.33555421233177185,
      "learning_rate": 7.47285969492923e-06,
      "loss": 0.7852,
      "step": 4564
    },
    {
      "epoch": 1.2540859772009338,
      "grad_norm": 0.3863696753978729,
      "learning_rate": 7.4701113096056075e-06,
      "loss": 0.7569,
      "step": 4565
    },
    {
      "epoch": 1.2543606647438539,
      "grad_norm": 0.40175729990005493,
      "learning_rate": 7.4673629242819845e-06,
      "loss": 0.8271,
      "step": 4566
    },
    {
      "epoch": 1.254635352286774,
      "grad_norm": 0.5109983682632446,
      "learning_rate": 7.464614538958362e-06,
      "loss": 0.9234,
      "step": 4567
    },
    {
      "epoch": 1.2549100398296937,
      "grad_norm": 0.4000880718231201,
      "learning_rate": 7.46186615363474e-06,
      "loss": 0.4795,
      "step": 4568
    },
    {
      "epoch": 1.2551847273726136,
      "grad_norm": 0.43118706345558167,
      "learning_rate": 7.459117768311118e-06,
      "loss": 0.6455,
      "step": 4569
    },
    {
      "epoch": 1.2554594149155336,
      "grad_norm": 0.5026120543479919,
      "learning_rate": 7.456369382987495e-06,
      "loss": 0.6153,
      "step": 4570
    },
    {
      "epoch": 1.2557341024584536,
      "grad_norm": 0.4500002861022949,
      "learning_rate": 7.453620997663873e-06,
      "loss": 0.8647,
      "step": 4571
    },
    {
      "epoch": 1.2560087900013734,
      "grad_norm": 0.39576756954193115,
      "learning_rate": 7.450872612340251e-06,
      "loss": 0.5995,
      "step": 4572
    },
    {
      "epoch": 1.2562834775442933,
      "grad_norm": 0.3618657886981964,
      "learning_rate": 7.448124227016629e-06,
      "loss": 0.7505,
      "step": 4573
    },
    {
      "epoch": 1.2565581650872133,
      "grad_norm": 0.48922139406204224,
      "learning_rate": 7.445375841693006e-06,
      "loss": 0.7294,
      "step": 4574
    },
    {
      "epoch": 1.2568328526301333,
      "grad_norm": 0.4580352306365967,
      "learning_rate": 7.442627456369384e-06,
      "loss": 0.8412,
      "step": 4575
    },
    {
      "epoch": 1.2571075401730532,
      "grad_norm": 0.5801017880439758,
      "learning_rate": 7.4398790710457615e-06,
      "loss": 0.7394,
      "step": 4576
    },
    {
      "epoch": 1.257382227715973,
      "grad_norm": 0.4098374843597412,
      "learning_rate": 7.437130685722139e-06,
      "loss": 0.8366,
      "step": 4577
    },
    {
      "epoch": 1.257656915258893,
      "grad_norm": 0.4948083758354187,
      "learning_rate": 7.434382300398516e-06,
      "loss": 0.7984,
      "step": 4578
    },
    {
      "epoch": 1.257931602801813,
      "grad_norm": 0.3257092833518982,
      "learning_rate": 7.431633915074894e-06,
      "loss": 0.5571,
      "step": 4579
    },
    {
      "epoch": 1.2582062903447329,
      "grad_norm": 0.41422584652900696,
      "learning_rate": 7.428885529751272e-06,
      "loss": 0.6477,
      "step": 4580
    },
    {
      "epoch": 1.2584809778876527,
      "grad_norm": 0.5215250849723816,
      "learning_rate": 7.42613714442765e-06,
      "loss": 0.6239,
      "step": 4581
    },
    {
      "epoch": 1.2587556654305727,
      "grad_norm": 0.3660622239112854,
      "learning_rate": 7.423388759104026e-06,
      "loss": 0.8529,
      "step": 4582
    },
    {
      "epoch": 1.2590303529734928,
      "grad_norm": 0.5556186437606812,
      "learning_rate": 7.420640373780405e-06,
      "loss": 0.8827,
      "step": 4583
    },
    {
      "epoch": 1.2593050405164126,
      "grad_norm": 0.5773485898971558,
      "learning_rate": 7.417891988456783e-06,
      "loss": 0.8113,
      "step": 4584
    },
    {
      "epoch": 1.2595797280593324,
      "grad_norm": 0.4794907867908478,
      "learning_rate": 7.415143603133161e-06,
      "loss": 0.7207,
      "step": 4585
    },
    {
      "epoch": 1.2598544156022524,
      "grad_norm": 0.38972270488739014,
      "learning_rate": 7.412395217809537e-06,
      "loss": 0.8794,
      "step": 4586
    },
    {
      "epoch": 1.2601291031451725,
      "grad_norm": 0.4801022708415985,
      "learning_rate": 7.409646832485915e-06,
      "loss": 0.7979,
      "step": 4587
    },
    {
      "epoch": 1.2604037906880923,
      "grad_norm": 0.44079405069351196,
      "learning_rate": 7.4068984471622925e-06,
      "loss": 0.6703,
      "step": 4588
    },
    {
      "epoch": 1.260678478231012,
      "grad_norm": 0.5065140128135681,
      "learning_rate": 7.40415006183867e-06,
      "loss": 0.6585,
      "step": 4589
    },
    {
      "epoch": 1.2609531657739321,
      "grad_norm": 0.4343850612640381,
      "learning_rate": 7.401401676515047e-06,
      "loss": 1.0108,
      "step": 4590
    },
    {
      "epoch": 1.2612278533168522,
      "grad_norm": 0.5314486622810364,
      "learning_rate": 7.398653291191425e-06,
      "loss": 0.6723,
      "step": 4591
    },
    {
      "epoch": 1.261502540859772,
      "grad_norm": 0.4440940022468567,
      "learning_rate": 7.395904905867803e-06,
      "loss": 0.8377,
      "step": 4592
    },
    {
      "epoch": 1.261777228402692,
      "grad_norm": 0.4359395205974579,
      "learning_rate": 7.393156520544181e-06,
      "loss": 0.8751,
      "step": 4593
    },
    {
      "epoch": 1.2620519159456118,
      "grad_norm": 0.5181637406349182,
      "learning_rate": 7.390408135220558e-06,
      "loss": 0.7889,
      "step": 4594
    },
    {
      "epoch": 1.2623266034885319,
      "grad_norm": 0.3689740002155304,
      "learning_rate": 7.387659749896936e-06,
      "loss": 0.4995,
      "step": 4595
    },
    {
      "epoch": 1.2626012910314517,
      "grad_norm": 0.3349186182022095,
      "learning_rate": 7.384911364573314e-06,
      "loss": 0.4969,
      "step": 4596
    },
    {
      "epoch": 1.2628759785743717,
      "grad_norm": 0.42664864659309387,
      "learning_rate": 7.382162979249692e-06,
      "loss": 0.8045,
      "step": 4597
    },
    {
      "epoch": 1.2631506661172915,
      "grad_norm": 0.3660520017147064,
      "learning_rate": 7.379414593926069e-06,
      "loss": 1.0922,
      "step": 4598
    },
    {
      "epoch": 1.2634253536602116,
      "grad_norm": 0.5452141761779785,
      "learning_rate": 7.3766662086024465e-06,
      "loss": 0.8556,
      "step": 4599
    },
    {
      "epoch": 1.2637000412031314,
      "grad_norm": 0.4163968861103058,
      "learning_rate": 7.373917823278824e-06,
      "loss": 0.5905,
      "step": 4600
    },
    {
      "epoch": 1.2639747287460514,
      "grad_norm": 0.3962389826774597,
      "learning_rate": 7.371169437955202e-06,
      "loss": 0.7292,
      "step": 4601
    },
    {
      "epoch": 1.2642494162889713,
      "grad_norm": 0.5856627821922302,
      "learning_rate": 7.368421052631579e-06,
      "loss": 1.0087,
      "step": 4602
    },
    {
      "epoch": 1.2645241038318913,
      "grad_norm": 0.48130765557289124,
      "learning_rate": 7.365672667307957e-06,
      "loss": 0.9983,
      "step": 4603
    },
    {
      "epoch": 1.264798791374811,
      "grad_norm": 0.4049580693244934,
      "learning_rate": 7.362924281984335e-06,
      "loss": 0.7825,
      "step": 4604
    },
    {
      "epoch": 1.2650734789177311,
      "grad_norm": 0.4788823425769806,
      "learning_rate": 7.360175896660713e-06,
      "loss": 0.8365,
      "step": 4605
    },
    {
      "epoch": 1.265348166460651,
      "grad_norm": 0.5439493656158447,
      "learning_rate": 7.35742751133709e-06,
      "loss": 0.7222,
      "step": 4606
    },
    {
      "epoch": 1.265622854003571,
      "grad_norm": 0.53118896484375,
      "learning_rate": 7.354679126013468e-06,
      "loss": 0.7306,
      "step": 4607
    },
    {
      "epoch": 1.2658975415464908,
      "grad_norm": 0.38613787293434143,
      "learning_rate": 7.351930740689846e-06,
      "loss": 0.6597,
      "step": 4608
    },
    {
      "epoch": 1.2661722290894109,
      "grad_norm": 0.3705192506313324,
      "learning_rate": 7.3491823553662235e-06,
      "loss": 0.7557,
      "step": 4609
    },
    {
      "epoch": 1.2664469166323307,
      "grad_norm": 0.38684454560279846,
      "learning_rate": 7.3464339700426005e-06,
      "loss": 0.7976,
      "step": 4610
    },
    {
      "epoch": 1.2667216041752507,
      "grad_norm": 0.38896897435188293,
      "learning_rate": 7.343685584718978e-06,
      "loss": 0.7487,
      "step": 4611
    },
    {
      "epoch": 1.2669962917181705,
      "grad_norm": 0.44071510434150696,
      "learning_rate": 7.340937199395356e-06,
      "loss": 0.6604,
      "step": 4612
    },
    {
      "epoch": 1.2672709792610906,
      "grad_norm": 0.4798683226108551,
      "learning_rate": 7.338188814071734e-06,
      "loss": 0.6678,
      "step": 4613
    },
    {
      "epoch": 1.2675456668040104,
      "grad_norm": 0.5353296995162964,
      "learning_rate": 7.33544042874811e-06,
      "loss": 0.9834,
      "step": 4614
    },
    {
      "epoch": 1.2678203543469304,
      "grad_norm": 0.40926092863082886,
      "learning_rate": 7.332692043424488e-06,
      "loss": 0.653,
      "step": 4615
    },
    {
      "epoch": 1.2680950418898502,
      "grad_norm": 0.3177875280380249,
      "learning_rate": 7.329943658100867e-06,
      "loss": 0.749,
      "step": 4616
    },
    {
      "epoch": 1.2683697294327703,
      "grad_norm": 0.35986271500587463,
      "learning_rate": 7.327195272777245e-06,
      "loss": 0.5954,
      "step": 4617
    },
    {
      "epoch": 1.26864441697569,
      "grad_norm": 0.45648837089538574,
      "learning_rate": 7.324446887453621e-06,
      "loss": 0.6996,
      "step": 4618
    },
    {
      "epoch": 1.2689191045186101,
      "grad_norm": 0.6034810543060303,
      "learning_rate": 7.321698502129999e-06,
      "loss": 0.8063,
      "step": 4619
    },
    {
      "epoch": 1.26919379206153,
      "grad_norm": 0.5033732056617737,
      "learning_rate": 7.318950116806377e-06,
      "loss": 0.6426,
      "step": 4620
    },
    {
      "epoch": 1.26946847960445,
      "grad_norm": 0.5288159847259521,
      "learning_rate": 7.3162017314827545e-06,
      "loss": 0.8862,
      "step": 4621
    },
    {
      "epoch": 1.2697431671473698,
      "grad_norm": 0.3282407522201538,
      "learning_rate": 7.3134533461591316e-06,
      "loss": 0.6881,
      "step": 4622
    },
    {
      "epoch": 1.2700178546902898,
      "grad_norm": 0.39590415358543396,
      "learning_rate": 7.3107049608355094e-06,
      "loss": 0.6135,
      "step": 4623
    },
    {
      "epoch": 1.2702925422332096,
      "grad_norm": 0.4740498960018158,
      "learning_rate": 7.307956575511887e-06,
      "loss": 0.6551,
      "step": 4624
    },
    {
      "epoch": 1.2705672297761297,
      "grad_norm": 0.43033960461616516,
      "learning_rate": 7.305208190188265e-06,
      "loss": 0.876,
      "step": 4625
    },
    {
      "epoch": 1.2708419173190495,
      "grad_norm": 0.4071500301361084,
      "learning_rate": 7.302459804864642e-06,
      "loss": 0.9046,
      "step": 4626
    },
    {
      "epoch": 1.2711166048619695,
      "grad_norm": 0.4316636025905609,
      "learning_rate": 7.29971141954102e-06,
      "loss": 0.6117,
      "step": 4627
    },
    {
      "epoch": 1.2713912924048896,
      "grad_norm": 0.40085211396217346,
      "learning_rate": 7.296963034217398e-06,
      "loss": 0.6692,
      "step": 4628
    },
    {
      "epoch": 1.2716659799478094,
      "grad_norm": 0.33670666813850403,
      "learning_rate": 7.294214648893776e-06,
      "loss": 0.7329,
      "step": 4629
    },
    {
      "epoch": 1.2719406674907292,
      "grad_norm": 0.39878320693969727,
      "learning_rate": 7.291466263570153e-06,
      "loss": 0.6054,
      "step": 4630
    },
    {
      "epoch": 1.2722153550336492,
      "grad_norm": 0.5569553971290588,
      "learning_rate": 7.288717878246531e-06,
      "loss": 0.72,
      "step": 4631
    },
    {
      "epoch": 1.2724900425765693,
      "grad_norm": 0.5175907015800476,
      "learning_rate": 7.2859694929229085e-06,
      "loss": 0.5838,
      "step": 4632
    },
    {
      "epoch": 1.272764730119489,
      "grad_norm": 0.5221782326698303,
      "learning_rate": 7.283221107599286e-06,
      "loss": 0.9262,
      "step": 4633
    },
    {
      "epoch": 1.273039417662409,
      "grad_norm": 0.3870822489261627,
      "learning_rate": 7.2804727222756634e-06,
      "loss": 0.9903,
      "step": 4634
    },
    {
      "epoch": 1.273314105205329,
      "grad_norm": 0.4987470805644989,
      "learning_rate": 7.277724336952041e-06,
      "loss": 0.8484,
      "step": 4635
    },
    {
      "epoch": 1.273588792748249,
      "grad_norm": 0.4460897445678711,
      "learning_rate": 7.274975951628419e-06,
      "loss": 0.7814,
      "step": 4636
    },
    {
      "epoch": 1.2738634802911688,
      "grad_norm": 0.36182868480682373,
      "learning_rate": 7.272227566304797e-06,
      "loss": 0.9541,
      "step": 4637
    },
    {
      "epoch": 1.2741381678340886,
      "grad_norm": 0.5427964925765991,
      "learning_rate": 7.269479180981174e-06,
      "loss": 0.8236,
      "step": 4638
    },
    {
      "epoch": 1.2744128553770087,
      "grad_norm": 0.3239089548587799,
      "learning_rate": 7.266730795657552e-06,
      "loss": 0.8721,
      "step": 4639
    },
    {
      "epoch": 1.2746875429199287,
      "grad_norm": 0.5683175325393677,
      "learning_rate": 7.26398241033393e-06,
      "loss": 0.6213,
      "step": 4640
    },
    {
      "epoch": 1.2749622304628485,
      "grad_norm": 0.3796648383140564,
      "learning_rate": 7.261234025010308e-06,
      "loss": 0.7778,
      "step": 4641
    },
    {
      "epoch": 1.2752369180057683,
      "grad_norm": 0.5058760643005371,
      "learning_rate": 7.258485639686685e-06,
      "loss": 0.7723,
      "step": 4642
    },
    {
      "epoch": 1.2755116055486884,
      "grad_norm": 0.34393638372421265,
      "learning_rate": 7.2557372543630625e-06,
      "loss": 0.7829,
      "step": 4643
    },
    {
      "epoch": 1.2757862930916084,
      "grad_norm": 0.3258192241191864,
      "learning_rate": 7.25298886903944e-06,
      "loss": 0.6912,
      "step": 4644
    },
    {
      "epoch": 1.2760609806345282,
      "grad_norm": 0.41808053851127625,
      "learning_rate": 7.250240483715818e-06,
      "loss": 0.8242,
      "step": 4645
    },
    {
      "epoch": 1.276335668177448,
      "grad_norm": 0.5221173763275146,
      "learning_rate": 7.2474920983921945e-06,
      "loss": 0.6803,
      "step": 4646
    },
    {
      "epoch": 1.276610355720368,
      "grad_norm": 0.5235178470611572,
      "learning_rate": 7.244743713068572e-06,
      "loss": 0.8738,
      "step": 4647
    },
    {
      "epoch": 1.276885043263288,
      "grad_norm": 0.3906387388706207,
      "learning_rate": 7.241995327744951e-06,
      "loss": 0.8074,
      "step": 4648
    },
    {
      "epoch": 1.277159730806208,
      "grad_norm": 0.3771061897277832,
      "learning_rate": 7.239246942421329e-06,
      "loss": 0.6124,
      "step": 4649
    },
    {
      "epoch": 1.2774344183491277,
      "grad_norm": 0.5593774318695068,
      "learning_rate": 7.236498557097705e-06,
      "loss": 0.7553,
      "step": 4650
    },
    {
      "epoch": 1.2777091058920478,
      "grad_norm": 0.6236807703971863,
      "learning_rate": 7.233750171774083e-06,
      "loss": 0.9451,
      "step": 4651
    },
    {
      "epoch": 1.2779837934349678,
      "grad_norm": 0.46942922472953796,
      "learning_rate": 7.231001786450461e-06,
      "loss": 1.0278,
      "step": 4652
    },
    {
      "epoch": 1.2782584809778876,
      "grad_norm": 0.4923854172229767,
      "learning_rate": 7.228253401126839e-06,
      "loss": 0.6809,
      "step": 4653
    },
    {
      "epoch": 1.2785331685208075,
      "grad_norm": 0.4660966992378235,
      "learning_rate": 7.225505015803216e-06,
      "loss": 0.6952,
      "step": 4654
    },
    {
      "epoch": 1.2788078560637275,
      "grad_norm": 0.5346837043762207,
      "learning_rate": 7.2227566304795936e-06,
      "loss": 0.6085,
      "step": 4655
    },
    {
      "epoch": 1.2790825436066475,
      "grad_norm": 0.5608029961585999,
      "learning_rate": 7.2200082451559714e-06,
      "loss": 0.8794,
      "step": 4656
    },
    {
      "epoch": 1.2793572311495673,
      "grad_norm": 0.32469481229782104,
      "learning_rate": 7.217259859832349e-06,
      "loss": 0.8256,
      "step": 4657
    },
    {
      "epoch": 1.2796319186924872,
      "grad_norm": 0.6628618240356445,
      "learning_rate": 7.214511474508726e-06,
      "loss": 0.8038,
      "step": 4658
    },
    {
      "epoch": 1.2799066062354072,
      "grad_norm": 0.406222939491272,
      "learning_rate": 7.211763089185104e-06,
      "loss": 0.5173,
      "step": 4659
    },
    {
      "epoch": 1.2801812937783272,
      "grad_norm": 0.3916313946247101,
      "learning_rate": 7.209014703861482e-06,
      "loss": 0.64,
      "step": 4660
    },
    {
      "epoch": 1.280455981321247,
      "grad_norm": 0.3535917103290558,
      "learning_rate": 7.20626631853786e-06,
      "loss": 0.6989,
      "step": 4661
    },
    {
      "epoch": 1.280730668864167,
      "grad_norm": 0.4737142026424408,
      "learning_rate": 7.203517933214237e-06,
      "loss": 0.718,
      "step": 4662
    },
    {
      "epoch": 1.281005356407087,
      "grad_norm": 0.39815813302993774,
      "learning_rate": 7.200769547890615e-06,
      "loss": 0.8443,
      "step": 4663
    },
    {
      "epoch": 1.281280043950007,
      "grad_norm": 0.45714902877807617,
      "learning_rate": 7.198021162566993e-06,
      "loss": 0.8189,
      "step": 4664
    },
    {
      "epoch": 1.2815547314929268,
      "grad_norm": 0.43695494532585144,
      "learning_rate": 7.1952727772433706e-06,
      "loss": 0.5686,
      "step": 4665
    },
    {
      "epoch": 1.2818294190358468,
      "grad_norm": 0.4643217623233795,
      "learning_rate": 7.1925243919197476e-06,
      "loss": 0.7051,
      "step": 4666
    },
    {
      "epoch": 1.2821041065787666,
      "grad_norm": 0.3982868194580078,
      "learning_rate": 7.1897760065961254e-06,
      "loss": 0.7341,
      "step": 4667
    },
    {
      "epoch": 1.2823787941216866,
      "grad_norm": 0.37828508019447327,
      "learning_rate": 7.187027621272503e-06,
      "loss": 0.5981,
      "step": 4668
    },
    {
      "epoch": 1.2826534816646065,
      "grad_norm": 0.48707106709480286,
      "learning_rate": 7.184279235948881e-06,
      "loss": 0.8238,
      "step": 4669
    },
    {
      "epoch": 1.2829281692075265,
      "grad_norm": 0.4852864444255829,
      "learning_rate": 7.181530850625258e-06,
      "loss": 0.9734,
      "step": 4670
    },
    {
      "epoch": 1.2832028567504463,
      "grad_norm": 0.6204916834831238,
      "learning_rate": 7.178782465301636e-06,
      "loss": 0.8174,
      "step": 4671
    },
    {
      "epoch": 1.2834775442933664,
      "grad_norm": 0.5357089638710022,
      "learning_rate": 7.176034079978014e-06,
      "loss": 0.6441,
      "step": 4672
    },
    {
      "epoch": 1.2837522318362862,
      "grad_norm": 0.45720183849334717,
      "learning_rate": 7.173285694654392e-06,
      "loss": 0.8429,
      "step": 4673
    },
    {
      "epoch": 1.2840269193792062,
      "grad_norm": 0.29122069478034973,
      "learning_rate": 7.170537309330769e-06,
      "loss": 0.7722,
      "step": 4674
    },
    {
      "epoch": 1.284301606922126,
      "grad_norm": 0.5896292328834534,
      "learning_rate": 7.167788924007147e-06,
      "loss": 0.884,
      "step": 4675
    },
    {
      "epoch": 1.284576294465046,
      "grad_norm": 0.5340958833694458,
      "learning_rate": 7.1650405386835246e-06,
      "loss": 0.6718,
      "step": 4676
    },
    {
      "epoch": 1.2848509820079659,
      "grad_norm": 0.45524489879608154,
      "learning_rate": 7.1622921533599024e-06,
      "loss": 0.6596,
      "step": 4677
    },
    {
      "epoch": 1.285125669550886,
      "grad_norm": 0.3008299469947815,
      "learning_rate": 7.159543768036279e-06,
      "loss": 0.5375,
      "step": 4678
    },
    {
      "epoch": 1.2854003570938057,
      "grad_norm": 0.5117905139923096,
      "learning_rate": 7.1567953827126565e-06,
      "loss": 0.5659,
      "step": 4679
    },
    {
      "epoch": 1.2856750446367258,
      "grad_norm": 0.48976263403892517,
      "learning_rate": 7.154046997389034e-06,
      "loss": 0.8604,
      "step": 4680
    },
    {
      "epoch": 1.2859497321796456,
      "grad_norm": 0.6311905384063721,
      "learning_rate": 7.151298612065413e-06,
      "loss": 0.6938,
      "step": 4681
    },
    {
      "epoch": 1.2862244197225656,
      "grad_norm": 0.38071030378341675,
      "learning_rate": 7.148550226741789e-06,
      "loss": 0.7762,
      "step": 4682
    },
    {
      "epoch": 1.2864991072654854,
      "grad_norm": 0.5923769474029541,
      "learning_rate": 7.145801841418167e-06,
      "loss": 0.6366,
      "step": 4683
    },
    {
      "epoch": 1.2867737948084055,
      "grad_norm": 0.33407557010650635,
      "learning_rate": 7.143053456094545e-06,
      "loss": 0.7323,
      "step": 4684
    },
    {
      "epoch": 1.2870484823513253,
      "grad_norm": 0.4358058273792267,
      "learning_rate": 7.140305070770923e-06,
      "loss": 0.6029,
      "step": 4685
    },
    {
      "epoch": 1.2873231698942453,
      "grad_norm": 0.5202435851097107,
      "learning_rate": 7.1375566854473e-06,
      "loss": 0.8225,
      "step": 4686
    },
    {
      "epoch": 1.2875978574371651,
      "grad_norm": 0.4575293958187103,
      "learning_rate": 7.134808300123678e-06,
      "loss": 0.7995,
      "step": 4687
    },
    {
      "epoch": 1.2878725449800852,
      "grad_norm": 0.511448323726654,
      "learning_rate": 7.132059914800056e-06,
      "loss": 0.9453,
      "step": 4688
    },
    {
      "epoch": 1.288147232523005,
      "grad_norm": 0.46067389845848083,
      "learning_rate": 7.1293115294764335e-06,
      "loss": 0.7463,
      "step": 4689
    },
    {
      "epoch": 1.288421920065925,
      "grad_norm": 0.4761159121990204,
      "learning_rate": 7.1265631441528105e-06,
      "loss": 0.8388,
      "step": 4690
    },
    {
      "epoch": 1.2886966076088449,
      "grad_norm": 0.44825324416160583,
      "learning_rate": 7.123814758829188e-06,
      "loss": 0.7017,
      "step": 4691
    },
    {
      "epoch": 1.288971295151765,
      "grad_norm": 0.45634105801582336,
      "learning_rate": 7.121066373505566e-06,
      "loss": 0.9578,
      "step": 4692
    },
    {
      "epoch": 1.289245982694685,
      "grad_norm": 0.5711625218391418,
      "learning_rate": 7.118317988181944e-06,
      "loss": 0.88,
      "step": 4693
    },
    {
      "epoch": 1.2895206702376047,
      "grad_norm": 0.42461681365966797,
      "learning_rate": 7.115569602858321e-06,
      "loss": 0.6263,
      "step": 4694
    },
    {
      "epoch": 1.2897953577805246,
      "grad_norm": 0.4467339515686035,
      "learning_rate": 7.112821217534699e-06,
      "loss": 0.5786,
      "step": 4695
    },
    {
      "epoch": 1.2900700453234446,
      "grad_norm": 0.49691206216812134,
      "learning_rate": 7.110072832211077e-06,
      "loss": 0.7037,
      "step": 4696
    },
    {
      "epoch": 1.2903447328663646,
      "grad_norm": 0.3542187511920929,
      "learning_rate": 7.107324446887455e-06,
      "loss": 0.6215,
      "step": 4697
    },
    {
      "epoch": 1.2906194204092845,
      "grad_norm": 0.597669780254364,
      "learning_rate": 7.104576061563832e-06,
      "loss": 0.803,
      "step": 4698
    },
    {
      "epoch": 1.2908941079522043,
      "grad_norm": 0.5401637554168701,
      "learning_rate": 7.10182767624021e-06,
      "loss": 0.8429,
      "step": 4699
    },
    {
      "epoch": 1.2911687954951243,
      "grad_norm": 0.4172574579715729,
      "learning_rate": 7.0990792909165875e-06,
      "loss": 0.8249,
      "step": 4700
    },
    {
      "epoch": 1.2914434830380443,
      "grad_norm": 0.3902255594730377,
      "learning_rate": 7.096330905592965e-06,
      "loss": 0.8137,
      "step": 4701
    },
    {
      "epoch": 1.2917181705809642,
      "grad_norm": 0.4453471601009369,
      "learning_rate": 7.093582520269342e-06,
      "loss": 0.6615,
      "step": 4702
    },
    {
      "epoch": 1.291992858123884,
      "grad_norm": 0.43067145347595215,
      "learning_rate": 7.09083413494572e-06,
      "loss": 0.5844,
      "step": 4703
    },
    {
      "epoch": 1.292267545666804,
      "grad_norm": 0.3869006037712097,
      "learning_rate": 7.088085749622098e-06,
      "loss": 0.5024,
      "step": 4704
    },
    {
      "epoch": 1.292542233209724,
      "grad_norm": 0.5230872631072998,
      "learning_rate": 7.085337364298476e-06,
      "loss": 0.9245,
      "step": 4705
    },
    {
      "epoch": 1.2928169207526439,
      "grad_norm": 0.5315709710121155,
      "learning_rate": 7.082588978974852e-06,
      "loss": 0.766,
      "step": 4706
    },
    {
      "epoch": 1.2930916082955637,
      "grad_norm": 0.39975348114967346,
      "learning_rate": 7.079840593651231e-06,
      "loss": 0.6762,
      "step": 4707
    },
    {
      "epoch": 1.2933662958384837,
      "grad_norm": 0.47710758447647095,
      "learning_rate": 7.077092208327609e-06,
      "loss": 0.8925,
      "step": 4708
    },
    {
      "epoch": 1.2936409833814038,
      "grad_norm": 0.53385990858078,
      "learning_rate": 7.0743438230039866e-06,
      "loss": 0.9482,
      "step": 4709
    },
    {
      "epoch": 1.2939156709243236,
      "grad_norm": 0.4443935751914978,
      "learning_rate": 7.071595437680363e-06,
      "loss": 0.728,
      "step": 4710
    },
    {
      "epoch": 1.2941903584672434,
      "grad_norm": 0.4742089807987213,
      "learning_rate": 7.068847052356741e-06,
      "loss": 0.9051,
      "step": 4711
    },
    {
      "epoch": 1.2944650460101634,
      "grad_norm": 0.485679030418396,
      "learning_rate": 7.0660986670331185e-06,
      "loss": 0.8518,
      "step": 4712
    },
    {
      "epoch": 1.2947397335530835,
      "grad_norm": 0.3830036520957947,
      "learning_rate": 7.063350281709496e-06,
      "loss": 0.7707,
      "step": 4713
    },
    {
      "epoch": 1.2950144210960033,
      "grad_norm": 0.37692025303840637,
      "learning_rate": 7.060601896385873e-06,
      "loss": 0.6787,
      "step": 4714
    },
    {
      "epoch": 1.295289108638923,
      "grad_norm": 0.45660173892974854,
      "learning_rate": 7.057853511062251e-06,
      "loss": 0.7405,
      "step": 4715
    },
    {
      "epoch": 1.2955637961818431,
      "grad_norm": 0.291375070810318,
      "learning_rate": 7.055105125738629e-06,
      "loss": 0.5289,
      "step": 4716
    },
    {
      "epoch": 1.2958384837247632,
      "grad_norm": 0.5906421542167664,
      "learning_rate": 7.052356740415007e-06,
      "loss": 0.8472,
      "step": 4717
    },
    {
      "epoch": 1.296113171267683,
      "grad_norm": 0.4185674786567688,
      "learning_rate": 7.049608355091384e-06,
      "loss": 0.6854,
      "step": 4718
    },
    {
      "epoch": 1.2963878588106028,
      "grad_norm": 0.3900362253189087,
      "learning_rate": 7.046859969767762e-06,
      "loss": 0.8572,
      "step": 4719
    },
    {
      "epoch": 1.2966625463535228,
      "grad_norm": 0.38594651222229004,
      "learning_rate": 7.04411158444414e-06,
      "loss": 0.8382,
      "step": 4720
    },
    {
      "epoch": 1.2969372338964429,
      "grad_norm": 0.3283836245536804,
      "learning_rate": 7.041363199120518e-06,
      "loss": 0.6623,
      "step": 4721
    },
    {
      "epoch": 1.2972119214393627,
      "grad_norm": 0.3572525978088379,
      "learning_rate": 7.038614813796895e-06,
      "loss": 0.5662,
      "step": 4722
    },
    {
      "epoch": 1.2974866089822825,
      "grad_norm": 0.4921099543571472,
      "learning_rate": 7.0358664284732725e-06,
      "loss": 0.6409,
      "step": 4723
    },
    {
      "epoch": 1.2977612965252026,
      "grad_norm": 0.43294861912727356,
      "learning_rate": 7.03311804314965e-06,
      "loss": 0.6562,
      "step": 4724
    },
    {
      "epoch": 1.2980359840681226,
      "grad_norm": 0.46324554085731506,
      "learning_rate": 7.030369657826028e-06,
      "loss": 0.6233,
      "step": 4725
    },
    {
      "epoch": 1.2983106716110424,
      "grad_norm": 0.5368472337722778,
      "learning_rate": 7.027621272502405e-06,
      "loss": 0.8966,
      "step": 4726
    },
    {
      "epoch": 1.2985853591539624,
      "grad_norm": 0.4639420211315155,
      "learning_rate": 7.024872887178783e-06,
      "loss": 0.7871,
      "step": 4727
    },
    {
      "epoch": 1.2988600466968823,
      "grad_norm": 0.6049048900604248,
      "learning_rate": 7.022124501855161e-06,
      "loss": 0.6396,
      "step": 4728
    },
    {
      "epoch": 1.2991347342398023,
      "grad_norm": 0.4527416229248047,
      "learning_rate": 7.019376116531539e-06,
      "loss": 0.8136,
      "step": 4729
    },
    {
      "epoch": 1.2994094217827221,
      "grad_norm": 0.43838077783584595,
      "learning_rate": 7.016627731207916e-06,
      "loss": 0.8352,
      "step": 4730
    },
    {
      "epoch": 1.2996841093256422,
      "grad_norm": 0.5456330180168152,
      "learning_rate": 7.013879345884294e-06,
      "loss": 0.5479,
      "step": 4731
    },
    {
      "epoch": 1.299958796868562,
      "grad_norm": 0.4923820495605469,
      "learning_rate": 7.011130960560672e-06,
      "loss": 0.667,
      "step": 4732
    },
    {
      "epoch": 1.300233484411482,
      "grad_norm": 0.5129911303520203,
      "learning_rate": 7.0083825752370495e-06,
      "loss": 1.0438,
      "step": 4733
    },
    {
      "epoch": 1.3005081719544018,
      "grad_norm": 0.48827335238456726,
      "learning_rate": 7.0056341899134265e-06,
      "loss": 0.6685,
      "step": 4734
    },
    {
      "epoch": 1.3007828594973219,
      "grad_norm": 0.3813888728618622,
      "learning_rate": 7.002885804589804e-06,
      "loss": 0.9066,
      "step": 4735
    },
    {
      "epoch": 1.3010575470402417,
      "grad_norm": 0.2812984585762024,
      "learning_rate": 7.000137419266182e-06,
      "loss": 0.4448,
      "step": 4736
    },
    {
      "epoch": 1.3013322345831617,
      "grad_norm": 0.4270212650299072,
      "learning_rate": 6.99738903394256e-06,
      "loss": 0.7803,
      "step": 4737
    },
    {
      "epoch": 1.3016069221260815,
      "grad_norm": 0.383296400308609,
      "learning_rate": 6.994640648618936e-06,
      "loss": 0.7415,
      "step": 4738
    },
    {
      "epoch": 1.3018816096690016,
      "grad_norm": 0.4535225033760071,
      "learning_rate": 6.991892263295315e-06,
      "loss": 0.7712,
      "step": 4739
    },
    {
      "epoch": 1.3021562972119214,
      "grad_norm": 0.43434086441993713,
      "learning_rate": 6.989143877971693e-06,
      "loss": 0.9569,
      "step": 4740
    },
    {
      "epoch": 1.3024309847548414,
      "grad_norm": 0.4629109799861908,
      "learning_rate": 6.986395492648071e-06,
      "loss": 0.9896,
      "step": 4741
    },
    {
      "epoch": 1.3027056722977612,
      "grad_norm": 0.41020119190216064,
      "learning_rate": 6.983647107324447e-06,
      "loss": 0.7371,
      "step": 4742
    },
    {
      "epoch": 1.3029803598406813,
      "grad_norm": 0.4840027391910553,
      "learning_rate": 6.980898722000825e-06,
      "loss": 0.6635,
      "step": 4743
    },
    {
      "epoch": 1.303255047383601,
      "grad_norm": 0.4071095585823059,
      "learning_rate": 6.978150336677203e-06,
      "loss": 0.6363,
      "step": 4744
    },
    {
      "epoch": 1.3035297349265211,
      "grad_norm": 0.45461198687553406,
      "learning_rate": 6.9754019513535805e-06,
      "loss": 0.7238,
      "step": 4745
    },
    {
      "epoch": 1.303804422469441,
      "grad_norm": 0.4119392931461334,
      "learning_rate": 6.9726535660299575e-06,
      "loss": 0.7062,
      "step": 4746
    },
    {
      "epoch": 1.304079110012361,
      "grad_norm": 0.4992806315422058,
      "learning_rate": 6.969905180706335e-06,
      "loss": 0.7325,
      "step": 4747
    },
    {
      "epoch": 1.3043537975552808,
      "grad_norm": 0.41643938422203064,
      "learning_rate": 6.967156795382713e-06,
      "loss": 1.0198,
      "step": 4748
    },
    {
      "epoch": 1.3046284850982008,
      "grad_norm": 0.5993474721908569,
      "learning_rate": 6.964408410059091e-06,
      "loss": 0.7689,
      "step": 4749
    },
    {
      "epoch": 1.3049031726411207,
      "grad_norm": 0.3427785336971283,
      "learning_rate": 6.961660024735468e-06,
      "loss": 0.725,
      "step": 4750
    },
    {
      "epoch": 1.3051778601840407,
      "grad_norm": 0.37036821246147156,
      "learning_rate": 6.958911639411846e-06,
      "loss": 0.6689,
      "step": 4751
    },
    {
      "epoch": 1.3054525477269605,
      "grad_norm": 0.5926555395126343,
      "learning_rate": 6.956163254088224e-06,
      "loss": 0.761,
      "step": 4752
    },
    {
      "epoch": 1.3057272352698805,
      "grad_norm": 0.6049032807350159,
      "learning_rate": 6.953414868764601e-06,
      "loss": 0.8716,
      "step": 4753
    },
    {
      "epoch": 1.3060019228128004,
      "grad_norm": 0.46743714809417725,
      "learning_rate": 6.950666483440979e-06,
      "loss": 0.6158,
      "step": 4754
    },
    {
      "epoch": 1.3062766103557204,
      "grad_norm": 0.587696373462677,
      "learning_rate": 6.947918098117357e-06,
      "loss": 0.6476,
      "step": 4755
    },
    {
      "epoch": 1.3065512978986402,
      "grad_norm": 0.3434399664402008,
      "learning_rate": 6.9451697127937345e-06,
      "loss": 0.7787,
      "step": 4756
    },
    {
      "epoch": 1.3068259854415603,
      "grad_norm": 0.5685856938362122,
      "learning_rate": 6.9424213274701115e-06,
      "loss": 0.641,
      "step": 4757
    },
    {
      "epoch": 1.30710067298448,
      "grad_norm": 0.4240616261959076,
      "learning_rate": 6.939672942146489e-06,
      "loss": 0.9611,
      "step": 4758
    },
    {
      "epoch": 1.3073753605274,
      "grad_norm": 0.5827029943466187,
      "learning_rate": 6.936924556822867e-06,
      "loss": 0.6318,
      "step": 4759
    },
    {
      "epoch": 1.30765004807032,
      "grad_norm": 0.4257844090461731,
      "learning_rate": 6.934176171499245e-06,
      "loss": 0.9593,
      "step": 4760
    },
    {
      "epoch": 1.30792473561324,
      "grad_norm": 0.5187095999717712,
      "learning_rate": 6.931427786175622e-06,
      "loss": 0.8924,
      "step": 4761
    },
    {
      "epoch": 1.30819942315616,
      "grad_norm": 0.4013836979866028,
      "learning_rate": 6.928679400852e-06,
      "loss": 1.0446,
      "step": 4762
    },
    {
      "epoch": 1.3084741106990798,
      "grad_norm": 0.3016560971736908,
      "learning_rate": 6.925931015528378e-06,
      "loss": 0.8809,
      "step": 4763
    },
    {
      "epoch": 1.3087487982419996,
      "grad_norm": 0.4517676830291748,
      "learning_rate": 6.923182630204756e-06,
      "loss": 0.8922,
      "step": 4764
    },
    {
      "epoch": 1.3090234857849197,
      "grad_norm": 0.5768283009529114,
      "learning_rate": 6.920434244881133e-06,
      "loss": 0.8048,
      "step": 4765
    },
    {
      "epoch": 1.3092981733278397,
      "grad_norm": 0.48729944229125977,
      "learning_rate": 6.917685859557511e-06,
      "loss": 0.9427,
      "step": 4766
    },
    {
      "epoch": 1.3095728608707595,
      "grad_norm": 0.5160335302352905,
      "learning_rate": 6.9149374742338885e-06,
      "loss": 0.7194,
      "step": 4767
    },
    {
      "epoch": 1.3098475484136793,
      "grad_norm": 0.45349735021591187,
      "learning_rate": 6.912189088910266e-06,
      "loss": 0.7873,
      "step": 4768
    },
    {
      "epoch": 1.3101222359565994,
      "grad_norm": 0.4160321354866028,
      "learning_rate": 6.9094407035866426e-06,
      "loss": 0.8009,
      "step": 4769
    },
    {
      "epoch": 1.3103969234995194,
      "grad_norm": 0.48577359318733215,
      "learning_rate": 6.90669231826302e-06,
      "loss": 0.6718,
      "step": 4770
    },
    {
      "epoch": 1.3106716110424392,
      "grad_norm": 0.49828940629959106,
      "learning_rate": 6.903943932939398e-06,
      "loss": 0.8374,
      "step": 4771
    },
    {
      "epoch": 1.310946298585359,
      "grad_norm": 0.39838871359825134,
      "learning_rate": 6.901195547615777e-06,
      "loss": 0.4539,
      "step": 4772
    },
    {
      "epoch": 1.311220986128279,
      "grad_norm": 0.49620741605758667,
      "learning_rate": 6.898447162292153e-06,
      "loss": 0.6547,
      "step": 4773
    },
    {
      "epoch": 1.3114956736711991,
      "grad_norm": 0.5048398971557617,
      "learning_rate": 6.895698776968531e-06,
      "loss": 0.6841,
      "step": 4774
    },
    {
      "epoch": 1.311770361214119,
      "grad_norm": 0.5697450041770935,
      "learning_rate": 6.892950391644909e-06,
      "loss": 0.7269,
      "step": 4775
    },
    {
      "epoch": 1.3120450487570388,
      "grad_norm": 0.43764349818229675,
      "learning_rate": 6.890202006321287e-06,
      "loss": 0.8074,
      "step": 4776
    },
    {
      "epoch": 1.3123197362999588,
      "grad_norm": 0.4022422730922699,
      "learning_rate": 6.887453620997664e-06,
      "loss": 0.784,
      "step": 4777
    },
    {
      "epoch": 1.3125944238428788,
      "grad_norm": 0.3401060700416565,
      "learning_rate": 6.884705235674042e-06,
      "loss": 0.5289,
      "step": 4778
    },
    {
      "epoch": 1.3128691113857986,
      "grad_norm": 0.37291979789733887,
      "learning_rate": 6.8819568503504195e-06,
      "loss": 0.5078,
      "step": 4779
    },
    {
      "epoch": 1.3131437989287185,
      "grad_norm": 0.5371974110603333,
      "learning_rate": 6.879208465026797e-06,
      "loss": 0.6213,
      "step": 4780
    },
    {
      "epoch": 1.3134184864716385,
      "grad_norm": 0.3761036992073059,
      "learning_rate": 6.876460079703174e-06,
      "loss": 0.4696,
      "step": 4781
    },
    {
      "epoch": 1.3136931740145585,
      "grad_norm": 0.4038563370704651,
      "learning_rate": 6.873711694379552e-06,
      "loss": 0.9249,
      "step": 4782
    },
    {
      "epoch": 1.3139678615574784,
      "grad_norm": 0.4572203755378723,
      "learning_rate": 6.87096330905593e-06,
      "loss": 0.8808,
      "step": 4783
    },
    {
      "epoch": 1.3142425491003982,
      "grad_norm": 0.48484838008880615,
      "learning_rate": 6.868214923732308e-06,
      "loss": 0.9,
      "step": 4784
    },
    {
      "epoch": 1.3145172366433182,
      "grad_norm": 0.4296209514141083,
      "learning_rate": 6.865466538408685e-06,
      "loss": 0.6665,
      "step": 4785
    },
    {
      "epoch": 1.3147919241862382,
      "grad_norm": 0.4456424415111542,
      "learning_rate": 6.862718153085063e-06,
      "loss": 0.5626,
      "step": 4786
    },
    {
      "epoch": 1.315066611729158,
      "grad_norm": 0.501412570476532,
      "learning_rate": 6.859969767761441e-06,
      "loss": 0.8033,
      "step": 4787
    },
    {
      "epoch": 1.3153412992720779,
      "grad_norm": 0.4588662087917328,
      "learning_rate": 6.857221382437819e-06,
      "loss": 0.715,
      "step": 4788
    },
    {
      "epoch": 1.315615986814998,
      "grad_norm": 0.54009610414505,
      "learning_rate": 6.854472997114196e-06,
      "loss": 0.7854,
      "step": 4789
    },
    {
      "epoch": 1.315890674357918,
      "grad_norm": 0.4181956350803375,
      "learning_rate": 6.8517246117905735e-06,
      "loss": 0.8812,
      "step": 4790
    },
    {
      "epoch": 1.3161653619008378,
      "grad_norm": 0.3657034933567047,
      "learning_rate": 6.848976226466951e-06,
      "loss": 0.7962,
      "step": 4791
    },
    {
      "epoch": 1.3164400494437576,
      "grad_norm": 0.32826805114746094,
      "learning_rate": 6.846227841143329e-06,
      "loss": 0.5954,
      "step": 4792
    },
    {
      "epoch": 1.3167147369866776,
      "grad_norm": 0.3688773512840271,
      "learning_rate": 6.843479455819706e-06,
      "loss": 1.1852,
      "step": 4793
    },
    {
      "epoch": 1.3169894245295977,
      "grad_norm": 0.4245795011520386,
      "learning_rate": 6.840731070496084e-06,
      "loss": 0.9224,
      "step": 4794
    },
    {
      "epoch": 1.3172641120725175,
      "grad_norm": 0.4878925681114197,
      "learning_rate": 6.837982685172462e-06,
      "loss": 0.7334,
      "step": 4795
    },
    {
      "epoch": 1.3175387996154375,
      "grad_norm": 0.4804989993572235,
      "learning_rate": 6.83523429984884e-06,
      "loss": 0.6048,
      "step": 4796
    },
    {
      "epoch": 1.3178134871583573,
      "grad_norm": 0.46878546476364136,
      "learning_rate": 6.832485914525216e-06,
      "loss": 0.902,
      "step": 4797
    },
    {
      "epoch": 1.3180881747012774,
      "grad_norm": 0.36165881156921387,
      "learning_rate": 6.829737529201595e-06,
      "loss": 0.7272,
      "step": 4798
    },
    {
      "epoch": 1.3183628622441972,
      "grad_norm": 0.655619204044342,
      "learning_rate": 6.826989143877973e-06,
      "loss": 0.8549,
      "step": 4799
    },
    {
      "epoch": 1.3186375497871172,
      "grad_norm": 0.43796083331108093,
      "learning_rate": 6.8242407585543505e-06,
      "loss": 0.74,
      "step": 4800
    },
    {
      "epoch": 1.318912237330037,
      "grad_norm": 0.36030542850494385,
      "learning_rate": 6.821492373230727e-06,
      "loss": 0.6343,
      "step": 4801
    },
    {
      "epoch": 1.319186924872957,
      "grad_norm": 0.4094589352607727,
      "learning_rate": 6.8187439879071046e-06,
      "loss": 0.8696,
      "step": 4802
    },
    {
      "epoch": 1.319461612415877,
      "grad_norm": 0.5010083317756653,
      "learning_rate": 6.8159956025834824e-06,
      "loss": 0.6772,
      "step": 4803
    },
    {
      "epoch": 1.319736299958797,
      "grad_norm": 0.5140247941017151,
      "learning_rate": 6.81324721725986e-06,
      "loss": 0.8415,
      "step": 4804
    },
    {
      "epoch": 1.3200109875017167,
      "grad_norm": 0.39422085881233215,
      "learning_rate": 6.810498831936237e-06,
      "loss": 0.4717,
      "step": 4805
    },
    {
      "epoch": 1.3202856750446368,
      "grad_norm": 0.3279789686203003,
      "learning_rate": 6.807750446612615e-06,
      "loss": 0.7899,
      "step": 4806
    },
    {
      "epoch": 1.3205603625875566,
      "grad_norm": 0.45146888494491577,
      "learning_rate": 6.805002061288993e-06,
      "loss": 0.9353,
      "step": 4807
    },
    {
      "epoch": 1.3208350501304766,
      "grad_norm": 0.3999648988246918,
      "learning_rate": 6.802253675965371e-06,
      "loss": 1.0213,
      "step": 4808
    },
    {
      "epoch": 1.3211097376733965,
      "grad_norm": 0.537138819694519,
      "learning_rate": 6.799505290641748e-06,
      "loss": 0.6434,
      "step": 4809
    },
    {
      "epoch": 1.3213844252163165,
      "grad_norm": 0.44214701652526855,
      "learning_rate": 6.796756905318126e-06,
      "loss": 0.6608,
      "step": 4810
    },
    {
      "epoch": 1.3216591127592363,
      "grad_norm": 0.5586408972740173,
      "learning_rate": 6.794008519994504e-06,
      "loss": 0.8419,
      "step": 4811
    },
    {
      "epoch": 1.3219338003021563,
      "grad_norm": 0.34968826174736023,
      "learning_rate": 6.7912601346708815e-06,
      "loss": 0.5493,
      "step": 4812
    },
    {
      "epoch": 1.3222084878450762,
      "grad_norm": 0.649865448474884,
      "learning_rate": 6.7885117493472586e-06,
      "loss": 0.9735,
      "step": 4813
    },
    {
      "epoch": 1.3224831753879962,
      "grad_norm": 0.43910568952560425,
      "learning_rate": 6.7857633640236364e-06,
      "loss": 0.9397,
      "step": 4814
    },
    {
      "epoch": 1.322757862930916,
      "grad_norm": 0.5162168145179749,
      "learning_rate": 6.783014978700014e-06,
      "loss": 1.0449,
      "step": 4815
    },
    {
      "epoch": 1.323032550473836,
      "grad_norm": 0.3541739284992218,
      "learning_rate": 6.780266593376392e-06,
      "loss": 0.7825,
      "step": 4816
    },
    {
      "epoch": 1.3233072380167559,
      "grad_norm": 0.4960581660270691,
      "learning_rate": 6.777518208052769e-06,
      "loss": 0.8017,
      "step": 4817
    },
    {
      "epoch": 1.323581925559676,
      "grad_norm": 0.5507283806800842,
      "learning_rate": 6.774769822729147e-06,
      "loss": 0.7529,
      "step": 4818
    },
    {
      "epoch": 1.3238566131025957,
      "grad_norm": 0.414959192276001,
      "learning_rate": 6.772021437405525e-06,
      "loss": 0.7064,
      "step": 4819
    },
    {
      "epoch": 1.3241313006455158,
      "grad_norm": 0.5318635106086731,
      "learning_rate": 6.769273052081903e-06,
      "loss": 0.7211,
      "step": 4820
    },
    {
      "epoch": 1.3244059881884356,
      "grad_norm": 0.4870814085006714,
      "learning_rate": 6.76652466675828e-06,
      "loss": 0.7253,
      "step": 4821
    },
    {
      "epoch": 1.3246806757313556,
      "grad_norm": 0.3998878002166748,
      "learning_rate": 6.763776281434658e-06,
      "loss": 0.7166,
      "step": 4822
    },
    {
      "epoch": 1.3249553632742754,
      "grad_norm": 0.5463903546333313,
      "learning_rate": 6.7610278961110356e-06,
      "loss": 0.6059,
      "step": 4823
    },
    {
      "epoch": 1.3252300508171955,
      "grad_norm": 0.387685626745224,
      "learning_rate": 6.758279510787413e-06,
      "loss": 0.7375,
      "step": 4824
    },
    {
      "epoch": 1.3255047383601153,
      "grad_norm": 0.4782569408416748,
      "learning_rate": 6.7555311254637904e-06,
      "loss": 0.8502,
      "step": 4825
    },
    {
      "epoch": 1.3257794259030353,
      "grad_norm": 0.5497481226921082,
      "learning_rate": 6.752782740140168e-06,
      "loss": 0.6348,
      "step": 4826
    },
    {
      "epoch": 1.3260541134459554,
      "grad_norm": 0.3819947838783264,
      "learning_rate": 6.750034354816546e-06,
      "loss": 0.8641,
      "step": 4827
    },
    {
      "epoch": 1.3263288009888752,
      "grad_norm": 0.5219348073005676,
      "learning_rate": 6.747285969492924e-06,
      "loss": 0.857,
      "step": 4828
    },
    {
      "epoch": 1.326603488531795,
      "grad_norm": 0.573885440826416,
      "learning_rate": 6.7445375841693e-06,
      "loss": 0.8149,
      "step": 4829
    },
    {
      "epoch": 1.326878176074715,
      "grad_norm": 0.4544476568698883,
      "learning_rate": 6.741789198845678e-06,
      "loss": 0.7109,
      "step": 4830
    },
    {
      "epoch": 1.327152863617635,
      "grad_norm": 0.3691796660423279,
      "learning_rate": 6.739040813522057e-06,
      "loss": 0.7276,
      "step": 4831
    },
    {
      "epoch": 1.3274275511605549,
      "grad_norm": 0.41929060220718384,
      "learning_rate": 6.736292428198435e-06,
      "loss": 0.5905,
      "step": 4832
    },
    {
      "epoch": 1.3277022387034747,
      "grad_norm": 0.4153493344783783,
      "learning_rate": 6.733544042874811e-06,
      "loss": 1.0264,
      "step": 4833
    },
    {
      "epoch": 1.3279769262463947,
      "grad_norm": 0.3188563883304596,
      "learning_rate": 6.730795657551189e-06,
      "loss": 0.4615,
      "step": 4834
    },
    {
      "epoch": 1.3282516137893148,
      "grad_norm": 0.6707417964935303,
      "learning_rate": 6.728047272227567e-06,
      "loss": 0.9101,
      "step": 4835
    },
    {
      "epoch": 1.3285263013322346,
      "grad_norm": 0.5631164908409119,
      "learning_rate": 6.7252988869039444e-06,
      "loss": 0.6453,
      "step": 4836
    },
    {
      "epoch": 1.3288009888751544,
      "grad_norm": 0.44053059816360474,
      "learning_rate": 6.7225505015803215e-06,
      "loss": 0.8148,
      "step": 4837
    },
    {
      "epoch": 1.3290756764180744,
      "grad_norm": 0.39918217062950134,
      "learning_rate": 6.719802116256699e-06,
      "loss": 0.8041,
      "step": 4838
    },
    {
      "epoch": 1.3293503639609945,
      "grad_norm": 0.5558343529701233,
      "learning_rate": 6.717053730933077e-06,
      "loss": 0.7569,
      "step": 4839
    },
    {
      "epoch": 1.3296250515039143,
      "grad_norm": 0.48606550693511963,
      "learning_rate": 6.714305345609455e-06,
      "loss": 0.8688,
      "step": 4840
    },
    {
      "epoch": 1.3298997390468341,
      "grad_norm": 0.3606182932853699,
      "learning_rate": 6.711556960285832e-06,
      "loss": 0.5015,
      "step": 4841
    },
    {
      "epoch": 1.3301744265897542,
      "grad_norm": 0.45835819840431213,
      "learning_rate": 6.70880857496221e-06,
      "loss": 0.6801,
      "step": 4842
    },
    {
      "epoch": 1.3304491141326742,
      "grad_norm": 0.4469321370124817,
      "learning_rate": 6.706060189638588e-06,
      "loss": 0.6585,
      "step": 4843
    },
    {
      "epoch": 1.330723801675594,
      "grad_norm": 0.6105084419250488,
      "learning_rate": 6.703311804314966e-06,
      "loss": 0.807,
      "step": 4844
    },
    {
      "epoch": 1.3309984892185138,
      "grad_norm": 0.5172963738441467,
      "learning_rate": 6.700563418991343e-06,
      "loss": 0.8612,
      "step": 4845
    },
    {
      "epoch": 1.3312731767614339,
      "grad_norm": 0.42274925112724304,
      "learning_rate": 6.697815033667721e-06,
      "loss": 0.5742,
      "step": 4846
    },
    {
      "epoch": 1.331547864304354,
      "grad_norm": 0.40676161646842957,
      "learning_rate": 6.6950666483440985e-06,
      "loss": 0.8622,
      "step": 4847
    },
    {
      "epoch": 1.3318225518472737,
      "grad_norm": 0.41435280442237854,
      "learning_rate": 6.692318263020476e-06,
      "loss": 0.8436,
      "step": 4848
    },
    {
      "epoch": 1.3320972393901935,
      "grad_norm": 0.46716803312301636,
      "learning_rate": 6.689569877696853e-06,
      "loss": 1.0117,
      "step": 4849
    },
    {
      "epoch": 1.3323719269331136,
      "grad_norm": 0.3534153997898102,
      "learning_rate": 6.686821492373231e-06,
      "loss": 0.7054,
      "step": 4850
    },
    {
      "epoch": 1.3326466144760336,
      "grad_norm": 0.4836193323135376,
      "learning_rate": 6.684073107049609e-06,
      "loss": 0.9587,
      "step": 4851
    },
    {
      "epoch": 1.3329213020189534,
      "grad_norm": 0.46342357993125916,
      "learning_rate": 6.681324721725987e-06,
      "loss": 0.7716,
      "step": 4852
    },
    {
      "epoch": 1.3331959895618732,
      "grad_norm": 0.5012251734733582,
      "learning_rate": 6.678576336402364e-06,
      "loss": 0.6859,
      "step": 4853
    },
    {
      "epoch": 1.3334706771047933,
      "grad_norm": 0.648066520690918,
      "learning_rate": 6.675827951078742e-06,
      "loss": 0.9367,
      "step": 4854
    },
    {
      "epoch": 1.3337453646477133,
      "grad_norm": 0.45804107189178467,
      "learning_rate": 6.67307956575512e-06,
      "loss": 0.9897,
      "step": 4855
    },
    {
      "epoch": 1.3340200521906331,
      "grad_norm": 0.5161862373352051,
      "learning_rate": 6.6703311804314976e-06,
      "loss": 0.72,
      "step": 4856
    },
    {
      "epoch": 1.334294739733553,
      "grad_norm": 0.411310076713562,
      "learning_rate": 6.667582795107875e-06,
      "loss": 0.923,
      "step": 4857
    },
    {
      "epoch": 1.334569427276473,
      "grad_norm": 0.34122234582901,
      "learning_rate": 6.6648344097842525e-06,
      "loss": 0.865,
      "step": 4858
    },
    {
      "epoch": 1.334844114819393,
      "grad_norm": 0.5636544227600098,
      "learning_rate": 6.66208602446063e-06,
      "loss": 0.8171,
      "step": 4859
    },
    {
      "epoch": 1.3351188023623128,
      "grad_norm": 0.6584670543670654,
      "learning_rate": 6.659337639137008e-06,
      "loss": 0.715,
      "step": 4860
    },
    {
      "epoch": 1.3353934899052329,
      "grad_norm": 0.40466421842575073,
      "learning_rate": 6.656589253813384e-06,
      "loss": 0.5423,
      "step": 4861
    },
    {
      "epoch": 1.3356681774481527,
      "grad_norm": 0.4192219376564026,
      "learning_rate": 6.653840868489762e-06,
      "loss": 0.6556,
      "step": 4862
    },
    {
      "epoch": 1.3359428649910727,
      "grad_norm": 0.3616863191127777,
      "learning_rate": 6.651092483166141e-06,
      "loss": 0.9963,
      "step": 4863
    },
    {
      "epoch": 1.3362175525339925,
      "grad_norm": 0.3929496705532074,
      "learning_rate": 6.648344097842519e-06,
      "loss": 0.9492,
      "step": 4864
    },
    {
      "epoch": 1.3364922400769126,
      "grad_norm": 0.4644421935081482,
      "learning_rate": 6.645595712518895e-06,
      "loss": 0.715,
      "step": 4865
    },
    {
      "epoch": 1.3367669276198324,
      "grad_norm": 0.640666127204895,
      "learning_rate": 6.642847327195273e-06,
      "loss": 0.8372,
      "step": 4866
    },
    {
      "epoch": 1.3370416151627524,
      "grad_norm": 0.4309883415699005,
      "learning_rate": 6.640098941871651e-06,
      "loss": 0.7159,
      "step": 4867
    },
    {
      "epoch": 1.3373163027056723,
      "grad_norm": 0.3928660750389099,
      "learning_rate": 6.637350556548029e-06,
      "loss": 0.803,
      "step": 4868
    },
    {
      "epoch": 1.3375909902485923,
      "grad_norm": 0.4712608754634857,
      "learning_rate": 6.634602171224406e-06,
      "loss": 0.7796,
      "step": 4869
    },
    {
      "epoch": 1.337865677791512,
      "grad_norm": 0.3882645070552826,
      "learning_rate": 6.6318537859007835e-06,
      "loss": 0.7031,
      "step": 4870
    },
    {
      "epoch": 1.3381403653344321,
      "grad_norm": 0.40992408990859985,
      "learning_rate": 6.629105400577161e-06,
      "loss": 0.5822,
      "step": 4871
    },
    {
      "epoch": 1.338415052877352,
      "grad_norm": 0.40458613634109497,
      "learning_rate": 6.626357015253539e-06,
      "loss": 0.7882,
      "step": 4872
    },
    {
      "epoch": 1.338689740420272,
      "grad_norm": 0.2880481481552124,
      "learning_rate": 6.623608629929916e-06,
      "loss": 0.719,
      "step": 4873
    },
    {
      "epoch": 1.3389644279631918,
      "grad_norm": 0.5715451836585999,
      "learning_rate": 6.620860244606294e-06,
      "loss": 0.9348,
      "step": 4874
    },
    {
      "epoch": 1.3392391155061119,
      "grad_norm": 0.44358041882514954,
      "learning_rate": 6.618111859282672e-06,
      "loss": 0.9081,
      "step": 4875
    },
    {
      "epoch": 1.3395138030490317,
      "grad_norm": 0.3778267800807953,
      "learning_rate": 6.61536347395905e-06,
      "loss": 0.8119,
      "step": 4876
    },
    {
      "epoch": 1.3397884905919517,
      "grad_norm": 0.5092009902000427,
      "learning_rate": 6.612615088635427e-06,
      "loss": 0.6918,
      "step": 4877
    },
    {
      "epoch": 1.3400631781348715,
      "grad_norm": 0.45460572838783264,
      "learning_rate": 6.609866703311805e-06,
      "loss": 0.4844,
      "step": 4878
    },
    {
      "epoch": 1.3403378656777916,
      "grad_norm": 0.46292394399642944,
      "learning_rate": 6.607118317988183e-06,
      "loss": 0.795,
      "step": 4879
    },
    {
      "epoch": 1.3406125532207114,
      "grad_norm": 0.3248820900917053,
      "learning_rate": 6.6043699326645605e-06,
      "loss": 0.6381,
      "step": 4880
    },
    {
      "epoch": 1.3408872407636314,
      "grad_norm": 0.5437043905258179,
      "learning_rate": 6.6016215473409375e-06,
      "loss": 0.7129,
      "step": 4881
    },
    {
      "epoch": 1.3411619283065512,
      "grad_norm": 0.4893501102924347,
      "learning_rate": 6.598873162017315e-06,
      "loss": 0.6872,
      "step": 4882
    },
    {
      "epoch": 1.3414366158494713,
      "grad_norm": 0.7246708869934082,
      "learning_rate": 6.596124776693693e-06,
      "loss": 0.7019,
      "step": 4883
    },
    {
      "epoch": 1.341711303392391,
      "grad_norm": 0.4933360517024994,
      "learning_rate": 6.593376391370071e-06,
      "loss": 0.8118,
      "step": 4884
    },
    {
      "epoch": 1.3419859909353111,
      "grad_norm": 0.3453083634376526,
      "learning_rate": 6.590628006046448e-06,
      "loss": 0.9526,
      "step": 4885
    },
    {
      "epoch": 1.342260678478231,
      "grad_norm": 0.4103948473930359,
      "learning_rate": 6.587879620722826e-06,
      "loss": 0.6739,
      "step": 4886
    },
    {
      "epoch": 1.342535366021151,
      "grad_norm": 0.6217215657234192,
      "learning_rate": 6.585131235399204e-06,
      "loss": 0.9675,
      "step": 4887
    },
    {
      "epoch": 1.3428100535640708,
      "grad_norm": 0.7000248432159424,
      "learning_rate": 6.582382850075582e-06,
      "loss": 0.6485,
      "step": 4888
    },
    {
      "epoch": 1.3430847411069908,
      "grad_norm": 0.4675997495651245,
      "learning_rate": 6.579634464751959e-06,
      "loss": 0.4534,
      "step": 4889
    },
    {
      "epoch": 1.3433594286499106,
      "grad_norm": 0.34933650493621826,
      "learning_rate": 6.576886079428337e-06,
      "loss": 0.7174,
      "step": 4890
    },
    {
      "epoch": 1.3436341161928307,
      "grad_norm": 0.5348476767539978,
      "learning_rate": 6.5741376941047145e-06,
      "loss": 1.005,
      "step": 4891
    },
    {
      "epoch": 1.3439088037357505,
      "grad_norm": 0.40992286801338196,
      "learning_rate": 6.571389308781092e-06,
      "loss": 0.7668,
      "step": 4892
    },
    {
      "epoch": 1.3441834912786705,
      "grad_norm": 0.4905542731285095,
      "learning_rate": 6.5686409234574685e-06,
      "loss": 0.6588,
      "step": 4893
    },
    {
      "epoch": 1.3444581788215904,
      "grad_norm": 0.6441194415092468,
      "learning_rate": 6.565892538133846e-06,
      "loss": 0.8238,
      "step": 4894
    },
    {
      "epoch": 1.3447328663645104,
      "grad_norm": 0.5329863429069519,
      "learning_rate": 6.563144152810224e-06,
      "loss": 0.609,
      "step": 4895
    },
    {
      "epoch": 1.3450075539074304,
      "grad_norm": 0.5274813771247864,
      "learning_rate": 6.560395767486603e-06,
      "loss": 0.5974,
      "step": 4896
    },
    {
      "epoch": 1.3452822414503502,
      "grad_norm": 0.2857544422149658,
      "learning_rate": 6.557647382162979e-06,
      "loss": 0.7918,
      "step": 4897
    },
    {
      "epoch": 1.34555692899327,
      "grad_norm": 0.4976291060447693,
      "learning_rate": 6.554898996839357e-06,
      "loss": 0.872,
      "step": 4898
    },
    {
      "epoch": 1.34583161653619,
      "grad_norm": 0.5513745546340942,
      "learning_rate": 6.552150611515735e-06,
      "loss": 0.9754,
      "step": 4899
    },
    {
      "epoch": 1.3461063040791101,
      "grad_norm": 0.47698068618774414,
      "learning_rate": 6.549402226192113e-06,
      "loss": 0.9813,
      "step": 4900
    },
    {
      "epoch": 1.34638099162203,
      "grad_norm": 0.39841437339782715,
      "learning_rate": 6.54665384086849e-06,
      "loss": 0.9399,
      "step": 4901
    },
    {
      "epoch": 1.3466556791649498,
      "grad_norm": 0.5908219814300537,
      "learning_rate": 6.543905455544868e-06,
      "loss": 0.6798,
      "step": 4902
    },
    {
      "epoch": 1.3469303667078698,
      "grad_norm": 0.5111178755760193,
      "learning_rate": 6.5411570702212455e-06,
      "loss": 0.7238,
      "step": 4903
    },
    {
      "epoch": 1.3472050542507898,
      "grad_norm": 0.570988118648529,
      "learning_rate": 6.538408684897623e-06,
      "loss": 0.8884,
      "step": 4904
    },
    {
      "epoch": 1.3474797417937097,
      "grad_norm": 0.4662353992462158,
      "learning_rate": 6.535660299574e-06,
      "loss": 0.5564,
      "step": 4905
    },
    {
      "epoch": 1.3477544293366295,
      "grad_norm": 0.4613804221153259,
      "learning_rate": 6.532911914250378e-06,
      "loss": 0.941,
      "step": 4906
    },
    {
      "epoch": 1.3480291168795495,
      "grad_norm": 0.5442581176757812,
      "learning_rate": 6.530163528926756e-06,
      "loss": 0.5939,
      "step": 4907
    },
    {
      "epoch": 1.3483038044224696,
      "grad_norm": 0.41382718086242676,
      "learning_rate": 6.527415143603134e-06,
      "loss": 0.8833,
      "step": 4908
    },
    {
      "epoch": 1.3485784919653894,
      "grad_norm": 0.4917092025279999,
      "learning_rate": 6.524666758279511e-06,
      "loss": 0.696,
      "step": 4909
    },
    {
      "epoch": 1.3488531795083092,
      "grad_norm": 0.4983059763908386,
      "learning_rate": 6.521918372955889e-06,
      "loss": 0.8506,
      "step": 4910
    },
    {
      "epoch": 1.3491278670512292,
      "grad_norm": 0.5832436680793762,
      "learning_rate": 6.519169987632267e-06,
      "loss": 0.6871,
      "step": 4911
    },
    {
      "epoch": 1.3494025545941493,
      "grad_norm": 0.4658409059047699,
      "learning_rate": 6.516421602308645e-06,
      "loss": 0.666,
      "step": 4912
    },
    {
      "epoch": 1.349677242137069,
      "grad_norm": 0.37939608097076416,
      "learning_rate": 6.513673216985022e-06,
      "loss": 0.6617,
      "step": 4913
    },
    {
      "epoch": 1.349951929679989,
      "grad_norm": 0.5832832455635071,
      "learning_rate": 6.5109248316613995e-06,
      "loss": 0.6876,
      "step": 4914
    },
    {
      "epoch": 1.350226617222909,
      "grad_norm": 0.5000505447387695,
      "learning_rate": 6.508176446337777e-06,
      "loss": 0.8936,
      "step": 4915
    },
    {
      "epoch": 1.350501304765829,
      "grad_norm": 0.47410160303115845,
      "learning_rate": 6.505428061014155e-06,
      "loss": 0.6792,
      "step": 4916
    },
    {
      "epoch": 1.3507759923087488,
      "grad_norm": 0.32741817831993103,
      "learning_rate": 6.502679675690532e-06,
      "loss": 0.5776,
      "step": 4917
    },
    {
      "epoch": 1.3510506798516686,
      "grad_norm": 0.4851546585559845,
      "learning_rate": 6.49993129036691e-06,
      "loss": 0.8171,
      "step": 4918
    },
    {
      "epoch": 1.3513253673945886,
      "grad_norm": 0.39364853501319885,
      "learning_rate": 6.497182905043288e-06,
      "loss": 0.6261,
      "step": 4919
    },
    {
      "epoch": 1.3516000549375087,
      "grad_norm": 0.5133031606674194,
      "learning_rate": 6.494434519719666e-06,
      "loss": 0.6519,
      "step": 4920
    },
    {
      "epoch": 1.3518747424804285,
      "grad_norm": 0.4166129529476166,
      "learning_rate": 6.491686134396042e-06,
      "loss": 0.8497,
      "step": 4921
    },
    {
      "epoch": 1.3521494300233483,
      "grad_norm": 0.46554818749427795,
      "learning_rate": 6.488937749072421e-06,
      "loss": 0.8888,
      "step": 4922
    },
    {
      "epoch": 1.3524241175662683,
      "grad_norm": 0.33868151903152466,
      "learning_rate": 6.486189363748799e-06,
      "loss": 0.6597,
      "step": 4923
    },
    {
      "epoch": 1.3526988051091884,
      "grad_norm": 0.43302300572395325,
      "learning_rate": 6.4834409784251765e-06,
      "loss": 0.8099,
      "step": 4924
    },
    {
      "epoch": 1.3529734926521082,
      "grad_norm": 0.4590868055820465,
      "learning_rate": 6.480692593101553e-06,
      "loss": 0.8443,
      "step": 4925
    },
    {
      "epoch": 1.3532481801950282,
      "grad_norm": 0.5858125686645508,
      "learning_rate": 6.4779442077779305e-06,
      "loss": 0.7321,
      "step": 4926
    },
    {
      "epoch": 1.353522867737948,
      "grad_norm": 0.4115515649318695,
      "learning_rate": 6.475195822454308e-06,
      "loss": 0.8236,
      "step": 4927
    },
    {
      "epoch": 1.353797555280868,
      "grad_norm": 0.31375497579574585,
      "learning_rate": 6.472447437130686e-06,
      "loss": 0.75,
      "step": 4928
    },
    {
      "epoch": 1.354072242823788,
      "grad_norm": 0.4648646116256714,
      "learning_rate": 6.469699051807063e-06,
      "loss": 0.6498,
      "step": 4929
    },
    {
      "epoch": 1.354346930366708,
      "grad_norm": 0.479520708322525,
      "learning_rate": 6.466950666483441e-06,
      "loss": 0.6906,
      "step": 4930
    },
    {
      "epoch": 1.3546216179096278,
      "grad_norm": 0.3487059473991394,
      "learning_rate": 6.464202281159819e-06,
      "loss": 0.946,
      "step": 4931
    },
    {
      "epoch": 1.3548963054525478,
      "grad_norm": 0.4735170304775238,
      "learning_rate": 6.461453895836197e-06,
      "loss": 0.5026,
      "step": 4932
    },
    {
      "epoch": 1.3551709929954676,
      "grad_norm": 0.43400314450263977,
      "learning_rate": 6.458705510512574e-06,
      "loss": 0.7728,
      "step": 4933
    },
    {
      "epoch": 1.3554456805383877,
      "grad_norm": 0.5450800657272339,
      "learning_rate": 6.455957125188952e-06,
      "loss": 0.6977,
      "step": 4934
    },
    {
      "epoch": 1.3557203680813075,
      "grad_norm": 0.41118115186691284,
      "learning_rate": 6.45320873986533e-06,
      "loss": 0.7968,
      "step": 4935
    },
    {
      "epoch": 1.3559950556242275,
      "grad_norm": 0.42205432057380676,
      "learning_rate": 6.4504603545417075e-06,
      "loss": 0.9476,
      "step": 4936
    },
    {
      "epoch": 1.3562697431671473,
      "grad_norm": 0.45956870913505554,
      "learning_rate": 6.4477119692180845e-06,
      "loss": 0.7785,
      "step": 4937
    },
    {
      "epoch": 1.3565444307100674,
      "grad_norm": 0.4957009553909302,
      "learning_rate": 6.444963583894462e-06,
      "loss": 0.7205,
      "step": 4938
    },
    {
      "epoch": 1.3568191182529872,
      "grad_norm": 0.6347135901451111,
      "learning_rate": 6.44221519857084e-06,
      "loss": 0.8175,
      "step": 4939
    },
    {
      "epoch": 1.3570938057959072,
      "grad_norm": 0.38118305802345276,
      "learning_rate": 6.439466813247218e-06,
      "loss": 0.6197,
      "step": 4940
    },
    {
      "epoch": 1.357368493338827,
      "grad_norm": 0.3850763738155365,
      "learning_rate": 6.436718427923595e-06,
      "loss": 1.0197,
      "step": 4941
    },
    {
      "epoch": 1.357643180881747,
      "grad_norm": 0.5097999572753906,
      "learning_rate": 6.433970042599973e-06,
      "loss": 0.6141,
      "step": 4942
    },
    {
      "epoch": 1.3579178684246669,
      "grad_norm": 0.46951112151145935,
      "learning_rate": 6.431221657276351e-06,
      "loss": 0.9967,
      "step": 4943
    },
    {
      "epoch": 1.358192555967587,
      "grad_norm": 0.3653852641582489,
      "learning_rate": 6.428473271952729e-06,
      "loss": 0.7952,
      "step": 4944
    },
    {
      "epoch": 1.3584672435105067,
      "grad_norm": 0.49609795212745667,
      "learning_rate": 6.425724886629106e-06,
      "loss": 0.6652,
      "step": 4945
    },
    {
      "epoch": 1.3587419310534268,
      "grad_norm": 0.38113826513290405,
      "learning_rate": 6.422976501305484e-06,
      "loss": 0.6141,
      "step": 4946
    },
    {
      "epoch": 1.3590166185963466,
      "grad_norm": 0.43601274490356445,
      "learning_rate": 6.4202281159818615e-06,
      "loss": 0.7703,
      "step": 4947
    },
    {
      "epoch": 1.3592913061392666,
      "grad_norm": 0.4395063519477844,
      "learning_rate": 6.417479730658239e-06,
      "loss": 0.7159,
      "step": 4948
    },
    {
      "epoch": 1.3595659936821864,
      "grad_norm": 0.3615662157535553,
      "learning_rate": 6.414731345334616e-06,
      "loss": 1.0285,
      "step": 4949
    },
    {
      "epoch": 1.3598406812251065,
      "grad_norm": 0.5726227164268494,
      "learning_rate": 6.411982960010994e-06,
      "loss": 0.8145,
      "step": 4950
    },
    {
      "epoch": 1.3601153687680263,
      "grad_norm": 0.525675892829895,
      "learning_rate": 6.409234574687372e-06,
      "loss": 0.8688,
      "step": 4951
    },
    {
      "epoch": 1.3603900563109463,
      "grad_norm": 0.3492642045021057,
      "learning_rate": 6.40648618936375e-06,
      "loss": 0.5555,
      "step": 4952
    },
    {
      "epoch": 1.3606647438538662,
      "grad_norm": 0.3582136332988739,
      "learning_rate": 6.403737804040126e-06,
      "loss": 0.6385,
      "step": 4953
    },
    {
      "epoch": 1.3609394313967862,
      "grad_norm": 0.29523780941963196,
      "learning_rate": 6.400989418716505e-06,
      "loss": 0.6469,
      "step": 4954
    },
    {
      "epoch": 1.361214118939706,
      "grad_norm": 0.574181079864502,
      "learning_rate": 6.398241033392883e-06,
      "loss": 0.6618,
      "step": 4955
    },
    {
      "epoch": 1.361488806482626,
      "grad_norm": 0.40773797035217285,
      "learning_rate": 6.395492648069261e-06,
      "loss": 0.456,
      "step": 4956
    },
    {
      "epoch": 1.3617634940255459,
      "grad_norm": 0.5055791139602661,
      "learning_rate": 6.392744262745637e-06,
      "loss": 0.6342,
      "step": 4957
    },
    {
      "epoch": 1.362038181568466,
      "grad_norm": 0.4887053072452545,
      "learning_rate": 6.389995877422015e-06,
      "loss": 0.5671,
      "step": 4958
    },
    {
      "epoch": 1.3623128691113857,
      "grad_norm": 0.4319467544555664,
      "learning_rate": 6.3872474920983925e-06,
      "loss": 0.6567,
      "step": 4959
    },
    {
      "epoch": 1.3625875566543058,
      "grad_norm": 0.3844691514968872,
      "learning_rate": 6.38449910677477e-06,
      "loss": 0.8012,
      "step": 4960
    },
    {
      "epoch": 1.3628622441972258,
      "grad_norm": 0.48158523440361023,
      "learning_rate": 6.3817507214511474e-06,
      "loss": 0.5815,
      "step": 4961
    },
    {
      "epoch": 1.3631369317401456,
      "grad_norm": 0.40951430797576904,
      "learning_rate": 6.379002336127525e-06,
      "loss": 0.8876,
      "step": 4962
    },
    {
      "epoch": 1.3634116192830654,
      "grad_norm": 0.6273190379142761,
      "learning_rate": 6.376253950803903e-06,
      "loss": 0.7155,
      "step": 4963
    },
    {
      "epoch": 1.3636863068259855,
      "grad_norm": 0.5837612748146057,
      "learning_rate": 6.373505565480281e-06,
      "loss": 0.7161,
      "step": 4964
    },
    {
      "epoch": 1.3639609943689055,
      "grad_norm": 0.5310952663421631,
      "learning_rate": 6.370757180156658e-06,
      "loss": 0.7014,
      "step": 4965
    },
    {
      "epoch": 1.3642356819118253,
      "grad_norm": 0.46121376752853394,
      "learning_rate": 6.368008794833036e-06,
      "loss": 0.8621,
      "step": 4966
    },
    {
      "epoch": 1.3645103694547451,
      "grad_norm": 0.5220460295677185,
      "learning_rate": 6.365260409509414e-06,
      "loss": 0.8239,
      "step": 4967
    },
    {
      "epoch": 1.3647850569976652,
      "grad_norm": 0.48254725337028503,
      "learning_rate": 6.362512024185792e-06,
      "loss": 0.9333,
      "step": 4968
    },
    {
      "epoch": 1.3650597445405852,
      "grad_norm": 0.4673576056957245,
      "learning_rate": 6.359763638862169e-06,
      "loss": 0.5354,
      "step": 4969
    },
    {
      "epoch": 1.365334432083505,
      "grad_norm": 0.46023130416870117,
      "learning_rate": 6.3570152535385465e-06,
      "loss": 0.6716,
      "step": 4970
    },
    {
      "epoch": 1.3656091196264248,
      "grad_norm": 0.5960204005241394,
      "learning_rate": 6.354266868214924e-06,
      "loss": 0.8087,
      "step": 4971
    },
    {
      "epoch": 1.3658838071693449,
      "grad_norm": 0.5381335020065308,
      "learning_rate": 6.351518482891302e-06,
      "loss": 0.8235,
      "step": 4972
    },
    {
      "epoch": 1.366158494712265,
      "grad_norm": 0.4195288121700287,
      "learning_rate": 6.348770097567679e-06,
      "loss": 0.7394,
      "step": 4973
    },
    {
      "epoch": 1.3664331822551847,
      "grad_norm": 0.5676498413085938,
      "learning_rate": 6.346021712244057e-06,
      "loss": 0.9028,
      "step": 4974
    },
    {
      "epoch": 1.3667078697981045,
      "grad_norm": 0.5068097114562988,
      "learning_rate": 6.343273326920435e-06,
      "loss": 0.6487,
      "step": 4975
    },
    {
      "epoch": 1.3669825573410246,
      "grad_norm": 0.36618006229400635,
      "learning_rate": 6.340524941596813e-06,
      "loss": 0.9096,
      "step": 4976
    },
    {
      "epoch": 1.3672572448839446,
      "grad_norm": 0.5696470737457275,
      "learning_rate": 6.33777655627319e-06,
      "loss": 0.6084,
      "step": 4977
    },
    {
      "epoch": 1.3675319324268644,
      "grad_norm": 0.6743308305740356,
      "learning_rate": 6.335028170949568e-06,
      "loss": 0.8925,
      "step": 4978
    },
    {
      "epoch": 1.3678066199697843,
      "grad_norm": 0.3920530676841736,
      "learning_rate": 6.332279785625946e-06,
      "loss": 0.6115,
      "step": 4979
    },
    {
      "epoch": 1.3680813075127043,
      "grad_norm": 0.427355021238327,
      "learning_rate": 6.3295314003023235e-06,
      "loss": 0.9086,
      "step": 4980
    },
    {
      "epoch": 1.3683559950556243,
      "grad_norm": 0.31983014941215515,
      "learning_rate": 6.3267830149787006e-06,
      "loss": 0.835,
      "step": 4981
    },
    {
      "epoch": 1.3686306825985441,
      "grad_norm": 0.44458162784576416,
      "learning_rate": 6.324034629655078e-06,
      "loss": 0.7534,
      "step": 4982
    },
    {
      "epoch": 1.368905370141464,
      "grad_norm": 0.3875182867050171,
      "learning_rate": 6.321286244331456e-06,
      "loss": 0.8083,
      "step": 4983
    },
    {
      "epoch": 1.369180057684384,
      "grad_norm": 0.4949052631855011,
      "learning_rate": 6.318537859007834e-06,
      "loss": 0.6904,
      "step": 4984
    },
    {
      "epoch": 1.369454745227304,
      "grad_norm": 0.5062209963798523,
      "learning_rate": 6.31578947368421e-06,
      "loss": 0.9493,
      "step": 4985
    },
    {
      "epoch": 1.3697294327702239,
      "grad_norm": 0.4487903416156769,
      "learning_rate": 6.313041088360588e-06,
      "loss": 0.6351,
      "step": 4986
    },
    {
      "epoch": 1.3700041203131437,
      "grad_norm": 0.41258931159973145,
      "learning_rate": 6.310292703036967e-06,
      "loss": 0.5069,
      "step": 4987
    },
    {
      "epoch": 1.3702788078560637,
      "grad_norm": 0.42891305685043335,
      "learning_rate": 6.307544317713345e-06,
      "loss": 0.9204,
      "step": 4988
    },
    {
      "epoch": 1.3705534953989837,
      "grad_norm": 0.45081987977027893,
      "learning_rate": 6.304795932389721e-06,
      "loss": 0.6897,
      "step": 4989
    },
    {
      "epoch": 1.3708281829419036,
      "grad_norm": 0.5515851378440857,
      "learning_rate": 6.302047547066099e-06,
      "loss": 0.6871,
      "step": 4990
    },
    {
      "epoch": 1.3711028704848234,
      "grad_norm": 0.4539810121059418,
      "learning_rate": 6.299299161742477e-06,
      "loss": 0.87,
      "step": 4991
    },
    {
      "epoch": 1.3713775580277434,
      "grad_norm": 0.4657541811466217,
      "learning_rate": 6.2965507764188546e-06,
      "loss": 0.6946,
      "step": 4992
    },
    {
      "epoch": 1.3716522455706635,
      "grad_norm": 0.5145890712738037,
      "learning_rate": 6.293802391095232e-06,
      "loss": 0.7495,
      "step": 4993
    },
    {
      "epoch": 1.3719269331135833,
      "grad_norm": 0.5265249013900757,
      "learning_rate": 6.2910540057716094e-06,
      "loss": 0.6939,
      "step": 4994
    },
    {
      "epoch": 1.3722016206565033,
      "grad_norm": 0.5206893086433411,
      "learning_rate": 6.288305620447987e-06,
      "loss": 0.8603,
      "step": 4995
    },
    {
      "epoch": 1.3724763081994231,
      "grad_norm": 0.5375477075576782,
      "learning_rate": 6.285557235124365e-06,
      "loss": 0.7944,
      "step": 4996
    },
    {
      "epoch": 1.3727509957423432,
      "grad_norm": 0.3572283089160919,
      "learning_rate": 6.282808849800742e-06,
      "loss": 0.7309,
      "step": 4997
    },
    {
      "epoch": 1.373025683285263,
      "grad_norm": 0.34535932540893555,
      "learning_rate": 6.28006046447712e-06,
      "loss": 0.6796,
      "step": 4998
    },
    {
      "epoch": 1.373300370828183,
      "grad_norm": 0.34907981753349304,
      "learning_rate": 6.277312079153498e-06,
      "loss": 0.833,
      "step": 4999
    },
    {
      "epoch": 1.3735750583711028,
      "grad_norm": 0.41091740131378174,
      "learning_rate": 6.274563693829876e-06,
      "loss": 0.5477,
      "step": 5000
    },
    {
      "epoch": 1.3738497459140229,
      "grad_norm": 0.4742948114871979,
      "learning_rate": 6.271815308506253e-06,
      "loss": 0.6499,
      "step": 5001
    },
    {
      "epoch": 1.3741244334569427,
      "grad_norm": 0.4108510911464691,
      "learning_rate": 6.269066923182631e-06,
      "loss": 0.8513,
      "step": 5002
    },
    {
      "epoch": 1.3743991209998627,
      "grad_norm": 0.380500465631485,
      "learning_rate": 6.2663185378590086e-06,
      "loss": 1.079,
      "step": 5003
    },
    {
      "epoch": 1.3746738085427825,
      "grad_norm": 0.4313468039035797,
      "learning_rate": 6.2635701525353864e-06,
      "loss": 0.8443,
      "step": 5004
    },
    {
      "epoch": 1.3749484960857026,
      "grad_norm": 0.5176811814308167,
      "learning_rate": 6.2608217672117635e-06,
      "loss": 0.8008,
      "step": 5005
    },
    {
      "epoch": 1.3752231836286224,
      "grad_norm": 0.45252540707588196,
      "learning_rate": 6.258073381888141e-06,
      "loss": 0.9593,
      "step": 5006
    },
    {
      "epoch": 1.3754978711715424,
      "grad_norm": 0.5200440883636475,
      "learning_rate": 6.255324996564519e-06,
      "loss": 0.8572,
      "step": 5007
    },
    {
      "epoch": 1.3757725587144622,
      "grad_norm": 0.4547209143638611,
      "learning_rate": 6.252576611240897e-06,
      "loss": 0.6732,
      "step": 5008
    },
    {
      "epoch": 1.3760472462573823,
      "grad_norm": 0.41927602887153625,
      "learning_rate": 6.249828225917274e-06,
      "loss": 0.7251,
      "step": 5009
    },
    {
      "epoch": 1.376321933800302,
      "grad_norm": 0.4020504355430603,
      "learning_rate": 6.247079840593652e-06,
      "loss": 0.6316,
      "step": 5010
    },
    {
      "epoch": 1.3765966213432221,
      "grad_norm": 0.5015017986297607,
      "learning_rate": 6.24433145527003e-06,
      "loss": 0.7066,
      "step": 5011
    },
    {
      "epoch": 1.376871308886142,
      "grad_norm": 0.3192053437232971,
      "learning_rate": 6.241583069946408e-06,
      "loss": 0.8673,
      "step": 5012
    },
    {
      "epoch": 1.377145996429062,
      "grad_norm": 0.5175102949142456,
      "learning_rate": 6.238834684622785e-06,
      "loss": 0.8404,
      "step": 5013
    },
    {
      "epoch": 1.3774206839719818,
      "grad_norm": 0.39353257417678833,
      "learning_rate": 6.2360862992991626e-06,
      "loss": 0.7398,
      "step": 5014
    },
    {
      "epoch": 1.3776953715149018,
      "grad_norm": 0.4928176701068878,
      "learning_rate": 6.2333379139755404e-06,
      "loss": 0.6152,
      "step": 5015
    },
    {
      "epoch": 1.3779700590578217,
      "grad_norm": 0.621627688407898,
      "learning_rate": 6.230589528651918e-06,
      "loss": 0.8422,
      "step": 5016
    },
    {
      "epoch": 1.3782447466007417,
      "grad_norm": 0.4394063651561737,
      "learning_rate": 6.2278411433282945e-06,
      "loss": 0.6715,
      "step": 5017
    },
    {
      "epoch": 1.3785194341436615,
      "grad_norm": 0.46305325627326965,
      "learning_rate": 6.225092758004672e-06,
      "loss": 0.8909,
      "step": 5018
    },
    {
      "epoch": 1.3787941216865816,
      "grad_norm": 0.4931298792362213,
      "learning_rate": 6.22234437268105e-06,
      "loss": 0.7838,
      "step": 5019
    },
    {
      "epoch": 1.3790688092295014,
      "grad_norm": 0.38729143142700195,
      "learning_rate": 6.219595987357429e-06,
      "loss": 0.9105,
      "step": 5020
    },
    {
      "epoch": 1.3793434967724214,
      "grad_norm": 0.4796910881996155,
      "learning_rate": 6.216847602033805e-06,
      "loss": 0.8353,
      "step": 5021
    },
    {
      "epoch": 1.3796181843153412,
      "grad_norm": 0.41322022676467896,
      "learning_rate": 6.214099216710183e-06,
      "loss": 0.918,
      "step": 5022
    },
    {
      "epoch": 1.3798928718582613,
      "grad_norm": 0.4220471680164337,
      "learning_rate": 6.211350831386561e-06,
      "loss": 0.704,
      "step": 5023
    },
    {
      "epoch": 1.380167559401181,
      "grad_norm": 0.42026594281196594,
      "learning_rate": 6.208602446062939e-06,
      "loss": 0.766,
      "step": 5024
    },
    {
      "epoch": 1.3804422469441011,
      "grad_norm": 0.560882568359375,
      "learning_rate": 6.205854060739316e-06,
      "loss": 0.8218,
      "step": 5025
    },
    {
      "epoch": 1.380716934487021,
      "grad_norm": 0.3842681050300598,
      "learning_rate": 6.203105675415694e-06,
      "loss": 0.9054,
      "step": 5026
    },
    {
      "epoch": 1.380991622029941,
      "grad_norm": 0.45010703802108765,
      "learning_rate": 6.2003572900920715e-06,
      "loss": 0.7685,
      "step": 5027
    },
    {
      "epoch": 1.3812663095728608,
      "grad_norm": 0.328060507774353,
      "learning_rate": 6.197608904768449e-06,
      "loss": 0.7092,
      "step": 5028
    },
    {
      "epoch": 1.3815409971157808,
      "grad_norm": 0.46480292081832886,
      "learning_rate": 6.194860519444826e-06,
      "loss": 0.7397,
      "step": 5029
    },
    {
      "epoch": 1.3818156846587009,
      "grad_norm": 0.5664406418800354,
      "learning_rate": 6.192112134121204e-06,
      "loss": 0.6207,
      "step": 5030
    },
    {
      "epoch": 1.3820903722016207,
      "grad_norm": 0.4489409923553467,
      "learning_rate": 6.189363748797582e-06,
      "loss": 0.8686,
      "step": 5031
    },
    {
      "epoch": 1.3823650597445405,
      "grad_norm": 0.43398091197013855,
      "learning_rate": 6.18661536347396e-06,
      "loss": 0.6135,
      "step": 5032
    },
    {
      "epoch": 1.3826397472874605,
      "grad_norm": 0.6333518028259277,
      "learning_rate": 6.183866978150337e-06,
      "loss": 0.8992,
      "step": 5033
    },
    {
      "epoch": 1.3829144348303806,
      "grad_norm": 0.642058253288269,
      "learning_rate": 6.181118592826715e-06,
      "loss": 0.6488,
      "step": 5034
    },
    {
      "epoch": 1.3831891223733004,
      "grad_norm": 0.49662864208221436,
      "learning_rate": 6.178370207503093e-06,
      "loss": 0.6847,
      "step": 5035
    },
    {
      "epoch": 1.3834638099162202,
      "grad_norm": 0.4001329243183136,
      "learning_rate": 6.175621822179471e-06,
      "loss": 0.8103,
      "step": 5036
    },
    {
      "epoch": 1.3837384974591402,
      "grad_norm": 0.4891694486141205,
      "learning_rate": 6.172873436855848e-06,
      "loss": 0.9473,
      "step": 5037
    },
    {
      "epoch": 1.3840131850020603,
      "grad_norm": 0.4459743797779083,
      "learning_rate": 6.1701250515322255e-06,
      "loss": 0.7607,
      "step": 5038
    },
    {
      "epoch": 1.38428787254498,
      "grad_norm": 0.48559585213661194,
      "learning_rate": 6.167376666208603e-06,
      "loss": 0.8797,
      "step": 5039
    },
    {
      "epoch": 1.3845625600879,
      "grad_norm": 0.5461605787277222,
      "learning_rate": 6.164628280884981e-06,
      "loss": 0.7511,
      "step": 5040
    },
    {
      "epoch": 1.38483724763082,
      "grad_norm": 0.36183249950408936,
      "learning_rate": 6.161879895561358e-06,
      "loss": 0.6813,
      "step": 5041
    },
    {
      "epoch": 1.38511193517374,
      "grad_norm": 0.4334149956703186,
      "learning_rate": 6.159131510237736e-06,
      "loss": 1.1169,
      "step": 5042
    },
    {
      "epoch": 1.3853866227166598,
      "grad_norm": 0.47109946608543396,
      "learning_rate": 6.156383124914114e-06,
      "loss": 0.8389,
      "step": 5043
    },
    {
      "epoch": 1.3856613102595796,
      "grad_norm": 0.3901388943195343,
      "learning_rate": 6.153634739590492e-06,
      "loss": 0.6567,
      "step": 5044
    },
    {
      "epoch": 1.3859359978024997,
      "grad_norm": 0.5398979187011719,
      "learning_rate": 6.150886354266868e-06,
      "loss": 0.6279,
      "step": 5045
    },
    {
      "epoch": 1.3862106853454197,
      "grad_norm": 0.37640970945358276,
      "learning_rate": 6.148137968943247e-06,
      "loss": 0.5803,
      "step": 5046
    },
    {
      "epoch": 1.3864853728883395,
      "grad_norm": 0.4661823809146881,
      "learning_rate": 6.145389583619625e-06,
      "loss": 0.936,
      "step": 5047
    },
    {
      "epoch": 1.3867600604312593,
      "grad_norm": 0.37705540657043457,
      "learning_rate": 6.1426411982960024e-06,
      "loss": 0.7859,
      "step": 5048
    },
    {
      "epoch": 1.3870347479741794,
      "grad_norm": 0.43941766023635864,
      "learning_rate": 6.139892812972379e-06,
      "loss": 0.5726,
      "step": 5049
    },
    {
      "epoch": 1.3873094355170994,
      "grad_norm": 0.5439789891242981,
      "learning_rate": 6.1371444276487565e-06,
      "loss": 0.8103,
      "step": 5050
    },
    {
      "epoch": 1.3875841230600192,
      "grad_norm": 0.5071553587913513,
      "learning_rate": 6.134396042325134e-06,
      "loss": 0.8311,
      "step": 5051
    },
    {
      "epoch": 1.387858810602939,
      "grad_norm": 0.6041333079338074,
      "learning_rate": 6.131647657001513e-06,
      "loss": 0.6341,
      "step": 5052
    },
    {
      "epoch": 1.388133498145859,
      "grad_norm": 0.38318386673927307,
      "learning_rate": 6.128899271677889e-06,
      "loss": 0.923,
      "step": 5053
    },
    {
      "epoch": 1.388408185688779,
      "grad_norm": 0.41656655073165894,
      "learning_rate": 6.126150886354267e-06,
      "loss": 0.5734,
      "step": 5054
    },
    {
      "epoch": 1.388682873231699,
      "grad_norm": 0.37265118956565857,
      "learning_rate": 6.123402501030645e-06,
      "loss": 0.8107,
      "step": 5055
    },
    {
      "epoch": 1.3889575607746187,
      "grad_norm": 0.4914625287055969,
      "learning_rate": 6.120654115707023e-06,
      "loss": 0.9717,
      "step": 5056
    },
    {
      "epoch": 1.3892322483175388,
      "grad_norm": 0.41186273097991943,
      "learning_rate": 6.1179057303834e-06,
      "loss": 0.8603,
      "step": 5057
    },
    {
      "epoch": 1.3895069358604588,
      "grad_norm": 0.4015207588672638,
      "learning_rate": 6.115157345059778e-06,
      "loss": 0.8201,
      "step": 5058
    },
    {
      "epoch": 1.3897816234033786,
      "grad_norm": 0.48986053466796875,
      "learning_rate": 6.112408959736156e-06,
      "loss": 0.5696,
      "step": 5059
    },
    {
      "epoch": 1.3900563109462987,
      "grad_norm": 0.5577121376991272,
      "learning_rate": 6.1096605744125335e-06,
      "loss": 0.8366,
      "step": 5060
    },
    {
      "epoch": 1.3903309984892185,
      "grad_norm": 0.6225824356079102,
      "learning_rate": 6.1069121890889105e-06,
      "loss": 0.6131,
      "step": 5061
    },
    {
      "epoch": 1.3906056860321385,
      "grad_norm": 0.5069544911384583,
      "learning_rate": 6.104163803765288e-06,
      "loss": 0.7752,
      "step": 5062
    },
    {
      "epoch": 1.3908803735750583,
      "grad_norm": 0.5027193427085876,
      "learning_rate": 6.101415418441666e-06,
      "loss": 0.6608,
      "step": 5063
    },
    {
      "epoch": 1.3911550611179784,
      "grad_norm": 0.33285778760910034,
      "learning_rate": 6.098667033118044e-06,
      "loss": 0.6141,
      "step": 5064
    },
    {
      "epoch": 1.3914297486608982,
      "grad_norm": 0.4549674093723297,
      "learning_rate": 6.095918647794421e-06,
      "loss": 0.6462,
      "step": 5065
    },
    {
      "epoch": 1.3917044362038182,
      "grad_norm": 0.6112603545188904,
      "learning_rate": 6.093170262470799e-06,
      "loss": 0.8007,
      "step": 5066
    },
    {
      "epoch": 1.391979123746738,
      "grad_norm": 0.5985937714576721,
      "learning_rate": 6.090421877147177e-06,
      "loss": 0.9382,
      "step": 5067
    },
    {
      "epoch": 1.392253811289658,
      "grad_norm": 0.43717169761657715,
      "learning_rate": 6.087673491823555e-06,
      "loss": 0.8607,
      "step": 5068
    },
    {
      "epoch": 1.392528498832578,
      "grad_norm": 0.34315165877342224,
      "learning_rate": 6.084925106499932e-06,
      "loss": 0.9617,
      "step": 5069
    },
    {
      "epoch": 1.392803186375498,
      "grad_norm": 0.45607611536979675,
      "learning_rate": 6.08217672117631e-06,
      "loss": 0.672,
      "step": 5070
    },
    {
      "epoch": 1.3930778739184178,
      "grad_norm": 0.36479026079177856,
      "learning_rate": 6.0794283358526875e-06,
      "loss": 0.7076,
      "step": 5071
    },
    {
      "epoch": 1.3933525614613378,
      "grad_norm": 0.3847627639770508,
      "learning_rate": 6.0766799505290645e-06,
      "loss": 0.7907,
      "step": 5072
    },
    {
      "epoch": 1.3936272490042576,
      "grad_norm": 0.4473748803138733,
      "learning_rate": 6.073931565205442e-06,
      "loss": 0.6479,
      "step": 5073
    },
    {
      "epoch": 1.3939019365471776,
      "grad_norm": 0.3574783205986023,
      "learning_rate": 6.07118317988182e-06,
      "loss": 0.6755,
      "step": 5074
    },
    {
      "epoch": 1.3941766240900975,
      "grad_norm": 0.451612263917923,
      "learning_rate": 6.068434794558198e-06,
      "loss": 0.957,
      "step": 5075
    },
    {
      "epoch": 1.3944513116330175,
      "grad_norm": 0.4053654074668884,
      "learning_rate": 6.065686409234574e-06,
      "loss": 0.8547,
      "step": 5076
    },
    {
      "epoch": 1.3947259991759373,
      "grad_norm": 0.4217512905597687,
      "learning_rate": 6.062938023910952e-06,
      "loss": 0.6106,
      "step": 5077
    },
    {
      "epoch": 1.3950006867188574,
      "grad_norm": 0.37033611536026,
      "learning_rate": 6.060189638587331e-06,
      "loss": 0.7496,
      "step": 5078
    },
    {
      "epoch": 1.3952753742617772,
      "grad_norm": 0.4991113841533661,
      "learning_rate": 6.057441253263709e-06,
      "loss": 0.7104,
      "step": 5079
    },
    {
      "epoch": 1.3955500618046972,
      "grad_norm": 0.4023694396018982,
      "learning_rate": 6.054692867940085e-06,
      "loss": 0.9069,
      "step": 5080
    },
    {
      "epoch": 1.395824749347617,
      "grad_norm": 0.5251494646072388,
      "learning_rate": 6.051944482616463e-06,
      "loss": 1.111,
      "step": 5081
    },
    {
      "epoch": 1.396099436890537,
      "grad_norm": 0.5540185570716858,
      "learning_rate": 6.049196097292841e-06,
      "loss": 0.8418,
      "step": 5082
    },
    {
      "epoch": 1.3963741244334569,
      "grad_norm": 0.43935754895210266,
      "learning_rate": 6.0464477119692185e-06,
      "loss": 0.8438,
      "step": 5083
    },
    {
      "epoch": 1.396648811976377,
      "grad_norm": 0.3752208948135376,
      "learning_rate": 6.0436993266455955e-06,
      "loss": 0.9317,
      "step": 5084
    },
    {
      "epoch": 1.3969234995192967,
      "grad_norm": 0.43614283204078674,
      "learning_rate": 6.040950941321973e-06,
      "loss": 0.8326,
      "step": 5085
    },
    {
      "epoch": 1.3971981870622168,
      "grad_norm": 0.4287751019001007,
      "learning_rate": 6.038202555998351e-06,
      "loss": 0.919,
      "step": 5086
    },
    {
      "epoch": 1.3974728746051366,
      "grad_norm": 0.40126803517341614,
      "learning_rate": 6.035454170674729e-06,
      "loss": 1.0144,
      "step": 5087
    },
    {
      "epoch": 1.3977475621480566,
      "grad_norm": 0.36518868803977966,
      "learning_rate": 6.032705785351106e-06,
      "loss": 0.8746,
      "step": 5088
    },
    {
      "epoch": 1.3980222496909764,
      "grad_norm": 0.49406588077545166,
      "learning_rate": 6.029957400027484e-06,
      "loss": 0.7705,
      "step": 5089
    },
    {
      "epoch": 1.3982969372338965,
      "grad_norm": 0.42571520805358887,
      "learning_rate": 6.027209014703862e-06,
      "loss": 0.4893,
      "step": 5090
    },
    {
      "epoch": 1.3985716247768163,
      "grad_norm": 0.5242347717285156,
      "learning_rate": 6.02446062938024e-06,
      "loss": 0.7136,
      "step": 5091
    },
    {
      "epoch": 1.3988463123197363,
      "grad_norm": 0.48355185985565186,
      "learning_rate": 6.021712244056617e-06,
      "loss": 0.7289,
      "step": 5092
    },
    {
      "epoch": 1.3991209998626561,
      "grad_norm": 0.3942504823207855,
      "learning_rate": 6.018963858732995e-06,
      "loss": 0.7896,
      "step": 5093
    },
    {
      "epoch": 1.3993956874055762,
      "grad_norm": 0.44668036699295044,
      "learning_rate": 6.0162154734093725e-06,
      "loss": 0.7175,
      "step": 5094
    },
    {
      "epoch": 1.3996703749484962,
      "grad_norm": 0.46826764941215515,
      "learning_rate": 6.01346708808575e-06,
      "loss": 0.5342,
      "step": 5095
    },
    {
      "epoch": 1.399945062491416,
      "grad_norm": 0.30839043855667114,
      "learning_rate": 6.010718702762127e-06,
      "loss": 1.0242,
      "step": 5096
    },
    {
      "epoch": 1.4002197500343359,
      "grad_norm": 0.5148879885673523,
      "learning_rate": 6.007970317438505e-06,
      "loss": 0.8148,
      "step": 5097
    },
    {
      "epoch": 1.400494437577256,
      "grad_norm": 0.4652245044708252,
      "learning_rate": 6.005221932114883e-06,
      "loss": 0.8081,
      "step": 5098
    },
    {
      "epoch": 1.400769125120176,
      "grad_norm": 0.45428070425987244,
      "learning_rate": 6.002473546791261e-06,
      "loss": 0.675,
      "step": 5099
    },
    {
      "epoch": 1.4010438126630957,
      "grad_norm": 0.5197811126708984,
      "learning_rate": 5.999725161467638e-06,
      "loss": 0.7234,
      "step": 5100
    },
    {
      "epoch": 1.4013185002060156,
      "grad_norm": 0.4934731125831604,
      "learning_rate": 5.996976776144016e-06,
      "loss": 0.6246,
      "step": 5101
    },
    {
      "epoch": 1.4015931877489356,
      "grad_norm": 0.3705839216709137,
      "learning_rate": 5.994228390820394e-06,
      "loss": 0.8922,
      "step": 5102
    },
    {
      "epoch": 1.4018678752918556,
      "grad_norm": 0.4190366864204407,
      "learning_rate": 5.991480005496772e-06,
      "loss": 0.5948,
      "step": 5103
    },
    {
      "epoch": 1.4021425628347755,
      "grad_norm": 0.47617393732070923,
      "learning_rate": 5.988731620173149e-06,
      "loss": 0.777,
      "step": 5104
    },
    {
      "epoch": 1.4024172503776953,
      "grad_norm": 0.40457549691200256,
      "learning_rate": 5.9859832348495265e-06,
      "loss": 1.0215,
      "step": 5105
    },
    {
      "epoch": 1.4026919379206153,
      "grad_norm": 0.34991979598999023,
      "learning_rate": 5.983234849525904e-06,
      "loss": 0.804,
      "step": 5106
    },
    {
      "epoch": 1.4029666254635353,
      "grad_norm": 0.4800872206687927,
      "learning_rate": 5.980486464202282e-06,
      "loss": 0.7776,
      "step": 5107
    },
    {
      "epoch": 1.4032413130064552,
      "grad_norm": 0.4621385335922241,
      "learning_rate": 5.9777380788786584e-06,
      "loss": 0.9051,
      "step": 5108
    },
    {
      "epoch": 1.403516000549375,
      "grad_norm": 0.5090343952178955,
      "learning_rate": 5.974989693555036e-06,
      "loss": 0.8141,
      "step": 5109
    },
    {
      "epoch": 1.403790688092295,
      "grad_norm": 0.3494471311569214,
      "learning_rate": 5.972241308231414e-06,
      "loss": 0.8056,
      "step": 5110
    },
    {
      "epoch": 1.404065375635215,
      "grad_norm": 0.4380625784397125,
      "learning_rate": 5.969492922907793e-06,
      "loss": 0.9053,
      "step": 5111
    },
    {
      "epoch": 1.4043400631781349,
      "grad_norm": 0.4188384413719177,
      "learning_rate": 5.966744537584169e-06,
      "loss": 0.9101,
      "step": 5112
    },
    {
      "epoch": 1.4046147507210547,
      "grad_norm": 0.4318563640117645,
      "learning_rate": 5.963996152260547e-06,
      "loss": 0.6646,
      "step": 5113
    },
    {
      "epoch": 1.4048894382639747,
      "grad_norm": 0.3405383229255676,
      "learning_rate": 5.961247766936925e-06,
      "loss": 0.6877,
      "step": 5114
    },
    {
      "epoch": 1.4051641258068948,
      "grad_norm": 0.49115893244743347,
      "learning_rate": 5.958499381613303e-06,
      "loss": 0.7735,
      "step": 5115
    },
    {
      "epoch": 1.4054388133498146,
      "grad_norm": 0.4717424809932709,
      "learning_rate": 5.95575099628968e-06,
      "loss": 0.838,
      "step": 5116
    },
    {
      "epoch": 1.4057135008927344,
      "grad_norm": 0.42680269479751587,
      "learning_rate": 5.9530026109660575e-06,
      "loss": 0.5939,
      "step": 5117
    },
    {
      "epoch": 1.4059881884356544,
      "grad_norm": 0.4599684774875641,
      "learning_rate": 5.950254225642435e-06,
      "loss": 0.8139,
      "step": 5118
    },
    {
      "epoch": 1.4062628759785745,
      "grad_norm": 0.34455475211143494,
      "learning_rate": 5.947505840318813e-06,
      "loss": 0.5999,
      "step": 5119
    },
    {
      "epoch": 1.4065375635214943,
      "grad_norm": 0.3551925718784332,
      "learning_rate": 5.94475745499519e-06,
      "loss": 0.8495,
      "step": 5120
    },
    {
      "epoch": 1.406812251064414,
      "grad_norm": 0.33606237173080444,
      "learning_rate": 5.942009069671568e-06,
      "loss": 0.8906,
      "step": 5121
    },
    {
      "epoch": 1.4070869386073341,
      "grad_norm": 0.3514360785484314,
      "learning_rate": 5.939260684347946e-06,
      "loss": 0.7892,
      "step": 5122
    },
    {
      "epoch": 1.4073616261502542,
      "grad_norm": 0.47679904103279114,
      "learning_rate": 5.936512299024324e-06,
      "loss": 0.8928,
      "step": 5123
    },
    {
      "epoch": 1.407636313693174,
      "grad_norm": 0.3798239231109619,
      "learning_rate": 5.933763913700701e-06,
      "loss": 0.7547,
      "step": 5124
    },
    {
      "epoch": 1.4079110012360938,
      "grad_norm": 0.45598310232162476,
      "learning_rate": 5.931015528377079e-06,
      "loss": 0.6785,
      "step": 5125
    },
    {
      "epoch": 1.4081856887790138,
      "grad_norm": 0.514822244644165,
      "learning_rate": 5.928267143053457e-06,
      "loss": 0.8531,
      "step": 5126
    },
    {
      "epoch": 1.4084603763219339,
      "grad_norm": 0.48211243748664856,
      "learning_rate": 5.9255187577298345e-06,
      "loss": 0.8409,
      "step": 5127
    },
    {
      "epoch": 1.4087350638648537,
      "grad_norm": 0.4752773642539978,
      "learning_rate": 5.9227703724062115e-06,
      "loss": 0.7942,
      "step": 5128
    },
    {
      "epoch": 1.4090097514077737,
      "grad_norm": 0.5246528387069702,
      "learning_rate": 5.920021987082589e-06,
      "loss": 0.6222,
      "step": 5129
    },
    {
      "epoch": 1.4092844389506936,
      "grad_norm": 0.45610153675079346,
      "learning_rate": 5.917273601758967e-06,
      "loss": 0.7741,
      "step": 5130
    },
    {
      "epoch": 1.4095591264936136,
      "grad_norm": 0.36854520440101624,
      "learning_rate": 5.914525216435345e-06,
      "loss": 0.9178,
      "step": 5131
    },
    {
      "epoch": 1.4098338140365334,
      "grad_norm": 0.4005920886993408,
      "learning_rate": 5.911776831111722e-06,
      "loss": 0.6938,
      "step": 5132
    },
    {
      "epoch": 1.4101085015794534,
      "grad_norm": 0.4209333062171936,
      "learning_rate": 5.9090284457881e-06,
      "loss": 0.6877,
      "step": 5133
    },
    {
      "epoch": 1.4103831891223733,
      "grad_norm": 0.5066643953323364,
      "learning_rate": 5.906280060464478e-06,
      "loss": 0.8424,
      "step": 5134
    },
    {
      "epoch": 1.4106578766652933,
      "grad_norm": 0.37763580679893494,
      "learning_rate": 5.903531675140856e-06,
      "loss": 0.7482,
      "step": 5135
    },
    {
      "epoch": 1.4109325642082131,
      "grad_norm": 0.452955037355423,
      "learning_rate": 5.900783289817232e-06,
      "loss": 0.7034,
      "step": 5136
    },
    {
      "epoch": 1.4112072517511332,
      "grad_norm": 0.5085922479629517,
      "learning_rate": 5.898034904493611e-06,
      "loss": 0.725,
      "step": 5137
    },
    {
      "epoch": 1.411481939294053,
      "grad_norm": 0.42204803228378296,
      "learning_rate": 5.8952865191699885e-06,
      "loss": 0.8536,
      "step": 5138
    },
    {
      "epoch": 1.411756626836973,
      "grad_norm": 0.377482533454895,
      "learning_rate": 5.892538133846366e-06,
      "loss": 0.6191,
      "step": 5139
    },
    {
      "epoch": 1.4120313143798928,
      "grad_norm": 0.4072999656200409,
      "learning_rate": 5.8897897485227426e-06,
      "loss": 0.6827,
      "step": 5140
    },
    {
      "epoch": 1.4123060019228129,
      "grad_norm": 0.4565677046775818,
      "learning_rate": 5.8870413631991204e-06,
      "loss": 0.6892,
      "step": 5141
    },
    {
      "epoch": 1.4125806894657327,
      "grad_norm": 0.49992871284484863,
      "learning_rate": 5.884292977875498e-06,
      "loss": 0.9244,
      "step": 5142
    },
    {
      "epoch": 1.4128553770086527,
      "grad_norm": 0.49834054708480835,
      "learning_rate": 5.881544592551876e-06,
      "loss": 0.6048,
      "step": 5143
    },
    {
      "epoch": 1.4131300645515725,
      "grad_norm": 0.5108608603477478,
      "learning_rate": 5.878796207228253e-06,
      "loss": 0.8426,
      "step": 5144
    },
    {
      "epoch": 1.4134047520944926,
      "grad_norm": 0.332543283700943,
      "learning_rate": 5.876047821904631e-06,
      "loss": 0.6303,
      "step": 5145
    },
    {
      "epoch": 1.4136794396374124,
      "grad_norm": 0.39825552701950073,
      "learning_rate": 5.873299436581009e-06,
      "loss": 0.9427,
      "step": 5146
    },
    {
      "epoch": 1.4139541271803324,
      "grad_norm": 0.4003773331642151,
      "learning_rate": 5.870551051257387e-06,
      "loss": 0.6933,
      "step": 5147
    },
    {
      "epoch": 1.4142288147232522,
      "grad_norm": 0.4808642268180847,
      "learning_rate": 5.867802665933764e-06,
      "loss": 0.7579,
      "step": 5148
    },
    {
      "epoch": 1.4145035022661723,
      "grad_norm": 0.5106691122055054,
      "learning_rate": 5.865054280610142e-06,
      "loss": 0.7849,
      "step": 5149
    },
    {
      "epoch": 1.414778189809092,
      "grad_norm": 0.5874350666999817,
      "learning_rate": 5.8623058952865196e-06,
      "loss": 0.7133,
      "step": 5150
    },
    {
      "epoch": 1.4150528773520121,
      "grad_norm": 0.47367605566978455,
      "learning_rate": 5.8595575099628974e-06,
      "loss": 0.8967,
      "step": 5151
    },
    {
      "epoch": 1.415327564894932,
      "grad_norm": 0.3825845420360565,
      "learning_rate": 5.8568091246392744e-06,
      "loss": 0.521,
      "step": 5152
    },
    {
      "epoch": 1.415602252437852,
      "grad_norm": 0.34429848194122314,
      "learning_rate": 5.854060739315652e-06,
      "loss": 0.7426,
      "step": 5153
    },
    {
      "epoch": 1.4158769399807718,
      "grad_norm": 0.48428937792778015,
      "learning_rate": 5.85131235399203e-06,
      "loss": 0.7855,
      "step": 5154
    },
    {
      "epoch": 1.4161516275236918,
      "grad_norm": 0.5365732908248901,
      "learning_rate": 5.848563968668408e-06,
      "loss": 0.6018,
      "step": 5155
    },
    {
      "epoch": 1.4164263150666117,
      "grad_norm": 0.43369218707084656,
      "learning_rate": 5.845815583344785e-06,
      "loss": 0.6704,
      "step": 5156
    },
    {
      "epoch": 1.4167010026095317,
      "grad_norm": 0.4134613275527954,
      "learning_rate": 5.843067198021163e-06,
      "loss": 0.8879,
      "step": 5157
    },
    {
      "epoch": 1.4169756901524515,
      "grad_norm": 0.45938655734062195,
      "learning_rate": 5.840318812697541e-06,
      "loss": 0.6918,
      "step": 5158
    },
    {
      "epoch": 1.4172503776953715,
      "grad_norm": 0.49212175607681274,
      "learning_rate": 5.837570427373919e-06,
      "loss": 0.7941,
      "step": 5159
    },
    {
      "epoch": 1.4175250652382916,
      "grad_norm": 0.49456173181533813,
      "learning_rate": 5.834822042050296e-06,
      "loss": 0.753,
      "step": 5160
    },
    {
      "epoch": 1.4177997527812114,
      "grad_norm": 0.48951849341392517,
      "learning_rate": 5.8320736567266736e-06,
      "loss": 0.8117,
      "step": 5161
    },
    {
      "epoch": 1.4180744403241312,
      "grad_norm": 0.4406321048736572,
      "learning_rate": 5.8293252714030514e-06,
      "loss": 0.7732,
      "step": 5162
    },
    {
      "epoch": 1.4183491278670513,
      "grad_norm": 0.41353511810302734,
      "learning_rate": 5.826576886079429e-06,
      "loss": 0.6557,
      "step": 5163
    },
    {
      "epoch": 1.4186238154099713,
      "grad_norm": 0.3773764371871948,
      "learning_rate": 5.823828500755806e-06,
      "loss": 0.7566,
      "step": 5164
    },
    {
      "epoch": 1.418898502952891,
      "grad_norm": 0.39127251505851746,
      "learning_rate": 5.821080115432184e-06,
      "loss": 0.7613,
      "step": 5165
    },
    {
      "epoch": 1.419173190495811,
      "grad_norm": 0.40069761872291565,
      "learning_rate": 5.818331730108562e-06,
      "loss": 0.7981,
      "step": 5166
    },
    {
      "epoch": 1.419447878038731,
      "grad_norm": 0.5707579851150513,
      "learning_rate": 5.81558334478494e-06,
      "loss": 0.8434,
      "step": 5167
    },
    {
      "epoch": 1.419722565581651,
      "grad_norm": 0.42584043741226196,
      "learning_rate": 5.812834959461316e-06,
      "loss": 0.6994,
      "step": 5168
    },
    {
      "epoch": 1.4199972531245708,
      "grad_norm": 0.4836571514606476,
      "learning_rate": 5.810086574137695e-06,
      "loss": 0.6463,
      "step": 5169
    },
    {
      "epoch": 1.4202719406674906,
      "grad_norm": 0.6673811674118042,
      "learning_rate": 5.807338188814073e-06,
      "loss": 0.646,
      "step": 5170
    },
    {
      "epoch": 1.4205466282104107,
      "grad_norm": 0.5091736912727356,
      "learning_rate": 5.8045898034904505e-06,
      "loss": 0.9261,
      "step": 5171
    },
    {
      "epoch": 1.4208213157533307,
      "grad_norm": 0.5719352960586548,
      "learning_rate": 5.801841418166827e-06,
      "loss": 0.9482,
      "step": 5172
    },
    {
      "epoch": 1.4210960032962505,
      "grad_norm": 0.37194979190826416,
      "learning_rate": 5.799093032843205e-06,
      "loss": 0.613,
      "step": 5173
    },
    {
      "epoch": 1.4213706908391703,
      "grad_norm": 0.5382468104362488,
      "learning_rate": 5.7963446475195825e-06,
      "loss": 0.6973,
      "step": 5174
    },
    {
      "epoch": 1.4216453783820904,
      "grad_norm": 0.5012110471725464,
      "learning_rate": 5.79359626219596e-06,
      "loss": 1.0335,
      "step": 5175
    },
    {
      "epoch": 1.4219200659250104,
      "grad_norm": 0.47468024492263794,
      "learning_rate": 5.790847876872337e-06,
      "loss": 0.6558,
      "step": 5176
    },
    {
      "epoch": 1.4221947534679302,
      "grad_norm": 0.48589593172073364,
      "learning_rate": 5.788099491548715e-06,
      "loss": 0.9025,
      "step": 5177
    },
    {
      "epoch": 1.42246944101085,
      "grad_norm": 0.5198528170585632,
      "learning_rate": 5.785351106225093e-06,
      "loss": 0.5515,
      "step": 5178
    },
    {
      "epoch": 1.42274412855377,
      "grad_norm": 0.4547770619392395,
      "learning_rate": 5.782602720901471e-06,
      "loss": 0.9385,
      "step": 5179
    },
    {
      "epoch": 1.4230188160966901,
      "grad_norm": 0.38904085755348206,
      "learning_rate": 5.779854335577848e-06,
      "loss": 1.076,
      "step": 5180
    },
    {
      "epoch": 1.42329350363961,
      "grad_norm": 0.45804107189178467,
      "learning_rate": 5.777105950254226e-06,
      "loss": 0.9743,
      "step": 5181
    },
    {
      "epoch": 1.4235681911825298,
      "grad_norm": 0.3750498592853546,
      "learning_rate": 5.774357564930604e-06,
      "loss": 0.4991,
      "step": 5182
    },
    {
      "epoch": 1.4238428787254498,
      "grad_norm": 0.4146634340286255,
      "learning_rate": 5.7716091796069816e-06,
      "loss": 0.9885,
      "step": 5183
    },
    {
      "epoch": 1.4241175662683698,
      "grad_norm": 0.4730978012084961,
      "learning_rate": 5.768860794283359e-06,
      "loss": 0.8466,
      "step": 5184
    },
    {
      "epoch": 1.4243922538112896,
      "grad_norm": 0.49704137444496155,
      "learning_rate": 5.7661124089597365e-06,
      "loss": 0.9291,
      "step": 5185
    },
    {
      "epoch": 1.4246669413542095,
      "grad_norm": 0.5855399370193481,
      "learning_rate": 5.763364023636114e-06,
      "loss": 0.8476,
      "step": 5186
    },
    {
      "epoch": 1.4249416288971295,
      "grad_norm": 0.5280583500862122,
      "learning_rate": 5.760615638312492e-06,
      "loss": 0.7416,
      "step": 5187
    },
    {
      "epoch": 1.4252163164400495,
      "grad_norm": 0.5794113874435425,
      "learning_rate": 5.757867252988869e-06,
      "loss": 0.8362,
      "step": 5188
    },
    {
      "epoch": 1.4254910039829694,
      "grad_norm": 0.35483574867248535,
      "learning_rate": 5.755118867665247e-06,
      "loss": 0.8581,
      "step": 5189
    },
    {
      "epoch": 1.4257656915258892,
      "grad_norm": 0.3738589882850647,
      "learning_rate": 5.752370482341625e-06,
      "loss": 0.6572,
      "step": 5190
    },
    {
      "epoch": 1.4260403790688092,
      "grad_norm": 0.37670648097991943,
      "learning_rate": 5.749622097018003e-06,
      "loss": 0.6858,
      "step": 5191
    },
    {
      "epoch": 1.4263150666117292,
      "grad_norm": 0.419594943523407,
      "learning_rate": 5.74687371169438e-06,
      "loss": 0.5052,
      "step": 5192
    },
    {
      "epoch": 1.426589754154649,
      "grad_norm": 0.4891067147254944,
      "learning_rate": 5.744125326370758e-06,
      "loss": 0.5831,
      "step": 5193
    },
    {
      "epoch": 1.426864441697569,
      "grad_norm": 0.57086580991745,
      "learning_rate": 5.7413769410471356e-06,
      "loss": 0.6686,
      "step": 5194
    },
    {
      "epoch": 1.427139129240489,
      "grad_norm": 0.4057351052761078,
      "learning_rate": 5.7386285557235134e-06,
      "loss": 0.898,
      "step": 5195
    },
    {
      "epoch": 1.427413816783409,
      "grad_norm": 0.5440675616264343,
      "learning_rate": 5.7358801703998905e-06,
      "loss": 0.7954,
      "step": 5196
    },
    {
      "epoch": 1.4276885043263288,
      "grad_norm": 0.5159278512001038,
      "learning_rate": 5.733131785076268e-06,
      "loss": 0.863,
      "step": 5197
    },
    {
      "epoch": 1.4279631918692488,
      "grad_norm": 0.47904348373413086,
      "learning_rate": 5.730383399752646e-06,
      "loss": 0.6171,
      "step": 5198
    },
    {
      "epoch": 1.4282378794121686,
      "grad_norm": 0.43527668714523315,
      "learning_rate": 5.727635014429024e-06,
      "loss": 0.9302,
      "step": 5199
    },
    {
      "epoch": 1.4285125669550887,
      "grad_norm": 0.5548836588859558,
      "learning_rate": 5.7248866291054e-06,
      "loss": 0.7732,
      "step": 5200
    },
    {
      "epoch": 1.4287872544980085,
      "grad_norm": 0.47256070375442505,
      "learning_rate": 5.722138243781778e-06,
      "loss": 0.8318,
      "step": 5201
    },
    {
      "epoch": 1.4290619420409285,
      "grad_norm": 0.5841325521469116,
      "learning_rate": 5.719389858458157e-06,
      "loss": 0.8075,
      "step": 5202
    },
    {
      "epoch": 1.4293366295838483,
      "grad_norm": 0.42504483461380005,
      "learning_rate": 5.716641473134535e-06,
      "loss": 0.6667,
      "step": 5203
    },
    {
      "epoch": 1.4296113171267684,
      "grad_norm": 0.3208548426628113,
      "learning_rate": 5.713893087810911e-06,
      "loss": 0.9179,
      "step": 5204
    },
    {
      "epoch": 1.4298860046696882,
      "grad_norm": 0.49384942650794983,
      "learning_rate": 5.711144702487289e-06,
      "loss": 0.6765,
      "step": 5205
    },
    {
      "epoch": 1.4301606922126082,
      "grad_norm": 0.5493744611740112,
      "learning_rate": 5.708396317163667e-06,
      "loss": 0.8862,
      "step": 5206
    },
    {
      "epoch": 1.430435379755528,
      "grad_norm": 0.6060923933982849,
      "learning_rate": 5.7056479318400445e-06,
      "loss": 0.6422,
      "step": 5207
    },
    {
      "epoch": 1.430710067298448,
      "grad_norm": 0.6128448843955994,
      "learning_rate": 5.7028995465164215e-06,
      "loss": 0.8581,
      "step": 5208
    },
    {
      "epoch": 1.430984754841368,
      "grad_norm": 0.45196133852005005,
      "learning_rate": 5.700151161192799e-06,
      "loss": 0.9332,
      "step": 5209
    },
    {
      "epoch": 1.431259442384288,
      "grad_norm": 0.5246838331222534,
      "learning_rate": 5.697402775869177e-06,
      "loss": 0.8582,
      "step": 5210
    },
    {
      "epoch": 1.4315341299272077,
      "grad_norm": 0.3236776888370514,
      "learning_rate": 5.694654390545555e-06,
      "loss": 0.9209,
      "step": 5211
    },
    {
      "epoch": 1.4318088174701278,
      "grad_norm": 0.4224337339401245,
      "learning_rate": 5.691906005221932e-06,
      "loss": 0.9099,
      "step": 5212
    },
    {
      "epoch": 1.4320835050130476,
      "grad_norm": 0.35315629839897156,
      "learning_rate": 5.68915761989831e-06,
      "loss": 0.7067,
      "step": 5213
    },
    {
      "epoch": 1.4323581925559676,
      "grad_norm": 0.5184800624847412,
      "learning_rate": 5.686409234574688e-06,
      "loss": 0.7826,
      "step": 5214
    },
    {
      "epoch": 1.4326328800988875,
      "grad_norm": 0.29439058899879456,
      "learning_rate": 5.683660849251066e-06,
      "loss": 0.7746,
      "step": 5215
    },
    {
      "epoch": 1.4329075676418075,
      "grad_norm": 0.49953725934028625,
      "learning_rate": 5.680912463927443e-06,
      "loss": 0.7255,
      "step": 5216
    },
    {
      "epoch": 1.4331822551847273,
      "grad_norm": 0.38382014632225037,
      "learning_rate": 5.678164078603821e-06,
      "loss": 0.7321,
      "step": 5217
    },
    {
      "epoch": 1.4334569427276473,
      "grad_norm": 0.41161009669303894,
      "learning_rate": 5.6754156932801985e-06,
      "loss": 0.8408,
      "step": 5218
    },
    {
      "epoch": 1.4337316302705672,
      "grad_norm": 0.37695369124412537,
      "learning_rate": 5.672667307956576e-06,
      "loss": 0.9004,
      "step": 5219
    },
    {
      "epoch": 1.4340063178134872,
      "grad_norm": 0.5571134686470032,
      "learning_rate": 5.669918922632953e-06,
      "loss": 0.8348,
      "step": 5220
    },
    {
      "epoch": 1.434281005356407,
      "grad_norm": 0.36402472853660583,
      "learning_rate": 5.667170537309331e-06,
      "loss": 0.8542,
      "step": 5221
    },
    {
      "epoch": 1.434555692899327,
      "grad_norm": 0.3572477400302887,
      "learning_rate": 5.664422151985709e-06,
      "loss": 0.518,
      "step": 5222
    },
    {
      "epoch": 1.4348303804422469,
      "grad_norm": 0.36760812997817993,
      "learning_rate": 5.661673766662087e-06,
      "loss": 0.6892,
      "step": 5223
    },
    {
      "epoch": 1.435105067985167,
      "grad_norm": 0.4241592586040497,
      "learning_rate": 5.658925381338464e-06,
      "loss": 0.7039,
      "step": 5224
    },
    {
      "epoch": 1.4353797555280867,
      "grad_norm": 0.4492739140987396,
      "learning_rate": 5.656176996014842e-06,
      "loss": 0.8097,
      "step": 5225
    },
    {
      "epoch": 1.4356544430710068,
      "grad_norm": 0.3754887282848358,
      "learning_rate": 5.65342861069122e-06,
      "loss": 0.5358,
      "step": 5226
    },
    {
      "epoch": 1.4359291306139266,
      "grad_norm": 0.40682253241539,
      "learning_rate": 5.650680225367598e-06,
      "loss": 0.6968,
      "step": 5227
    },
    {
      "epoch": 1.4362038181568466,
      "grad_norm": 0.3963531255722046,
      "learning_rate": 5.647931840043975e-06,
      "loss": 0.8612,
      "step": 5228
    },
    {
      "epoch": 1.4364785056997667,
      "grad_norm": 0.604079008102417,
      "learning_rate": 5.6451834547203525e-06,
      "loss": 0.8999,
      "step": 5229
    },
    {
      "epoch": 1.4367531932426865,
      "grad_norm": 0.3992083668708801,
      "learning_rate": 5.64243506939673e-06,
      "loss": 0.9514,
      "step": 5230
    },
    {
      "epoch": 1.4370278807856063,
      "grad_norm": 0.43577006459236145,
      "learning_rate": 5.639686684073108e-06,
      "loss": 0.9539,
      "step": 5231
    },
    {
      "epoch": 1.4373025683285263,
      "grad_norm": 0.3694327175617218,
      "learning_rate": 5.636938298749484e-06,
      "loss": 0.856,
      "step": 5232
    },
    {
      "epoch": 1.4375772558714464,
      "grad_norm": 0.49002566933631897,
      "learning_rate": 5.634189913425862e-06,
      "loss": 0.7197,
      "step": 5233
    },
    {
      "epoch": 1.4378519434143662,
      "grad_norm": 0.3923005163669586,
      "learning_rate": 5.63144152810224e-06,
      "loss": 0.6764,
      "step": 5234
    },
    {
      "epoch": 1.438126630957286,
      "grad_norm": 0.45694831013679504,
      "learning_rate": 5.628693142778619e-06,
      "loss": 0.9344,
      "step": 5235
    },
    {
      "epoch": 1.438401318500206,
      "grad_norm": 0.5214462876319885,
      "learning_rate": 5.625944757454995e-06,
      "loss": 0.8924,
      "step": 5236
    },
    {
      "epoch": 1.438676006043126,
      "grad_norm": 0.466590017080307,
      "learning_rate": 5.623196372131373e-06,
      "loss": 0.865,
      "step": 5237
    },
    {
      "epoch": 1.4389506935860459,
      "grad_norm": 0.41561299562454224,
      "learning_rate": 5.620447986807751e-06,
      "loss": 0.9024,
      "step": 5238
    },
    {
      "epoch": 1.4392253811289657,
      "grad_norm": 0.38628044724464417,
      "learning_rate": 5.617699601484129e-06,
      "loss": 0.8743,
      "step": 5239
    },
    {
      "epoch": 1.4395000686718857,
      "grad_norm": 0.5373130440711975,
      "learning_rate": 5.614951216160506e-06,
      "loss": 0.8152,
      "step": 5240
    },
    {
      "epoch": 1.4397747562148058,
      "grad_norm": 0.5861842632293701,
      "learning_rate": 5.6122028308368835e-06,
      "loss": 0.7945,
      "step": 5241
    },
    {
      "epoch": 1.4400494437577256,
      "grad_norm": 0.4903906285762787,
      "learning_rate": 5.609454445513261e-06,
      "loss": 0.8889,
      "step": 5242
    },
    {
      "epoch": 1.4403241313006454,
      "grad_norm": 0.5149016976356506,
      "learning_rate": 5.606706060189639e-06,
      "loss": 0.7694,
      "step": 5243
    },
    {
      "epoch": 1.4405988188435654,
      "grad_norm": 0.4219248294830322,
      "learning_rate": 5.603957674866016e-06,
      "loss": 0.7243,
      "step": 5244
    },
    {
      "epoch": 1.4408735063864855,
      "grad_norm": 0.3877178430557251,
      "learning_rate": 5.601209289542394e-06,
      "loss": 0.9138,
      "step": 5245
    },
    {
      "epoch": 1.4411481939294053,
      "grad_norm": 0.4042515158653259,
      "learning_rate": 5.598460904218772e-06,
      "loss": 1.0325,
      "step": 5246
    },
    {
      "epoch": 1.4414228814723251,
      "grad_norm": 0.3626459240913391,
      "learning_rate": 5.59571251889515e-06,
      "loss": 0.5842,
      "step": 5247
    },
    {
      "epoch": 1.4416975690152452,
      "grad_norm": 0.3521331548690796,
      "learning_rate": 5.592964133571527e-06,
      "loss": 0.4887,
      "step": 5248
    },
    {
      "epoch": 1.4419722565581652,
      "grad_norm": 0.31134429574012756,
      "learning_rate": 5.590215748247905e-06,
      "loss": 0.985,
      "step": 5249
    },
    {
      "epoch": 1.442246944101085,
      "grad_norm": 0.4576683044433594,
      "learning_rate": 5.587467362924283e-06,
      "loss": 0.603,
      "step": 5250
    },
    {
      "epoch": 1.4425216316440048,
      "grad_norm": 0.4517243802547455,
      "learning_rate": 5.5847189776006605e-06,
      "loss": 0.6025,
      "step": 5251
    },
    {
      "epoch": 1.4427963191869249,
      "grad_norm": 0.44267237186431885,
      "learning_rate": 5.5819705922770375e-06,
      "loss": 0.6473,
      "step": 5252
    },
    {
      "epoch": 1.443071006729845,
      "grad_norm": 0.514480471611023,
      "learning_rate": 5.579222206953415e-06,
      "loss": 0.8498,
      "step": 5253
    },
    {
      "epoch": 1.4433456942727647,
      "grad_norm": 0.41541793942451477,
      "learning_rate": 5.576473821629793e-06,
      "loss": 0.5918,
      "step": 5254
    },
    {
      "epoch": 1.4436203818156845,
      "grad_norm": 0.45484182238578796,
      "learning_rate": 5.573725436306171e-06,
      "loss": 0.6974,
      "step": 5255
    },
    {
      "epoch": 1.4438950693586046,
      "grad_norm": 0.4364394247531891,
      "learning_rate": 5.570977050982548e-06,
      "loss": 0.5812,
      "step": 5256
    },
    {
      "epoch": 1.4441697569015246,
      "grad_norm": 0.5155108571052551,
      "learning_rate": 5.568228665658926e-06,
      "loss": 0.8165,
      "step": 5257
    },
    {
      "epoch": 1.4444444444444444,
      "grad_norm": 0.40523388981819153,
      "learning_rate": 5.565480280335304e-06,
      "loss": 0.6464,
      "step": 5258
    },
    {
      "epoch": 1.4447191319873642,
      "grad_norm": 0.3657989799976349,
      "learning_rate": 5.562731895011682e-06,
      "loss": 0.8561,
      "step": 5259
    },
    {
      "epoch": 1.4449938195302843,
      "grad_norm": 0.446084201335907,
      "learning_rate": 5.559983509688058e-06,
      "loss": 0.725,
      "step": 5260
    },
    {
      "epoch": 1.4452685070732043,
      "grad_norm": 0.39225468039512634,
      "learning_rate": 5.557235124364437e-06,
      "loss": 0.8503,
      "step": 5261
    },
    {
      "epoch": 1.4455431946161241,
      "grad_norm": 0.46391570568084717,
      "learning_rate": 5.5544867390408145e-06,
      "loss": 0.8498,
      "step": 5262
    },
    {
      "epoch": 1.4458178821590442,
      "grad_norm": 0.592948853969574,
      "learning_rate": 5.551738353717192e-06,
      "loss": 0.9752,
      "step": 5263
    },
    {
      "epoch": 1.446092569701964,
      "grad_norm": 0.7058988213539124,
      "learning_rate": 5.5489899683935685e-06,
      "loss": 0.8158,
      "step": 5264
    },
    {
      "epoch": 1.446367257244884,
      "grad_norm": 0.38417404890060425,
      "learning_rate": 5.546241583069946e-06,
      "loss": 0.7839,
      "step": 5265
    },
    {
      "epoch": 1.4466419447878038,
      "grad_norm": 0.589886486530304,
      "learning_rate": 5.543493197746324e-06,
      "loss": 0.8268,
      "step": 5266
    },
    {
      "epoch": 1.4469166323307239,
      "grad_norm": 0.5500078201293945,
      "learning_rate": 5.540744812422703e-06,
      "loss": 1.01,
      "step": 5267
    },
    {
      "epoch": 1.4471913198736437,
      "grad_norm": 0.30514973402023315,
      "learning_rate": 5.537996427099079e-06,
      "loss": 0.5723,
      "step": 5268
    },
    {
      "epoch": 1.4474660074165637,
      "grad_norm": 0.5041224956512451,
      "learning_rate": 5.535248041775457e-06,
      "loss": 0.5889,
      "step": 5269
    },
    {
      "epoch": 1.4477406949594835,
      "grad_norm": 0.48664039373397827,
      "learning_rate": 5.532499656451835e-06,
      "loss": 0.5982,
      "step": 5270
    },
    {
      "epoch": 1.4480153825024036,
      "grad_norm": 0.5079206824302673,
      "learning_rate": 5.529751271128213e-06,
      "loss": 0.86,
      "step": 5271
    },
    {
      "epoch": 1.4482900700453234,
      "grad_norm": 0.5011540055274963,
      "learning_rate": 5.52700288580459e-06,
      "loss": 1.0108,
      "step": 5272
    },
    {
      "epoch": 1.4485647575882434,
      "grad_norm": 0.41748735308647156,
      "learning_rate": 5.524254500480968e-06,
      "loss": 0.6439,
      "step": 5273
    },
    {
      "epoch": 1.4488394451311633,
      "grad_norm": 0.4861394762992859,
      "learning_rate": 5.5215061151573455e-06,
      "loss": 0.7211,
      "step": 5274
    },
    {
      "epoch": 1.4491141326740833,
      "grad_norm": 0.40236690640449524,
      "learning_rate": 5.518757729833723e-06,
      "loss": 0.7548,
      "step": 5275
    },
    {
      "epoch": 1.449388820217003,
      "grad_norm": 0.4517991840839386,
      "learning_rate": 5.5160093445101e-06,
      "loss": 0.7167,
      "step": 5276
    },
    {
      "epoch": 1.4496635077599231,
      "grad_norm": 0.5046599507331848,
      "learning_rate": 5.513260959186478e-06,
      "loss": 0.8253,
      "step": 5277
    },
    {
      "epoch": 1.449938195302843,
      "grad_norm": 0.40421077609062195,
      "learning_rate": 5.510512573862856e-06,
      "loss": 0.8287,
      "step": 5278
    },
    {
      "epoch": 1.450212882845763,
      "grad_norm": 0.6273903250694275,
      "learning_rate": 5.507764188539234e-06,
      "loss": 0.8842,
      "step": 5279
    },
    {
      "epoch": 1.4504875703886828,
      "grad_norm": 0.3925383687019348,
      "learning_rate": 5.505015803215611e-06,
      "loss": 0.9175,
      "step": 5280
    },
    {
      "epoch": 1.4507622579316028,
      "grad_norm": 0.4679775536060333,
      "learning_rate": 5.502267417891989e-06,
      "loss": 0.8079,
      "step": 5281
    },
    {
      "epoch": 1.4510369454745227,
      "grad_norm": 0.5502325296401978,
      "learning_rate": 5.499519032568367e-06,
      "loss": 0.7432,
      "step": 5282
    },
    {
      "epoch": 1.4513116330174427,
      "grad_norm": 0.37806203961372375,
      "learning_rate": 5.496770647244745e-06,
      "loss": 0.5999,
      "step": 5283
    },
    {
      "epoch": 1.4515863205603625,
      "grad_norm": 0.4417617917060852,
      "learning_rate": 5.494022261921122e-06,
      "loss": 0.7197,
      "step": 5284
    },
    {
      "epoch": 1.4518610081032826,
      "grad_norm": 0.44513049721717834,
      "learning_rate": 5.4912738765974995e-06,
      "loss": 0.8419,
      "step": 5285
    },
    {
      "epoch": 1.4521356956462024,
      "grad_norm": 0.45192015171051025,
      "learning_rate": 5.488525491273877e-06,
      "loss": 0.7121,
      "step": 5286
    },
    {
      "epoch": 1.4524103831891224,
      "grad_norm": 0.4541471302509308,
      "learning_rate": 5.485777105950255e-06,
      "loss": 0.6259,
      "step": 5287
    },
    {
      "epoch": 1.4526850707320422,
      "grad_norm": 0.4229124188423157,
      "learning_rate": 5.483028720626632e-06,
      "loss": 0.6404,
      "step": 5288
    },
    {
      "epoch": 1.4529597582749623,
      "grad_norm": 0.5468158721923828,
      "learning_rate": 5.48028033530301e-06,
      "loss": 0.6643,
      "step": 5289
    },
    {
      "epoch": 1.453234445817882,
      "grad_norm": 0.461365669965744,
      "learning_rate": 5.477531949979388e-06,
      "loss": 0.7583,
      "step": 5290
    },
    {
      "epoch": 1.4535091333608021,
      "grad_norm": 0.4643438756465912,
      "learning_rate": 5.474783564655766e-06,
      "loss": 0.7087,
      "step": 5291
    },
    {
      "epoch": 1.453783820903722,
      "grad_norm": 0.42953869700431824,
      "learning_rate": 5.472035179332142e-06,
      "loss": 0.7916,
      "step": 5292
    },
    {
      "epoch": 1.454058508446642,
      "grad_norm": 0.46795353293418884,
      "learning_rate": 5.469286794008521e-06,
      "loss": 0.5602,
      "step": 5293
    },
    {
      "epoch": 1.454333195989562,
      "grad_norm": 0.4761100113391876,
      "learning_rate": 5.466538408684899e-06,
      "loss": 0.7187,
      "step": 5294
    },
    {
      "epoch": 1.4546078835324818,
      "grad_norm": 0.5503353476524353,
      "learning_rate": 5.4637900233612765e-06,
      "loss": 0.682,
      "step": 5295
    },
    {
      "epoch": 1.4548825710754016,
      "grad_norm": 0.5568586587905884,
      "learning_rate": 5.461041638037653e-06,
      "loss": 0.8296,
      "step": 5296
    },
    {
      "epoch": 1.4551572586183217,
      "grad_norm": 0.5895821452140808,
      "learning_rate": 5.4582932527140305e-06,
      "loss": 0.8763,
      "step": 5297
    },
    {
      "epoch": 1.4554319461612417,
      "grad_norm": 0.31119418144226074,
      "learning_rate": 5.455544867390408e-06,
      "loss": 0.8881,
      "step": 5298
    },
    {
      "epoch": 1.4557066337041615,
      "grad_norm": 0.369761198759079,
      "learning_rate": 5.452796482066786e-06,
      "loss": 0.999,
      "step": 5299
    },
    {
      "epoch": 1.4559813212470814,
      "grad_norm": 0.40478232502937317,
      "learning_rate": 5.450048096743163e-06,
      "loss": 0.8671,
      "step": 5300
    },
    {
      "epoch": 1.4562560087900014,
      "grad_norm": 0.48931869864463806,
      "learning_rate": 5.447299711419541e-06,
      "loss": 0.6889,
      "step": 5301
    },
    {
      "epoch": 1.4565306963329214,
      "grad_norm": 0.48732876777648926,
      "learning_rate": 5.444551326095919e-06,
      "loss": 0.7562,
      "step": 5302
    },
    {
      "epoch": 1.4568053838758412,
      "grad_norm": 0.615598201751709,
      "learning_rate": 5.441802940772297e-06,
      "loss": 0.694,
      "step": 5303
    },
    {
      "epoch": 1.457080071418761,
      "grad_norm": 0.5606292486190796,
      "learning_rate": 5.439054555448674e-06,
      "loss": 0.7309,
      "step": 5304
    },
    {
      "epoch": 1.457354758961681,
      "grad_norm": 0.5758853554725647,
      "learning_rate": 5.436306170125052e-06,
      "loss": 1.0278,
      "step": 5305
    },
    {
      "epoch": 1.4576294465046011,
      "grad_norm": 0.4276863634586334,
      "learning_rate": 5.43355778480143e-06,
      "loss": 0.8082,
      "step": 5306
    },
    {
      "epoch": 1.457904134047521,
      "grad_norm": 0.4336002469062805,
      "learning_rate": 5.4308093994778075e-06,
      "loss": 0.9955,
      "step": 5307
    },
    {
      "epoch": 1.4581788215904408,
      "grad_norm": 0.4262015223503113,
      "learning_rate": 5.4280610141541846e-06,
      "loss": 0.6124,
      "step": 5308
    },
    {
      "epoch": 1.4584535091333608,
      "grad_norm": 0.5932504534721375,
      "learning_rate": 5.425312628830562e-06,
      "loss": 0.7383,
      "step": 5309
    },
    {
      "epoch": 1.4587281966762808,
      "grad_norm": 0.3997665047645569,
      "learning_rate": 5.42256424350694e-06,
      "loss": 0.5031,
      "step": 5310
    },
    {
      "epoch": 1.4590028842192007,
      "grad_norm": 0.5458210706710815,
      "learning_rate": 5.419815858183318e-06,
      "loss": 0.9413,
      "step": 5311
    },
    {
      "epoch": 1.4592775717621205,
      "grad_norm": 0.5184202194213867,
      "learning_rate": 5.417067472859695e-06,
      "loss": 0.8969,
      "step": 5312
    },
    {
      "epoch": 1.4595522593050405,
      "grad_norm": 0.4231388568878174,
      "learning_rate": 5.414319087536073e-06,
      "loss": 0.5861,
      "step": 5313
    },
    {
      "epoch": 1.4598269468479605,
      "grad_norm": 0.5669137835502625,
      "learning_rate": 5.411570702212451e-06,
      "loss": 0.6614,
      "step": 5314
    },
    {
      "epoch": 1.4601016343908804,
      "grad_norm": 0.3940035402774811,
      "learning_rate": 5.408822316888829e-06,
      "loss": 0.6953,
      "step": 5315
    },
    {
      "epoch": 1.4603763219338002,
      "grad_norm": 0.4532488286495209,
      "learning_rate": 5.406073931565206e-06,
      "loss": 0.63,
      "step": 5316
    },
    {
      "epoch": 1.4606510094767202,
      "grad_norm": 0.39350613951683044,
      "learning_rate": 5.403325546241584e-06,
      "loss": 0.5472,
      "step": 5317
    },
    {
      "epoch": 1.4609256970196403,
      "grad_norm": 0.5048090219497681,
      "learning_rate": 5.4005771609179615e-06,
      "loss": 0.6909,
      "step": 5318
    },
    {
      "epoch": 1.46120038456256,
      "grad_norm": 0.40792933106422424,
      "learning_rate": 5.397828775594339e-06,
      "loss": 0.665,
      "step": 5319
    },
    {
      "epoch": 1.4614750721054799,
      "grad_norm": 0.5645368695259094,
      "learning_rate": 5.3950803902707164e-06,
      "loss": 0.5227,
      "step": 5320
    },
    {
      "epoch": 1.4617497596484,
      "grad_norm": 0.5576413869857788,
      "learning_rate": 5.392332004947094e-06,
      "loss": 0.702,
      "step": 5321
    },
    {
      "epoch": 1.46202444719132,
      "grad_norm": 0.3772546648979187,
      "learning_rate": 5.389583619623472e-06,
      "loss": 0.9725,
      "step": 5322
    },
    {
      "epoch": 1.4622991347342398,
      "grad_norm": 0.5760698914527893,
      "learning_rate": 5.38683523429985e-06,
      "loss": 0.866,
      "step": 5323
    },
    {
      "epoch": 1.4625738222771596,
      "grad_norm": 0.5484282374382019,
      "learning_rate": 5.384086848976226e-06,
      "loss": 1.0182,
      "step": 5324
    },
    {
      "epoch": 1.4628485098200796,
      "grad_norm": 0.4724949300289154,
      "learning_rate": 5.381338463652604e-06,
      "loss": 0.8642,
      "step": 5325
    },
    {
      "epoch": 1.4631231973629997,
      "grad_norm": 0.47787371277809143,
      "learning_rate": 5.378590078328983e-06,
      "loss": 0.7513,
      "step": 5326
    },
    {
      "epoch": 1.4633978849059195,
      "grad_norm": 0.508989691734314,
      "learning_rate": 5.375841693005361e-06,
      "loss": 0.7892,
      "step": 5327
    },
    {
      "epoch": 1.4636725724488395,
      "grad_norm": 0.5332726240158081,
      "learning_rate": 5.373093307681737e-06,
      "loss": 0.8204,
      "step": 5328
    },
    {
      "epoch": 1.4639472599917593,
      "grad_norm": 0.35693004727363586,
      "learning_rate": 5.370344922358115e-06,
      "loss": 0.5918,
      "step": 5329
    },
    {
      "epoch": 1.4642219475346794,
      "grad_norm": 0.35515856742858887,
      "learning_rate": 5.3675965370344926e-06,
      "loss": 0.5672,
      "step": 5330
    },
    {
      "epoch": 1.4644966350775992,
      "grad_norm": 0.456237256526947,
      "learning_rate": 5.3648481517108704e-06,
      "loss": 0.7051,
      "step": 5331
    },
    {
      "epoch": 1.4647713226205192,
      "grad_norm": 0.6603469848632812,
      "learning_rate": 5.3620997663872475e-06,
      "loss": 0.8125,
      "step": 5332
    },
    {
      "epoch": 1.465046010163439,
      "grad_norm": 0.47372663021087646,
      "learning_rate": 5.359351381063625e-06,
      "loss": 0.8983,
      "step": 5333
    },
    {
      "epoch": 1.465320697706359,
      "grad_norm": 0.538330078125,
      "learning_rate": 5.356602995740003e-06,
      "loss": 0.7304,
      "step": 5334
    },
    {
      "epoch": 1.465595385249279,
      "grad_norm": 0.428754061460495,
      "learning_rate": 5.353854610416381e-06,
      "loss": 0.647,
      "step": 5335
    },
    {
      "epoch": 1.465870072792199,
      "grad_norm": 0.364980548620224,
      "learning_rate": 5.351106225092758e-06,
      "loss": 0.8975,
      "step": 5336
    },
    {
      "epoch": 1.4661447603351188,
      "grad_norm": 0.4550382196903229,
      "learning_rate": 5.348357839769136e-06,
      "loss": 1.0812,
      "step": 5337
    },
    {
      "epoch": 1.4664194478780388,
      "grad_norm": 0.41411539912223816,
      "learning_rate": 5.345609454445514e-06,
      "loss": 1.0877,
      "step": 5338
    },
    {
      "epoch": 1.4666941354209586,
      "grad_norm": 0.545085072517395,
      "learning_rate": 5.342861069121892e-06,
      "loss": 0.7531,
      "step": 5339
    },
    {
      "epoch": 1.4669688229638786,
      "grad_norm": 0.5447960495948792,
      "learning_rate": 5.340112683798269e-06,
      "loss": 0.663,
      "step": 5340
    },
    {
      "epoch": 1.4672435105067985,
      "grad_norm": 0.3244243264198303,
      "learning_rate": 5.3373642984746466e-06,
      "loss": 0.7871,
      "step": 5341
    },
    {
      "epoch": 1.4675181980497185,
      "grad_norm": 0.3383794128894806,
      "learning_rate": 5.3346159131510244e-06,
      "loss": 1.0215,
      "step": 5342
    },
    {
      "epoch": 1.4677928855926383,
      "grad_norm": 0.4148211181163788,
      "learning_rate": 5.331867527827402e-06,
      "loss": 0.628,
      "step": 5343
    },
    {
      "epoch": 1.4680675731355584,
      "grad_norm": 0.5072391033172607,
      "learning_rate": 5.329119142503779e-06,
      "loss": 0.6937,
      "step": 5344
    },
    {
      "epoch": 1.4683422606784782,
      "grad_norm": 0.5497791767120361,
      "learning_rate": 5.326370757180157e-06,
      "loss": 0.7601,
      "step": 5345
    },
    {
      "epoch": 1.4686169482213982,
      "grad_norm": 0.5503769516944885,
      "learning_rate": 5.323622371856535e-06,
      "loss": 0.768,
      "step": 5346
    },
    {
      "epoch": 1.468891635764318,
      "grad_norm": 0.8402937054634094,
      "learning_rate": 5.320873986532913e-06,
      "loss": 0.8709,
      "step": 5347
    },
    {
      "epoch": 1.469166323307238,
      "grad_norm": 0.4329729676246643,
      "learning_rate": 5.31812560120929e-06,
      "loss": 0.7304,
      "step": 5348
    },
    {
      "epoch": 1.4694410108501579,
      "grad_norm": 0.4651384651660919,
      "learning_rate": 5.315377215885668e-06,
      "loss": 0.6843,
      "step": 5349
    },
    {
      "epoch": 1.469715698393078,
      "grad_norm": 0.5779486894607544,
      "learning_rate": 5.312628830562046e-06,
      "loss": 0.7628,
      "step": 5350
    },
    {
      "epoch": 1.4699903859359977,
      "grad_norm": 0.47399768233299255,
      "learning_rate": 5.3098804452384236e-06,
      "loss": 0.9355,
      "step": 5351
    },
    {
      "epoch": 1.4702650734789178,
      "grad_norm": 0.4951716363430023,
      "learning_rate": 5.3071320599148006e-06,
      "loss": 0.7004,
      "step": 5352
    },
    {
      "epoch": 1.4705397610218376,
      "grad_norm": 0.44234949350357056,
      "learning_rate": 5.3043836745911784e-06,
      "loss": 0.7536,
      "step": 5353
    },
    {
      "epoch": 1.4708144485647576,
      "grad_norm": 0.42342281341552734,
      "learning_rate": 5.301635289267556e-06,
      "loss": 0.7625,
      "step": 5354
    },
    {
      "epoch": 1.4710891361076774,
      "grad_norm": 0.38861095905303955,
      "learning_rate": 5.298886903943934e-06,
      "loss": 0.7191,
      "step": 5355
    },
    {
      "epoch": 1.4713638236505975,
      "grad_norm": 0.5400751233100891,
      "learning_rate": 5.29613851862031e-06,
      "loss": 0.6591,
      "step": 5356
    },
    {
      "epoch": 1.4716385111935173,
      "grad_norm": 0.3441079258918762,
      "learning_rate": 5.293390133296688e-06,
      "loss": 0.7946,
      "step": 5357
    },
    {
      "epoch": 1.4719131987364373,
      "grad_norm": 0.5559800267219543,
      "learning_rate": 5.290641747973067e-06,
      "loss": 0.8893,
      "step": 5358
    },
    {
      "epoch": 1.4721878862793571,
      "grad_norm": 0.41306769847869873,
      "learning_rate": 5.287893362649445e-06,
      "loss": 0.8665,
      "step": 5359
    },
    {
      "epoch": 1.4724625738222772,
      "grad_norm": 0.40270110964775085,
      "learning_rate": 5.285144977325821e-06,
      "loss": 0.8079,
      "step": 5360
    },
    {
      "epoch": 1.472737261365197,
      "grad_norm": 0.35779738426208496,
      "learning_rate": 5.282396592002199e-06,
      "loss": 0.6966,
      "step": 5361
    },
    {
      "epoch": 1.473011948908117,
      "grad_norm": 0.3401602506637573,
      "learning_rate": 5.279648206678577e-06,
      "loss": 0.8177,
      "step": 5362
    },
    {
      "epoch": 1.473286636451037,
      "grad_norm": 0.35309094190597534,
      "learning_rate": 5.276899821354955e-06,
      "loss": 0.6744,
      "step": 5363
    },
    {
      "epoch": 1.473561323993957,
      "grad_norm": 0.4061456322669983,
      "learning_rate": 5.274151436031332e-06,
      "loss": 0.6026,
      "step": 5364
    },
    {
      "epoch": 1.4738360115368767,
      "grad_norm": 0.5500430464744568,
      "learning_rate": 5.2714030507077095e-06,
      "loss": 0.8029,
      "step": 5365
    },
    {
      "epoch": 1.4741106990797967,
      "grad_norm": 0.5461018681526184,
      "learning_rate": 5.268654665384087e-06,
      "loss": 0.9736,
      "step": 5366
    },
    {
      "epoch": 1.4743853866227168,
      "grad_norm": 0.505563497543335,
      "learning_rate": 5.265906280060465e-06,
      "loss": 0.8489,
      "step": 5367
    },
    {
      "epoch": 1.4746600741656366,
      "grad_norm": 0.36116382479667664,
      "learning_rate": 5.263157894736842e-06,
      "loss": 0.6099,
      "step": 5368
    },
    {
      "epoch": 1.4749347617085564,
      "grad_norm": 0.5669950842857361,
      "learning_rate": 5.26040950941322e-06,
      "loss": 0.9658,
      "step": 5369
    },
    {
      "epoch": 1.4752094492514765,
      "grad_norm": 0.5153294205665588,
      "learning_rate": 5.257661124089598e-06,
      "loss": 0.7845,
      "step": 5370
    },
    {
      "epoch": 1.4754841367943965,
      "grad_norm": 0.4178479015827179,
      "learning_rate": 5.254912738765976e-06,
      "loss": 0.581,
      "step": 5371
    },
    {
      "epoch": 1.4757588243373163,
      "grad_norm": 0.4732635021209717,
      "learning_rate": 5.252164353442353e-06,
      "loss": 0.7415,
      "step": 5372
    },
    {
      "epoch": 1.4760335118802361,
      "grad_norm": 0.46183133125305176,
      "learning_rate": 5.249415968118731e-06,
      "loss": 0.6852,
      "step": 5373
    },
    {
      "epoch": 1.4763081994231562,
      "grad_norm": 0.43871617317199707,
      "learning_rate": 5.246667582795109e-06,
      "loss": 0.9163,
      "step": 5374
    },
    {
      "epoch": 1.4765828869660762,
      "grad_norm": 0.5581013560295105,
      "learning_rate": 5.2439191974714865e-06,
      "loss": 0.9834,
      "step": 5375
    },
    {
      "epoch": 1.476857574508996,
      "grad_norm": 0.308408260345459,
      "learning_rate": 5.2411708121478635e-06,
      "loss": 0.7786,
      "step": 5376
    },
    {
      "epoch": 1.4771322620519158,
      "grad_norm": 0.5237141251564026,
      "learning_rate": 5.238422426824241e-06,
      "loss": 0.5731,
      "step": 5377
    },
    {
      "epoch": 1.4774069495948359,
      "grad_norm": 0.3609352111816406,
      "learning_rate": 5.235674041500619e-06,
      "loss": 0.6881,
      "step": 5378
    },
    {
      "epoch": 1.477681637137756,
      "grad_norm": 0.4368118345737457,
      "learning_rate": 5.232925656176997e-06,
      "loss": 0.7119,
      "step": 5379
    },
    {
      "epoch": 1.4779563246806757,
      "grad_norm": 0.6762322187423706,
      "learning_rate": 5.230177270853374e-06,
      "loss": 0.7703,
      "step": 5380
    },
    {
      "epoch": 1.4782310122235955,
      "grad_norm": 0.40360578894615173,
      "learning_rate": 5.227428885529752e-06,
      "loss": 0.6266,
      "step": 5381
    },
    {
      "epoch": 1.4785056997665156,
      "grad_norm": 0.4232671856880188,
      "learning_rate": 5.22468050020613e-06,
      "loss": 0.7896,
      "step": 5382
    },
    {
      "epoch": 1.4787803873094356,
      "grad_norm": 0.4818667471408844,
      "learning_rate": 5.221932114882508e-06,
      "loss": 0.8317,
      "step": 5383
    },
    {
      "epoch": 1.4790550748523554,
      "grad_norm": 0.4725593030452728,
      "learning_rate": 5.219183729558885e-06,
      "loss": 0.627,
      "step": 5384
    },
    {
      "epoch": 1.4793297623952752,
      "grad_norm": 0.4183611571788788,
      "learning_rate": 5.216435344235263e-06,
      "loss": 0.538,
      "step": 5385
    },
    {
      "epoch": 1.4796044499381953,
      "grad_norm": 0.36618852615356445,
      "learning_rate": 5.2136869589116405e-06,
      "loss": 0.7591,
      "step": 5386
    },
    {
      "epoch": 1.4798791374811153,
      "grad_norm": 0.38823947310447693,
      "learning_rate": 5.210938573588017e-06,
      "loss": 0.6238,
      "step": 5387
    },
    {
      "epoch": 1.4801538250240351,
      "grad_norm": 0.4622204005718231,
      "learning_rate": 5.2081901882643945e-06,
      "loss": 0.5535,
      "step": 5388
    },
    {
      "epoch": 1.480428512566955,
      "grad_norm": 0.5577043890953064,
      "learning_rate": 5.205441802940772e-06,
      "loss": 0.7311,
      "step": 5389
    },
    {
      "epoch": 1.480703200109875,
      "grad_norm": 0.5829969048500061,
      "learning_rate": 5.20269341761715e-06,
      "loss": 0.9099,
      "step": 5390
    },
    {
      "epoch": 1.480977887652795,
      "grad_norm": 0.41943359375,
      "learning_rate": 5.199945032293527e-06,
      "loss": 0.7065,
      "step": 5391
    },
    {
      "epoch": 1.4812525751957148,
      "grad_norm": 0.45941492915153503,
      "learning_rate": 5.197196646969905e-06,
      "loss": 0.7909,
      "step": 5392
    },
    {
      "epoch": 1.4815272627386347,
      "grad_norm": 0.5035371780395508,
      "learning_rate": 5.194448261646283e-06,
      "loss": 0.8699,
      "step": 5393
    },
    {
      "epoch": 1.4818019502815547,
      "grad_norm": 0.46593207120895386,
      "learning_rate": 5.191699876322661e-06,
      "loss": 0.8626,
      "step": 5394
    },
    {
      "epoch": 1.4820766378244747,
      "grad_norm": 0.46566876769065857,
      "learning_rate": 5.188951490999038e-06,
      "loss": 0.7782,
      "step": 5395
    },
    {
      "epoch": 1.4823513253673946,
      "grad_norm": 0.43660542368888855,
      "learning_rate": 5.186203105675416e-06,
      "loss": 0.8862,
      "step": 5396
    },
    {
      "epoch": 1.4826260129103146,
      "grad_norm": 0.497550368309021,
      "learning_rate": 5.183454720351794e-06,
      "loss": 0.9298,
      "step": 5397
    },
    {
      "epoch": 1.4829007004532344,
      "grad_norm": 0.4637840688228607,
      "learning_rate": 5.1807063350281715e-06,
      "loss": 0.7556,
      "step": 5398
    },
    {
      "epoch": 1.4831753879961544,
      "grad_norm": 0.4163012206554413,
      "learning_rate": 5.1779579497045485e-06,
      "loss": 0.7969,
      "step": 5399
    },
    {
      "epoch": 1.4834500755390743,
      "grad_norm": 0.36227965354919434,
      "learning_rate": 5.175209564380926e-06,
      "loss": 0.7734,
      "step": 5400
    },
    {
      "epoch": 1.4837247630819943,
      "grad_norm": 0.39357346296310425,
      "learning_rate": 5.172461179057304e-06,
      "loss": 0.7256,
      "step": 5401
    },
    {
      "epoch": 1.4839994506249141,
      "grad_norm": 0.42741015553474426,
      "learning_rate": 5.169712793733682e-06,
      "loss": 0.6374,
      "step": 5402
    },
    {
      "epoch": 1.4842741381678342,
      "grad_norm": 0.4876604378223419,
      "learning_rate": 5.166964408410059e-06,
      "loss": 0.8902,
      "step": 5403
    },
    {
      "epoch": 1.484548825710754,
      "grad_norm": 0.4076707363128662,
      "learning_rate": 5.164216023086437e-06,
      "loss": 0.8022,
      "step": 5404
    },
    {
      "epoch": 1.484823513253674,
      "grad_norm": 0.5937197804450989,
      "learning_rate": 5.161467637762815e-06,
      "loss": 0.5999,
      "step": 5405
    },
    {
      "epoch": 1.4850982007965938,
      "grad_norm": 0.40493646264076233,
      "learning_rate": 5.158719252439193e-06,
      "loss": 0.6209,
      "step": 5406
    },
    {
      "epoch": 1.4853728883395139,
      "grad_norm": 0.5928286910057068,
      "learning_rate": 5.15597086711557e-06,
      "loss": 0.7101,
      "step": 5407
    },
    {
      "epoch": 1.4856475758824337,
      "grad_norm": 0.5112078189849854,
      "learning_rate": 5.153222481791948e-06,
      "loss": 1.054,
      "step": 5408
    },
    {
      "epoch": 1.4859222634253537,
      "grad_norm": 0.5704174041748047,
      "learning_rate": 5.1504740964683255e-06,
      "loss": 0.8871,
      "step": 5409
    },
    {
      "epoch": 1.4861969509682735,
      "grad_norm": 0.4259001314640045,
      "learning_rate": 5.147725711144703e-06,
      "loss": 0.7673,
      "step": 5410
    },
    {
      "epoch": 1.4864716385111936,
      "grad_norm": 0.38475584983825684,
      "learning_rate": 5.14497732582108e-06,
      "loss": 0.7783,
      "step": 5411
    },
    {
      "epoch": 1.4867463260541134,
      "grad_norm": 0.5142020583152771,
      "learning_rate": 5.142228940497458e-06,
      "loss": 0.8247,
      "step": 5412
    },
    {
      "epoch": 1.4870210135970334,
      "grad_norm": 0.5181363821029663,
      "learning_rate": 5.139480555173836e-06,
      "loss": 0.6401,
      "step": 5413
    },
    {
      "epoch": 1.4872957011399532,
      "grad_norm": 0.4866337180137634,
      "learning_rate": 5.136732169850214e-06,
      "loss": 0.6674,
      "step": 5414
    },
    {
      "epoch": 1.4875703886828733,
      "grad_norm": 0.3910055458545685,
      "learning_rate": 5.13398378452659e-06,
      "loss": 0.6579,
      "step": 5415
    },
    {
      "epoch": 1.487845076225793,
      "grad_norm": 0.4350123405456543,
      "learning_rate": 5.131235399202968e-06,
      "loss": 0.767,
      "step": 5416
    },
    {
      "epoch": 1.4881197637687131,
      "grad_norm": 0.4131951630115509,
      "learning_rate": 5.128487013879347e-06,
      "loss": 0.8079,
      "step": 5417
    },
    {
      "epoch": 1.488394451311633,
      "grad_norm": 0.4636397957801819,
      "learning_rate": 5.125738628555725e-06,
      "loss": 0.6539,
      "step": 5418
    },
    {
      "epoch": 1.488669138854553,
      "grad_norm": 0.5380955338478088,
      "learning_rate": 5.122990243232101e-06,
      "loss": 0.6922,
      "step": 5419
    },
    {
      "epoch": 1.4889438263974728,
      "grad_norm": 0.5332962274551392,
      "learning_rate": 5.120241857908479e-06,
      "loss": 0.6209,
      "step": 5420
    },
    {
      "epoch": 1.4892185139403928,
      "grad_norm": 0.4019162952899933,
      "learning_rate": 5.1174934725848565e-06,
      "loss": 0.682,
      "step": 5421
    },
    {
      "epoch": 1.4894932014833127,
      "grad_norm": 0.3923324942588806,
      "learning_rate": 5.114745087261234e-06,
      "loss": 0.6744,
      "step": 5422
    },
    {
      "epoch": 1.4897678890262327,
      "grad_norm": 0.4396432042121887,
      "learning_rate": 5.111996701937611e-06,
      "loss": 0.8723,
      "step": 5423
    },
    {
      "epoch": 1.4900425765691525,
      "grad_norm": 0.4224303066730499,
      "learning_rate": 5.109248316613989e-06,
      "loss": 0.6269,
      "step": 5424
    },
    {
      "epoch": 1.4903172641120725,
      "grad_norm": 0.3835831582546234,
      "learning_rate": 5.106499931290367e-06,
      "loss": 0.7294,
      "step": 5425
    },
    {
      "epoch": 1.4905919516549924,
      "grad_norm": 0.34620773792266846,
      "learning_rate": 5.103751545966745e-06,
      "loss": 0.7862,
      "step": 5426
    },
    {
      "epoch": 1.4908666391979124,
      "grad_norm": 0.6392489075660706,
      "learning_rate": 5.101003160643122e-06,
      "loss": 0.7954,
      "step": 5427
    },
    {
      "epoch": 1.4911413267408324,
      "grad_norm": 0.4733431041240692,
      "learning_rate": 5.0982547753195e-06,
      "loss": 0.731,
      "step": 5428
    },
    {
      "epoch": 1.4914160142837523,
      "grad_norm": 0.2983608543872833,
      "learning_rate": 5.095506389995878e-06,
      "loss": 0.582,
      "step": 5429
    },
    {
      "epoch": 1.491690701826672,
      "grad_norm": 0.41206681728363037,
      "learning_rate": 5.092758004672256e-06,
      "loss": 1.0259,
      "step": 5430
    },
    {
      "epoch": 1.491965389369592,
      "grad_norm": 0.4663045108318329,
      "learning_rate": 5.090009619348633e-06,
      "loss": 0.6101,
      "step": 5431
    },
    {
      "epoch": 1.4922400769125121,
      "grad_norm": 0.5056527853012085,
      "learning_rate": 5.0872612340250105e-06,
      "loss": 0.8026,
      "step": 5432
    },
    {
      "epoch": 1.492514764455432,
      "grad_norm": 0.3756473660469055,
      "learning_rate": 5.084512848701388e-06,
      "loss": 0.7266,
      "step": 5433
    },
    {
      "epoch": 1.4927894519983518,
      "grad_norm": 0.4345811903476715,
      "learning_rate": 5.081764463377766e-06,
      "loss": 0.8924,
      "step": 5434
    },
    {
      "epoch": 1.4930641395412718,
      "grad_norm": 0.48516845703125,
      "learning_rate": 5.079016078054143e-06,
      "loss": 0.8454,
      "step": 5435
    },
    {
      "epoch": 1.4933388270841919,
      "grad_norm": 0.4734036922454834,
      "learning_rate": 5.076267692730521e-06,
      "loss": 0.8115,
      "step": 5436
    },
    {
      "epoch": 1.4936135146271117,
      "grad_norm": 0.4000236392021179,
      "learning_rate": 5.073519307406899e-06,
      "loss": 0.7211,
      "step": 5437
    },
    {
      "epoch": 1.4938882021700315,
      "grad_norm": 0.5165472626686096,
      "learning_rate": 5.070770922083277e-06,
      "loss": 0.9365,
      "step": 5438
    },
    {
      "epoch": 1.4941628897129515,
      "grad_norm": 0.5281567573547363,
      "learning_rate": 5.068022536759654e-06,
      "loss": 0.6671,
      "step": 5439
    },
    {
      "epoch": 1.4944375772558716,
      "grad_norm": 0.44905179738998413,
      "learning_rate": 5.065274151436032e-06,
      "loss": 0.7617,
      "step": 5440
    },
    {
      "epoch": 1.4947122647987914,
      "grad_norm": 0.44474416971206665,
      "learning_rate": 5.06252576611241e-06,
      "loss": 0.9487,
      "step": 5441
    },
    {
      "epoch": 1.4949869523417112,
      "grad_norm": 0.3914794623851776,
      "learning_rate": 5.0597773807887875e-06,
      "loss": 1.0414,
      "step": 5442
    },
    {
      "epoch": 1.4952616398846312,
      "grad_norm": 0.4910506010055542,
      "learning_rate": 5.0570289954651645e-06,
      "loss": 0.758,
      "step": 5443
    },
    {
      "epoch": 1.4955363274275513,
      "grad_norm": 0.475253701210022,
      "learning_rate": 5.054280610141542e-06,
      "loss": 0.8137,
      "step": 5444
    },
    {
      "epoch": 1.495811014970471,
      "grad_norm": 0.42678171396255493,
      "learning_rate": 5.05153222481792e-06,
      "loss": 0.7854,
      "step": 5445
    },
    {
      "epoch": 1.496085702513391,
      "grad_norm": 0.4131637513637543,
      "learning_rate": 5.048783839494298e-06,
      "loss": 0.8042,
      "step": 5446
    },
    {
      "epoch": 1.496360390056311,
      "grad_norm": 0.38258659839630127,
      "learning_rate": 5.046035454170674e-06,
      "loss": 0.661,
      "step": 5447
    },
    {
      "epoch": 1.496635077599231,
      "grad_norm": 0.4898398518562317,
      "learning_rate": 5.043287068847052e-06,
      "loss": 0.7613,
      "step": 5448
    },
    {
      "epoch": 1.4969097651421508,
      "grad_norm": 0.5800288915634155,
      "learning_rate": 5.04053868352343e-06,
      "loss": 0.756,
      "step": 5449
    },
    {
      "epoch": 1.4971844526850706,
      "grad_norm": 0.3969309628009796,
      "learning_rate": 5.037790298199809e-06,
      "loss": 0.9372,
      "step": 5450
    },
    {
      "epoch": 1.4974591402279906,
      "grad_norm": 0.3880132734775543,
      "learning_rate": 5.035041912876185e-06,
      "loss": 0.8522,
      "step": 5451
    },
    {
      "epoch": 1.4977338277709107,
      "grad_norm": 0.5293967723846436,
      "learning_rate": 5.032293527552563e-06,
      "loss": 0.7606,
      "step": 5452
    },
    {
      "epoch": 1.4980085153138305,
      "grad_norm": 0.3579423725605011,
      "learning_rate": 5.029545142228941e-06,
      "loss": 0.5951,
      "step": 5453
    },
    {
      "epoch": 1.4982832028567503,
      "grad_norm": 0.4671151638031006,
      "learning_rate": 5.0267967569053185e-06,
      "loss": 0.7167,
      "step": 5454
    },
    {
      "epoch": 1.4985578903996704,
      "grad_norm": 0.5408095717430115,
      "learning_rate": 5.0240483715816955e-06,
      "loss": 0.5711,
      "step": 5455
    },
    {
      "epoch": 1.4988325779425904,
      "grad_norm": 0.39074909687042236,
      "learning_rate": 5.021299986258073e-06,
      "loss": 0.8017,
      "step": 5456
    },
    {
      "epoch": 1.4991072654855102,
      "grad_norm": 0.3886057436466217,
      "learning_rate": 5.018551600934451e-06,
      "loss": 0.7324,
      "step": 5457
    },
    {
      "epoch": 1.49938195302843,
      "grad_norm": 0.4958634376525879,
      "learning_rate": 5.015803215610829e-06,
      "loss": 0.8401,
      "step": 5458
    },
    {
      "epoch": 1.49965664057135,
      "grad_norm": 0.40522927045822144,
      "learning_rate": 5.013054830287206e-06,
      "loss": 0.7052,
      "step": 5459
    },
    {
      "epoch": 1.49993132811427,
      "grad_norm": 0.4248192608356476,
      "learning_rate": 5.010306444963584e-06,
      "loss": 0.7341,
      "step": 5460
    },
    {
      "epoch": 1.50020601565719,
      "grad_norm": 0.38761311769485474,
      "learning_rate": 5.007558059639962e-06,
      "loss": 0.722,
      "step": 5461
    },
    {
      "epoch": 1.5004807032001097,
      "grad_norm": 0.4701148271560669,
      "learning_rate": 5.00480967431634e-06,
      "loss": 0.761,
      "step": 5462
    },
    {
      "epoch": 1.5007553907430298,
      "grad_norm": 0.5767157077789307,
      "learning_rate": 5.002061288992717e-06,
      "loss": 0.8113,
      "step": 5463
    },
    {
      "epoch": 1.5010300782859498,
      "grad_norm": 0.5434339642524719,
      "learning_rate": 4.999312903669095e-06,
      "loss": 0.7289,
      "step": 5464
    },
    {
      "epoch": 1.5013047658288696,
      "grad_norm": 0.44853660464286804,
      "learning_rate": 4.9965645183454725e-06,
      "loss": 0.7767,
      "step": 5465
    },
    {
      "epoch": 1.5015794533717894,
      "grad_norm": 0.5234737396240234,
      "learning_rate": 4.9938161330218496e-06,
      "loss": 0.8706,
      "step": 5466
    },
    {
      "epoch": 1.5018541409147095,
      "grad_norm": 0.45487266778945923,
      "learning_rate": 4.991067747698227e-06,
      "loss": 0.685,
      "step": 5467
    },
    {
      "epoch": 1.5021288284576295,
      "grad_norm": 0.40602973103523254,
      "learning_rate": 4.988319362374605e-06,
      "loss": 0.7104,
      "step": 5468
    },
    {
      "epoch": 1.5024035160005493,
      "grad_norm": 0.5173311233520508,
      "learning_rate": 4.985570977050983e-06,
      "loss": 0.9448,
      "step": 5469
    },
    {
      "epoch": 1.5026782035434691,
      "grad_norm": 0.4152161478996277,
      "learning_rate": 4.98282259172736e-06,
      "loss": 0.79,
      "step": 5470
    },
    {
      "epoch": 1.5029528910863892,
      "grad_norm": 0.5663454532623291,
      "learning_rate": 4.980074206403738e-06,
      "loss": 1.0384,
      "step": 5471
    },
    {
      "epoch": 1.5032275786293092,
      "grad_norm": 0.46266528964042664,
      "learning_rate": 4.977325821080116e-06,
      "loss": 0.9114,
      "step": 5472
    },
    {
      "epoch": 1.503502266172229,
      "grad_norm": 0.36643803119659424,
      "learning_rate": 4.974577435756494e-06,
      "loss": 0.7513,
      "step": 5473
    },
    {
      "epoch": 1.5037769537151489,
      "grad_norm": 0.5141777992248535,
      "learning_rate": 4.971829050432871e-06,
      "loss": 0.5876,
      "step": 5474
    },
    {
      "epoch": 1.504051641258069,
      "grad_norm": 0.5137673020362854,
      "learning_rate": 4.969080665109249e-06,
      "loss": 0.6326,
      "step": 5475
    },
    {
      "epoch": 1.504326328800989,
      "grad_norm": 0.39933696389198303,
      "learning_rate": 4.9663322797856265e-06,
      "loss": 0.8823,
      "step": 5476
    },
    {
      "epoch": 1.504601016343909,
      "grad_norm": 0.46700847148895264,
      "learning_rate": 4.963583894462004e-06,
      "loss": 0.9301,
      "step": 5477
    },
    {
      "epoch": 1.5048757038868288,
      "grad_norm": 0.5491647720336914,
      "learning_rate": 4.9608355091383814e-06,
      "loss": 0.7893,
      "step": 5478
    },
    {
      "epoch": 1.5051503914297486,
      "grad_norm": 0.45979467034339905,
      "learning_rate": 4.958087123814759e-06,
      "loss": 0.8631,
      "step": 5479
    },
    {
      "epoch": 1.5054250789726686,
      "grad_norm": 0.48543989658355713,
      "learning_rate": 4.955338738491136e-06,
      "loss": 0.6795,
      "step": 5480
    },
    {
      "epoch": 1.5056997665155887,
      "grad_norm": 0.4170393943786621,
      "learning_rate": 4.952590353167514e-06,
      "loss": 0.7601,
      "step": 5481
    },
    {
      "epoch": 1.5059744540585085,
      "grad_norm": 0.5819951891899109,
      "learning_rate": 4.949841967843892e-06,
      "loss": 0.8384,
      "step": 5482
    },
    {
      "epoch": 1.5062491416014283,
      "grad_norm": 0.6044164896011353,
      "learning_rate": 4.94709358252027e-06,
      "loss": 0.8178,
      "step": 5483
    },
    {
      "epoch": 1.5065238291443483,
      "grad_norm": 0.4424345791339874,
      "learning_rate": 4.944345197196647e-06,
      "loss": 0.6853,
      "step": 5484
    },
    {
      "epoch": 1.5067985166872684,
      "grad_norm": 0.5499774217605591,
      "learning_rate": 4.941596811873025e-06,
      "loss": 0.7001,
      "step": 5485
    },
    {
      "epoch": 1.5070732042301882,
      "grad_norm": 0.4000018239021301,
      "learning_rate": 4.938848426549403e-06,
      "loss": 0.7251,
      "step": 5486
    },
    {
      "epoch": 1.507347891773108,
      "grad_norm": 0.37080100178718567,
      "learning_rate": 4.9361000412257805e-06,
      "loss": 0.5943,
      "step": 5487
    },
    {
      "epoch": 1.507622579316028,
      "grad_norm": 0.4914731979370117,
      "learning_rate": 4.9333516559021576e-06,
      "loss": 0.8564,
      "step": 5488
    },
    {
      "epoch": 1.507897266858948,
      "grad_norm": 0.3909311890602112,
      "learning_rate": 4.9306032705785354e-06,
      "loss": 0.5357,
      "step": 5489
    },
    {
      "epoch": 1.508171954401868,
      "grad_norm": 0.36872398853302,
      "learning_rate": 4.927854885254913e-06,
      "loss": 0.7966,
      "step": 5490
    },
    {
      "epoch": 1.5084466419447877,
      "grad_norm": 0.4378650486469269,
      "learning_rate": 4.925106499931291e-06,
      "loss": 0.8757,
      "step": 5491
    },
    {
      "epoch": 1.5087213294877078,
      "grad_norm": 0.5165675282478333,
      "learning_rate": 4.922358114607668e-06,
      "loss": 0.9705,
      "step": 5492
    },
    {
      "epoch": 1.5089960170306278,
      "grad_norm": 0.4312388300895691,
      "learning_rate": 4.919609729284046e-06,
      "loss": 0.5999,
      "step": 5493
    },
    {
      "epoch": 1.5092707045735476,
      "grad_norm": 0.412106454372406,
      "learning_rate": 4.916861343960423e-06,
      "loss": 0.6861,
      "step": 5494
    },
    {
      "epoch": 1.5095453921164674,
      "grad_norm": 0.4015195369720459,
      "learning_rate": 4.914112958636802e-06,
      "loss": 0.7646,
      "step": 5495
    },
    {
      "epoch": 1.5098200796593875,
      "grad_norm": 0.49584469199180603,
      "learning_rate": 4.911364573313179e-06,
      "loss": 0.6752,
      "step": 5496
    },
    {
      "epoch": 1.5100947672023075,
      "grad_norm": 0.49366307258605957,
      "learning_rate": 4.908616187989557e-06,
      "loss": 0.9724,
      "step": 5497
    },
    {
      "epoch": 1.5103694547452273,
      "grad_norm": 0.5350185036659241,
      "learning_rate": 4.905867802665934e-06,
      "loss": 0.5018,
      "step": 5498
    },
    {
      "epoch": 1.5106441422881471,
      "grad_norm": 0.4943978786468506,
      "learning_rate": 4.9031194173423116e-06,
      "loss": 0.742,
      "step": 5499
    },
    {
      "epoch": 1.5109188298310672,
      "grad_norm": 0.418623685836792,
      "learning_rate": 4.9003710320186894e-06,
      "loss": 0.643,
      "step": 5500
    },
    {
      "epoch": 1.5111935173739872,
      "grad_norm": 0.4620453417301178,
      "learning_rate": 4.897622646695067e-06,
      "loss": 0.6261,
      "step": 5501
    },
    {
      "epoch": 1.511468204916907,
      "grad_norm": 0.44007158279418945,
      "learning_rate": 4.894874261371444e-06,
      "loss": 0.8918,
      "step": 5502
    },
    {
      "epoch": 1.5117428924598268,
      "grad_norm": 0.31314200162887573,
      "learning_rate": 4.892125876047822e-06,
      "loss": 0.7409,
      "step": 5503
    },
    {
      "epoch": 1.5120175800027469,
      "grad_norm": 0.5403143763542175,
      "learning_rate": 4.8893774907242e-06,
      "loss": 0.679,
      "step": 5504
    },
    {
      "epoch": 1.512292267545667,
      "grad_norm": 0.47006136178970337,
      "learning_rate": 4.886629105400578e-06,
      "loss": 0.6578,
      "step": 5505
    },
    {
      "epoch": 1.5125669550885867,
      "grad_norm": 0.41009458899497986,
      "learning_rate": 4.883880720076955e-06,
      "loss": 0.5432,
      "step": 5506
    },
    {
      "epoch": 1.5128416426315066,
      "grad_norm": 0.362861305475235,
      "learning_rate": 4.881132334753333e-06,
      "loss": 0.8606,
      "step": 5507
    },
    {
      "epoch": 1.5131163301744266,
      "grad_norm": 0.5114688873291016,
      "learning_rate": 4.878383949429711e-06,
      "loss": 0.8376,
      "step": 5508
    },
    {
      "epoch": 1.5133910177173466,
      "grad_norm": 0.4915665090084076,
      "learning_rate": 4.8756355641060886e-06,
      "loss": 0.9184,
      "step": 5509
    },
    {
      "epoch": 1.5136657052602664,
      "grad_norm": 0.4321082830429077,
      "learning_rate": 4.8728871787824656e-06,
      "loss": 0.755,
      "step": 5510
    },
    {
      "epoch": 1.5139403928031863,
      "grad_norm": 0.6047990322113037,
      "learning_rate": 4.8701387934588434e-06,
      "loss": 0.9397,
      "step": 5511
    },
    {
      "epoch": 1.5142150803461063,
      "grad_norm": 0.43872854113578796,
      "learning_rate": 4.8673904081352205e-06,
      "loss": 0.8889,
      "step": 5512
    },
    {
      "epoch": 1.5144897678890263,
      "grad_norm": 0.4148659110069275,
      "learning_rate": 4.864642022811598e-06,
      "loss": 0.5888,
      "step": 5513
    },
    {
      "epoch": 1.5147644554319462,
      "grad_norm": 0.43397483229637146,
      "learning_rate": 4.861893637487976e-06,
      "loss": 0.6209,
      "step": 5514
    },
    {
      "epoch": 1.515039142974866,
      "grad_norm": 0.5580003261566162,
      "learning_rate": 4.859145252164354e-06,
      "loss": 0.7488,
      "step": 5515
    },
    {
      "epoch": 1.515313830517786,
      "grad_norm": 0.5526689887046814,
      "learning_rate": 4.856396866840731e-06,
      "loss": 0.653,
      "step": 5516
    },
    {
      "epoch": 1.515588518060706,
      "grad_norm": 0.38723278045654297,
      "learning_rate": 4.853648481517109e-06,
      "loss": 0.7241,
      "step": 5517
    },
    {
      "epoch": 1.5158632056036259,
      "grad_norm": 0.49174708127975464,
      "learning_rate": 4.850900096193487e-06,
      "loss": 0.8607,
      "step": 5518
    },
    {
      "epoch": 1.5161378931465457,
      "grad_norm": 0.4620455205440521,
      "learning_rate": 4.848151710869865e-06,
      "loss": 0.8963,
      "step": 5519
    },
    {
      "epoch": 1.5164125806894657,
      "grad_norm": 0.4878450632095337,
      "learning_rate": 4.845403325546242e-06,
      "loss": 0.8198,
      "step": 5520
    },
    {
      "epoch": 1.5166872682323858,
      "grad_norm": 0.4566137492656708,
      "learning_rate": 4.84265494022262e-06,
      "loss": 0.7265,
      "step": 5521
    },
    {
      "epoch": 1.5169619557753056,
      "grad_norm": 0.5118844509124756,
      "learning_rate": 4.8399065548989974e-06,
      "loss": 0.7786,
      "step": 5522
    },
    {
      "epoch": 1.5172366433182254,
      "grad_norm": 0.5696452260017395,
      "learning_rate": 4.837158169575375e-06,
      "loss": 0.8088,
      "step": 5523
    },
    {
      "epoch": 1.5175113308611454,
      "grad_norm": 0.337125301361084,
      "learning_rate": 4.834409784251752e-06,
      "loss": 0.9087,
      "step": 5524
    },
    {
      "epoch": 1.5177860184040655,
      "grad_norm": 0.4851342439651489,
      "learning_rate": 4.83166139892813e-06,
      "loss": 0.7787,
      "step": 5525
    },
    {
      "epoch": 1.5180607059469853,
      "grad_norm": 0.45952311158180237,
      "learning_rate": 4.828913013604507e-06,
      "loss": 0.6995,
      "step": 5526
    },
    {
      "epoch": 1.518335393489905,
      "grad_norm": 0.43666085600852966,
      "learning_rate": 4.826164628280885e-06,
      "loss": 0.9738,
      "step": 5527
    },
    {
      "epoch": 1.5186100810328251,
      "grad_norm": 0.5697729587554932,
      "learning_rate": 4.823416242957263e-06,
      "loss": 0.9675,
      "step": 5528
    },
    {
      "epoch": 1.5188847685757452,
      "grad_norm": 0.445248544216156,
      "learning_rate": 4.820667857633641e-06,
      "loss": 0.8287,
      "step": 5529
    },
    {
      "epoch": 1.519159456118665,
      "grad_norm": 0.4596867859363556,
      "learning_rate": 4.817919472310018e-06,
      "loss": 0.7531,
      "step": 5530
    },
    {
      "epoch": 1.5194341436615848,
      "grad_norm": 0.49937155842781067,
      "learning_rate": 4.815171086986396e-06,
      "loss": 0.7237,
      "step": 5531
    },
    {
      "epoch": 1.5197088312045048,
      "grad_norm": 0.4250665009021759,
      "learning_rate": 4.812422701662774e-06,
      "loss": 0.9659,
      "step": 5532
    },
    {
      "epoch": 1.5199835187474249,
      "grad_norm": 0.541818380355835,
      "learning_rate": 4.8096743163391514e-06,
      "loss": 0.5793,
      "step": 5533
    },
    {
      "epoch": 1.5202582062903447,
      "grad_norm": 0.3891429603099823,
      "learning_rate": 4.8069259310155285e-06,
      "loss": 0.6847,
      "step": 5534
    },
    {
      "epoch": 1.5205328938332645,
      "grad_norm": 0.5423728227615356,
      "learning_rate": 4.804177545691906e-06,
      "loss": 0.7648,
      "step": 5535
    },
    {
      "epoch": 1.5208075813761845,
      "grad_norm": 0.39551910758018494,
      "learning_rate": 4.801429160368284e-06,
      "loss": 0.6601,
      "step": 5536
    },
    {
      "epoch": 1.5210822689191046,
      "grad_norm": 0.42419928312301636,
      "learning_rate": 4.798680775044662e-06,
      "loss": 0.7395,
      "step": 5537
    },
    {
      "epoch": 1.5213569564620244,
      "grad_norm": 0.5405192375183105,
      "learning_rate": 4.795932389721039e-06,
      "loss": 0.8732,
      "step": 5538
    },
    {
      "epoch": 1.5216316440049442,
      "grad_norm": 0.4575425386428833,
      "learning_rate": 4.793184004397417e-06,
      "loss": 0.7752,
      "step": 5539
    },
    {
      "epoch": 1.5219063315478643,
      "grad_norm": 0.6944116353988647,
      "learning_rate": 4.790435619073794e-06,
      "loss": 0.7787,
      "step": 5540
    },
    {
      "epoch": 1.5221810190907843,
      "grad_norm": 0.42856565117836,
      "learning_rate": 4.787687233750173e-06,
      "loss": 0.7656,
      "step": 5541
    },
    {
      "epoch": 1.522455706633704,
      "grad_norm": 0.5419394969940186,
      "learning_rate": 4.78493884842655e-06,
      "loss": 0.5652,
      "step": 5542
    },
    {
      "epoch": 1.522730394176624,
      "grad_norm": 0.4489867091178894,
      "learning_rate": 4.782190463102928e-06,
      "loss": 0.6164,
      "step": 5543
    },
    {
      "epoch": 1.523005081719544,
      "grad_norm": 0.5257501602172852,
      "learning_rate": 4.779442077779305e-06,
      "loss": 0.6992,
      "step": 5544
    },
    {
      "epoch": 1.523279769262464,
      "grad_norm": 0.35113680362701416,
      "learning_rate": 4.7766936924556825e-06,
      "loss": 0.5364,
      "step": 5545
    },
    {
      "epoch": 1.523554456805384,
      "grad_norm": 0.40414008498191833,
      "learning_rate": 4.77394530713206e-06,
      "loss": 0.9032,
      "step": 5546
    },
    {
      "epoch": 1.5238291443483039,
      "grad_norm": 0.4347968101501465,
      "learning_rate": 4.771196921808438e-06,
      "loss": 0.7561,
      "step": 5547
    },
    {
      "epoch": 1.5241038318912237,
      "grad_norm": 0.4164765179157257,
      "learning_rate": 4.768448536484815e-06,
      "loss": 0.6989,
      "step": 5548
    },
    {
      "epoch": 1.5243785194341437,
      "grad_norm": 0.5291433930397034,
      "learning_rate": 4.765700151161193e-06,
      "loss": 0.7046,
      "step": 5549
    },
    {
      "epoch": 1.5246532069770637,
      "grad_norm": 0.5631390810012817,
      "learning_rate": 4.762951765837571e-06,
      "loss": 0.7471,
      "step": 5550
    },
    {
      "epoch": 1.5249278945199836,
      "grad_norm": 0.4492050111293793,
      "learning_rate": 4.760203380513949e-06,
      "loss": 0.8617,
      "step": 5551
    },
    {
      "epoch": 1.5252025820629034,
      "grad_norm": 0.5286779999732971,
      "learning_rate": 4.757454995190326e-06,
      "loss": 0.8716,
      "step": 5552
    },
    {
      "epoch": 1.5254772696058234,
      "grad_norm": 0.5081833600997925,
      "learning_rate": 4.754706609866704e-06,
      "loss": 0.9511,
      "step": 5553
    },
    {
      "epoch": 1.5257519571487435,
      "grad_norm": 0.6088399291038513,
      "learning_rate": 4.751958224543082e-06,
      "loss": 0.7088,
      "step": 5554
    },
    {
      "epoch": 1.5260266446916633,
      "grad_norm": 0.4341203570365906,
      "learning_rate": 4.7492098392194595e-06,
      "loss": 0.5507,
      "step": 5555
    },
    {
      "epoch": 1.526301332234583,
      "grad_norm": 0.4745272397994995,
      "learning_rate": 4.7464614538958365e-06,
      "loss": 0.7543,
      "step": 5556
    },
    {
      "epoch": 1.5265760197775031,
      "grad_norm": 0.349913626909256,
      "learning_rate": 4.743713068572214e-06,
      "loss": 0.9495,
      "step": 5557
    },
    {
      "epoch": 1.5268507073204232,
      "grad_norm": 0.4971676170825958,
      "learning_rate": 4.740964683248591e-06,
      "loss": 0.746,
      "step": 5558
    },
    {
      "epoch": 1.527125394863343,
      "grad_norm": 0.4121992588043213,
      "learning_rate": 4.738216297924969e-06,
      "loss": 0.6349,
      "step": 5559
    },
    {
      "epoch": 1.5274000824062628,
      "grad_norm": 0.4963414967060089,
      "learning_rate": 4.735467912601347e-06,
      "loss": 0.6512,
      "step": 5560
    },
    {
      "epoch": 1.5276747699491828,
      "grad_norm": 0.3873559236526489,
      "learning_rate": 4.732719527277725e-06,
      "loss": 0.7771,
      "step": 5561
    },
    {
      "epoch": 1.5279494574921029,
      "grad_norm": 0.6288356184959412,
      "learning_rate": 4.729971141954102e-06,
      "loss": 0.6834,
      "step": 5562
    },
    {
      "epoch": 1.5282241450350227,
      "grad_norm": 0.49871131777763367,
      "learning_rate": 4.72722275663048e-06,
      "loss": 0.8755,
      "step": 5563
    },
    {
      "epoch": 1.5284988325779425,
      "grad_norm": 0.5421341061592102,
      "learning_rate": 4.724474371306858e-06,
      "loss": 0.7234,
      "step": 5564
    },
    {
      "epoch": 1.5287735201208625,
      "grad_norm": 0.43100377917289734,
      "learning_rate": 4.721725985983236e-06,
      "loss": 0.8446,
      "step": 5565
    },
    {
      "epoch": 1.5290482076637826,
      "grad_norm": 0.366163969039917,
      "learning_rate": 4.718977600659613e-06,
      "loss": 0.9636,
      "step": 5566
    },
    {
      "epoch": 1.5293228952067024,
      "grad_norm": 0.4189234972000122,
      "learning_rate": 4.7162292153359905e-06,
      "loss": 0.6234,
      "step": 5567
    },
    {
      "epoch": 1.5295975827496222,
      "grad_norm": 0.38919174671173096,
      "learning_rate": 4.713480830012368e-06,
      "loss": 0.9912,
      "step": 5568
    },
    {
      "epoch": 1.5298722702925422,
      "grad_norm": 0.6894047856330872,
      "learning_rate": 4.710732444688746e-06,
      "loss": 0.6507,
      "step": 5569
    },
    {
      "epoch": 1.5301469578354623,
      "grad_norm": 0.48897144198417664,
      "learning_rate": 4.707984059365123e-06,
      "loss": 0.8327,
      "step": 5570
    },
    {
      "epoch": 1.530421645378382,
      "grad_norm": 0.5810003876686096,
      "learning_rate": 4.705235674041501e-06,
      "loss": 0.9644,
      "step": 5571
    },
    {
      "epoch": 1.530696332921302,
      "grad_norm": 0.43007585406303406,
      "learning_rate": 4.702487288717878e-06,
      "loss": 0.8532,
      "step": 5572
    },
    {
      "epoch": 1.530971020464222,
      "grad_norm": 0.47998499870300293,
      "learning_rate": 4.699738903394257e-06,
      "loss": 0.7818,
      "step": 5573
    },
    {
      "epoch": 1.531245708007142,
      "grad_norm": 0.3826018273830414,
      "learning_rate": 4.696990518070634e-06,
      "loss": 0.8018,
      "step": 5574
    },
    {
      "epoch": 1.5315203955500618,
      "grad_norm": 0.5045073628425598,
      "learning_rate": 4.694242132747012e-06,
      "loss": 0.7627,
      "step": 5575
    },
    {
      "epoch": 1.5317950830929816,
      "grad_norm": 0.4612228274345398,
      "learning_rate": 4.691493747423389e-06,
      "loss": 1.123,
      "step": 5576
    },
    {
      "epoch": 1.5320697706359017,
      "grad_norm": 0.554922342300415,
      "learning_rate": 4.688745362099767e-06,
      "loss": 0.7409,
      "step": 5577
    },
    {
      "epoch": 1.5323444581788217,
      "grad_norm": 0.4592605233192444,
      "learning_rate": 4.6859969767761445e-06,
      "loss": 0.8364,
      "step": 5578
    },
    {
      "epoch": 1.5326191457217415,
      "grad_norm": 0.432743638753891,
      "learning_rate": 4.683248591452522e-06,
      "loss": 0.7779,
      "step": 5579
    },
    {
      "epoch": 1.5328938332646613,
      "grad_norm": 0.5685707330703735,
      "learning_rate": 4.680500206128899e-06,
      "loss": 0.8919,
      "step": 5580
    },
    {
      "epoch": 1.5331685208075814,
      "grad_norm": 0.5153523087501526,
      "learning_rate": 4.677751820805277e-06,
      "loss": 0.7106,
      "step": 5581
    },
    {
      "epoch": 1.5334432083505014,
      "grad_norm": 0.661437451839447,
      "learning_rate": 4.675003435481655e-06,
      "loss": 0.7498,
      "step": 5582
    },
    {
      "epoch": 1.5337178958934212,
      "grad_norm": 0.5347268581390381,
      "learning_rate": 4.672255050158033e-06,
      "loss": 0.8738,
      "step": 5583
    },
    {
      "epoch": 1.533992583436341,
      "grad_norm": 1.8114510774612427,
      "learning_rate": 4.66950666483441e-06,
      "loss": 0.744,
      "step": 5584
    },
    {
      "epoch": 1.534267270979261,
      "grad_norm": 0.35810038447380066,
      "learning_rate": 4.666758279510788e-06,
      "loss": 0.6346,
      "step": 5585
    },
    {
      "epoch": 1.5345419585221811,
      "grad_norm": 0.4578748941421509,
      "learning_rate": 4.664009894187166e-06,
      "loss": 0.7012,
      "step": 5586
    },
    {
      "epoch": 1.534816646065101,
      "grad_norm": 0.4460470676422119,
      "learning_rate": 4.661261508863544e-06,
      "loss": 0.6261,
      "step": 5587
    },
    {
      "epoch": 1.5350913336080207,
      "grad_norm": 0.5008774399757385,
      "learning_rate": 4.658513123539921e-06,
      "loss": 0.9201,
      "step": 5588
    },
    {
      "epoch": 1.5353660211509408,
      "grad_norm": 0.44502362608909607,
      "learning_rate": 4.6557647382162985e-06,
      "loss": 0.804,
      "step": 5589
    },
    {
      "epoch": 1.5356407086938608,
      "grad_norm": 0.49077534675598145,
      "learning_rate": 4.6530163528926755e-06,
      "loss": 0.995,
      "step": 5590
    },
    {
      "epoch": 1.5359153962367806,
      "grad_norm": 0.42476704716682434,
      "learning_rate": 4.650267967569053e-06,
      "loss": 1.0089,
      "step": 5591
    },
    {
      "epoch": 1.5361900837797005,
      "grad_norm": 0.4046640694141388,
      "learning_rate": 4.647519582245431e-06,
      "loss": 0.9612,
      "step": 5592
    },
    {
      "epoch": 1.5364647713226205,
      "grad_norm": 0.46006888151168823,
      "learning_rate": 4.644771196921809e-06,
      "loss": 0.8048,
      "step": 5593
    },
    {
      "epoch": 1.5367394588655405,
      "grad_norm": 0.5845133662223816,
      "learning_rate": 4.642022811598186e-06,
      "loss": 0.6918,
      "step": 5594
    },
    {
      "epoch": 1.5370141464084603,
      "grad_norm": 0.4422937035560608,
      "learning_rate": 4.639274426274564e-06,
      "loss": 0.8886,
      "step": 5595
    },
    {
      "epoch": 1.5372888339513802,
      "grad_norm": 0.39785036444664,
      "learning_rate": 4.636526040950942e-06,
      "loss": 0.6923,
      "step": 5596
    },
    {
      "epoch": 1.5375635214943002,
      "grad_norm": 0.44027408957481384,
      "learning_rate": 4.63377765562732e-06,
      "loss": 0.5277,
      "step": 5597
    },
    {
      "epoch": 1.5378382090372202,
      "grad_norm": 0.43150588870048523,
      "learning_rate": 4.631029270303697e-06,
      "loss": 0.505,
      "step": 5598
    },
    {
      "epoch": 1.53811289658014,
      "grad_norm": 0.5964921116828918,
      "learning_rate": 4.628280884980075e-06,
      "loss": 0.8847,
      "step": 5599
    },
    {
      "epoch": 1.5383875841230599,
      "grad_norm": 0.42486581206321716,
      "learning_rate": 4.6255324996564525e-06,
      "loss": 0.8729,
      "step": 5600
    },
    {
      "epoch": 1.53866227166598,
      "grad_norm": 0.4563451409339905,
      "learning_rate": 4.62278411433283e-06,
      "loss": 0.6745,
      "step": 5601
    },
    {
      "epoch": 1.5389369592089,
      "grad_norm": 0.6552596092224121,
      "learning_rate": 4.620035729009207e-06,
      "loss": 0.7853,
      "step": 5602
    },
    {
      "epoch": 1.5392116467518198,
      "grad_norm": 0.5289437770843506,
      "learning_rate": 4.617287343685585e-06,
      "loss": 0.6622,
      "step": 5603
    },
    {
      "epoch": 1.5394863342947396,
      "grad_norm": 0.4757509231567383,
      "learning_rate": 4.614538958361962e-06,
      "loss": 0.7736,
      "step": 5604
    },
    {
      "epoch": 1.5397610218376596,
      "grad_norm": 0.419852077960968,
      "learning_rate": 4.61179057303834e-06,
      "loss": 0.6875,
      "step": 5605
    },
    {
      "epoch": 1.5400357093805797,
      "grad_norm": 0.49759232997894287,
      "learning_rate": 4.609042187714718e-06,
      "loss": 0.6474,
      "step": 5606
    },
    {
      "epoch": 1.5403103969234995,
      "grad_norm": 0.43117308616638184,
      "learning_rate": 4.606293802391096e-06,
      "loss": 1.0308,
      "step": 5607
    },
    {
      "epoch": 1.5405850844664193,
      "grad_norm": 0.39209070801734924,
      "learning_rate": 4.603545417067473e-06,
      "loss": 0.8571,
      "step": 5608
    },
    {
      "epoch": 1.5408597720093393,
      "grad_norm": 0.5445133447647095,
      "learning_rate": 4.600797031743851e-06,
      "loss": 0.8594,
      "step": 5609
    },
    {
      "epoch": 1.5411344595522594,
      "grad_norm": 0.542805552482605,
      "learning_rate": 4.598048646420229e-06,
      "loss": 0.8307,
      "step": 5610
    },
    {
      "epoch": 1.5414091470951794,
      "grad_norm": 0.45290353894233704,
      "learning_rate": 4.5953002610966065e-06,
      "loss": 0.8996,
      "step": 5611
    },
    {
      "epoch": 1.5416838346380992,
      "grad_norm": 0.538456380367279,
      "learning_rate": 4.5925518757729835e-06,
      "loss": 0.8443,
      "step": 5612
    },
    {
      "epoch": 1.541958522181019,
      "grad_norm": 0.5527083277702332,
      "learning_rate": 4.589803490449361e-06,
      "loss": 0.6889,
      "step": 5613
    },
    {
      "epoch": 1.542233209723939,
      "grad_norm": 0.593413770198822,
      "learning_rate": 4.587055105125739e-06,
      "loss": 0.9232,
      "step": 5614
    },
    {
      "epoch": 1.542507897266859,
      "grad_norm": 0.39173707365989685,
      "learning_rate": 4.584306719802117e-06,
      "loss": 0.9509,
      "step": 5615
    },
    {
      "epoch": 1.542782584809779,
      "grad_norm": 0.505922257900238,
      "learning_rate": 4.581558334478494e-06,
      "loss": 0.639,
      "step": 5616
    },
    {
      "epoch": 1.5430572723526987,
      "grad_norm": 0.5414848327636719,
      "learning_rate": 4.578809949154872e-06,
      "loss": 0.7685,
      "step": 5617
    },
    {
      "epoch": 1.5433319598956188,
      "grad_norm": 0.5283029079437256,
      "learning_rate": 4.576061563831249e-06,
      "loss": 0.9503,
      "step": 5618
    },
    {
      "epoch": 1.5436066474385388,
      "grad_norm": 0.40111812949180603,
      "learning_rate": 4.573313178507628e-06,
      "loss": 0.8319,
      "step": 5619
    },
    {
      "epoch": 1.5438813349814586,
      "grad_norm": 0.37385496497154236,
      "learning_rate": 4.570564793184005e-06,
      "loss": 0.625,
      "step": 5620
    },
    {
      "epoch": 1.5441560225243784,
      "grad_norm": 0.5466792583465576,
      "learning_rate": 4.567816407860383e-06,
      "loss": 0.8572,
      "step": 5621
    },
    {
      "epoch": 1.5444307100672985,
      "grad_norm": 0.46429792046546936,
      "learning_rate": 4.56506802253676e-06,
      "loss": 0.7259,
      "step": 5622
    },
    {
      "epoch": 1.5447053976102185,
      "grad_norm": 0.4345654845237732,
      "learning_rate": 4.5623196372131375e-06,
      "loss": 0.9692,
      "step": 5623
    },
    {
      "epoch": 1.5449800851531383,
      "grad_norm": 0.6299007534980774,
      "learning_rate": 4.559571251889515e-06,
      "loss": 0.8517,
      "step": 5624
    },
    {
      "epoch": 1.5452547726960582,
      "grad_norm": 0.5249785780906677,
      "learning_rate": 4.556822866565893e-06,
      "loss": 0.6232,
      "step": 5625
    },
    {
      "epoch": 1.5455294602389782,
      "grad_norm": 0.44623813033103943,
      "learning_rate": 4.55407448124227e-06,
      "loss": 0.7389,
      "step": 5626
    },
    {
      "epoch": 1.5458041477818982,
      "grad_norm": 0.6267253756523132,
      "learning_rate": 4.551326095918648e-06,
      "loss": 0.6195,
      "step": 5627
    },
    {
      "epoch": 1.546078835324818,
      "grad_norm": 0.47230076789855957,
      "learning_rate": 4.548577710595026e-06,
      "loss": 0.6449,
      "step": 5628
    },
    {
      "epoch": 1.5463535228677379,
      "grad_norm": 0.42976802587509155,
      "learning_rate": 4.545829325271404e-06,
      "loss": 0.6592,
      "step": 5629
    },
    {
      "epoch": 1.546628210410658,
      "grad_norm": 0.39420202374458313,
      "learning_rate": 4.543080939947781e-06,
      "loss": 0.5242,
      "step": 5630
    },
    {
      "epoch": 1.546902897953578,
      "grad_norm": 0.5185912251472473,
      "learning_rate": 4.540332554624159e-06,
      "loss": 0.649,
      "step": 5631
    },
    {
      "epoch": 1.5471775854964978,
      "grad_norm": 0.4277348518371582,
      "learning_rate": 4.537584169300537e-06,
      "loss": 0.6045,
      "step": 5632
    },
    {
      "epoch": 1.5474522730394176,
      "grad_norm": 0.47717446088790894,
      "learning_rate": 4.5348357839769145e-06,
      "loss": 0.6416,
      "step": 5633
    },
    {
      "epoch": 1.5477269605823376,
      "grad_norm": 0.4232649803161621,
      "learning_rate": 4.5320873986532915e-06,
      "loss": 0.8959,
      "step": 5634
    },
    {
      "epoch": 1.5480016481252576,
      "grad_norm": 0.3835744857788086,
      "learning_rate": 4.529339013329669e-06,
      "loss": 0.6273,
      "step": 5635
    },
    {
      "epoch": 1.5482763356681775,
      "grad_norm": 0.4809291660785675,
      "learning_rate": 4.5265906280060464e-06,
      "loss": 0.8485,
      "step": 5636
    },
    {
      "epoch": 1.5485510232110973,
      "grad_norm": 0.47333094477653503,
      "learning_rate": 4.523842242682424e-06,
      "loss": 0.8299,
      "step": 5637
    },
    {
      "epoch": 1.5488257107540173,
      "grad_norm": 0.4427059292793274,
      "learning_rate": 4.521093857358802e-06,
      "loss": 0.8102,
      "step": 5638
    },
    {
      "epoch": 1.5491003982969374,
      "grad_norm": 0.5191941857337952,
      "learning_rate": 4.51834547203518e-06,
      "loss": 0.7327,
      "step": 5639
    },
    {
      "epoch": 1.5493750858398572,
      "grad_norm": 0.4399535357952118,
      "learning_rate": 4.515597086711557e-06,
      "loss": 0.8864,
      "step": 5640
    },
    {
      "epoch": 1.549649773382777,
      "grad_norm": 0.5376676321029663,
      "learning_rate": 4.512848701387935e-06,
      "loss": 0.6011,
      "step": 5641
    },
    {
      "epoch": 1.549924460925697,
      "grad_norm": 0.4254341721534729,
      "learning_rate": 4.510100316064313e-06,
      "loss": 0.6263,
      "step": 5642
    },
    {
      "epoch": 1.550199148468617,
      "grad_norm": 0.42912957072257996,
      "learning_rate": 4.507351930740691e-06,
      "loss": 0.5525,
      "step": 5643
    },
    {
      "epoch": 1.5504738360115369,
      "grad_norm": 0.4743158221244812,
      "learning_rate": 4.504603545417068e-06,
      "loss": 0.8761,
      "step": 5644
    },
    {
      "epoch": 1.5507485235544567,
      "grad_norm": 0.5503890514373779,
      "learning_rate": 4.5018551600934455e-06,
      "loss": 0.6623,
      "step": 5645
    },
    {
      "epoch": 1.5510232110973767,
      "grad_norm": 0.6173017024993896,
      "learning_rate": 4.499106774769823e-06,
      "loss": 0.6069,
      "step": 5646
    },
    {
      "epoch": 1.5512978986402968,
      "grad_norm": 0.4846153259277344,
      "learning_rate": 4.496358389446201e-06,
      "loss": 0.8017,
      "step": 5647
    },
    {
      "epoch": 1.5515725861832166,
      "grad_norm": 0.3925398588180542,
      "learning_rate": 4.493610004122578e-06,
      "loss": 0.7808,
      "step": 5648
    },
    {
      "epoch": 1.5518472737261364,
      "grad_norm": 0.45735082030296326,
      "learning_rate": 4.490861618798956e-06,
      "loss": 0.7275,
      "step": 5649
    },
    {
      "epoch": 1.5521219612690564,
      "grad_norm": 0.6296531558036804,
      "learning_rate": 4.488113233475333e-06,
      "loss": 0.8238,
      "step": 5650
    },
    {
      "epoch": 1.5523966488119765,
      "grad_norm": 0.4080139696598053,
      "learning_rate": 4.485364848151711e-06,
      "loss": 1.0683,
      "step": 5651
    },
    {
      "epoch": 1.5526713363548963,
      "grad_norm": 0.3976295590400696,
      "learning_rate": 4.482616462828089e-06,
      "loss": 0.6199,
      "step": 5652
    },
    {
      "epoch": 1.552946023897816,
      "grad_norm": 0.47055673599243164,
      "learning_rate": 4.479868077504467e-06,
      "loss": 0.8663,
      "step": 5653
    },
    {
      "epoch": 1.5532207114407361,
      "grad_norm": 0.5115500688552856,
      "learning_rate": 4.477119692180844e-06,
      "loss": 0.8458,
      "step": 5654
    },
    {
      "epoch": 1.5534953989836562,
      "grad_norm": 0.3988882303237915,
      "learning_rate": 4.474371306857222e-06,
      "loss": 0.6638,
      "step": 5655
    },
    {
      "epoch": 1.553770086526576,
      "grad_norm": 0.44192758202552795,
      "learning_rate": 4.4716229215335995e-06,
      "loss": 0.6687,
      "step": 5656
    },
    {
      "epoch": 1.5540447740694958,
      "grad_norm": 0.4838665723800659,
      "learning_rate": 4.468874536209977e-06,
      "loss": 0.5723,
      "step": 5657
    },
    {
      "epoch": 1.5543194616124159,
      "grad_norm": 0.45621928572654724,
      "learning_rate": 4.4661261508863544e-06,
      "loss": 0.9072,
      "step": 5658
    },
    {
      "epoch": 1.554594149155336,
      "grad_norm": 0.5620595216751099,
      "learning_rate": 4.463377765562732e-06,
      "loss": 0.7814,
      "step": 5659
    },
    {
      "epoch": 1.5548688366982557,
      "grad_norm": 0.40597769618034363,
      "learning_rate": 4.46062938023911e-06,
      "loss": 0.6003,
      "step": 5660
    },
    {
      "epoch": 1.5551435242411755,
      "grad_norm": 0.42922061681747437,
      "learning_rate": 4.457880994915488e-06,
      "loss": 0.597,
      "step": 5661
    },
    {
      "epoch": 1.5554182117840956,
      "grad_norm": 0.5032103061676025,
      "learning_rate": 4.455132609591865e-06,
      "loss": 0.5701,
      "step": 5662
    },
    {
      "epoch": 1.5556928993270156,
      "grad_norm": 0.39987823367118835,
      "learning_rate": 4.452384224268243e-06,
      "loss": 0.6371,
      "step": 5663
    },
    {
      "epoch": 1.5559675868699354,
      "grad_norm": 0.48125335574150085,
      "learning_rate": 4.44963583894462e-06,
      "loss": 0.7065,
      "step": 5664
    },
    {
      "epoch": 1.5562422744128552,
      "grad_norm": 0.5584878325462341,
      "learning_rate": 4.446887453620999e-06,
      "loss": 0.6849,
      "step": 5665
    },
    {
      "epoch": 1.5565169619557753,
      "grad_norm": 0.3886931836605072,
      "learning_rate": 4.444139068297376e-06,
      "loss": 0.8703,
      "step": 5666
    },
    {
      "epoch": 1.5567916494986953,
      "grad_norm": 0.44070932269096375,
      "learning_rate": 4.4413906829737536e-06,
      "loss": 0.7997,
      "step": 5667
    },
    {
      "epoch": 1.5570663370416151,
      "grad_norm": 0.31382572650909424,
      "learning_rate": 4.4386422976501306e-06,
      "loss": 0.6237,
      "step": 5668
    },
    {
      "epoch": 1.557341024584535,
      "grad_norm": 0.4733794629573822,
      "learning_rate": 4.4358939123265084e-06,
      "loss": 0.8801,
      "step": 5669
    },
    {
      "epoch": 1.557615712127455,
      "grad_norm": 0.4862426519393921,
      "learning_rate": 4.433145527002886e-06,
      "loss": 0.8376,
      "step": 5670
    },
    {
      "epoch": 1.557890399670375,
      "grad_norm": 0.4843164384365082,
      "learning_rate": 4.430397141679264e-06,
      "loss": 0.8191,
      "step": 5671
    },
    {
      "epoch": 1.5581650872132948,
      "grad_norm": 0.4894069731235504,
      "learning_rate": 4.427648756355641e-06,
      "loss": 0.7358,
      "step": 5672
    },
    {
      "epoch": 1.5584397747562146,
      "grad_norm": 0.46596559882164,
      "learning_rate": 4.424900371032019e-06,
      "loss": 0.7216,
      "step": 5673
    },
    {
      "epoch": 1.5587144622991347,
      "grad_norm": 0.5392791628837585,
      "learning_rate": 4.422151985708397e-06,
      "loss": 0.8236,
      "step": 5674
    },
    {
      "epoch": 1.5589891498420547,
      "grad_norm": 0.46981489658355713,
      "learning_rate": 4.419403600384775e-06,
      "loss": 0.7017,
      "step": 5675
    },
    {
      "epoch": 1.5592638373849745,
      "grad_norm": 0.4114433526992798,
      "learning_rate": 4.416655215061152e-06,
      "loss": 0.7219,
      "step": 5676
    },
    {
      "epoch": 1.5595385249278944,
      "grad_norm": 0.45712435245513916,
      "learning_rate": 4.41390682973753e-06,
      "loss": 0.703,
      "step": 5677
    },
    {
      "epoch": 1.5598132124708144,
      "grad_norm": 0.36442282795906067,
      "learning_rate": 4.4111584444139076e-06,
      "loss": 0.6167,
      "step": 5678
    },
    {
      "epoch": 1.5600879000137344,
      "grad_norm": 0.5247613787651062,
      "learning_rate": 4.408410059090285e-06,
      "loss": 0.7664,
      "step": 5679
    },
    {
      "epoch": 1.5603625875566545,
      "grad_norm": 0.4201558232307434,
      "learning_rate": 4.4056616737666624e-06,
      "loss": 0.6011,
      "step": 5680
    },
    {
      "epoch": 1.5606372750995743,
      "grad_norm": 0.43468472361564636,
      "learning_rate": 4.40291328844304e-06,
      "loss": 0.8592,
      "step": 5681
    },
    {
      "epoch": 1.560911962642494,
      "grad_norm": 0.45223256945610046,
      "learning_rate": 4.400164903119417e-06,
      "loss": 0.7077,
      "step": 5682
    },
    {
      "epoch": 1.5611866501854141,
      "grad_norm": 0.4653623104095459,
      "learning_rate": 4.397416517795795e-06,
      "loss": 0.4966,
      "step": 5683
    },
    {
      "epoch": 1.5614613377283342,
      "grad_norm": 0.3959925174713135,
      "learning_rate": 4.394668132472173e-06,
      "loss": 0.7738,
      "step": 5684
    },
    {
      "epoch": 1.561736025271254,
      "grad_norm": 0.47435280680656433,
      "learning_rate": 4.391919747148551e-06,
      "loss": 0.8394,
      "step": 5685
    },
    {
      "epoch": 1.5620107128141738,
      "grad_norm": 0.4947885572910309,
      "learning_rate": 4.389171361824928e-06,
      "loss": 0.8544,
      "step": 5686
    },
    {
      "epoch": 1.5622854003570938,
      "grad_norm": 0.41972360014915466,
      "learning_rate": 4.386422976501306e-06,
      "loss": 0.8245,
      "step": 5687
    },
    {
      "epoch": 1.5625600879000139,
      "grad_norm": 0.5349785089492798,
      "learning_rate": 4.383674591177684e-06,
      "loss": 0.685,
      "step": 5688
    },
    {
      "epoch": 1.5628347754429337,
      "grad_norm": 0.40236696600914,
      "learning_rate": 4.3809262058540616e-06,
      "loss": 0.7488,
      "step": 5689
    },
    {
      "epoch": 1.5631094629858535,
      "grad_norm": 0.5690762400627136,
      "learning_rate": 4.378177820530439e-06,
      "loss": 0.6599,
      "step": 5690
    },
    {
      "epoch": 1.5633841505287736,
      "grad_norm": 0.5314335823059082,
      "learning_rate": 4.3754294352068164e-06,
      "loss": 0.9206,
      "step": 5691
    },
    {
      "epoch": 1.5636588380716936,
      "grad_norm": 0.4243451654911041,
      "learning_rate": 4.372681049883194e-06,
      "loss": 0.7902,
      "step": 5692
    },
    {
      "epoch": 1.5639335256146134,
      "grad_norm": 0.3843523859977722,
      "learning_rate": 4.369932664559572e-06,
      "loss": 0.8104,
      "step": 5693
    },
    {
      "epoch": 1.5642082131575332,
      "grad_norm": 0.48872900009155273,
      "learning_rate": 4.367184279235949e-06,
      "loss": 0.8928,
      "step": 5694
    },
    {
      "epoch": 1.5644829007004533,
      "grad_norm": 0.5233150720596313,
      "learning_rate": 4.364435893912327e-06,
      "loss": 0.8942,
      "step": 5695
    },
    {
      "epoch": 1.5647575882433733,
      "grad_norm": 0.5885780453681946,
      "learning_rate": 4.361687508588704e-06,
      "loss": 0.7097,
      "step": 5696
    },
    {
      "epoch": 1.5650322757862931,
      "grad_norm": 0.47873654961586,
      "learning_rate": 4.358939123265083e-06,
      "loss": 1.0126,
      "step": 5697
    },
    {
      "epoch": 1.565306963329213,
      "grad_norm": 0.4308883249759674,
      "learning_rate": 4.35619073794146e-06,
      "loss": 0.7463,
      "step": 5698
    },
    {
      "epoch": 1.565581650872133,
      "grad_norm": 0.3563311994075775,
      "learning_rate": 4.353442352617838e-06,
      "loss": 0.657,
      "step": 5699
    },
    {
      "epoch": 1.565856338415053,
      "grad_norm": 0.5359319448471069,
      "learning_rate": 4.350693967294215e-06,
      "loss": 0.6028,
      "step": 5700
    },
    {
      "epoch": 1.5661310259579728,
      "grad_norm": 0.4617803990840912,
      "learning_rate": 4.347945581970593e-06,
      "loss": 0.8414,
      "step": 5701
    },
    {
      "epoch": 1.5664057135008926,
      "grad_norm": 0.5472536087036133,
      "learning_rate": 4.3451971966469705e-06,
      "loss": 0.9136,
      "step": 5702
    },
    {
      "epoch": 1.5666804010438127,
      "grad_norm": 0.38016217947006226,
      "learning_rate": 4.3424488113233475e-06,
      "loss": 0.765,
      "step": 5703
    },
    {
      "epoch": 1.5669550885867327,
      "grad_norm": 0.5543139576911926,
      "learning_rate": 4.339700425999725e-06,
      "loss": 0.6729,
      "step": 5704
    },
    {
      "epoch": 1.5672297761296525,
      "grad_norm": 0.42755043506622314,
      "learning_rate": 4.336952040676103e-06,
      "loss": 0.9318,
      "step": 5705
    },
    {
      "epoch": 1.5675044636725723,
      "grad_norm": 0.47328075766563416,
      "learning_rate": 4.334203655352481e-06,
      "loss": 0.8126,
      "step": 5706
    },
    {
      "epoch": 1.5677791512154924,
      "grad_norm": 0.4724140167236328,
      "learning_rate": 4.331455270028858e-06,
      "loss": 0.7495,
      "step": 5707
    },
    {
      "epoch": 1.5680538387584124,
      "grad_norm": 0.446387380361557,
      "learning_rate": 4.328706884705236e-06,
      "loss": 0.4621,
      "step": 5708
    },
    {
      "epoch": 1.5683285263013322,
      "grad_norm": 0.4034940004348755,
      "learning_rate": 4.325958499381613e-06,
      "loss": 0.4983,
      "step": 5709
    },
    {
      "epoch": 1.568603213844252,
      "grad_norm": 0.3860255479812622,
      "learning_rate": 4.323210114057992e-06,
      "loss": 0.6582,
      "step": 5710
    },
    {
      "epoch": 1.568877901387172,
      "grad_norm": 0.39560002088546753,
      "learning_rate": 4.320461728734369e-06,
      "loss": 1.0754,
      "step": 5711
    },
    {
      "epoch": 1.5691525889300921,
      "grad_norm": 0.4594748020172119,
      "learning_rate": 4.317713343410747e-06,
      "loss": 0.936,
      "step": 5712
    },
    {
      "epoch": 1.569427276473012,
      "grad_norm": 0.3786577880382538,
      "learning_rate": 4.314964958087124e-06,
      "loss": 0.8126,
      "step": 5713
    },
    {
      "epoch": 1.5697019640159318,
      "grad_norm": 0.412414014339447,
      "learning_rate": 4.3122165727635015e-06,
      "loss": 0.5577,
      "step": 5714
    },
    {
      "epoch": 1.5699766515588518,
      "grad_norm": 0.42544806003570557,
      "learning_rate": 4.309468187439879e-06,
      "loss": 0.7227,
      "step": 5715
    },
    {
      "epoch": 1.5702513391017718,
      "grad_norm": 0.400787353515625,
      "learning_rate": 4.306719802116257e-06,
      "loss": 0.9213,
      "step": 5716
    },
    {
      "epoch": 1.5705260266446917,
      "grad_norm": 0.4002636969089508,
      "learning_rate": 4.303971416792634e-06,
      "loss": 0.5491,
      "step": 5717
    },
    {
      "epoch": 1.5708007141876115,
      "grad_norm": 0.42293795943260193,
      "learning_rate": 4.301223031469012e-06,
      "loss": 1.0447,
      "step": 5718
    },
    {
      "epoch": 1.5710754017305315,
      "grad_norm": 0.48324456810951233,
      "learning_rate": 4.29847464614539e-06,
      "loss": 0.9344,
      "step": 5719
    },
    {
      "epoch": 1.5713500892734515,
      "grad_norm": 0.3309319019317627,
      "learning_rate": 4.295726260821768e-06,
      "loss": 0.8024,
      "step": 5720
    },
    {
      "epoch": 1.5716247768163714,
      "grad_norm": 0.4065021574497223,
      "learning_rate": 4.292977875498145e-06,
      "loss": 0.6953,
      "step": 5721
    },
    {
      "epoch": 1.5718994643592912,
      "grad_norm": 0.42703041434288025,
      "learning_rate": 4.290229490174523e-06,
      "loss": 0.7052,
      "step": 5722
    },
    {
      "epoch": 1.5721741519022112,
      "grad_norm": 0.4233732223510742,
      "learning_rate": 4.287481104850901e-06,
      "loss": 0.8053,
      "step": 5723
    },
    {
      "epoch": 1.5724488394451313,
      "grad_norm": 0.41935062408447266,
      "learning_rate": 4.2847327195272785e-06,
      "loss": 0.869,
      "step": 5724
    },
    {
      "epoch": 1.572723526988051,
      "grad_norm": 0.626695990562439,
      "learning_rate": 4.2819843342036555e-06,
      "loss": 0.7204,
      "step": 5725
    },
    {
      "epoch": 1.5729982145309709,
      "grad_norm": 0.6049544811248779,
      "learning_rate": 4.279235948880033e-06,
      "loss": 0.7868,
      "step": 5726
    },
    {
      "epoch": 1.573272902073891,
      "grad_norm": 0.3918648958206177,
      "learning_rate": 4.27648756355641e-06,
      "loss": 0.9834,
      "step": 5727
    },
    {
      "epoch": 1.573547589616811,
      "grad_norm": 0.4757474362850189,
      "learning_rate": 4.273739178232788e-06,
      "loss": 0.9248,
      "step": 5728
    },
    {
      "epoch": 1.5738222771597308,
      "grad_norm": 0.4297725558280945,
      "learning_rate": 4.270990792909166e-06,
      "loss": 0.5812,
      "step": 5729
    },
    {
      "epoch": 1.5740969647026506,
      "grad_norm": 0.6314119696617126,
      "learning_rate": 4.268242407585544e-06,
      "loss": 0.8019,
      "step": 5730
    },
    {
      "epoch": 1.5743716522455706,
      "grad_norm": 0.45389261841773987,
      "learning_rate": 4.265494022261921e-06,
      "loss": 0.7671,
      "step": 5731
    },
    {
      "epoch": 1.5746463397884907,
      "grad_norm": 0.5490732192993164,
      "learning_rate": 4.262745636938299e-06,
      "loss": 0.8446,
      "step": 5732
    },
    {
      "epoch": 1.5749210273314105,
      "grad_norm": 0.4030367136001587,
      "learning_rate": 4.259997251614677e-06,
      "loss": 0.9171,
      "step": 5733
    },
    {
      "epoch": 1.5751957148743303,
      "grad_norm": 0.46297189593315125,
      "learning_rate": 4.257248866291055e-06,
      "loss": 0.7559,
      "step": 5734
    },
    {
      "epoch": 1.5754704024172503,
      "grad_norm": 0.5005112290382385,
      "learning_rate": 4.254500480967432e-06,
      "loss": 0.6899,
      "step": 5735
    },
    {
      "epoch": 1.5757450899601704,
      "grad_norm": 0.46670547127723694,
      "learning_rate": 4.2517520956438095e-06,
      "loss": 0.624,
      "step": 5736
    },
    {
      "epoch": 1.5760197775030902,
      "grad_norm": 0.5959976315498352,
      "learning_rate": 4.249003710320187e-06,
      "loss": 0.6025,
      "step": 5737
    },
    {
      "epoch": 1.57629446504601,
      "grad_norm": 0.5421279668807983,
      "learning_rate": 4.246255324996565e-06,
      "loss": 0.9542,
      "step": 5738
    },
    {
      "epoch": 1.57656915258893,
      "grad_norm": 0.4595278203487396,
      "learning_rate": 4.243506939672942e-06,
      "loss": 0.7695,
      "step": 5739
    },
    {
      "epoch": 1.57684384013185,
      "grad_norm": 0.4828765392303467,
      "learning_rate": 4.24075855434932e-06,
      "loss": 0.7732,
      "step": 5740
    },
    {
      "epoch": 1.57711852767477,
      "grad_norm": 0.4971148669719696,
      "learning_rate": 4.238010169025697e-06,
      "loss": 0.9602,
      "step": 5741
    },
    {
      "epoch": 1.5773932152176897,
      "grad_norm": 0.5028340220451355,
      "learning_rate": 4.235261783702075e-06,
      "loss": 0.6856,
      "step": 5742
    },
    {
      "epoch": 1.5776679027606098,
      "grad_norm": 0.9370529651641846,
      "learning_rate": 4.232513398378453e-06,
      "loss": 0.8799,
      "step": 5743
    },
    {
      "epoch": 1.5779425903035298,
      "grad_norm": 0.36055758595466614,
      "learning_rate": 4.229765013054831e-06,
      "loss": 0.7021,
      "step": 5744
    },
    {
      "epoch": 1.5782172778464498,
      "grad_norm": 0.5287807583808899,
      "learning_rate": 4.227016627731208e-06,
      "loss": 0.6167,
      "step": 5745
    },
    {
      "epoch": 1.5784919653893696,
      "grad_norm": 0.49759888648986816,
      "learning_rate": 4.224268242407586e-06,
      "loss": 0.8273,
      "step": 5746
    },
    {
      "epoch": 1.5787666529322895,
      "grad_norm": 0.6066952347755432,
      "learning_rate": 4.2215198570839635e-06,
      "loss": 0.7313,
      "step": 5747
    },
    {
      "epoch": 1.5790413404752095,
      "grad_norm": 0.43862876296043396,
      "learning_rate": 4.218771471760341e-06,
      "loss": 0.7529,
      "step": 5748
    },
    {
      "epoch": 1.5793160280181295,
      "grad_norm": 0.5669203400611877,
      "learning_rate": 4.216023086436718e-06,
      "loss": 0.7663,
      "step": 5749
    },
    {
      "epoch": 1.5795907155610494,
      "grad_norm": 0.40793129801750183,
      "learning_rate": 4.213274701113096e-06,
      "loss": 0.8157,
      "step": 5750
    },
    {
      "epoch": 1.5798654031039692,
      "grad_norm": 0.532331645488739,
      "learning_rate": 4.210526315789474e-06,
      "loss": 0.8488,
      "step": 5751
    },
    {
      "epoch": 1.5801400906468892,
      "grad_norm": 0.4046843647956848,
      "learning_rate": 4.207777930465852e-06,
      "loss": 0.7613,
      "step": 5752
    },
    {
      "epoch": 1.5804147781898092,
      "grad_norm": 0.3614465892314911,
      "learning_rate": 4.205029545142229e-06,
      "loss": 0.7734,
      "step": 5753
    },
    {
      "epoch": 1.580689465732729,
      "grad_norm": 0.3780049979686737,
      "learning_rate": 4.202281159818607e-06,
      "loss": 0.6267,
      "step": 5754
    },
    {
      "epoch": 1.5809641532756489,
      "grad_norm": 0.40593063831329346,
      "learning_rate": 4.199532774494984e-06,
      "loss": 0.6521,
      "step": 5755
    },
    {
      "epoch": 1.581238840818569,
      "grad_norm": 0.4816504120826721,
      "learning_rate": 4.196784389171363e-06,
      "loss": 0.5778,
      "step": 5756
    },
    {
      "epoch": 1.581513528361489,
      "grad_norm": 0.5015629529953003,
      "learning_rate": 4.19403600384774e-06,
      "loss": 0.7089,
      "step": 5757
    },
    {
      "epoch": 1.5817882159044088,
      "grad_norm": 0.5015655755996704,
      "learning_rate": 4.1912876185241175e-06,
      "loss": 0.7719,
      "step": 5758
    },
    {
      "epoch": 1.5820629034473286,
      "grad_norm": 0.4971967041492462,
      "learning_rate": 4.1885392332004945e-06,
      "loss": 0.8483,
      "step": 5759
    },
    {
      "epoch": 1.5823375909902486,
      "grad_norm": 0.4496043026447296,
      "learning_rate": 4.185790847876872e-06,
      "loss": 0.6797,
      "step": 5760
    },
    {
      "epoch": 1.5826122785331687,
      "grad_norm": 0.4454754590988159,
      "learning_rate": 4.18304246255325e-06,
      "loss": 0.7733,
      "step": 5761
    },
    {
      "epoch": 1.5828869660760885,
      "grad_norm": 0.5003180503845215,
      "learning_rate": 4.180294077229628e-06,
      "loss": 0.6977,
      "step": 5762
    },
    {
      "epoch": 1.5831616536190083,
      "grad_norm": 0.5005028247833252,
      "learning_rate": 4.177545691906005e-06,
      "loss": 0.7959,
      "step": 5763
    },
    {
      "epoch": 1.5834363411619283,
      "grad_norm": 0.4299142360687256,
      "learning_rate": 4.174797306582383e-06,
      "loss": 0.6613,
      "step": 5764
    },
    {
      "epoch": 1.5837110287048484,
      "grad_norm": 0.43123796582221985,
      "learning_rate": 4.172048921258761e-06,
      "loss": 0.791,
      "step": 5765
    },
    {
      "epoch": 1.5839857162477682,
      "grad_norm": 0.6137955784797668,
      "learning_rate": 4.169300535935139e-06,
      "loss": 0.7321,
      "step": 5766
    },
    {
      "epoch": 1.584260403790688,
      "grad_norm": 0.4272039830684662,
      "learning_rate": 4.166552150611516e-06,
      "loss": 0.6388,
      "step": 5767
    },
    {
      "epoch": 1.584535091333608,
      "grad_norm": 0.39076513051986694,
      "learning_rate": 4.163803765287894e-06,
      "loss": 0.9717,
      "step": 5768
    },
    {
      "epoch": 1.584809778876528,
      "grad_norm": 0.5087603330612183,
      "learning_rate": 4.1610553799642715e-06,
      "loss": 0.803,
      "step": 5769
    },
    {
      "epoch": 1.585084466419448,
      "grad_norm": 0.3619186580181122,
      "learning_rate": 4.158306994640649e-06,
      "loss": 0.7109,
      "step": 5770
    },
    {
      "epoch": 1.5853591539623677,
      "grad_norm": 0.4352250099182129,
      "learning_rate": 4.155558609317026e-06,
      "loss": 0.9065,
      "step": 5771
    },
    {
      "epoch": 1.5856338415052877,
      "grad_norm": 0.4310373365879059,
      "learning_rate": 4.152810223993404e-06,
      "loss": 0.6402,
      "step": 5772
    },
    {
      "epoch": 1.5859085290482078,
      "grad_norm": 0.42808935046195984,
      "learning_rate": 4.150061838669781e-06,
      "loss": 0.8184,
      "step": 5773
    },
    {
      "epoch": 1.5861832165911276,
      "grad_norm": 0.4356672465801239,
      "learning_rate": 4.147313453346159e-06,
      "loss": 0.7457,
      "step": 5774
    },
    {
      "epoch": 1.5864579041340474,
      "grad_norm": 0.466528981924057,
      "learning_rate": 4.144565068022537e-06,
      "loss": 0.8372,
      "step": 5775
    },
    {
      "epoch": 1.5867325916769675,
      "grad_norm": 0.506734311580658,
      "learning_rate": 4.141816682698915e-06,
      "loss": 0.8783,
      "step": 5776
    },
    {
      "epoch": 1.5870072792198875,
      "grad_norm": 0.5858774781227112,
      "learning_rate": 4.139068297375292e-06,
      "loss": 0.8268,
      "step": 5777
    },
    {
      "epoch": 1.5872819667628073,
      "grad_norm": 0.4214038550853729,
      "learning_rate": 4.13631991205167e-06,
      "loss": 0.7772,
      "step": 5778
    },
    {
      "epoch": 1.5875566543057271,
      "grad_norm": 0.5466591715812683,
      "learning_rate": 4.133571526728048e-06,
      "loss": 0.6415,
      "step": 5779
    },
    {
      "epoch": 1.5878313418486472,
      "grad_norm": 0.5342857241630554,
      "learning_rate": 4.1308231414044255e-06,
      "loss": 0.5748,
      "step": 5780
    },
    {
      "epoch": 1.5881060293915672,
      "grad_norm": 0.43652376532554626,
      "learning_rate": 4.1280747560808025e-06,
      "loss": 0.6409,
      "step": 5781
    },
    {
      "epoch": 1.588380716934487,
      "grad_norm": 0.599263072013855,
      "learning_rate": 4.12532637075718e-06,
      "loss": 0.7422,
      "step": 5782
    },
    {
      "epoch": 1.5886554044774068,
      "grad_norm": 0.5224542021751404,
      "learning_rate": 4.122577985433558e-06,
      "loss": 0.9236,
      "step": 5783
    },
    {
      "epoch": 1.5889300920203269,
      "grad_norm": 0.45316991209983826,
      "learning_rate": 4.119829600109936e-06,
      "loss": 0.4924,
      "step": 5784
    },
    {
      "epoch": 1.589204779563247,
      "grad_norm": 0.48102033138275146,
      "learning_rate": 4.117081214786313e-06,
      "loss": 0.6404,
      "step": 5785
    },
    {
      "epoch": 1.5894794671061667,
      "grad_norm": 0.5612022280693054,
      "learning_rate": 4.114332829462691e-06,
      "loss": 0.7578,
      "step": 5786
    },
    {
      "epoch": 1.5897541546490865,
      "grad_norm": 0.4503215551376343,
      "learning_rate": 4.111584444139068e-06,
      "loss": 0.8304,
      "step": 5787
    },
    {
      "epoch": 1.5900288421920066,
      "grad_norm": 0.4050649106502533,
      "learning_rate": 4.108836058815447e-06,
      "loss": 0.6608,
      "step": 5788
    },
    {
      "epoch": 1.5903035297349266,
      "grad_norm": 0.4254535734653473,
      "learning_rate": 4.106087673491824e-06,
      "loss": 0.757,
      "step": 5789
    },
    {
      "epoch": 1.5905782172778464,
      "grad_norm": 0.3902583420276642,
      "learning_rate": 4.103339288168202e-06,
      "loss": 0.7977,
      "step": 5790
    },
    {
      "epoch": 1.5908529048207662,
      "grad_norm": 0.40646374225616455,
      "learning_rate": 4.100590902844579e-06,
      "loss": 0.7937,
      "step": 5791
    },
    {
      "epoch": 1.5911275923636863,
      "grad_norm": 0.5049079656600952,
      "learning_rate": 4.0978425175209565e-06,
      "loss": 0.7226,
      "step": 5792
    },
    {
      "epoch": 1.5914022799066063,
      "grad_norm": 0.5628886222839355,
      "learning_rate": 4.095094132197334e-06,
      "loss": 0.8235,
      "step": 5793
    },
    {
      "epoch": 1.5916769674495261,
      "grad_norm": 0.565263569355011,
      "learning_rate": 4.092345746873712e-06,
      "loss": 0.7617,
      "step": 5794
    },
    {
      "epoch": 1.591951654992446,
      "grad_norm": 0.49574020504951477,
      "learning_rate": 4.089597361550089e-06,
      "loss": 0.8846,
      "step": 5795
    },
    {
      "epoch": 1.592226342535366,
      "grad_norm": 0.540187656879425,
      "learning_rate": 4.086848976226467e-06,
      "loss": 0.7221,
      "step": 5796
    },
    {
      "epoch": 1.592501030078286,
      "grad_norm": 0.4395543336868286,
      "learning_rate": 4.084100590902845e-06,
      "loss": 0.6484,
      "step": 5797
    },
    {
      "epoch": 1.5927757176212058,
      "grad_norm": 0.3816060423851013,
      "learning_rate": 4.081352205579223e-06,
      "loss": 0.8709,
      "step": 5798
    },
    {
      "epoch": 1.5930504051641257,
      "grad_norm": 0.4281570315361023,
      "learning_rate": 4.0786038202556e-06,
      "loss": 0.8898,
      "step": 5799
    },
    {
      "epoch": 1.5933250927070457,
      "grad_norm": 0.4307633638381958,
      "learning_rate": 4.075855434931978e-06,
      "loss": 0.7876,
      "step": 5800
    },
    {
      "epoch": 1.5935997802499657,
      "grad_norm": 0.5828182101249695,
      "learning_rate": 4.073107049608356e-06,
      "loss": 0.7324,
      "step": 5801
    },
    {
      "epoch": 1.5938744677928856,
      "grad_norm": 0.5326460599899292,
      "learning_rate": 4.0703586642847335e-06,
      "loss": 0.9638,
      "step": 5802
    },
    {
      "epoch": 1.5941491553358054,
      "grad_norm": 0.5201151967048645,
      "learning_rate": 4.0676102789611105e-06,
      "loss": 0.915,
      "step": 5803
    },
    {
      "epoch": 1.5944238428787254,
      "grad_norm": 0.4360696077346802,
      "learning_rate": 4.064861893637488e-06,
      "loss": 0.7906,
      "step": 5804
    },
    {
      "epoch": 1.5946985304216454,
      "grad_norm": 0.4128274619579315,
      "learning_rate": 4.0621135083138654e-06,
      "loss": 0.5392,
      "step": 5805
    },
    {
      "epoch": 1.5949732179645653,
      "grad_norm": 0.3418784439563751,
      "learning_rate": 4.059365122990243e-06,
      "loss": 0.7051,
      "step": 5806
    },
    {
      "epoch": 1.595247905507485,
      "grad_norm": 0.5869203805923462,
      "learning_rate": 4.056616737666621e-06,
      "loss": 0.7771,
      "step": 5807
    },
    {
      "epoch": 1.5955225930504051,
      "grad_norm": 0.3919033408164978,
      "learning_rate": 4.053868352342999e-06,
      "loss": 0.8672,
      "step": 5808
    },
    {
      "epoch": 1.5957972805933252,
      "grad_norm": 0.49680590629577637,
      "learning_rate": 4.051119967019376e-06,
      "loss": 0.8586,
      "step": 5809
    },
    {
      "epoch": 1.596071968136245,
      "grad_norm": 0.33630430698394775,
      "learning_rate": 4.048371581695754e-06,
      "loss": 0.4342,
      "step": 5810
    },
    {
      "epoch": 1.596346655679165,
      "grad_norm": 0.5780985951423645,
      "learning_rate": 4.045623196372132e-06,
      "loss": 0.8026,
      "step": 5811
    },
    {
      "epoch": 1.5966213432220848,
      "grad_norm": 0.4322604835033417,
      "learning_rate": 4.04287481104851e-06,
      "loss": 0.6637,
      "step": 5812
    },
    {
      "epoch": 1.5968960307650049,
      "grad_norm": 0.4930916428565979,
      "learning_rate": 4.040126425724887e-06,
      "loss": 0.7278,
      "step": 5813
    },
    {
      "epoch": 1.597170718307925,
      "grad_norm": 0.4773244261741638,
      "learning_rate": 4.0373780404012645e-06,
      "loss": 0.6409,
      "step": 5814
    },
    {
      "epoch": 1.5974454058508447,
      "grad_norm": 0.5181475281715393,
      "learning_rate": 4.034629655077642e-06,
      "loss": 0.8503,
      "step": 5815
    },
    {
      "epoch": 1.5977200933937645,
      "grad_norm": 0.42115506529808044,
      "learning_rate": 4.03188126975402e-06,
      "loss": 0.7754,
      "step": 5816
    },
    {
      "epoch": 1.5979947809366846,
      "grad_norm": 0.5683130621910095,
      "learning_rate": 4.029132884430397e-06,
      "loss": 0.6373,
      "step": 5817
    },
    {
      "epoch": 1.5982694684796046,
      "grad_norm": 0.4238966703414917,
      "learning_rate": 4.026384499106775e-06,
      "loss": 0.8389,
      "step": 5818
    },
    {
      "epoch": 1.5985441560225244,
      "grad_norm": 0.35399943590164185,
      "learning_rate": 4.023636113783152e-06,
      "loss": 0.443,
      "step": 5819
    },
    {
      "epoch": 1.5988188435654442,
      "grad_norm": 0.5220634937286377,
      "learning_rate": 4.02088772845953e-06,
      "loss": 0.7797,
      "step": 5820
    },
    {
      "epoch": 1.5990935311083643,
      "grad_norm": 0.5358604788780212,
      "learning_rate": 4.018139343135908e-06,
      "loss": 0.8067,
      "step": 5821
    },
    {
      "epoch": 1.5993682186512843,
      "grad_norm": 0.5250703692436218,
      "learning_rate": 4.015390957812286e-06,
      "loss": 0.6815,
      "step": 5822
    },
    {
      "epoch": 1.5996429061942041,
      "grad_norm": 0.40500006079673767,
      "learning_rate": 4.012642572488663e-06,
      "loss": 1.0504,
      "step": 5823
    },
    {
      "epoch": 1.599917593737124,
      "grad_norm": 0.5416041612625122,
      "learning_rate": 4.009894187165041e-06,
      "loss": 0.9933,
      "step": 5824
    },
    {
      "epoch": 1.600192281280044,
      "grad_norm": 0.48495012521743774,
      "learning_rate": 4.0071458018414185e-06,
      "loss": 0.7417,
      "step": 5825
    },
    {
      "epoch": 1.600466968822964,
      "grad_norm": 0.43337175250053406,
      "learning_rate": 4.004397416517796e-06,
      "loss": 0.5553,
      "step": 5826
    },
    {
      "epoch": 1.6007416563658838,
      "grad_norm": 0.5382654666900635,
      "learning_rate": 4.0016490311941734e-06,
      "loss": 0.7949,
      "step": 5827
    },
    {
      "epoch": 1.6010163439088037,
      "grad_norm": 0.6563187837600708,
      "learning_rate": 3.998900645870551e-06,
      "loss": 0.5314,
      "step": 5828
    },
    {
      "epoch": 1.6012910314517237,
      "grad_norm": 0.5928195118904114,
      "learning_rate": 3.996152260546929e-06,
      "loss": 0.7931,
      "step": 5829
    },
    {
      "epoch": 1.6015657189946437,
      "grad_norm": 0.4621230959892273,
      "learning_rate": 3.993403875223307e-06,
      "loss": 0.7432,
      "step": 5830
    },
    {
      "epoch": 1.6018404065375635,
      "grad_norm": 0.5149057507514954,
      "learning_rate": 3.990655489899684e-06,
      "loss": 0.5254,
      "step": 5831
    },
    {
      "epoch": 1.6021150940804834,
      "grad_norm": 0.48359808325767517,
      "learning_rate": 3.987907104576062e-06,
      "loss": 0.7184,
      "step": 5832
    },
    {
      "epoch": 1.6023897816234034,
      "grad_norm": 0.39496976137161255,
      "learning_rate": 3.985158719252439e-06,
      "loss": 0.4753,
      "step": 5833
    },
    {
      "epoch": 1.6026644691663234,
      "grad_norm": 0.4751993417739868,
      "learning_rate": 3.982410333928818e-06,
      "loss": 0.7768,
      "step": 5834
    },
    {
      "epoch": 1.6029391567092433,
      "grad_norm": 0.43794775009155273,
      "learning_rate": 3.979661948605195e-06,
      "loss": 0.6092,
      "step": 5835
    },
    {
      "epoch": 1.603213844252163,
      "grad_norm": 0.4573882222175598,
      "learning_rate": 3.9769135632815726e-06,
      "loss": 0.7937,
      "step": 5836
    },
    {
      "epoch": 1.603488531795083,
      "grad_norm": 0.3996470868587494,
      "learning_rate": 3.9741651779579496e-06,
      "loss": 0.8472,
      "step": 5837
    },
    {
      "epoch": 1.6037632193380031,
      "grad_norm": 0.4051952660083771,
      "learning_rate": 3.9714167926343274e-06,
      "loss": 0.8713,
      "step": 5838
    },
    {
      "epoch": 1.604037906880923,
      "grad_norm": 0.41290339827537537,
      "learning_rate": 3.968668407310705e-06,
      "loss": 0.7789,
      "step": 5839
    },
    {
      "epoch": 1.6043125944238428,
      "grad_norm": 0.38741549849510193,
      "learning_rate": 3.965920021987083e-06,
      "loss": 0.7016,
      "step": 5840
    },
    {
      "epoch": 1.6045872819667628,
      "grad_norm": 0.5329941511154175,
      "learning_rate": 3.96317163666346e-06,
      "loss": 0.7591,
      "step": 5841
    },
    {
      "epoch": 1.6048619695096829,
      "grad_norm": 0.4437634348869324,
      "learning_rate": 3.960423251339838e-06,
      "loss": 0.6127,
      "step": 5842
    },
    {
      "epoch": 1.6051366570526027,
      "grad_norm": 0.4213385283946991,
      "learning_rate": 3.957674866016216e-06,
      "loss": 0.5178,
      "step": 5843
    },
    {
      "epoch": 1.6054113445955225,
      "grad_norm": 0.45275431871414185,
      "learning_rate": 3.954926480692594e-06,
      "loss": 0.9276,
      "step": 5844
    },
    {
      "epoch": 1.6056860321384425,
      "grad_norm": 0.6022258400917053,
      "learning_rate": 3.952178095368971e-06,
      "loss": 0.7164,
      "step": 5845
    },
    {
      "epoch": 1.6059607196813626,
      "grad_norm": 0.44600650668144226,
      "learning_rate": 3.949429710045349e-06,
      "loss": 0.7299,
      "step": 5846
    },
    {
      "epoch": 1.6062354072242824,
      "grad_norm": 0.4298762083053589,
      "learning_rate": 3.9466813247217266e-06,
      "loss": 0.7303,
      "step": 5847
    },
    {
      "epoch": 1.6065100947672022,
      "grad_norm": 0.49929293990135193,
      "learning_rate": 3.9439329393981044e-06,
      "loss": 0.7866,
      "step": 5848
    },
    {
      "epoch": 1.6067847823101222,
      "grad_norm": 0.43289920687675476,
      "learning_rate": 3.9411845540744814e-06,
      "loss": 0.7214,
      "step": 5849
    },
    {
      "epoch": 1.6070594698530423,
      "grad_norm": 0.4142580032348633,
      "learning_rate": 3.938436168750859e-06,
      "loss": 0.5833,
      "step": 5850
    },
    {
      "epoch": 1.607334157395962,
      "grad_norm": 0.518233060836792,
      "learning_rate": 3.935687783427236e-06,
      "loss": 0.7163,
      "step": 5851
    },
    {
      "epoch": 1.607608844938882,
      "grad_norm": 0.42199012637138367,
      "learning_rate": 3.932939398103614e-06,
      "loss": 0.7588,
      "step": 5852
    },
    {
      "epoch": 1.607883532481802,
      "grad_norm": 0.34938254952430725,
      "learning_rate": 3.930191012779992e-06,
      "loss": 0.6965,
      "step": 5853
    },
    {
      "epoch": 1.608158220024722,
      "grad_norm": 0.4638349413871765,
      "learning_rate": 3.92744262745637e-06,
      "loss": 0.6876,
      "step": 5854
    },
    {
      "epoch": 1.6084329075676418,
      "grad_norm": 0.5121871829032898,
      "learning_rate": 3.924694242132747e-06,
      "loss": 0.8221,
      "step": 5855
    },
    {
      "epoch": 1.6087075951105616,
      "grad_norm": 0.41203173995018005,
      "learning_rate": 3.921945856809125e-06,
      "loss": 0.6167,
      "step": 5856
    },
    {
      "epoch": 1.6089822826534816,
      "grad_norm": 0.49416518211364746,
      "learning_rate": 3.919197471485503e-06,
      "loss": 0.7753,
      "step": 5857
    },
    {
      "epoch": 1.6092569701964017,
      "grad_norm": 0.4984978437423706,
      "learning_rate": 3.9164490861618806e-06,
      "loss": 0.7477,
      "step": 5858
    },
    {
      "epoch": 1.6095316577393215,
      "grad_norm": 0.5660351514816284,
      "learning_rate": 3.913700700838258e-06,
      "loss": 0.8392,
      "step": 5859
    },
    {
      "epoch": 1.6098063452822413,
      "grad_norm": 0.4104899764060974,
      "learning_rate": 3.9109523155146355e-06,
      "loss": 0.533,
      "step": 5860
    },
    {
      "epoch": 1.6100810328251614,
      "grad_norm": 0.42781251668930054,
      "learning_rate": 3.908203930191013e-06,
      "loss": 0.6154,
      "step": 5861
    },
    {
      "epoch": 1.6103557203680814,
      "grad_norm": 0.4304501414299011,
      "learning_rate": 3.905455544867391e-06,
      "loss": 0.5422,
      "step": 5862
    },
    {
      "epoch": 1.6106304079110012,
      "grad_norm": 0.48282673954963684,
      "learning_rate": 3.902707159543768e-06,
      "loss": 0.7353,
      "step": 5863
    },
    {
      "epoch": 1.610905095453921,
      "grad_norm": 0.47815176844596863,
      "learning_rate": 3.899958774220146e-06,
      "loss": 0.5869,
      "step": 5864
    },
    {
      "epoch": 1.611179782996841,
      "grad_norm": 0.6333217024803162,
      "learning_rate": 3.897210388896523e-06,
      "loss": 0.7905,
      "step": 5865
    },
    {
      "epoch": 1.611454470539761,
      "grad_norm": 0.580082356929779,
      "learning_rate": 3.894462003572901e-06,
      "loss": 0.6698,
      "step": 5866
    },
    {
      "epoch": 1.611729158082681,
      "grad_norm": 0.45607244968414307,
      "learning_rate": 3.891713618249279e-06,
      "loss": 0.768,
      "step": 5867
    },
    {
      "epoch": 1.6120038456256007,
      "grad_norm": 0.4022911787033081,
      "learning_rate": 3.888965232925657e-06,
      "loss": 0.8854,
      "step": 5868
    },
    {
      "epoch": 1.6122785331685208,
      "grad_norm": 0.6575073003768921,
      "learning_rate": 3.886216847602034e-06,
      "loss": 0.9147,
      "step": 5869
    },
    {
      "epoch": 1.6125532207114408,
      "grad_norm": 0.365483820438385,
      "learning_rate": 3.883468462278412e-06,
      "loss": 0.5627,
      "step": 5870
    },
    {
      "epoch": 1.6128279082543606,
      "grad_norm": 0.486639142036438,
      "learning_rate": 3.8807200769547895e-06,
      "loss": 0.6947,
      "step": 5871
    },
    {
      "epoch": 1.6131025957972804,
      "grad_norm": 0.5683345794677734,
      "learning_rate": 3.877971691631167e-06,
      "loss": 0.7578,
      "step": 5872
    },
    {
      "epoch": 1.6133772833402005,
      "grad_norm": 0.3498998284339905,
      "learning_rate": 3.875223306307544e-06,
      "loss": 0.8531,
      "step": 5873
    },
    {
      "epoch": 1.6136519708831205,
      "grad_norm": 0.4768506586551666,
      "learning_rate": 3.872474920983922e-06,
      "loss": 0.6644,
      "step": 5874
    },
    {
      "epoch": 1.6139266584260403,
      "grad_norm": 0.39376166462898254,
      "learning_rate": 3.8697265356603e-06,
      "loss": 0.8724,
      "step": 5875
    },
    {
      "epoch": 1.6142013459689601,
      "grad_norm": 0.3714107275009155,
      "learning_rate": 3.866978150336678e-06,
      "loss": 0.6748,
      "step": 5876
    },
    {
      "epoch": 1.6144760335118802,
      "grad_norm": 0.5075357556343079,
      "learning_rate": 3.864229765013055e-06,
      "loss": 0.8855,
      "step": 5877
    },
    {
      "epoch": 1.6147507210548002,
      "grad_norm": 0.46010303497314453,
      "learning_rate": 3.861481379689433e-06,
      "loss": 0.721,
      "step": 5878
    },
    {
      "epoch": 1.6150254085977203,
      "grad_norm": 0.3937021791934967,
      "learning_rate": 3.85873299436581e-06,
      "loss": 0.5123,
      "step": 5879
    },
    {
      "epoch": 1.61530009614064,
      "grad_norm": 0.47774094343185425,
      "learning_rate": 3.8559846090421886e-06,
      "loss": 0.9976,
      "step": 5880
    },
    {
      "epoch": 1.61557478368356,
      "grad_norm": 0.4303920269012451,
      "learning_rate": 3.853236223718566e-06,
      "loss": 0.8821,
      "step": 5881
    },
    {
      "epoch": 1.61584947122648,
      "grad_norm": 0.5377100110054016,
      "learning_rate": 3.8504878383949435e-06,
      "loss": 1.0295,
      "step": 5882
    },
    {
      "epoch": 1.6161241587694,
      "grad_norm": 0.5058221817016602,
      "learning_rate": 3.8477394530713205e-06,
      "loss": 0.8562,
      "step": 5883
    },
    {
      "epoch": 1.6163988463123198,
      "grad_norm": 0.5337246060371399,
      "learning_rate": 3.844991067747698e-06,
      "loss": 0.7208,
      "step": 5884
    },
    {
      "epoch": 1.6166735338552396,
      "grad_norm": 0.49568891525268555,
      "learning_rate": 3.842242682424076e-06,
      "loss": 0.8282,
      "step": 5885
    },
    {
      "epoch": 1.6169482213981596,
      "grad_norm": 0.5690940022468567,
      "learning_rate": 3.839494297100454e-06,
      "loss": 0.7162,
      "step": 5886
    },
    {
      "epoch": 1.6172229089410797,
      "grad_norm": 0.6386249661445618,
      "learning_rate": 3.836745911776831e-06,
      "loss": 0.6505,
      "step": 5887
    },
    {
      "epoch": 1.6174975964839995,
      "grad_norm": 0.406144380569458,
      "learning_rate": 3.833997526453209e-06,
      "loss": 0.818,
      "step": 5888
    },
    {
      "epoch": 1.6177722840269193,
      "grad_norm": 0.5297405123710632,
      "learning_rate": 3.831249141129587e-06,
      "loss": 0.8173,
      "step": 5889
    },
    {
      "epoch": 1.6180469715698393,
      "grad_norm": 0.3946855366230011,
      "learning_rate": 3.828500755805965e-06,
      "loss": 0.7292,
      "step": 5890
    },
    {
      "epoch": 1.6183216591127594,
      "grad_norm": 0.4641486406326294,
      "learning_rate": 3.825752370482342e-06,
      "loss": 0.6178,
      "step": 5891
    },
    {
      "epoch": 1.6185963466556792,
      "grad_norm": 0.41498711705207825,
      "learning_rate": 3.82300398515872e-06,
      "loss": 1.1071,
      "step": 5892
    },
    {
      "epoch": 1.618871034198599,
      "grad_norm": 0.4660887122154236,
      "learning_rate": 3.8202555998350975e-06,
      "loss": 0.7301,
      "step": 5893
    },
    {
      "epoch": 1.619145721741519,
      "grad_norm": 0.4548601508140564,
      "learning_rate": 3.817507214511475e-06,
      "loss": 0.5493,
      "step": 5894
    },
    {
      "epoch": 1.619420409284439,
      "grad_norm": 0.46510443091392517,
      "learning_rate": 3.814758829187852e-06,
      "loss": 0.8126,
      "step": 5895
    },
    {
      "epoch": 1.619695096827359,
      "grad_norm": 0.4107329845428467,
      "learning_rate": 3.8120104438642302e-06,
      "loss": 0.5211,
      "step": 5896
    },
    {
      "epoch": 1.6199697843702787,
      "grad_norm": 0.6390679478645325,
      "learning_rate": 3.8092620585406077e-06,
      "loss": 0.7695,
      "step": 5897
    },
    {
      "epoch": 1.6202444719131988,
      "grad_norm": 0.5466967225074768,
      "learning_rate": 3.8065136732169855e-06,
      "loss": 0.7236,
      "step": 5898
    },
    {
      "epoch": 1.6205191594561188,
      "grad_norm": 0.35227012634277344,
      "learning_rate": 3.803765287893363e-06,
      "loss": 0.7389,
      "step": 5899
    },
    {
      "epoch": 1.6207938469990386,
      "grad_norm": 0.37601909041404724,
      "learning_rate": 3.801016902569741e-06,
      "loss": 0.5891,
      "step": 5900
    },
    {
      "epoch": 1.6210685345419584,
      "grad_norm": 0.48196089267730713,
      "learning_rate": 3.7982685172461183e-06,
      "loss": 0.4565,
      "step": 5901
    },
    {
      "epoch": 1.6213432220848785,
      "grad_norm": 0.4789588451385498,
      "learning_rate": 3.795520131922496e-06,
      "loss": 0.6952,
      "step": 5902
    },
    {
      "epoch": 1.6216179096277985,
      "grad_norm": 0.5660492777824402,
      "learning_rate": 3.792771746598873e-06,
      "loss": 0.6447,
      "step": 5903
    },
    {
      "epoch": 1.6218925971707183,
      "grad_norm": 0.3837989866733551,
      "learning_rate": 3.7900233612752515e-06,
      "loss": 0.5256,
      "step": 5904
    },
    {
      "epoch": 1.6221672847136381,
      "grad_norm": 0.42445796728134155,
      "learning_rate": 3.7872749759516285e-06,
      "loss": 0.6597,
      "step": 5905
    },
    {
      "epoch": 1.6224419722565582,
      "grad_norm": 0.5334020256996155,
      "learning_rate": 3.7845265906280064e-06,
      "loss": 0.78,
      "step": 5906
    },
    {
      "epoch": 1.6227166597994782,
      "grad_norm": 0.3877374827861786,
      "learning_rate": 3.781778205304384e-06,
      "loss": 0.9179,
      "step": 5907
    },
    {
      "epoch": 1.622991347342398,
      "grad_norm": 0.43809860944747925,
      "learning_rate": 3.7790298199807617e-06,
      "loss": 0.7312,
      "step": 5908
    },
    {
      "epoch": 1.6232660348853178,
      "grad_norm": 0.5061585307121277,
      "learning_rate": 3.776281434657139e-06,
      "loss": 0.8495,
      "step": 5909
    },
    {
      "epoch": 1.6235407224282379,
      "grad_norm": 0.5366806983947754,
      "learning_rate": 3.773533049333517e-06,
      "loss": 0.6644,
      "step": 5910
    },
    {
      "epoch": 1.623815409971158,
      "grad_norm": 0.5357295870780945,
      "learning_rate": 3.7707846640098944e-06,
      "loss": 0.6771,
      "step": 5911
    },
    {
      "epoch": 1.6240900975140777,
      "grad_norm": 0.4312531054019928,
      "learning_rate": 3.7680362786862723e-06,
      "loss": 0.7883,
      "step": 5912
    },
    {
      "epoch": 1.6243647850569976,
      "grad_norm": 0.3827122449874878,
      "learning_rate": 3.7652878933626497e-06,
      "loss": 0.6106,
      "step": 5913
    },
    {
      "epoch": 1.6246394725999176,
      "grad_norm": 0.6545654535293579,
      "learning_rate": 3.7625395080390276e-06,
      "loss": 0.7866,
      "step": 5914
    },
    {
      "epoch": 1.6249141601428376,
      "grad_norm": 0.39699608087539673,
      "learning_rate": 3.759791122715405e-06,
      "loss": 0.6569,
      "step": 5915
    },
    {
      "epoch": 1.6251888476857574,
      "grad_norm": 0.39960429072380066,
      "learning_rate": 3.757042737391783e-06,
      "loss": 0.6347,
      "step": 5916
    },
    {
      "epoch": 1.6254635352286773,
      "grad_norm": 0.41062021255493164,
      "learning_rate": 3.7542943520681604e-06,
      "loss": 0.6504,
      "step": 5917
    },
    {
      "epoch": 1.6257382227715973,
      "grad_norm": 0.41100335121154785,
      "learning_rate": 3.7515459667445382e-06,
      "loss": 0.6951,
      "step": 5918
    },
    {
      "epoch": 1.6260129103145173,
      "grad_norm": 0.4201267957687378,
      "learning_rate": 3.7487975814209153e-06,
      "loss": 0.7689,
      "step": 5919
    },
    {
      "epoch": 1.6262875978574372,
      "grad_norm": 0.5667842626571655,
      "learning_rate": 3.7460491960972935e-06,
      "loss": 0.8731,
      "step": 5920
    },
    {
      "epoch": 1.626562285400357,
      "grad_norm": 0.33064377307891846,
      "learning_rate": 3.7433008107736706e-06,
      "loss": 0.6324,
      "step": 5921
    },
    {
      "epoch": 1.626836972943277,
      "grad_norm": 0.5134676694869995,
      "learning_rate": 3.7405524254500484e-06,
      "loss": 1.0361,
      "step": 5922
    },
    {
      "epoch": 1.627111660486197,
      "grad_norm": 0.47911423444747925,
      "learning_rate": 3.737804040126426e-06,
      "loss": 0.596,
      "step": 5923
    },
    {
      "epoch": 1.6273863480291169,
      "grad_norm": 0.521214485168457,
      "learning_rate": 3.7350556548028037e-06,
      "loss": 0.948,
      "step": 5924
    },
    {
      "epoch": 1.6276610355720367,
      "grad_norm": 0.4472801387310028,
      "learning_rate": 3.732307269479181e-06,
      "loss": 0.5896,
      "step": 5925
    },
    {
      "epoch": 1.6279357231149567,
      "grad_norm": 0.3497093915939331,
      "learning_rate": 3.729558884155559e-06,
      "loss": 0.553,
      "step": 5926
    },
    {
      "epoch": 1.6282104106578767,
      "grad_norm": 0.5906078815460205,
      "learning_rate": 3.7268104988319365e-06,
      "loss": 0.658,
      "step": 5927
    },
    {
      "epoch": 1.6284850982007966,
      "grad_norm": 0.5634285807609558,
      "learning_rate": 3.7240621135083144e-06,
      "loss": 0.7575,
      "step": 5928
    },
    {
      "epoch": 1.6287597857437164,
      "grad_norm": 0.5355270504951477,
      "learning_rate": 3.721313728184692e-06,
      "loss": 0.7181,
      "step": 5929
    },
    {
      "epoch": 1.6290344732866364,
      "grad_norm": 0.6154586672782898,
      "learning_rate": 3.7185653428610697e-06,
      "loss": 0.6398,
      "step": 5930
    },
    {
      "epoch": 1.6293091608295565,
      "grad_norm": 0.4611974358558655,
      "learning_rate": 3.715816957537447e-06,
      "loss": 0.9164,
      "step": 5931
    },
    {
      "epoch": 1.6295838483724763,
      "grad_norm": 0.4198198914527893,
      "learning_rate": 3.713068572213825e-06,
      "loss": 0.9694,
      "step": 5932
    },
    {
      "epoch": 1.629858535915396,
      "grad_norm": 0.5410616397857666,
      "learning_rate": 3.7103201868902024e-06,
      "loss": 0.7067,
      "step": 5933
    },
    {
      "epoch": 1.6301332234583161,
      "grad_norm": 0.4950576424598694,
      "learning_rate": 3.7075718015665803e-06,
      "loss": 0.8073,
      "step": 5934
    },
    {
      "epoch": 1.6304079110012362,
      "grad_norm": 0.5866024494171143,
      "learning_rate": 3.7048234162429573e-06,
      "loss": 0.8801,
      "step": 5935
    },
    {
      "epoch": 1.630682598544156,
      "grad_norm": 0.4700744152069092,
      "learning_rate": 3.702075030919335e-06,
      "loss": 0.9237,
      "step": 5936
    },
    {
      "epoch": 1.6309572860870758,
      "grad_norm": 0.483046293258667,
      "learning_rate": 3.6993266455957126e-06,
      "loss": 0.9345,
      "step": 5937
    },
    {
      "epoch": 1.6312319736299958,
      "grad_norm": 0.5522273778915405,
      "learning_rate": 3.6965782602720905e-06,
      "loss": 0.7881,
      "step": 5938
    },
    {
      "epoch": 1.6315066611729159,
      "grad_norm": 0.5603979229927063,
      "learning_rate": 3.693829874948468e-06,
      "loss": 0.5894,
      "step": 5939
    },
    {
      "epoch": 1.6317813487158357,
      "grad_norm": 0.5375959277153015,
      "learning_rate": 3.691081489624846e-06,
      "loss": 0.868,
      "step": 5940
    },
    {
      "epoch": 1.6320560362587555,
      "grad_norm": 0.37530869245529175,
      "learning_rate": 3.6883331043012233e-06,
      "loss": 0.6294,
      "step": 5941
    },
    {
      "epoch": 1.6323307238016755,
      "grad_norm": 0.508410632610321,
      "learning_rate": 3.685584718977601e-06,
      "loss": 0.9104,
      "step": 5942
    },
    {
      "epoch": 1.6326054113445956,
      "grad_norm": 0.5829809308052063,
      "learning_rate": 3.6828363336539786e-06,
      "loss": 0.7396,
      "step": 5943
    },
    {
      "epoch": 1.6328800988875154,
      "grad_norm": 0.38751092553138733,
      "learning_rate": 3.6800879483303564e-06,
      "loss": 0.737,
      "step": 5944
    },
    {
      "epoch": 1.6331547864304354,
      "grad_norm": 0.2962881624698639,
      "learning_rate": 3.677339563006734e-06,
      "loss": 1.078,
      "step": 5945
    },
    {
      "epoch": 1.6334294739733553,
      "grad_norm": 0.3947311341762543,
      "learning_rate": 3.6745911776831118e-06,
      "loss": 0.572,
      "step": 5946
    },
    {
      "epoch": 1.6337041615162753,
      "grad_norm": 0.43530312180519104,
      "learning_rate": 3.671842792359489e-06,
      "loss": 0.6306,
      "step": 5947
    },
    {
      "epoch": 1.6339788490591953,
      "grad_norm": 0.3924047648906708,
      "learning_rate": 3.669094407035867e-06,
      "loss": 0.7315,
      "step": 5948
    },
    {
      "epoch": 1.6342535366021151,
      "grad_norm": 0.4453970491886139,
      "learning_rate": 3.666346021712244e-06,
      "loss": 0.7367,
      "step": 5949
    },
    {
      "epoch": 1.634528224145035,
      "grad_norm": 0.48806536197662354,
      "learning_rate": 3.6635976363886224e-06,
      "loss": 0.7811,
      "step": 5950
    },
    {
      "epoch": 1.634802911687955,
      "grad_norm": 0.4107206165790558,
      "learning_rate": 3.6608492510649994e-06,
      "loss": 0.847,
      "step": 5951
    },
    {
      "epoch": 1.635077599230875,
      "grad_norm": 0.47424328327178955,
      "learning_rate": 3.6581008657413773e-06,
      "loss": 0.7068,
      "step": 5952
    },
    {
      "epoch": 1.6353522867737948,
      "grad_norm": 0.6546381115913391,
      "learning_rate": 3.6553524804177547e-06,
      "loss": 0.7361,
      "step": 5953
    },
    {
      "epoch": 1.6356269743167147,
      "grad_norm": 0.5167074799537659,
      "learning_rate": 3.6526040950941326e-06,
      "loss": 0.7987,
      "step": 5954
    },
    {
      "epoch": 1.6359016618596347,
      "grad_norm": 0.5111149549484253,
      "learning_rate": 3.64985570977051e-06,
      "loss": 0.7661,
      "step": 5955
    },
    {
      "epoch": 1.6361763494025547,
      "grad_norm": 0.6186175346374512,
      "learning_rate": 3.647107324446888e-06,
      "loss": 0.6868,
      "step": 5956
    },
    {
      "epoch": 1.6364510369454746,
      "grad_norm": 0.4715898931026459,
      "learning_rate": 3.6443589391232653e-06,
      "loss": 1.0341,
      "step": 5957
    },
    {
      "epoch": 1.6367257244883944,
      "grad_norm": 0.3689751625061035,
      "learning_rate": 3.641610553799643e-06,
      "loss": 0.7518,
      "step": 5958
    },
    {
      "epoch": 1.6370004120313144,
      "grad_norm": 0.4948360025882721,
      "learning_rate": 3.6388621684760206e-06,
      "loss": 0.6503,
      "step": 5959
    },
    {
      "epoch": 1.6372750995742344,
      "grad_norm": 0.3490253984928131,
      "learning_rate": 3.6361137831523985e-06,
      "loss": 0.8695,
      "step": 5960
    },
    {
      "epoch": 1.6375497871171543,
      "grad_norm": 0.5163859724998474,
      "learning_rate": 3.633365397828776e-06,
      "loss": 0.815,
      "step": 5961
    },
    {
      "epoch": 1.637824474660074,
      "grad_norm": 0.4372492730617523,
      "learning_rate": 3.630617012505154e-06,
      "loss": 0.8986,
      "step": 5962
    },
    {
      "epoch": 1.6380991622029941,
      "grad_norm": 0.5535122156143188,
      "learning_rate": 3.6278686271815313e-06,
      "loss": 0.5883,
      "step": 5963
    },
    {
      "epoch": 1.6383738497459142,
      "grad_norm": 0.4004654884338379,
      "learning_rate": 3.625120241857909e-06,
      "loss": 0.5708,
      "step": 5964
    },
    {
      "epoch": 1.638648537288834,
      "grad_norm": 0.3653557002544403,
      "learning_rate": 3.622371856534286e-06,
      "loss": 0.6468,
      "step": 5965
    },
    {
      "epoch": 1.6389232248317538,
      "grad_norm": 0.4777287542819977,
      "learning_rate": 3.6196234712106645e-06,
      "loss": 0.8508,
      "step": 5966
    },
    {
      "epoch": 1.6391979123746738,
      "grad_norm": 0.40693873167037964,
      "learning_rate": 3.6168750858870415e-06,
      "loss": 0.9062,
      "step": 5967
    },
    {
      "epoch": 1.6394725999175939,
      "grad_norm": 0.40409165620803833,
      "learning_rate": 3.6141267005634193e-06,
      "loss": 0.7093,
      "step": 5968
    },
    {
      "epoch": 1.6397472874605137,
      "grad_norm": 0.399631142616272,
      "learning_rate": 3.6113783152397968e-06,
      "loss": 0.5895,
      "step": 5969
    },
    {
      "epoch": 1.6400219750034335,
      "grad_norm": 0.4917164444923401,
      "learning_rate": 3.6086299299161747e-06,
      "loss": 0.7767,
      "step": 5970
    },
    {
      "epoch": 1.6402966625463535,
      "grad_norm": 0.40960413217544556,
      "learning_rate": 3.605881544592552e-06,
      "loss": 0.5948,
      "step": 5971
    },
    {
      "epoch": 1.6405713500892736,
      "grad_norm": 0.4614546597003937,
      "learning_rate": 3.60313315926893e-06,
      "loss": 0.8496,
      "step": 5972
    },
    {
      "epoch": 1.6408460376321934,
      "grad_norm": 0.44394171237945557,
      "learning_rate": 3.6003847739453074e-06,
      "loss": 0.7875,
      "step": 5973
    },
    {
      "epoch": 1.6411207251751132,
      "grad_norm": 0.5405049324035645,
      "learning_rate": 3.5976363886216853e-06,
      "loss": 0.5743,
      "step": 5974
    },
    {
      "epoch": 1.6413954127180332,
      "grad_norm": 0.6246528625488281,
      "learning_rate": 3.5948880032980627e-06,
      "loss": 0.6491,
      "step": 5975
    },
    {
      "epoch": 1.6416701002609533,
      "grad_norm": 0.49184441566467285,
      "learning_rate": 3.5921396179744406e-06,
      "loss": 0.7182,
      "step": 5976
    },
    {
      "epoch": 1.641944787803873,
      "grad_norm": 0.4353300631046295,
      "learning_rate": 3.589391232650818e-06,
      "loss": 0.6579,
      "step": 5977
    },
    {
      "epoch": 1.642219475346793,
      "grad_norm": 0.4574357271194458,
      "learning_rate": 3.586642847327196e-06,
      "loss": 0.6156,
      "step": 5978
    },
    {
      "epoch": 1.642494162889713,
      "grad_norm": 0.5714036822319031,
      "learning_rate": 3.5838944620035733e-06,
      "loss": 0.874,
      "step": 5979
    },
    {
      "epoch": 1.642768850432633,
      "grad_norm": 0.4143763780593872,
      "learning_rate": 3.5811460766799512e-06,
      "loss": 0.762,
      "step": 5980
    },
    {
      "epoch": 1.6430435379755528,
      "grad_norm": 0.7077419757843018,
      "learning_rate": 3.5783976913563282e-06,
      "loss": 0.7754,
      "step": 5981
    },
    {
      "epoch": 1.6433182255184726,
      "grad_norm": 0.5184651017189026,
      "learning_rate": 3.5756493060327065e-06,
      "loss": 0.8088,
      "step": 5982
    },
    {
      "epoch": 1.6435929130613927,
      "grad_norm": 0.41885051131248474,
      "learning_rate": 3.5729009207090835e-06,
      "loss": 0.5743,
      "step": 5983
    },
    {
      "epoch": 1.6438676006043127,
      "grad_norm": 0.4005752503871918,
      "learning_rate": 3.5701525353854614e-06,
      "loss": 0.6067,
      "step": 5984
    },
    {
      "epoch": 1.6441422881472325,
      "grad_norm": 0.46120741963386536,
      "learning_rate": 3.567404150061839e-06,
      "loss": 0.962,
      "step": 5985
    },
    {
      "epoch": 1.6444169756901523,
      "grad_norm": 0.43471887707710266,
      "learning_rate": 3.5646557647382167e-06,
      "loss": 0.5863,
      "step": 5986
    },
    {
      "epoch": 1.6446916632330724,
      "grad_norm": 0.45777028799057007,
      "learning_rate": 3.561907379414594e-06,
      "loss": 0.7485,
      "step": 5987
    },
    {
      "epoch": 1.6449663507759924,
      "grad_norm": 0.3430735170841217,
      "learning_rate": 3.559158994090972e-06,
      "loss": 0.9279,
      "step": 5988
    },
    {
      "epoch": 1.6452410383189122,
      "grad_norm": 0.38140690326690674,
      "learning_rate": 3.5564106087673495e-06,
      "loss": 0.6871,
      "step": 5989
    },
    {
      "epoch": 1.645515725861832,
      "grad_norm": 0.48528483510017395,
      "learning_rate": 3.5536622234437274e-06,
      "loss": 0.6321,
      "step": 5990
    },
    {
      "epoch": 1.645790413404752,
      "grad_norm": 0.38236647844314575,
      "learning_rate": 3.550913838120105e-06,
      "loss": 0.6305,
      "step": 5991
    },
    {
      "epoch": 1.646065100947672,
      "grad_norm": 0.4548792839050293,
      "learning_rate": 3.5481654527964827e-06,
      "loss": 0.8515,
      "step": 5992
    },
    {
      "epoch": 1.646339788490592,
      "grad_norm": 0.30026209354400635,
      "learning_rate": 3.54541706747286e-06,
      "loss": 0.4945,
      "step": 5993
    },
    {
      "epoch": 1.6466144760335117,
      "grad_norm": 0.5492051243782043,
      "learning_rate": 3.542668682149238e-06,
      "loss": 0.6698,
      "step": 5994
    },
    {
      "epoch": 1.6468891635764318,
      "grad_norm": 0.6041578650474548,
      "learning_rate": 3.5399202968256154e-06,
      "loss": 0.904,
      "step": 5995
    },
    {
      "epoch": 1.6471638511193518,
      "grad_norm": 0.6044710278511047,
      "learning_rate": 3.5371719115019933e-06,
      "loss": 0.7969,
      "step": 5996
    },
    {
      "epoch": 1.6474385386622716,
      "grad_norm": 0.4054805636405945,
      "learning_rate": 3.5344235261783703e-06,
      "loss": 1.0049,
      "step": 5997
    },
    {
      "epoch": 1.6477132262051915,
      "grad_norm": 0.4508221447467804,
      "learning_rate": 3.531675140854748e-06,
      "loss": 0.6602,
      "step": 5998
    },
    {
      "epoch": 1.6479879137481115,
      "grad_norm": 0.3312844932079315,
      "learning_rate": 3.5289267555311256e-06,
      "loss": 0.678,
      "step": 5999
    },
    {
      "epoch": 1.6482626012910315,
      "grad_norm": 0.462042897939682,
      "learning_rate": 3.5261783702075035e-06,
      "loss": 0.5523,
      "step": 6000
    },
    {
      "epoch": 1.6485372888339513,
      "grad_norm": 0.4966147541999817,
      "learning_rate": 3.523429984883881e-06,
      "loss": 0.6914,
      "step": 6001
    },
    {
      "epoch": 1.6488119763768712,
      "grad_norm": 0.4918941557407379,
      "learning_rate": 3.520681599560259e-06,
      "loss": 0.7538,
      "step": 6002
    },
    {
      "epoch": 1.6490866639197912,
      "grad_norm": 0.44634929299354553,
      "learning_rate": 3.5179332142366362e-06,
      "loss": 0.7584,
      "step": 6003
    },
    {
      "epoch": 1.6493613514627112,
      "grad_norm": 0.3812210261821747,
      "learning_rate": 3.515184828913014e-06,
      "loss": 0.844,
      "step": 6004
    },
    {
      "epoch": 1.649636039005631,
      "grad_norm": 0.6711182594299316,
      "learning_rate": 3.5124364435893916e-06,
      "loss": 0.6295,
      "step": 6005
    },
    {
      "epoch": 1.6499107265485509,
      "grad_norm": 0.5229851603507996,
      "learning_rate": 3.5096880582657694e-06,
      "loss": 0.6028,
      "step": 6006
    },
    {
      "epoch": 1.650185414091471,
      "grad_norm": 0.6182683706283569,
      "learning_rate": 3.506939672942147e-06,
      "loss": 0.7238,
      "step": 6007
    },
    {
      "epoch": 1.650460101634391,
      "grad_norm": 0.4726530611515045,
      "learning_rate": 3.5041912876185247e-06,
      "loss": 0.548,
      "step": 6008
    },
    {
      "epoch": 1.6507347891773108,
      "grad_norm": 0.5265671610832214,
      "learning_rate": 3.501442902294902e-06,
      "loss": 0.6643,
      "step": 6009
    },
    {
      "epoch": 1.6510094767202306,
      "grad_norm": 0.42503637075424194,
      "learning_rate": 3.49869451697128e-06,
      "loss": 0.6814,
      "step": 6010
    },
    {
      "epoch": 1.6512841642631506,
      "grad_norm": 0.423902690410614,
      "learning_rate": 3.4959461316476575e-06,
      "loss": 0.7304,
      "step": 6011
    },
    {
      "epoch": 1.6515588518060706,
      "grad_norm": 0.6062108874320984,
      "learning_rate": 3.4931977463240354e-06,
      "loss": 0.7872,
      "step": 6012
    },
    {
      "epoch": 1.6518335393489907,
      "grad_norm": 0.4270547032356262,
      "learning_rate": 3.4904493610004124e-06,
      "loss": 0.6788,
      "step": 6013
    },
    {
      "epoch": 1.6521082268919105,
      "grad_norm": 0.46758633852005005,
      "learning_rate": 3.4877009756767902e-06,
      "loss": 0.8036,
      "step": 6014
    },
    {
      "epoch": 1.6523829144348303,
      "grad_norm": 0.42420443892478943,
      "learning_rate": 3.4849525903531677e-06,
      "loss": 0.587,
      "step": 6015
    },
    {
      "epoch": 1.6526576019777504,
      "grad_norm": 0.5072359442710876,
      "learning_rate": 3.4822042050295456e-06,
      "loss": 0.9396,
      "step": 6016
    },
    {
      "epoch": 1.6529322895206704,
      "grad_norm": 0.4977132976055145,
      "learning_rate": 3.479455819705923e-06,
      "loss": 0.578,
      "step": 6017
    },
    {
      "epoch": 1.6532069770635902,
      "grad_norm": 0.45562508702278137,
      "learning_rate": 3.4767074343823005e-06,
      "loss": 0.7864,
      "step": 6018
    },
    {
      "epoch": 1.65348166460651,
      "grad_norm": 0.3036643862724304,
      "learning_rate": 3.4739590490586783e-06,
      "loss": 0.5985,
      "step": 6019
    },
    {
      "epoch": 1.65375635214943,
      "grad_norm": 0.4211210608482361,
      "learning_rate": 3.4712106637350558e-06,
      "loss": 0.5334,
      "step": 6020
    },
    {
      "epoch": 1.65403103969235,
      "grad_norm": 0.4297163188457489,
      "learning_rate": 3.4684622784114336e-06,
      "loss": 0.7978,
      "step": 6021
    },
    {
      "epoch": 1.65430572723527,
      "grad_norm": 0.3888187110424042,
      "learning_rate": 3.465713893087811e-06,
      "loss": 0.7008,
      "step": 6022
    },
    {
      "epoch": 1.6545804147781897,
      "grad_norm": 0.4012954831123352,
      "learning_rate": 3.462965507764189e-06,
      "loss": 0.9948,
      "step": 6023
    },
    {
      "epoch": 1.6548551023211098,
      "grad_norm": 0.4919377863407135,
      "learning_rate": 3.4602171224405664e-06,
      "loss": 0.6639,
      "step": 6024
    },
    {
      "epoch": 1.6551297898640298,
      "grad_norm": 0.557762622833252,
      "learning_rate": 3.4574687371169443e-06,
      "loss": 0.7598,
      "step": 6025
    },
    {
      "epoch": 1.6554044774069496,
      "grad_norm": 0.5079896450042725,
      "learning_rate": 3.4547203517933213e-06,
      "loss": 0.8721,
      "step": 6026
    },
    {
      "epoch": 1.6556791649498694,
      "grad_norm": 0.44684308767318726,
      "learning_rate": 3.451971966469699e-06,
      "loss": 0.6496,
      "step": 6027
    },
    {
      "epoch": 1.6559538524927895,
      "grad_norm": 0.48272961378097534,
      "learning_rate": 3.4492235811460766e-06,
      "loss": 0.7137,
      "step": 6028
    },
    {
      "epoch": 1.6562285400357095,
      "grad_norm": 0.3975396156311035,
      "learning_rate": 3.4464751958224545e-06,
      "loss": 0.8491,
      "step": 6029
    },
    {
      "epoch": 1.6565032275786293,
      "grad_norm": 0.47560587525367737,
      "learning_rate": 3.443726810498832e-06,
      "loss": 0.6949,
      "step": 6030
    },
    {
      "epoch": 1.6567779151215491,
      "grad_norm": 0.5208978056907654,
      "learning_rate": 3.4409784251752098e-06,
      "loss": 0.6868,
      "step": 6031
    },
    {
      "epoch": 1.6570526026644692,
      "grad_norm": 0.4125770926475525,
      "learning_rate": 3.438230039851587e-06,
      "loss": 0.4656,
      "step": 6032
    },
    {
      "epoch": 1.6573272902073892,
      "grad_norm": 0.39515218138694763,
      "learning_rate": 3.435481654527965e-06,
      "loss": 0.8142,
      "step": 6033
    },
    {
      "epoch": 1.657601977750309,
      "grad_norm": 0.5735121369361877,
      "learning_rate": 3.4327332692043425e-06,
      "loss": 0.9274,
      "step": 6034
    },
    {
      "epoch": 1.6578766652932289,
      "grad_norm": 0.49000290036201477,
      "learning_rate": 3.4299848838807204e-06,
      "loss": 0.7149,
      "step": 6035
    },
    {
      "epoch": 1.658151352836149,
      "grad_norm": 0.47160038352012634,
      "learning_rate": 3.427236498557098e-06,
      "loss": 0.6377,
      "step": 6036
    },
    {
      "epoch": 1.658426040379069,
      "grad_norm": 0.5258443355560303,
      "learning_rate": 3.4244881132334757e-06,
      "loss": 0.9087,
      "step": 6037
    },
    {
      "epoch": 1.6587007279219887,
      "grad_norm": 0.5037051439285278,
      "learning_rate": 3.421739727909853e-06,
      "loss": 0.9526,
      "step": 6038
    },
    {
      "epoch": 1.6589754154649086,
      "grad_norm": 0.6532189249992371,
      "learning_rate": 3.418991342586231e-06,
      "loss": 0.9192,
      "step": 6039
    },
    {
      "epoch": 1.6592501030078286,
      "grad_norm": 0.6794821619987488,
      "learning_rate": 3.416242957262608e-06,
      "loss": 0.7288,
      "step": 6040
    },
    {
      "epoch": 1.6595247905507486,
      "grad_norm": 0.47917860746383667,
      "learning_rate": 3.4134945719389863e-06,
      "loss": 0.5392,
      "step": 6041
    },
    {
      "epoch": 1.6597994780936685,
      "grad_norm": 0.45437631011009216,
      "learning_rate": 3.4107461866153633e-06,
      "loss": 0.8108,
      "step": 6042
    },
    {
      "epoch": 1.6600741656365883,
      "grad_norm": 0.40560483932495117,
      "learning_rate": 3.4079978012917412e-06,
      "loss": 0.8475,
      "step": 6043
    },
    {
      "epoch": 1.6603488531795083,
      "grad_norm": 0.5401124358177185,
      "learning_rate": 3.4052494159681187e-06,
      "loss": 0.9317,
      "step": 6044
    },
    {
      "epoch": 1.6606235407224283,
      "grad_norm": 0.6158139109611511,
      "learning_rate": 3.4025010306444965e-06,
      "loss": 0.9776,
      "step": 6045
    },
    {
      "epoch": 1.6608982282653482,
      "grad_norm": 0.507872462272644,
      "learning_rate": 3.399752645320874e-06,
      "loss": 0.7026,
      "step": 6046
    },
    {
      "epoch": 1.661172915808268,
      "grad_norm": 0.3939599394798279,
      "learning_rate": 3.397004259997252e-06,
      "loss": 0.6213,
      "step": 6047
    },
    {
      "epoch": 1.661447603351188,
      "grad_norm": 0.44196459650993347,
      "learning_rate": 3.3942558746736293e-06,
      "loss": 0.4561,
      "step": 6048
    },
    {
      "epoch": 1.661722290894108,
      "grad_norm": 0.49934306740760803,
      "learning_rate": 3.391507489350007e-06,
      "loss": 0.9669,
      "step": 6049
    },
    {
      "epoch": 1.6619969784370279,
      "grad_norm": 0.5359998941421509,
      "learning_rate": 3.3887591040263846e-06,
      "loss": 0.9091,
      "step": 6050
    },
    {
      "epoch": 1.6622716659799477,
      "grad_norm": 0.5034804940223694,
      "learning_rate": 3.3860107187027625e-06,
      "loss": 0.7793,
      "step": 6051
    },
    {
      "epoch": 1.6625463535228677,
      "grad_norm": 0.33284440636634827,
      "learning_rate": 3.38326233337914e-06,
      "loss": 0.5627,
      "step": 6052
    },
    {
      "epoch": 1.6628210410657878,
      "grad_norm": 0.5320950150489807,
      "learning_rate": 3.3805139480555178e-06,
      "loss": 0.832,
      "step": 6053
    },
    {
      "epoch": 1.6630957286087076,
      "grad_norm": 0.4890627861022949,
      "learning_rate": 3.3777655627318952e-06,
      "loss": 0.75,
      "step": 6054
    },
    {
      "epoch": 1.6633704161516274,
      "grad_norm": 0.5280139446258545,
      "learning_rate": 3.375017177408273e-06,
      "loss": 0.8156,
      "step": 6055
    },
    {
      "epoch": 1.6636451036945474,
      "grad_norm": 0.5015754103660583,
      "learning_rate": 3.37226879208465e-06,
      "loss": 0.7813,
      "step": 6056
    },
    {
      "epoch": 1.6639197912374675,
      "grad_norm": 0.3631986379623413,
      "learning_rate": 3.3695204067610284e-06,
      "loss": 0.9935,
      "step": 6057
    },
    {
      "epoch": 1.6641944787803873,
      "grad_norm": 0.6777940988540649,
      "learning_rate": 3.3667720214374054e-06,
      "loss": 0.7046,
      "step": 6058
    },
    {
      "epoch": 1.664469166323307,
      "grad_norm": 0.48021015524864197,
      "learning_rate": 3.3640236361137833e-06,
      "loss": 0.6092,
      "step": 6059
    },
    {
      "epoch": 1.6647438538662271,
      "grad_norm": 0.46390482783317566,
      "learning_rate": 3.3612752507901607e-06,
      "loss": 0.9172,
      "step": 6060
    },
    {
      "epoch": 1.6650185414091472,
      "grad_norm": 0.623595654964447,
      "learning_rate": 3.3585268654665386e-06,
      "loss": 0.8207,
      "step": 6061
    },
    {
      "epoch": 1.665293228952067,
      "grad_norm": 0.5273749828338623,
      "learning_rate": 3.355778480142916e-06,
      "loss": 0.5784,
      "step": 6062
    },
    {
      "epoch": 1.6655679164949868,
      "grad_norm": 0.4886229336261749,
      "learning_rate": 3.353030094819294e-06,
      "loss": 0.7786,
      "step": 6063
    },
    {
      "epoch": 1.6658426040379068,
      "grad_norm": 0.5126197338104248,
      "learning_rate": 3.3502817094956714e-06,
      "loss": 0.7175,
      "step": 6064
    },
    {
      "epoch": 1.6661172915808269,
      "grad_norm": 0.5285618901252747,
      "learning_rate": 3.3475333241720492e-06,
      "loss": 0.8097,
      "step": 6065
    },
    {
      "epoch": 1.6663919791237467,
      "grad_norm": 0.46832531690597534,
      "learning_rate": 3.3447849388484267e-06,
      "loss": 1.0354,
      "step": 6066
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.4949914813041687,
      "learning_rate": 3.3420365535248045e-06,
      "loss": 0.5353,
      "step": 6067
    },
    {
      "epoch": 1.6669413542095866,
      "grad_norm": 0.6981638073921204,
      "learning_rate": 3.339288168201182e-06,
      "loss": 0.8419,
      "step": 6068
    },
    {
      "epoch": 1.6672160417525066,
      "grad_norm": 0.3919355273246765,
      "learning_rate": 3.33653978287756e-06,
      "loss": 0.5598,
      "step": 6069
    },
    {
      "epoch": 1.6674907292954264,
      "grad_norm": 0.3373871147632599,
      "learning_rate": 3.3337913975539373e-06,
      "loss": 0.7328,
      "step": 6070
    },
    {
      "epoch": 1.6677654168383462,
      "grad_norm": 0.450834184885025,
      "learning_rate": 3.331043012230315e-06,
      "loss": 0.9907,
      "step": 6071
    },
    {
      "epoch": 1.6680401043812663,
      "grad_norm": 0.3909827470779419,
      "learning_rate": 3.328294626906692e-06,
      "loss": 0.9029,
      "step": 6072
    },
    {
      "epoch": 1.6683147919241863,
      "grad_norm": 0.40659138560295105,
      "learning_rate": 3.3255462415830705e-06,
      "loss": 0.8218,
      "step": 6073
    },
    {
      "epoch": 1.6685894794671061,
      "grad_norm": 0.45592793822288513,
      "learning_rate": 3.3227978562594475e-06,
      "loss": 0.8698,
      "step": 6074
    },
    {
      "epoch": 1.668864167010026,
      "grad_norm": 0.4501058757305145,
      "learning_rate": 3.3200494709358254e-06,
      "loss": 0.8903,
      "step": 6075
    },
    {
      "epoch": 1.669138854552946,
      "grad_norm": 0.43635308742523193,
      "learning_rate": 3.317301085612203e-06,
      "loss": 0.7428,
      "step": 6076
    },
    {
      "epoch": 1.669413542095866,
      "grad_norm": 0.43688008189201355,
      "learning_rate": 3.3145527002885807e-06,
      "loss": 1.0528,
      "step": 6077
    },
    {
      "epoch": 1.6696882296387858,
      "grad_norm": 0.49377474188804626,
      "learning_rate": 3.311804314964958e-06,
      "loss": 0.5765,
      "step": 6078
    },
    {
      "epoch": 1.6699629171817059,
      "grad_norm": 0.3122542202472687,
      "learning_rate": 3.309055929641336e-06,
      "loss": 0.8444,
      "step": 6079
    },
    {
      "epoch": 1.6702376047246257,
      "grad_norm": 0.44089311361312866,
      "learning_rate": 3.3063075443177134e-06,
      "loss": 0.7966,
      "step": 6080
    },
    {
      "epoch": 1.6705122922675457,
      "grad_norm": 0.579459547996521,
      "learning_rate": 3.3035591589940913e-06,
      "loss": 0.8103,
      "step": 6081
    },
    {
      "epoch": 1.6707869798104658,
      "grad_norm": 0.40723663568496704,
      "learning_rate": 3.3008107736704687e-06,
      "loss": 0.9311,
      "step": 6082
    },
    {
      "epoch": 1.6710616673533856,
      "grad_norm": 0.32647764682769775,
      "learning_rate": 3.2980623883468466e-06,
      "loss": 0.5275,
      "step": 6083
    },
    {
      "epoch": 1.6713363548963054,
      "grad_norm": 0.349981427192688,
      "learning_rate": 3.295314003023224e-06,
      "loss": 0.4929,
      "step": 6084
    },
    {
      "epoch": 1.6716110424392254,
      "grad_norm": 0.4530639052391052,
      "learning_rate": 3.292565617699602e-06,
      "loss": 0.736,
      "step": 6085
    },
    {
      "epoch": 1.6718857299821455,
      "grad_norm": 0.46037405729293823,
      "learning_rate": 3.2898172323759794e-06,
      "loss": 0.6629,
      "step": 6086
    },
    {
      "epoch": 1.6721604175250653,
      "grad_norm": 0.3706494867801666,
      "learning_rate": 3.2870688470523572e-06,
      "loss": 0.9425,
      "step": 6087
    },
    {
      "epoch": 1.672435105067985,
      "grad_norm": 0.6153491735458374,
      "learning_rate": 3.2843204617287343e-06,
      "loss": 0.78,
      "step": 6088
    },
    {
      "epoch": 1.6727097926109051,
      "grad_norm": 0.46789950132369995,
      "learning_rate": 3.281572076405112e-06,
      "loss": 0.7545,
      "step": 6089
    },
    {
      "epoch": 1.6729844801538252,
      "grad_norm": 0.420376181602478,
      "learning_rate": 3.2788236910814896e-06,
      "loss": 0.6772,
      "step": 6090
    },
    {
      "epoch": 1.673259167696745,
      "grad_norm": 0.3684846758842468,
      "learning_rate": 3.2760753057578674e-06,
      "loss": 0.5582,
      "step": 6091
    },
    {
      "epoch": 1.6735338552396648,
      "grad_norm": 0.5229328274726868,
      "learning_rate": 3.273326920434245e-06,
      "loss": 0.762,
      "step": 6092
    },
    {
      "epoch": 1.6738085427825848,
      "grad_norm": 0.45272648334503174,
      "learning_rate": 3.2705785351106227e-06,
      "loss": 0.8803,
      "step": 6093
    },
    {
      "epoch": 1.6740832303255049,
      "grad_norm": 0.5516226291656494,
      "learning_rate": 3.267830149787e-06,
      "loss": 0.6156,
      "step": 6094
    },
    {
      "epoch": 1.6743579178684247,
      "grad_norm": 0.40238600969314575,
      "learning_rate": 3.265081764463378e-06,
      "loss": 0.5133,
      "step": 6095
    },
    {
      "epoch": 1.6746326054113445,
      "grad_norm": 0.4031648635864258,
      "learning_rate": 3.2623333791397555e-06,
      "loss": 0.8493,
      "step": 6096
    },
    {
      "epoch": 1.6749072929542645,
      "grad_norm": 0.417434960603714,
      "learning_rate": 3.2595849938161334e-06,
      "loss": 0.9972,
      "step": 6097
    },
    {
      "epoch": 1.6751819804971846,
      "grad_norm": 0.4228263795375824,
      "learning_rate": 3.256836608492511e-06,
      "loss": 0.7042,
      "step": 6098
    },
    {
      "epoch": 1.6754566680401044,
      "grad_norm": 0.5076915621757507,
      "learning_rate": 3.2540882231688887e-06,
      "loss": 1.0078,
      "step": 6099
    },
    {
      "epoch": 1.6757313555830242,
      "grad_norm": 0.5616434812545776,
      "learning_rate": 3.251339837845266e-06,
      "loss": 0.6621,
      "step": 6100
    },
    {
      "epoch": 1.6760060431259443,
      "grad_norm": 0.4962306320667267,
      "learning_rate": 3.248591452521644e-06,
      "loss": 0.5946,
      "step": 6101
    },
    {
      "epoch": 1.6762807306688643,
      "grad_norm": 0.38410162925720215,
      "learning_rate": 3.245843067198021e-06,
      "loss": 0.4759,
      "step": 6102
    },
    {
      "epoch": 1.676555418211784,
      "grad_norm": 0.3889616131782532,
      "learning_rate": 3.2430946818743993e-06,
      "loss": 0.7766,
      "step": 6103
    },
    {
      "epoch": 1.676830105754704,
      "grad_norm": 0.5087049007415771,
      "learning_rate": 3.2403462965507763e-06,
      "loss": 0.9355,
      "step": 6104
    },
    {
      "epoch": 1.677104793297624,
      "grad_norm": 0.3957724869251251,
      "learning_rate": 3.237597911227154e-06,
      "loss": 0.8162,
      "step": 6105
    },
    {
      "epoch": 1.677379480840544,
      "grad_norm": 0.3349362015724182,
      "learning_rate": 3.2348495259035316e-06,
      "loss": 0.7014,
      "step": 6106
    },
    {
      "epoch": 1.6776541683834638,
      "grad_norm": 0.3224674463272095,
      "learning_rate": 3.2321011405799095e-06,
      "loss": 0.4997,
      "step": 6107
    },
    {
      "epoch": 1.6779288559263836,
      "grad_norm": 0.5522698163986206,
      "learning_rate": 3.229352755256287e-06,
      "loss": 0.6806,
      "step": 6108
    },
    {
      "epoch": 1.6782035434693037,
      "grad_norm": 0.4620193839073181,
      "learning_rate": 3.226604369932665e-06,
      "loss": 0.9176,
      "step": 6109
    },
    {
      "epoch": 1.6784782310122237,
      "grad_norm": 0.741051197052002,
      "learning_rate": 3.2238559846090423e-06,
      "loss": 0.7577,
      "step": 6110
    },
    {
      "epoch": 1.6787529185551435,
      "grad_norm": 0.5154979228973389,
      "learning_rate": 3.22110759928542e-06,
      "loss": 0.8242,
      "step": 6111
    },
    {
      "epoch": 1.6790276060980633,
      "grad_norm": 0.40946412086486816,
      "learning_rate": 3.2183592139617976e-06,
      "loss": 0.9976,
      "step": 6112
    },
    {
      "epoch": 1.6793022936409834,
      "grad_norm": 0.5117462873458862,
      "learning_rate": 3.2156108286381754e-06,
      "loss": 0.8418,
      "step": 6113
    },
    {
      "epoch": 1.6795769811839034,
      "grad_norm": 0.39169421792030334,
      "learning_rate": 3.212862443314553e-06,
      "loss": 1.0212,
      "step": 6114
    },
    {
      "epoch": 1.6798516687268232,
      "grad_norm": 0.5158714056015015,
      "learning_rate": 3.2101140579909308e-06,
      "loss": 0.4773,
      "step": 6115
    },
    {
      "epoch": 1.680126356269743,
      "grad_norm": 0.4924823045730591,
      "learning_rate": 3.207365672667308e-06,
      "loss": 0.7624,
      "step": 6116
    },
    {
      "epoch": 1.680401043812663,
      "grad_norm": 0.34125521779060364,
      "learning_rate": 3.204617287343686e-06,
      "loss": 0.635,
      "step": 6117
    },
    {
      "epoch": 1.6806757313555831,
      "grad_norm": 0.44719183444976807,
      "learning_rate": 3.201868902020063e-06,
      "loss": 0.7562,
      "step": 6118
    },
    {
      "epoch": 1.680950418898503,
      "grad_norm": 0.597598135471344,
      "learning_rate": 3.1991205166964414e-06,
      "loss": 0.8145,
      "step": 6119
    },
    {
      "epoch": 1.6812251064414228,
      "grad_norm": 0.48824355006217957,
      "learning_rate": 3.1963721313728184e-06,
      "loss": 0.7348,
      "step": 6120
    },
    {
      "epoch": 1.6814997939843428,
      "grad_norm": 0.3344753682613373,
      "learning_rate": 3.1936237460491963e-06,
      "loss": 0.6551,
      "step": 6121
    },
    {
      "epoch": 1.6817744815272628,
      "grad_norm": 0.4915921688079834,
      "learning_rate": 3.1908753607255737e-06,
      "loss": 0.8168,
      "step": 6122
    },
    {
      "epoch": 1.6820491690701826,
      "grad_norm": 0.44363221526145935,
      "learning_rate": 3.1881269754019516e-06,
      "loss": 0.8865,
      "step": 6123
    },
    {
      "epoch": 1.6823238566131025,
      "grad_norm": 0.5745982527732849,
      "learning_rate": 3.185378590078329e-06,
      "loss": 0.8346,
      "step": 6124
    },
    {
      "epoch": 1.6825985441560225,
      "grad_norm": 0.4089939594268799,
      "learning_rate": 3.182630204754707e-06,
      "loss": 0.5957,
      "step": 6125
    },
    {
      "epoch": 1.6828732316989425,
      "grad_norm": 0.47984498739242554,
      "learning_rate": 3.1798818194310843e-06,
      "loss": 0.6813,
      "step": 6126
    },
    {
      "epoch": 1.6831479192418624,
      "grad_norm": 0.49705150723457336,
      "learning_rate": 3.177133434107462e-06,
      "loss": 0.9668,
      "step": 6127
    },
    {
      "epoch": 1.6834226067847822,
      "grad_norm": 0.5042861700057983,
      "learning_rate": 3.1743850487838397e-06,
      "loss": 0.5643,
      "step": 6128
    },
    {
      "epoch": 1.6836972943277022,
      "grad_norm": 0.502717912197113,
      "learning_rate": 3.1716366634602175e-06,
      "loss": 0.8824,
      "step": 6129
    },
    {
      "epoch": 1.6839719818706222,
      "grad_norm": 0.4290507137775421,
      "learning_rate": 3.168888278136595e-06,
      "loss": 0.7598,
      "step": 6130
    },
    {
      "epoch": 1.684246669413542,
      "grad_norm": 0.44850829243659973,
      "learning_rate": 3.166139892812973e-06,
      "loss": 0.704,
      "step": 6131
    },
    {
      "epoch": 1.6845213569564619,
      "grad_norm": 0.522720992565155,
      "learning_rate": 3.1633915074893503e-06,
      "loss": 0.77,
      "step": 6132
    },
    {
      "epoch": 1.684796044499382,
      "grad_norm": 0.5051789879798889,
      "learning_rate": 3.160643122165728e-06,
      "loss": 0.977,
      "step": 6133
    },
    {
      "epoch": 1.685070732042302,
      "grad_norm": 0.46806347370147705,
      "learning_rate": 3.157894736842105e-06,
      "loss": 0.6601,
      "step": 6134
    },
    {
      "epoch": 1.6853454195852218,
      "grad_norm": 0.49069833755493164,
      "learning_rate": 3.1551463515184835e-06,
      "loss": 0.8161,
      "step": 6135
    },
    {
      "epoch": 1.6856201071281416,
      "grad_norm": 0.5877727270126343,
      "learning_rate": 3.1523979661948605e-06,
      "loss": 0.7339,
      "step": 6136
    },
    {
      "epoch": 1.6858947946710616,
      "grad_norm": 0.2809796929359436,
      "learning_rate": 3.1496495808712383e-06,
      "loss": 0.3784,
      "step": 6137
    },
    {
      "epoch": 1.6861694822139817,
      "grad_norm": 0.4120803475379944,
      "learning_rate": 3.146901195547616e-06,
      "loss": 0.9903,
      "step": 6138
    },
    {
      "epoch": 1.6864441697569015,
      "grad_norm": 0.4144696593284607,
      "learning_rate": 3.1441528102239937e-06,
      "loss": 0.7328,
      "step": 6139
    },
    {
      "epoch": 1.6867188572998213,
      "grad_norm": 0.4975622892379761,
      "learning_rate": 3.141404424900371e-06,
      "loss": 0.9517,
      "step": 6140
    },
    {
      "epoch": 1.6869935448427413,
      "grad_norm": 0.48649898171424866,
      "learning_rate": 3.138656039576749e-06,
      "loss": 0.6675,
      "step": 6141
    },
    {
      "epoch": 1.6872682323856614,
      "grad_norm": 0.4280545711517334,
      "learning_rate": 3.1359076542531264e-06,
      "loss": 0.8596,
      "step": 6142
    },
    {
      "epoch": 1.6875429199285812,
      "grad_norm": 0.41153937578201294,
      "learning_rate": 3.1331592689295043e-06,
      "loss": 0.718,
      "step": 6143
    },
    {
      "epoch": 1.687817607471501,
      "grad_norm": 0.49252834916114807,
      "learning_rate": 3.1304108836058817e-06,
      "loss": 0.8337,
      "step": 6144
    },
    {
      "epoch": 1.688092295014421,
      "grad_norm": 0.4934068024158478,
      "learning_rate": 3.1276624982822596e-06,
      "loss": 0.8941,
      "step": 6145
    },
    {
      "epoch": 1.688366982557341,
      "grad_norm": 0.5701661705970764,
      "learning_rate": 3.124914112958637e-06,
      "loss": 0.7525,
      "step": 6146
    },
    {
      "epoch": 1.6886416701002611,
      "grad_norm": 0.38291794061660767,
      "learning_rate": 3.122165727635015e-06,
      "loss": 0.6803,
      "step": 6147
    },
    {
      "epoch": 1.688916357643181,
      "grad_norm": 0.45131486654281616,
      "learning_rate": 3.1194173423113923e-06,
      "loss": 0.5236,
      "step": 6148
    },
    {
      "epoch": 1.6891910451861007,
      "grad_norm": 0.48010149598121643,
      "learning_rate": 3.1166689569877702e-06,
      "loss": 0.8398,
      "step": 6149
    },
    {
      "epoch": 1.6894657327290208,
      "grad_norm": 0.6064929962158203,
      "learning_rate": 3.1139205716641472e-06,
      "loss": 0.8215,
      "step": 6150
    },
    {
      "epoch": 1.6897404202719408,
      "grad_norm": 0.45573338866233826,
      "learning_rate": 3.111172186340525e-06,
      "loss": 0.8672,
      "step": 6151
    },
    {
      "epoch": 1.6900151078148606,
      "grad_norm": 0.4347550570964813,
      "learning_rate": 3.1084238010169026e-06,
      "loss": 0.5892,
      "step": 6152
    },
    {
      "epoch": 1.6902897953577805,
      "grad_norm": 0.4584459960460663,
      "learning_rate": 3.1056754156932804e-06,
      "loss": 0.8717,
      "step": 6153
    },
    {
      "epoch": 1.6905644829007005,
      "grad_norm": 0.39636752009391785,
      "learning_rate": 3.102927030369658e-06,
      "loss": 0.9093,
      "step": 6154
    },
    {
      "epoch": 1.6908391704436205,
      "grad_norm": 0.5656508207321167,
      "learning_rate": 3.1001786450460357e-06,
      "loss": 0.6149,
      "step": 6155
    },
    {
      "epoch": 1.6911138579865403,
      "grad_norm": 0.4144599735736847,
      "learning_rate": 3.097430259722413e-06,
      "loss": 0.9554,
      "step": 6156
    },
    {
      "epoch": 1.6913885455294602,
      "grad_norm": 0.3749900758266449,
      "learning_rate": 3.094681874398791e-06,
      "loss": 0.9905,
      "step": 6157
    },
    {
      "epoch": 1.6916632330723802,
      "grad_norm": 0.5283666253089905,
      "learning_rate": 3.0919334890751685e-06,
      "loss": 0.7973,
      "step": 6158
    },
    {
      "epoch": 1.6919379206153002,
      "grad_norm": 0.4514632225036621,
      "learning_rate": 3.0891851037515464e-06,
      "loss": 0.6898,
      "step": 6159
    },
    {
      "epoch": 1.69221260815822,
      "grad_norm": 0.40470099449157715,
      "learning_rate": 3.086436718427924e-06,
      "loss": 0.8335,
      "step": 6160
    },
    {
      "epoch": 1.6924872957011399,
      "grad_norm": 0.3890436291694641,
      "learning_rate": 3.0836883331043017e-06,
      "loss": 0.8143,
      "step": 6161
    },
    {
      "epoch": 1.69276198324406,
      "grad_norm": 0.41294610500335693,
      "learning_rate": 3.080939947780679e-06,
      "loss": 0.8806,
      "step": 6162
    },
    {
      "epoch": 1.69303667078698,
      "grad_norm": 0.5820196866989136,
      "learning_rate": 3.078191562457057e-06,
      "loss": 0.9146,
      "step": 6163
    },
    {
      "epoch": 1.6933113583298998,
      "grad_norm": 0.3635047674179077,
      "learning_rate": 3.075443177133434e-06,
      "loss": 0.8678,
      "step": 6164
    },
    {
      "epoch": 1.6935860458728196,
      "grad_norm": 0.3675835132598877,
      "learning_rate": 3.0726947918098123e-06,
      "loss": 1.0414,
      "step": 6165
    },
    {
      "epoch": 1.6938607334157396,
      "grad_norm": 0.47551047801971436,
      "learning_rate": 3.0699464064861893e-06,
      "loss": 0.6147,
      "step": 6166
    },
    {
      "epoch": 1.6941354209586597,
      "grad_norm": 0.4521588981151581,
      "learning_rate": 3.067198021162567e-06,
      "loss": 0.7211,
      "step": 6167
    },
    {
      "epoch": 1.6944101085015795,
      "grad_norm": 0.4009525179862976,
      "learning_rate": 3.0644496358389446e-06,
      "loss": 0.884,
      "step": 6168
    },
    {
      "epoch": 1.6946847960444993,
      "grad_norm": 0.5761016011238098,
      "learning_rate": 3.0617012505153225e-06,
      "loss": 0.7497,
      "step": 6169
    },
    {
      "epoch": 1.6949594835874193,
      "grad_norm": 0.46992412209510803,
      "learning_rate": 3.0589528651917e-06,
      "loss": 0.6764,
      "step": 6170
    },
    {
      "epoch": 1.6952341711303394,
      "grad_norm": 0.28563135862350464,
      "learning_rate": 3.056204479868078e-06,
      "loss": 0.8663,
      "step": 6171
    },
    {
      "epoch": 1.6955088586732592,
      "grad_norm": 0.46017563343048096,
      "learning_rate": 3.0534560945444552e-06,
      "loss": 0.6741,
      "step": 6172
    },
    {
      "epoch": 1.695783546216179,
      "grad_norm": 0.41115421056747437,
      "learning_rate": 3.050707709220833e-06,
      "loss": 0.6635,
      "step": 6173
    },
    {
      "epoch": 1.696058233759099,
      "grad_norm": 0.47881531715393066,
      "learning_rate": 3.0479593238972106e-06,
      "loss": 0.5326,
      "step": 6174
    },
    {
      "epoch": 1.696332921302019,
      "grad_norm": 0.5191245675086975,
      "learning_rate": 3.0452109385735884e-06,
      "loss": 0.7199,
      "step": 6175
    },
    {
      "epoch": 1.6966076088449389,
      "grad_norm": 0.5367603898048401,
      "learning_rate": 3.042462553249966e-06,
      "loss": 0.9498,
      "step": 6176
    },
    {
      "epoch": 1.6968822963878587,
      "grad_norm": 0.4966566860675812,
      "learning_rate": 3.0397141679263437e-06,
      "loss": 0.6834,
      "step": 6177
    },
    {
      "epoch": 1.6971569839307787,
      "grad_norm": 0.4778011441230774,
      "learning_rate": 3.036965782602721e-06,
      "loss": 0.6441,
      "step": 6178
    },
    {
      "epoch": 1.6974316714736988,
      "grad_norm": 0.42681336402893066,
      "learning_rate": 3.034217397279099e-06,
      "loss": 0.5459,
      "step": 6179
    },
    {
      "epoch": 1.6977063590166186,
      "grad_norm": 0.3938364088535309,
      "learning_rate": 3.031469011955476e-06,
      "loss": 0.6575,
      "step": 6180
    },
    {
      "epoch": 1.6979810465595384,
      "grad_norm": 0.4970420300960541,
      "learning_rate": 3.0287206266318544e-06,
      "loss": 0.6963,
      "step": 6181
    },
    {
      "epoch": 1.6982557341024584,
      "grad_norm": 0.5060737729072571,
      "learning_rate": 3.0259722413082314e-06,
      "loss": 0.7566,
      "step": 6182
    },
    {
      "epoch": 1.6985304216453785,
      "grad_norm": 0.5710163116455078,
      "learning_rate": 3.0232238559846093e-06,
      "loss": 0.8405,
      "step": 6183
    },
    {
      "epoch": 1.6988051091882983,
      "grad_norm": 0.432682067155838,
      "learning_rate": 3.0204754706609867e-06,
      "loss": 0.7904,
      "step": 6184
    },
    {
      "epoch": 1.6990797967312181,
      "grad_norm": 0.41961508989334106,
      "learning_rate": 3.0177270853373646e-06,
      "loss": 0.9887,
      "step": 6185
    },
    {
      "epoch": 1.6993544842741382,
      "grad_norm": 0.3937271237373352,
      "learning_rate": 3.014978700013742e-06,
      "loss": 1.0672,
      "step": 6186
    },
    {
      "epoch": 1.6996291718170582,
      "grad_norm": 0.5582990050315857,
      "learning_rate": 3.01223031469012e-06,
      "loss": 0.8306,
      "step": 6187
    },
    {
      "epoch": 1.699903859359978,
      "grad_norm": 0.4721944332122803,
      "learning_rate": 3.0094819293664973e-06,
      "loss": 0.6917,
      "step": 6188
    },
    {
      "epoch": 1.7001785469028978,
      "grad_norm": 0.5857105851173401,
      "learning_rate": 3.006733544042875e-06,
      "loss": 0.6205,
      "step": 6189
    },
    {
      "epoch": 1.7004532344458179,
      "grad_norm": 0.629859209060669,
      "learning_rate": 3.0039851587192526e-06,
      "loss": 0.8307,
      "step": 6190
    },
    {
      "epoch": 1.700727921988738,
      "grad_norm": 0.4591158330440521,
      "learning_rate": 3.0012367733956305e-06,
      "loss": 0.5516,
      "step": 6191
    },
    {
      "epoch": 1.7010026095316577,
      "grad_norm": 0.4579952359199524,
      "learning_rate": 2.998488388072008e-06,
      "loss": 0.5166,
      "step": 6192
    },
    {
      "epoch": 1.7012772970745775,
      "grad_norm": 0.5434670448303223,
      "learning_rate": 2.995740002748386e-06,
      "loss": 0.6126,
      "step": 6193
    },
    {
      "epoch": 1.7015519846174976,
      "grad_norm": 0.4574035108089447,
      "learning_rate": 2.9929916174247633e-06,
      "loss": 0.8103,
      "step": 6194
    },
    {
      "epoch": 1.7018266721604176,
      "grad_norm": 0.5741196274757385,
      "learning_rate": 2.990243232101141e-06,
      "loss": 0.8872,
      "step": 6195
    },
    {
      "epoch": 1.7021013597033374,
      "grad_norm": 0.4015485942363739,
      "learning_rate": 2.987494846777518e-06,
      "loss": 1.0484,
      "step": 6196
    },
    {
      "epoch": 1.7023760472462572,
      "grad_norm": 0.5330895185470581,
      "learning_rate": 2.9847464614538964e-06,
      "loss": 0.9038,
      "step": 6197
    },
    {
      "epoch": 1.7026507347891773,
      "grad_norm": 0.4886147975921631,
      "learning_rate": 2.9819980761302735e-06,
      "loss": 0.7028,
      "step": 6198
    },
    {
      "epoch": 1.7029254223320973,
      "grad_norm": 0.2990241050720215,
      "learning_rate": 2.9792496908066513e-06,
      "loss": 0.5516,
      "step": 6199
    },
    {
      "epoch": 1.7032001098750171,
      "grad_norm": 0.5393416285514832,
      "learning_rate": 2.9765013054830288e-06,
      "loss": 0.8491,
      "step": 6200
    },
    {
      "epoch": 1.703474797417937,
      "grad_norm": 0.5193343758583069,
      "learning_rate": 2.9737529201594066e-06,
      "loss": 0.7441,
      "step": 6201
    },
    {
      "epoch": 1.703749484960857,
      "grad_norm": 0.4783848822116852,
      "learning_rate": 2.971004534835784e-06,
      "loss": 0.8405,
      "step": 6202
    },
    {
      "epoch": 1.704024172503777,
      "grad_norm": 0.41611340641975403,
      "learning_rate": 2.968256149512162e-06,
      "loss": 0.9868,
      "step": 6203
    },
    {
      "epoch": 1.7042988600466968,
      "grad_norm": 0.44905781745910645,
      "learning_rate": 2.9655077641885394e-06,
      "loss": 0.7205,
      "step": 6204
    },
    {
      "epoch": 1.7045735475896167,
      "grad_norm": 0.4937903881072998,
      "learning_rate": 2.9627593788649173e-06,
      "loss": 0.6074,
      "step": 6205
    },
    {
      "epoch": 1.7048482351325367,
      "grad_norm": 0.3265456259250641,
      "learning_rate": 2.9600109935412947e-06,
      "loss": 0.5905,
      "step": 6206
    },
    {
      "epoch": 1.7051229226754567,
      "grad_norm": 0.35910719633102417,
      "learning_rate": 2.9572626082176726e-06,
      "loss": 0.9205,
      "step": 6207
    },
    {
      "epoch": 1.7053976102183765,
      "grad_norm": 0.44654515385627747,
      "learning_rate": 2.95451422289405e-06,
      "loss": 0.728,
      "step": 6208
    },
    {
      "epoch": 1.7056722977612964,
      "grad_norm": 0.5379891991615295,
      "learning_rate": 2.951765837570428e-06,
      "loss": 0.6358,
      "step": 6209
    },
    {
      "epoch": 1.7059469853042164,
      "grad_norm": 0.3610733449459076,
      "learning_rate": 2.9490174522468053e-06,
      "loss": 0.7915,
      "step": 6210
    },
    {
      "epoch": 1.7062216728471364,
      "grad_norm": 0.43514084815979004,
      "learning_rate": 2.946269066923183e-06,
      "loss": 0.6036,
      "step": 6211
    },
    {
      "epoch": 1.7064963603900565,
      "grad_norm": 0.38929158449172974,
      "learning_rate": 2.9435206815995602e-06,
      "loss": 0.8117,
      "step": 6212
    },
    {
      "epoch": 1.7067710479329763,
      "grad_norm": 0.5639654397964478,
      "learning_rate": 2.940772296275938e-06,
      "loss": 0.923,
      "step": 6213
    },
    {
      "epoch": 1.707045735475896,
      "grad_norm": 0.4773043096065521,
      "learning_rate": 2.9380239109523155e-06,
      "loss": 0.8758,
      "step": 6214
    },
    {
      "epoch": 1.7073204230188161,
      "grad_norm": 0.42878103256225586,
      "learning_rate": 2.9352755256286934e-06,
      "loss": 0.6686,
      "step": 6215
    },
    {
      "epoch": 1.7075951105617362,
      "grad_norm": 0.3042755722999573,
      "learning_rate": 2.932527140305071e-06,
      "loss": 0.7392,
      "step": 6216
    },
    {
      "epoch": 1.707869798104656,
      "grad_norm": 0.3854241967201233,
      "learning_rate": 2.9297787549814487e-06,
      "loss": 0.6905,
      "step": 6217
    },
    {
      "epoch": 1.7081444856475758,
      "grad_norm": 0.5822088718414307,
      "learning_rate": 2.927030369657826e-06,
      "loss": 0.9002,
      "step": 6218
    },
    {
      "epoch": 1.7084191731904959,
      "grad_norm": 0.4917292296886444,
      "learning_rate": 2.924281984334204e-06,
      "loss": 0.8394,
      "step": 6219
    },
    {
      "epoch": 1.708693860733416,
      "grad_norm": 0.507908284664154,
      "learning_rate": 2.9215335990105815e-06,
      "loss": 0.7573,
      "step": 6220
    },
    {
      "epoch": 1.7089685482763357,
      "grad_norm": 0.4711517095565796,
      "learning_rate": 2.9187852136869593e-06,
      "loss": 0.9431,
      "step": 6221
    },
    {
      "epoch": 1.7092432358192555,
      "grad_norm": 0.44792357087135315,
      "learning_rate": 2.9160368283633368e-06,
      "loss": 0.5987,
      "step": 6222
    },
    {
      "epoch": 1.7095179233621756,
      "grad_norm": 0.5855966806411743,
      "learning_rate": 2.9132884430397146e-06,
      "loss": 0.8176,
      "step": 6223
    },
    {
      "epoch": 1.7097926109050956,
      "grad_norm": 0.544685959815979,
      "learning_rate": 2.910540057716092e-06,
      "loss": 1.0003,
      "step": 6224
    },
    {
      "epoch": 1.7100672984480154,
      "grad_norm": 0.46642300486564636,
      "learning_rate": 2.90779167239247e-06,
      "loss": 0.7309,
      "step": 6225
    },
    {
      "epoch": 1.7103419859909352,
      "grad_norm": 0.27780312299728394,
      "learning_rate": 2.9050432870688474e-06,
      "loss": 0.4641,
      "step": 6226
    },
    {
      "epoch": 1.7106166735338553,
      "grad_norm": 0.4978647530078888,
      "learning_rate": 2.9022949017452253e-06,
      "loss": 0.8332,
      "step": 6227
    },
    {
      "epoch": 1.7108913610767753,
      "grad_norm": 0.6024034023284912,
      "learning_rate": 2.8995465164216023e-06,
      "loss": 0.884,
      "step": 6228
    },
    {
      "epoch": 1.7111660486196951,
      "grad_norm": 0.4319961667060852,
      "learning_rate": 2.89679813109798e-06,
      "loss": 0.8724,
      "step": 6229
    },
    {
      "epoch": 1.711440736162615,
      "grad_norm": 0.39522239565849304,
      "learning_rate": 2.8940497457743576e-06,
      "loss": 0.5543,
      "step": 6230
    },
    {
      "epoch": 1.711715423705535,
      "grad_norm": 0.4097713530063629,
      "learning_rate": 2.8913013604507355e-06,
      "loss": 0.4502,
      "step": 6231
    },
    {
      "epoch": 1.711990111248455,
      "grad_norm": 0.4512462019920349,
      "learning_rate": 2.888552975127113e-06,
      "loss": 0.6213,
      "step": 6232
    },
    {
      "epoch": 1.7122647987913748,
      "grad_norm": 0.38720807433128357,
      "learning_rate": 2.8858045898034908e-06,
      "loss": 0.8299,
      "step": 6233
    },
    {
      "epoch": 1.7125394863342946,
      "grad_norm": 0.3948467969894409,
      "learning_rate": 2.8830562044798682e-06,
      "loss": 0.5198,
      "step": 6234
    },
    {
      "epoch": 1.7128141738772147,
      "grad_norm": 0.4989863932132721,
      "learning_rate": 2.880307819156246e-06,
      "loss": 0.7156,
      "step": 6235
    },
    {
      "epoch": 1.7130888614201347,
      "grad_norm": 0.45753344893455505,
      "learning_rate": 2.8775594338326235e-06,
      "loss": 0.7769,
      "step": 6236
    },
    {
      "epoch": 1.7133635489630545,
      "grad_norm": 0.6397256851196289,
      "learning_rate": 2.8748110485090014e-06,
      "loss": 0.8336,
      "step": 6237
    },
    {
      "epoch": 1.7136382365059744,
      "grad_norm": 0.5538741946220398,
      "learning_rate": 2.872062663185379e-06,
      "loss": 0.5569,
      "step": 6238
    },
    {
      "epoch": 1.7139129240488944,
      "grad_norm": 0.4690588116645813,
      "learning_rate": 2.8693142778617567e-06,
      "loss": 0.8166,
      "step": 6239
    },
    {
      "epoch": 1.7141876115918144,
      "grad_norm": 0.3838154375553131,
      "learning_rate": 2.866565892538134e-06,
      "loss": 0.6489,
      "step": 6240
    },
    {
      "epoch": 1.7144622991347342,
      "grad_norm": 0.555220365524292,
      "learning_rate": 2.863817507214512e-06,
      "loss": 0.8236,
      "step": 6241
    },
    {
      "epoch": 1.714736986677654,
      "grad_norm": 0.4833832085132599,
      "learning_rate": 2.861069121890889e-06,
      "loss": 0.8286,
      "step": 6242
    },
    {
      "epoch": 1.715011674220574,
      "grad_norm": 0.5605918765068054,
      "learning_rate": 2.8583207365672673e-06,
      "loss": 0.5725,
      "step": 6243
    },
    {
      "epoch": 1.7152863617634941,
      "grad_norm": 0.527079701423645,
      "learning_rate": 2.8555723512436444e-06,
      "loss": 0.592,
      "step": 6244
    },
    {
      "epoch": 1.715561049306414,
      "grad_norm": 0.5399224758148193,
      "learning_rate": 2.8528239659200222e-06,
      "loss": 0.8038,
      "step": 6245
    },
    {
      "epoch": 1.7158357368493338,
      "grad_norm": 0.33524051308631897,
      "learning_rate": 2.8500755805963997e-06,
      "loss": 0.5703,
      "step": 6246
    },
    {
      "epoch": 1.7161104243922538,
      "grad_norm": 0.38154876232147217,
      "learning_rate": 2.8473271952727775e-06,
      "loss": 0.667,
      "step": 6247
    },
    {
      "epoch": 1.7163851119351738,
      "grad_norm": 0.4447574019432068,
      "learning_rate": 2.844578809949155e-06,
      "loss": 0.9242,
      "step": 6248
    },
    {
      "epoch": 1.7166597994780937,
      "grad_norm": 0.33674201369285583,
      "learning_rate": 2.841830424625533e-06,
      "loss": 0.8966,
      "step": 6249
    },
    {
      "epoch": 1.7169344870210135,
      "grad_norm": 0.45490601658821106,
      "learning_rate": 2.8390820393019103e-06,
      "loss": 0.7672,
      "step": 6250
    },
    {
      "epoch": 1.7172091745639335,
      "grad_norm": 0.4678831100463867,
      "learning_rate": 2.836333653978288e-06,
      "loss": 0.9591,
      "step": 6251
    },
    {
      "epoch": 1.7174838621068536,
      "grad_norm": 0.45207566022872925,
      "learning_rate": 2.8335852686546656e-06,
      "loss": 0.5692,
      "step": 6252
    },
    {
      "epoch": 1.7177585496497734,
      "grad_norm": 0.46852874755859375,
      "learning_rate": 2.8308368833310435e-06,
      "loss": 0.6385,
      "step": 6253
    },
    {
      "epoch": 1.7180332371926932,
      "grad_norm": 0.48950082063674927,
      "learning_rate": 2.828088498007421e-06,
      "loss": 0.689,
      "step": 6254
    },
    {
      "epoch": 1.7183079247356132,
      "grad_norm": 0.4680221676826477,
      "learning_rate": 2.825340112683799e-06,
      "loss": 0.6375,
      "step": 6255
    },
    {
      "epoch": 1.7185826122785333,
      "grad_norm": 0.3984021842479706,
      "learning_rate": 2.8225917273601762e-06,
      "loss": 0.7377,
      "step": 6256
    },
    {
      "epoch": 1.718857299821453,
      "grad_norm": 0.41457584500312805,
      "learning_rate": 2.819843342036554e-06,
      "loss": 0.789,
      "step": 6257
    },
    {
      "epoch": 1.719131987364373,
      "grad_norm": 0.5926675796508789,
      "learning_rate": 2.817094956712931e-06,
      "loss": 0.5445,
      "step": 6258
    },
    {
      "epoch": 1.719406674907293,
      "grad_norm": 0.5272892117500305,
      "learning_rate": 2.8143465713893094e-06,
      "loss": 0.5997,
      "step": 6259
    },
    {
      "epoch": 1.719681362450213,
      "grad_norm": 0.6077450513839722,
      "learning_rate": 2.8115981860656864e-06,
      "loss": 0.8849,
      "step": 6260
    },
    {
      "epoch": 1.7199560499931328,
      "grad_norm": 0.4452582001686096,
      "learning_rate": 2.8088498007420643e-06,
      "loss": 0.8649,
      "step": 6261
    },
    {
      "epoch": 1.7202307375360526,
      "grad_norm": 0.34623369574546814,
      "learning_rate": 2.8061014154184418e-06,
      "loss": 0.8637,
      "step": 6262
    },
    {
      "epoch": 1.7205054250789726,
      "grad_norm": 0.5053703784942627,
      "learning_rate": 2.8033530300948196e-06,
      "loss": 0.6657,
      "step": 6263
    },
    {
      "epoch": 1.7207801126218927,
      "grad_norm": 0.5617427229881287,
      "learning_rate": 2.800604644771197e-06,
      "loss": 0.8108,
      "step": 6264
    },
    {
      "epoch": 1.7210548001648125,
      "grad_norm": 0.4299059808254242,
      "learning_rate": 2.797856259447575e-06,
      "loss": 0.8426,
      "step": 6265
    },
    {
      "epoch": 1.7213294877077323,
      "grad_norm": 0.5722403526306152,
      "learning_rate": 2.7951078741239524e-06,
      "loss": 0.8354,
      "step": 6266
    },
    {
      "epoch": 1.7216041752506523,
      "grad_norm": 0.5158604979515076,
      "learning_rate": 2.7923594888003302e-06,
      "loss": 0.7103,
      "step": 6267
    },
    {
      "epoch": 1.7218788627935724,
      "grad_norm": 0.4299401044845581,
      "learning_rate": 2.7896111034767077e-06,
      "loss": 0.68,
      "step": 6268
    },
    {
      "epoch": 1.7221535503364922,
      "grad_norm": 0.5788416862487793,
      "learning_rate": 2.7868627181530856e-06,
      "loss": 0.6602,
      "step": 6269
    },
    {
      "epoch": 1.722428237879412,
      "grad_norm": 0.39289161562919617,
      "learning_rate": 2.784114332829463e-06,
      "loss": 0.8472,
      "step": 6270
    },
    {
      "epoch": 1.722702925422332,
      "grad_norm": 0.4258892834186554,
      "learning_rate": 2.781365947505841e-06,
      "loss": 0.6367,
      "step": 6271
    },
    {
      "epoch": 1.722977612965252,
      "grad_norm": 0.5484829545021057,
      "learning_rate": 2.7786175621822183e-06,
      "loss": 0.7671,
      "step": 6272
    },
    {
      "epoch": 1.723252300508172,
      "grad_norm": 0.5108931660652161,
      "learning_rate": 2.775869176858596e-06,
      "loss": 0.7396,
      "step": 6273
    },
    {
      "epoch": 1.7235269880510917,
      "grad_norm": 0.410184770822525,
      "learning_rate": 2.773120791534973e-06,
      "loss": 0.9223,
      "step": 6274
    },
    {
      "epoch": 1.7238016755940118,
      "grad_norm": 0.4041116535663605,
      "learning_rate": 2.7703724062113515e-06,
      "loss": 0.8416,
      "step": 6275
    },
    {
      "epoch": 1.7240763631369318,
      "grad_norm": 0.4723641574382782,
      "learning_rate": 2.7676240208877285e-06,
      "loss": 0.6095,
      "step": 6276
    },
    {
      "epoch": 1.7243510506798516,
      "grad_norm": 0.5257006287574768,
      "learning_rate": 2.7648756355641064e-06,
      "loss": 0.6312,
      "step": 6277
    },
    {
      "epoch": 1.7246257382227714,
      "grad_norm": 0.5567312240600586,
      "learning_rate": 2.762127250240484e-06,
      "loss": 0.7499,
      "step": 6278
    },
    {
      "epoch": 1.7249004257656915,
      "grad_norm": 0.369474858045578,
      "learning_rate": 2.7593788649168617e-06,
      "loss": 0.6282,
      "step": 6279
    },
    {
      "epoch": 1.7251751133086115,
      "grad_norm": 0.39042234420776367,
      "learning_rate": 2.756630479593239e-06,
      "loss": 0.7191,
      "step": 6280
    },
    {
      "epoch": 1.7254498008515315,
      "grad_norm": 0.5605190992355347,
      "learning_rate": 2.753882094269617e-06,
      "loss": 0.7441,
      "step": 6281
    },
    {
      "epoch": 1.7257244883944514,
      "grad_norm": 0.5084329843521118,
      "learning_rate": 2.7511337089459944e-06,
      "loss": 0.7221,
      "step": 6282
    },
    {
      "epoch": 1.7259991759373712,
      "grad_norm": 0.5486838221549988,
      "learning_rate": 2.7483853236223723e-06,
      "loss": 0.8247,
      "step": 6283
    },
    {
      "epoch": 1.7262738634802912,
      "grad_norm": 0.40444493293762207,
      "learning_rate": 2.7456369382987498e-06,
      "loss": 0.6239,
      "step": 6284
    },
    {
      "epoch": 1.7265485510232113,
      "grad_norm": 0.3882560431957245,
      "learning_rate": 2.7428885529751276e-06,
      "loss": 0.6166,
      "step": 6285
    },
    {
      "epoch": 1.726823238566131,
      "grad_norm": 0.5439590811729431,
      "learning_rate": 2.740140167651505e-06,
      "loss": 0.6753,
      "step": 6286
    },
    {
      "epoch": 1.7270979261090509,
      "grad_norm": 0.3573680818080902,
      "learning_rate": 2.737391782327883e-06,
      "loss": 0.4834,
      "step": 6287
    },
    {
      "epoch": 1.727372613651971,
      "grad_norm": 0.3409899175167084,
      "learning_rate": 2.7346433970042604e-06,
      "loss": 0.4963,
      "step": 6288
    },
    {
      "epoch": 1.727647301194891,
      "grad_norm": 0.5213469862937927,
      "learning_rate": 2.7318950116806383e-06,
      "loss": 0.7107,
      "step": 6289
    },
    {
      "epoch": 1.7279219887378108,
      "grad_norm": 0.3916848599910736,
      "learning_rate": 2.7291466263570153e-06,
      "loss": 0.7783,
      "step": 6290
    },
    {
      "epoch": 1.7281966762807306,
      "grad_norm": 0.3676528036594391,
      "learning_rate": 2.726398241033393e-06,
      "loss": 0.9866,
      "step": 6291
    },
    {
      "epoch": 1.7284713638236506,
      "grad_norm": 0.34711596369743347,
      "learning_rate": 2.7236498557097706e-06,
      "loss": 0.7258,
      "step": 6292
    },
    {
      "epoch": 1.7287460513665707,
      "grad_norm": 0.5790194272994995,
      "learning_rate": 2.7209014703861485e-06,
      "loss": 0.7387,
      "step": 6293
    },
    {
      "epoch": 1.7290207389094905,
      "grad_norm": 0.5355509519577026,
      "learning_rate": 2.718153085062526e-06,
      "loss": 0.6688,
      "step": 6294
    },
    {
      "epoch": 1.7292954264524103,
      "grad_norm": 0.4263617992401123,
      "learning_rate": 2.7154046997389038e-06,
      "loss": 1.0948,
      "step": 6295
    },
    {
      "epoch": 1.7295701139953303,
      "grad_norm": 0.3835175335407257,
      "learning_rate": 2.712656314415281e-06,
      "loss": 0.8994,
      "step": 6296
    },
    {
      "epoch": 1.7298448015382504,
      "grad_norm": 0.49522510170936584,
      "learning_rate": 2.709907929091659e-06,
      "loss": 0.8712,
      "step": 6297
    },
    {
      "epoch": 1.7301194890811702,
      "grad_norm": 0.40767011046409607,
      "learning_rate": 2.7071595437680365e-06,
      "loss": 0.7639,
      "step": 6298
    },
    {
      "epoch": 1.73039417662409,
      "grad_norm": 0.4746874272823334,
      "learning_rate": 2.7044111584444144e-06,
      "loss": 0.6825,
      "step": 6299
    },
    {
      "epoch": 1.73066886416701,
      "grad_norm": 0.5541802644729614,
      "learning_rate": 2.701662773120792e-06,
      "loss": 0.7312,
      "step": 6300
    },
    {
      "epoch": 1.73094355170993,
      "grad_norm": 0.6984193325042725,
      "learning_rate": 2.6989143877971697e-06,
      "loss": 0.6511,
      "step": 6301
    },
    {
      "epoch": 1.73121823925285,
      "grad_norm": 0.5879589915275574,
      "learning_rate": 2.696166002473547e-06,
      "loss": 0.7769,
      "step": 6302
    },
    {
      "epoch": 1.7314929267957697,
      "grad_norm": 0.42972710728645325,
      "learning_rate": 2.693417617149925e-06,
      "loss": 0.6232,
      "step": 6303
    },
    {
      "epoch": 1.7317676143386898,
      "grad_norm": 0.47406429052352905,
      "learning_rate": 2.690669231826302e-06,
      "loss": 0.6683,
      "step": 6304
    },
    {
      "epoch": 1.7320423018816098,
      "grad_norm": 0.5771329402923584,
      "learning_rate": 2.6879208465026803e-06,
      "loss": 0.6665,
      "step": 6305
    },
    {
      "epoch": 1.7323169894245296,
      "grad_norm": 0.30592697858810425,
      "learning_rate": 2.6851724611790573e-06,
      "loss": 0.5124,
      "step": 6306
    },
    {
      "epoch": 1.7325916769674494,
      "grad_norm": 0.5522046089172363,
      "learning_rate": 2.6824240758554352e-06,
      "loss": 0.7251,
      "step": 6307
    },
    {
      "epoch": 1.7328663645103695,
      "grad_norm": 0.4397338032722473,
      "learning_rate": 2.6796756905318127e-06,
      "loss": 0.8104,
      "step": 6308
    },
    {
      "epoch": 1.7331410520532895,
      "grad_norm": 0.5321007966995239,
      "learning_rate": 2.6769273052081905e-06,
      "loss": 0.6687,
      "step": 6309
    },
    {
      "epoch": 1.7334157395962093,
      "grad_norm": 0.38716068863868713,
      "learning_rate": 2.674178919884568e-06,
      "loss": 0.8859,
      "step": 6310
    },
    {
      "epoch": 1.7336904271391291,
      "grad_norm": 0.38782718777656555,
      "learning_rate": 2.671430534560946e-06,
      "loss": 0.9951,
      "step": 6311
    },
    {
      "epoch": 1.7339651146820492,
      "grad_norm": 0.7116121053695679,
      "learning_rate": 2.6686821492373233e-06,
      "loss": 0.811,
      "step": 6312
    },
    {
      "epoch": 1.7342398022249692,
      "grad_norm": 0.40067484974861145,
      "learning_rate": 2.665933763913701e-06,
      "loss": 0.5306,
      "step": 6313
    },
    {
      "epoch": 1.734514489767889,
      "grad_norm": 0.45895591378211975,
      "learning_rate": 2.6631853785900786e-06,
      "loss": 0.8028,
      "step": 6314
    },
    {
      "epoch": 1.7347891773108088,
      "grad_norm": 0.5707567930221558,
      "learning_rate": 2.6604369932664565e-06,
      "loss": 0.7612,
      "step": 6315
    },
    {
      "epoch": 1.7350638648537289,
      "grad_norm": 0.4178702235221863,
      "learning_rate": 2.657688607942834e-06,
      "loss": 0.5074,
      "step": 6316
    },
    {
      "epoch": 1.735338552396649,
      "grad_norm": 0.39099857211112976,
      "learning_rate": 2.6549402226192118e-06,
      "loss": 0.8302,
      "step": 6317
    },
    {
      "epoch": 1.7356132399395687,
      "grad_norm": 0.44055578112602234,
      "learning_rate": 2.6521918372955892e-06,
      "loss": 0.7672,
      "step": 6318
    },
    {
      "epoch": 1.7358879274824885,
      "grad_norm": 0.5123927593231201,
      "learning_rate": 2.649443451971967e-06,
      "loss": 0.9267,
      "step": 6319
    },
    {
      "epoch": 1.7361626150254086,
      "grad_norm": 0.5482785701751709,
      "learning_rate": 2.646695066648344e-06,
      "loss": 0.6611,
      "step": 6320
    },
    {
      "epoch": 1.7364373025683286,
      "grad_norm": 0.6028702855110168,
      "learning_rate": 2.6439466813247224e-06,
      "loss": 0.6089,
      "step": 6321
    },
    {
      "epoch": 1.7367119901112484,
      "grad_norm": 0.5534782409667969,
      "learning_rate": 2.6411982960010994e-06,
      "loss": 0.7517,
      "step": 6322
    },
    {
      "epoch": 1.7369866776541683,
      "grad_norm": 0.4899964928627014,
      "learning_rate": 2.6384499106774773e-06,
      "loss": 0.7939,
      "step": 6323
    },
    {
      "epoch": 1.7372613651970883,
      "grad_norm": 0.5329838395118713,
      "learning_rate": 2.6357015253538547e-06,
      "loss": 0.7789,
      "step": 6324
    },
    {
      "epoch": 1.7375360527400083,
      "grad_norm": 0.5444923639297485,
      "learning_rate": 2.6329531400302326e-06,
      "loss": 0.7944,
      "step": 6325
    },
    {
      "epoch": 1.7378107402829281,
      "grad_norm": 0.5303479433059692,
      "learning_rate": 2.63020475470661e-06,
      "loss": 0.729,
      "step": 6326
    },
    {
      "epoch": 1.738085427825848,
      "grad_norm": 0.37688130140304565,
      "learning_rate": 2.627456369382988e-06,
      "loss": 0.749,
      "step": 6327
    },
    {
      "epoch": 1.738360115368768,
      "grad_norm": 0.5877695083618164,
      "learning_rate": 2.6247079840593654e-06,
      "loss": 0.7079,
      "step": 6328
    },
    {
      "epoch": 1.738634802911688,
      "grad_norm": 0.46197211742401123,
      "learning_rate": 2.6219595987357432e-06,
      "loss": 0.6751,
      "step": 6329
    },
    {
      "epoch": 1.7389094904546079,
      "grad_norm": 0.40481260418891907,
      "learning_rate": 2.6192112134121207e-06,
      "loss": 0.8646,
      "step": 6330
    },
    {
      "epoch": 1.7391841779975277,
      "grad_norm": 0.4897391200065613,
      "learning_rate": 2.6164628280884985e-06,
      "loss": 0.8349,
      "step": 6331
    },
    {
      "epoch": 1.7394588655404477,
      "grad_norm": 0.5427249670028687,
      "learning_rate": 2.613714442764876e-06,
      "loss": 0.919,
      "step": 6332
    },
    {
      "epoch": 1.7397335530833677,
      "grad_norm": 0.5778437852859497,
      "learning_rate": 2.610966057441254e-06,
      "loss": 0.8125,
      "step": 6333
    },
    {
      "epoch": 1.7400082406262876,
      "grad_norm": 0.38529184460639954,
      "learning_rate": 2.6082176721176313e-06,
      "loss": 0.8625,
      "step": 6334
    },
    {
      "epoch": 1.7402829281692074,
      "grad_norm": 0.4473329782485962,
      "learning_rate": 2.6054692867940083e-06,
      "loss": 0.74,
      "step": 6335
    },
    {
      "epoch": 1.7405576157121274,
      "grad_norm": 0.4908509850502014,
      "learning_rate": 2.602720901470386e-06,
      "loss": 0.8577,
      "step": 6336
    },
    {
      "epoch": 1.7408323032550475,
      "grad_norm": 0.3150959312915802,
      "learning_rate": 2.5999725161467636e-06,
      "loss": 0.7871,
      "step": 6337
    },
    {
      "epoch": 1.7411069907979673,
      "grad_norm": 0.4632014334201813,
      "learning_rate": 2.5972241308231415e-06,
      "loss": 0.8784,
      "step": 6338
    },
    {
      "epoch": 1.741381678340887,
      "grad_norm": 0.4679774343967438,
      "learning_rate": 2.594475745499519e-06,
      "loss": 0.6903,
      "step": 6339
    },
    {
      "epoch": 1.7416563658838071,
      "grad_norm": 0.5062248706817627,
      "learning_rate": 2.591727360175897e-06,
      "loss": 0.806,
      "step": 6340
    },
    {
      "epoch": 1.7419310534267272,
      "grad_norm": 0.55002760887146,
      "learning_rate": 2.5889789748522743e-06,
      "loss": 0.606,
      "step": 6341
    },
    {
      "epoch": 1.742205740969647,
      "grad_norm": 0.38049396872520447,
      "learning_rate": 2.586230589528652e-06,
      "loss": 0.7302,
      "step": 6342
    },
    {
      "epoch": 1.7424804285125668,
      "grad_norm": 0.6658844351768494,
      "learning_rate": 2.5834822042050296e-06,
      "loss": 0.7084,
      "step": 6343
    },
    {
      "epoch": 1.7427551160554868,
      "grad_norm": 0.33578160405158997,
      "learning_rate": 2.5807338188814074e-06,
      "loss": 0.4608,
      "step": 6344
    },
    {
      "epoch": 1.7430298035984069,
      "grad_norm": 0.3644588589668274,
      "learning_rate": 2.577985433557785e-06,
      "loss": 0.7704,
      "step": 6345
    },
    {
      "epoch": 1.743304491141327,
      "grad_norm": 0.586098313331604,
      "learning_rate": 2.5752370482341627e-06,
      "loss": 0.6631,
      "step": 6346
    },
    {
      "epoch": 1.7435791786842467,
      "grad_norm": 0.49756425619125366,
      "learning_rate": 2.57248866291054e-06,
      "loss": 0.5876,
      "step": 6347
    },
    {
      "epoch": 1.7438538662271665,
      "grad_norm": 0.3793402910232544,
      "learning_rate": 2.569740277586918e-06,
      "loss": 0.6274,
      "step": 6348
    },
    {
      "epoch": 1.7441285537700866,
      "grad_norm": 0.3632967174053192,
      "learning_rate": 2.566991892263295e-06,
      "loss": 0.7742,
      "step": 6349
    },
    {
      "epoch": 1.7444032413130066,
      "grad_norm": 0.44907015562057495,
      "learning_rate": 2.5642435069396734e-06,
      "loss": 0.7858,
      "step": 6350
    },
    {
      "epoch": 1.7446779288559264,
      "grad_norm": 0.47353795170783997,
      "learning_rate": 2.5614951216160504e-06,
      "loss": 0.7865,
      "step": 6351
    },
    {
      "epoch": 1.7449526163988462,
      "grad_norm": 0.4124012887477875,
      "learning_rate": 2.5587467362924283e-06,
      "loss": 0.7294,
      "step": 6352
    },
    {
      "epoch": 1.7452273039417663,
      "grad_norm": 0.5240453481674194,
      "learning_rate": 2.5559983509688057e-06,
      "loss": 1.1125,
      "step": 6353
    },
    {
      "epoch": 1.7455019914846863,
      "grad_norm": 0.4153597056865692,
      "learning_rate": 2.5532499656451836e-06,
      "loss": 0.9319,
      "step": 6354
    },
    {
      "epoch": 1.7457766790276061,
      "grad_norm": 0.4576231837272644,
      "learning_rate": 2.550501580321561e-06,
      "loss": 0.6338,
      "step": 6355
    },
    {
      "epoch": 1.746051366570526,
      "grad_norm": 0.5368174314498901,
      "learning_rate": 2.547753194997939e-06,
      "loss": 0.6735,
      "step": 6356
    },
    {
      "epoch": 1.746326054113446,
      "grad_norm": 0.5363152027130127,
      "learning_rate": 2.5450048096743163e-06,
      "loss": 0.688,
      "step": 6357
    },
    {
      "epoch": 1.746600741656366,
      "grad_norm": 0.5539938807487488,
      "learning_rate": 2.542256424350694e-06,
      "loss": 0.6377,
      "step": 6358
    },
    {
      "epoch": 1.7468754291992858,
      "grad_norm": 0.5488123297691345,
      "learning_rate": 2.5395080390270716e-06,
      "loss": 0.8209,
      "step": 6359
    },
    {
      "epoch": 1.7471501167422057,
      "grad_norm": 0.40650421380996704,
      "learning_rate": 2.5367596537034495e-06,
      "loss": 0.8579,
      "step": 6360
    },
    {
      "epoch": 1.7474248042851257,
      "grad_norm": 0.3899066150188446,
      "learning_rate": 2.534011268379827e-06,
      "loss": 0.4833,
      "step": 6361
    },
    {
      "epoch": 1.7476994918280457,
      "grad_norm": 0.43105342984199524,
      "learning_rate": 2.531262883056205e-06,
      "loss": 0.6807,
      "step": 6362
    },
    {
      "epoch": 1.7479741793709656,
      "grad_norm": 0.5526109337806702,
      "learning_rate": 2.5285144977325823e-06,
      "loss": 0.9561,
      "step": 6363
    },
    {
      "epoch": 1.7482488669138854,
      "grad_norm": 0.3704740107059479,
      "learning_rate": 2.52576611240896e-06,
      "loss": 0.7178,
      "step": 6364
    },
    {
      "epoch": 1.7485235544568054,
      "grad_norm": 0.45476940274238586,
      "learning_rate": 2.523017727085337e-06,
      "loss": 0.7769,
      "step": 6365
    },
    {
      "epoch": 1.7487982419997254,
      "grad_norm": 0.44168272614479065,
      "learning_rate": 2.520269341761715e-06,
      "loss": 0.7186,
      "step": 6366
    },
    {
      "epoch": 1.7490729295426453,
      "grad_norm": 0.6415305137634277,
      "learning_rate": 2.5175209564380925e-06,
      "loss": 0.6786,
      "step": 6367
    },
    {
      "epoch": 1.749347617085565,
      "grad_norm": 0.49180230498313904,
      "learning_rate": 2.5147725711144703e-06,
      "loss": 0.7874,
      "step": 6368
    },
    {
      "epoch": 1.7496223046284851,
      "grad_norm": 0.4219960570335388,
      "learning_rate": 2.5120241857908478e-06,
      "loss": 0.74,
      "step": 6369
    },
    {
      "epoch": 1.7498969921714052,
      "grad_norm": 0.37955033779144287,
      "learning_rate": 2.5092758004672256e-06,
      "loss": 0.5808,
      "step": 6370
    },
    {
      "epoch": 1.750171679714325,
      "grad_norm": 0.5694497227668762,
      "learning_rate": 2.506527415143603e-06,
      "loss": 0.6688,
      "step": 6371
    },
    {
      "epoch": 1.7504463672572448,
      "grad_norm": 0.37983039021492004,
      "learning_rate": 2.503779029819981e-06,
      "loss": 0.5627,
      "step": 6372
    },
    {
      "epoch": 1.7507210548001648,
      "grad_norm": 0.41544339060783386,
      "learning_rate": 2.5010306444963584e-06,
      "loss": 0.7369,
      "step": 6373
    },
    {
      "epoch": 1.7509957423430849,
      "grad_norm": 0.45842495560646057,
      "learning_rate": 2.4982822591727363e-06,
      "loss": 0.8355,
      "step": 6374
    },
    {
      "epoch": 1.7512704298860047,
      "grad_norm": 0.4294470548629761,
      "learning_rate": 2.4955338738491137e-06,
      "loss": 0.7764,
      "step": 6375
    },
    {
      "epoch": 1.7515451174289245,
      "grad_norm": 0.4208733141422272,
      "learning_rate": 2.4927854885254916e-06,
      "loss": 0.8431,
      "step": 6376
    },
    {
      "epoch": 1.7518198049718445,
      "grad_norm": 0.4710511267185211,
      "learning_rate": 2.490037103201869e-06,
      "loss": 0.7004,
      "step": 6377
    },
    {
      "epoch": 1.7520944925147646,
      "grad_norm": 0.5494781136512756,
      "learning_rate": 2.487288717878247e-06,
      "loss": 0.4781,
      "step": 6378
    },
    {
      "epoch": 1.7523691800576844,
      "grad_norm": 0.3771415054798126,
      "learning_rate": 2.4845403325546243e-06,
      "loss": 0.6541,
      "step": 6379
    },
    {
      "epoch": 1.7526438676006042,
      "grad_norm": 0.4634026288986206,
      "learning_rate": 2.481791947231002e-06,
      "loss": 0.5879,
      "step": 6380
    },
    {
      "epoch": 1.7529185551435242,
      "grad_norm": 0.5324101448059082,
      "learning_rate": 2.4790435619073796e-06,
      "loss": 0.7879,
      "step": 6381
    },
    {
      "epoch": 1.7531932426864443,
      "grad_norm": 0.37765413522720337,
      "learning_rate": 2.476295176583757e-06,
      "loss": 0.5314,
      "step": 6382
    },
    {
      "epoch": 1.753467930229364,
      "grad_norm": 0.532014012336731,
      "learning_rate": 2.473546791260135e-06,
      "loss": 0.6415,
      "step": 6383
    },
    {
      "epoch": 1.753742617772284,
      "grad_norm": 0.40021267533302307,
      "learning_rate": 2.4707984059365124e-06,
      "loss": 0.8978,
      "step": 6384
    },
    {
      "epoch": 1.754017305315204,
      "grad_norm": 0.4596175253391266,
      "learning_rate": 2.4680500206128903e-06,
      "loss": 0.8564,
      "step": 6385
    },
    {
      "epoch": 1.754291992858124,
      "grad_norm": 0.40943029522895813,
      "learning_rate": 2.4653016352892677e-06,
      "loss": 0.8227,
      "step": 6386
    },
    {
      "epoch": 1.7545666804010438,
      "grad_norm": 0.5782352685928345,
      "learning_rate": 2.4625532499656456e-06,
      "loss": 0.6547,
      "step": 6387
    },
    {
      "epoch": 1.7548413679439636,
      "grad_norm": 0.5158299803733826,
      "learning_rate": 2.459804864642023e-06,
      "loss": 0.6518,
      "step": 6388
    },
    {
      "epoch": 1.7551160554868837,
      "grad_norm": 0.4603872001171112,
      "learning_rate": 2.457056479318401e-06,
      "loss": 0.6844,
      "step": 6389
    },
    {
      "epoch": 1.7553907430298037,
      "grad_norm": 0.40671640634536743,
      "learning_rate": 2.4543080939947783e-06,
      "loss": 0.8468,
      "step": 6390
    },
    {
      "epoch": 1.7556654305727235,
      "grad_norm": 0.5344324707984924,
      "learning_rate": 2.4515597086711558e-06,
      "loss": 0.8504,
      "step": 6391
    },
    {
      "epoch": 1.7559401181156433,
      "grad_norm": 0.4776952862739563,
      "learning_rate": 2.4488113233475337e-06,
      "loss": 0.6598,
      "step": 6392
    },
    {
      "epoch": 1.7562148056585634,
      "grad_norm": 0.448288232088089,
      "learning_rate": 2.446062938023911e-06,
      "loss": 0.7671,
      "step": 6393
    },
    {
      "epoch": 1.7564894932014834,
      "grad_norm": 0.4116191267967224,
      "learning_rate": 2.443314552700289e-06,
      "loss": 0.6825,
      "step": 6394
    },
    {
      "epoch": 1.7567641807444032,
      "grad_norm": 0.46797826886177063,
      "learning_rate": 2.4405661673766664e-06,
      "loss": 0.9139,
      "step": 6395
    },
    {
      "epoch": 1.757038868287323,
      "grad_norm": 0.3782229721546173,
      "learning_rate": 2.4378177820530443e-06,
      "loss": 0.9267,
      "step": 6396
    },
    {
      "epoch": 1.757313555830243,
      "grad_norm": 0.5958751440048218,
      "learning_rate": 2.4350693967294217e-06,
      "loss": 0.9152,
      "step": 6397
    },
    {
      "epoch": 1.757588243373163,
      "grad_norm": 0.5423263907432556,
      "learning_rate": 2.432321011405799e-06,
      "loss": 0.6924,
      "step": 6398
    },
    {
      "epoch": 1.757862930916083,
      "grad_norm": 0.5135052800178528,
      "learning_rate": 2.429572626082177e-06,
      "loss": 0.8934,
      "step": 6399
    },
    {
      "epoch": 1.7581376184590027,
      "grad_norm": 0.45127394795417786,
      "learning_rate": 2.4268242407585545e-06,
      "loss": 0.7169,
      "step": 6400
    },
    {
      "epoch": 1.7584123060019228,
      "grad_norm": 0.5291427969932556,
      "learning_rate": 2.4240758554349323e-06,
      "loss": 0.8463,
      "step": 6401
    },
    {
      "epoch": 1.7586869935448428,
      "grad_norm": 0.4316544234752655,
      "learning_rate": 2.42132747011131e-06,
      "loss": 0.8849,
      "step": 6402
    },
    {
      "epoch": 1.7589616810877626,
      "grad_norm": 0.5965449213981628,
      "learning_rate": 2.4185790847876877e-06,
      "loss": 0.9583,
      "step": 6403
    },
    {
      "epoch": 1.7592363686306824,
      "grad_norm": 0.3827684819698334,
      "learning_rate": 2.415830699464065e-06,
      "loss": 0.4846,
      "step": 6404
    },
    {
      "epoch": 1.7595110561736025,
      "grad_norm": 0.43693044781684875,
      "learning_rate": 2.4130823141404425e-06,
      "loss": 0.5818,
      "step": 6405
    },
    {
      "epoch": 1.7597857437165225,
      "grad_norm": 0.5056318044662476,
      "learning_rate": 2.4103339288168204e-06,
      "loss": 0.7437,
      "step": 6406
    },
    {
      "epoch": 1.7600604312594423,
      "grad_norm": 0.4664318859577179,
      "learning_rate": 2.407585543493198e-06,
      "loss": 0.5184,
      "step": 6407
    },
    {
      "epoch": 1.7603351188023622,
      "grad_norm": 0.44625476002693176,
      "learning_rate": 2.4048371581695757e-06,
      "loss": 0.781,
      "step": 6408
    },
    {
      "epoch": 1.7606098063452822,
      "grad_norm": 0.5199610590934753,
      "learning_rate": 2.402088772845953e-06,
      "loss": 0.7997,
      "step": 6409
    },
    {
      "epoch": 1.7608844938882022,
      "grad_norm": 0.47256189584732056,
      "learning_rate": 2.399340387522331e-06,
      "loss": 0.6131,
      "step": 6410
    },
    {
      "epoch": 1.761159181431122,
      "grad_norm": 0.4744185507297516,
      "learning_rate": 2.3965920021987085e-06,
      "loss": 0.828,
      "step": 6411
    },
    {
      "epoch": 1.7614338689740419,
      "grad_norm": 0.36334657669067383,
      "learning_rate": 2.3938436168750863e-06,
      "loss": 0.6967,
      "step": 6412
    },
    {
      "epoch": 1.761708556516962,
      "grad_norm": 0.49378469586372375,
      "learning_rate": 2.391095231551464e-06,
      "loss": 0.7131,
      "step": 6413
    },
    {
      "epoch": 1.761983244059882,
      "grad_norm": 0.47690796852111816,
      "learning_rate": 2.3883468462278412e-06,
      "loss": 0.6234,
      "step": 6414
    },
    {
      "epoch": 1.762257931602802,
      "grad_norm": 0.25860872864723206,
      "learning_rate": 2.385598460904219e-06,
      "loss": 0.7104,
      "step": 6415
    },
    {
      "epoch": 1.7625326191457218,
      "grad_norm": 0.2800036370754242,
      "learning_rate": 2.3828500755805965e-06,
      "loss": 0.459,
      "step": 6416
    },
    {
      "epoch": 1.7628073066886416,
      "grad_norm": 0.3509023189544678,
      "learning_rate": 2.3801016902569744e-06,
      "loss": 0.7206,
      "step": 6417
    },
    {
      "epoch": 1.7630819942315616,
      "grad_norm": 0.6002823710441589,
      "learning_rate": 2.377353304933352e-06,
      "loss": 0.6,
      "step": 6418
    },
    {
      "epoch": 1.7633566817744817,
      "grad_norm": 0.39452242851257324,
      "learning_rate": 2.3746049196097297e-06,
      "loss": 0.8109,
      "step": 6419
    },
    {
      "epoch": 1.7636313693174015,
      "grad_norm": 0.556774377822876,
      "learning_rate": 2.371856534286107e-06,
      "loss": 0.9385,
      "step": 6420
    },
    {
      "epoch": 1.7639060568603213,
      "grad_norm": 0.4048871397972107,
      "learning_rate": 2.3691081489624846e-06,
      "loss": 0.9081,
      "step": 6421
    },
    {
      "epoch": 1.7641807444032414,
      "grad_norm": 0.3969275951385498,
      "learning_rate": 2.3663597636388625e-06,
      "loss": 0.7543,
      "step": 6422
    },
    {
      "epoch": 1.7644554319461614,
      "grad_norm": 0.3395959138870239,
      "learning_rate": 2.36361137831524e-06,
      "loss": 0.7926,
      "step": 6423
    },
    {
      "epoch": 1.7647301194890812,
      "grad_norm": 0.3337366580963135,
      "learning_rate": 2.360862992991618e-06,
      "loss": 0.6064,
      "step": 6424
    },
    {
      "epoch": 1.765004807032001,
      "grad_norm": 0.35216569900512695,
      "learning_rate": 2.3581146076679952e-06,
      "loss": 0.9891,
      "step": 6425
    },
    {
      "epoch": 1.765279494574921,
      "grad_norm": 0.4757758378982544,
      "learning_rate": 2.355366222344373e-06,
      "loss": 0.8736,
      "step": 6426
    },
    {
      "epoch": 1.765554182117841,
      "grad_norm": 0.4207930266857147,
      "learning_rate": 2.3526178370207506e-06,
      "loss": 1.0082,
      "step": 6427
    },
    {
      "epoch": 1.765828869660761,
      "grad_norm": 0.43539127707481384,
      "learning_rate": 2.3498694516971284e-06,
      "loss": 0.6692,
      "step": 6428
    },
    {
      "epoch": 1.7661035572036807,
      "grad_norm": 0.392057329416275,
      "learning_rate": 2.347121066373506e-06,
      "loss": 0.7309,
      "step": 6429
    },
    {
      "epoch": 1.7663782447466008,
      "grad_norm": 0.5728806257247925,
      "learning_rate": 2.3443726810498833e-06,
      "loss": 0.8429,
      "step": 6430
    },
    {
      "epoch": 1.7666529322895208,
      "grad_norm": 0.4076404273509979,
      "learning_rate": 2.341624295726261e-06,
      "loss": 0.5249,
      "step": 6431
    },
    {
      "epoch": 1.7669276198324406,
      "grad_norm": 0.48735684156417847,
      "learning_rate": 2.3388759104026386e-06,
      "loss": 0.6394,
      "step": 6432
    },
    {
      "epoch": 1.7672023073753604,
      "grad_norm": 0.44064632058143616,
      "learning_rate": 2.3361275250790165e-06,
      "loss": 0.7135,
      "step": 6433
    },
    {
      "epoch": 1.7674769949182805,
      "grad_norm": 0.6878612041473389,
      "learning_rate": 2.333379139755394e-06,
      "loss": 0.6585,
      "step": 6434
    },
    {
      "epoch": 1.7677516824612005,
      "grad_norm": 0.5521544814109802,
      "learning_rate": 2.330630754431772e-06,
      "loss": 0.6563,
      "step": 6435
    },
    {
      "epoch": 1.7680263700041203,
      "grad_norm": 0.4604271948337555,
      "learning_rate": 2.3278823691081492e-06,
      "loss": 0.8331,
      "step": 6436
    },
    {
      "epoch": 1.7683010575470401,
      "grad_norm": 0.5457484126091003,
      "learning_rate": 2.3251339837845267e-06,
      "loss": 0.6458,
      "step": 6437
    },
    {
      "epoch": 1.7685757450899602,
      "grad_norm": 0.5179538130760193,
      "learning_rate": 2.3223855984609046e-06,
      "loss": 0.81,
      "step": 6438
    },
    {
      "epoch": 1.7688504326328802,
      "grad_norm": 0.48642539978027344,
      "learning_rate": 2.319637213137282e-06,
      "loss": 1.0273,
      "step": 6439
    },
    {
      "epoch": 1.7691251201758,
      "grad_norm": 0.4659065902233124,
      "learning_rate": 2.31688882781366e-06,
      "loss": 0.5637,
      "step": 6440
    },
    {
      "epoch": 1.7693998077187199,
      "grad_norm": 0.3938581645488739,
      "learning_rate": 2.3141404424900373e-06,
      "loss": 0.6156,
      "step": 6441
    },
    {
      "epoch": 1.76967449526164,
      "grad_norm": 0.3577747941017151,
      "learning_rate": 2.311392057166415e-06,
      "loss": 0.8108,
      "step": 6442
    },
    {
      "epoch": 1.76994918280456,
      "grad_norm": 0.43394792079925537,
      "learning_rate": 2.3086436718427926e-06,
      "loss": 0.7937,
      "step": 6443
    },
    {
      "epoch": 1.7702238703474797,
      "grad_norm": 0.4914436936378479,
      "learning_rate": 2.30589528651917e-06,
      "loss": 0.8514,
      "step": 6444
    },
    {
      "epoch": 1.7704985578903996,
      "grad_norm": 0.5242630243301392,
      "learning_rate": 2.303146901195548e-06,
      "loss": 0.9136,
      "step": 6445
    },
    {
      "epoch": 1.7707732454333196,
      "grad_norm": 0.4812999665737152,
      "learning_rate": 2.3003985158719254e-06,
      "loss": 0.7412,
      "step": 6446
    },
    {
      "epoch": 1.7710479329762396,
      "grad_norm": 0.6055417656898499,
      "learning_rate": 2.2976501305483033e-06,
      "loss": 0.7819,
      "step": 6447
    },
    {
      "epoch": 1.7713226205191595,
      "grad_norm": 0.5159626007080078,
      "learning_rate": 2.2949017452246807e-06,
      "loss": 0.7066,
      "step": 6448
    },
    {
      "epoch": 1.7715973080620793,
      "grad_norm": 0.36583104729652405,
      "learning_rate": 2.2921533599010586e-06,
      "loss": 0.7856,
      "step": 6449
    },
    {
      "epoch": 1.7718719956049993,
      "grad_norm": 0.3550235629081726,
      "learning_rate": 2.289404974577436e-06,
      "loss": 0.5961,
      "step": 6450
    },
    {
      "epoch": 1.7721466831479193,
      "grad_norm": 0.43724948167800903,
      "learning_rate": 2.286656589253814e-06,
      "loss": 0.6058,
      "step": 6451
    },
    {
      "epoch": 1.7724213706908392,
      "grad_norm": 0.4969959557056427,
      "learning_rate": 2.2839082039301913e-06,
      "loss": 0.7739,
      "step": 6452
    },
    {
      "epoch": 1.772696058233759,
      "grad_norm": 0.45676180720329285,
      "learning_rate": 2.2811598186065688e-06,
      "loss": 0.7575,
      "step": 6453
    },
    {
      "epoch": 1.772970745776679,
      "grad_norm": 0.37214046716690063,
      "learning_rate": 2.2784114332829466e-06,
      "loss": 0.9164,
      "step": 6454
    },
    {
      "epoch": 1.773245433319599,
      "grad_norm": 0.5622801780700684,
      "learning_rate": 2.275663047959324e-06,
      "loss": 0.7997,
      "step": 6455
    },
    {
      "epoch": 1.7735201208625189,
      "grad_norm": 0.5380580425262451,
      "learning_rate": 2.272914662635702e-06,
      "loss": 0.8219,
      "step": 6456
    },
    {
      "epoch": 1.7737948084054387,
      "grad_norm": 0.4846932590007782,
      "learning_rate": 2.2701662773120794e-06,
      "loss": 0.4777,
      "step": 6457
    },
    {
      "epoch": 1.7740694959483587,
      "grad_norm": 0.5219236612319946,
      "learning_rate": 2.2674178919884573e-06,
      "loss": 0.6908,
      "step": 6458
    },
    {
      "epoch": 1.7743441834912788,
      "grad_norm": 0.4388272762298584,
      "learning_rate": 2.2646695066648347e-06,
      "loss": 0.7285,
      "step": 6459
    },
    {
      "epoch": 1.7746188710341986,
      "grad_norm": 0.47121936082839966,
      "learning_rate": 2.261921121341212e-06,
      "loss": 0.6198,
      "step": 6460
    },
    {
      "epoch": 1.7748935585771184,
      "grad_norm": 0.3548250198364258,
      "learning_rate": 2.25917273601759e-06,
      "loss": 0.6039,
      "step": 6461
    },
    {
      "epoch": 1.7751682461200384,
      "grad_norm": 0.5040969252586365,
      "learning_rate": 2.2564243506939675e-06,
      "loss": 0.7223,
      "step": 6462
    },
    {
      "epoch": 1.7754429336629585,
      "grad_norm": 0.3558255732059479,
      "learning_rate": 2.2536759653703453e-06,
      "loss": 0.5729,
      "step": 6463
    },
    {
      "epoch": 1.7757176212058783,
      "grad_norm": 0.4442790746688843,
      "learning_rate": 2.2509275800467228e-06,
      "loss": 0.875,
      "step": 6464
    },
    {
      "epoch": 1.775992308748798,
      "grad_norm": 0.5713890790939331,
      "learning_rate": 2.2481791947231006e-06,
      "loss": 0.7948,
      "step": 6465
    },
    {
      "epoch": 1.7762669962917181,
      "grad_norm": 0.3514251708984375,
      "learning_rate": 2.245430809399478e-06,
      "loss": 0.6418,
      "step": 6466
    },
    {
      "epoch": 1.7765416838346382,
      "grad_norm": 0.5598773956298828,
      "learning_rate": 2.2426824240758555e-06,
      "loss": 0.8021,
      "step": 6467
    },
    {
      "epoch": 1.776816371377558,
      "grad_norm": 0.44644400477409363,
      "learning_rate": 2.2399340387522334e-06,
      "loss": 0.8341,
      "step": 6468
    },
    {
      "epoch": 1.7770910589204778,
      "grad_norm": 0.4710898995399475,
      "learning_rate": 2.237185653428611e-06,
      "loss": 0.7321,
      "step": 6469
    },
    {
      "epoch": 1.7773657464633978,
      "grad_norm": 0.4585421681404114,
      "learning_rate": 2.2344372681049887e-06,
      "loss": 0.7796,
      "step": 6470
    },
    {
      "epoch": 1.7776404340063179,
      "grad_norm": 0.4675769507884979,
      "learning_rate": 2.231688882781366e-06,
      "loss": 0.6409,
      "step": 6471
    },
    {
      "epoch": 1.7779151215492377,
      "grad_norm": 0.3682332932949066,
      "learning_rate": 2.228940497457744e-06,
      "loss": 1.0722,
      "step": 6472
    },
    {
      "epoch": 1.7781898090921575,
      "grad_norm": 0.37285077571868896,
      "learning_rate": 2.2261921121341215e-06,
      "loss": 0.8448,
      "step": 6473
    },
    {
      "epoch": 1.7784644966350776,
      "grad_norm": 0.6878227591514587,
      "learning_rate": 2.2234437268104993e-06,
      "loss": 0.8204,
      "step": 6474
    },
    {
      "epoch": 1.7787391841779976,
      "grad_norm": 0.6383132338523865,
      "learning_rate": 2.2206953414868768e-06,
      "loss": 0.7694,
      "step": 6475
    },
    {
      "epoch": 1.7790138717209174,
      "grad_norm": 0.42029109597206116,
      "learning_rate": 2.2179469561632542e-06,
      "loss": 1.0268,
      "step": 6476
    },
    {
      "epoch": 1.7792885592638372,
      "grad_norm": 0.44565069675445557,
      "learning_rate": 2.215198570839632e-06,
      "loss": 0.9031,
      "step": 6477
    },
    {
      "epoch": 1.7795632468067573,
      "grad_norm": 0.5522034764289856,
      "learning_rate": 2.2124501855160095e-06,
      "loss": 0.8099,
      "step": 6478
    },
    {
      "epoch": 1.7798379343496773,
      "grad_norm": 0.5339244604110718,
      "learning_rate": 2.2097018001923874e-06,
      "loss": 0.6173,
      "step": 6479
    },
    {
      "epoch": 1.7801126218925973,
      "grad_norm": 0.4828771948814392,
      "learning_rate": 2.206953414868765e-06,
      "loss": 0.778,
      "step": 6480
    },
    {
      "epoch": 1.7803873094355172,
      "grad_norm": 0.4665094316005707,
      "learning_rate": 2.2042050295451427e-06,
      "loss": 0.7661,
      "step": 6481
    },
    {
      "epoch": 1.780661996978437,
      "grad_norm": 0.5154409408569336,
      "learning_rate": 2.20145664422152e-06,
      "loss": 0.6871,
      "step": 6482
    },
    {
      "epoch": 1.780936684521357,
      "grad_norm": 0.48203960061073303,
      "learning_rate": 2.1987082588978976e-06,
      "loss": 0.7055,
      "step": 6483
    },
    {
      "epoch": 1.781211372064277,
      "grad_norm": 0.4288630783557892,
      "learning_rate": 2.1959598735742755e-06,
      "loss": 0.5726,
      "step": 6484
    },
    {
      "epoch": 1.7814860596071969,
      "grad_norm": 0.564259946346283,
      "learning_rate": 2.193211488250653e-06,
      "loss": 0.907,
      "step": 6485
    },
    {
      "epoch": 1.7817607471501167,
      "grad_norm": 0.6584871411323547,
      "learning_rate": 2.1904631029270308e-06,
      "loss": 0.8252,
      "step": 6486
    },
    {
      "epoch": 1.7820354346930367,
      "grad_norm": 0.4074375331401825,
      "learning_rate": 2.1877147176034082e-06,
      "loss": 0.6276,
      "step": 6487
    },
    {
      "epoch": 1.7823101222359568,
      "grad_norm": 0.47269487380981445,
      "learning_rate": 2.184966332279786e-06,
      "loss": 0.7016,
      "step": 6488
    },
    {
      "epoch": 1.7825848097788766,
      "grad_norm": 0.56583571434021,
      "learning_rate": 2.1822179469561635e-06,
      "loss": 0.7047,
      "step": 6489
    },
    {
      "epoch": 1.7828594973217964,
      "grad_norm": 0.3455188572406769,
      "learning_rate": 2.1794695616325414e-06,
      "loss": 0.4621,
      "step": 6490
    },
    {
      "epoch": 1.7831341848647164,
      "grad_norm": 0.5489561557769775,
      "learning_rate": 2.176721176308919e-06,
      "loss": 0.7277,
      "step": 6491
    },
    {
      "epoch": 1.7834088724076365,
      "grad_norm": 0.566676139831543,
      "learning_rate": 2.1739727909852963e-06,
      "loss": 0.7292,
      "step": 6492
    },
    {
      "epoch": 1.7836835599505563,
      "grad_norm": 0.44596773386001587,
      "learning_rate": 2.1712244056616737e-06,
      "loss": 0.8998,
      "step": 6493
    },
    {
      "epoch": 1.783958247493476,
      "grad_norm": 0.40684667229652405,
      "learning_rate": 2.1684760203380516e-06,
      "loss": 0.8347,
      "step": 6494
    },
    {
      "epoch": 1.7842329350363961,
      "grad_norm": 0.4485931992530823,
      "learning_rate": 2.165727635014429e-06,
      "loss": 0.9203,
      "step": 6495
    },
    {
      "epoch": 1.7845076225793162,
      "grad_norm": 0.5765759348869324,
      "learning_rate": 2.1629792496908065e-06,
      "loss": 0.7947,
      "step": 6496
    },
    {
      "epoch": 1.784782310122236,
      "grad_norm": 0.4710905849933624,
      "learning_rate": 2.1602308643671844e-06,
      "loss": 0.7351,
      "step": 6497
    },
    {
      "epoch": 1.7850569976651558,
      "grad_norm": 0.5633881688117981,
      "learning_rate": 2.157482479043562e-06,
      "loss": 0.6849,
      "step": 6498
    },
    {
      "epoch": 1.7853316852080758,
      "grad_norm": 0.6200318932533264,
      "learning_rate": 2.1547340937199397e-06,
      "loss": 0.7924,
      "step": 6499
    },
    {
      "epoch": 1.7856063727509959,
      "grad_norm": 0.6424962878227234,
      "learning_rate": 2.151985708396317e-06,
      "loss": 0.8588,
      "step": 6500
    },
    {
      "epoch": 1.7858810602939157,
      "grad_norm": 0.5943189859390259,
      "learning_rate": 2.149237323072695e-06,
      "loss": 0.6079,
      "step": 6501
    },
    {
      "epoch": 1.7861557478368355,
      "grad_norm": 0.5216742157936096,
      "learning_rate": 2.1464889377490724e-06,
      "loss": 0.5846,
      "step": 6502
    },
    {
      "epoch": 1.7864304353797555,
      "grad_norm": 0.3992663025856018,
      "learning_rate": 2.1437405524254503e-06,
      "loss": 0.7957,
      "step": 6503
    },
    {
      "epoch": 1.7867051229226756,
      "grad_norm": 0.44697433710098267,
      "learning_rate": 2.1409921671018277e-06,
      "loss": 0.7427,
      "step": 6504
    },
    {
      "epoch": 1.7869798104655954,
      "grad_norm": 0.4478335678577423,
      "learning_rate": 2.138243781778205e-06,
      "loss": 1.088,
      "step": 6505
    },
    {
      "epoch": 1.7872544980085152,
      "grad_norm": 0.5003899335861206,
      "learning_rate": 2.135495396454583e-06,
      "loss": 0.8201,
      "step": 6506
    },
    {
      "epoch": 1.7875291855514353,
      "grad_norm": 0.4292943477630615,
      "learning_rate": 2.1327470111309605e-06,
      "loss": 0.6826,
      "step": 6507
    },
    {
      "epoch": 1.7878038730943553,
      "grad_norm": 0.5409943461418152,
      "learning_rate": 2.1299986258073384e-06,
      "loss": 0.6524,
      "step": 6508
    },
    {
      "epoch": 1.788078560637275,
      "grad_norm": 0.4319935441017151,
      "learning_rate": 2.127250240483716e-06,
      "loss": 0.899,
      "step": 6509
    },
    {
      "epoch": 1.788353248180195,
      "grad_norm": 0.3872225284576416,
      "learning_rate": 2.1245018551600937e-06,
      "loss": 0.7357,
      "step": 6510
    },
    {
      "epoch": 1.788627935723115,
      "grad_norm": 0.6317558884620667,
      "learning_rate": 2.121753469836471e-06,
      "loss": 0.6534,
      "step": 6511
    },
    {
      "epoch": 1.788902623266035,
      "grad_norm": 0.41221287846565247,
      "learning_rate": 2.1190050845128486e-06,
      "loss": 0.5972,
      "step": 6512
    },
    {
      "epoch": 1.7891773108089548,
      "grad_norm": 0.43843314051628113,
      "learning_rate": 2.1162566991892264e-06,
      "loss": 0.6155,
      "step": 6513
    },
    {
      "epoch": 1.7894519983518746,
      "grad_norm": 0.48292917013168335,
      "learning_rate": 2.113508313865604e-06,
      "loss": 0.8783,
      "step": 6514
    },
    {
      "epoch": 1.7897266858947947,
      "grad_norm": 0.4433574378490448,
      "learning_rate": 2.1107599285419817e-06,
      "loss": 0.7336,
      "step": 6515
    },
    {
      "epoch": 1.7900013734377147,
      "grad_norm": 0.386355459690094,
      "learning_rate": 2.108011543218359e-06,
      "loss": 0.5879,
      "step": 6516
    },
    {
      "epoch": 1.7902760609806345,
      "grad_norm": 0.5551638603210449,
      "learning_rate": 2.105263157894737e-06,
      "loss": 0.8834,
      "step": 6517
    },
    {
      "epoch": 1.7905507485235543,
      "grad_norm": 0.4390048682689667,
      "learning_rate": 2.1025147725711145e-06,
      "loss": 0.9745,
      "step": 6518
    },
    {
      "epoch": 1.7908254360664744,
      "grad_norm": 0.4480077624320984,
      "learning_rate": 2.099766387247492e-06,
      "loss": 0.8189,
      "step": 6519
    },
    {
      "epoch": 1.7911001236093944,
      "grad_norm": 0.46362990140914917,
      "learning_rate": 2.09701800192387e-06,
      "loss": 0.6262,
      "step": 6520
    },
    {
      "epoch": 1.7913748111523142,
      "grad_norm": 0.4333224296569824,
      "learning_rate": 2.0942696166002473e-06,
      "loss": 0.7583,
      "step": 6521
    },
    {
      "epoch": 1.791649498695234,
      "grad_norm": 0.472622275352478,
      "learning_rate": 2.091521231276625e-06,
      "loss": 0.7095,
      "step": 6522
    },
    {
      "epoch": 1.791924186238154,
      "grad_norm": 0.5555925965309143,
      "learning_rate": 2.0887728459530026e-06,
      "loss": 0.6919,
      "step": 6523
    },
    {
      "epoch": 1.7921988737810741,
      "grad_norm": 0.5468027591705322,
      "learning_rate": 2.0860244606293804e-06,
      "loss": 0.697,
      "step": 6524
    },
    {
      "epoch": 1.792473561323994,
      "grad_norm": 0.5030918121337891,
      "learning_rate": 2.083276075305758e-06,
      "loss": 0.5933,
      "step": 6525
    },
    {
      "epoch": 1.7927482488669138,
      "grad_norm": 0.4898500144481659,
      "learning_rate": 2.0805276899821358e-06,
      "loss": 0.8256,
      "step": 6526
    },
    {
      "epoch": 1.7930229364098338,
      "grad_norm": 0.49314549565315247,
      "learning_rate": 2.077779304658513e-06,
      "loss": 0.7755,
      "step": 6527
    },
    {
      "epoch": 1.7932976239527538,
      "grad_norm": 0.49693015217781067,
      "learning_rate": 2.0750309193348906e-06,
      "loss": 0.742,
      "step": 6528
    },
    {
      "epoch": 1.7935723114956736,
      "grad_norm": 0.47889021039009094,
      "learning_rate": 2.0722825340112685e-06,
      "loss": 0.9067,
      "step": 6529
    },
    {
      "epoch": 1.7938469990385935,
      "grad_norm": 0.5615435242652893,
      "learning_rate": 2.069534148687646e-06,
      "loss": 0.7316,
      "step": 6530
    },
    {
      "epoch": 1.7941216865815135,
      "grad_norm": 0.4606127142906189,
      "learning_rate": 2.066785763364024e-06,
      "loss": 0.7584,
      "step": 6531
    },
    {
      "epoch": 1.7943963741244335,
      "grad_norm": 0.4818347096443176,
      "learning_rate": 2.0640373780404013e-06,
      "loss": 0.9828,
      "step": 6532
    },
    {
      "epoch": 1.7946710616673534,
      "grad_norm": 0.43605902791023254,
      "learning_rate": 2.061288992716779e-06,
      "loss": 0.9801,
      "step": 6533
    },
    {
      "epoch": 1.7949457492102732,
      "grad_norm": 0.5881408452987671,
      "learning_rate": 2.0585406073931566e-06,
      "loss": 0.79,
      "step": 6534
    },
    {
      "epoch": 1.7952204367531932,
      "grad_norm": 0.6037408113479614,
      "learning_rate": 2.055792222069534e-06,
      "loss": 0.9697,
      "step": 6535
    },
    {
      "epoch": 1.7954951242961132,
      "grad_norm": 0.38265901803970337,
      "learning_rate": 2.053043836745912e-06,
      "loss": 0.6774,
      "step": 6536
    },
    {
      "epoch": 1.795769811839033,
      "grad_norm": 0.4221583902835846,
      "learning_rate": 2.0502954514222893e-06,
      "loss": 1.0579,
      "step": 6537
    },
    {
      "epoch": 1.7960444993819529,
      "grad_norm": 0.46329265832901,
      "learning_rate": 2.047547066098667e-06,
      "loss": 0.6382,
      "step": 6538
    },
    {
      "epoch": 1.796319186924873,
      "grad_norm": 0.5706669092178345,
      "learning_rate": 2.0447986807750446e-06,
      "loss": 0.828,
      "step": 6539
    },
    {
      "epoch": 1.796593874467793,
      "grad_norm": 0.4502868950366974,
      "learning_rate": 2.0420502954514225e-06,
      "loss": 0.8004,
      "step": 6540
    },
    {
      "epoch": 1.7968685620107128,
      "grad_norm": 0.47719845175743103,
      "learning_rate": 2.0393019101278e-06,
      "loss": 0.7229,
      "step": 6541
    },
    {
      "epoch": 1.7971432495536326,
      "grad_norm": 0.50881427526474,
      "learning_rate": 2.036553524804178e-06,
      "loss": 0.7616,
      "step": 6542
    },
    {
      "epoch": 1.7974179370965526,
      "grad_norm": 0.3909013867378235,
      "learning_rate": 2.0338051394805553e-06,
      "loss": 0.7102,
      "step": 6543
    },
    {
      "epoch": 1.7976926246394727,
      "grad_norm": 0.44534143805503845,
      "learning_rate": 2.0310567541569327e-06,
      "loss": 0.878,
      "step": 6544
    },
    {
      "epoch": 1.7979673121823925,
      "grad_norm": 0.34172821044921875,
      "learning_rate": 2.0283083688333106e-06,
      "loss": 0.9275,
      "step": 6545
    },
    {
      "epoch": 1.7982419997253125,
      "grad_norm": 0.4815152883529663,
      "learning_rate": 2.025559983509688e-06,
      "loss": 0.7742,
      "step": 6546
    },
    {
      "epoch": 1.7985166872682323,
      "grad_norm": 0.4871101379394531,
      "learning_rate": 2.022811598186066e-06,
      "loss": 0.5328,
      "step": 6547
    },
    {
      "epoch": 1.7987913748111524,
      "grad_norm": 0.37864622473716736,
      "learning_rate": 2.0200632128624433e-06,
      "loss": 0.8394,
      "step": 6548
    },
    {
      "epoch": 1.7990660623540724,
      "grad_norm": 0.48064735531806946,
      "learning_rate": 2.017314827538821e-06,
      "loss": 0.7689,
      "step": 6549
    },
    {
      "epoch": 1.7993407498969922,
      "grad_norm": 0.4752042591571808,
      "learning_rate": 2.0145664422151987e-06,
      "loss": 0.9447,
      "step": 6550
    },
    {
      "epoch": 1.799615437439912,
      "grad_norm": 0.6308935284614563,
      "learning_rate": 2.011818056891576e-06,
      "loss": 0.8511,
      "step": 6551
    },
    {
      "epoch": 1.799890124982832,
      "grad_norm": 0.6228872537612915,
      "learning_rate": 2.009069671567954e-06,
      "loss": 0.8171,
      "step": 6552
    },
    {
      "epoch": 1.8001648125257521,
      "grad_norm": 0.4437654912471771,
      "learning_rate": 2.0063212862443314e-06,
      "loss": 0.9953,
      "step": 6553
    },
    {
      "epoch": 1.800439500068672,
      "grad_norm": 0.5464659929275513,
      "learning_rate": 2.0035729009207093e-06,
      "loss": 0.7345,
      "step": 6554
    },
    {
      "epoch": 1.8007141876115917,
      "grad_norm": 0.39849451184272766,
      "learning_rate": 2.0008245155970867e-06,
      "loss": 0.5812,
      "step": 6555
    },
    {
      "epoch": 1.8009888751545118,
      "grad_norm": 0.4954953193664551,
      "learning_rate": 1.9980761302734646e-06,
      "loss": 0.6965,
      "step": 6556
    },
    {
      "epoch": 1.8012635626974318,
      "grad_norm": 0.4800529181957245,
      "learning_rate": 1.995327744949842e-06,
      "loss": 0.5322,
      "step": 6557
    },
    {
      "epoch": 1.8015382502403516,
      "grad_norm": 0.6075348854064941,
      "learning_rate": 1.9925793596262195e-06,
      "loss": 0.953,
      "step": 6558
    },
    {
      "epoch": 1.8018129377832715,
      "grad_norm": 0.3653543293476105,
      "learning_rate": 1.9898309743025973e-06,
      "loss": 0.6948,
      "step": 6559
    },
    {
      "epoch": 1.8020876253261915,
      "grad_norm": 0.6997180581092834,
      "learning_rate": 1.9870825889789748e-06,
      "loss": 0.6302,
      "step": 6560
    },
    {
      "epoch": 1.8023623128691115,
      "grad_norm": 0.5241941213607788,
      "learning_rate": 1.9843342036553527e-06,
      "loss": 0.7536,
      "step": 6561
    },
    {
      "epoch": 1.8026370004120313,
      "grad_norm": 0.6371385455131531,
      "learning_rate": 1.98158581833173e-06,
      "loss": 0.6757,
      "step": 6562
    },
    {
      "epoch": 1.8029116879549512,
      "grad_norm": 0.4274626076221466,
      "learning_rate": 1.978837433008108e-06,
      "loss": 0.6526,
      "step": 6563
    },
    {
      "epoch": 1.8031863754978712,
      "grad_norm": 0.36158105731010437,
      "learning_rate": 1.9760890476844854e-06,
      "loss": 1.009,
      "step": 6564
    },
    {
      "epoch": 1.8034610630407912,
      "grad_norm": 0.532478392124176,
      "learning_rate": 1.9733406623608633e-06,
      "loss": 0.7146,
      "step": 6565
    },
    {
      "epoch": 1.803735750583711,
      "grad_norm": 0.5056813955307007,
      "learning_rate": 1.9705922770372407e-06,
      "loss": 0.8175,
      "step": 6566
    },
    {
      "epoch": 1.8040104381266309,
      "grad_norm": 0.49778661131858826,
      "learning_rate": 1.967843891713618e-06,
      "loss": 0.6871,
      "step": 6567
    },
    {
      "epoch": 1.804285125669551,
      "grad_norm": 0.5376777052879333,
      "learning_rate": 1.965095506389996e-06,
      "loss": 0.8174,
      "step": 6568
    },
    {
      "epoch": 1.804559813212471,
      "grad_norm": 0.47994375228881836,
      "learning_rate": 1.9623471210663735e-06,
      "loss": 0.8544,
      "step": 6569
    },
    {
      "epoch": 1.8048345007553908,
      "grad_norm": 0.4159001111984253,
      "learning_rate": 1.9595987357427513e-06,
      "loss": 0.7082,
      "step": 6570
    },
    {
      "epoch": 1.8051091882983106,
      "grad_norm": 0.6819661855697632,
      "learning_rate": 1.956850350419129e-06,
      "loss": 0.5987,
      "step": 6571
    },
    {
      "epoch": 1.8053838758412306,
      "grad_norm": 0.49790239334106445,
      "learning_rate": 1.9541019650955067e-06,
      "loss": 0.8393,
      "step": 6572
    },
    {
      "epoch": 1.8056585633841506,
      "grad_norm": 0.45083290338516235,
      "learning_rate": 1.951353579771884e-06,
      "loss": 0.6993,
      "step": 6573
    },
    {
      "epoch": 1.8059332509270705,
      "grad_norm": 0.4513593316078186,
      "learning_rate": 1.9486051944482615e-06,
      "loss": 0.7487,
      "step": 6574
    },
    {
      "epoch": 1.8062079384699903,
      "grad_norm": 0.47037631273269653,
      "learning_rate": 1.9458568091246394e-06,
      "loss": 0.8139,
      "step": 6575
    },
    {
      "epoch": 1.8064826260129103,
      "grad_norm": 0.5317729711532593,
      "learning_rate": 1.943108423801017e-06,
      "loss": 0.8105,
      "step": 6576
    },
    {
      "epoch": 1.8067573135558304,
      "grad_norm": 0.4874773621559143,
      "learning_rate": 1.9403600384773947e-06,
      "loss": 0.7783,
      "step": 6577
    },
    {
      "epoch": 1.8070320010987502,
      "grad_norm": 0.4028353691101074,
      "learning_rate": 1.937611653153772e-06,
      "loss": 1.137,
      "step": 6578
    },
    {
      "epoch": 1.80730668864167,
      "grad_norm": 0.3565140664577484,
      "learning_rate": 1.93486326783015e-06,
      "loss": 0.7212,
      "step": 6579
    },
    {
      "epoch": 1.80758137618459,
      "grad_norm": 0.5772522687911987,
      "learning_rate": 1.9321148825065275e-06,
      "loss": 0.6823,
      "step": 6580
    },
    {
      "epoch": 1.80785606372751,
      "grad_norm": 0.46603575348854065,
      "learning_rate": 1.929366497182905e-06,
      "loss": 0.827,
      "step": 6581
    },
    {
      "epoch": 1.8081307512704299,
      "grad_norm": 0.4251099228858948,
      "learning_rate": 1.926618111859283e-06,
      "loss": 0.618,
      "step": 6582
    },
    {
      "epoch": 1.8084054388133497,
      "grad_norm": 0.363720178604126,
      "learning_rate": 1.9238697265356602e-06,
      "loss": 0.7898,
      "step": 6583
    },
    {
      "epoch": 1.8086801263562697,
      "grad_norm": 0.3426339626312256,
      "learning_rate": 1.921121341212038e-06,
      "loss": 0.8084,
      "step": 6584
    },
    {
      "epoch": 1.8089548138991898,
      "grad_norm": 0.5877065062522888,
      "learning_rate": 1.9183729558884156e-06,
      "loss": 0.849,
      "step": 6585
    },
    {
      "epoch": 1.8092295014421096,
      "grad_norm": 0.5365882515907288,
      "learning_rate": 1.9156245705647934e-06,
      "loss": 0.6834,
      "step": 6586
    },
    {
      "epoch": 1.8095041889850294,
      "grad_norm": 0.440982848405838,
      "learning_rate": 1.912876185241171e-06,
      "loss": 0.8999,
      "step": 6587
    },
    {
      "epoch": 1.8097788765279494,
      "grad_norm": 0.356940895318985,
      "learning_rate": 1.9101277999175487e-06,
      "loss": 0.6335,
      "step": 6588
    },
    {
      "epoch": 1.8100535640708695,
      "grad_norm": 0.4729122817516327,
      "learning_rate": 1.907379414593926e-06,
      "loss": 0.5241,
      "step": 6589
    },
    {
      "epoch": 1.8103282516137893,
      "grad_norm": 0.6187028884887695,
      "learning_rate": 1.9046310292703038e-06,
      "loss": 0.7386,
      "step": 6590
    },
    {
      "epoch": 1.8106029391567091,
      "grad_norm": 0.5581074953079224,
      "learning_rate": 1.9018826439466815e-06,
      "loss": 0.6445,
      "step": 6591
    },
    {
      "epoch": 1.8108776266996292,
      "grad_norm": 0.6467830538749695,
      "learning_rate": 1.8991342586230591e-06,
      "loss": 0.7324,
      "step": 6592
    },
    {
      "epoch": 1.8111523142425492,
      "grad_norm": 0.3990406095981598,
      "learning_rate": 1.8963858732994366e-06,
      "loss": 0.947,
      "step": 6593
    },
    {
      "epoch": 1.811427001785469,
      "grad_norm": 0.45702502131462097,
      "learning_rate": 1.8936374879758142e-06,
      "loss": 0.7783,
      "step": 6594
    },
    {
      "epoch": 1.8117016893283888,
      "grad_norm": 0.8024113774299622,
      "learning_rate": 1.890889102652192e-06,
      "loss": 0.8339,
      "step": 6595
    },
    {
      "epoch": 1.8119763768713089,
      "grad_norm": 0.48568665981292725,
      "learning_rate": 1.8881407173285696e-06,
      "loss": 0.5779,
      "step": 6596
    },
    {
      "epoch": 1.812251064414229,
      "grad_norm": 0.466304749250412,
      "learning_rate": 1.8853923320049472e-06,
      "loss": 0.6937,
      "step": 6597
    },
    {
      "epoch": 1.8125257519571487,
      "grad_norm": 0.5000752806663513,
      "learning_rate": 1.8826439466813249e-06,
      "loss": 0.9341,
      "step": 6598
    },
    {
      "epoch": 1.8128004395000685,
      "grad_norm": 0.620032787322998,
      "learning_rate": 1.8798955613577025e-06,
      "loss": 0.6014,
      "step": 6599
    },
    {
      "epoch": 1.8130751270429886,
      "grad_norm": 0.5516341924667358,
      "learning_rate": 1.8771471760340802e-06,
      "loss": 0.8093,
      "step": 6600
    },
    {
      "epoch": 1.8133498145859086,
      "grad_norm": 0.40929973125457764,
      "learning_rate": 1.8743987907104576e-06,
      "loss": 0.541,
      "step": 6601
    },
    {
      "epoch": 1.8136245021288284,
      "grad_norm": 0.4906820058822632,
      "learning_rate": 1.8716504053868353e-06,
      "loss": 0.7456,
      "step": 6602
    },
    {
      "epoch": 1.8138991896717482,
      "grad_norm": 0.5794925093650818,
      "learning_rate": 1.868902020063213e-06,
      "loss": 0.75,
      "step": 6603
    },
    {
      "epoch": 1.8141738772146683,
      "grad_norm": 0.4677768349647522,
      "learning_rate": 1.8661536347395906e-06,
      "loss": 0.8469,
      "step": 6604
    },
    {
      "epoch": 1.8144485647575883,
      "grad_norm": 0.3206622898578644,
      "learning_rate": 1.8634052494159683e-06,
      "loss": 0.8136,
      "step": 6605
    },
    {
      "epoch": 1.8147232523005081,
      "grad_norm": 0.5284033417701721,
      "learning_rate": 1.860656864092346e-06,
      "loss": 0.8026,
      "step": 6606
    },
    {
      "epoch": 1.814997939843428,
      "grad_norm": 0.4371177852153778,
      "learning_rate": 1.8579084787687236e-06,
      "loss": 0.7206,
      "step": 6607
    },
    {
      "epoch": 1.815272627386348,
      "grad_norm": 0.5050565004348755,
      "learning_rate": 1.8551600934451012e-06,
      "loss": 0.6624,
      "step": 6608
    },
    {
      "epoch": 1.815547314929268,
      "grad_norm": 0.4496941864490509,
      "learning_rate": 1.8524117081214787e-06,
      "loss": 0.8495,
      "step": 6609
    },
    {
      "epoch": 1.8158220024721878,
      "grad_norm": 0.44156312942504883,
      "learning_rate": 1.8496633227978563e-06,
      "loss": 1.0464,
      "step": 6610
    },
    {
      "epoch": 1.8160966900151077,
      "grad_norm": 0.41942158341407776,
      "learning_rate": 1.846914937474234e-06,
      "loss": 0.807,
      "step": 6611
    },
    {
      "epoch": 1.8163713775580277,
      "grad_norm": 0.4145399034023285,
      "learning_rate": 1.8441665521506116e-06,
      "loss": 0.7184,
      "step": 6612
    },
    {
      "epoch": 1.8166460651009477,
      "grad_norm": 0.4551543593406677,
      "learning_rate": 1.8414181668269893e-06,
      "loss": 0.7876,
      "step": 6613
    },
    {
      "epoch": 1.8169207526438678,
      "grad_norm": 0.5706730484962463,
      "learning_rate": 1.838669781503367e-06,
      "loss": 0.7735,
      "step": 6614
    },
    {
      "epoch": 1.8171954401867876,
      "grad_norm": 0.3083389401435852,
      "learning_rate": 1.8359213961797446e-06,
      "loss": 0.4699,
      "step": 6615
    },
    {
      "epoch": 1.8174701277297074,
      "grad_norm": 0.5044680833816528,
      "learning_rate": 1.833173010856122e-06,
      "loss": 0.6971,
      "step": 6616
    },
    {
      "epoch": 1.8177448152726274,
      "grad_norm": 0.40149402618408203,
      "learning_rate": 1.8304246255324997e-06,
      "loss": 0.9879,
      "step": 6617
    },
    {
      "epoch": 1.8180195028155475,
      "grad_norm": 0.39477697014808655,
      "learning_rate": 1.8276762402088774e-06,
      "loss": 0.7235,
      "step": 6618
    },
    {
      "epoch": 1.8182941903584673,
      "grad_norm": 0.5947336554527283,
      "learning_rate": 1.824927854885255e-06,
      "loss": 0.767,
      "step": 6619
    },
    {
      "epoch": 1.818568877901387,
      "grad_norm": 0.46730101108551025,
      "learning_rate": 1.8221794695616327e-06,
      "loss": 0.6986,
      "step": 6620
    },
    {
      "epoch": 1.8188435654443071,
      "grad_norm": 0.3436514437198639,
      "learning_rate": 1.8194310842380103e-06,
      "loss": 0.8317,
      "step": 6621
    },
    {
      "epoch": 1.8191182529872272,
      "grad_norm": 0.36981621384620667,
      "learning_rate": 1.816682698914388e-06,
      "loss": 0.6702,
      "step": 6622
    },
    {
      "epoch": 1.819392940530147,
      "grad_norm": 0.5115891695022583,
      "learning_rate": 1.8139343135907656e-06,
      "loss": 0.7566,
      "step": 6623
    },
    {
      "epoch": 1.8196676280730668,
      "grad_norm": 0.48971667885780334,
      "learning_rate": 1.811185928267143e-06,
      "loss": 0.7417,
      "step": 6624
    },
    {
      "epoch": 1.8199423156159868,
      "grad_norm": 0.6194664239883423,
      "learning_rate": 1.8084375429435207e-06,
      "loss": 0.8554,
      "step": 6625
    },
    {
      "epoch": 1.8202170031589069,
      "grad_norm": 0.5087242126464844,
      "learning_rate": 1.8056891576198984e-06,
      "loss": 0.607,
      "step": 6626
    },
    {
      "epoch": 1.8204916907018267,
      "grad_norm": 0.5803731679916382,
      "learning_rate": 1.802940772296276e-06,
      "loss": 0.7632,
      "step": 6627
    },
    {
      "epoch": 1.8207663782447465,
      "grad_norm": 0.44101274013519287,
      "learning_rate": 1.8001923869726537e-06,
      "loss": 0.7562,
      "step": 6628
    },
    {
      "epoch": 1.8210410657876666,
      "grad_norm": 0.510285496711731,
      "learning_rate": 1.7974440016490314e-06,
      "loss": 0.6118,
      "step": 6629
    },
    {
      "epoch": 1.8213157533305866,
      "grad_norm": 0.48419156670570374,
      "learning_rate": 1.794695616325409e-06,
      "loss": 0.7848,
      "step": 6630
    },
    {
      "epoch": 1.8215904408735064,
      "grad_norm": 0.3772200345993042,
      "learning_rate": 1.7919472310017867e-06,
      "loss": 0.6247,
      "step": 6631
    },
    {
      "epoch": 1.8218651284164262,
      "grad_norm": 0.5203094482421875,
      "learning_rate": 1.7891988456781641e-06,
      "loss": 0.7995,
      "step": 6632
    },
    {
      "epoch": 1.8221398159593463,
      "grad_norm": 0.40662211179733276,
      "learning_rate": 1.7864504603545418e-06,
      "loss": 0.5899,
      "step": 6633
    },
    {
      "epoch": 1.8224145035022663,
      "grad_norm": 0.3930540978908539,
      "learning_rate": 1.7837020750309194e-06,
      "loss": 0.673,
      "step": 6634
    },
    {
      "epoch": 1.8226891910451861,
      "grad_norm": 0.4337243139743805,
      "learning_rate": 1.780953689707297e-06,
      "loss": 0.935,
      "step": 6635
    },
    {
      "epoch": 1.822963878588106,
      "grad_norm": 0.5648255348205566,
      "learning_rate": 1.7782053043836747e-06,
      "loss": 0.8482,
      "step": 6636
    },
    {
      "epoch": 1.823238566131026,
      "grad_norm": 0.569860577583313,
      "learning_rate": 1.7754569190600524e-06,
      "loss": 0.6449,
      "step": 6637
    },
    {
      "epoch": 1.823513253673946,
      "grad_norm": 0.4837545156478882,
      "learning_rate": 1.77270853373643e-06,
      "loss": 0.4842,
      "step": 6638
    },
    {
      "epoch": 1.8237879412168658,
      "grad_norm": 0.4547354578971863,
      "learning_rate": 1.7699601484128077e-06,
      "loss": 0.7024,
      "step": 6639
    },
    {
      "epoch": 1.8240626287597856,
      "grad_norm": 0.447981059551239,
      "learning_rate": 1.7672117630891852e-06,
      "loss": 0.8932,
      "step": 6640
    },
    {
      "epoch": 1.8243373163027057,
      "grad_norm": 0.64500492811203,
      "learning_rate": 1.7644633777655628e-06,
      "loss": 0.649,
      "step": 6641
    },
    {
      "epoch": 1.8246120038456257,
      "grad_norm": 0.47503969073295593,
      "learning_rate": 1.7617149924419405e-06,
      "loss": 0.4864,
      "step": 6642
    },
    {
      "epoch": 1.8248866913885455,
      "grad_norm": 0.3712707757949829,
      "learning_rate": 1.7589666071183181e-06,
      "loss": 0.9899,
      "step": 6643
    },
    {
      "epoch": 1.8251613789314654,
      "grad_norm": 0.5218545198440552,
      "learning_rate": 1.7562182217946958e-06,
      "loss": 0.6116,
      "step": 6644
    },
    {
      "epoch": 1.8254360664743854,
      "grad_norm": 0.5760801434516907,
      "learning_rate": 1.7534698364710734e-06,
      "loss": 0.9974,
      "step": 6645
    },
    {
      "epoch": 1.8257107540173054,
      "grad_norm": 0.47238707542419434,
      "learning_rate": 1.750721451147451e-06,
      "loss": 0.8913,
      "step": 6646
    },
    {
      "epoch": 1.8259854415602252,
      "grad_norm": 0.5391637682914734,
      "learning_rate": 1.7479730658238287e-06,
      "loss": 0.7805,
      "step": 6647
    },
    {
      "epoch": 1.826260129103145,
      "grad_norm": 0.4549553692340851,
      "learning_rate": 1.7452246805002062e-06,
      "loss": 0.7729,
      "step": 6648
    },
    {
      "epoch": 1.826534816646065,
      "grad_norm": 0.4028136134147644,
      "learning_rate": 1.7424762951765838e-06,
      "loss": 0.679,
      "step": 6649
    },
    {
      "epoch": 1.8268095041889851,
      "grad_norm": 0.45557376742362976,
      "learning_rate": 1.7397279098529615e-06,
      "loss": 0.6915,
      "step": 6650
    },
    {
      "epoch": 1.827084191731905,
      "grad_norm": 0.5503435134887695,
      "learning_rate": 1.7369795245293392e-06,
      "loss": 0.7897,
      "step": 6651
    },
    {
      "epoch": 1.8273588792748248,
      "grad_norm": 0.3709627389907837,
      "learning_rate": 1.7342311392057168e-06,
      "loss": 0.5323,
      "step": 6652
    },
    {
      "epoch": 1.8276335668177448,
      "grad_norm": 0.44921696186065674,
      "learning_rate": 1.7314827538820945e-06,
      "loss": 0.9222,
      "step": 6653
    },
    {
      "epoch": 1.8279082543606648,
      "grad_norm": 0.49523627758026123,
      "learning_rate": 1.7287343685584721e-06,
      "loss": 0.9101,
      "step": 6654
    },
    {
      "epoch": 1.8281829419035847,
      "grad_norm": 0.40017229318618774,
      "learning_rate": 1.7259859832348496e-06,
      "loss": 0.8178,
      "step": 6655
    },
    {
      "epoch": 1.8284576294465045,
      "grad_norm": 0.3866109848022461,
      "learning_rate": 1.7232375979112272e-06,
      "loss": 0.8887,
      "step": 6656
    },
    {
      "epoch": 1.8287323169894245,
      "grad_norm": 0.5328197479248047,
      "learning_rate": 1.7204892125876049e-06,
      "loss": 0.8292,
      "step": 6657
    },
    {
      "epoch": 1.8290070045323445,
      "grad_norm": 0.5334430932998657,
      "learning_rate": 1.7177408272639825e-06,
      "loss": 0.8897,
      "step": 6658
    },
    {
      "epoch": 1.8292816920752644,
      "grad_norm": 0.4147275686264038,
      "learning_rate": 1.7149924419403602e-06,
      "loss": 0.6955,
      "step": 6659
    },
    {
      "epoch": 1.8295563796181842,
      "grad_norm": 0.5051692724227905,
      "learning_rate": 1.7122440566167379e-06,
      "loss": 0.7975,
      "step": 6660
    },
    {
      "epoch": 1.8298310671611042,
      "grad_norm": 0.6112477779388428,
      "learning_rate": 1.7094956712931155e-06,
      "loss": 0.8196,
      "step": 6661
    },
    {
      "epoch": 1.8301057547040243,
      "grad_norm": 0.525896430015564,
      "learning_rate": 1.7067472859694932e-06,
      "loss": 0.8297,
      "step": 6662
    },
    {
      "epoch": 1.830380442246944,
      "grad_norm": 0.44240134954452515,
      "learning_rate": 1.7039989006458706e-06,
      "loss": 0.7347,
      "step": 6663
    },
    {
      "epoch": 1.8306551297898639,
      "grad_norm": 0.46368035674095154,
      "learning_rate": 1.7012505153222483e-06,
      "loss": 0.4837,
      "step": 6664
    },
    {
      "epoch": 1.830929817332784,
      "grad_norm": 0.4852820038795471,
      "learning_rate": 1.698502129998626e-06,
      "loss": 0.7659,
      "step": 6665
    },
    {
      "epoch": 1.831204504875704,
      "grad_norm": 0.4473765790462494,
      "learning_rate": 1.6957537446750036e-06,
      "loss": 0.6611,
      "step": 6666
    },
    {
      "epoch": 1.8314791924186238,
      "grad_norm": 0.4322218894958496,
      "learning_rate": 1.6930053593513812e-06,
      "loss": 0.8452,
      "step": 6667
    },
    {
      "epoch": 1.8317538799615436,
      "grad_norm": 0.5094895958900452,
      "learning_rate": 1.6902569740277589e-06,
      "loss": 1.0885,
      "step": 6668
    },
    {
      "epoch": 1.8320285675044636,
      "grad_norm": 0.5500903129577637,
      "learning_rate": 1.6875085887041365e-06,
      "loss": 0.8606,
      "step": 6669
    },
    {
      "epoch": 1.8323032550473837,
      "grad_norm": 0.44006553292274475,
      "learning_rate": 1.6847602033805142e-06,
      "loss": 0.9675,
      "step": 6670
    },
    {
      "epoch": 1.8325779425903035,
      "grad_norm": 0.49835509061813354,
      "learning_rate": 1.6820118180568916e-06,
      "loss": 0.9129,
      "step": 6671
    },
    {
      "epoch": 1.8328526301332233,
      "grad_norm": 0.4368208050727844,
      "learning_rate": 1.6792634327332693e-06,
      "loss": 0.5859,
      "step": 6672
    },
    {
      "epoch": 1.8331273176761433,
      "grad_norm": 0.37689247727394104,
      "learning_rate": 1.676515047409647e-06,
      "loss": 0.772,
      "step": 6673
    },
    {
      "epoch": 1.8334020052190634,
      "grad_norm": 0.5253175497055054,
      "learning_rate": 1.6737666620860246e-06,
      "loss": 0.6283,
      "step": 6674
    },
    {
      "epoch": 1.8336766927619832,
      "grad_norm": 0.47437697649002075,
      "learning_rate": 1.6710182767624023e-06,
      "loss": 0.7929,
      "step": 6675
    },
    {
      "epoch": 1.833951380304903,
      "grad_norm": 0.4154324531555176,
      "learning_rate": 1.66826989143878e-06,
      "loss": 0.7463,
      "step": 6676
    },
    {
      "epoch": 1.834226067847823,
      "grad_norm": 0.5320785641670227,
      "learning_rate": 1.6655215061151576e-06,
      "loss": 0.6235,
      "step": 6677
    },
    {
      "epoch": 1.834500755390743,
      "grad_norm": 0.46656709909439087,
      "learning_rate": 1.6627731207915352e-06,
      "loss": 0.9379,
      "step": 6678
    },
    {
      "epoch": 1.834775442933663,
      "grad_norm": 0.4065811038017273,
      "learning_rate": 1.6600247354679127e-06,
      "loss": 0.5572,
      "step": 6679
    },
    {
      "epoch": 1.835050130476583,
      "grad_norm": 0.5120657086372375,
      "learning_rate": 1.6572763501442903e-06,
      "loss": 1.0705,
      "step": 6680
    },
    {
      "epoch": 1.8353248180195028,
      "grad_norm": 0.49364203214645386,
      "learning_rate": 1.654527964820668e-06,
      "loss": 0.7617,
      "step": 6681
    },
    {
      "epoch": 1.8355995055624228,
      "grad_norm": 0.4172072112560272,
      "learning_rate": 1.6517795794970456e-06,
      "loss": 1.0066,
      "step": 6682
    },
    {
      "epoch": 1.8358741931053428,
      "grad_norm": 0.43306827545166016,
      "learning_rate": 1.6490311941734233e-06,
      "loss": 0.8435,
      "step": 6683
    },
    {
      "epoch": 1.8361488806482626,
      "grad_norm": 0.418770432472229,
      "learning_rate": 1.646282808849801e-06,
      "loss": 0.8622,
      "step": 6684
    },
    {
      "epoch": 1.8364235681911825,
      "grad_norm": 0.4016958475112915,
      "learning_rate": 1.6435344235261786e-06,
      "loss": 0.8348,
      "step": 6685
    },
    {
      "epoch": 1.8366982557341025,
      "grad_norm": 0.5301036834716797,
      "learning_rate": 1.640786038202556e-06,
      "loss": 0.5989,
      "step": 6686
    },
    {
      "epoch": 1.8369729432770225,
      "grad_norm": 0.4507203996181488,
      "learning_rate": 1.6380376528789337e-06,
      "loss": 0.6989,
      "step": 6687
    },
    {
      "epoch": 1.8372476308199424,
      "grad_norm": 0.6071709394454956,
      "learning_rate": 1.6352892675553114e-06,
      "loss": 1.0054,
      "step": 6688
    },
    {
      "epoch": 1.8375223183628622,
      "grad_norm": 0.4849522113800049,
      "learning_rate": 1.632540882231689e-06,
      "loss": 0.4471,
      "step": 6689
    },
    {
      "epoch": 1.8377970059057822,
      "grad_norm": 0.3804212212562561,
      "learning_rate": 1.6297924969080667e-06,
      "loss": 0.6138,
      "step": 6690
    },
    {
      "epoch": 1.8380716934487022,
      "grad_norm": 0.4218100607395172,
      "learning_rate": 1.6270441115844443e-06,
      "loss": 0.8091,
      "step": 6691
    },
    {
      "epoch": 1.838346380991622,
      "grad_norm": 0.5008241534233093,
      "learning_rate": 1.624295726260822e-06,
      "loss": 0.6862,
      "step": 6692
    },
    {
      "epoch": 1.8386210685345419,
      "grad_norm": 0.5048027634620667,
      "learning_rate": 1.6215473409371997e-06,
      "loss": 0.55,
      "step": 6693
    },
    {
      "epoch": 1.838895756077462,
      "grad_norm": 0.4741087555885315,
      "learning_rate": 1.618798955613577e-06,
      "loss": 0.6376,
      "step": 6694
    },
    {
      "epoch": 1.839170443620382,
      "grad_norm": 0.45852911472320557,
      "learning_rate": 1.6160505702899548e-06,
      "loss": 0.8994,
      "step": 6695
    },
    {
      "epoch": 1.8394451311633018,
      "grad_norm": 0.3720400035381317,
      "learning_rate": 1.6133021849663324e-06,
      "loss": 0.8433,
      "step": 6696
    },
    {
      "epoch": 1.8397198187062216,
      "grad_norm": 0.5432929992675781,
      "learning_rate": 1.61055379964271e-06,
      "loss": 0.7225,
      "step": 6697
    },
    {
      "epoch": 1.8399945062491416,
      "grad_norm": 0.34624040126800537,
      "learning_rate": 1.6078054143190877e-06,
      "loss": 0.8313,
      "step": 6698
    },
    {
      "epoch": 1.8402691937920617,
      "grad_norm": 0.4056749641895294,
      "learning_rate": 1.6050570289954654e-06,
      "loss": 1.0056,
      "step": 6699
    },
    {
      "epoch": 1.8405438813349815,
      "grad_norm": 0.46965768933296204,
      "learning_rate": 1.602308643671843e-06,
      "loss": 0.7455,
      "step": 6700
    },
    {
      "epoch": 1.8408185688779013,
      "grad_norm": 0.37877461314201355,
      "learning_rate": 1.5995602583482207e-06,
      "loss": 0.7446,
      "step": 6701
    },
    {
      "epoch": 1.8410932564208213,
      "grad_norm": 0.4511500895023346,
      "learning_rate": 1.5968118730245981e-06,
      "loss": 0.9095,
      "step": 6702
    },
    {
      "epoch": 1.8413679439637414,
      "grad_norm": 0.5810932517051697,
      "learning_rate": 1.5940634877009758e-06,
      "loss": 0.8341,
      "step": 6703
    },
    {
      "epoch": 1.8416426315066612,
      "grad_norm": 0.4850318431854248,
      "learning_rate": 1.5913151023773534e-06,
      "loss": 0.8271,
      "step": 6704
    },
    {
      "epoch": 1.841917319049581,
      "grad_norm": 0.5023086071014404,
      "learning_rate": 1.588566717053731e-06,
      "loss": 0.8892,
      "step": 6705
    },
    {
      "epoch": 1.842192006592501,
      "grad_norm": 0.504112184047699,
      "learning_rate": 1.5858183317301088e-06,
      "loss": 0.5011,
      "step": 6706
    },
    {
      "epoch": 1.842466694135421,
      "grad_norm": 0.480415016412735,
      "learning_rate": 1.5830699464064864e-06,
      "loss": 0.8353,
      "step": 6707
    },
    {
      "epoch": 1.842741381678341,
      "grad_norm": 0.37331125140190125,
      "learning_rate": 1.580321561082864e-06,
      "loss": 0.6151,
      "step": 6708
    },
    {
      "epoch": 1.8430160692212607,
      "grad_norm": 0.6354605555534363,
      "learning_rate": 1.5775731757592417e-06,
      "loss": 0.7186,
      "step": 6709
    },
    {
      "epoch": 1.8432907567641807,
      "grad_norm": 0.5144596695899963,
      "learning_rate": 1.5748247904356192e-06,
      "loss": 0.8671,
      "step": 6710
    },
    {
      "epoch": 1.8435654443071008,
      "grad_norm": 0.40765684843063354,
      "learning_rate": 1.5720764051119968e-06,
      "loss": 0.8821,
      "step": 6711
    },
    {
      "epoch": 1.8438401318500206,
      "grad_norm": 0.49268800020217896,
      "learning_rate": 1.5693280197883745e-06,
      "loss": 0.9802,
      "step": 6712
    },
    {
      "epoch": 1.8441148193929404,
      "grad_norm": 0.3998929262161255,
      "learning_rate": 1.5665796344647521e-06,
      "loss": 0.4386,
      "step": 6713
    },
    {
      "epoch": 1.8443895069358605,
      "grad_norm": 0.3509516716003418,
      "learning_rate": 1.5638312491411298e-06,
      "loss": 0.6847,
      "step": 6714
    },
    {
      "epoch": 1.8446641944787805,
      "grad_norm": 0.41875144839286804,
      "learning_rate": 1.5610828638175075e-06,
      "loss": 0.7239,
      "step": 6715
    },
    {
      "epoch": 1.8449388820217003,
      "grad_norm": 0.6118956804275513,
      "learning_rate": 1.5583344784938851e-06,
      "loss": 0.9464,
      "step": 6716
    },
    {
      "epoch": 1.8452135695646201,
      "grad_norm": 0.3467423617839813,
      "learning_rate": 1.5555860931702626e-06,
      "loss": 0.637,
      "step": 6717
    },
    {
      "epoch": 1.8454882571075402,
      "grad_norm": 0.42636343836784363,
      "learning_rate": 1.5528377078466402e-06,
      "loss": 0.7008,
      "step": 6718
    },
    {
      "epoch": 1.8457629446504602,
      "grad_norm": 0.41753485798835754,
      "learning_rate": 1.5500893225230179e-06,
      "loss": 0.5301,
      "step": 6719
    },
    {
      "epoch": 1.84603763219338,
      "grad_norm": 0.4915897846221924,
      "learning_rate": 1.5473409371993955e-06,
      "loss": 0.7161,
      "step": 6720
    },
    {
      "epoch": 1.8463123197362998,
      "grad_norm": 0.5512756109237671,
      "learning_rate": 1.5445925518757732e-06,
      "loss": 0.7649,
      "step": 6721
    },
    {
      "epoch": 1.8465870072792199,
      "grad_norm": 0.5637492537498474,
      "learning_rate": 1.5418441665521508e-06,
      "loss": 0.8231,
      "step": 6722
    },
    {
      "epoch": 1.84686169482214,
      "grad_norm": 0.5765686631202698,
      "learning_rate": 1.5390957812285285e-06,
      "loss": 0.8538,
      "step": 6723
    },
    {
      "epoch": 1.8471363823650597,
      "grad_norm": 0.4953571557998657,
      "learning_rate": 1.5363473959049061e-06,
      "loss": 0.602,
      "step": 6724
    },
    {
      "epoch": 1.8474110699079795,
      "grad_norm": 0.40673550963401794,
      "learning_rate": 1.5335990105812836e-06,
      "loss": 0.7488,
      "step": 6725
    },
    {
      "epoch": 1.8476857574508996,
      "grad_norm": 0.5102822780609131,
      "learning_rate": 1.5308506252576612e-06,
      "loss": 0.7659,
      "step": 6726
    },
    {
      "epoch": 1.8479604449938196,
      "grad_norm": 0.39010122418403625,
      "learning_rate": 1.528102239934039e-06,
      "loss": 0.6853,
      "step": 6727
    },
    {
      "epoch": 1.8482351325367394,
      "grad_norm": 0.5127353668212891,
      "learning_rate": 1.5253538546104166e-06,
      "loss": 0.8308,
      "step": 6728
    },
    {
      "epoch": 1.8485098200796592,
      "grad_norm": 0.37926366925239563,
      "learning_rate": 1.5226054692867942e-06,
      "loss": 0.7788,
      "step": 6729
    },
    {
      "epoch": 1.8487845076225793,
      "grad_norm": 0.48449715971946716,
      "learning_rate": 1.5198570839631719e-06,
      "loss": 0.6601,
      "step": 6730
    },
    {
      "epoch": 1.8490591951654993,
      "grad_norm": 0.6022487282752991,
      "learning_rate": 1.5171086986395495e-06,
      "loss": 0.767,
      "step": 6731
    },
    {
      "epoch": 1.8493338827084191,
      "grad_norm": 0.4303298890590668,
      "learning_rate": 1.5143603133159272e-06,
      "loss": 0.6951,
      "step": 6732
    },
    {
      "epoch": 1.849608570251339,
      "grad_norm": 0.47326892614364624,
      "learning_rate": 1.5116119279923046e-06,
      "loss": 0.7278,
      "step": 6733
    },
    {
      "epoch": 1.849883257794259,
      "grad_norm": 0.5149480104446411,
      "learning_rate": 1.5088635426686823e-06,
      "loss": 0.7369,
      "step": 6734
    },
    {
      "epoch": 1.850157945337179,
      "grad_norm": 0.4155031144618988,
      "learning_rate": 1.50611515734506e-06,
      "loss": 0.793,
      "step": 6735
    },
    {
      "epoch": 1.8504326328800988,
      "grad_norm": 0.4268549084663391,
      "learning_rate": 1.5033667720214376e-06,
      "loss": 0.8703,
      "step": 6736
    },
    {
      "epoch": 1.8507073204230187,
      "grad_norm": 0.3798433542251587,
      "learning_rate": 1.5006183866978153e-06,
      "loss": 0.9123,
      "step": 6737
    },
    {
      "epoch": 1.8509820079659387,
      "grad_norm": 0.4948151111602783,
      "learning_rate": 1.497870001374193e-06,
      "loss": 0.9787,
      "step": 6738
    },
    {
      "epoch": 1.8512566955088587,
      "grad_norm": 0.5434207320213318,
      "learning_rate": 1.4951216160505706e-06,
      "loss": 0.8101,
      "step": 6739
    },
    {
      "epoch": 1.8515313830517786,
      "grad_norm": 0.5492603182792664,
      "learning_rate": 1.4923732307269482e-06,
      "loss": 0.6863,
      "step": 6740
    },
    {
      "epoch": 1.8518060705946984,
      "grad_norm": 0.5184614062309265,
      "learning_rate": 1.4896248454033257e-06,
      "loss": 0.8119,
      "step": 6741
    },
    {
      "epoch": 1.8520807581376184,
      "grad_norm": 0.6023529767990112,
      "learning_rate": 1.4868764600797033e-06,
      "loss": 0.7933,
      "step": 6742
    },
    {
      "epoch": 1.8523554456805384,
      "grad_norm": 0.4213932454586029,
      "learning_rate": 1.484128074756081e-06,
      "loss": 0.6284,
      "step": 6743
    },
    {
      "epoch": 1.8526301332234583,
      "grad_norm": 0.5001868009567261,
      "learning_rate": 1.4813796894324586e-06,
      "loss": 0.7917,
      "step": 6744
    },
    {
      "epoch": 1.852904820766378,
      "grad_norm": 0.4658079445362091,
      "learning_rate": 1.4786313041088363e-06,
      "loss": 1.051,
      "step": 6745
    },
    {
      "epoch": 1.8531795083092981,
      "grad_norm": 0.3912467956542969,
      "learning_rate": 1.475882918785214e-06,
      "loss": 0.7453,
      "step": 6746
    },
    {
      "epoch": 1.8534541958522182,
      "grad_norm": 0.5171387791633606,
      "learning_rate": 1.4731345334615916e-06,
      "loss": 0.8107,
      "step": 6747
    },
    {
      "epoch": 1.8537288833951382,
      "grad_norm": 0.4645439684391022,
      "learning_rate": 1.470386148137969e-06,
      "loss": 0.7828,
      "step": 6748
    },
    {
      "epoch": 1.854003570938058,
      "grad_norm": 0.6104767918586731,
      "learning_rate": 1.4676377628143467e-06,
      "loss": 0.5599,
      "step": 6749
    },
    {
      "epoch": 1.8542782584809778,
      "grad_norm": 0.4795092046260834,
      "learning_rate": 1.4648893774907244e-06,
      "loss": 0.7204,
      "step": 6750
    },
    {
      "epoch": 1.8545529460238979,
      "grad_norm": 0.5637977719306946,
      "learning_rate": 1.462140992167102e-06,
      "loss": 0.6658,
      "step": 6751
    },
    {
      "epoch": 1.854827633566818,
      "grad_norm": 0.3502955436706543,
      "learning_rate": 1.4593926068434797e-06,
      "loss": 0.6255,
      "step": 6752
    },
    {
      "epoch": 1.8551023211097377,
      "grad_norm": 0.4871920049190521,
      "learning_rate": 1.4566442215198573e-06,
      "loss": 0.879,
      "step": 6753
    },
    {
      "epoch": 1.8553770086526575,
      "grad_norm": 0.4318540394306183,
      "learning_rate": 1.453895836196235e-06,
      "loss": 0.6025,
      "step": 6754
    },
    {
      "epoch": 1.8556516961955776,
      "grad_norm": 0.45124879479408264,
      "learning_rate": 1.4511474508726126e-06,
      "loss": 0.8103,
      "step": 6755
    },
    {
      "epoch": 1.8559263837384976,
      "grad_norm": 0.46068164706230164,
      "learning_rate": 1.44839906554899e-06,
      "loss": 0.803,
      "step": 6756
    },
    {
      "epoch": 1.8562010712814174,
      "grad_norm": 0.43738847970962524,
      "learning_rate": 1.4456506802253677e-06,
      "loss": 1.0127,
      "step": 6757
    },
    {
      "epoch": 1.8564757588243372,
      "grad_norm": 0.4878472089767456,
      "learning_rate": 1.4429022949017454e-06,
      "loss": 0.7686,
      "step": 6758
    },
    {
      "epoch": 1.8567504463672573,
      "grad_norm": 0.5381679534912109,
      "learning_rate": 1.440153909578123e-06,
      "loss": 0.577,
      "step": 6759
    },
    {
      "epoch": 1.8570251339101773,
      "grad_norm": 0.46171489357948303,
      "learning_rate": 1.4374055242545007e-06,
      "loss": 0.7439,
      "step": 6760
    },
    {
      "epoch": 1.8572998214530971,
      "grad_norm": 0.4059096872806549,
      "learning_rate": 1.4346571389308784e-06,
      "loss": 0.5561,
      "step": 6761
    },
    {
      "epoch": 1.857574508996017,
      "grad_norm": 0.49572303891181946,
      "learning_rate": 1.431908753607256e-06,
      "loss": 0.8596,
      "step": 6762
    },
    {
      "epoch": 1.857849196538937,
      "grad_norm": 0.44611260294914246,
      "learning_rate": 1.4291603682836337e-06,
      "loss": 0.982,
      "step": 6763
    },
    {
      "epoch": 1.858123884081857,
      "grad_norm": 0.4365690350532532,
      "learning_rate": 1.4264119829600111e-06,
      "loss": 0.9234,
      "step": 6764
    },
    {
      "epoch": 1.8583985716247768,
      "grad_norm": 0.4417438805103302,
      "learning_rate": 1.4236635976363888e-06,
      "loss": 0.7551,
      "step": 6765
    },
    {
      "epoch": 1.8586732591676967,
      "grad_norm": 0.4105669856071472,
      "learning_rate": 1.4209152123127664e-06,
      "loss": 0.8395,
      "step": 6766
    },
    {
      "epoch": 1.8589479467106167,
      "grad_norm": 0.41522911190986633,
      "learning_rate": 1.418166826989144e-06,
      "loss": 0.9426,
      "step": 6767
    },
    {
      "epoch": 1.8592226342535367,
      "grad_norm": 0.545627236366272,
      "learning_rate": 1.4154184416655217e-06,
      "loss": 0.8024,
      "step": 6768
    },
    {
      "epoch": 1.8594973217964565,
      "grad_norm": 0.5315762758255005,
      "learning_rate": 1.4126700563418994e-06,
      "loss": 0.8828,
      "step": 6769
    },
    {
      "epoch": 1.8597720093393764,
      "grad_norm": 0.4454786777496338,
      "learning_rate": 1.409921671018277e-06,
      "loss": 0.733,
      "step": 6770
    },
    {
      "epoch": 1.8600466968822964,
      "grad_norm": 0.3884722888469696,
      "learning_rate": 1.4071732856946547e-06,
      "loss": 0.8856,
      "step": 6771
    },
    {
      "epoch": 1.8603213844252164,
      "grad_norm": 0.4790688157081604,
      "learning_rate": 1.4044249003710322e-06,
      "loss": 0.956,
      "step": 6772
    },
    {
      "epoch": 1.8605960719681363,
      "grad_norm": 0.583221435546875,
      "learning_rate": 1.4016765150474098e-06,
      "loss": 0.7606,
      "step": 6773
    },
    {
      "epoch": 1.860870759511056,
      "grad_norm": 0.3795173764228821,
      "learning_rate": 1.3989281297237875e-06,
      "loss": 0.7461,
      "step": 6774
    },
    {
      "epoch": 1.861145447053976,
      "grad_norm": 0.36491289734840393,
      "learning_rate": 1.3961797444001651e-06,
      "loss": 0.8873,
      "step": 6775
    },
    {
      "epoch": 1.8614201345968961,
      "grad_norm": 0.4014540910720825,
      "learning_rate": 1.3934313590765428e-06,
      "loss": 0.5465,
      "step": 6776
    },
    {
      "epoch": 1.861694822139816,
      "grad_norm": 0.4206380248069763,
      "learning_rate": 1.3906829737529204e-06,
      "loss": 0.6066,
      "step": 6777
    },
    {
      "epoch": 1.8619695096827358,
      "grad_norm": 0.4395750164985657,
      "learning_rate": 1.387934588429298e-06,
      "loss": 0.9319,
      "step": 6778
    },
    {
      "epoch": 1.8622441972256558,
      "grad_norm": 0.47188106179237366,
      "learning_rate": 1.3851862031056757e-06,
      "loss": 0.7385,
      "step": 6779
    },
    {
      "epoch": 1.8625188847685759,
      "grad_norm": 0.38099080324172974,
      "learning_rate": 1.3824378177820532e-06,
      "loss": 0.7101,
      "step": 6780
    },
    {
      "epoch": 1.8627935723114957,
      "grad_norm": 0.42658358812332153,
      "learning_rate": 1.3796894324584308e-06,
      "loss": 0.833,
      "step": 6781
    },
    {
      "epoch": 1.8630682598544155,
      "grad_norm": 0.4553048014640808,
      "learning_rate": 1.3769410471348085e-06,
      "loss": 0.8601,
      "step": 6782
    },
    {
      "epoch": 1.8633429473973355,
      "grad_norm": 0.39055269956588745,
      "learning_rate": 1.3741926618111862e-06,
      "loss": 0.6669,
      "step": 6783
    },
    {
      "epoch": 1.8636176349402556,
      "grad_norm": 0.3977274000644684,
      "learning_rate": 1.3714442764875638e-06,
      "loss": 0.8168,
      "step": 6784
    },
    {
      "epoch": 1.8638923224831754,
      "grad_norm": 0.6095190644264221,
      "learning_rate": 1.3686958911639415e-06,
      "loss": 0.757,
      "step": 6785
    },
    {
      "epoch": 1.8641670100260952,
      "grad_norm": 0.4662819504737854,
      "learning_rate": 1.3659475058403191e-06,
      "loss": 0.8868,
      "step": 6786
    },
    {
      "epoch": 1.8644416975690152,
      "grad_norm": 0.2968122661113739,
      "learning_rate": 1.3631991205166966e-06,
      "loss": 0.493,
      "step": 6787
    },
    {
      "epoch": 1.8647163851119353,
      "grad_norm": 0.5317868590354919,
      "learning_rate": 1.3604507351930742e-06,
      "loss": 0.9211,
      "step": 6788
    },
    {
      "epoch": 1.864991072654855,
      "grad_norm": 0.36260610818862915,
      "learning_rate": 1.3577023498694519e-06,
      "loss": 0.5063,
      "step": 6789
    },
    {
      "epoch": 1.865265760197775,
      "grad_norm": 0.7659564018249512,
      "learning_rate": 1.3549539645458295e-06,
      "loss": 0.7108,
      "step": 6790
    },
    {
      "epoch": 1.865540447740695,
      "grad_norm": 0.37922966480255127,
      "learning_rate": 1.3522055792222072e-06,
      "loss": 0.915,
      "step": 6791
    },
    {
      "epoch": 1.865815135283615,
      "grad_norm": 0.4697253704071045,
      "learning_rate": 1.3494571938985849e-06,
      "loss": 0.7668,
      "step": 6792
    },
    {
      "epoch": 1.8660898228265348,
      "grad_norm": 0.4447784125804901,
      "learning_rate": 1.3467088085749625e-06,
      "loss": 0.9957,
      "step": 6793
    },
    {
      "epoch": 1.8663645103694546,
      "grad_norm": 0.41906771063804626,
      "learning_rate": 1.3439604232513402e-06,
      "loss": 0.6322,
      "step": 6794
    },
    {
      "epoch": 1.8666391979123746,
      "grad_norm": 0.4499603807926178,
      "learning_rate": 1.3412120379277176e-06,
      "loss": 0.868,
      "step": 6795
    },
    {
      "epoch": 1.8669138854552947,
      "grad_norm": 0.5113299489021301,
      "learning_rate": 1.3384636526040953e-06,
      "loss": 0.7628,
      "step": 6796
    },
    {
      "epoch": 1.8671885729982145,
      "grad_norm": 0.41360750794410706,
      "learning_rate": 1.335715267280473e-06,
      "loss": 0.7891,
      "step": 6797
    },
    {
      "epoch": 1.8674632605411343,
      "grad_norm": 0.4025169909000397,
      "learning_rate": 1.3329668819568506e-06,
      "loss": 0.6385,
      "step": 6798
    },
    {
      "epoch": 1.8677379480840544,
      "grad_norm": 0.4873040020465851,
      "learning_rate": 1.3302184966332282e-06,
      "loss": 0.7657,
      "step": 6799
    },
    {
      "epoch": 1.8680126356269744,
      "grad_norm": 0.3919154107570648,
      "learning_rate": 1.3274701113096059e-06,
      "loss": 0.6283,
      "step": 6800
    },
    {
      "epoch": 1.8682873231698942,
      "grad_norm": 0.32967954874038696,
      "learning_rate": 1.3247217259859835e-06,
      "loss": 0.5451,
      "step": 6801
    },
    {
      "epoch": 1.868562010712814,
      "grad_norm": 0.47452080249786377,
      "learning_rate": 1.3219733406623612e-06,
      "loss": 0.6959,
      "step": 6802
    },
    {
      "epoch": 1.868836698255734,
      "grad_norm": 0.4410618245601654,
      "learning_rate": 1.3192249553387386e-06,
      "loss": 0.6761,
      "step": 6803
    },
    {
      "epoch": 1.869111385798654,
      "grad_norm": 0.6087415218353271,
      "learning_rate": 1.3164765700151163e-06,
      "loss": 0.9915,
      "step": 6804
    },
    {
      "epoch": 1.869386073341574,
      "grad_norm": 0.44184231758117676,
      "learning_rate": 1.313728184691494e-06,
      "loss": 0.7695,
      "step": 6805
    },
    {
      "epoch": 1.8696607608844937,
      "grad_norm": 0.45755672454833984,
      "learning_rate": 1.3109797993678716e-06,
      "loss": 0.8711,
      "step": 6806
    },
    {
      "epoch": 1.8699354484274138,
      "grad_norm": 0.35848209261894226,
      "learning_rate": 1.3082314140442493e-06,
      "loss": 0.7289,
      "step": 6807
    },
    {
      "epoch": 1.8702101359703338,
      "grad_norm": 0.44321393966674805,
      "learning_rate": 1.305483028720627e-06,
      "loss": 0.795,
      "step": 6808
    },
    {
      "epoch": 1.8704848235132536,
      "grad_norm": 0.4076327383518219,
      "learning_rate": 1.3027346433970042e-06,
      "loss": 0.9229,
      "step": 6809
    },
    {
      "epoch": 1.8707595110561734,
      "grad_norm": 0.37990063428878784,
      "learning_rate": 1.2999862580733818e-06,
      "loss": 0.8279,
      "step": 6810
    },
    {
      "epoch": 1.8710341985990935,
      "grad_norm": 0.4760870337486267,
      "learning_rate": 1.2972378727497595e-06,
      "loss": 0.7847,
      "step": 6811
    },
    {
      "epoch": 1.8713088861420135,
      "grad_norm": 0.6126416325569153,
      "learning_rate": 1.2944894874261371e-06,
      "loss": 0.8393,
      "step": 6812
    },
    {
      "epoch": 1.8715835736849336,
      "grad_norm": 0.4348177909851074,
      "learning_rate": 1.2917411021025148e-06,
      "loss": 0.5443,
      "step": 6813
    },
    {
      "epoch": 1.8718582612278534,
      "grad_norm": 0.5716005563735962,
      "learning_rate": 1.2889927167788924e-06,
      "loss": 0.8468,
      "step": 6814
    },
    {
      "epoch": 1.8721329487707732,
      "grad_norm": 0.35004478693008423,
      "learning_rate": 1.28624433145527e-06,
      "loss": 0.9369,
      "step": 6815
    },
    {
      "epoch": 1.8724076363136932,
      "grad_norm": 0.5481693148612976,
      "learning_rate": 1.2834959461316475e-06,
      "loss": 0.9018,
      "step": 6816
    },
    {
      "epoch": 1.8726823238566133,
      "grad_norm": 0.5037670135498047,
      "learning_rate": 1.2807475608080252e-06,
      "loss": 0.8366,
      "step": 6817
    },
    {
      "epoch": 1.872957011399533,
      "grad_norm": 0.3904690742492676,
      "learning_rate": 1.2779991754844029e-06,
      "loss": 1.0323,
      "step": 6818
    },
    {
      "epoch": 1.873231698942453,
      "grad_norm": 0.5076096653938293,
      "learning_rate": 1.2752507901607805e-06,
      "loss": 0.7108,
      "step": 6819
    },
    {
      "epoch": 1.873506386485373,
      "grad_norm": 0.3357529044151306,
      "learning_rate": 1.2725024048371582e-06,
      "loss": 0.9118,
      "step": 6820
    },
    {
      "epoch": 1.873781074028293,
      "grad_norm": 0.5104172229766846,
      "learning_rate": 1.2697540195135358e-06,
      "loss": 0.7583,
      "step": 6821
    },
    {
      "epoch": 1.8740557615712128,
      "grad_norm": 0.3750895857810974,
      "learning_rate": 1.2670056341899135e-06,
      "loss": 0.6678,
      "step": 6822
    },
    {
      "epoch": 1.8743304491141326,
      "grad_norm": 0.378162682056427,
      "learning_rate": 1.2642572488662911e-06,
      "loss": 0.4984,
      "step": 6823
    },
    {
      "epoch": 1.8746051366570526,
      "grad_norm": 0.39374974370002747,
      "learning_rate": 1.2615088635426686e-06,
      "loss": 0.749,
      "step": 6824
    },
    {
      "epoch": 1.8748798241999727,
      "grad_norm": 0.5275006294250488,
      "learning_rate": 1.2587604782190462e-06,
      "loss": 0.8278,
      "step": 6825
    },
    {
      "epoch": 1.8751545117428925,
      "grad_norm": 0.5842604637145996,
      "learning_rate": 1.2560120928954239e-06,
      "loss": 0.8448,
      "step": 6826
    },
    {
      "epoch": 1.8754291992858123,
      "grad_norm": 0.49629345536231995,
      "learning_rate": 1.2532637075718015e-06,
      "loss": 1.0364,
      "step": 6827
    },
    {
      "epoch": 1.8757038868287323,
      "grad_norm": 0.5387627482414246,
      "learning_rate": 1.2505153222481792e-06,
      "loss": 0.7566,
      "step": 6828
    },
    {
      "epoch": 1.8759785743716524,
      "grad_norm": 0.5848492980003357,
      "learning_rate": 1.2477669369245569e-06,
      "loss": 0.8526,
      "step": 6829
    },
    {
      "epoch": 1.8762532619145722,
      "grad_norm": 0.4226507842540741,
      "learning_rate": 1.2450185516009345e-06,
      "loss": 0.6684,
      "step": 6830
    },
    {
      "epoch": 1.876527949457492,
      "grad_norm": 0.44343554973602295,
      "learning_rate": 1.2422701662773122e-06,
      "loss": 1.0057,
      "step": 6831
    },
    {
      "epoch": 1.876802637000412,
      "grad_norm": 0.40060707926750183,
      "learning_rate": 1.2395217809536898e-06,
      "loss": 0.5525,
      "step": 6832
    },
    {
      "epoch": 1.877077324543332,
      "grad_norm": 0.5234181880950928,
      "learning_rate": 1.2367733956300675e-06,
      "loss": 0.9179,
      "step": 6833
    },
    {
      "epoch": 1.877352012086252,
      "grad_norm": 0.47203224897384644,
      "learning_rate": 1.2340250103064451e-06,
      "loss": 0.6595,
      "step": 6834
    },
    {
      "epoch": 1.8776266996291717,
      "grad_norm": 0.47343140840530396,
      "learning_rate": 1.2312766249828228e-06,
      "loss": 0.7251,
      "step": 6835
    },
    {
      "epoch": 1.8779013871720918,
      "grad_norm": 0.5541365146636963,
      "learning_rate": 1.2285282396592004e-06,
      "loss": 0.785,
      "step": 6836
    },
    {
      "epoch": 1.8781760747150118,
      "grad_norm": 0.5203399658203125,
      "learning_rate": 1.2257798543355779e-06,
      "loss": 0.7267,
      "step": 6837
    },
    {
      "epoch": 1.8784507622579316,
      "grad_norm": 0.5578840970993042,
      "learning_rate": 1.2230314690119555e-06,
      "loss": 0.6109,
      "step": 6838
    },
    {
      "epoch": 1.8787254498008514,
      "grad_norm": 0.4171781837940216,
      "learning_rate": 1.2202830836883332e-06,
      "loss": 0.8085,
      "step": 6839
    },
    {
      "epoch": 1.8790001373437715,
      "grad_norm": 0.5567190051078796,
      "learning_rate": 1.2175346983647109e-06,
      "loss": 0.9659,
      "step": 6840
    },
    {
      "epoch": 1.8792748248866915,
      "grad_norm": 0.5489189028739929,
      "learning_rate": 1.2147863130410885e-06,
      "loss": 0.6999,
      "step": 6841
    },
    {
      "epoch": 1.8795495124296113,
      "grad_norm": 0.35405924916267395,
      "learning_rate": 1.2120379277174662e-06,
      "loss": 0.8002,
      "step": 6842
    },
    {
      "epoch": 1.8798241999725311,
      "grad_norm": 0.2871133089065552,
      "learning_rate": 1.2092895423938438e-06,
      "loss": 0.451,
      "step": 6843
    },
    {
      "epoch": 1.8800988875154512,
      "grad_norm": 0.4934852421283722,
      "learning_rate": 1.2065411570702213e-06,
      "loss": 0.7289,
      "step": 6844
    },
    {
      "epoch": 1.8803735750583712,
      "grad_norm": 0.4324808716773987,
      "learning_rate": 1.203792771746599e-06,
      "loss": 0.858,
      "step": 6845
    },
    {
      "epoch": 1.880648262601291,
      "grad_norm": 0.43556034564971924,
      "learning_rate": 1.2010443864229766e-06,
      "loss": 0.9766,
      "step": 6846
    },
    {
      "epoch": 1.8809229501442108,
      "grad_norm": 0.45063114166259766,
      "learning_rate": 1.1982960010993542e-06,
      "loss": 0.7775,
      "step": 6847
    },
    {
      "epoch": 1.8811976376871309,
      "grad_norm": 0.6123026609420776,
      "learning_rate": 1.195547615775732e-06,
      "loss": 0.8752,
      "step": 6848
    },
    {
      "epoch": 1.881472325230051,
      "grad_norm": 0.4122432470321655,
      "learning_rate": 1.1927992304521096e-06,
      "loss": 0.8217,
      "step": 6849
    },
    {
      "epoch": 1.8817470127729707,
      "grad_norm": 0.5595595836639404,
      "learning_rate": 1.1900508451284872e-06,
      "loss": 0.6918,
      "step": 6850
    },
    {
      "epoch": 1.8820217003158906,
      "grad_norm": 0.6137750148773193,
      "learning_rate": 1.1873024598048649e-06,
      "loss": 0.7634,
      "step": 6851
    },
    {
      "epoch": 1.8822963878588106,
      "grad_norm": 0.36473652720451355,
      "learning_rate": 1.1845540744812423e-06,
      "loss": 0.9737,
      "step": 6852
    },
    {
      "epoch": 1.8825710754017306,
      "grad_norm": 0.5468374490737915,
      "learning_rate": 1.18180568915762e-06,
      "loss": 0.9093,
      "step": 6853
    },
    {
      "epoch": 1.8828457629446504,
      "grad_norm": 0.5371907353401184,
      "learning_rate": 1.1790573038339976e-06,
      "loss": 0.6498,
      "step": 6854
    },
    {
      "epoch": 1.8831204504875703,
      "grad_norm": 0.4695816934108734,
      "learning_rate": 1.1763089185103753e-06,
      "loss": 0.74,
      "step": 6855
    },
    {
      "epoch": 1.8833951380304903,
      "grad_norm": 0.4983128607273102,
      "learning_rate": 1.173560533186753e-06,
      "loss": 0.7293,
      "step": 6856
    },
    {
      "epoch": 1.8836698255734103,
      "grad_norm": 0.43958792090415955,
      "learning_rate": 1.1708121478631306e-06,
      "loss": 0.8605,
      "step": 6857
    },
    {
      "epoch": 1.8839445131163302,
      "grad_norm": 0.38697460293769836,
      "learning_rate": 1.1680637625395082e-06,
      "loss": 0.879,
      "step": 6858
    },
    {
      "epoch": 1.88421920065925,
      "grad_norm": 0.5978444814682007,
      "learning_rate": 1.165315377215886e-06,
      "loss": 0.6776,
      "step": 6859
    },
    {
      "epoch": 1.88449388820217,
      "grad_norm": 0.3245832324028015,
      "learning_rate": 1.1625669918922633e-06,
      "loss": 0.5086,
      "step": 6860
    },
    {
      "epoch": 1.88476857574509,
      "grad_norm": 0.39487889409065247,
      "learning_rate": 1.159818606568641e-06,
      "loss": 0.7409,
      "step": 6861
    },
    {
      "epoch": 1.8850432632880099,
      "grad_norm": 0.40455880761146545,
      "learning_rate": 1.1570702212450187e-06,
      "loss": 0.615,
      "step": 6862
    },
    {
      "epoch": 1.8853179508309297,
      "grad_norm": 0.475130558013916,
      "learning_rate": 1.1543218359213963e-06,
      "loss": 0.7837,
      "step": 6863
    },
    {
      "epoch": 1.8855926383738497,
      "grad_norm": 0.45665472745895386,
      "learning_rate": 1.151573450597774e-06,
      "loss": 0.8514,
      "step": 6864
    },
    {
      "epoch": 1.8858673259167698,
      "grad_norm": 0.4475814700126648,
      "learning_rate": 1.1488250652741516e-06,
      "loss": 0.6228,
      "step": 6865
    },
    {
      "epoch": 1.8861420134596896,
      "grad_norm": 0.4148368537425995,
      "learning_rate": 1.1460766799505293e-06,
      "loss": 0.9108,
      "step": 6866
    },
    {
      "epoch": 1.8864167010026094,
      "grad_norm": 0.43740108609199524,
      "learning_rate": 1.143328294626907e-06,
      "loss": 0.7458,
      "step": 6867
    },
    {
      "epoch": 1.8866913885455294,
      "grad_norm": 0.39545002579689026,
      "learning_rate": 1.1405799093032844e-06,
      "loss": 0.8867,
      "step": 6868
    },
    {
      "epoch": 1.8869660760884495,
      "grad_norm": 0.32630497217178345,
      "learning_rate": 1.137831523979662e-06,
      "loss": 0.5783,
      "step": 6869
    },
    {
      "epoch": 1.8872407636313693,
      "grad_norm": 0.5657339096069336,
      "learning_rate": 1.1350831386560397e-06,
      "loss": 0.6736,
      "step": 6870
    },
    {
      "epoch": 1.887515451174289,
      "grad_norm": 0.5276301503181458,
      "learning_rate": 1.1323347533324174e-06,
      "loss": 0.7315,
      "step": 6871
    },
    {
      "epoch": 1.8877901387172091,
      "grad_norm": 0.4446544349193573,
      "learning_rate": 1.129586368008795e-06,
      "loss": 0.8342,
      "step": 6872
    },
    {
      "epoch": 1.8880648262601292,
      "grad_norm": 0.375026136636734,
      "learning_rate": 1.1268379826851727e-06,
      "loss": 0.6801,
      "step": 6873
    },
    {
      "epoch": 1.888339513803049,
      "grad_norm": 0.4676097333431244,
      "learning_rate": 1.1240895973615503e-06,
      "loss": 0.6996,
      "step": 6874
    },
    {
      "epoch": 1.8886142013459688,
      "grad_norm": 0.5105862021446228,
      "learning_rate": 1.1213412120379278e-06,
      "loss": 0.6784,
      "step": 6875
    },
    {
      "epoch": 1.8888888888888888,
      "grad_norm": 0.39569151401519775,
      "learning_rate": 1.1185928267143054e-06,
      "loss": 0.6752,
      "step": 6876
    },
    {
      "epoch": 1.8891635764318089,
      "grad_norm": 0.48595407605171204,
      "learning_rate": 1.115844441390683e-06,
      "loss": 0.6547,
      "step": 6877
    },
    {
      "epoch": 1.8894382639747287,
      "grad_norm": 0.5265291333198547,
      "learning_rate": 1.1130960560670607e-06,
      "loss": 0.8465,
      "step": 6878
    },
    {
      "epoch": 1.8897129515176485,
      "grad_norm": 0.542370617389679,
      "learning_rate": 1.1103476707434384e-06,
      "loss": 0.7961,
      "step": 6879
    },
    {
      "epoch": 1.8899876390605685,
      "grad_norm": 0.4787590205669403,
      "learning_rate": 1.107599285419816e-06,
      "loss": 0.8482,
      "step": 6880
    },
    {
      "epoch": 1.8902623266034886,
      "grad_norm": 0.44325971603393555,
      "learning_rate": 1.1048509000961937e-06,
      "loss": 0.9313,
      "step": 6881
    },
    {
      "epoch": 1.8905370141464086,
      "grad_norm": 0.3453201949596405,
      "learning_rate": 1.1021025147725714e-06,
      "loss": 0.5033,
      "step": 6882
    },
    {
      "epoch": 1.8908117016893284,
      "grad_norm": 0.42634525895118713,
      "learning_rate": 1.0993541294489488e-06,
      "loss": 0.9053,
      "step": 6883
    },
    {
      "epoch": 1.8910863892322483,
      "grad_norm": 0.5705235004425049,
      "learning_rate": 1.0966057441253265e-06,
      "loss": 0.7653,
      "step": 6884
    },
    {
      "epoch": 1.8913610767751683,
      "grad_norm": 0.45057597756385803,
      "learning_rate": 1.0938573588017041e-06,
      "loss": 0.6527,
      "step": 6885
    },
    {
      "epoch": 1.8916357643180883,
      "grad_norm": 0.5039416551589966,
      "learning_rate": 1.0911089734780818e-06,
      "loss": 0.9494,
      "step": 6886
    },
    {
      "epoch": 1.8919104518610081,
      "grad_norm": 0.402743399143219,
      "learning_rate": 1.0883605881544594e-06,
      "loss": 0.8817,
      "step": 6887
    },
    {
      "epoch": 1.892185139403928,
      "grad_norm": 0.7503561973571777,
      "learning_rate": 1.0856122028308369e-06,
      "loss": 0.6726,
      "step": 6888
    },
    {
      "epoch": 1.892459826946848,
      "grad_norm": 0.5338152050971985,
      "learning_rate": 1.0828638175072145e-06,
      "loss": 0.7243,
      "step": 6889
    },
    {
      "epoch": 1.892734514489768,
      "grad_norm": 0.4831852316856384,
      "learning_rate": 1.0801154321835922e-06,
      "loss": 0.6975,
      "step": 6890
    },
    {
      "epoch": 1.8930092020326879,
      "grad_norm": 0.3767944276332855,
      "learning_rate": 1.0773670468599698e-06,
      "loss": 0.6001,
      "step": 6891
    },
    {
      "epoch": 1.8932838895756077,
      "grad_norm": 0.44602441787719727,
      "learning_rate": 1.0746186615363475e-06,
      "loss": 0.7636,
      "step": 6892
    },
    {
      "epoch": 1.8935585771185277,
      "grad_norm": 0.3907627761363983,
      "learning_rate": 1.0718702762127251e-06,
      "loss": 0.6115,
      "step": 6893
    },
    {
      "epoch": 1.8938332646614477,
      "grad_norm": 0.35133132338523865,
      "learning_rate": 1.0691218908891026e-06,
      "loss": 0.6005,
      "step": 6894
    },
    {
      "epoch": 1.8941079522043676,
      "grad_norm": 0.35516592860221863,
      "learning_rate": 1.0663735055654802e-06,
      "loss": 0.5634,
      "step": 6895
    },
    {
      "epoch": 1.8943826397472874,
      "grad_norm": 0.3962392210960388,
      "learning_rate": 1.063625120241858e-06,
      "loss": 0.722,
      "step": 6896
    },
    {
      "epoch": 1.8946573272902074,
      "grad_norm": 0.425265371799469,
      "learning_rate": 1.0608767349182356e-06,
      "loss": 0.7155,
      "step": 6897
    },
    {
      "epoch": 1.8949320148331275,
      "grad_norm": 0.3699781000614166,
      "learning_rate": 1.0581283495946132e-06,
      "loss": 0.6985,
      "step": 6898
    },
    {
      "epoch": 1.8952067023760473,
      "grad_norm": 0.4376623034477234,
      "learning_rate": 1.0553799642709909e-06,
      "loss": 0.8829,
      "step": 6899
    },
    {
      "epoch": 1.895481389918967,
      "grad_norm": 0.5243950486183167,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 0.7201,
      "step": 6900
    },
    {
      "epoch": 1.8957560774618871,
      "grad_norm": 0.5456399321556091,
      "learning_rate": 1.049883193623746e-06,
      "loss": 0.6048,
      "step": 6901
    },
    {
      "epoch": 1.8960307650048072,
      "grad_norm": 0.5794591903686523,
      "learning_rate": 1.0471348083001236e-06,
      "loss": 0.8828,
      "step": 6902
    },
    {
      "epoch": 1.896305452547727,
      "grad_norm": 0.33150386810302734,
      "learning_rate": 1.0443864229765013e-06,
      "loss": 0.6968,
      "step": 6903
    },
    {
      "epoch": 1.8965801400906468,
      "grad_norm": 0.3552091121673584,
      "learning_rate": 1.041638037652879e-06,
      "loss": 0.6641,
      "step": 6904
    },
    {
      "epoch": 1.8968548276335668,
      "grad_norm": 0.4562767744064331,
      "learning_rate": 1.0388896523292566e-06,
      "loss": 0.6794,
      "step": 6905
    },
    {
      "epoch": 1.8971295151764869,
      "grad_norm": 0.6071734428405762,
      "learning_rate": 1.0361412670056343e-06,
      "loss": 0.6889,
      "step": 6906
    },
    {
      "epoch": 1.8974042027194067,
      "grad_norm": 0.48104923963546753,
      "learning_rate": 1.033392881682012e-06,
      "loss": 0.8653,
      "step": 6907
    },
    {
      "epoch": 1.8976788902623265,
      "grad_norm": 0.41238752007484436,
      "learning_rate": 1.0306444963583896e-06,
      "loss": 0.5195,
      "step": 6908
    },
    {
      "epoch": 1.8979535778052465,
      "grad_norm": 0.33859938383102417,
      "learning_rate": 1.027896111034767e-06,
      "loss": 0.9817,
      "step": 6909
    },
    {
      "epoch": 1.8982282653481666,
      "grad_norm": 0.5605043768882751,
      "learning_rate": 1.0251477257111447e-06,
      "loss": 0.9333,
      "step": 6910
    },
    {
      "epoch": 1.8985029528910864,
      "grad_norm": 0.45993587374687195,
      "learning_rate": 1.0223993403875223e-06,
      "loss": 0.8908,
      "step": 6911
    },
    {
      "epoch": 1.8987776404340062,
      "grad_norm": 0.39788350462913513,
      "learning_rate": 1.0196509550639e-06,
      "loss": 0.8051,
      "step": 6912
    },
    {
      "epoch": 1.8990523279769262,
      "grad_norm": 0.5313208699226379,
      "learning_rate": 1.0169025697402776e-06,
      "loss": 0.5253,
      "step": 6913
    },
    {
      "epoch": 1.8993270155198463,
      "grad_norm": 0.4522089958190918,
      "learning_rate": 1.0141541844166553e-06,
      "loss": 0.8351,
      "step": 6914
    },
    {
      "epoch": 1.899601703062766,
      "grad_norm": 0.4405219554901123,
      "learning_rate": 1.011405799093033e-06,
      "loss": 0.7253,
      "step": 6915
    },
    {
      "epoch": 1.899876390605686,
      "grad_norm": 0.4644671380519867,
      "learning_rate": 1.0086574137694106e-06,
      "loss": 0.9312,
      "step": 6916
    },
    {
      "epoch": 1.900151078148606,
      "grad_norm": 0.3244141936302185,
      "learning_rate": 1.005909028445788e-06,
      "loss": 0.6734,
      "step": 6917
    },
    {
      "epoch": 1.900425765691526,
      "grad_norm": 0.590843915939331,
      "learning_rate": 1.0031606431221657e-06,
      "loss": 0.891,
      "step": 6918
    },
    {
      "epoch": 1.9007004532344458,
      "grad_norm": 0.34949326515197754,
      "learning_rate": 1.0004122577985434e-06,
      "loss": 0.4774,
      "step": 6919
    },
    {
      "epoch": 1.9009751407773656,
      "grad_norm": 0.4493802785873413,
      "learning_rate": 9.97663872474921e-07,
      "loss": 0.7927,
      "step": 6920
    },
    {
      "epoch": 1.9012498283202857,
      "grad_norm": 0.4087413549423218,
      "learning_rate": 9.949154871512987e-07,
      "loss": 0.9347,
      "step": 6921
    },
    {
      "epoch": 1.9015245158632057,
      "grad_norm": 0.5937046408653259,
      "learning_rate": 9.921671018276763e-07,
      "loss": 0.7516,
      "step": 6922
    },
    {
      "epoch": 1.9017992034061255,
      "grad_norm": 0.5154469013214111,
      "learning_rate": 9.89418716504054e-07,
      "loss": 0.7246,
      "step": 6923
    },
    {
      "epoch": 1.9020738909490453,
      "grad_norm": 0.6415454745292664,
      "learning_rate": 9.866703311804316e-07,
      "loss": 0.6923,
      "step": 6924
    },
    {
      "epoch": 1.9023485784919654,
      "grad_norm": 0.4359308183193207,
      "learning_rate": 9.83921945856809e-07,
      "loss": 0.8676,
      "step": 6925
    },
    {
      "epoch": 1.9026232660348854,
      "grad_norm": 0.5474462509155273,
      "learning_rate": 9.811735605331867e-07,
      "loss": 0.8559,
      "step": 6926
    },
    {
      "epoch": 1.9028979535778052,
      "grad_norm": 0.5096530318260193,
      "learning_rate": 9.784251752095644e-07,
      "loss": 0.8695,
      "step": 6927
    },
    {
      "epoch": 1.903172641120725,
      "grad_norm": 0.450874924659729,
      "learning_rate": 9.75676789885942e-07,
      "loss": 0.6907,
      "step": 6928
    },
    {
      "epoch": 1.903447328663645,
      "grad_norm": 0.5016416907310486,
      "learning_rate": 9.729284045623197e-07,
      "loss": 0.7289,
      "step": 6929
    },
    {
      "epoch": 1.9037220162065651,
      "grad_norm": 0.508916974067688,
      "learning_rate": 9.701800192386974e-07,
      "loss": 0.7477,
      "step": 6930
    },
    {
      "epoch": 1.903996703749485,
      "grad_norm": 0.5870039463043213,
      "learning_rate": 9.67431633915075e-07,
      "loss": 0.6149,
      "step": 6931
    },
    {
      "epoch": 1.9042713912924047,
      "grad_norm": 0.5984142422676086,
      "learning_rate": 9.646832485914525e-07,
      "loss": 0.7226,
      "step": 6932
    },
    {
      "epoch": 1.9045460788353248,
      "grad_norm": 0.39882922172546387,
      "learning_rate": 9.619348632678301e-07,
      "loss": 0.8476,
      "step": 6933
    },
    {
      "epoch": 1.9048207663782448,
      "grad_norm": 0.42213648557662964,
      "learning_rate": 9.591864779442078e-07,
      "loss": 0.7481,
      "step": 6934
    },
    {
      "epoch": 1.9050954539211646,
      "grad_norm": 0.531394898891449,
      "learning_rate": 9.564380926205854e-07,
      "loss": 0.6862,
      "step": 6935
    },
    {
      "epoch": 1.9053701414640845,
      "grad_norm": 0.4164160192012787,
      "learning_rate": 9.53689707296963e-07,
      "loss": 0.539,
      "step": 6936
    },
    {
      "epoch": 1.9056448290070045,
      "grad_norm": 0.5258790850639343,
      "learning_rate": 9.509413219733407e-07,
      "loss": 0.6834,
      "step": 6937
    },
    {
      "epoch": 1.9059195165499245,
      "grad_norm": 0.5016978979110718,
      "learning_rate": 9.481929366497183e-07,
      "loss": 0.7638,
      "step": 6938
    },
    {
      "epoch": 1.9061942040928443,
      "grad_norm": 0.36052465438842773,
      "learning_rate": 9.45444551326096e-07,
      "loss": 0.5537,
      "step": 6939
    },
    {
      "epoch": 1.9064688916357642,
      "grad_norm": 0.5393620729446411,
      "learning_rate": 9.426961660024736e-07,
      "loss": 0.8185,
      "step": 6940
    },
    {
      "epoch": 1.9067435791786842,
      "grad_norm": 0.564990758895874,
      "learning_rate": 9.399477806788513e-07,
      "loss": 0.6982,
      "step": 6941
    },
    {
      "epoch": 1.9070182667216042,
      "grad_norm": 0.6751354932785034,
      "learning_rate": 9.371993953552288e-07,
      "loss": 0.5724,
      "step": 6942
    },
    {
      "epoch": 1.907292954264524,
      "grad_norm": 0.394196480512619,
      "learning_rate": 9.344510100316065e-07,
      "loss": 0.6059,
      "step": 6943
    },
    {
      "epoch": 1.9075676418074439,
      "grad_norm": 0.5987599492073059,
      "learning_rate": 9.317026247079841e-07,
      "loss": 0.7263,
      "step": 6944
    },
    {
      "epoch": 1.907842329350364,
      "grad_norm": 0.5154045224189758,
      "learning_rate": 9.289542393843618e-07,
      "loss": 0.852,
      "step": 6945
    },
    {
      "epoch": 1.908117016893284,
      "grad_norm": 0.6123310327529907,
      "learning_rate": 9.262058540607393e-07,
      "loss": 0.7016,
      "step": 6946
    },
    {
      "epoch": 1.908391704436204,
      "grad_norm": 0.40636223554611206,
      "learning_rate": 9.23457468737117e-07,
      "loss": 0.521,
      "step": 6947
    },
    {
      "epoch": 1.9086663919791238,
      "grad_norm": 0.36979398131370544,
      "learning_rate": 9.207090834134946e-07,
      "loss": 0.6502,
      "step": 6948
    },
    {
      "epoch": 1.9089410795220436,
      "grad_norm": 0.423136442899704,
      "learning_rate": 9.179606980898723e-07,
      "loss": 0.6735,
      "step": 6949
    },
    {
      "epoch": 1.9092157670649637,
      "grad_norm": 0.5653222799301147,
      "learning_rate": 9.152123127662499e-07,
      "loss": 0.9313,
      "step": 6950
    },
    {
      "epoch": 1.9094904546078837,
      "grad_norm": 0.43096789717674255,
      "learning_rate": 9.124639274426275e-07,
      "loss": 0.737,
      "step": 6951
    },
    {
      "epoch": 1.9097651421508035,
      "grad_norm": 0.5539238452911377,
      "learning_rate": 9.097155421190052e-07,
      "loss": 0.6533,
      "step": 6952
    },
    {
      "epoch": 1.9100398296937233,
      "grad_norm": 0.5242297053337097,
      "learning_rate": 9.069671567953828e-07,
      "loss": 0.6308,
      "step": 6953
    },
    {
      "epoch": 1.9103145172366434,
      "grad_norm": 0.6877854466438293,
      "learning_rate": 9.042187714717604e-07,
      "loss": 0.7619,
      "step": 6954
    },
    {
      "epoch": 1.9105892047795634,
      "grad_norm": 0.4603343904018402,
      "learning_rate": 9.01470386148138e-07,
      "loss": 0.9069,
      "step": 6955
    },
    {
      "epoch": 1.9108638923224832,
      "grad_norm": 0.6423461437225342,
      "learning_rate": 8.987220008245157e-07,
      "loss": 0.9751,
      "step": 6956
    },
    {
      "epoch": 1.911138579865403,
      "grad_norm": 0.4251330494880676,
      "learning_rate": 8.959736155008933e-07,
      "loss": 0.5238,
      "step": 6957
    },
    {
      "epoch": 1.911413267408323,
      "grad_norm": 0.4469367265701294,
      "learning_rate": 8.932252301772709e-07,
      "loss": 0.6295,
      "step": 6958
    },
    {
      "epoch": 1.911687954951243,
      "grad_norm": 0.39607468247413635,
      "learning_rate": 8.904768448536485e-07,
      "loss": 0.7393,
      "step": 6959
    },
    {
      "epoch": 1.911962642494163,
      "grad_norm": 0.3959270417690277,
      "learning_rate": 8.877284595300262e-07,
      "loss": 0.9623,
      "step": 6960
    },
    {
      "epoch": 1.9122373300370827,
      "grad_norm": 0.6323999762535095,
      "learning_rate": 8.849800742064039e-07,
      "loss": 0.6743,
      "step": 6961
    },
    {
      "epoch": 1.9125120175800028,
      "grad_norm": 0.551388144493103,
      "learning_rate": 8.822316888827814e-07,
      "loss": 0.6282,
      "step": 6962
    },
    {
      "epoch": 1.9127867051229228,
      "grad_norm": 0.5448858737945557,
      "learning_rate": 8.794833035591591e-07,
      "loss": 0.6617,
      "step": 6963
    },
    {
      "epoch": 1.9130613926658426,
      "grad_norm": 0.37920117378234863,
      "learning_rate": 8.767349182355367e-07,
      "loss": 0.6426,
      "step": 6964
    },
    {
      "epoch": 1.9133360802087624,
      "grad_norm": 0.5382164120674133,
      "learning_rate": 8.739865329119144e-07,
      "loss": 0.6058,
      "step": 6965
    },
    {
      "epoch": 1.9136107677516825,
      "grad_norm": 0.5099905729293823,
      "learning_rate": 8.712381475882919e-07,
      "loss": 0.7618,
      "step": 6966
    },
    {
      "epoch": 1.9138854552946025,
      "grad_norm": 0.4099205732345581,
      "learning_rate": 8.684897622646696e-07,
      "loss": 0.7058,
      "step": 6967
    },
    {
      "epoch": 1.9141601428375223,
      "grad_norm": 0.3086361587047577,
      "learning_rate": 8.657413769410472e-07,
      "loss": 0.7964,
      "step": 6968
    },
    {
      "epoch": 1.9144348303804422,
      "grad_norm": 0.35485345125198364,
      "learning_rate": 8.629929916174248e-07,
      "loss": 0.8455,
      "step": 6969
    },
    {
      "epoch": 1.9147095179233622,
      "grad_norm": 0.4326930344104767,
      "learning_rate": 8.602446062938024e-07,
      "loss": 0.7347,
      "step": 6970
    },
    {
      "epoch": 1.9149842054662822,
      "grad_norm": 0.6124758124351501,
      "learning_rate": 8.574962209701801e-07,
      "loss": 0.8369,
      "step": 6971
    },
    {
      "epoch": 1.915258893009202,
      "grad_norm": 0.38550105690956116,
      "learning_rate": 8.547478356465578e-07,
      "loss": 0.7484,
      "step": 6972
    },
    {
      "epoch": 1.9155335805521219,
      "grad_norm": 0.4447147250175476,
      "learning_rate": 8.519994503229353e-07,
      "loss": 0.8142,
      "step": 6973
    },
    {
      "epoch": 1.915808268095042,
      "grad_norm": 0.45867908000946045,
      "learning_rate": 8.49251064999313e-07,
      "loss": 0.7435,
      "step": 6974
    },
    {
      "epoch": 1.916082955637962,
      "grad_norm": 0.46390044689178467,
      "learning_rate": 8.465026796756906e-07,
      "loss": 0.6788,
      "step": 6975
    },
    {
      "epoch": 1.9163576431808818,
      "grad_norm": 0.4388512074947357,
      "learning_rate": 8.437542943520683e-07,
      "loss": 0.6487,
      "step": 6976
    },
    {
      "epoch": 1.9166323307238016,
      "grad_norm": 0.4265596866607666,
      "learning_rate": 8.410059090284458e-07,
      "loss": 0.8136,
      "step": 6977
    },
    {
      "epoch": 1.9169070182667216,
      "grad_norm": 0.41089150309562683,
      "learning_rate": 8.382575237048235e-07,
      "loss": 0.8154,
      "step": 6978
    },
    {
      "epoch": 1.9171817058096416,
      "grad_norm": 0.40169423818588257,
      "learning_rate": 8.355091383812011e-07,
      "loss": 0.6869,
      "step": 6979
    },
    {
      "epoch": 1.9174563933525615,
      "grad_norm": 0.4899655282497406,
      "learning_rate": 8.327607530575788e-07,
      "loss": 0.8522,
      "step": 6980
    },
    {
      "epoch": 1.9177310808954813,
      "grad_norm": 0.47718754410743713,
      "learning_rate": 8.300123677339563e-07,
      "loss": 0.7006,
      "step": 6981
    },
    {
      "epoch": 1.9180057684384013,
      "grad_norm": 0.4294368028640747,
      "learning_rate": 8.27263982410334e-07,
      "loss": 0.6097,
      "step": 6982
    },
    {
      "epoch": 1.9182804559813214,
      "grad_norm": 0.4780611991882324,
      "learning_rate": 8.245155970867117e-07,
      "loss": 1.0183,
      "step": 6983
    },
    {
      "epoch": 1.9185551435242412,
      "grad_norm": 0.5445194244384766,
      "learning_rate": 8.217672117630893e-07,
      "loss": 0.8214,
      "step": 6984
    },
    {
      "epoch": 1.918829831067161,
      "grad_norm": 0.46899908781051636,
      "learning_rate": 8.190188264394669e-07,
      "loss": 0.8871,
      "step": 6985
    },
    {
      "epoch": 1.919104518610081,
      "grad_norm": 0.5337770581245422,
      "learning_rate": 8.162704411158445e-07,
      "loss": 0.6665,
      "step": 6986
    },
    {
      "epoch": 1.919379206153001,
      "grad_norm": 0.5659521222114563,
      "learning_rate": 8.135220557922222e-07,
      "loss": 0.697,
      "step": 6987
    },
    {
      "epoch": 1.9196538936959209,
      "grad_norm": 0.4505114257335663,
      "learning_rate": 8.107736704685998e-07,
      "loss": 0.7172,
      "step": 6988
    },
    {
      "epoch": 1.9199285812388407,
      "grad_norm": 0.40043601393699646,
      "learning_rate": 8.080252851449774e-07,
      "loss": 0.4789,
      "step": 6989
    },
    {
      "epoch": 1.9202032687817607,
      "grad_norm": 0.3594653904438019,
      "learning_rate": 8.05276899821355e-07,
      "loss": 0.7733,
      "step": 6990
    },
    {
      "epoch": 1.9204779563246808,
      "grad_norm": 0.4126918613910675,
      "learning_rate": 8.025285144977327e-07,
      "loss": 0.5933,
      "step": 6991
    },
    {
      "epoch": 1.9207526438676006,
      "grad_norm": 0.43908748030662537,
      "learning_rate": 7.997801291741103e-07,
      "loss": 0.8451,
      "step": 6992
    },
    {
      "epoch": 1.9210273314105204,
      "grad_norm": 0.4527234137058258,
      "learning_rate": 7.970317438504879e-07,
      "loss": 0.933,
      "step": 6993
    },
    {
      "epoch": 1.9213020189534404,
      "grad_norm": 0.40892964601516724,
      "learning_rate": 7.942833585268656e-07,
      "loss": 0.8313,
      "step": 6994
    },
    {
      "epoch": 1.9215767064963605,
      "grad_norm": 0.4950719475746155,
      "learning_rate": 7.915349732032432e-07,
      "loss": 0.8131,
      "step": 6995
    },
    {
      "epoch": 1.9218513940392803,
      "grad_norm": 0.3998623490333557,
      "learning_rate": 7.887865878796209e-07,
      "loss": 0.7611,
      "step": 6996
    },
    {
      "epoch": 1.9221260815822,
      "grad_norm": 0.578318178653717,
      "learning_rate": 7.860382025559984e-07,
      "loss": 0.8103,
      "step": 6997
    },
    {
      "epoch": 1.9224007691251201,
      "grad_norm": 0.4585724174976349,
      "learning_rate": 7.832898172323761e-07,
      "loss": 0.7725,
      "step": 6998
    },
    {
      "epoch": 1.9226754566680402,
      "grad_norm": 0.7237281203269958,
      "learning_rate": 7.805414319087537e-07,
      "loss": 0.7981,
      "step": 6999
    },
    {
      "epoch": 1.92295014421096,
      "grad_norm": 0.483789324760437,
      "learning_rate": 7.777930465851313e-07,
      "loss": 0.7403,
      "step": 7000
    }
  ],
  "logging_steps": 1,
  "max_steps": 7282,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.6105327780775678e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}