{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5090353779587682,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.18668322684139083,
      "learning_rate": 0.0,
      "loss": 1.8394,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.20650988735469475,
      "learning_rate": 0.0001,
      "loss": 2.3141,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.22046919046056535,
      "learning_rate": 0.0001,
      "loss": 2.1404,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.23873050737446988,
      "learning_rate": 9.997453526865293e-05,
      "loss": 2.1978,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2081567472868006,
      "learning_rate": 9.994907053730583e-05,
      "loss": 1.797,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1886675070025996,
      "learning_rate": 9.992360580595875e-05,
      "loss": 1.6845,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.1767301801295483,
      "learning_rate": 9.989814107461167e-05,
      "loss": 2.0519,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.22052457220349336,
      "learning_rate": 9.987267634326458e-05,
      "loss": 1.9004,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2561394854396737,
      "learning_rate": 9.98472116119175e-05,
      "loss": 2.2788,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2309558457067922,
      "learning_rate": 9.982174688057042e-05,
      "loss": 2.036,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2795391365887652,
      "learning_rate": 9.979628214922333e-05,
      "loss": 1.9411,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2718455156145512,
      "learning_rate": 9.977081741787625e-05,
      "loss": 1.8239,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.24375797038618635,
      "learning_rate": 9.974535268652915e-05,
      "loss": 1.7713,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.3212734322963607,
      "learning_rate": 9.971988795518208e-05,
      "loss": 2.0731,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2949670415991085,
      "learning_rate": 9.9694423223835e-05,
      "loss": 1.7618,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.26650378771339317,
      "learning_rate": 9.96689584924879e-05,
      "loss": 2.0073,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.2303602561325778,
      "learning_rate": 9.964349376114083e-05,
      "loss": 1.8402,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.19071474503273808,
      "learning_rate": 9.961802902979374e-05,
      "loss": 1.6371,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.21247058517565695,
      "learning_rate": 9.959256429844665e-05,
      "loss": 1.7291,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.29788490484004526,
      "learning_rate": 9.956709956709958e-05,
      "loss": 2.1516,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24952736523685068,
      "learning_rate": 9.954163483575249e-05,
      "loss": 1.8062,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24339158164984467,
      "learning_rate": 9.95161701044054e-05,
      "loss": 2.0696,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24682448902556342,
      "learning_rate": 9.949070537305833e-05,
      "loss": 1.8974,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2520444837431572,
      "learning_rate": 9.946524064171123e-05,
      "loss": 1.9982,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21667345503265834,
      "learning_rate": 9.943977591036415e-05,
      "loss": 1.8855,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20205030040213595,
      "learning_rate": 9.941431117901706e-05,
      "loss": 1.6719,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.25663378051735036,
      "learning_rate": 9.938884644766998e-05,
      "loss": 1.8823,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24494613515141275,
      "learning_rate": 9.93633817163229e-05,
      "loss": 1.7899,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22728921000088043,
      "learning_rate": 9.933791698497581e-05,
      "loss": 2.0576,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20033400778675428,
      "learning_rate": 9.931245225362873e-05,
      "loss": 1.8655,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2458925257086536,
      "learning_rate": 9.928698752228165e-05,
      "loss": 1.9585,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1861874815322032,
      "learning_rate": 9.926152279093456e-05,
      "loss": 1.9062,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2199191542164078,
      "learning_rate": 9.923605805958748e-05,
      "loss": 2.0163,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2168328882496767,
      "learning_rate": 9.921059332824039e-05,
      "loss": 1.9842,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21647678836894843,
      "learning_rate": 9.91851285968933e-05,
      "loss": 1.7001,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20657337587150454,
      "learning_rate": 9.915966386554623e-05,
      "loss": 1.9808,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22285460119825284,
      "learning_rate": 9.913419913419914e-05,
      "loss": 2.1209,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2037563233184191,
      "learning_rate": 9.910873440285205e-05,
      "loss": 1.6725,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21991517343016573,
      "learning_rate": 9.908326967150498e-05,
      "loss": 2.0868,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21562313548954337,
      "learning_rate": 9.905780494015789e-05,
      "loss": 2.0301,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.26596374509967907,
      "learning_rate": 9.90323402088108e-05,
      "loss": 2.0505,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.24973537924167893,
      "learning_rate": 9.900687547746372e-05,
      "loss": 2.4615,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21510668184289308,
      "learning_rate": 9.898141074611664e-05,
      "loss": 1.7368,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1723812623411351,
      "learning_rate": 9.895594601476955e-05,
      "loss": 1.6354,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.22746169121427556,
      "learning_rate": 9.893048128342246e-05,
      "loss": 2.1088,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.27176215691032246,
      "learning_rate": 9.890501655207539e-05,
      "loss": 2.0887,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2113572994267603,
      "learning_rate": 9.88795518207283e-05,
      "loss": 1.8923,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2311921082666697,
      "learning_rate": 9.885408708938121e-05,
      "loss": 2.0053,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.21631334700001129,
      "learning_rate": 9.882862235803412e-05,
      "loss": 1.8493,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2131940009966438,
      "learning_rate": 9.880315762668705e-05,
      "loss": 1.8507,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.26949683259896795,
      "learning_rate": 9.877769289533996e-05,
      "loss": 1.9211,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.20666923901728515,
      "learning_rate": 9.875222816399287e-05,
      "loss": 1.7817,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.17479038712810302,
      "learning_rate": 9.872676343264578e-05,
      "loss": 1.7987,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.26824054724737734,
      "learning_rate": 9.870129870129871e-05,
      "loss": 1.9868,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2544263940578613,
      "learning_rate": 9.867583396995162e-05,
      "loss": 2.0314,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2806195637276212,
      "learning_rate": 9.865036923860453e-05,
      "loss": 2.0125,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.19544621658649786,
      "learning_rate": 9.862490450725746e-05,
      "loss": 1.7809,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2504147724761243,
      "learning_rate": 9.859943977591037e-05,
      "loss": 2.1766,
      "step": 58
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22152374260830437,
      "learning_rate": 9.857397504456328e-05,
      "loss": 2.0083,
      "step": 59
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20102838380117335,
      "learning_rate": 9.85485103132162e-05,
      "loss": 1.9817,
      "step": 60
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21677533463700935,
      "learning_rate": 9.852304558186911e-05,
      "loss": 1.9689,
      "step": 61
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20905116609320215,
      "learning_rate": 9.849758085052203e-05,
      "loss": 2.0173,
      "step": 62
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21550938056512234,
      "learning_rate": 9.847211611917495e-05,
      "loss": 2.1043,
      "step": 63
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23724001721462154,
      "learning_rate": 9.844665138782786e-05,
      "loss": 2.0639,
      "step": 64
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2201372504716533,
      "learning_rate": 9.842118665648078e-05,
      "loss": 2.085,
      "step": 65
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2355750751853312,
      "learning_rate": 9.83957219251337e-05,
      "loss": 1.9968,
      "step": 66
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.17637378642261511,
      "learning_rate": 9.837025719378661e-05,
      "loss": 1.771,
      "step": 67
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.25906351758805374,
      "learning_rate": 9.834479246243953e-05,
      "loss": 2.2,
      "step": 68
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2275259890095465,
      "learning_rate": 9.831932773109243e-05,
      "loss": 1.8159,
      "step": 69
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1980881491845839,
      "learning_rate": 9.829386299974536e-05,
      "loss": 1.9643,
      "step": 70
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2058717539357096,
      "learning_rate": 9.826839826839827e-05,
      "loss": 2.1604,
      "step": 71
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2143307195813568,
      "learning_rate": 9.824293353705118e-05,
      "loss": 1.7573,
      "step": 72
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23338330590904638,
      "learning_rate": 9.821746880570411e-05,
      "loss": 2.0012,
      "step": 73
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.27815596429569944,
      "learning_rate": 9.819200407435702e-05,
      "loss": 1.9551,
      "step": 74
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22362637406455313,
      "learning_rate": 9.816653934300993e-05,
      "loss": 1.9162,
      "step": 75
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21896329766757727,
      "learning_rate": 9.814107461166286e-05,
      "loss": 2.0692,
      "step": 76
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22342664885681365,
      "learning_rate": 9.811560988031576e-05,
      "loss": 1.8841,
      "step": 77
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22785719541652857,
      "learning_rate": 9.809014514896868e-05,
      "loss": 1.9784,
      "step": 78
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.26447957975744196,
      "learning_rate": 9.806468041762161e-05,
      "loss": 1.7973,
      "step": 79
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.27247251807318934,
      "learning_rate": 9.80392156862745e-05,
      "loss": 1.9147,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20979533207130613,
      "learning_rate": 9.801375095492743e-05,
      "loss": 1.9589,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20882101001884892,
      "learning_rate": 9.798828622358034e-05,
      "loss": 1.8915,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2303846254211127,
      "learning_rate": 9.796282149223326e-05,
      "loss": 1.8688,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.195074823723581,
      "learning_rate": 9.793735676088618e-05,
      "loss": 1.9336,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21544466079041769,
      "learning_rate": 9.791189202953909e-05,
      "loss": 1.9476,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21979704680042544,
      "learning_rate": 9.7886427298192e-05,
      "loss": 1.8772,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2468756032888067,
      "learning_rate": 9.786096256684493e-05,
      "loss": 1.9661,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21893561526354924,
      "learning_rate": 9.783549783549783e-05,
      "loss": 1.8254,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.22456850426384328,
      "learning_rate": 9.781003310415076e-05,
      "loss": 1.9367,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21911173951413468,
      "learning_rate": 9.778456837280368e-05,
      "loss": 1.823,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.18737492760853453,
      "learning_rate": 9.775910364145658e-05,
      "loss": 1.7855,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23796009872049736,
      "learning_rate": 9.77336389101095e-05,
      "loss": 2.1084,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.24563292848492027,
      "learning_rate": 9.770817417876243e-05,
      "loss": 2.1699,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.20949684127939133,
      "learning_rate": 9.768270944741533e-05,
      "loss": 2.1912,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.19708490119220384,
      "learning_rate": 9.765724471606825e-05,
      "loss": 1.4705,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2343314289283105,
      "learning_rate": 9.763177998472117e-05,
      "loss": 1.696,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21503323955083936,
      "learning_rate": 9.760631525337408e-05,
      "loss": 1.8542,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2317012546697141,
      "learning_rate": 9.7580850522027e-05,
      "loss": 1.7464,
      "step": 98
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20571972527548762,
      "learning_rate": 9.75553857906799e-05,
      "loss": 1.8839,
      "step": 99
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.26962345039017876,
      "learning_rate": 9.752992105933283e-05,
      "loss": 2.1152,
      "step": 100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21136096551098874,
      "learning_rate": 9.750445632798575e-05,
      "loss": 1.794,
      "step": 101
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2624788065710887,
      "learning_rate": 9.747899159663865e-05,
      "loss": 2.0013,
      "step": 102
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20572013271793682,
      "learning_rate": 9.745352686529158e-05,
      "loss": 2.051,
      "step": 103
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2034459055059183,
      "learning_rate": 9.742806213394449e-05,
      "loss": 1.8117,
      "step": 104
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.18901260521522967,
      "learning_rate": 9.74025974025974e-05,
      "loss": 1.8273,
      "step": 105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23189693763581237,
      "learning_rate": 9.737713267125033e-05,
      "loss": 1.762,
      "step": 106
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22780526433581438,
      "learning_rate": 9.735166793990324e-05,
      "loss": 1.7569,
      "step": 107
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.18899027335403254,
      "learning_rate": 9.732620320855615e-05,
      "loss": 1.7129,
      "step": 108
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19854782296756943,
      "learning_rate": 9.730073847720908e-05,
      "loss": 1.9007,
      "step": 109
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.20392060357214004,
      "learning_rate": 9.727527374586199e-05,
      "loss": 1.6141,
      "step": 110
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2378622341910083,
      "learning_rate": 9.72498090145149e-05,
      "loss": 1.9192,
      "step": 111
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2833162936645229,
      "learning_rate": 9.722434428316781e-05,
      "loss": 1.9424,
      "step": 112
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.195262101346659,
      "learning_rate": 9.719887955182073e-05,
      "loss": 1.8432,
      "step": 113
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.18626690744618216,
      "learning_rate": 9.717341482047365e-05,
      "loss": 1.7522,
      "step": 114
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.25144789555295355,
      "learning_rate": 9.714795008912656e-05,
      "loss": 2.1025,
      "step": 115
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2507526720110009,
      "learning_rate": 9.712248535777948e-05,
      "loss": 1.9353,
      "step": 116
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2398695928710626,
      "learning_rate": 9.70970206264324e-05,
      "loss": 1.7189,
      "step": 117
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2049264892192432,
      "learning_rate": 9.707155589508531e-05,
      "loss": 1.81,
      "step": 118
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21239921304434395,
      "learning_rate": 9.704609116373823e-05,
      "loss": 1.8867,
      "step": 119
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.19948118352409933,
      "learning_rate": 9.702062643239115e-05,
      "loss": 1.7714,
      "step": 120
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23275057205842845,
      "learning_rate": 9.699516170104406e-05,
      "loss": 2.0321,
      "step": 121
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.25282056951198856,
      "learning_rate": 9.696969696969698e-05,
      "loss": 2.1878,
      "step": 122
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.22077724596241188,
      "learning_rate": 9.694423223834989e-05,
      "loss": 1.7711,
      "step": 123
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24358276719249283,
      "learning_rate": 9.69187675070028e-05,
      "loss": 1.7919,
      "step": 124
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23687386356473691,
      "learning_rate": 9.689330277565573e-05,
      "loss": 1.8732,
      "step": 125
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1961942510175504,
      "learning_rate": 9.686783804430864e-05,
      "loss": 1.8143,
      "step": 126
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24989519306549782,
      "learning_rate": 9.684237331296155e-05,
      "loss": 1.9983,
      "step": 127
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.18367867838900498,
      "learning_rate": 9.681690858161448e-05,
      "loss": 1.8044,
      "step": 128
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1851171637376302,
      "learning_rate": 9.679144385026739e-05,
      "loss": 1.5508,
      "step": 129
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1981646287407315,
      "learning_rate": 9.67659791189203e-05,
      "loss": 1.6866,
      "step": 130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.25168898879511503,
      "learning_rate": 9.674051438757321e-05,
      "loss": 1.9691,
      "step": 131
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2406960292737266,
      "learning_rate": 9.671504965622614e-05,
      "loss": 1.8737,
      "step": 132
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23760573204480365,
      "learning_rate": 9.668958492487905e-05,
      "loss": 2.0618,
      "step": 133
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.21245060514928146,
      "learning_rate": 9.666412019353196e-05,
      "loss": 1.7938,
      "step": 134
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.23616623653756544,
      "learning_rate": 9.663865546218487e-05,
      "loss": 2.2237,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.24385343826643593,
      "learning_rate": 9.66131907308378e-05,
      "loss": 2.1726,
      "step": 136
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.2086523305951323,
      "learning_rate": 9.658772599949071e-05,
      "loss": 2.0386,
      "step": 137
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2209783795615669,
      "learning_rate": 9.656226126814362e-05,
      "loss": 1.8839,
      "step": 138
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20788942469293714,
      "learning_rate": 9.653679653679654e-05,
      "loss": 1.7717,
      "step": 139
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24552151589399218,
      "learning_rate": 9.651133180544946e-05,
      "loss": 1.9754,
      "step": 140
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2176532927132165,
      "learning_rate": 9.648586707410237e-05,
      "loss": 2.0004,
      "step": 141
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23477900176569336,
      "learning_rate": 9.646040234275528e-05,
      "loss": 2.0929,
      "step": 142
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2599706772628387,
      "learning_rate": 9.643493761140821e-05,
      "loss": 1.9682,
      "step": 143
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2654960824368493,
      "learning_rate": 9.640947288006112e-05,
      "loss": 1.927,
      "step": 144
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.20820691425532545,
      "learning_rate": 9.638400814871403e-05,
      "loss": 1.8643,
      "step": 145
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.25697305594483716,
      "learning_rate": 9.635854341736695e-05,
      "loss": 1.9165,
      "step": 146
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2342038483361337,
      "learning_rate": 9.633307868601986e-05,
      "loss": 1.8247,
      "step": 147
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22760589993222088,
      "learning_rate": 9.630761395467278e-05,
      "loss": 1.7226,
      "step": 148
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24795127888830662,
      "learning_rate": 9.62821492233257e-05,
      "loss": 1.926,
      "step": 149
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21953914702795194,
      "learning_rate": 9.625668449197861e-05,
      "loss": 1.7582,
      "step": 150
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2197624516317337,
      "learning_rate": 9.623121976063153e-05,
      "loss": 1.656,
      "step": 151
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2309760958845721,
      "learning_rate": 9.620575502928445e-05,
      "loss": 1.8533,
      "step": 152
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2568432331862067,
      "learning_rate": 9.618029029793736e-05,
      "loss": 2.1763,
      "step": 153
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24345830556671422,
      "learning_rate": 9.615482556659028e-05,
      "loss": 1.9257,
      "step": 154
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22500400705213244,
      "learning_rate": 9.612936083524318e-05,
      "loss": 1.7517,
      "step": 155
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2244433577284875,
      "learning_rate": 9.610389610389611e-05,
      "loss": 1.8401,
      "step": 156
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2966025758924672,
      "learning_rate": 9.607843137254903e-05,
      "loss": 2.049,
      "step": 157
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2554714815186226,
      "learning_rate": 9.605296664120193e-05,
      "loss": 1.9473,
      "step": 158
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2942129291965485,
      "learning_rate": 9.602750190985486e-05,
      "loss": 2.2011,
      "step": 159
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2525935056081843,
      "learning_rate": 9.600203717850777e-05,
      "loss": 1.8666,
      "step": 160
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.28258844953158013,
      "learning_rate": 9.597657244716068e-05,
      "loss": 1.9187,
      "step": 161
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.24764972341222816,
      "learning_rate": 9.595110771581361e-05,
      "loss": 1.9439,
      "step": 162
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22459914142681472,
      "learning_rate": 9.592564298446652e-05,
      "loss": 2.0468,
      "step": 163
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2383724960346664,
      "learning_rate": 9.590017825311943e-05,
      "loss": 1.8381,
      "step": 164
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2579963779731197,
      "learning_rate": 9.587471352177236e-05,
      "loss": 2.0741,
      "step": 165
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22364171605939087,
      "learning_rate": 9.584924879042526e-05,
      "loss": 1.6662,
      "step": 166
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.23417480183675768,
      "learning_rate": 9.582378405907818e-05,
      "loss": 1.7069,
      "step": 167
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.25343272730311545,
      "learning_rate": 9.579831932773111e-05,
      "loss": 1.763,
      "step": 168
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2517659726829119,
      "learning_rate": 9.5772854596384e-05,
      "loss": 2.1275,
      "step": 169
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.25888382238424834,
      "learning_rate": 9.574738986503693e-05,
      "loss": 1.7395,
      "step": 170
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.21638117276537386,
      "learning_rate": 9.572192513368984e-05,
      "loss": 1.9478,
      "step": 171
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19808260127052518,
      "learning_rate": 9.569646040234276e-05,
      "loss": 1.6967,
      "step": 172
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19907375744019035,
      "learning_rate": 9.567099567099568e-05,
      "loss": 1.7707,
      "step": 173
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2459730124614947,
      "learning_rate": 9.56455309396486e-05,
      "loss": 1.8615,
      "step": 174
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.25489061716341954,
      "learning_rate": 9.56200662083015e-05,
      "loss": 2.1669,
      "step": 175
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.22452811554198862,
      "learning_rate": 9.559460147695443e-05,
      "loss": 2.1389,
      "step": 176
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23382342125125047,
      "learning_rate": 9.556913674560733e-05,
      "loss": 1.9017,
      "step": 177
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.19184570079482094,
      "learning_rate": 9.554367201426026e-05,
      "loss": 1.586,
      "step": 178
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23668167371557072,
      "learning_rate": 9.551820728291318e-05,
      "loss": 2.0936,
      "step": 179
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20898807379756024,
      "learning_rate": 9.549274255156608e-05,
      "loss": 1.8818,
      "step": 180
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2396599251528385,
      "learning_rate": 9.5467277820219e-05,
      "loss": 2.07,
      "step": 181
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20789926230727251,
      "learning_rate": 9.544181308887192e-05,
      "loss": 1.963,
      "step": 182
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.26476631495390657,
      "learning_rate": 9.541634835752483e-05,
      "loss": 1.8445,
      "step": 183
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2273176941900955,
      "learning_rate": 9.539088362617775e-05,
      "loss": 1.9145,
      "step": 184
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23125805905191937,
      "learning_rate": 9.536541889483067e-05,
      "loss": 1.9857,
      "step": 185
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.25186137813337384,
      "learning_rate": 9.533995416348358e-05,
      "loss": 2.0566,
      "step": 186
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.30429619334858266,
      "learning_rate": 9.53144894321365e-05,
      "loss": 1.9218,
      "step": 187
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2087098395097207,
      "learning_rate": 9.52890247007894e-05,
      "loss": 2.0563,
      "step": 188
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.210832825933067,
      "learning_rate": 9.526355996944233e-05,
      "loss": 1.6114,
      "step": 189
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.26623973116973887,
      "learning_rate": 9.523809523809524e-05,
      "loss": 2.305,
      "step": 190
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23251965012881762,
      "learning_rate": 9.521263050674815e-05,
      "loss": 2.0812,
      "step": 191
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23477609011259198,
      "learning_rate": 9.518716577540108e-05,
      "loss": 1.8897,
      "step": 192
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2121841537706127,
      "learning_rate": 9.516170104405399e-05,
      "loss": 1.8104,
      "step": 193
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21000993733508805,
      "learning_rate": 9.51362363127069e-05,
      "loss": 1.761,
      "step": 194
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2412983353771996,
      "learning_rate": 9.511077158135983e-05,
      "loss": 1.9648,
      "step": 195
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21809671979675835,
      "learning_rate": 9.508530685001274e-05,
      "loss": 1.9521,
      "step": 196
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2196701570803616,
      "learning_rate": 9.505984211866565e-05,
      "loss": 1.962,
      "step": 197
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20178135884970153,
      "learning_rate": 9.503437738731858e-05,
      "loss": 1.8271,
      "step": 198
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2165889430175184,
      "learning_rate": 9.500891265597148e-05,
      "loss": 1.8558,
      "step": 199
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21204500503733067,
      "learning_rate": 9.49834479246244e-05,
      "loss": 1.8737,
      "step": 200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2448504390910478,
      "learning_rate": 9.495798319327731e-05,
      "loss": 1.7742,
      "step": 201
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24421357400135402,
      "learning_rate": 9.493251846193023e-05,
      "loss": 2.0813,
      "step": 202
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24034686672484812,
      "learning_rate": 9.490705373058315e-05,
      "loss": 1.6847,
      "step": 203
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22251871110699203,
      "learning_rate": 9.488158899923606e-05,
      "loss": 2.0008,
      "step": 204
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.20788696969656037,
      "learning_rate": 9.485612426788898e-05,
      "loss": 1.825,
      "step": 205
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24342274215879736,
      "learning_rate": 9.48306595365419e-05,
      "loss": 2.137,
      "step": 206
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21797141240999446,
      "learning_rate": 9.480519480519481e-05,
      "loss": 1.919,
      "step": 207
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.22818316214645618,
      "learning_rate": 9.477973007384773e-05,
      "loss": 1.6708,
      "step": 208
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.24104874289079398,
      "learning_rate": 9.475426534250064e-05,
      "loss": 1.8872,
      "step": 209
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.27131811031770453,
      "learning_rate": 9.472880061115355e-05,
      "loss": 2.1951,
      "step": 210
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23612631701298523,
      "learning_rate": 9.470333587980648e-05,
      "loss": 1.8685,
      "step": 211
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.25111361668915844,
      "learning_rate": 9.467787114845939e-05,
      "loss": 1.9868,
      "step": 212
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23129340711654747,
      "learning_rate": 9.46524064171123e-05,
      "loss": 1.8138,
      "step": 213
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.23683270270522913,
      "learning_rate": 9.462694168576523e-05,
      "loss": 1.9155,
      "step": 214
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2231374860523671,
      "learning_rate": 9.460147695441814e-05,
      "loss": 1.6972,
      "step": 215
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.21331695425466865,
      "learning_rate": 9.457601222307105e-05,
      "loss": 1.8439,
      "step": 216
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.21181957054324804,
      "learning_rate": 9.455054749172396e-05,
      "loss": 1.7157,
      "step": 217
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.229617283591777,
      "learning_rate": 9.452508276037689e-05,
      "loss": 2.1186,
      "step": 218
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19437555647276678,
      "learning_rate": 9.44996180290298e-05,
      "loss": 1.7184,
      "step": 219
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23963701987329625,
      "learning_rate": 9.447415329768271e-05,
      "loss": 1.8568,
      "step": 220
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22261811649298022,
      "learning_rate": 9.444868856633562e-05,
      "loss": 1.6762,
      "step": 221
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22519402085899096,
      "learning_rate": 9.442322383498855e-05,
      "loss": 1.978,
      "step": 222
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.26627643028672393,
      "learning_rate": 9.439775910364146e-05,
      "loss": 2.1728,
      "step": 223
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2309815472420271,
      "learning_rate": 9.437229437229437e-05,
      "loss": 1.9574,
      "step": 224
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23051474807252864,
      "learning_rate": 9.434682964094729e-05,
      "loss": 1.995,
      "step": 225
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.30308261653684976,
      "learning_rate": 9.432136490960021e-05,
      "loss": 2.0055,
      "step": 226
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22150569759310887,
      "learning_rate": 9.429590017825312e-05,
      "loss": 1.8448,
      "step": 227
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23139463747136307,
      "learning_rate": 9.427043544690604e-05,
      "loss": 2.0128,
      "step": 228
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22266980180155999,
      "learning_rate": 9.424497071555896e-05,
      "loss": 1.8328,
      "step": 229
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22092674539443577,
      "learning_rate": 9.421950598421187e-05,
      "loss": 1.4841,
      "step": 230
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20078682390780883,
      "learning_rate": 9.419404125286478e-05,
      "loss": 1.7392,
      "step": 231
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2692318787918702,
      "learning_rate": 9.416857652151771e-05,
      "loss": 1.8473,
      "step": 232
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22854627909015088,
      "learning_rate": 9.414311179017061e-05,
      "loss": 2.0171,
      "step": 233
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.24774161651527243,
      "learning_rate": 9.411764705882353e-05,
      "loss": 1.9486,
      "step": 234
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.24102782430332428,
      "learning_rate": 9.409218232747645e-05,
      "loss": 1.9989,
      "step": 235
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2010424021392836,
      "learning_rate": 9.406671759612936e-05,
      "loss": 1.7616,
      "step": 236
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23034735084682545,
      "learning_rate": 9.404125286478228e-05,
      "loss": 1.7533,
      "step": 237
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.21964697350073992,
      "learning_rate": 9.40157881334352e-05,
      "loss": 1.7249,
      "step": 238
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2577106823861686,
      "learning_rate": 9.399032340208811e-05,
      "loss": 2.1802,
      "step": 239
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2224282134304929,
      "learning_rate": 9.396485867074103e-05,
      "loss": 1.7209,
      "step": 240
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.25699890246231155,
      "learning_rate": 9.393939393939395e-05,
      "loss": 1.8835,
      "step": 241
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3005056604932864,
      "learning_rate": 9.391392920804686e-05,
      "loss": 2.058,
      "step": 242
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23921523980493628,
      "learning_rate": 9.388846447669978e-05,
      "loss": 1.8772,
      "step": 243
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2118256907628859,
      "learning_rate": 9.386299974535268e-05,
      "loss": 1.9704,
      "step": 244
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.20004277814897858,
      "learning_rate": 9.383753501400561e-05,
      "loss": 1.6381,
      "step": 245
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23023427738670452,
      "learning_rate": 9.381207028265852e-05,
      "loss": 1.918,
      "step": 246
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.22430066994725595,
      "learning_rate": 9.378660555131143e-05,
      "loss": 1.8846,
      "step": 247
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2543002666006967,
      "learning_rate": 9.376114081996436e-05,
      "loss": 2.1089,
      "step": 248
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2595002101959818,
      "learning_rate": 9.373567608861727e-05,
      "loss": 1.9058,
      "step": 249
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23391445529028587,
      "learning_rate": 9.371021135727018e-05,
      "loss": 1.964,
      "step": 250
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37687783156386134,
      "learning_rate": 9.368474662592311e-05,
      "loss": 1.8056,
      "step": 251
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.25918034177581145,
      "learning_rate": 9.3659281894576e-05,
      "loss": 2.3148,
      "step": 252
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.24306255351316375,
      "learning_rate": 9.363381716322893e-05,
      "loss": 1.9831,
      "step": 253
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2575646567417516,
      "learning_rate": 9.360835243188186e-05,
      "loss": 1.8934,
      "step": 254
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.23062400114675347,
      "learning_rate": 9.358288770053476e-05,
      "loss": 2.0416,
      "step": 255
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24386803477315933,
      "learning_rate": 9.355742296918768e-05,
      "loss": 1.7916,
      "step": 256
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21974957663492034,
      "learning_rate": 9.35319582378406e-05,
      "loss": 1.9653,
      "step": 257
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21492461503340388,
      "learning_rate": 9.35064935064935e-05,
      "loss": 1.7282,
      "step": 258
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22698493741125245,
      "learning_rate": 9.348102877514643e-05,
      "loss": 1.8432,
      "step": 259
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20285954001886503,
      "learning_rate": 9.345556404379934e-05,
      "loss": 1.7529,
      "step": 260
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22456986452974662,
      "learning_rate": 9.343009931245226e-05,
      "loss": 1.8794,
      "step": 261
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22394667177371683,
      "learning_rate": 9.340463458110518e-05,
      "loss": 1.8886,
      "step": 262
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22092161082335537,
      "learning_rate": 9.337916984975808e-05,
      "loss": 2.0836,
      "step": 263
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20578946766995015,
      "learning_rate": 9.3353705118411e-05,
      "loss": 1.5635,
      "step": 264
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.20675349046696734,
      "learning_rate": 9.332824038706393e-05,
      "loss": 1.9682,
      "step": 265
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.23204721980255197,
      "learning_rate": 9.330277565571683e-05,
      "loss": 2.1915,
      "step": 266
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2634111457529946,
      "learning_rate": 9.327731092436976e-05,
      "loss": 1.7796,
      "step": 267
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2743533600157014,
      "learning_rate": 9.325184619302267e-05,
      "loss": 1.6329,
      "step": 268
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2420317702906553,
      "learning_rate": 9.322638146167558e-05,
      "loss": 1.6665,
      "step": 269
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.19690034112871688,
      "learning_rate": 9.32009167303285e-05,
      "loss": 1.8188,
      "step": 270
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2416787341135894,
      "learning_rate": 9.317545199898142e-05,
      "loss": 1.9407,
      "step": 271
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21190598468063523,
      "learning_rate": 9.314998726763433e-05,
      "loss": 1.8868,
      "step": 272
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24950943440028028,
      "learning_rate": 9.312452253628725e-05,
      "loss": 1.9476,
      "step": 273
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21485197745088533,
      "learning_rate": 9.309905780494015e-05,
      "loss": 1.6875,
      "step": 274
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2398399666478477,
      "learning_rate": 9.307359307359308e-05,
      "loss": 1.8274,
      "step": 275
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24735280047478256,
      "learning_rate": 9.3048128342246e-05,
      "loss": 2.009,
      "step": 276
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2631694952704255,
      "learning_rate": 9.30226636108989e-05,
      "loss": 1.97,
      "step": 277
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.19990147682229312,
      "learning_rate": 9.299719887955183e-05,
      "loss": 1.5918,
      "step": 278
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.19055219768739767,
      "learning_rate": 9.297173414820474e-05,
      "loss": 1.5943,
      "step": 279
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.23155827723062083,
      "learning_rate": 9.294626941685765e-05,
      "loss": 1.7937,
      "step": 280
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2576038643469748,
      "learning_rate": 9.292080468551058e-05,
      "loss": 2.0919,
      "step": 281
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22399345203923643,
      "learning_rate": 9.289533995416349e-05,
      "loss": 1.6301,
      "step": 282
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.22107518323709327,
      "learning_rate": 9.28698752228164e-05,
      "loss": 1.6966,
      "step": 283
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2091183455822762,
      "learning_rate": 9.284441049146933e-05,
      "loss": 1.8305,
      "step": 284
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.21341806222699425,
      "learning_rate": 9.281894576012223e-05,
      "loss": 1.7999,
      "step": 285
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2196306400731007,
      "learning_rate": 9.279348102877515e-05,
      "loss": 1.9407,
      "step": 286
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24222282951836133,
      "learning_rate": 9.276801629742806e-05,
      "loss": 1.7496,
      "step": 287
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2120531741834089,
      "learning_rate": 9.274255156608098e-05,
      "loss": 1.4856,
      "step": 288
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.25255465595815,
      "learning_rate": 9.27170868347339e-05,
      "loss": 1.9091,
      "step": 289
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.23932122149332355,
      "learning_rate": 9.269162210338681e-05,
      "loss": 1.9319,
      "step": 290
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2251063529297804,
      "learning_rate": 9.266615737203973e-05,
      "loss": 1.8085,
      "step": 291
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.24473398201937743,
      "learning_rate": 9.264069264069265e-05,
      "loss": 2.005,
      "step": 292
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2465717275751802,
      "learning_rate": 9.261522790934556e-05,
      "loss": 2.1889,
      "step": 293
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2842685521193732,
      "learning_rate": 9.258976317799848e-05,
      "loss": 1.9895,
      "step": 294
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2375850983047579,
      "learning_rate": 9.256429844665139e-05,
      "loss": 1.8794,
      "step": 295
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22515910503528352,
      "learning_rate": 9.253883371530431e-05,
      "loss": 1.8471,
      "step": 296
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.20256419894910435,
      "learning_rate": 9.251336898395723e-05,
      "loss": 1.9121,
      "step": 297
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19414163585376648,
      "learning_rate": 9.248790425261014e-05,
      "loss": 1.754,
      "step": 298
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2590600520249542,
      "learning_rate": 9.246243952126305e-05,
      "loss": 1.9198,
      "step": 299
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2186561110505484,
      "learning_rate": 9.243697478991598e-05,
      "loss": 1.8622,
      "step": 300
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2325676052442486,
      "learning_rate": 9.241151005856889e-05,
      "loss": 2.038,
      "step": 301
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3049684598514628,
      "learning_rate": 9.23860453272218e-05,
      "loss": 2.2992,
      "step": 302
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2632339372769938,
      "learning_rate": 9.236058059587471e-05,
      "loss": 1.9476,
      "step": 303
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21988150326001385,
      "learning_rate": 9.233511586452764e-05,
      "loss": 1.8327,
      "step": 304
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2642056043525351,
      "learning_rate": 9.230965113318055e-05,
      "loss": 1.8121,
      "step": 305
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2752051032919026,
      "learning_rate": 9.228418640183346e-05,
      "loss": 1.9429,
      "step": 306
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23849414476318845,
      "learning_rate": 9.225872167048639e-05,
      "loss": 1.858,
      "step": 307
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.264730461894566,
      "learning_rate": 9.22332569391393e-05,
      "loss": 2.2006,
      "step": 308
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23811792028971387,
      "learning_rate": 9.220779220779221e-05,
      "loss": 2.0294,
      "step": 309
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24963177444652385,
      "learning_rate": 9.218232747644512e-05,
      "loss": 1.8812,
      "step": 310
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21743851296807606,
      "learning_rate": 9.215686274509804e-05,
      "loss": 1.8752,
      "step": 311
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.207702098647325,
      "learning_rate": 9.213139801375096e-05,
      "loss": 1.8431,
      "step": 312
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2536897469848907,
      "learning_rate": 9.210593328240387e-05,
      "loss": 2.344,
      "step": 313
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2527882537720398,
      "learning_rate": 9.208046855105679e-05,
      "loss": 2.1259,
      "step": 314
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22217569861745942,
      "learning_rate": 9.205500381970971e-05,
      "loss": 1.9742,
      "step": 315
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23149605689696912,
      "learning_rate": 9.202953908836262e-05,
      "loss": 1.8699,
      "step": 316
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2558552399512293,
      "learning_rate": 9.200407435701554e-05,
      "loss": 2.0121,
      "step": 317
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2815150495507308,
      "learning_rate": 9.197860962566846e-05,
      "loss": 1.8827,
      "step": 318
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24109724971829488,
      "learning_rate": 9.195314489432137e-05,
      "loss": 2.034,
      "step": 319
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21261453662280155,
      "learning_rate": 9.192768016297429e-05,
      "loss": 1.5434,
      "step": 320
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23790910458001532,
      "learning_rate": 9.19022154316272e-05,
      "loss": 1.9696,
      "step": 321
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.21656851727106363,
      "learning_rate": 9.187675070028011e-05,
      "loss": 1.7906,
      "step": 322
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24685226589261666,
      "learning_rate": 9.185128596893303e-05,
      "loss": 2.0424,
      "step": 323
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22259924811476584,
      "learning_rate": 9.182582123758595e-05,
      "loss": 1.8244,
      "step": 324
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2196586079950527,
      "learning_rate": 9.180035650623886e-05,
      "loss": 1.8852,
      "step": 325
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24882911130988142,
      "learning_rate": 9.177489177489178e-05,
      "loss": 2.1478,
      "step": 326
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.26072995732231113,
      "learning_rate": 9.17494270435447e-05,
      "loss": 1.7825,
      "step": 327
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.22901788762404968,
      "learning_rate": 9.172396231219761e-05,
      "loss": 2.0263,
      "step": 328
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3027309786698346,
      "learning_rate": 9.169849758085053e-05,
      "loss": 1.6272,
      "step": 329
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2426624716595467,
      "learning_rate": 9.167303284950343e-05,
      "loss": 1.8666,
      "step": 330
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.30318419969720933,
      "learning_rate": 9.164756811815636e-05,
      "loss": 1.8957,
      "step": 331
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2620787351070465,
      "learning_rate": 9.162210338680927e-05,
      "loss": 1.9753,
      "step": 332
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.23925396663143897,
      "learning_rate": 9.159663865546218e-05,
      "loss": 1.8427,
      "step": 333
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.25543502386679007,
      "learning_rate": 9.157117392411511e-05,
      "loss": 2.0556,
      "step": 334
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.25514597844638615,
      "learning_rate": 9.154570919276802e-05,
      "loss": 1.8451,
      "step": 335
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2341678180502237,
      "learning_rate": 9.152024446142093e-05,
      "loss": 1.9502,
      "step": 336
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2529748559630748,
      "learning_rate": 9.149477973007386e-05,
      "loss": 1.9665,
      "step": 337
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.26268342852744336,
      "learning_rate": 9.146931499872676e-05,
      "loss": 2.045,
      "step": 338
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22052807744368427,
      "learning_rate": 9.144385026737968e-05,
      "loss": 1.8048,
      "step": 339
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2763289570061149,
      "learning_rate": 9.141838553603261e-05,
      "loss": 2.1017,
      "step": 340
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3086075236469837,
      "learning_rate": 9.13929208046855e-05,
      "loss": 1.8693,
      "step": 341
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24003803945126273,
      "learning_rate": 9.136745607333843e-05,
      "loss": 1.8391,
      "step": 342
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23133864540209823,
      "learning_rate": 9.134199134199136e-05,
      "loss": 1.9012,
      "step": 343
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2897567659629809,
      "learning_rate": 9.131652661064426e-05,
      "loss": 1.831,
      "step": 344
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22979690936728908,
      "learning_rate": 9.129106187929718e-05,
      "loss": 1.7455,
      "step": 345
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24403021795538454,
      "learning_rate": 9.12655971479501e-05,
      "loss": 1.7353,
      "step": 346
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21211903411341312,
      "learning_rate": 9.1240132416603e-05,
      "loss": 1.9091,
      "step": 347
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2465935581647895,
      "learning_rate": 9.121466768525593e-05,
      "loss": 1.9653,
      "step": 348
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22337100266582047,
      "learning_rate": 9.118920295390883e-05,
      "loss": 1.9857,
      "step": 349
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19762650141546803,
      "learning_rate": 9.116373822256176e-05,
      "loss": 1.6218,
      "step": 350
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.27535143363690945,
      "learning_rate": 9.113827349121468e-05,
      "loss": 1.8376,
      "step": 351
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24252525742486392,
      "learning_rate": 9.111280875986758e-05,
      "loss": 1.8176,
      "step": 352
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23695886790494372,
      "learning_rate": 9.10873440285205e-05,
      "loss": 1.7958,
      "step": 353
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21746293413915188,
      "learning_rate": 9.106187929717343e-05,
      "loss": 1.667,
      "step": 354
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3088650054487904,
      "learning_rate": 9.103641456582633e-05,
      "loss": 1.9392,
      "step": 355
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1965171096562478,
      "learning_rate": 9.101094983447926e-05,
      "loss": 1.7628,
      "step": 356
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2238052486293197,
      "learning_rate": 9.098548510313217e-05,
      "loss": 1.8061,
      "step": 357
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23181591059915962,
      "learning_rate": 9.096002037178508e-05,
      "loss": 1.9798,
      "step": 358
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2392610666481906,
      "learning_rate": 9.0934555640438e-05,
      "loss": 1.9227,
      "step": 359
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.250920285695026,
      "learning_rate": 9.090909090909092e-05,
      "loss": 2.0199,
      "step": 360
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.20542773464715974,
      "learning_rate": 9.088362617774383e-05,
      "loss": 2.0731,
      "step": 361
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19969186515394957,
      "learning_rate": 9.085816144639675e-05,
      "loss": 1.8649,
      "step": 362
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21939794762728568,
      "learning_rate": 9.083269671504965e-05,
      "loss": 1.9547,
      "step": 363
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18498324065396723,
      "learning_rate": 9.080723198370258e-05,
      "loss": 1.6511,
      "step": 364
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2395442415908442,
      "learning_rate": 9.078176725235549e-05,
      "loss": 2.0866,
      "step": 365
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2867061222424313,
      "learning_rate": 9.07563025210084e-05,
      "loss": 2.1337,
      "step": 366
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24312892338202968,
      "learning_rate": 9.073083778966133e-05,
      "loss": 2.0018,
      "step": 367
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.21800031946097598,
      "learning_rate": 9.070537305831424e-05,
      "loss": 1.7939,
      "step": 368
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22455248734256647,
      "learning_rate": 9.067990832696715e-05,
      "loss": 1.8707,
      "step": 369
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.19183669055751298,
      "learning_rate": 9.065444359562008e-05,
      "loss": 1.7357,
      "step": 370
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22691866404601918,
      "learning_rate": 9.062897886427299e-05,
      "loss": 1.7986,
      "step": 371
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.30994248506537475,
      "learning_rate": 9.06035141329259e-05,
      "loss": 2.1003,
      "step": 372
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.22813953523184213,
      "learning_rate": 9.057804940157881e-05,
      "loss": 1.9194,
      "step": 373
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23784922692149396,
      "learning_rate": 9.055258467023173e-05,
      "loss": 2.0995,
      "step": 374
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25676292571039067,
      "learning_rate": 9.052711993888465e-05,
      "loss": 2.1612,
      "step": 375
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25134486326781896,
      "learning_rate": 9.050165520753756e-05,
      "loss": 2.0511,
      "step": 376
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.24854067963044174,
      "learning_rate": 9.047619047619048e-05,
      "loss": 1.9804,
      "step": 377
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23139343807085166,
      "learning_rate": 9.04507257448434e-05,
      "loss": 1.5953,
      "step": 378
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.19988237438709283,
      "learning_rate": 9.042526101349631e-05,
      "loss": 1.7571,
      "step": 379
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23942742077516901,
      "learning_rate": 9.039979628214923e-05,
      "loss": 2.0776,
      "step": 380
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2313486291648339,
      "learning_rate": 9.037433155080214e-05,
      "loss": 1.6489,
      "step": 381
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21412353306850734,
      "learning_rate": 9.034886681945506e-05,
      "loss": 1.7505,
      "step": 382
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2854456804430116,
      "learning_rate": 9.032340208810798e-05,
      "loss": 2.1109,
      "step": 383
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.253768766301195,
      "learning_rate": 9.029793735676089e-05,
      "loss": 1.8257,
      "step": 384
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2703400790748331,
      "learning_rate": 9.02724726254138e-05,
      "loss": 1.9159,
      "step": 385
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.24359989557350126,
      "learning_rate": 9.024700789406673e-05,
      "loss": 1.9214,
      "step": 386
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25899432696845703,
      "learning_rate": 9.022154316271964e-05,
      "loss": 1.9816,
      "step": 387
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3254011608126227,
      "learning_rate": 9.019607843137255e-05,
      "loss": 1.7581,
      "step": 388
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2031342889422471,
      "learning_rate": 9.017061370002546e-05,
      "loss": 1.7662,
      "step": 389
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3026088700555451,
      "learning_rate": 9.014514896867839e-05,
      "loss": 2.2481,
      "step": 390
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21346682694721678,
      "learning_rate": 9.01196842373313e-05,
      "loss": 1.9334,
      "step": 391
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25777758737626477,
      "learning_rate": 9.009421950598421e-05,
      "loss": 2.2532,
      "step": 392
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.241481143173268,
      "learning_rate": 9.006875477463714e-05,
      "loss": 1.8002,
      "step": 393
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.20162168232777278,
      "learning_rate": 9.004329004329005e-05,
      "loss": 1.8179,
      "step": 394
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.24928963588818537,
      "learning_rate": 9.001782531194296e-05,
      "loss": 1.9853,
      "step": 395
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2562772090934246,
      "learning_rate": 8.999236058059587e-05,
      "loss": 1.991,
      "step": 396
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.265292759285601,
      "learning_rate": 8.99668958492488e-05,
      "loss": 1.777,
      "step": 397
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22762612081280048,
      "learning_rate": 8.994143111790171e-05,
      "loss": 1.8467,
      "step": 398
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2363791292496596,
      "learning_rate": 8.991596638655462e-05,
      "loss": 1.7783,
      "step": 399
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.24695737617258487,
      "learning_rate": 8.989050165520754e-05,
      "loss": 1.8617,
      "step": 400
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25795499158430346,
      "learning_rate": 8.986503692386046e-05,
      "loss": 2.1168,
      "step": 401
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2367588799205056,
      "learning_rate": 8.983957219251337e-05,
      "loss": 1.686,
      "step": 402
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2486136187076788,
      "learning_rate": 8.981410746116629e-05,
      "loss": 1.6311,
      "step": 403
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23212428234792606,
      "learning_rate": 8.978864272981921e-05,
      "loss": 1.8845,
      "step": 404
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.21297822252507279,
      "learning_rate": 8.976317799847212e-05,
      "loss": 1.6802,
      "step": 405
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22666768394766693,
      "learning_rate": 8.973771326712504e-05,
      "loss": 1.8839,
      "step": 406
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22721848395946773,
      "learning_rate": 8.971224853577795e-05,
      "loss": 1.7933,
      "step": 407
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23387883437850063,
      "learning_rate": 8.968678380443086e-05,
      "loss": 1.8834,
      "step": 408
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.25587524481325125,
      "learning_rate": 8.966131907308379e-05,
      "loss": 1.9019,
      "step": 409
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.22077238633383342,
      "learning_rate": 8.96358543417367e-05,
      "loss": 1.9587,
      "step": 410
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.269133675406636,
      "learning_rate": 8.961038961038961e-05,
      "loss": 1.8813,
      "step": 411
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.23159155269564086,
      "learning_rate": 8.958492487904253e-05,
      "loss": 2.0881,
      "step": 412
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.28728695003647925,
      "learning_rate": 8.955946014769545e-05,
      "loss": 2.0738,
      "step": 413
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22292837445688052,
      "learning_rate": 8.953399541634836e-05,
      "loss": 1.7848,
      "step": 414
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.25611957394245294,
      "learning_rate": 8.950853068500128e-05,
      "loss": 1.9309,
      "step": 415
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2808692394535083,
      "learning_rate": 8.948306595365418e-05,
      "loss": 2.0691,
      "step": 416
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.24805891203021524,
      "learning_rate": 8.945760122230711e-05,
      "loss": 1.8176,
      "step": 417
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22788913957793705,
      "learning_rate": 8.943213649096003e-05,
      "loss": 2.0164,
      "step": 418
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.28556400044197944,
      "learning_rate": 8.940667175961293e-05,
      "loss": 2.1959,
      "step": 419
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2865977497114976,
      "learning_rate": 8.938120702826586e-05,
      "loss": 1.9149,
      "step": 420
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23410808146142997,
      "learning_rate": 8.935574229691877e-05,
      "loss": 1.8804,
      "step": 421
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2320700474936451,
      "learning_rate": 8.933027756557168e-05,
      "loss": 1.6851,
      "step": 422
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23821295869621667,
      "learning_rate": 8.930481283422461e-05,
      "loss": 2.059,
      "step": 423
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22035404888955523,
      "learning_rate": 8.927934810287751e-05,
      "loss": 2.0662,
      "step": 424
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.25753879467837276,
      "learning_rate": 8.925388337153043e-05,
      "loss": 1.7946,
      "step": 425
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23174398602498045,
      "learning_rate": 8.922841864018336e-05,
      "loss": 1.9452,
      "step": 426
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21423272026802384,
      "learning_rate": 8.920295390883626e-05,
      "loss": 1.8668,
      "step": 427
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.23524905778157035,
      "learning_rate": 8.917748917748918e-05,
      "loss": 1.9177,
      "step": 428
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22128603765671084,
      "learning_rate": 8.915202444614211e-05,
      "loss": 1.6931,
      "step": 429
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.30417705228572045,
      "learning_rate": 8.9126559714795e-05,
      "loss": 1.7556,
      "step": 430
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22680455464532032,
      "learning_rate": 8.910109498344793e-05,
      "loss": 1.8978,
      "step": 431
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22491518846108297,
      "learning_rate": 8.907563025210084e-05,
      "loss": 1.8708,
      "step": 432
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.24514940513649147,
      "learning_rate": 8.905016552075376e-05,
      "loss": 2.0945,
      "step": 433
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3211911292812862,
      "learning_rate": 8.902470078940668e-05,
      "loss": 2.1676,
      "step": 434
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21672812222249246,
      "learning_rate": 8.89992360580596e-05,
      "loss": 1.9561,
      "step": 435
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.24722686150361434,
      "learning_rate": 8.89737713267125e-05,
      "loss": 2.0444,
      "step": 436
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2407730235551704,
      "learning_rate": 8.894830659536543e-05,
      "loss": 2.0372,
      "step": 437
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21103154840876165,
      "learning_rate": 8.892284186401833e-05,
      "loss": 1.786,
      "step": 438
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2513228969486743,
      "learning_rate": 8.889737713267126e-05,
      "loss": 1.9824,
      "step": 439
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.258544590475214,
      "learning_rate": 8.887191240132418e-05,
      "loss": 1.972,
      "step": 440
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22888813467895552,
      "learning_rate": 8.884644766997708e-05,
      "loss": 2.0106,
      "step": 441
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.21185224484030166,
      "learning_rate": 8.882098293863e-05,
      "loss": 1.6985,
      "step": 442
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2505140384273364,
      "learning_rate": 8.879551820728292e-05,
      "loss": 1.7832,
      "step": 443
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2634044704116914,
      "learning_rate": 8.877005347593583e-05,
      "loss": 1.9525,
      "step": 444
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.26359759096740165,
      "learning_rate": 8.874458874458876e-05,
      "loss": 1.8602,
      "step": 445
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22000572273007202,
      "learning_rate": 8.871912401324167e-05,
      "loss": 1.6984,
      "step": 446
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.26066085175478015,
      "learning_rate": 8.869365928189458e-05,
      "loss": 2.0568,
      "step": 447
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2773800812686037,
      "learning_rate": 8.86681945505475e-05,
      "loss": 1.9775,
      "step": 448
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.22358819774715658,
      "learning_rate": 8.86427298192004e-05,
      "loss": 1.8704,
      "step": 449
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3077196828448956,
      "learning_rate": 8.861726508785333e-05,
      "loss": 2.1512,
      "step": 450
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2786371669431707,
      "learning_rate": 8.859180035650624e-05,
      "loss": 2.0763,
      "step": 451
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22067992333181366,
      "learning_rate": 8.856633562515915e-05,
      "loss": 1.7935,
      "step": 452
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23456282242402895,
      "learning_rate": 8.854087089381208e-05,
      "loss": 1.8671,
      "step": 453
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.26017027006661725,
      "learning_rate": 8.851540616246499e-05,
      "loss": 1.5871,
      "step": 454
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2605339390232834,
      "learning_rate": 8.84899414311179e-05,
      "loss": 2.0417,
      "step": 455
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2249779620901503,
      "learning_rate": 8.846447669977083e-05,
      "loss": 1.8832,
      "step": 456
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23915839952016046,
      "learning_rate": 8.843901196842374e-05,
      "loss": 1.8563,
      "step": 457
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2620073406978356,
      "learning_rate": 8.841354723707665e-05,
      "loss": 1.9188,
      "step": 458
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.21269945433115764,
      "learning_rate": 8.838808250572957e-05,
      "loss": 1.5572,
      "step": 459
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.28633311164541697,
      "learning_rate": 8.836261777438248e-05,
      "loss": 2.0076,
      "step": 460
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2578739179800228,
      "learning_rate": 8.83371530430354e-05,
      "loss": 2.0649,
      "step": 461
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23703436505578596,
      "learning_rate": 8.831168831168831e-05,
      "loss": 1.8394,
      "step": 462
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.27268398548268413,
      "learning_rate": 8.828622358034123e-05,
      "loss": 2.0113,
      "step": 463
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.22900004271804628,
      "learning_rate": 8.826075884899415e-05,
      "loss": 1.8828,
      "step": 464
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2886673959221963,
      "learning_rate": 8.823529411764706e-05,
      "loss": 1.8799,
      "step": 465
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2320470672893286,
      "learning_rate": 8.820982938629998e-05,
      "loss": 1.9301,
      "step": 466
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.26738534373052875,
      "learning_rate": 8.818436465495289e-05,
      "loss": 2.0133,
      "step": 467
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25844804859889603,
      "learning_rate": 8.815889992360581e-05,
      "loss": 1.9833,
      "step": 468
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25046136127694163,
      "learning_rate": 8.813343519225873e-05,
      "loss": 1.7024,
      "step": 469
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2430617795253286,
      "learning_rate": 8.810797046091164e-05,
      "loss": 1.801,
      "step": 470
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2419599648636819,
      "learning_rate": 8.808250572956455e-05,
      "loss": 1.8624,
      "step": 471
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2659463341273293,
      "learning_rate": 8.805704099821748e-05,
      "loss": 2.0745,
      "step": 472
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.24013448980226984,
      "learning_rate": 8.803157626687039e-05,
      "loss": 1.7037,
      "step": 473
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23022342815075916,
      "learning_rate": 8.80061115355233e-05,
      "loss": 1.8988,
      "step": 474
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2367004267836987,
      "learning_rate": 8.798064680417623e-05,
      "loss": 1.8279,
      "step": 475
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2620297477283304,
      "learning_rate": 8.795518207282914e-05,
      "loss": 1.8255,
      "step": 476
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2547679218327907,
      "learning_rate": 8.792971734148205e-05,
      "loss": 1.9405,
      "step": 477
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.24895355889731416,
      "learning_rate": 8.790425261013496e-05,
      "loss": 1.9529,
      "step": 478
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.28745180586441543,
      "learning_rate": 8.787878787878789e-05,
      "loss": 1.8011,
      "step": 479
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2878862108904096,
      "learning_rate": 8.78533231474408e-05,
      "loss": 2.0354,
      "step": 480
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25603030952198,
      "learning_rate": 8.782785841609371e-05,
      "loss": 2.0772,
      "step": 481
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.23827494941261018,
      "learning_rate": 8.780239368474664e-05,
      "loss": 2.2084,
      "step": 482
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.20837960126816166,
      "learning_rate": 8.777692895339955e-05,
      "loss": 1.5153,
      "step": 483
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.20977901295253784,
      "learning_rate": 8.775146422205246e-05,
      "loss": 1.953,
      "step": 484
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2803722831995815,
      "learning_rate": 8.772599949070537e-05,
      "loss": 2.0062,
      "step": 485
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.31592634092672583,
      "learning_rate": 8.770053475935829e-05,
      "loss": 1.9259,
      "step": 486
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.20853917661705262,
      "learning_rate": 8.767507002801121e-05,
      "loss": 1.7843,
      "step": 487
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2622256707469382,
      "learning_rate": 8.764960529666412e-05,
      "loss": 1.6427,
      "step": 488
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.26821743683336324,
      "learning_rate": 8.762414056531704e-05,
      "loss": 1.797,
      "step": 489
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.29751845947665584,
      "learning_rate": 8.759867583396996e-05,
      "loss": 2.0772,
      "step": 490
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2499209815555211,
      "learning_rate": 8.757321110262287e-05,
      "loss": 2.2188,
      "step": 491
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2432774199108946,
      "learning_rate": 8.754774637127579e-05,
      "loss": 1.7888,
      "step": 492
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2342067273383554,
      "learning_rate": 8.752228163992871e-05,
      "loss": 1.9258,
      "step": 493
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.262556869159515,
      "learning_rate": 8.749681690858161e-05,
      "loss": 2.0776,
      "step": 494
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21427865705052013,
      "learning_rate": 8.747135217723454e-05,
      "loss": 1.9552,
      "step": 495
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2385128880858244,
      "learning_rate": 8.744588744588745e-05,
      "loss": 2.0085,
      "step": 496
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.19959484390847743,
      "learning_rate": 8.742042271454036e-05,
      "loss": 1.6707,
      "step": 497
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2590217264205479,
      "learning_rate": 8.739495798319329e-05,
      "loss": 2.0314,
      "step": 498
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24982669903351035,
      "learning_rate": 8.73694932518462e-05,
      "loss": 1.6773,
      "step": 499
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.26589731674888417,
      "learning_rate": 8.734402852049911e-05,
      "loss": 2.1827,
      "step": 500
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2413060236399928,
      "learning_rate": 8.731856378915203e-05,
      "loss": 1.9234,
      "step": 501
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.25943996695878224,
      "learning_rate": 8.729309905780493e-05,
      "loss": 2.064,
      "step": 502
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2362646853207089,
      "learning_rate": 8.726763432645786e-05,
      "loss": 1.736,
      "step": 503
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5992785606723764,
      "learning_rate": 8.724216959511078e-05,
      "loss": 2.0825,
      "step": 504
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2540455949327468,
      "learning_rate": 8.721670486376368e-05,
      "loss": 1.9849,
      "step": 505
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.28364373494321304,
      "learning_rate": 8.719124013241661e-05,
      "loss": 1.9012,
      "step": 506
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24171865365853143,
      "learning_rate": 8.716577540106952e-05,
      "loss": 2.0105,
      "step": 507
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.26652123215640866,
      "learning_rate": 8.714031066972243e-05,
      "loss": 1.7502,
      "step": 508
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24173515900322068,
      "learning_rate": 8.711484593837536e-05,
      "loss": 1.9814,
      "step": 509
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.23354274486346174,
      "learning_rate": 8.708938120702827e-05,
      "loss": 1.7613,
      "step": 510
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2443504042577395,
      "learning_rate": 8.706391647568118e-05,
      "loss": 1.6964,
      "step": 511
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24122925352912084,
      "learning_rate": 8.703845174433411e-05,
      "loss": 1.8006,
      "step": 512
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2333847584566521,
      "learning_rate": 8.701298701298701e-05,
      "loss": 1.8593,
      "step": 513
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2279094660924481,
      "learning_rate": 8.698752228163993e-05,
      "loss": 1.8878,
      "step": 514
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2527167996599817,
      "learning_rate": 8.696205755029286e-05,
      "loss": 1.8256,
      "step": 515
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2681148186553472,
      "learning_rate": 8.693659281894576e-05,
      "loss": 1.8707,
      "step": 516
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.23937145570416912,
      "learning_rate": 8.691112808759868e-05,
      "loss": 1.9254,
      "step": 517
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2501936549921692,
      "learning_rate": 8.68856633562516e-05,
      "loss": 1.9475,
      "step": 518
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24358617767152815,
      "learning_rate": 8.68601986249045e-05,
      "loss": 2.0011,
      "step": 519
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.24964833858406257,
      "learning_rate": 8.683473389355743e-05,
      "loss": 1.9767,
      "step": 520
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22841990332662865,
      "learning_rate": 8.680926916221034e-05,
      "loss": 1.8258,
      "step": 521
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2216953490518328,
      "learning_rate": 8.678380443086326e-05,
      "loss": 1.8854,
      "step": 522
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.21155130303192776,
      "learning_rate": 8.675833969951618e-05,
      "loss": 1.6807,
      "step": 523
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2564998792365461,
      "learning_rate": 8.673287496816908e-05,
      "loss": 2.0952,
      "step": 524
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.32649142208193577,
      "learning_rate": 8.6707410236822e-05,
      "loss": 2.3346,
      "step": 525
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22465641524281874,
      "learning_rate": 8.668194550547493e-05,
      "loss": 1.876,
      "step": 526
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.23566365423282545,
      "learning_rate": 8.665648077412783e-05,
      "loss": 1.7912,
      "step": 527
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2426874706745744,
      "learning_rate": 8.663101604278076e-05,
      "loss": 2.1064,
      "step": 528
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.256102526492817,
      "learning_rate": 8.660555131143367e-05,
      "loss": 1.7732,
      "step": 529
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.22674630371643414,
      "learning_rate": 8.658008658008658e-05,
      "loss": 1.8903,
      "step": 530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3223009379141864,
      "learning_rate": 8.65546218487395e-05,
      "loss": 2.0436,
      "step": 531
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.253161114600626,
      "learning_rate": 8.652915711739242e-05,
      "loss": 1.9741,
      "step": 532
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23017841575516462,
      "learning_rate": 8.650369238604533e-05,
      "loss": 1.7193,
      "step": 533
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.23793766389765506,
      "learning_rate": 8.647822765469826e-05,
      "loss": 1.8982,
      "step": 534
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24108078568832444,
      "learning_rate": 8.645276292335115e-05,
      "loss": 1.8601,
      "step": 535
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21032110859794031,
      "learning_rate": 8.642729819200408e-05,
      "loss": 1.7736,
      "step": 536
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25852270274014505,
      "learning_rate": 8.640183346065699e-05,
      "loss": 2.2393,
      "step": 537
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21256717118119425,
      "learning_rate": 8.63763687293099e-05,
      "loss": 1.7333,
      "step": 538
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24096739207525672,
      "learning_rate": 8.635090399796283e-05,
      "loss": 1.8944,
      "step": 539
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25287979694385604,
      "learning_rate": 8.632543926661574e-05,
      "loss": 1.9336,
      "step": 540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2716083517972064,
      "learning_rate": 8.629997453526865e-05,
      "loss": 2.0506,
      "step": 541
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2370221846524856,
      "learning_rate": 8.627450980392158e-05,
      "loss": 1.8359,
      "step": 542
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24195282856938768,
      "learning_rate": 8.624904507257449e-05,
      "loss": 1.8325,
      "step": 543
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.28289711302928944,
      "learning_rate": 8.62235803412274e-05,
      "loss": 1.9292,
      "step": 544
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24438103092781144,
      "learning_rate": 8.619811560988032e-05,
      "loss": 1.5135,
      "step": 545
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2214814446606962,
      "learning_rate": 8.617265087853324e-05,
      "loss": 1.8107,
      "step": 546
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2495619616900341,
      "learning_rate": 8.614718614718615e-05,
      "loss": 1.8602,
      "step": 547
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25771399263311523,
      "learning_rate": 8.612172141583907e-05,
      "loss": 1.8007,
      "step": 548
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24994041060517444,
      "learning_rate": 8.609625668449198e-05,
      "loss": 1.9156,
      "step": 549
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22366784167075848,
      "learning_rate": 8.60707919531449e-05,
      "loss": 1.9989,
      "step": 550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2630253246918491,
      "learning_rate": 8.604532722179781e-05,
      "loss": 1.6313,
      "step": 551
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.26379510331153455,
      "learning_rate": 8.601986249045073e-05,
      "loss": 1.7552,
      "step": 552
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2079340754806867,
      "learning_rate": 8.599439775910365e-05,
      "loss": 2.0048,
      "step": 553
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.21692329020273132,
      "learning_rate": 8.596893302775656e-05,
      "loss": 1.9039,
      "step": 554
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22066896731286834,
      "learning_rate": 8.594346829640948e-05,
      "loss": 1.7687,
      "step": 555
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22092093632239163,
      "learning_rate": 8.591800356506239e-05,
      "loss": 1.5822,
      "step": 556
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.26549194313206226,
      "learning_rate": 8.589253883371531e-05,
      "loss": 1.9684,
      "step": 557
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2781915333279139,
      "learning_rate": 8.586707410236823e-05,
      "loss": 2.1979,
      "step": 558
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2516189251283423,
      "learning_rate": 8.584160937102114e-05,
      "loss": 2.0177,
      "step": 559
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22633635810464564,
      "learning_rate": 8.581614463967405e-05,
      "loss": 1.7088,
      "step": 560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25571341391614216,
      "learning_rate": 8.579067990832698e-05,
      "loss": 1.666,
      "step": 561
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24154109213497876,
      "learning_rate": 8.576521517697989e-05,
      "loss": 1.9523,
      "step": 562
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.198164384354011,
      "learning_rate": 8.57397504456328e-05,
      "loss": 1.8036,
      "step": 563
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25184206030165557,
      "learning_rate": 8.571428571428571e-05,
      "loss": 1.9446,
      "step": 564
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2573172407271812,
      "learning_rate": 8.568882098293864e-05,
      "loss": 1.9315,
      "step": 565
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2696812392046249,
      "learning_rate": 8.566335625159155e-05,
      "loss": 1.8225,
      "step": 566
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25286593425375464,
      "learning_rate": 8.563789152024446e-05,
      "loss": 2.0231,
      "step": 567
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.22767392856341184,
      "learning_rate": 8.561242678889739e-05,
      "loss": 1.7285,
      "step": 568
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.24801175927731736,
      "learning_rate": 8.55869620575503e-05,
      "loss": 1.6252,
      "step": 569
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21808987878527664,
      "learning_rate": 8.556149732620321e-05,
      "loss": 1.7553,
      "step": 570
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2573754623320004,
      "learning_rate": 8.553603259485612e-05,
      "loss": 2.1009,
      "step": 571
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.18939076213266787,
      "learning_rate": 8.551056786350904e-05,
      "loss": 1.686,
      "step": 572
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.24812637016436956,
      "learning_rate": 8.548510313216196e-05,
      "loss": 1.9609,
      "step": 573
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2338966733848831,
      "learning_rate": 8.545963840081487e-05,
      "loss": 1.9818,
      "step": 574
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2404840149226457,
      "learning_rate": 8.543417366946779e-05,
      "loss": 1.8081,
      "step": 575
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.25798725216628743,
      "learning_rate": 8.540870893812071e-05,
      "loss": 1.7854,
      "step": 576
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.27750663272560383,
      "learning_rate": 8.538324420677362e-05,
      "loss": 2.0337,
      "step": 577
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2769580899803715,
      "learning_rate": 8.535777947542654e-05,
      "loss": 1.9486,
      "step": 578
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.25306861331816644,
      "learning_rate": 8.533231474407946e-05,
      "loss": 1.7939,
      "step": 579
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22845769620715287,
      "learning_rate": 8.530685001273236e-05,
      "loss": 1.8121,
      "step": 580
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2595561480765673,
      "learning_rate": 8.528138528138529e-05,
      "loss": 1.8217,
      "step": 581
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2548756068588673,
      "learning_rate": 8.52559205500382e-05,
      "loss": 1.7951,
      "step": 582
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23753990478065184,
      "learning_rate": 8.523045581869111e-05,
      "loss": 1.827,
      "step": 583
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2795844698118675,
      "learning_rate": 8.520499108734404e-05,
      "loss": 1.8995,
      "step": 584
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2650106702986557,
      "learning_rate": 8.517952635599695e-05,
      "loss": 1.8459,
      "step": 585
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2583049636819499,
      "learning_rate": 8.515406162464986e-05,
      "loss": 2.0852,
      "step": 586
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2544859357696881,
      "learning_rate": 8.512859689330279e-05,
      "loss": 1.7271,
      "step": 587
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.27203970995110843,
      "learning_rate": 8.51031321619557e-05,
      "loss": 2.1576,
      "step": 588
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2760773611682643,
      "learning_rate": 8.507766743060861e-05,
      "loss": 2.1948,
      "step": 589
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.26657512379993753,
      "learning_rate": 8.505220269926153e-05,
      "loss": 1.952,
      "step": 590
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22449921933656483,
      "learning_rate": 8.502673796791443e-05,
      "loss": 1.8616,
      "step": 591
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2269948090503817,
      "learning_rate": 8.500127323656736e-05,
      "loss": 1.7622,
      "step": 592
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.225806033942239,
      "learning_rate": 8.497580850522027e-05,
      "loss": 1.679,
      "step": 593
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.27226095482913315,
      "learning_rate": 8.495034377387318e-05,
      "loss": 2.2212,
      "step": 594
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.25834770713831146,
      "learning_rate": 8.492487904252611e-05,
      "loss": 1.905,
      "step": 595
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.22415664490617518,
      "learning_rate": 8.489941431117902e-05,
      "loss": 1.9853,
      "step": 596
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2614588372097758,
      "learning_rate": 8.487394957983193e-05,
      "loss": 1.7831,
      "step": 597
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2513907906833273,
      "learning_rate": 8.484848484848486e-05,
      "loss": 2.1041,
      "step": 598
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21971733073868202,
      "learning_rate": 8.482302011713776e-05,
      "loss": 1.6445,
      "step": 599
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2197734189405905,
      "learning_rate": 8.479755538579068e-05,
      "loss": 1.5833,
      "step": 600
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.26087178356434226,
      "learning_rate": 8.477209065444361e-05,
      "loss": 1.8881,
      "step": 601
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.27040403833766596,
      "learning_rate": 8.474662592309651e-05,
      "loss": 1.9293,
      "step": 602
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2602180613727054,
      "learning_rate": 8.472116119174943e-05,
      "loss": 2.1294,
      "step": 603
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2234191039988877,
      "learning_rate": 8.469569646040236e-05,
      "loss": 1.7762,
      "step": 604
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3153772694403617,
      "learning_rate": 8.467023172905526e-05,
      "loss": 1.9273,
      "step": 605
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23455438107172347,
      "learning_rate": 8.464476699770818e-05,
      "loss": 1.9544,
      "step": 606
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2658116862672306,
      "learning_rate": 8.46193022663611e-05,
      "loss": 2.2271,
      "step": 607
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.32975917281160555,
      "learning_rate": 8.4593837535014e-05,
      "loss": 1.7337,
      "step": 608
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2228088608801704,
      "learning_rate": 8.456837280366693e-05,
      "loss": 2.0129,
      "step": 609
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.26447431178004766,
      "learning_rate": 8.454290807231983e-05,
      "loss": 1.9728,
      "step": 610
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21976201089381797,
      "learning_rate": 8.451744334097276e-05,
      "loss": 1.8064,
      "step": 611
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3072809470726187,
      "learning_rate": 8.449197860962568e-05,
      "loss": 1.835,
      "step": 612
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24577672914201923,
      "learning_rate": 8.446651387827858e-05,
      "loss": 1.9267,
      "step": 613
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2236691741034138,
      "learning_rate": 8.44410491469315e-05,
      "loss": 1.8683,
      "step": 614
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23482479010040266,
      "learning_rate": 8.441558441558442e-05,
      "loss": 2.0112,
      "step": 615
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.25066078299216493,
      "learning_rate": 8.439011968423733e-05,
      "loss": 1.835,
      "step": 616
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.264604515569423,
      "learning_rate": 8.436465495289026e-05,
      "loss": 1.9542,
      "step": 617
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.26229058609109396,
      "learning_rate": 8.433919022154317e-05,
      "loss": 1.8818,
      "step": 618
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2535097988499568,
      "learning_rate": 8.431372549019608e-05,
      "loss": 1.8894,
      "step": 619
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23424679110516414,
      "learning_rate": 8.4288260758849e-05,
      "loss": 1.6131,
      "step": 620
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.26691813415164034,
      "learning_rate": 8.426279602750192e-05,
      "loss": 2.029,
      "step": 621
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.22172670243930442,
      "learning_rate": 8.423733129615483e-05,
      "loss": 1.6191,
      "step": 622
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.27667095622410315,
      "learning_rate": 8.421186656480774e-05,
      "loss": 2.2586,
      "step": 623
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.301753010140633,
      "learning_rate": 8.418640183346065e-05,
      "loss": 1.9892,
      "step": 624
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1897230388556837,
      "learning_rate": 8.416093710211358e-05,
      "loss": 1.7812,
      "step": 625
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24970974314637662,
      "learning_rate": 8.413547237076649e-05,
      "loss": 2.0589,
      "step": 626
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3032015856119192,
      "learning_rate": 8.41100076394194e-05,
      "loss": 2.2445,
      "step": 627
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2108389659614189,
      "learning_rate": 8.408454290807233e-05,
      "loss": 1.6547,
      "step": 628
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.33296527758702654,
      "learning_rate": 8.405907817672524e-05,
      "loss": 2.0687,
      "step": 629
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.25872450689146614,
      "learning_rate": 8.403361344537815e-05,
      "loss": 2.1157,
      "step": 630
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24882848251532677,
      "learning_rate": 8.400814871403108e-05,
      "loss": 2.1495,
      "step": 631
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24377256496837985,
      "learning_rate": 8.398268398268399e-05,
      "loss": 1.6708,
      "step": 632
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.27653964427752414,
      "learning_rate": 8.39572192513369e-05,
      "loss": 2.0369,
      "step": 633
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23087998248198688,
      "learning_rate": 8.393175451998982e-05,
      "loss": 1.9132,
      "step": 634
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2343100298794321,
      "learning_rate": 8.390628978864273e-05,
      "loss": 2.0154,
      "step": 635
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.27080108224361626,
      "learning_rate": 8.388082505729565e-05,
      "loss": 1.9704,
      "step": 636
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23709343506276895,
      "learning_rate": 8.385536032594857e-05,
      "loss": 1.9831,
      "step": 637
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2233678922733808,
      "learning_rate": 8.382989559460148e-05,
      "loss": 1.5794,
      "step": 638
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.22254628259973275,
      "learning_rate": 8.38044308632544e-05,
      "loss": 1.9141,
      "step": 639
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.22679648977175332,
      "learning_rate": 8.377896613190732e-05,
      "loss": 1.869,
      "step": 640
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24650163756314084,
      "learning_rate": 8.375350140056023e-05,
      "loss": 1.975,
      "step": 641
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2356487471404544,
      "learning_rate": 8.372803666921314e-05,
      "loss": 2.0583,
      "step": 642
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23967999170522092,
      "learning_rate": 8.370257193786606e-05,
      "loss": 1.7817,
      "step": 643
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.20266040525683782,
      "learning_rate": 8.367710720651898e-05,
      "loss": 1.7697,
      "step": 644
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2594234966991798,
      "learning_rate": 8.365164247517189e-05,
      "loss": 1.8042,
      "step": 645
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2562563488336597,
      "learning_rate": 8.36261777438248e-05,
      "loss": 1.9621,
      "step": 646
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.27401385191984207,
      "learning_rate": 8.360071301247773e-05,
      "loss": 1.9699,
      "step": 647
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2675310914964076,
      "learning_rate": 8.357524828113064e-05,
      "loss": 1.8894,
      "step": 648
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2242659555646828,
      "learning_rate": 8.354978354978355e-05,
      "loss": 1.8756,
      "step": 649
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.254521578001409,
      "learning_rate": 8.352431881843646e-05,
      "loss": 1.8776,
      "step": 650
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24220297364056145,
      "learning_rate": 8.349885408708939e-05,
      "loss": 1.887,
      "step": 651
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2440556645382515,
      "learning_rate": 8.34733893557423e-05,
      "loss": 1.7679,
      "step": 652
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22591067139033177,
      "learning_rate": 8.344792462439521e-05,
      "loss": 1.9008,
      "step": 653
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3398909700407971,
      "learning_rate": 8.342245989304814e-05,
      "loss": 1.8229,
      "step": 654
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24230125278447373,
      "learning_rate": 8.339699516170105e-05,
      "loss": 1.7633,
      "step": 655
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20022248804750795,
      "learning_rate": 8.337153043035396e-05,
      "loss": 1.6237,
      "step": 656
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25412426905003854,
      "learning_rate": 8.334606569900687e-05,
      "loss": 1.8433,
      "step": 657
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22552341569093054,
      "learning_rate": 8.332060096765979e-05,
      "loss": 1.7831,
      "step": 658
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22664716388325745,
      "learning_rate": 8.329513623631271e-05,
      "loss": 1.7294,
      "step": 659
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25906233843900756,
      "learning_rate": 8.326967150496562e-05,
      "loss": 1.8616,
      "step": 660
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3111900770192149,
      "learning_rate": 8.324420677361854e-05,
      "loss": 2.0473,
      "step": 661
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24845573190451736,
      "learning_rate": 8.321874204227146e-05,
      "loss": 2.0842,
      "step": 662
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2721528256093899,
      "learning_rate": 8.319327731092437e-05,
      "loss": 2.01,
      "step": 663
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.29973802850597847,
      "learning_rate": 8.316781257957729e-05,
      "loss": 2.0272,
      "step": 664
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22714114288136683,
      "learning_rate": 8.314234784823021e-05,
      "loss": 1.7836,
      "step": 665
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23256989582074034,
      "learning_rate": 8.311688311688312e-05,
      "loss": 1.6346,
      "step": 666
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23821437397510137,
      "learning_rate": 8.309141838553604e-05,
      "loss": 1.7571,
      "step": 667
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2412285199880779,
      "learning_rate": 8.306595365418896e-05,
      "loss": 2.0028,
      "step": 668
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.26433168437428606,
      "learning_rate": 8.304048892284186e-05,
      "loss": 1.7625,
      "step": 669
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2744294023129235,
      "learning_rate": 8.301502419149479e-05,
      "loss": 2.0854,
      "step": 670
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.26885424466275476,
      "learning_rate": 8.29895594601477e-05,
      "loss": 1.9209,
      "step": 671
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2709272142457605,
      "learning_rate": 8.296409472880061e-05,
      "loss": 2.1129,
      "step": 672
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23999185648900634,
      "learning_rate": 8.293862999745354e-05,
      "loss": 1.6473,
      "step": 673
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25770608555351354,
      "learning_rate": 8.291316526610645e-05,
      "loss": 1.7928,
      "step": 674
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2537587397578314,
      "learning_rate": 8.288770053475936e-05,
      "loss": 1.8933,
      "step": 675
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23704349601861344,
      "learning_rate": 8.286223580341229e-05,
      "loss": 1.8767,
      "step": 676
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2513192208858804,
      "learning_rate": 8.283677107206518e-05,
      "loss": 1.7624,
      "step": 677
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2537437065405792,
      "learning_rate": 8.281130634071811e-05,
      "loss": 1.9107,
      "step": 678
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25124285630845544,
      "learning_rate": 8.278584160937104e-05,
      "loss": 2.2631,
      "step": 679
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2830311483235791,
      "learning_rate": 8.276037687802393e-05,
      "loss": 1.9933,
      "step": 680
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22526784656744517,
      "learning_rate": 8.273491214667686e-05,
      "loss": 1.745,
      "step": 681
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20692988379313565,
      "learning_rate": 8.270944741532977e-05,
      "loss": 1.8224,
      "step": 682
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3125523880915381,
      "learning_rate": 8.268398268398268e-05,
      "loss": 2.1634,
      "step": 683
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.26933932727564425,
      "learning_rate": 8.265851795263561e-05,
      "loss": 2.0392,
      "step": 684
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2341672214756589,
      "learning_rate": 8.263305322128852e-05,
      "loss": 1.8573,
      "step": 685
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24637431847948854,
      "learning_rate": 8.260758848994143e-05,
      "loss": 1.7781,
      "step": 686
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.27247735797800854,
      "learning_rate": 8.258212375859436e-05,
      "loss": 2.0406,
      "step": 687
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2509004499748774,
      "learning_rate": 8.255665902724726e-05,
      "loss": 2.0022,
      "step": 688
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2490574233855169,
      "learning_rate": 8.253119429590018e-05,
      "loss": 1.9605,
      "step": 689
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21044261594535146,
      "learning_rate": 8.250572956455311e-05,
      "loss": 1.8126,
      "step": 690
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2536984840725562,
      "learning_rate": 8.248026483320601e-05,
      "loss": 1.9673,
      "step": 691
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.20622765029844037,
      "learning_rate": 8.245480010185893e-05,
      "loss": 1.7606,
      "step": 692
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2542420147904985,
      "learning_rate": 8.242933537051184e-05,
      "loss": 2.0577,
      "step": 693
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.26315616047124607,
      "learning_rate": 8.240387063916476e-05,
      "loss": 1.9809,
      "step": 694
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22211583961688874,
      "learning_rate": 8.237840590781768e-05,
      "loss": 1.8249,
      "step": 695
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.24345249090692983,
      "learning_rate": 8.23529411764706e-05,
      "loss": 2.1196,
      "step": 696
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2437022889955659,
      "learning_rate": 8.23274764451235e-05,
      "loss": 2.2124,
      "step": 697
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21587595502844253,
      "learning_rate": 8.230201171377643e-05,
      "loss": 1.7082,
      "step": 698
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.19780382068027816,
      "learning_rate": 8.227654698242933e-05,
      "loss": 1.6877,
      "step": 699
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2606709700917808,
      "learning_rate": 8.225108225108226e-05,
      "loss": 1.6861,
      "step": 700
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22780511715912988,
      "learning_rate": 8.222561751973517e-05,
      "loss": 1.8206,
      "step": 701
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21568730083147977,
      "learning_rate": 8.220015278838808e-05,
      "loss": 1.7652,
      "step": 702
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.23477539194517308,
      "learning_rate": 8.2174688057041e-05,
      "loss": 1.8679,
      "step": 703
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2474139165125321,
      "learning_rate": 8.214922332569392e-05,
      "loss": 1.7812,
      "step": 704
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21314623125660764,
      "learning_rate": 8.212375859434683e-05,
      "loss": 1.7575,
      "step": 705
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2562433957202535,
      "learning_rate": 8.209829386299976e-05,
      "loss": 1.9478,
      "step": 706
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.29203077845895087,
      "learning_rate": 8.207282913165267e-05,
      "loss": 2.145,
      "step": 707
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.23551961035637994,
      "learning_rate": 8.204736440030558e-05,
      "loss": 2.0507,
      "step": 708
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2874919133500298,
      "learning_rate": 8.20218996689585e-05,
      "loss": 1.8884,
      "step": 709
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2435714419184239,
      "learning_rate": 8.19964349376114e-05,
      "loss": 1.7244,
      "step": 710
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.25461303933038876,
      "learning_rate": 8.197097020626433e-05,
      "loss": 1.8317,
      "step": 711
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3174139389033714,
      "learning_rate": 8.194550547491724e-05,
      "loss": 2.4544,
      "step": 712
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.260350536850943,
      "learning_rate": 8.192004074357015e-05,
      "loss": 1.7694,
      "step": 713
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.26749580397835515,
      "learning_rate": 8.189457601222308e-05,
      "loss": 1.8921,
      "step": 714
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21515415482896466,
      "learning_rate": 8.186911128087599e-05,
      "loss": 1.6123,
      "step": 715
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2814437675566034,
      "learning_rate": 8.18436465495289e-05,
      "loss": 1.6636,
      "step": 716
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.25757055778830756,
      "learning_rate": 8.181818181818183e-05,
      "loss": 2.0113,
      "step": 717
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.26431323567719006,
      "learning_rate": 8.179271708683474e-05,
      "loss": 1.9948,
      "step": 718
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.264939728509053,
      "learning_rate": 8.176725235548765e-05,
      "loss": 1.823,
      "step": 719
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22863864847566018,
      "learning_rate": 8.174178762414057e-05,
      "loss": 1.7145,
      "step": 720
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2926476308161338,
      "learning_rate": 8.171632289279348e-05,
      "loss": 1.9253,
      "step": 721
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2567930149430035,
      "learning_rate": 8.16908581614464e-05,
      "loss": 1.9469,
      "step": 722
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.21388350820757424,
      "learning_rate": 8.166539343009932e-05,
      "loss": 1.7821,
      "step": 723
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2390070476919383,
      "learning_rate": 8.163992869875223e-05,
      "loss": 1.7102,
      "step": 724
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.24249325187378476,
      "learning_rate": 8.161446396740515e-05,
      "loss": 1.8183,
      "step": 725
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22973235529222133,
      "learning_rate": 8.158899923605807e-05,
      "loss": 1.8443,
      "step": 726
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2278244128468002,
      "learning_rate": 8.156353450471098e-05,
      "loss": 1.8287,
      "step": 727
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21975864600021108,
      "learning_rate": 8.153806977336389e-05,
      "loss": 1.7561,
      "step": 728
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.24673932789949182,
      "learning_rate": 8.151260504201682e-05,
      "loss": 1.8798,
      "step": 729
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23689722471477442,
      "learning_rate": 8.148714031066973e-05,
      "loss": 1.9101,
      "step": 730
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23534223925853134,
      "learning_rate": 8.146167557932264e-05,
      "loss": 2.0234,
      "step": 731
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23786358891087073,
      "learning_rate": 8.143621084797556e-05,
      "loss": 1.8746,
      "step": 732
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2531245449438832,
      "learning_rate": 8.141074611662848e-05,
      "loss": 1.7922,
      "step": 733
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.28806956616452595,
      "learning_rate": 8.138528138528139e-05,
      "loss": 1.9351,
      "step": 734
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2245345944850363,
      "learning_rate": 8.13598166539343e-05,
      "loss": 1.7899,
      "step": 735
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.262884773654068,
      "learning_rate": 8.133435192258721e-05,
      "loss": 1.9936,
      "step": 736
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2660954600675136,
      "learning_rate": 8.130888719124014e-05,
      "loss": 1.9235,
      "step": 737
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2544210936284766,
      "learning_rate": 8.128342245989305e-05,
      "loss": 2.0572,
      "step": 738
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21115268222074945,
      "learning_rate": 8.125795772854596e-05,
      "loss": 1.6713,
      "step": 739
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23399606907937723,
      "learning_rate": 8.123249299719889e-05,
      "loss": 1.7741,
      "step": 740
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2413301829518615,
      "learning_rate": 8.12070282658518e-05,
      "loss": 1.617,
      "step": 741
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.24363345300377096,
      "learning_rate": 8.118156353450471e-05,
      "loss": 1.8219,
      "step": 742
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.21322552997622987,
      "learning_rate": 8.115609880315764e-05,
      "loss": 1.8923,
      "step": 743
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2648814959339343,
      "learning_rate": 8.113063407181055e-05,
      "loss": 1.8955,
      "step": 744
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.24604419936013278,
      "learning_rate": 8.110516934046346e-05,
      "loss": 1.7781,
      "step": 745
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.31400117322830134,
      "learning_rate": 8.107970460911637e-05,
      "loss": 2.0413,
      "step": 746
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23840782055804374,
      "learning_rate": 8.105423987776929e-05,
      "loss": 1.8463,
      "step": 747
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2444438272270324,
      "learning_rate": 8.102877514642221e-05,
      "loss": 1.7236,
      "step": 748
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2796474737157944,
      "learning_rate": 8.100331041507512e-05,
      "loss": 2.1684,
      "step": 749
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2133773175143318,
      "learning_rate": 8.097784568372804e-05,
      "loss": 1.707,
      "step": 750
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2839241058327468,
      "learning_rate": 8.095238095238096e-05,
      "loss": 1.8488,
      "step": 751
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2177899957357643,
      "learning_rate": 8.092691622103387e-05,
      "loss": 1.5671,
      "step": 752
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.28396641702030523,
      "learning_rate": 8.090145148968679e-05,
      "loss": 1.9648,
      "step": 753
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.22902053902743993,
      "learning_rate": 8.087598675833971e-05,
      "loss": 1.8343,
      "step": 754
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.270526183017562,
      "learning_rate": 8.085052202699261e-05,
      "loss": 2.0001,
      "step": 755
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2427759405177403,
      "learning_rate": 8.082505729564554e-05,
      "loss": 1.8584,
      "step": 756
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20730834087964428,
      "learning_rate": 8.079959256429845e-05,
      "loss": 1.8038,
      "step": 757
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23676530738979948,
      "learning_rate": 8.077412783295136e-05,
      "loss": 1.8173,
      "step": 758
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.27498343645411766,
      "learning_rate": 8.074866310160429e-05,
      "loss": 2.1532,
      "step": 759
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.30573909192110055,
      "learning_rate": 8.07231983702572e-05,
      "loss": 1.7967,
      "step": 760
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.25874172713465876,
      "learning_rate": 8.069773363891011e-05,
      "loss": 1.9587,
      "step": 761
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23840994564446671,
      "learning_rate": 8.067226890756304e-05,
      "loss": 1.8361,
      "step": 762
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.20338547035408697,
      "learning_rate": 8.064680417621593e-05,
      "loss": 1.8748,
      "step": 763
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.22107511583394449,
      "learning_rate": 8.062133944486886e-05,
      "loss": 1.7538,
      "step": 764
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23126567033564033,
      "learning_rate": 8.059587471352179e-05,
      "loss": 1.7758,
      "step": 765
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2623222140707853,
      "learning_rate": 8.057040998217468e-05,
      "loss": 2.0597,
      "step": 766
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2645590655894485,
      "learning_rate": 8.054494525082761e-05,
      "loss": 1.8437,
      "step": 767
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2017017811080249,
      "learning_rate": 8.051948051948052e-05,
      "loss": 1.7988,
      "step": 768
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.23507420829129272,
      "learning_rate": 8.049401578813343e-05,
      "loss": 1.9862,
      "step": 769
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.26716449538128145,
      "learning_rate": 8.046855105678636e-05,
      "loss": 1.6237,
      "step": 770
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37869065827033116,
      "learning_rate": 8.044308632543927e-05,
      "loss": 1.8673,
      "step": 771
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.23080535327611468,
      "learning_rate": 8.041762159409218e-05,
      "loss": 1.7489,
      "step": 772
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.25052884909217044,
      "learning_rate": 8.039215686274511e-05,
      "loss": 1.9734,
      "step": 773
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.20729400047988472,
      "learning_rate": 8.036669213139801e-05,
      "loss": 1.6804,
      "step": 774
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24687693118293738,
      "learning_rate": 8.034122740005093e-05,
      "loss": 2.0605,
      "step": 775
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24239256083813954,
      "learning_rate": 8.031576266870386e-05,
      "loss": 1.818,
      "step": 776
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2385267259605302,
      "learning_rate": 8.029029793735676e-05,
      "loss": 1.9034,
      "step": 777
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2326704907462259,
      "learning_rate": 8.026483320600968e-05,
      "loss": 1.7703,
      "step": 778
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.26165915992218536,
      "learning_rate": 8.02393684746626e-05,
      "loss": 2.0577,
      "step": 779
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.19764839465849784,
      "learning_rate": 8.021390374331551e-05,
      "loss": 1.954,
      "step": 780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22493088144049952,
      "learning_rate": 8.018843901196843e-05,
      "loss": 2.1367,
      "step": 781
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2341060764533811,
      "learning_rate": 8.016297428062134e-05,
      "loss": 1.5133,
      "step": 782
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38908282077529777,
      "learning_rate": 8.013750954927426e-05,
      "loss": 1.8469,
      "step": 783
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.25293407657305794,
      "learning_rate": 8.011204481792718e-05,
      "loss": 1.8941,
      "step": 784
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3140019918389227,
      "learning_rate": 8.008658008658008e-05,
      "loss": 1.6617,
      "step": 785
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2559226154475287,
      "learning_rate": 8.0061115355233e-05,
      "loss": 1.6582,
      "step": 786
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.26485367590384346,
      "learning_rate": 8.003565062388593e-05,
      "loss": 1.9837,
      "step": 787
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.25715533221066317,
      "learning_rate": 8.001018589253883e-05,
      "loss": 1.6776,
      "step": 788
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.27671433607358126,
      "learning_rate": 7.998472116119176e-05,
      "loss": 2.0062,
      "step": 789
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2609719967453421,
      "learning_rate": 7.995925642984467e-05,
      "loss": 2.0136,
      "step": 790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.26557595136731144,
      "learning_rate": 7.993379169849758e-05,
      "loss": 1.9437,
      "step": 791
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2711309423960604,
      "learning_rate": 7.99083269671505e-05,
      "loss": 1.9532,
      "step": 792
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2546122638764411,
      "learning_rate": 7.988286223580342e-05,
      "loss": 1.8557,
      "step": 793
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2632866906982076,
      "learning_rate": 7.985739750445633e-05,
      "loss": 1.935,
      "step": 794
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.28892742042280245,
      "learning_rate": 7.983193277310926e-05,
      "loss": 2.0196,
      "step": 795
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2155105819372084,
      "learning_rate": 7.980646804176215e-05,
      "loss": 1.5858,
      "step": 796
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.23811179529959206,
      "learning_rate": 7.978100331041508e-05,
      "loss": 1.9028,
      "step": 797
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22173199477634228,
      "learning_rate": 7.975553857906799e-05,
      "loss": 1.874,
      "step": 798
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2670009960404223,
      "learning_rate": 7.97300738477209e-05,
      "loss": 1.8718,
      "step": 799
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24864604252498942,
      "learning_rate": 7.970460911637383e-05,
      "loss": 2.1092,
      "step": 800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22167578874428115,
      "learning_rate": 7.967914438502674e-05,
      "loss": 1.7823,
      "step": 801
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2647763467499256,
      "learning_rate": 7.965367965367965e-05,
      "loss": 1.8657,
      "step": 802
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32196599729557707,
      "learning_rate": 7.962821492233258e-05,
      "loss": 1.9988,
      "step": 803
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24083791298757853,
      "learning_rate": 7.960275019098549e-05,
      "loss": 1.6123,
      "step": 804
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2793392736091423,
      "learning_rate": 7.95772854596384e-05,
      "loss": 1.9157,
      "step": 805
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23601069294013594,
      "learning_rate": 7.955182072829132e-05,
      "loss": 1.8695,
      "step": 806
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2372446337405536,
      "learning_rate": 7.952635599694424e-05,
      "loss": 1.9903,
      "step": 807
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2515041727957943,
      "learning_rate": 7.950089126559715e-05,
      "loss": 1.9621,
      "step": 808
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2531873061180079,
      "learning_rate": 7.947542653425007e-05,
      "loss": 1.9211,
      "step": 809
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21761360632504378,
      "learning_rate": 7.944996180290298e-05,
      "loss": 1.5598,
      "step": 810
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.22808846056954318,
      "learning_rate": 7.94244970715559e-05,
      "loss": 1.7554,
      "step": 811
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23077816317385094,
      "learning_rate": 7.939903234020882e-05,
      "loss": 1.5468,
      "step": 812
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2215089182231589,
      "learning_rate": 7.937356760886173e-05,
      "loss": 1.652,
      "step": 813
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23311367779561595,
      "learning_rate": 7.934810287751464e-05,
      "loss": 1.7622,
      "step": 814
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2955128136695084,
      "learning_rate": 7.932263814616757e-05,
      "loss": 2.0124,
      "step": 815
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25773817493774576,
      "learning_rate": 7.929717341482048e-05,
      "loss": 2.0789,
      "step": 816
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34237355221347887,
      "learning_rate": 7.927170868347339e-05,
      "loss": 2.0829,
      "step": 817
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.28839753966260834,
      "learning_rate": 7.924624395212632e-05,
      "loss": 1.9994,
      "step": 818
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21212286265223146,
      "learning_rate": 7.922077922077923e-05,
      "loss": 1.7568,
      "step": 819
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2551081892350359,
      "learning_rate": 7.919531448943214e-05,
      "loss": 1.8875,
      "step": 820
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25490814024680464,
      "learning_rate": 7.916984975808505e-05,
      "loss": 1.8609,
      "step": 821
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2462200037724237,
      "learning_rate": 7.914438502673798e-05,
      "loss": 1.8756,
      "step": 822
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2495452947850181,
      "learning_rate": 7.911892029539089e-05,
      "loss": 1.987,
      "step": 823
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20413445933126373,
      "learning_rate": 7.90934555640438e-05,
      "loss": 1.5711,
      "step": 824
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2280167080933151,
      "learning_rate": 7.906799083269671e-05,
      "loss": 1.793,
      "step": 825
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.287282709750182,
      "learning_rate": 7.904252610134964e-05,
      "loss": 1.904,
      "step": 826
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.23544746439409528,
      "learning_rate": 7.901706137000255e-05,
      "loss": 1.7651,
      "step": 827
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2623434440061183,
      "learning_rate": 7.899159663865546e-05,
      "loss": 2.0431,
      "step": 828
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.29813009277728214,
      "learning_rate": 7.896613190730839e-05,
      "loss": 2.17,
      "step": 829
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24173203062575552,
      "learning_rate": 7.89406671759613e-05,
      "loss": 1.8406,
      "step": 830
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2366699237638536,
      "learning_rate": 7.891520244461421e-05,
      "loss": 1.8798,
      "step": 831
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2797406812007206,
      "learning_rate": 7.888973771326712e-05,
      "loss": 1.9852,
      "step": 832
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2099481977989102,
      "learning_rate": 7.886427298192004e-05,
      "loss": 1.871,
      "step": 833
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.22644585862553684,
      "learning_rate": 7.883880825057296e-05,
      "loss": 1.8052,
      "step": 834
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24565428898932076,
      "learning_rate": 7.881334351922587e-05,
      "loss": 2.1414,
      "step": 835
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2989449989300256,
      "learning_rate": 7.878787878787879e-05,
      "loss": 2.0666,
      "step": 836
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24542240881854216,
      "learning_rate": 7.876241405653171e-05,
      "loss": 1.954,
      "step": 837
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2672418304886155,
      "learning_rate": 7.873694932518462e-05,
      "loss": 2.0811,
      "step": 838
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2628618442827805,
      "learning_rate": 7.871148459383754e-05,
      "loss": 1.9559,
      "step": 839
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25650753303623913,
      "learning_rate": 7.868601986249046e-05,
      "loss": 2.2492,
      "step": 840
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.25022188413316765,
      "learning_rate": 7.866055513114336e-05,
      "loss": 2.0061,
      "step": 841
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2295783023375302,
      "learning_rate": 7.863509039979629e-05,
      "loss": 1.876,
      "step": 842
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3438795452503926,
      "learning_rate": 7.86096256684492e-05,
      "loss": 2.2212,
      "step": 843
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.20534516160381372,
      "learning_rate": 7.858416093710211e-05,
      "loss": 1.8619,
      "step": 844
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22899939201151417,
      "learning_rate": 7.855869620575504e-05,
      "loss": 1.7904,
      "step": 845
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.26192343948953833,
      "learning_rate": 7.853323147440795e-05,
      "loss": 1.8803,
      "step": 846
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2402411049000976,
      "learning_rate": 7.850776674306086e-05,
      "loss": 1.7429,
      "step": 847
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23155189023956613,
      "learning_rate": 7.848230201171379e-05,
      "loss": 1.7725,
      "step": 848
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24159559089848437,
      "learning_rate": 7.845683728036668e-05,
      "loss": 1.8425,
      "step": 849
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22127225800264405,
      "learning_rate": 7.843137254901961e-05,
      "loss": 1.6196,
      "step": 850
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2329639876433451,
      "learning_rate": 7.840590781767254e-05,
      "loss": 2.1553,
      "step": 851
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2841636941610758,
      "learning_rate": 7.838044308632543e-05,
      "loss": 2.0216,
      "step": 852
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23058339678000966,
      "learning_rate": 7.835497835497836e-05,
      "loss": 1.7883,
      "step": 853
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24237110500048858,
      "learning_rate": 7.832951362363129e-05,
      "loss": 1.8807,
      "step": 854
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24655649789483783,
      "learning_rate": 7.830404889228418e-05,
      "loss": 1.8507,
      "step": 855
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.28644569397987674,
      "learning_rate": 7.827858416093711e-05,
      "loss": 2.051,
      "step": 856
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23892097117174657,
      "learning_rate": 7.825311942959002e-05,
      "loss": 1.8824,
      "step": 857
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24085696112468405,
      "learning_rate": 7.822765469824293e-05,
      "loss": 1.6903,
      "step": 858
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23898549055005605,
      "learning_rate": 7.820218996689586e-05,
      "loss": 1.8986,
      "step": 859
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23549339168027972,
      "learning_rate": 7.817672523554876e-05,
      "loss": 1.8983,
      "step": 860
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2747127224535722,
      "learning_rate": 7.815126050420168e-05,
      "loss": 2.2242,
      "step": 861
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2970101274849217,
      "learning_rate": 7.812579577285461e-05,
      "loss": 2.0758,
      "step": 862
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2472828339287725,
      "learning_rate": 7.810033104150751e-05,
      "loss": 1.7372,
      "step": 863
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2623491807640201,
      "learning_rate": 7.807486631016043e-05,
      "loss": 1.6525,
      "step": 864
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2441195059224979,
      "learning_rate": 7.804940157881336e-05,
      "loss": 1.6852,
      "step": 865
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2572716038458357,
      "learning_rate": 7.802393684746626e-05,
      "loss": 2.002,
      "step": 866
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.19699396179900933,
      "learning_rate": 7.799847211611918e-05,
      "loss": 1.7233,
      "step": 867
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.20002757232567733,
      "learning_rate": 7.79730073847721e-05,
      "loss": 1.7233,
      "step": 868
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.25667728121646555,
      "learning_rate": 7.794754265342501e-05,
      "loss": 1.9715,
      "step": 869
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24319349057248757,
      "learning_rate": 7.792207792207793e-05,
      "loss": 1.7304,
      "step": 870
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24726911696006718,
      "learning_rate": 7.789661319073084e-05,
      "loss": 1.7162,
      "step": 871
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.28918886000356936,
      "learning_rate": 7.787114845938376e-05,
      "loss": 1.9583,
      "step": 872
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2577005200960701,
      "learning_rate": 7.784568372803668e-05,
      "loss": 1.9514,
      "step": 873
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24211230956641286,
      "learning_rate": 7.782021899668958e-05,
      "loss": 2.0073,
      "step": 874
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2986523229538993,
      "learning_rate": 7.779475426534251e-05,
      "loss": 1.9991,
      "step": 875
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2378687101520814,
      "learning_rate": 7.776928953399542e-05,
      "loss": 1.6956,
      "step": 876
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2448411504294662,
      "learning_rate": 7.774382480264833e-05,
      "loss": 1.8645,
      "step": 877
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.260326811865287,
      "learning_rate": 7.771836007130126e-05,
      "loss": 1.8771,
      "step": 878
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23711140143142848,
      "learning_rate": 7.769289533995417e-05,
      "loss": 1.7325,
      "step": 879
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.25312316152116904,
      "learning_rate": 7.766743060860708e-05,
      "loss": 1.8855,
      "step": 880
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2590127374184677,
      "learning_rate": 7.764196587726e-05,
      "loss": 1.7209,
      "step": 881
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24535401746192478,
      "learning_rate": 7.761650114591292e-05,
      "loss": 1.9677,
      "step": 882
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2612033865607243,
      "learning_rate": 7.759103641456583e-05,
      "loss": 1.9636,
      "step": 883
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2672248234251033,
      "learning_rate": 7.756557168321874e-05,
      "loss": 1.7793,
      "step": 884
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24691604804164172,
      "learning_rate": 7.754010695187165e-05,
      "loss": 2.0508,
      "step": 885
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24205617266569543,
      "learning_rate": 7.751464222052458e-05,
      "loss": 1.9556,
      "step": 886
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23950097320076194,
      "learning_rate": 7.748917748917749e-05,
      "loss": 1.8823,
      "step": 887
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23454313605150898,
      "learning_rate": 7.74637127578304e-05,
      "loss": 2.0957,
      "step": 888
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2518741276822946,
      "learning_rate": 7.743824802648333e-05,
      "loss": 1.984,
      "step": 889
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23913511906105006,
      "learning_rate": 7.741278329513624e-05,
      "loss": 1.6676,
      "step": 890
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23528027543073698,
      "learning_rate": 7.738731856378915e-05,
      "loss": 1.6761,
      "step": 891
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22746885632778302,
      "learning_rate": 7.736185383244207e-05,
      "loss": 1.9228,
      "step": 892
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2389799489749061,
      "learning_rate": 7.733638910109499e-05,
      "loss": 1.9622,
      "step": 893
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22927869240513388,
      "learning_rate": 7.73109243697479e-05,
      "loss": 1.7597,
      "step": 894
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24275816319261406,
      "learning_rate": 7.728545963840082e-05,
      "loss": 1.9038,
      "step": 895
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23010774399109485,
      "learning_rate": 7.725999490705373e-05,
      "loss": 1.9066,
      "step": 896
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.25068412991723865,
      "learning_rate": 7.723453017570665e-05,
      "loss": 1.863,
      "step": 897
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23238635392792073,
      "learning_rate": 7.720906544435957e-05,
      "loss": 1.7561,
      "step": 898
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22639256714411624,
      "learning_rate": 7.718360071301248e-05,
      "loss": 1.8246,
      "step": 899
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23030938279670504,
      "learning_rate": 7.71581359816654e-05,
      "loss": 1.936,
      "step": 900
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2617661305141856,
      "learning_rate": 7.713267125031832e-05,
      "loss": 1.7431,
      "step": 901
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24547250231249076,
      "learning_rate": 7.710720651897123e-05,
      "loss": 1.8947,
      "step": 902
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2328105609608234,
      "learning_rate": 7.708174178762414e-05,
      "loss": 1.7748,
      "step": 903
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2186369177899911,
      "learning_rate": 7.705627705627707e-05,
      "loss": 1.6294,
      "step": 904
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23367491909838675,
      "learning_rate": 7.703081232492998e-05,
      "loss": 1.8374,
      "step": 905
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.208741519321703,
      "learning_rate": 7.700534759358289e-05,
      "loss": 1.6709,
      "step": 906
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24236544870401433,
      "learning_rate": 7.69798828622358e-05,
      "loss": 1.911,
      "step": 907
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.26782105549040786,
      "learning_rate": 7.695441813088873e-05,
      "loss": 2.0718,
      "step": 908
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23612976419328677,
      "learning_rate": 7.692895339954164e-05,
      "loss": 1.8405,
      "step": 909
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24821660099782578,
      "learning_rate": 7.690348866819455e-05,
      "loss": 2.1939,
      "step": 910
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2448911954899524,
      "learning_rate": 7.687802393684746e-05,
      "loss": 2.1059,
      "step": 911
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23867366871261234,
      "learning_rate": 7.685255920550039e-05,
      "loss": 1.8553,
      "step": 912
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2541855083319077,
      "learning_rate": 7.68270944741533e-05,
      "loss": 1.7859,
      "step": 913
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3405341009591459,
      "learning_rate": 7.680162974280621e-05,
      "loss": 1.9631,
      "step": 914
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23780941759435273,
      "learning_rate": 7.677616501145914e-05,
      "loss": 1.8548,
      "step": 915
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2670959667222279,
      "learning_rate": 7.675070028011205e-05,
      "loss": 2.1335,
      "step": 916
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2149395295702038,
      "learning_rate": 7.672523554876496e-05,
      "loss": 1.7969,
      "step": 917
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.21035019919291034,
      "learning_rate": 7.669977081741789e-05,
      "loss": 1.7388,
      "step": 918
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.26288329988564973,
      "learning_rate": 7.667430608607079e-05,
      "loss": 2.0539,
      "step": 919
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22031332719255275,
      "learning_rate": 7.664884135472371e-05,
      "loss": 1.7581,
      "step": 920
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2339150206578499,
      "learning_rate": 7.662337662337662e-05,
      "loss": 1.74,
      "step": 921
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2909098439332173,
      "learning_rate": 7.659791189202954e-05,
      "loss": 2.0619,
      "step": 922
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.26884009714027207,
      "learning_rate": 7.657244716068246e-05,
      "loss": 1.7417,
      "step": 923
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2233060923878725,
      "learning_rate": 7.654698242933537e-05,
      "loss": 1.7564,
      "step": 924
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21819773852088117,
      "learning_rate": 7.652151769798829e-05,
      "loss": 1.8333,
      "step": 925
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2604109413789183,
      "learning_rate": 7.649605296664121e-05,
      "loss": 2.1359,
      "step": 926
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2632367252089379,
      "learning_rate": 7.647058823529411e-05,
      "loss": 1.8613,
      "step": 927
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2586751291202257,
      "learning_rate": 7.644512350394704e-05,
      "loss": 1.7107,
      "step": 928
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21636884551989266,
      "learning_rate": 7.641965877259996e-05,
      "loss": 2.013,
      "step": 929
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2280765292524814,
      "learning_rate": 7.639419404125286e-05,
      "loss": 1.9611,
      "step": 930
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24032350762980376,
      "learning_rate": 7.636872930990579e-05,
      "loss": 1.7703,
      "step": 931
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24868118103332415,
      "learning_rate": 7.63432645785587e-05,
      "loss": 1.9188,
      "step": 932
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19774779322088548,
      "learning_rate": 7.631779984721161e-05,
      "loss": 1.7019,
      "step": 933
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.23046990572106194,
      "learning_rate": 7.629233511586454e-05,
      "loss": 1.9484,
      "step": 934
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24550959723771484,
      "learning_rate": 7.626687038451745e-05,
      "loss": 1.7939,
      "step": 935
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2809890755314992,
      "learning_rate": 7.624140565317036e-05,
      "loss": 1.9596,
      "step": 936
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.31237752421726567,
      "learning_rate": 7.621594092182329e-05,
      "loss": 2.0557,
      "step": 937
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.23455083926617307,
      "learning_rate": 7.619047619047618e-05,
      "loss": 1.617,
      "step": 938
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24271565192298242,
      "learning_rate": 7.616501145912911e-05,
      "loss": 1.9186,
      "step": 939
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3234705161782687,
      "learning_rate": 7.613954672778204e-05,
      "loss": 2.2312,
      "step": 940
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.20938309397638855,
      "learning_rate": 7.611408199643493e-05,
      "loss": 1.8662,
      "step": 941
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.26719664786668124,
      "learning_rate": 7.608861726508786e-05,
      "loss": 1.9545,
      "step": 942
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24819446288048688,
      "learning_rate": 7.606315253374077e-05,
      "loss": 1.9658,
      "step": 943
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2756837137447019,
      "learning_rate": 7.603768780239368e-05,
      "loss": 1.7558,
      "step": 944
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.242656553492998,
      "learning_rate": 7.601222307104661e-05,
      "loss": 2.151,
      "step": 945
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.26971822970237047,
      "learning_rate": 7.598675833969952e-05,
      "loss": 1.9022,
      "step": 946
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2519256813420929,
      "learning_rate": 7.596129360835243e-05,
      "loss": 1.9729,
      "step": 947
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.29239166643883435,
      "learning_rate": 7.593582887700536e-05,
      "loss": 2.0671,
      "step": 948
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.25956244876566215,
      "learning_rate": 7.591036414565826e-05,
      "loss": 2.201,
      "step": 949
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.27114146689378416,
      "learning_rate": 7.588489941431118e-05,
      "loss": 2.095,
      "step": 950
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.25017787148928233,
      "learning_rate": 7.585943468296411e-05,
      "loss": 1.9677,
      "step": 951
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21798966454418292,
      "learning_rate": 7.583396995161701e-05,
      "loss": 1.9648,
      "step": 952
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3408898443872379,
      "learning_rate": 7.580850522026993e-05,
      "loss": 1.8321,
      "step": 953
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22591479387853222,
      "learning_rate": 7.578304048892285e-05,
      "loss": 1.9954,
      "step": 954
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2328924736172243,
      "learning_rate": 7.575757575757576e-05,
      "loss": 1.9178,
      "step": 955
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2173435089965625,
      "learning_rate": 7.573211102622868e-05,
      "loss": 1.7205,
      "step": 956
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2456469263841274,
      "learning_rate": 7.57066462948816e-05,
      "loss": 1.6118,
      "step": 957
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2887778298066372,
      "learning_rate": 7.568118156353451e-05,
      "loss": 2.1119,
      "step": 958
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.242980964697057,
      "learning_rate": 7.565571683218743e-05,
      "loss": 1.9016,
      "step": 959
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38571292428187637,
      "learning_rate": 7.563025210084033e-05,
      "loss": 2.0429,
      "step": 960
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.25195720580588754,
      "learning_rate": 7.560478736949326e-05,
      "loss": 1.8982,
      "step": 961
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.320485347539328,
      "learning_rate": 7.557932263814617e-05,
      "loss": 2.1381,
      "step": 962
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2251185492296154,
      "learning_rate": 7.555385790679908e-05,
      "loss": 1.8812,
      "step": 963
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2365123251612771,
      "learning_rate": 7.552839317545201e-05,
      "loss": 1.6989,
      "step": 964
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2251034485574401,
      "learning_rate": 7.550292844410492e-05,
      "loss": 1.5589,
      "step": 965
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1988388721881555,
      "learning_rate": 7.547746371275783e-05,
      "loss": 1.8252,
      "step": 966
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28923282244887305,
      "learning_rate": 7.545199898141076e-05,
      "loss": 2.1709,
      "step": 967
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2132909755096321,
      "learning_rate": 7.542653425006367e-05,
      "loss": 1.7329,
      "step": 968
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22508147837329856,
      "learning_rate": 7.540106951871658e-05,
      "loss": 1.5974,
      "step": 969
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2600908323181835,
      "learning_rate": 7.537560478736949e-05,
      "loss": 2.1381,
      "step": 970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.322659985241271,
      "learning_rate": 7.53501400560224e-05,
      "loss": 1.8803,
      "step": 971
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2474414841879655,
      "learning_rate": 7.532467532467533e-05,
      "loss": 1.8566,
      "step": 972
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.25302935076454597,
      "learning_rate": 7.529921059332824e-05,
      "loss": 1.9085,
      "step": 973
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22927825371223473,
      "learning_rate": 7.527374586198115e-05,
      "loss": 1.5849,
      "step": 974
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23052096987578194,
      "learning_rate": 7.524828113063408e-05,
      "loss": 1.7076,
      "step": 975
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22751792514838975,
      "learning_rate": 7.522281639928699e-05,
      "loss": 1.7537,
      "step": 976
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2147746581907753,
      "learning_rate": 7.51973516679399e-05,
      "loss": 1.741,
      "step": 977
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24550996140637407,
      "learning_rate": 7.517188693659283e-05,
      "loss": 1.784,
      "step": 978
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2880578625463606,
      "learning_rate": 7.514642220524574e-05,
      "loss": 2.117,
      "step": 979
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2504203808461966,
      "learning_rate": 7.512095747389865e-05,
      "loss": 1.7171,
      "step": 980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28360856072709995,
      "learning_rate": 7.509549274255157e-05,
      "loss": 1.8599,
      "step": 981
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.27146575369882403,
      "learning_rate": 7.507002801120448e-05,
      "loss": 1.873,
      "step": 982
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24146670322185135,
      "learning_rate": 7.50445632798574e-05,
      "loss": 1.6318,
      "step": 983
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22634116410765562,
      "learning_rate": 7.501909854851032e-05,
      "loss": 1.7617,
      "step": 984
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22952185094142133,
      "learning_rate": 7.499363381716323e-05,
      "loss": 1.8974,
      "step": 985
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.25222705764804615,
      "learning_rate": 7.496816908581615e-05,
      "loss": 1.6948,
      "step": 986
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.25059176978610787,
      "learning_rate": 7.494270435446907e-05,
      "loss": 1.8281,
      "step": 987
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23918773636957016,
      "learning_rate": 7.491723962312198e-05,
      "loss": 1.6135,
      "step": 988
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2602172596736344,
      "learning_rate": 7.489177489177489e-05,
      "loss": 1.861,
      "step": 989
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23004126099624014,
      "learning_rate": 7.486631016042782e-05,
      "loss": 1.9447,
      "step": 990
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22437029844262085,
      "learning_rate": 7.484084542908073e-05,
      "loss": 1.9113,
      "step": 991
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22822509947135397,
      "learning_rate": 7.481538069773364e-05,
      "loss": 1.7565,
      "step": 992
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22656387295800237,
      "learning_rate": 7.478991596638657e-05,
      "loss": 1.5697,
      "step": 993
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2272064577690763,
      "learning_rate": 7.476445123503948e-05,
      "loss": 1.8419,
      "step": 994
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2374931027013204,
      "learning_rate": 7.473898650369239e-05,
      "loss": 1.8254,
      "step": 995
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2804124733189521,
      "learning_rate": 7.47135217723453e-05,
      "loss": 1.9465,
      "step": 996
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.21066527991753883,
      "learning_rate": 7.468805704099821e-05,
      "loss": 1.9101,
      "step": 997
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24474048924987085,
      "learning_rate": 7.466259230965114e-05,
      "loss": 1.5658,
      "step": 998
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.22675961108484224,
      "learning_rate": 7.463712757830405e-05,
      "loss": 1.9399,
      "step": 999
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24537736830780577,
      "learning_rate": 7.461166284695696e-05,
      "loss": 1.9266,
      "step": 1000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2873054405840476,
      "learning_rate": 7.458619811560989e-05,
      "loss": 1.8692,
      "step": 1001
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.26074663079865634,
      "learning_rate": 7.45607333842628e-05,
      "loss": 1.6851,
      "step": 1002
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.29185094574968984,
      "learning_rate": 7.453526865291571e-05,
      "loss": 1.7844,
      "step": 1003
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2667254903977434,
      "learning_rate": 7.450980392156864e-05,
      "loss": 1.9728,
      "step": 1004
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.23175351028139793,
      "learning_rate": 7.448433919022154e-05,
      "loss": 1.8085,
      "step": 1005
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.23395886035315197,
      "learning_rate": 7.445887445887446e-05,
      "loss": 1.8175,
      "step": 1006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3028474658033109,
      "learning_rate": 7.443340972752738e-05,
      "loss": 2.0443,
      "step": 1007
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2580042039617424,
      "learning_rate": 7.440794499618029e-05,
      "loss": 1.898,
      "step": 1008
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20881989503877474,
      "learning_rate": 7.438248026483321e-05,
      "loss": 1.7365,
      "step": 1009
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21758219956532768,
      "learning_rate": 7.435701553348613e-05,
      "loss": 1.8344,
      "step": 1010
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30945228709449574,
      "learning_rate": 7.433155080213904e-05,
      "loss": 1.803,
      "step": 1011
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2778470482815492,
      "learning_rate": 7.430608607079196e-05,
      "loss": 1.9486,
      "step": 1012
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.27796491719991984,
      "learning_rate": 7.428062133944486e-05,
      "loss": 1.9901,
      "step": 1013
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2823655601272822,
      "learning_rate": 7.425515660809779e-05,
      "loss": 2.0025,
      "step": 1014
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21818991040931096,
      "learning_rate": 7.422969187675071e-05,
      "loss": 1.824,
      "step": 1015
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2402002884167473,
      "learning_rate": 7.420422714540361e-05,
      "loss": 1.8006,
      "step": 1016
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.23589217748546118,
      "learning_rate": 7.417876241405654e-05,
      "loss": 1.8656,
      "step": 1017
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.23281470528140585,
      "learning_rate": 7.415329768270945e-05,
      "loss": 1.5889,
      "step": 1018
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.29580881001459797,
      "learning_rate": 7.412783295136236e-05,
      "loss": 1.7995,
      "step": 1019
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.22434470297566775,
      "learning_rate": 7.410236822001529e-05,
      "loss": 1.8127,
      "step": 1020
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2914740516799641,
      "learning_rate": 7.40769034886682e-05,
      "loss": 2.2744,
      "step": 1021
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2742825121529459,
      "learning_rate": 7.405143875732111e-05,
      "loss": 1.9129,
      "step": 1022
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2334532812955576,
      "learning_rate": 7.402597402597404e-05,
      "loss": 1.7953,
      "step": 1023
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24967774841210677,
      "learning_rate": 7.400050929462693e-05,
      "loss": 1.8679,
      "step": 1024
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2595031961508815,
      "learning_rate": 7.397504456327986e-05,
      "loss": 1.9454,
      "step": 1025
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2521915933685689,
      "learning_rate": 7.394957983193279e-05,
      "loss": 2.1809,
      "step": 1026
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.26113880010800444,
      "learning_rate": 7.392411510058568e-05,
      "loss": 1.7928,
      "step": 1027
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24503990870092038,
      "learning_rate": 7.389865036923861e-05,
      "loss": 1.8939,
      "step": 1028
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.22210336940722358,
      "learning_rate": 7.387318563789152e-05,
      "loss": 1.7962,
      "step": 1029
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2868276758158762,
      "learning_rate": 7.384772090654443e-05,
      "loss": 2.1045,
      "step": 1030
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24116223727815847,
      "learning_rate": 7.382225617519736e-05,
      "loss": 1.7254,
      "step": 1031
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2789040316966339,
      "learning_rate": 7.379679144385027e-05,
      "loss": 1.9977,
      "step": 1032
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24868935259042466,
      "learning_rate": 7.377132671250318e-05,
      "loss": 1.7413,
      "step": 1033
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2558503040132623,
      "learning_rate": 7.374586198115611e-05,
      "loss": 1.9594,
      "step": 1034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2372110362556985,
      "learning_rate": 7.372039724980901e-05,
      "loss": 1.6742,
      "step": 1035
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2215575331170209,
      "learning_rate": 7.369493251846193e-05,
      "loss": 1.7478,
      "step": 1036
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.29277068754588087,
      "learning_rate": 7.366946778711486e-05,
      "loss": 1.8855,
      "step": 1037
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.247853489079965,
      "learning_rate": 7.364400305576776e-05,
      "loss": 1.9515,
      "step": 1038
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3179033441781071,
      "learning_rate": 7.361853832442068e-05,
      "loss": 1.9779,
      "step": 1039
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21068073469156284,
      "learning_rate": 7.35930735930736e-05,
      "loss": 1.7137,
      "step": 1040
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31388598886043856,
      "learning_rate": 7.356760886172651e-05,
      "loss": 1.9984,
      "step": 1041
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26587855635014485,
      "learning_rate": 7.354214413037943e-05,
      "loss": 1.9973,
      "step": 1042
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2843791746524781,
      "learning_rate": 7.351667939903235e-05,
      "loss": 2.2078,
      "step": 1043
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22613839212935044,
      "learning_rate": 7.349121466768526e-05,
      "loss": 2.0572,
      "step": 1044
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24078092198779658,
      "learning_rate": 7.346574993633818e-05,
      "loss": 2.0019,
      "step": 1045
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22878717117869515,
      "learning_rate": 7.344028520499108e-05,
      "loss": 2.0114,
      "step": 1046
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.28529572668418474,
      "learning_rate": 7.341482047364401e-05,
      "loss": 2.0328,
      "step": 1047
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3037342908902211,
      "learning_rate": 7.338935574229692e-05,
      "loss": 1.9793,
      "step": 1048
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2976567866603389,
      "learning_rate": 7.336389101094983e-05,
      "loss": 2.0043,
      "step": 1049
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22623327604414598,
      "learning_rate": 7.333842627960276e-05,
      "loss": 1.8501,
      "step": 1050
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22202210720923746,
      "learning_rate": 7.331296154825567e-05,
      "loss": 1.7602,
      "step": 1051
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.25556327342706764,
      "learning_rate": 7.328749681690858e-05,
      "loss": 1.9428,
      "step": 1052
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30017478291905664,
      "learning_rate": 7.326203208556151e-05,
      "loss": 1.9114,
      "step": 1053
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2530886618115332,
      "learning_rate": 7.323656735421442e-05,
      "loss": 1.6587,
      "step": 1054
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26379455255653006,
      "learning_rate": 7.321110262286733e-05,
      "loss": 1.8647,
      "step": 1055
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30701511136802717,
      "learning_rate": 7.318563789152026e-05,
      "loss": 2.1388,
      "step": 1056
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2851724881291049,
      "learning_rate": 7.316017316017317e-05,
      "loss": 1.888,
      "step": 1057
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26338541930182763,
      "learning_rate": 7.313470842882608e-05,
      "loss": 2.1868,
      "step": 1058
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24415074899663258,
      "learning_rate": 7.310924369747899e-05,
      "loss": 1.6828,
      "step": 1059
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2765224546330694,
      "learning_rate": 7.30837789661319e-05,
      "loss": 1.9021,
      "step": 1060
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26298709533401354,
      "learning_rate": 7.305831423478483e-05,
      "loss": 1.7669,
      "step": 1061
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21868878597983632,
      "learning_rate": 7.303284950343774e-05,
      "loss": 1.8156,
      "step": 1062
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2383054502111241,
      "learning_rate": 7.300738477209065e-05,
      "loss": 1.8936,
      "step": 1063
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3029936387854745,
      "learning_rate": 7.298192004074358e-05,
      "loss": 2.1997,
      "step": 1064
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2240249078331532,
      "learning_rate": 7.295645530939649e-05,
      "loss": 1.8388,
      "step": 1065
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30748644275552534,
      "learning_rate": 7.29309905780494e-05,
      "loss": 2.1551,
      "step": 1066
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31235704728599206,
      "learning_rate": 7.290552584670232e-05,
      "loss": 2.044,
      "step": 1067
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24152539871409162,
      "learning_rate": 7.288006111535524e-05,
      "loss": 1.8208,
      "step": 1068
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26192652582359693,
      "learning_rate": 7.285459638400815e-05,
      "loss": 1.9027,
      "step": 1069
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2731294531644669,
      "learning_rate": 7.282913165266107e-05,
      "loss": 1.9331,
      "step": 1070
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.25505936744970453,
      "learning_rate": 7.280366692131398e-05,
      "loss": 1.903,
      "step": 1071
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2282500476257383,
      "learning_rate": 7.27782021899669e-05,
      "loss": 1.7402,
      "step": 1072
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2620146487850985,
      "learning_rate": 7.275273745861982e-05,
      "loss": 1.9199,
      "step": 1073
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23315122911921574,
      "learning_rate": 7.272727272727273e-05,
      "loss": 1.7629,
      "step": 1074
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23948199609946433,
      "learning_rate": 7.270180799592564e-05,
      "loss": 1.7816,
      "step": 1075
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21465318204237038,
      "learning_rate": 7.267634326457857e-05,
      "loss": 1.7512,
      "step": 1076
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.27432777704428774,
      "learning_rate": 7.265087853323148e-05,
      "loss": 1.8063,
      "step": 1077
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2613009919134964,
      "learning_rate": 7.262541380188439e-05,
      "loss": 1.9957,
      "step": 1078
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24222462892491128,
      "learning_rate": 7.259994907053732e-05,
      "loss": 1.9051,
      "step": 1079
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23796910023553916,
      "learning_rate": 7.257448433919023e-05,
      "loss": 1.6754,
      "step": 1080
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27225403124782055,
      "learning_rate": 7.254901960784314e-05,
      "loss": 1.9675,
      "step": 1081
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23667078161663896,
      "learning_rate": 7.252355487649605e-05,
      "loss": 1.8707,
      "step": 1082
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24004472053880488,
      "learning_rate": 7.249809014514896e-05,
      "loss": 1.6981,
      "step": 1083
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2393421569384702,
      "learning_rate": 7.247262541380189e-05,
      "loss": 1.8638,
      "step": 1084
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24813366670630538,
      "learning_rate": 7.24471606824548e-05,
      "loss": 2.007,
      "step": 1085
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.25643140605516485,
      "learning_rate": 7.242169595110771e-05,
      "loss": 2.1365,
      "step": 1086
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2989812844911387,
      "learning_rate": 7.239623121976064e-05,
      "loss": 1.7626,
      "step": 1087
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2979038380259418,
      "learning_rate": 7.237076648841355e-05,
      "loss": 2.0305,
      "step": 1088
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24981192130202395,
      "learning_rate": 7.234530175706646e-05,
      "loss": 1.6851,
      "step": 1089
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24425436040945306,
      "learning_rate": 7.231983702571939e-05,
      "loss": 1.8165,
      "step": 1090
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2700889456790734,
      "learning_rate": 7.229437229437229e-05,
      "loss": 2.0746,
      "step": 1091
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3365204875725978,
      "learning_rate": 7.226890756302521e-05,
      "loss": 1.7851,
      "step": 1092
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2675215111309902,
      "learning_rate": 7.224344283167813e-05,
      "loss": 1.8756,
      "step": 1093
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.25048149947416654,
      "learning_rate": 7.221797810033104e-05,
      "loss": 1.7192,
      "step": 1094
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23477026669805706,
      "learning_rate": 7.219251336898396e-05,
      "loss": 1.859,
      "step": 1095
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.28073662687213674,
      "learning_rate": 7.216704863763688e-05,
      "loss": 1.8868,
      "step": 1096
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2989689364308601,
      "learning_rate": 7.214158390628979e-05,
      "loss": 1.8827,
      "step": 1097
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2810212105436412,
      "learning_rate": 7.211611917494271e-05,
      "loss": 2.016,
      "step": 1098
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2543432436622935,
      "learning_rate": 7.209065444359563e-05,
      "loss": 1.7812,
      "step": 1099
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3613659400741098,
      "learning_rate": 7.206518971224854e-05,
      "loss": 2.0044,
      "step": 1100
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.22297225245001823,
      "learning_rate": 7.203972498090146e-05,
      "loss": 1.6883,
      "step": 1101
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23228286946036356,
      "learning_rate": 7.201426024955436e-05,
      "loss": 1.8655,
      "step": 1102
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.25222815059535875,
      "learning_rate": 7.198879551820729e-05,
      "loss": 1.9635,
      "step": 1103
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26860423237937797,
      "learning_rate": 7.196333078686021e-05,
      "loss": 2.0223,
      "step": 1104
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.22766323548846765,
      "learning_rate": 7.193786605551311e-05,
      "loss": 1.8319,
      "step": 1105
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26320673570112335,
      "learning_rate": 7.191240132416604e-05,
      "loss": 1.8235,
      "step": 1106
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2472466981342383,
      "learning_rate": 7.188693659281895e-05,
      "loss": 1.8836,
      "step": 1107
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2930360716717708,
      "learning_rate": 7.186147186147186e-05,
      "loss": 2.036,
      "step": 1108
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29167954666218077,
      "learning_rate": 7.183600713012479e-05,
      "loss": 1.9315,
      "step": 1109
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2794854387898721,
      "learning_rate": 7.181054239877769e-05,
      "loss": 1.9765,
      "step": 1110
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23546131629828576,
      "learning_rate": 7.178507766743061e-05,
      "loss": 1.7479,
      "step": 1111
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2848010921499858,
      "learning_rate": 7.175961293608354e-05,
      "loss": 2.0008,
      "step": 1112
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23285082495229736,
      "learning_rate": 7.173414820473643e-05,
      "loss": 1.7227,
      "step": 1113
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23277238653860835,
      "learning_rate": 7.170868347338936e-05,
      "loss": 1.7846,
      "step": 1114
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24855975945486786,
      "learning_rate": 7.168321874204229e-05,
      "loss": 1.8656,
      "step": 1115
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23104977509797836,
      "learning_rate": 7.165775401069518e-05,
      "loss": 1.84,
      "step": 1116
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2543793452264319,
      "learning_rate": 7.163228927934811e-05,
      "loss": 1.9373,
      "step": 1117
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29724953265750936,
      "learning_rate": 7.160682454800102e-05,
      "loss": 2.0002,
      "step": 1118
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.25150347662887024,
      "learning_rate": 7.158135981665393e-05,
      "loss": 2.0944,
      "step": 1119
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30103818645027586,
      "learning_rate": 7.155589508530686e-05,
      "loss": 2.1487,
      "step": 1120
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2686274065919044,
      "learning_rate": 7.153043035395977e-05,
      "loss": 2.0224,
      "step": 1121
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2566219497226242,
      "learning_rate": 7.150496562261268e-05,
      "loss": 1.6538,
      "step": 1122
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25612670095278256,
      "learning_rate": 7.147950089126561e-05,
      "loss": 1.7372,
      "step": 1123
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2550233621507488,
      "learning_rate": 7.145403615991851e-05,
      "loss": 2.0321,
      "step": 1124
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24179696312626972,
      "learning_rate": 7.142857142857143e-05,
      "loss": 1.7023,
      "step": 1125
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2377412493034981,
      "learning_rate": 7.140310669722435e-05,
      "loss": 1.9684,
      "step": 1126
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23372052525767906,
      "learning_rate": 7.137764196587726e-05,
      "loss": 1.787,
      "step": 1127
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1985849320909872,
      "learning_rate": 7.135217723453018e-05,
      "loss": 1.716,
      "step": 1128
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2725190541622667,
      "learning_rate": 7.13267125031831e-05,
      "loss": 2.0011,
      "step": 1129
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.20734841865394643,
      "learning_rate": 7.130124777183601e-05,
      "loss": 1.7306,
      "step": 1130
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24358986338218003,
      "learning_rate": 7.127578304048893e-05,
      "loss": 1.655,
      "step": 1131
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2673702685671822,
      "learning_rate": 7.125031830914185e-05,
      "loss": 2.3778,
      "step": 1132
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2530655072787947,
      "learning_rate": 7.122485357779476e-05,
      "loss": 1.8649,
      "step": 1133
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32523596569646346,
      "learning_rate": 7.119938884644768e-05,
      "loss": 2.0269,
      "step": 1134
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.26169970428856776,
      "learning_rate": 7.117392411510058e-05,
      "loss": 1.8534,
      "step": 1135
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3037747135234151,
      "learning_rate": 7.114845938375351e-05,
      "loss": 2.1786,
      "step": 1136
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2299678580355851,
      "learning_rate": 7.112299465240642e-05,
      "loss": 1.8511,
      "step": 1137
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23206102584752836,
      "learning_rate": 7.109752992105933e-05,
      "loss": 1.9227,
      "step": 1138
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2537273657151455,
      "learning_rate": 7.107206518971226e-05,
      "loss": 1.8559,
      "step": 1139
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23767428403068216,
      "learning_rate": 7.104660045836517e-05,
      "loss": 2.222,
      "step": 1140
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23592912078024372,
      "learning_rate": 7.102113572701808e-05,
      "loss": 1.8312,
      "step": 1141
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30055754876376845,
      "learning_rate": 7.099567099567101e-05,
      "loss": 1.8955,
      "step": 1142
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.265817656493249,
      "learning_rate": 7.097020626432392e-05,
      "loss": 1.9888,
      "step": 1143
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25581988546790446,
      "learning_rate": 7.094474153297683e-05,
      "loss": 2.0585,
      "step": 1144
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2462609495110378,
      "learning_rate": 7.091927680162974e-05,
      "loss": 1.7901,
      "step": 1145
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2748599112846291,
      "learning_rate": 7.089381207028266e-05,
      "loss": 1.9851,
      "step": 1146
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2309607894623946,
      "learning_rate": 7.086834733893558e-05,
      "loss": 2.0558,
      "step": 1147
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24386882147790764,
      "learning_rate": 7.084288260758849e-05,
      "loss": 1.7379,
      "step": 1148
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.27708527784214965,
      "learning_rate": 7.08174178762414e-05,
      "loss": 1.8403,
      "step": 1149
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3354518396089165,
      "learning_rate": 7.079195314489433e-05,
      "loss": 1.7877,
      "step": 1150
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22490205354791917,
      "learning_rate": 7.076648841354724e-05,
      "loss": 1.6953,
      "step": 1151
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25284629536951586,
      "learning_rate": 7.074102368220015e-05,
      "loss": 1.8155,
      "step": 1152
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2068117165364758,
      "learning_rate": 7.071555895085307e-05,
      "loss": 1.6563,
      "step": 1153
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22447663406807988,
      "learning_rate": 7.069009421950599e-05,
      "loss": 1.8194,
      "step": 1154
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23613657163494348,
      "learning_rate": 7.06646294881589e-05,
      "loss": 1.9006,
      "step": 1155
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24702024894989047,
      "learning_rate": 7.063916475681182e-05,
      "loss": 2.1066,
      "step": 1156
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3056091771193659,
      "learning_rate": 7.061370002546473e-05,
      "loss": 1.9301,
      "step": 1157
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2747597332958069,
      "learning_rate": 7.058823529411765e-05,
      "loss": 2.1646,
      "step": 1158
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24583240285929006,
      "learning_rate": 7.056277056277057e-05,
      "loss": 1.9277,
      "step": 1159
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.331506392830848,
      "learning_rate": 7.053730583142348e-05,
      "loss": 2.0941,
      "step": 1160
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3055671318828273,
      "learning_rate": 7.051184110007639e-05,
      "loss": 1.7977,
      "step": 1161
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23832312200391165,
      "learning_rate": 7.048637636872932e-05,
      "loss": 1.8091,
      "step": 1162
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24203766525563156,
      "learning_rate": 7.046091163738223e-05,
      "loss": 1.7537,
      "step": 1163
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2523149952286734,
      "learning_rate": 7.043544690603514e-05,
      "loss": 1.7045,
      "step": 1164
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24369301007467567,
      "learning_rate": 7.040998217468807e-05,
      "loss": 1.7281,
      "step": 1165
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2259776689902922,
      "learning_rate": 7.038451744334098e-05,
      "loss": 1.5897,
      "step": 1166
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2994312483004138,
      "learning_rate": 7.035905271199389e-05,
      "loss": 1.9248,
      "step": 1167
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2526797086621746,
      "learning_rate": 7.03335879806468e-05,
      "loss": 1.7435,
      "step": 1168
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23277674758895478,
      "learning_rate": 7.030812324929971e-05,
      "loss": 1.72,
      "step": 1169
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26770775154189086,
      "learning_rate": 7.028265851795264e-05,
      "loss": 1.7953,
      "step": 1170
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2760582265145269,
      "learning_rate": 7.025719378660555e-05,
      "loss": 1.9682,
      "step": 1171
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2984211962176011,
      "learning_rate": 7.023172905525846e-05,
      "loss": 1.8924,
      "step": 1172
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2490923393519683,
      "learning_rate": 7.020626432391139e-05,
      "loss": 2.0529,
      "step": 1173
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23055860416536486,
      "learning_rate": 7.01807995925643e-05,
      "loss": 2.0508,
      "step": 1174
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24071860164918554,
      "learning_rate": 7.015533486121721e-05,
      "loss": 1.7432,
      "step": 1175
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23464291677367424,
      "learning_rate": 7.012987012987014e-05,
      "loss": 1.8712,
      "step": 1176
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.25136838377673104,
      "learning_rate": 7.010440539852305e-05,
      "loss": 1.8357,
      "step": 1177
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23352658572465643,
      "learning_rate": 7.007894066717596e-05,
      "loss": 1.9611,
      "step": 1178
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2741285583327767,
      "learning_rate": 7.005347593582889e-05,
      "loss": 1.9042,
      "step": 1179
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23900251196893887,
      "learning_rate": 7.002801120448179e-05,
      "loss": 1.9515,
      "step": 1180
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2060822834500915,
      "learning_rate": 7.000254647313471e-05,
      "loss": 1.6298,
      "step": 1181
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23316114325732196,
      "learning_rate": 6.997708174178763e-05,
      "loss": 1.6614,
      "step": 1182
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26262615669137385,
      "learning_rate": 6.995161701044054e-05,
      "loss": 1.9487,
      "step": 1183
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2601703130226013,
      "learning_rate": 6.992615227909346e-05,
      "loss": 1.9944,
      "step": 1184
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2698487333136885,
      "learning_rate": 6.990068754774638e-05,
      "loss": 2.1823,
      "step": 1185
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2781873686650244,
      "learning_rate": 6.987522281639929e-05,
      "loss": 2.2884,
      "step": 1186
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2159810309276435,
      "learning_rate": 6.984975808505221e-05,
      "loss": 1.5913,
      "step": 1187
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.28527333194417825,
      "learning_rate": 6.982429335370511e-05,
      "loss": 1.8822,
      "step": 1188
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23135515863468417,
      "learning_rate": 6.979882862235804e-05,
      "loss": 1.7036,
      "step": 1189
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24407147763453632,
      "learning_rate": 6.977336389101096e-05,
      "loss": 1.7656,
      "step": 1190
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27642317565367014,
      "learning_rate": 6.974789915966386e-05,
      "loss": 1.8132,
      "step": 1191
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3566106334497433,
      "learning_rate": 6.972243442831679e-05,
      "loss": 2.1099,
      "step": 1192
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3242742766665087,
      "learning_rate": 6.96969696969697e-05,
      "loss": 2.0686,
      "step": 1193
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2382209576481061,
      "learning_rate": 6.967150496562261e-05,
      "loss": 1.6727,
      "step": 1194
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24684876471560604,
      "learning_rate": 6.964604023427554e-05,
      "loss": 1.9016,
      "step": 1195
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2424328622470283,
      "learning_rate": 6.962057550292845e-05,
      "loss": 1.6665,
      "step": 1196
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.21481593558855225,
      "learning_rate": 6.959511077158136e-05,
      "loss": 1.5815,
      "step": 1197
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26915990430990216,
      "learning_rate": 6.956964604023429e-05,
      "loss": 2.0001,
      "step": 1198
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24361440787524696,
      "learning_rate": 6.954418130888719e-05,
      "loss": 1.7438,
      "step": 1199
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.25051539164691516,
      "learning_rate": 6.951871657754011e-05,
      "loss": 1.7314,
      "step": 1200
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2484958370261083,
      "learning_rate": 6.949325184619304e-05,
      "loss": 1.9663,
      "step": 1201
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.255854613864076,
      "learning_rate": 6.946778711484593e-05,
      "loss": 1.9106,
      "step": 1202
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22296479246373885,
      "learning_rate": 6.944232238349886e-05,
      "loss": 1.753,
      "step": 1203
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2380491361269653,
      "learning_rate": 6.941685765215177e-05,
      "loss": 1.5306,
      "step": 1204
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2523746365284298,
      "learning_rate": 6.939139292080468e-05,
      "loss": 1.7611,
      "step": 1205
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2216426297441028,
      "learning_rate": 6.936592818945761e-05,
      "loss": 1.8845,
      "step": 1206
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.21847890335976045,
      "learning_rate": 6.934046345811052e-05,
      "loss": 1.6605,
      "step": 1207
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.26162861937449305,
      "learning_rate": 6.931499872676343e-05,
      "loss": 2.0477,
      "step": 1208
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2900063467358922,
      "learning_rate": 6.928953399541636e-05,
      "loss": 1.9684,
      "step": 1209
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22609536749703432,
      "learning_rate": 6.926406926406926e-05,
      "loss": 1.6459,
      "step": 1210
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29145205498250837,
      "learning_rate": 6.923860453272218e-05,
      "loss": 2.1166,
      "step": 1211
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2553795253317233,
      "learning_rate": 6.921313980137511e-05,
      "loss": 1.9199,
      "step": 1212
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.25504394348961323,
      "learning_rate": 6.918767507002801e-05,
      "loss": 1.9561,
      "step": 1213
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29217266004279324,
      "learning_rate": 6.916221033868093e-05,
      "loss": 2.0036,
      "step": 1214
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23271868690970354,
      "learning_rate": 6.913674560733385e-05,
      "loss": 1.7754,
      "step": 1215
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2711700568417082,
      "learning_rate": 6.911128087598676e-05,
      "loss": 1.987,
      "step": 1216
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.25555659717339474,
      "learning_rate": 6.908581614463968e-05,
      "loss": 1.8564,
      "step": 1217
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23641129283480852,
      "learning_rate": 6.90603514132926e-05,
      "loss": 2.0455,
      "step": 1218
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2748935081457799,
      "learning_rate": 6.903488668194551e-05,
      "loss": 2.0183,
      "step": 1219
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.236873454665123,
      "learning_rate": 6.900942195059843e-05,
      "loss": 1.728,
      "step": 1220
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1917179741723266,
      "learning_rate": 6.898395721925133e-05,
      "loss": 1.5001,
      "step": 1221
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2337431576563751,
      "learning_rate": 6.895849248790426e-05,
      "loss": 1.9901,
      "step": 1222
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2622288813801724,
      "learning_rate": 6.893302775655717e-05,
      "loss": 1.8896,
      "step": 1223
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.27413476869769615,
      "learning_rate": 6.890756302521008e-05,
      "loss": 1.801,
      "step": 1224
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30057635078237704,
      "learning_rate": 6.888209829386301e-05,
      "loss": 2.0947,
      "step": 1225
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23108785525032752,
      "learning_rate": 6.885663356251592e-05,
      "loss": 1.7747,
      "step": 1226
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.26140156784470325,
      "learning_rate": 6.883116883116883e-05,
      "loss": 1.9075,
      "step": 1227
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.25422769885968144,
      "learning_rate": 6.880570409982176e-05,
      "loss": 1.9126,
      "step": 1228
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24198688381192876,
      "learning_rate": 6.878023936847467e-05,
      "loss": 1.7064,
      "step": 1229
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22749023568504842,
      "learning_rate": 6.875477463712758e-05,
      "loss": 1.5782,
      "step": 1230
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24970716223565978,
      "learning_rate": 6.87293099057805e-05,
      "loss": 1.7543,
      "step": 1231
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24069863716059806,
      "learning_rate": 6.87038451744334e-05,
      "loss": 1.9083,
      "step": 1232
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.21938296261489496,
      "learning_rate": 6.867838044308633e-05,
      "loss": 1.8046,
      "step": 1233
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29344740866119606,
      "learning_rate": 6.865291571173924e-05,
      "loss": 2.2228,
      "step": 1234
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2403872476488659,
      "learning_rate": 6.862745098039216e-05,
      "loss": 1.8867,
      "step": 1235
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20991083478840128,
      "learning_rate": 6.860198624904508e-05,
      "loss": 1.5654,
      "step": 1236
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24375700757150615,
      "learning_rate": 6.857652151769799e-05,
      "loss": 1.8848,
      "step": 1237
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21450749873567568,
      "learning_rate": 6.85510567863509e-05,
      "loss": 1.9196,
      "step": 1238
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.28072297202065793,
      "learning_rate": 6.852559205500382e-05,
      "loss": 1.9887,
      "step": 1239
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2647998278270836,
      "learning_rate": 6.850012732365674e-05,
      "loss": 1.8879,
      "step": 1240
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31102485579797085,
      "learning_rate": 6.847466259230965e-05,
      "loss": 1.8028,
      "step": 1241
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2752105042682146,
      "learning_rate": 6.844919786096257e-05,
      "loss": 1.9422,
      "step": 1242
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2793974395248586,
      "learning_rate": 6.842373312961549e-05,
      "loss": 1.9473,
      "step": 1243
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.291715101353079,
      "learning_rate": 6.83982683982684e-05,
      "loss": 2.0187,
      "step": 1244
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25682453666882127,
      "learning_rate": 6.837280366692132e-05,
      "loss": 2.0103,
      "step": 1245
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.26515443577592535,
      "learning_rate": 6.834733893557423e-05,
      "loss": 1.7907,
      "step": 1246
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25147888736953206,
      "learning_rate": 6.832187420422714e-05,
      "loss": 1.863,
      "step": 1247
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2366004509763539,
      "learning_rate": 6.829640947288007e-05,
      "loss": 1.8877,
      "step": 1248
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22844394961748615,
      "learning_rate": 6.827094474153298e-05,
      "loss": 1.8736,
      "step": 1249
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2543467149075416,
      "learning_rate": 6.824548001018589e-05,
      "loss": 2.0736,
      "step": 1250
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25818288091804553,
      "learning_rate": 6.822001527883882e-05,
      "loss": 1.8744,
      "step": 1251
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23945688023819678,
      "learning_rate": 6.819455054749173e-05,
      "loss": 1.9383,
      "step": 1252
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2378999128861847,
      "learning_rate": 6.816908581614464e-05,
      "loss": 1.8548,
      "step": 1253
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21585408977785395,
      "learning_rate": 6.814362108479757e-05,
      "loss": 1.7908,
      "step": 1254
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2908614089243041,
      "learning_rate": 6.811815635345048e-05,
      "loss": 2.259,
      "step": 1255
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25735586359063733,
      "learning_rate": 6.809269162210339e-05,
      "loss": 2.001,
      "step": 1256
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29593359861047985,
      "learning_rate": 6.80672268907563e-05,
      "loss": 2.1095,
      "step": 1257
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2446240179064592,
      "learning_rate": 6.804176215940921e-05,
      "loss": 1.9325,
      "step": 1258
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2967514609234207,
      "learning_rate": 6.801629742806214e-05,
      "loss": 1.9994,
      "step": 1259
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25424482806186377,
      "learning_rate": 6.799083269671505e-05,
      "loss": 1.8767,
      "step": 1260
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21837488436682204,
      "learning_rate": 6.796536796536796e-05,
      "loss": 1.7035,
      "step": 1261
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2392078035990844,
      "learning_rate": 6.793990323402089e-05,
      "loss": 1.8737,
      "step": 1262
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.27081493474137447,
      "learning_rate": 6.79144385026738e-05,
      "loss": 1.8451,
      "step": 1263
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2154631615481794,
      "learning_rate": 6.788897377132671e-05,
      "loss": 1.8449,
      "step": 1264
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22887887367874743,
      "learning_rate": 6.786350903997964e-05,
      "loss": 1.934,
      "step": 1265
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2634533435722663,
      "learning_rate": 6.783804430863254e-05,
      "loss": 1.7119,
      "step": 1266
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23048658625232663,
      "learning_rate": 6.781257957728546e-05,
      "loss": 1.923,
      "step": 1267
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2676248022423503,
      "learning_rate": 6.778711484593838e-05,
      "loss": 2.1183,
      "step": 1268
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24552379182876227,
      "learning_rate": 6.776165011459129e-05,
      "loss": 2.1152,
      "step": 1269
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24533812000343835,
      "learning_rate": 6.773618538324421e-05,
      "loss": 2.067,
      "step": 1270
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22815212454290484,
      "learning_rate": 6.771072065189713e-05,
      "loss": 1.6645,
      "step": 1271
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22504858942695816,
      "learning_rate": 6.768525592055004e-05,
      "loss": 1.671,
      "step": 1272
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25821809941671164,
      "learning_rate": 6.765979118920296e-05,
      "loss": 1.9414,
      "step": 1273
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2580070628430365,
      "learning_rate": 6.763432645785586e-05,
      "loss": 2.1002,
      "step": 1274
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2534860067596615,
      "learning_rate": 6.760886172650879e-05,
      "loss": 1.7564,
      "step": 1275
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29466220969681545,
      "learning_rate": 6.758339699516171e-05,
      "loss": 2.1683,
      "step": 1276
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24849163941014,
      "learning_rate": 6.755793226381461e-05,
      "loss": 1.9333,
      "step": 1277
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2715774897881854,
      "learning_rate": 6.753246753246754e-05,
      "loss": 2.1641,
      "step": 1278
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2773423127687851,
      "learning_rate": 6.750700280112045e-05,
      "loss": 2.1609,
      "step": 1279
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18887337730702833,
      "learning_rate": 6.748153806977336e-05,
      "loss": 1.4509,
      "step": 1280
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24272226698149396,
      "learning_rate": 6.745607333842629e-05,
      "loss": 2.0887,
      "step": 1281
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2505609209948652,
      "learning_rate": 6.74306086070792e-05,
      "loss": 1.7977,
      "step": 1282
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2514602068130961,
      "learning_rate": 6.740514387573211e-05,
      "loss": 2.1257,
      "step": 1283
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.25730321172276827,
      "learning_rate": 6.737967914438504e-05,
      "loss": 1.7093,
      "step": 1284
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27653387859111245,
      "learning_rate": 6.735421441303794e-05,
      "loss": 1.8013,
      "step": 1285
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.25129774152467227,
      "learning_rate": 6.732874968169086e-05,
      "loss": 2.0237,
      "step": 1286
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.309722381690592,
      "learning_rate": 6.730328495034379e-05,
      "loss": 2.0883,
      "step": 1287
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23143122022516766,
      "learning_rate": 6.727782021899669e-05,
      "loss": 1.6351,
      "step": 1288
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.20390221566195668,
      "learning_rate": 6.725235548764961e-05,
      "loss": 1.5702,
      "step": 1289
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22749916840233614,
      "learning_rate": 6.722689075630254e-05,
      "loss": 1.9343,
      "step": 1290
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33009673523214367,
      "learning_rate": 6.720142602495543e-05,
      "loss": 1.6855,
      "step": 1291
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27403768337901185,
      "learning_rate": 6.717596129360836e-05,
      "loss": 1.8071,
      "step": 1292
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2502055366326991,
      "learning_rate": 6.715049656226127e-05,
      "loss": 1.7944,
      "step": 1293
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23232143700286442,
      "learning_rate": 6.712503183091418e-05,
      "loss": 1.822,
      "step": 1294
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2513763568335079,
      "learning_rate": 6.709956709956711e-05,
      "loss": 1.8895,
      "step": 1295
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23929699946645303,
      "learning_rate": 6.707410236822001e-05,
      "loss": 1.8651,
      "step": 1296
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2972379523692023,
      "learning_rate": 6.704863763687293e-05,
      "loss": 1.9477,
      "step": 1297
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24284732826570454,
      "learning_rate": 6.702317290552586e-05,
      "loss": 1.9075,
      "step": 1298
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23334467440246193,
      "learning_rate": 6.699770817417876e-05,
      "loss": 1.7954,
      "step": 1299
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2592609350037409,
      "learning_rate": 6.697224344283168e-05,
      "loss": 1.7029,
      "step": 1300
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.25435686473262703,
      "learning_rate": 6.69467787114846e-05,
      "loss": 2.005,
      "step": 1301
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27128846489090047,
      "learning_rate": 6.692131398013751e-05,
      "loss": 1.9411,
      "step": 1302
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27775912006952264,
      "learning_rate": 6.689584924879043e-05,
      "loss": 1.741,
      "step": 1303
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3025675036838691,
      "learning_rate": 6.687038451744335e-05,
      "loss": 1.8248,
      "step": 1304
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.30757336985638156,
      "learning_rate": 6.684491978609626e-05,
      "loss": 2.0837,
      "step": 1305
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24105919764034295,
      "learning_rate": 6.681945505474918e-05,
      "loss": 1.8052,
      "step": 1306
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22409585216388958,
      "learning_rate": 6.67939903234021e-05,
      "loss": 1.6513,
      "step": 1307
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2285779721231332,
      "learning_rate": 6.676852559205501e-05,
      "loss": 1.7821,
      "step": 1308
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26111613115485566,
      "learning_rate": 6.674306086070792e-05,
      "loss": 1.9924,
      "step": 1309
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24233445969365314,
      "learning_rate": 6.671759612936083e-05,
      "loss": 1.7896,
      "step": 1310
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26812528092251803,
      "learning_rate": 6.669213139801376e-05,
      "loss": 1.8401,
      "step": 1311
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26731033568969936,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.8132,
      "step": 1312
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27340666733340496,
      "learning_rate": 6.664120193531958e-05,
      "loss": 1.9519,
      "step": 1313
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.25663399825692723,
      "learning_rate": 6.661573720397251e-05,
      "loss": 1.8158,
      "step": 1314
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24286383359922728,
      "learning_rate": 6.659027247262542e-05,
      "loss": 1.734,
      "step": 1315
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.22275185623797666,
      "learning_rate": 6.656480774127833e-05,
      "loss": 1.8296,
      "step": 1316
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22560261586284144,
      "learning_rate": 6.653934300993124e-05,
      "loss": 1.8353,
      "step": 1317
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2689726888315691,
      "learning_rate": 6.651387827858417e-05,
      "loss": 1.7082,
      "step": 1318
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2559405918656532,
      "learning_rate": 6.648841354723708e-05,
      "loss": 1.7235,
      "step": 1319
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.29269610115819655,
      "learning_rate": 6.646294881589e-05,
      "loss": 1.969,
      "step": 1320
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2800083701338728,
      "learning_rate": 6.64374840845429e-05,
      "loss": 2.0688,
      "step": 1321
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2775607401940869,
      "learning_rate": 6.641201935319583e-05,
      "loss": 1.9894,
      "step": 1322
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.28656169009328225,
      "learning_rate": 6.638655462184874e-05,
      "loss": 1.9262,
      "step": 1323
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22717866475058626,
      "learning_rate": 6.636108989050166e-05,
      "loss": 1.8858,
      "step": 1324
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22732265151658854,
      "learning_rate": 6.633562515915458e-05,
      "loss": 1.783,
      "step": 1325
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27420903756528914,
      "learning_rate": 6.631016042780749e-05,
      "loss": 2.0469,
      "step": 1326
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.259177481462969,
      "learning_rate": 6.62846956964604e-05,
      "loss": 2.0167,
      "step": 1327
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.31581149304212625,
      "learning_rate": 6.625923096511332e-05,
      "loss": 1.9306,
      "step": 1328
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2304666405920705,
      "learning_rate": 6.623376623376624e-05,
      "loss": 1.9596,
      "step": 1329
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.28666622969728994,
      "learning_rate": 6.620830150241916e-05,
      "loss": 2.1321,
      "step": 1330
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22961063980423926,
      "learning_rate": 6.618283677107207e-05,
      "loss": 1.7907,
      "step": 1331
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2070538940282064,
      "learning_rate": 6.615737203972498e-05,
      "loss": 1.6409,
      "step": 1332
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24414366910943996,
      "learning_rate": 6.61319073083779e-05,
      "loss": 1.9651,
      "step": 1333
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22810236740192402,
      "learning_rate": 6.610644257703082e-05,
      "loss": 1.5453,
      "step": 1334
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2658117843706197,
      "learning_rate": 6.608097784568373e-05,
      "loss": 1.9749,
      "step": 1335
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2468581966524379,
      "learning_rate": 6.605551311433664e-05,
      "loss": 2.0176,
      "step": 1336
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2165745634927306,
      "learning_rate": 6.603004838298957e-05,
      "loss": 1.8768,
      "step": 1337
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35145263014693984,
      "learning_rate": 6.600458365164248e-05,
      "loss": 2.1447,
      "step": 1338
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2898264532763385,
      "learning_rate": 6.597911892029539e-05,
      "loss": 2.0813,
      "step": 1339
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.25197292220531337,
      "learning_rate": 6.595365418894832e-05,
      "loss": 1.7399,
      "step": 1340
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24448733316268856,
      "learning_rate": 6.592818945760123e-05,
      "loss": 1.9932,
      "step": 1341
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24476118946409367,
      "learning_rate": 6.590272472625414e-05,
      "loss": 1.9617,
      "step": 1342
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2640236803446789,
      "learning_rate": 6.587725999490705e-05,
      "loss": 2.2631,
      "step": 1343
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2591299870319825,
      "learning_rate": 6.585179526355996e-05,
      "loss": 1.9671,
      "step": 1344
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2675207034682249,
      "learning_rate": 6.582633053221289e-05,
      "loss": 1.693,
      "step": 1345
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24760865376865063,
      "learning_rate": 6.58008658008658e-05,
      "loss": 1.9212,
      "step": 1346
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.21617154845504996,
      "learning_rate": 6.577540106951871e-05,
      "loss": 1.7881,
      "step": 1347
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2375275072054588,
      "learning_rate": 6.574993633817164e-05,
      "loss": 1.7963,
      "step": 1348
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2585299940545612,
      "learning_rate": 6.572447160682455e-05,
      "loss": 1.8754,
      "step": 1349
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.25376249792299027,
      "learning_rate": 6.569900687547746e-05,
      "loss": 1.8009,
      "step": 1350
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.25025286403430946,
      "learning_rate": 6.567354214413039e-05,
      "loss": 1.9266,
      "step": 1351
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.25199458054739865,
      "learning_rate": 6.564807741278329e-05,
      "loss": 1.9279,
      "step": 1352
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27266634783040256,
      "learning_rate": 6.562261268143621e-05,
      "loss": 1.9831,
      "step": 1353
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27669504351658974,
      "learning_rate": 6.559714795008913e-05,
      "loss": 1.9848,
      "step": 1354
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24953412069247122,
      "learning_rate": 6.557168321874204e-05,
      "loss": 1.8536,
      "step": 1355
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2702238885808487,
      "learning_rate": 6.554621848739496e-05,
      "loss": 2.0665,
      "step": 1356
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2321969715255644,
      "learning_rate": 6.552075375604788e-05,
      "loss": 1.9268,
      "step": 1357
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2798104799004717,
      "learning_rate": 6.549528902470079e-05,
      "loss": 1.8838,
      "step": 1358
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21752493840056586,
      "learning_rate": 6.546982429335371e-05,
      "loss": 1.6784,
      "step": 1359
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2513490132339607,
      "learning_rate": 6.544435956200661e-05,
      "loss": 1.8271,
      "step": 1360
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2365064814921405,
      "learning_rate": 6.541889483065954e-05,
      "loss": 2.0017,
      "step": 1361
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2366556782309439,
      "learning_rate": 6.539343009931246e-05,
      "loss": 1.559,
      "step": 1362
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2503095885499184,
      "learning_rate": 6.536796536796536e-05,
      "loss": 2.0942,
      "step": 1363
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23861207821690975,
      "learning_rate": 6.534250063661829e-05,
      "loss": 1.9378,
      "step": 1364
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21152157628586954,
      "learning_rate": 6.531703590527121e-05,
      "loss": 1.8343,
      "step": 1365
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23847317392502101,
      "learning_rate": 6.529157117392411e-05,
      "loss": 1.7983,
      "step": 1366
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25079873757873594,
      "learning_rate": 6.526610644257704e-05,
      "loss": 1.7489,
      "step": 1367
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2991405549227378,
      "learning_rate": 6.524064171122995e-05,
      "loss": 2.134,
      "step": 1368
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.20564232373526892,
      "learning_rate": 6.521517697988286e-05,
      "loss": 1.6942,
      "step": 1369
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25851456100968884,
      "learning_rate": 6.518971224853579e-05,
      "loss": 1.7756,
      "step": 1370
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3344556149784563,
      "learning_rate": 6.516424751718869e-05,
      "loss": 1.8869,
      "step": 1371
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22107715477010328,
      "learning_rate": 6.513878278584161e-05,
      "loss": 1.7634,
      "step": 1372
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2764487533710954,
      "learning_rate": 6.511331805449454e-05,
      "loss": 1.7338,
      "step": 1373
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29681104673896164,
      "learning_rate": 6.508785332314744e-05,
      "loss": 1.8667,
      "step": 1374
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2723612722690003,
      "learning_rate": 6.506238859180036e-05,
      "loss": 2.0227,
      "step": 1375
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23207059327593188,
      "learning_rate": 6.503692386045329e-05,
      "loss": 1.8282,
      "step": 1376
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29552727261424494,
      "learning_rate": 6.501145912910619e-05,
      "loss": 1.9478,
      "step": 1377
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24430187240969053,
      "learning_rate": 6.498599439775911e-05,
      "loss": 1.6939,
      "step": 1378
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23147228743696965,
      "learning_rate": 6.496052966641202e-05,
      "loss": 2.0028,
      "step": 1379
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2138479563806518,
      "learning_rate": 6.493506493506494e-05,
      "loss": 1.8769,
      "step": 1380
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23960372675608838,
      "learning_rate": 6.490960020371786e-05,
      "loss": 1.7108,
      "step": 1381
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.266942808416875,
      "learning_rate": 6.488413547237077e-05,
      "loss": 1.8276,
      "step": 1382
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23638051610892089,
      "learning_rate": 6.485867074102368e-05,
      "loss": 1.737,
      "step": 1383
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24774945818584873,
      "learning_rate": 6.483320600967661e-05,
      "loss": 1.9648,
      "step": 1384
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24892074861224348,
      "learning_rate": 6.480774127832951e-05,
      "loss": 1.7782,
      "step": 1385
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23232209443976776,
      "learning_rate": 6.478227654698243e-05,
      "loss": 1.6269,
      "step": 1386
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2334740009649208,
      "learning_rate": 6.475681181563535e-05,
      "loss": 1.8128,
      "step": 1387
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2576680645798872,
      "learning_rate": 6.473134708428826e-05,
      "loss": 2.1066,
      "step": 1388
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2638698682218262,
      "learning_rate": 6.470588235294118e-05,
      "loss": 1.7516,
      "step": 1389
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.27446292971529473,
      "learning_rate": 6.46804176215941e-05,
      "loss": 1.9318,
      "step": 1390
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24366698287989227,
      "learning_rate": 6.465495289024701e-05,
      "loss": 2.0153,
      "step": 1391
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25689162894301576,
      "learning_rate": 6.462948815889993e-05,
      "loss": 1.9566,
      "step": 1392
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2711576100748685,
      "learning_rate": 6.460402342755285e-05,
      "loss": 1.6647,
      "step": 1393
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2581664314686578,
      "learning_rate": 6.457855869620576e-05,
      "loss": 1.8281,
      "step": 1394
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23620730062151213,
      "learning_rate": 6.455309396485867e-05,
      "loss": 1.7709,
      "step": 1395
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27636056898839073,
      "learning_rate": 6.452762923351158e-05,
      "loss": 1.8426,
      "step": 1396
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.21443328661435834,
      "learning_rate": 6.450216450216451e-05,
      "loss": 1.6392,
      "step": 1397
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2087630945370244,
      "learning_rate": 6.447669977081742e-05,
      "loss": 1.6107,
      "step": 1398
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2534485726812814,
      "learning_rate": 6.445123503947033e-05,
      "loss": 1.7775,
      "step": 1399
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3154169558385601,
      "learning_rate": 6.442577030812326e-05,
      "loss": 2.1625,
      "step": 1400
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27588868250446313,
      "learning_rate": 6.440030557677617e-05,
      "loss": 1.7711,
      "step": 1401
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2575743471145686,
      "learning_rate": 6.437484084542908e-05,
      "loss": 1.7165,
      "step": 1402
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32638933162569,
      "learning_rate": 6.434937611408201e-05,
      "loss": 2.0342,
      "step": 1403
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.215914382111131,
      "learning_rate": 6.432391138273492e-05,
      "loss": 1.6111,
      "step": 1404
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2793878260490901,
      "learning_rate": 6.429844665138783e-05,
      "loss": 1.8474,
      "step": 1405
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2793958928576041,
      "learning_rate": 6.427298192004074e-05,
      "loss": 1.9646,
      "step": 1406
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30662599759536363,
      "learning_rate": 6.424751718869366e-05,
      "loss": 1.9238,
      "step": 1407
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2886978246152014,
      "learning_rate": 6.422205245734658e-05,
      "loss": 2.0002,
      "step": 1408
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2338112886678382,
      "learning_rate": 6.41965877259995e-05,
      "loss": 1.8845,
      "step": 1409
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25610246830845196,
      "learning_rate": 6.41711229946524e-05,
      "loss": 2.0396,
      "step": 1410
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2975151413382255,
      "learning_rate": 6.414565826330533e-05,
      "loss": 1.8425,
      "step": 1411
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24863671586657765,
      "learning_rate": 6.412019353195824e-05,
      "loss": 1.8178,
      "step": 1412
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25070527551001476,
      "learning_rate": 6.409472880061116e-05,
      "loss": 1.8697,
      "step": 1413
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3037780491315251,
      "learning_rate": 6.406926406926407e-05,
      "loss": 2.1406,
      "step": 1414
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.21667416854550103,
      "learning_rate": 6.404379933791699e-05,
      "loss": 1.9225,
      "step": 1415
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23434381277488175,
      "learning_rate": 6.40183346065699e-05,
      "loss": 2.0665,
      "step": 1416
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27467134571420476,
      "learning_rate": 6.399286987522282e-05,
      "loss": 1.9336,
      "step": 1417
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2859665709971017,
      "learning_rate": 6.396740514387573e-05,
      "loss": 2.042,
      "step": 1418
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2540574137329875,
      "learning_rate": 6.394194041252866e-05,
      "loss": 1.8025,
      "step": 1419
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.26581656336330206,
      "learning_rate": 6.391647568118157e-05,
      "loss": 1.6173,
      "step": 1420
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2626761248646801,
      "learning_rate": 6.389101094983448e-05,
      "loss": 2.1385,
      "step": 1421
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23861266367853373,
      "learning_rate": 6.386554621848739e-05,
      "loss": 1.7302,
      "step": 1422
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2662721632074192,
      "learning_rate": 6.384008148714032e-05,
      "loss": 2.0314,
      "step": 1423
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24905956978551994,
      "learning_rate": 6.381461675579323e-05,
      "loss": 1.8828,
      "step": 1424
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24006534460468518,
      "learning_rate": 6.378915202444614e-05,
      "loss": 1.6141,
      "step": 1425
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25605440350712466,
      "learning_rate": 6.376368729309907e-05,
      "loss": 1.8383,
      "step": 1426
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2328022400944262,
      "learning_rate": 6.373822256175198e-05,
      "loss": 1.7449,
      "step": 1427
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27644689374328363,
      "learning_rate": 6.371275783040489e-05,
      "loss": 2.0059,
      "step": 1428
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25915403720670466,
      "learning_rate": 6.368729309905782e-05,
      "loss": 1.8923,
      "step": 1429
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2270547517818991,
      "learning_rate": 6.366182836771072e-05,
      "loss": 1.7948,
      "step": 1430
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2355183054234252,
      "learning_rate": 6.363636363636364e-05,
      "loss": 1.9153,
      "step": 1431
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.22841443160614022,
      "learning_rate": 6.361089890501655e-05,
      "loss": 1.7276,
      "step": 1432
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23286847875282332,
      "learning_rate": 6.358543417366946e-05,
      "loss": 1.9285,
      "step": 1433
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24938555549328892,
      "learning_rate": 6.355996944232239e-05,
      "loss": 1.7281,
      "step": 1434
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2657884767907746,
      "learning_rate": 6.35345047109753e-05,
      "loss": 1.9137,
      "step": 1435
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2920189657020723,
      "learning_rate": 6.350903997962821e-05,
      "loss": 2.0625,
      "step": 1436
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2713019904179645,
      "learning_rate": 6.348357524828114e-05,
      "loss": 1.9509,
      "step": 1437
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2445235491871537,
      "learning_rate": 6.345811051693404e-05,
      "loss": 1.8115,
      "step": 1438
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2702311812446719,
      "learning_rate": 6.343264578558696e-05,
      "loss": 1.9981,
      "step": 1439
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22968169159584015,
      "learning_rate": 6.340718105423989e-05,
      "loss": 1.8995,
      "step": 1440
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22471574669439587,
      "learning_rate": 6.338171632289279e-05,
      "loss": 1.7872,
      "step": 1441
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23636276211974966,
      "learning_rate": 6.335625159154571e-05,
      "loss": 1.6673,
      "step": 1442
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.24773997747702176,
      "learning_rate": 6.333078686019863e-05,
      "loss": 1.8048,
      "step": 1443
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20969578168066197,
      "learning_rate": 6.330532212885154e-05,
      "loss": 1.5605,
      "step": 1444
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.26764932784946216,
      "learning_rate": 6.327985739750446e-05,
      "loss": 1.7746,
      "step": 1445
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23415979996101344,
      "learning_rate": 6.325439266615738e-05,
      "loss": 1.9106,
      "step": 1446
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2819685499073677,
      "learning_rate": 6.322892793481029e-05,
      "loss": 1.9338,
      "step": 1447
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3109154703013635,
      "learning_rate": 6.320346320346321e-05,
      "loss": 2.104,
      "step": 1448
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25161196654742346,
      "learning_rate": 6.317799847211611e-05,
      "loss": 1.9693,
      "step": 1449
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29015773053836613,
      "learning_rate": 6.315253374076904e-05,
      "loss": 1.7219,
      "step": 1450
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2714891776339336,
      "learning_rate": 6.312706900942196e-05,
      "loss": 2.1159,
      "step": 1451
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2761804603634261,
      "learning_rate": 6.310160427807486e-05,
      "loss": 1.9664,
      "step": 1452
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2757466497139924,
      "learning_rate": 6.307613954672779e-05,
      "loss": 1.921,
      "step": 1453
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2914477602563562,
      "learning_rate": 6.30506748153807e-05,
      "loss": 1.8062,
      "step": 1454
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2544853795063896,
      "learning_rate": 6.302521008403361e-05,
      "loss": 1.8595,
      "step": 1455
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22936712875165363,
      "learning_rate": 6.299974535268654e-05,
      "loss": 1.5139,
      "step": 1456
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.24532316300124118,
      "learning_rate": 6.297428062133945e-05,
      "loss": 1.9526,
      "step": 1457
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3077624749909794,
      "learning_rate": 6.294881588999236e-05,
      "loss": 2.0753,
      "step": 1458
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.26036970986766894,
      "learning_rate": 6.292335115864529e-05,
      "loss": 2.0096,
      "step": 1459
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2603575051181066,
      "learning_rate": 6.289788642729819e-05,
      "loss": 1.7578,
      "step": 1460
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2399714977754378,
      "learning_rate": 6.287242169595111e-05,
      "loss": 1.878,
      "step": 1461
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2849326236664881,
      "learning_rate": 6.284695696460404e-05,
      "loss": 1.814,
      "step": 1462
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23877184033983795,
      "learning_rate": 6.282149223325694e-05,
      "loss": 1.9115,
      "step": 1463
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20668147788374516,
      "learning_rate": 6.279602750190986e-05,
      "loss": 1.6837,
      "step": 1464
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25040498356110313,
      "learning_rate": 6.277056277056277e-05,
      "loss": 1.9307,
      "step": 1465
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2542878584427889,
      "learning_rate": 6.274509803921569e-05,
      "loss": 2.0858,
      "step": 1466
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.26215346373467596,
      "learning_rate": 6.271963330786861e-05,
      "loss": 2.105,
      "step": 1467
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21783654192462465,
      "learning_rate": 6.269416857652152e-05,
      "loss": 1.6093,
      "step": 1468
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3033084471593143,
      "learning_rate": 6.266870384517444e-05,
      "loss": 2.3007,
      "step": 1469
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2241801431361778,
      "learning_rate": 6.264323911382736e-05,
      "loss": 1.8179,
      "step": 1470
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20758916247638143,
      "learning_rate": 6.261777438248026e-05,
      "loss": 1.7694,
      "step": 1471
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.24665891751691102,
      "learning_rate": 6.259230965113318e-05,
      "loss": 1.5666,
      "step": 1472
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20837654420638607,
      "learning_rate": 6.25668449197861e-05,
      "loss": 1.7046,
      "step": 1473
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22807501023098187,
      "learning_rate": 6.254138018843901e-05,
      "loss": 1.8228,
      "step": 1474
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2185941720158155,
      "learning_rate": 6.251591545709193e-05,
      "loss": 1.7609,
      "step": 1475
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35387071932188074,
      "learning_rate": 6.249045072574485e-05,
      "loss": 2.0526,
      "step": 1476
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2554499429668722,
      "learning_rate": 6.246498599439776e-05,
      "loss": 1.5325,
      "step": 1477
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23743890961682723,
      "learning_rate": 6.243952126305068e-05,
      "loss": 1.7867,
      "step": 1478
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24947721985606203,
      "learning_rate": 6.24140565317036e-05,
      "loss": 1.7074,
      "step": 1479
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20516978440742267,
      "learning_rate": 6.238859180035651e-05,
      "loss": 1.5472,
      "step": 1480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21535132012508432,
      "learning_rate": 6.236312706900943e-05,
      "loss": 1.7486,
      "step": 1481
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23653666697894052,
      "learning_rate": 6.233766233766233e-05,
      "loss": 1.9899,
      "step": 1482
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2437997803840399,
      "learning_rate": 6.231219760631526e-05,
      "loss": 1.8934,
      "step": 1483
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22951134946649115,
      "learning_rate": 6.228673287496817e-05,
      "loss": 1.7781,
      "step": 1484
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.27430318309600316,
      "learning_rate": 6.226126814362108e-05,
      "loss": 1.893,
      "step": 1485
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2974475935249174,
      "learning_rate": 6.223580341227401e-05,
      "loss": 1.8601,
      "step": 1486
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25028619180316575,
      "learning_rate": 6.221033868092692e-05,
      "loss": 1.7759,
      "step": 1487
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2989686373796846,
      "learning_rate": 6.218487394957983e-05,
      "loss": 2.0212,
      "step": 1488
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2613443572016678,
      "learning_rate": 6.215940921823276e-05,
      "loss": 1.6686,
      "step": 1489
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2409582010917778,
      "learning_rate": 6.213394448688567e-05,
      "loss": 1.6892,
      "step": 1490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3053278307740795,
      "learning_rate": 6.210847975553858e-05,
      "loss": 1.8391,
      "step": 1491
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.27683640133493354,
      "learning_rate": 6.20830150241915e-05,
      "loss": 1.9558,
      "step": 1492
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26100866580786997,
      "learning_rate": 6.205755029284442e-05,
      "loss": 2.0114,
      "step": 1493
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25144925856208833,
      "learning_rate": 6.203208556149733e-05,
      "loss": 1.7074,
      "step": 1494
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2786330356808251,
      "learning_rate": 6.200662083015024e-05,
      "loss": 2.0112,
      "step": 1495
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25566249290316717,
      "learning_rate": 6.198115609880316e-05,
      "loss": 1.699,
      "step": 1496
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23609661553724798,
      "learning_rate": 6.195569136745608e-05,
      "loss": 1.9024,
      "step": 1497
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.315676495156524,
      "learning_rate": 6.1930226636109e-05,
      "loss": 2.0619,
      "step": 1498
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26074795948740676,
      "learning_rate": 6.19047619047619e-05,
      "loss": 1.9099,
      "step": 1499
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2535138105142371,
      "learning_rate": 6.187929717341482e-05,
      "loss": 1.7467,
      "step": 1500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22753225979438801,
      "learning_rate": 6.185383244206774e-05,
      "loss": 1.6906,
      "step": 1501
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2415540394035451,
      "learning_rate": 6.182836771072066e-05,
      "loss": 2.1264,
      "step": 1502
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24931738465143685,
      "learning_rate": 6.180290297937357e-05,
      "loss": 1.913,
      "step": 1503
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.262043395741899,
      "learning_rate": 6.17774382480265e-05,
      "loss": 1.9871,
      "step": 1504
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24457982802573416,
      "learning_rate": 6.17519735166794e-05,
      "loss": 2.0219,
      "step": 1505
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20040165571522914,
      "learning_rate": 6.172650878533232e-05,
      "loss": 1.7238,
      "step": 1506
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2316379067710165,
      "learning_rate": 6.170104405398523e-05,
      "loss": 1.6758,
      "step": 1507
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25894208018998377,
      "learning_rate": 6.167557932263814e-05,
      "loss": 1.6759,
      "step": 1508
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25088205302140765,
      "learning_rate": 6.165011459129107e-05,
      "loss": 1.9603,
      "step": 1509
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25628582890009516,
      "learning_rate": 6.162464985994398e-05,
      "loss": 2.0707,
      "step": 1510
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26404782090778656,
      "learning_rate": 6.159918512859689e-05,
      "loss": 1.956,
      "step": 1511
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23154477907500365,
      "learning_rate": 6.157372039724982e-05,
      "loss": 1.7662,
      "step": 1512
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24666147746258535,
      "learning_rate": 6.154825566590273e-05,
      "loss": 1.7528,
      "step": 1513
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25819520288915415,
      "learning_rate": 6.152279093455564e-05,
      "loss": 1.8696,
      "step": 1514
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23331428351437192,
      "learning_rate": 6.149732620320857e-05,
      "loss": 1.8525,
      "step": 1515
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2463173153529251,
      "learning_rate": 6.147186147186147e-05,
      "loss": 1.8842,
      "step": 1516
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22932427939527678,
      "learning_rate": 6.144639674051439e-05,
      "loss": 1.6951,
      "step": 1517
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25462386617656213,
      "learning_rate": 6.14209320091673e-05,
      "loss": 1.9397,
      "step": 1518
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2704000017182009,
      "learning_rate": 6.139546727782022e-05,
      "loss": 2.014,
      "step": 1519
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22640643828642856,
      "learning_rate": 6.137000254647314e-05,
      "loss": 1.7801,
      "step": 1520
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2801141585363283,
      "learning_rate": 6.134453781512605e-05,
      "loss": 2.0162,
      "step": 1521
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37284208338634134,
      "learning_rate": 6.131907308377896e-05,
      "loss": 2.4535,
      "step": 1522
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2206019954737265,
      "learning_rate": 6.129360835243189e-05,
      "loss": 1.758,
      "step": 1523
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24222795087111626,
      "learning_rate": 6.12681436210848e-05,
      "loss": 1.9262,
      "step": 1524
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2623245998615822,
      "learning_rate": 6.124267888973771e-05,
      "loss": 2.1177,
      "step": 1525
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2149816677267667,
      "learning_rate": 6.121721415839064e-05,
      "loss": 1.8284,
      "step": 1526
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25969795550289637,
      "learning_rate": 6.119174942704354e-05,
      "loss": 2.0281,
      "step": 1527
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22986253115787428,
      "learning_rate": 6.116628469569646e-05,
      "loss": 1.7601,
      "step": 1528
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26093874891062196,
      "learning_rate": 6.114081996434938e-05,
      "loss": 1.9096,
      "step": 1529
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2656012692790266,
      "learning_rate": 6.111535523300229e-05,
      "loss": 1.9157,
      "step": 1530
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.22004615429597213,
      "learning_rate": 6.108989050165521e-05,
      "loss": 1.7553,
      "step": 1531
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2721988415462892,
      "learning_rate": 6.106442577030813e-05,
      "loss": 1.8849,
      "step": 1532
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25852281801944327,
      "learning_rate": 6.103896103896104e-05,
      "loss": 1.8063,
      "step": 1533
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24761462659063677,
      "learning_rate": 6.101349630761396e-05,
      "loss": 1.8574,
      "step": 1534
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2726207797525334,
      "learning_rate": 6.098803157626687e-05,
      "loss": 1.7744,
      "step": 1535
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31156736919152733,
      "learning_rate": 6.096256684491979e-05,
      "loss": 1.692,
      "step": 1536
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29589856339757276,
      "learning_rate": 6.093710211357271e-05,
      "loss": 2.0424,
      "step": 1537
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25697746293815316,
      "learning_rate": 6.091163738222562e-05,
      "loss": 2.0985,
      "step": 1538
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2519730996190904,
      "learning_rate": 6.088617265087854e-05,
      "loss": 1.9054,
      "step": 1539
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2178550105568459,
      "learning_rate": 6.086070791953146e-05,
      "loss": 1.6298,
      "step": 1540
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24681784052612069,
      "learning_rate": 6.083524318818436e-05,
      "loss": 1.7629,
      "step": 1541
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23825988517410313,
      "learning_rate": 6.080977845683729e-05,
      "loss": 1.8602,
      "step": 1542
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2615875940351599,
      "learning_rate": 6.078431372549019e-05,
      "loss": 1.8776,
      "step": 1543
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24157319299390567,
      "learning_rate": 6.075884899414311e-05,
      "loss": 1.8031,
      "step": 1544
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2521081048574686,
      "learning_rate": 6.073338426279603e-05,
      "loss": 1.8563,
      "step": 1545
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2536205004287772,
      "learning_rate": 6.070791953144894e-05,
      "loss": 1.9038,
      "step": 1546
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.28644482262857046,
      "learning_rate": 6.068245480010186e-05,
      "loss": 1.9221,
      "step": 1547
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2645469273778994,
      "learning_rate": 6.065699006875478e-05,
      "loss": 1.8193,
      "step": 1548
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25973897811628416,
      "learning_rate": 6.063152533740769e-05,
      "loss": 1.8508,
      "step": 1549
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24847994567745993,
      "learning_rate": 6.060606060606061e-05,
      "loss": 1.9156,
      "step": 1550
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2641550370317926,
      "learning_rate": 6.058059587471352e-05,
      "loss": 1.8653,
      "step": 1551
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24168907683701302,
      "learning_rate": 6.055513114336644e-05,
      "loss": 1.9855,
      "step": 1552
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22317099049165945,
      "learning_rate": 6.052966641201936e-05,
      "loss": 1.7703,
      "step": 1553
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2616989498334913,
      "learning_rate": 6.0504201680672267e-05,
      "loss": 1.8338,
      "step": 1554
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.26024510295357467,
      "learning_rate": 6.0478736949325185e-05,
      "loss": 1.8048,
      "step": 1555
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.19852627285411298,
      "learning_rate": 6.0453272217978104e-05,
      "loss": 1.8041,
      "step": 1556
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2628471899781459,
      "learning_rate": 6.0427807486631016e-05,
      "loss": 1.9684,
      "step": 1557
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22506245236447928,
      "learning_rate": 6.0402342755283935e-05,
      "loss": 1.6453,
      "step": 1558
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2571811604095075,
      "learning_rate": 6.0376878023936854e-05,
      "loss": 1.9955,
      "step": 1559
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30268943206907745,
      "learning_rate": 6.0351413292589766e-05,
      "loss": 2.0422,
      "step": 1560
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3091237791772314,
      "learning_rate": 6.0325948561242685e-05,
      "loss": 2.1414,
      "step": 1561
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.23776443905612016,
      "learning_rate": 6.030048382989559e-05,
      "loss": 1.8843,
      "step": 1562
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27552521575346095,
      "learning_rate": 6.0275019098548516e-05,
      "loss": 1.6124,
      "step": 1563
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3014428924388266,
      "learning_rate": 6.0249554367201435e-05,
      "loss": 2.197,
      "step": 1564
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2699297433194896,
      "learning_rate": 6.022408963585434e-05,
      "loss": 1.7632,
      "step": 1565
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3797500496036178,
      "learning_rate": 6.019862490450726e-05,
      "loss": 2.0411,
      "step": 1566
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2750966010786085,
      "learning_rate": 6.0173160173160184e-05,
      "loss": 2.1269,
      "step": 1567
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.23276078334622122,
      "learning_rate": 6.014769544181309e-05,
      "loss": 1.7726,
      "step": 1568
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2908467180242655,
      "learning_rate": 6.012223071046601e-05,
      "loss": 2.187,
      "step": 1569
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.26803748525452153,
      "learning_rate": 6.009676597911892e-05,
      "loss": 1.9399,
      "step": 1570
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2333609499014553,
      "learning_rate": 6.007130124777184e-05,
      "loss": 1.9887,
      "step": 1571
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3065819896837221,
      "learning_rate": 6.004583651642476e-05,
      "loss": 1.9615,
      "step": 1572
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3629978469109588,
      "learning_rate": 6.0020371785077664e-05,
      "loss": 1.8352,
      "step": 1573
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27024806803772333,
      "learning_rate": 5.999490705373059e-05,
      "loss": 1.7978,
      "step": 1574
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3267840733438547,
      "learning_rate": 5.996944232238351e-05,
      "loss": 2.4383,
      "step": 1575
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24161056279754708,
      "learning_rate": 5.9943977591036414e-05,
      "loss": 1.9505,
      "step": 1576
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2705996663831965,
      "learning_rate": 5.991851285968933e-05,
      "loss": 2.0719,
      "step": 1577
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22459827893256726,
      "learning_rate": 5.9893048128342244e-05,
      "loss": 1.8081,
      "step": 1578
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27083593911360143,
      "learning_rate": 5.986758339699516e-05,
      "loss": 1.8431,
      "step": 1579
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27101741397604806,
      "learning_rate": 5.984211866564808e-05,
      "loss": 1.8037,
      "step": 1580
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2545983784616469,
      "learning_rate": 5.9816653934300994e-05,
      "loss": 1.9613,
      "step": 1581
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2262129059569068,
      "learning_rate": 5.979118920295391e-05,
      "loss": 1.8274,
      "step": 1582
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2673398230624889,
      "learning_rate": 5.976572447160683e-05,
      "loss": 1.9489,
      "step": 1583
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20774983809392522,
      "learning_rate": 5.9740259740259744e-05,
      "loss": 1.4933,
      "step": 1584
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24907516241492358,
      "learning_rate": 5.971479500891266e-05,
      "loss": 1.6926,
      "step": 1585
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24478498493706932,
      "learning_rate": 5.968933027756557e-05,
      "loss": 1.8847,
      "step": 1586
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28013622101707153,
      "learning_rate": 5.966386554621849e-05,
      "loss": 1.8548,
      "step": 1587
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2725771310995344,
      "learning_rate": 5.9638400814871406e-05,
      "loss": 2.1802,
      "step": 1588
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.23403779249588136,
      "learning_rate": 5.961293608352432e-05,
      "loss": 1.9072,
      "step": 1589
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2146086880756573,
      "learning_rate": 5.958747135217724e-05,
      "loss": 1.7345,
      "step": 1590
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.25779612807106644,
      "learning_rate": 5.9562006620830156e-05,
      "loss": 1.8754,
      "step": 1591
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24281365452004675,
      "learning_rate": 5.953654188948307e-05,
      "loss": 1.8136,
      "step": 1592
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23911587923286357,
      "learning_rate": 5.9511077158135987e-05,
      "loss": 1.9183,
      "step": 1593
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21796275578395394,
      "learning_rate": 5.948561242678889e-05,
      "loss": 1.8331,
      "step": 1594
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.258296021848505,
      "learning_rate": 5.946014769544182e-05,
      "loss": 1.7855,
      "step": 1595
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23530867094746857,
      "learning_rate": 5.9434682964094736e-05,
      "loss": 1.6778,
      "step": 1596
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2208216863535817,
      "learning_rate": 5.940921823274764e-05,
      "loss": 1.7123,
      "step": 1597
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23626065670430846,
      "learning_rate": 5.938375350140056e-05,
      "loss": 1.7934,
      "step": 1598
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2570483848402682,
      "learning_rate": 5.9358288770053486e-05,
      "loss": 1.6343,
      "step": 1599
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24978742651698496,
      "learning_rate": 5.933282403870639e-05,
      "loss": 1.8905,
      "step": 1600
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25463877428295967,
      "learning_rate": 5.930735930735931e-05,
      "loss": 2.0177,
      "step": 1601
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2477609758943463,
      "learning_rate": 5.928189457601223e-05,
      "loss": 1.6774,
      "step": 1602
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.26060204901142225,
      "learning_rate": 5.925642984466514e-05,
      "loss": 1.967,
      "step": 1603
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3095171548829742,
      "learning_rate": 5.923096511331806e-05,
      "loss": 1.8638,
      "step": 1604
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2518297825021641,
      "learning_rate": 5.9205500381970965e-05,
      "loss": 1.819,
      "step": 1605
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2939462249843084,
      "learning_rate": 5.918003565062389e-05,
      "loss": 2.0621,
      "step": 1606
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2626615736613039,
      "learning_rate": 5.915457091927681e-05,
      "loss": 1.9616,
      "step": 1607
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2905264589953374,
      "learning_rate": 5.9129106187929715e-05,
      "loss": 1.9486,
      "step": 1608
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29541525087613635,
      "learning_rate": 5.9103641456582634e-05,
      "loss": 2.1314,
      "step": 1609
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24192291080967857,
      "learning_rate": 5.907817672523556e-05,
      "loss": 1.7453,
      "step": 1610
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30470200039520245,
      "learning_rate": 5.9052711993888465e-05,
      "loss": 2.0023,
      "step": 1611
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2788043180495265,
      "learning_rate": 5.9027247262541384e-05,
      "loss": 1.9452,
      "step": 1612
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24726742205938537,
      "learning_rate": 5.9001782531194296e-05,
      "loss": 1.8762,
      "step": 1613
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.32697497651468815,
      "learning_rate": 5.8976317799847215e-05,
      "loss": 2.0121,
      "step": 1614
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.26092398659897753,
      "learning_rate": 5.8950853068500134e-05,
      "loss": 2.0001,
      "step": 1615
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2902436225413904,
      "learning_rate": 5.8925388337153046e-05,
      "loss": 2.0014,
      "step": 1616
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.28148702064899217,
      "learning_rate": 5.8899923605805964e-05,
      "loss": 1.888,
      "step": 1617
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2474122527178358,
      "learning_rate": 5.887445887445888e-05,
      "loss": 2.0164,
      "step": 1618
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24043364130359207,
      "learning_rate": 5.884899414311179e-05,
      "loss": 1.8548,
      "step": 1619
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22881301048532374,
      "learning_rate": 5.882352941176471e-05,
      "loss": 1.8628,
      "step": 1620
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25063492198815107,
      "learning_rate": 5.879806468041762e-05,
      "loss": 2.0986,
      "step": 1621
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2323713167482444,
      "learning_rate": 5.877259994907054e-05,
      "loss": 2.1365,
      "step": 1622
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24691177076215523,
      "learning_rate": 5.874713521772346e-05,
      "loss": 1.9383,
      "step": 1623
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22673647054803103,
      "learning_rate": 5.872167048637637e-05,
      "loss": 1.8255,
      "step": 1624
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24692504743712895,
      "learning_rate": 5.869620575502929e-05,
      "loss": 2.0798,
      "step": 1625
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2667273200367276,
      "learning_rate": 5.867074102368221e-05,
      "loss": 1.8868,
      "step": 1626
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22908787113265164,
      "learning_rate": 5.864527629233512e-05,
      "loss": 1.956,
      "step": 1627
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2246979414433047,
      "learning_rate": 5.861981156098804e-05,
      "loss": 2.1121,
      "step": 1628
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22929133289924503,
      "learning_rate": 5.859434682964094e-05,
      "loss": 1.8558,
      "step": 1629
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.22917793347376852,
      "learning_rate": 5.856888209829386e-05,
      "loss": 1.6781,
      "step": 1630
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25641446650774763,
      "learning_rate": 5.854341736694679e-05,
      "loss": 1.7482,
      "step": 1631
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.23393537298015707,
      "learning_rate": 5.851795263559969e-05,
      "loss": 1.8537,
      "step": 1632
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24914347429581077,
      "learning_rate": 5.849248790425261e-05,
      "loss": 1.888,
      "step": 1633
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24546671260082917,
      "learning_rate": 5.846702317290553e-05,
      "loss": 2.0078,
      "step": 1634
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.21882815327315605,
      "learning_rate": 5.844155844155844e-05,
      "loss": 1.8693,
      "step": 1635
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2596485332949677,
      "learning_rate": 5.841609371021136e-05,
      "loss": 1.9116,
      "step": 1636
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.22340332983736189,
      "learning_rate": 5.839062897886428e-05,
      "loss": 1.6942,
      "step": 1637
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30014327124952705,
      "learning_rate": 5.836516424751719e-05,
      "loss": 2.0411,
      "step": 1638
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25882642944038486,
      "learning_rate": 5.833969951617011e-05,
      "loss": 1.6902,
      "step": 1639
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25289569169685344,
      "learning_rate": 5.831423478482302e-05,
      "loss": 1.9374,
      "step": 1640
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27586659091704435,
      "learning_rate": 5.8288770053475936e-05,
      "loss": 1.6424,
      "step": 1641
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27927541316876636,
      "learning_rate": 5.826330532212886e-05,
      "loss": 1.9862,
      "step": 1642
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3011801539172539,
      "learning_rate": 5.8237840590781767e-05,
      "loss": 2.32,
      "step": 1643
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3050488261319834,
      "learning_rate": 5.8212375859434685e-05,
      "loss": 1.9606,
      "step": 1644
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2567523341578881,
      "learning_rate": 5.8186911128087604e-05,
      "loss": 2.035,
      "step": 1645
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.22761743857463332,
      "learning_rate": 5.8161446396740516e-05,
      "loss": 1.8363,
      "step": 1646
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2328384256968828,
      "learning_rate": 5.8135981665393435e-05,
      "loss": 1.8999,
      "step": 1647
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32891412485440624,
      "learning_rate": 5.811051693404635e-05,
      "loss": 2.0234,
      "step": 1648
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2773328026773537,
      "learning_rate": 5.8085052202699266e-05,
      "loss": 1.807,
      "step": 1649
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26592009645438053,
      "learning_rate": 5.8059587471352185e-05,
      "loss": 1.7819,
      "step": 1650
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2350560307412541,
      "learning_rate": 5.803412274000509e-05,
      "loss": 1.854,
      "step": 1651
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2211039962011883,
      "learning_rate": 5.800865800865801e-05,
      "loss": 2.0175,
      "step": 1652
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2956828222523623,
      "learning_rate": 5.7983193277310935e-05,
      "loss": 2.0187,
      "step": 1653
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24018430575744007,
      "learning_rate": 5.795772854596384e-05,
      "loss": 1.6413,
      "step": 1654
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2851889082091988,
      "learning_rate": 5.793226381461676e-05,
      "loss": 2.0801,
      "step": 1655
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26336700333348967,
      "learning_rate": 5.790679908326967e-05,
      "loss": 1.831,
      "step": 1656
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24335249528299224,
      "learning_rate": 5.788133435192259e-05,
      "loss": 1.9065,
      "step": 1657
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3074932999384943,
      "learning_rate": 5.785586962057551e-05,
      "loss": 1.6846,
      "step": 1658
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2929328642165607,
      "learning_rate": 5.783040488922842e-05,
      "loss": 2.117,
      "step": 1659
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3113969049616286,
      "learning_rate": 5.780494015788134e-05,
      "loss": 2.0645,
      "step": 1660
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2937729796085543,
      "learning_rate": 5.777947542653426e-05,
      "loss": 1.9968,
      "step": 1661
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26455967107729744,
      "learning_rate": 5.7754010695187164e-05,
      "loss": 1.9099,
      "step": 1662
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2721072673688603,
      "learning_rate": 5.772854596384009e-05,
      "loss": 2.1461,
      "step": 1663
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.23609306531007804,
      "learning_rate": 5.7703081232492995e-05,
      "loss": 2.0404,
      "step": 1664
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24975060152577083,
      "learning_rate": 5.7677616501145914e-05,
      "loss": 1.8813,
      "step": 1665
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2617133125765185,
      "learning_rate": 5.765215176979883e-05,
      "loss": 2.016,
      "step": 1666
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.276917385060399,
      "learning_rate": 5.7626687038451745e-05,
      "loss": 1.8029,
      "step": 1667
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2426905483611426,
      "learning_rate": 5.760122230710466e-05,
      "loss": 1.8359,
      "step": 1668
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24819638409704883,
      "learning_rate": 5.757575757575758e-05,
      "loss": 2.1224,
      "step": 1669
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2285441111880456,
      "learning_rate": 5.7550292844410494e-05,
      "loss": 1.7729,
      "step": 1670
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21992829281266243,
      "learning_rate": 5.752482811306341e-05,
      "loss": 1.8653,
      "step": 1671
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2455672657189347,
      "learning_rate": 5.749936338171632e-05,
      "loss": 1.8892,
      "step": 1672
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21163598726865365,
      "learning_rate": 5.747389865036924e-05,
      "loss": 1.7959,
      "step": 1673
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24640105754241015,
      "learning_rate": 5.744843391902216e-05,
      "loss": 1.9364,
      "step": 1674
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26734456080587427,
      "learning_rate": 5.742296918767507e-05,
      "loss": 1.9096,
      "step": 1675
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26069813616960297,
      "learning_rate": 5.739750445632799e-05,
      "loss": 1.9483,
      "step": 1676
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23182803512512878,
      "learning_rate": 5.7372039724980906e-05,
      "loss": 1.6035,
      "step": 1677
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24343677532886382,
      "learning_rate": 5.734657499363382e-05,
      "loss": 1.854,
      "step": 1678
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2559604154069802,
      "learning_rate": 5.732111026228674e-05,
      "loss": 2.1057,
      "step": 1679
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24903703540252603,
      "learning_rate": 5.7295645530939656e-05,
      "loss": 1.9747,
      "step": 1680
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24292430011120034,
      "learning_rate": 5.727018079959257e-05,
      "loss": 1.6008,
      "step": 1681
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2597732829635861,
      "learning_rate": 5.724471606824549e-05,
      "loss": 1.8885,
      "step": 1682
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23894214436574637,
      "learning_rate": 5.721925133689839e-05,
      "loss": 1.8221,
      "step": 1683
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29846146472358814,
      "learning_rate": 5.719378660555131e-05,
      "loss": 1.7687,
      "step": 1684
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2435618750544854,
      "learning_rate": 5.7168321874204236e-05,
      "loss": 1.9643,
      "step": 1685
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.20434998720844322,
      "learning_rate": 5.714285714285714e-05,
      "loss": 1.806,
      "step": 1686
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28126539082271323,
      "learning_rate": 5.711739241151006e-05,
      "loss": 2.0432,
      "step": 1687
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23609697055702877,
      "learning_rate": 5.709192768016298e-05,
      "loss": 1.7398,
      "step": 1688
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2619756036039603,
      "learning_rate": 5.706646294881589e-05,
      "loss": 2.0893,
      "step": 1689
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23441992170260872,
      "learning_rate": 5.704099821746881e-05,
      "loss": 1.808,
      "step": 1690
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28369231232122544,
      "learning_rate": 5.701553348612172e-05,
      "loss": 2.1341,
      "step": 1691
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2109833773420096,
      "learning_rate": 5.699006875477464e-05,
      "loss": 1.6876,
      "step": 1692
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2108497083722984,
      "learning_rate": 5.696460402342756e-05,
      "loss": 1.6995,
      "step": 1693
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2194012246664097,
      "learning_rate": 5.6939139292080465e-05,
      "loss": 1.9067,
      "step": 1694
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2713182613514185,
      "learning_rate": 5.6913674560733384e-05,
      "loss": 1.8971,
      "step": 1695
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.296808887950505,
      "learning_rate": 5.688820982938631e-05,
      "loss": 2.255,
      "step": 1696
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2619363961593609,
      "learning_rate": 5.6862745098039215e-05,
      "loss": 1.9257,
      "step": 1697
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2310624315603793,
      "learning_rate": 5.6837280366692134e-05,
      "loss": 1.7678,
      "step": 1698
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28676857476407935,
      "learning_rate": 5.6811815635345046e-05,
      "loss": 1.8109,
      "step": 1699
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2540912101214253,
      "learning_rate": 5.6786350903997965e-05,
      "loss": 1.822,
      "step": 1700
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2529094201882292,
      "learning_rate": 5.6760886172650884e-05,
      "loss": 1.9275,
      "step": 1701
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2851786147452933,
      "learning_rate": 5.6735421441303796e-05,
      "loss": 1.9064,
      "step": 1702
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21299800439580308,
      "learning_rate": 5.6709956709956715e-05,
      "loss": 1.6424,
      "step": 1703
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23520012094612563,
      "learning_rate": 5.6684491978609634e-05,
      "loss": 1.8498,
      "step": 1704
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3675517446718101,
      "learning_rate": 5.665902724726254e-05,
      "loss": 1.991,
      "step": 1705
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2517874416912386,
      "learning_rate": 5.6633562515915465e-05,
      "loss": 1.7943,
      "step": 1706
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27320188180075,
      "learning_rate": 5.660809778456837e-05,
      "loss": 1.7237,
      "step": 1707
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.221588208317137,
      "learning_rate": 5.658263305322129e-05,
      "loss": 1.6462,
      "step": 1708
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22910092053725895,
      "learning_rate": 5.655716832187421e-05,
      "loss": 1.935,
      "step": 1709
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24057608237244552,
      "learning_rate": 5.653170359052712e-05,
      "loss": 1.691,
      "step": 1710
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23637377085333372,
      "learning_rate": 5.650623885918004e-05,
      "loss": 1.9726,
      "step": 1711
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24854888578313664,
      "learning_rate": 5.648077412783296e-05,
      "loss": 1.9453,
      "step": 1712
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2606071093705677,
      "learning_rate": 5.645530939648587e-05,
      "loss": 1.7513,
      "step": 1713
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33451192583541084,
      "learning_rate": 5.642984466513879e-05,
      "loss": 1.6622,
      "step": 1714
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28049286284435937,
      "learning_rate": 5.640437993379171e-05,
      "loss": 1.9776,
      "step": 1715
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23603799050412522,
      "learning_rate": 5.637891520244461e-05,
      "loss": 1.8003,
      "step": 1716
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2525222619317922,
      "learning_rate": 5.635345047109754e-05,
      "loss": 1.7598,
      "step": 1717
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.25108993878703484,
      "learning_rate": 5.632798573975044e-05,
      "loss": 1.6963,
      "step": 1718
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23628532400404173,
      "learning_rate": 5.630252100840336e-05,
      "loss": 1.6703,
      "step": 1719
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2617072771888496,
      "learning_rate": 5.627705627705628e-05,
      "loss": 1.8925,
      "step": 1720
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23966982651995303,
      "learning_rate": 5.625159154570919e-05,
      "loss": 1.9139,
      "step": 1721
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32751289677160395,
      "learning_rate": 5.622612681436211e-05,
      "loss": 1.8088,
      "step": 1722
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21991901870146513,
      "learning_rate": 5.620066208301503e-05,
      "loss": 1.8793,
      "step": 1723
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.304056051250577,
      "learning_rate": 5.617519735166794e-05,
      "loss": 2.3134,
      "step": 1724
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28237868700326474,
      "learning_rate": 5.614973262032086e-05,
      "loss": 2.0754,
      "step": 1725
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28944437474500934,
      "learning_rate": 5.612426788897377e-05,
      "loss": 1.8642,
      "step": 1726
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2932810708087583,
      "learning_rate": 5.6098803157626686e-05,
      "loss": 1.7588,
      "step": 1727
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3131703101856893,
      "learning_rate": 5.607333842627961e-05,
      "loss": 2.2586,
      "step": 1728
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21605552180737272,
      "learning_rate": 5.604787369493252e-05,
      "loss": 1.6997,
      "step": 1729
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2918401341372067,
      "learning_rate": 5.6022408963585436e-05,
      "loss": 2.0401,
      "step": 1730
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24155233524645017,
      "learning_rate": 5.5996944232238355e-05,
      "loss": 1.8575,
      "step": 1731
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.26069640711381625,
      "learning_rate": 5.597147950089127e-05,
      "loss": 1.8647,
      "step": 1732
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2329740136854815,
      "learning_rate": 5.5946014769544186e-05,
      "loss": 1.7878,
      "step": 1733
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.22908269181710408,
      "learning_rate": 5.59205500381971e-05,
      "loss": 1.776,
      "step": 1734
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31914934779074683,
      "learning_rate": 5.5895085306850016e-05,
      "loss": 1.7529,
      "step": 1735
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23060766171717703,
      "learning_rate": 5.5869620575502935e-05,
      "loss": 1.6373,
      "step": 1736
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2634189805776267,
      "learning_rate": 5.584415584415584e-05,
      "loss": 1.8803,
      "step": 1737
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2214872810805536,
      "learning_rate": 5.5818691112808766e-05,
      "loss": 1.7981,
      "step": 1738
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2261818779823524,
      "learning_rate": 5.5793226381461685e-05,
      "loss": 1.9758,
      "step": 1739
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2376195130053332,
      "learning_rate": 5.576776165011459e-05,
      "loss": 2.0915,
      "step": 1740
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2637069820061634,
      "learning_rate": 5.574229691876751e-05,
      "loss": 1.758,
      "step": 1741
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23249804426724283,
      "learning_rate": 5.571683218742042e-05,
      "loss": 1.7041,
      "step": 1742
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.25433403071838684,
      "learning_rate": 5.569136745607334e-05,
      "loss": 1.8649,
      "step": 1743
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24052920509203618,
      "learning_rate": 5.566590272472626e-05,
      "loss": 1.9441,
      "step": 1744
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2567302065595584,
      "learning_rate": 5.564043799337917e-05,
      "loss": 1.9928,
      "step": 1745
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21436613920934142,
      "learning_rate": 5.561497326203209e-05,
      "loss": 1.5653,
      "step": 1746
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3135895688776553,
      "learning_rate": 5.558950853068501e-05,
      "loss": 2.0952,
      "step": 1747
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24846845380016735,
      "learning_rate": 5.5564043799337914e-05,
      "loss": 1.8637,
      "step": 1748
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2653113926121906,
      "learning_rate": 5.553857906799084e-05,
      "loss": 1.7786,
      "step": 1749
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.25425882069394706,
      "learning_rate": 5.5513114336643745e-05,
      "loss": 2.2434,
      "step": 1750
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2466540543020976,
      "learning_rate": 5.5487649605296664e-05,
      "loss": 1.8173,
      "step": 1751
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23269323327898367,
      "learning_rate": 5.546218487394958e-05,
      "loss": 1.8893,
      "step": 1752
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19195040406288671,
      "learning_rate": 5.5436720142602495e-05,
      "loss": 1.7546,
      "step": 1753
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2605782182591182,
      "learning_rate": 5.5411255411255414e-05,
      "loss": 2.0306,
      "step": 1754
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.266186513361073,
      "learning_rate": 5.538579067990833e-05,
      "loss": 1.7673,
      "step": 1755
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27585153359520465,
      "learning_rate": 5.5360325948561245e-05,
      "loss": 1.8655,
      "step": 1756
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3288340400930531,
      "learning_rate": 5.5334861217214163e-05,
      "loss": 2.2879,
      "step": 1757
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2589555456763858,
      "learning_rate": 5.530939648586708e-05,
      "loss": 1.8638,
      "step": 1758
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23793950353880625,
      "learning_rate": 5.528393175451999e-05,
      "loss": 1.8433,
      "step": 1759
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24227639074433197,
      "learning_rate": 5.525846702317291e-05,
      "loss": 1.8814,
      "step": 1760
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24922419664655437,
      "learning_rate": 5.523300229182582e-05,
      "loss": 1.8689,
      "step": 1761
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24870996379466973,
      "learning_rate": 5.520753756047874e-05,
      "loss": 1.7643,
      "step": 1762
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2596777573742192,
      "learning_rate": 5.5182072829131656e-05,
      "loss": 2.0185,
      "step": 1763
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2499650915927096,
      "learning_rate": 5.515660809778457e-05,
      "loss": 2.0008,
      "step": 1764
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2318288787564209,
      "learning_rate": 5.513114336643749e-05,
      "loss": 1.8108,
      "step": 1765
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27257386468568495,
      "learning_rate": 5.5105678635090406e-05,
      "loss": 1.7635,
      "step": 1766
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2992005987613024,
      "learning_rate": 5.508021390374332e-05,
      "loss": 1.9779,
      "step": 1767
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23588669747860982,
      "learning_rate": 5.505474917239624e-05,
      "loss": 1.847,
      "step": 1768
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2188394993526254,
      "learning_rate": 5.502928444104914e-05,
      "loss": 1.6067,
      "step": 1769
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2695534600215359,
      "learning_rate": 5.500381970970207e-05,
      "loss": 2.0661,
      "step": 1770
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22736149518737195,
      "learning_rate": 5.497835497835499e-05,
      "loss": 1.6571,
      "step": 1771
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.25814689277838104,
      "learning_rate": 5.495289024700789e-05,
      "loss": 1.9552,
      "step": 1772
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2238504358725983,
      "learning_rate": 5.492742551566081e-05,
      "loss": 1.5043,
      "step": 1773
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.268667772932315,
      "learning_rate": 5.490196078431373e-05,
      "loss": 1.7679,
      "step": 1774
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27460783772153585,
      "learning_rate": 5.487649605296664e-05,
      "loss": 2.1356,
      "step": 1775
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2458523368494746,
      "learning_rate": 5.485103132161956e-05,
      "loss": 1.7842,
      "step": 1776
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.25243776123248696,
      "learning_rate": 5.482556659027247e-05,
      "loss": 1.9312,
      "step": 1777
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.265612531818153,
      "learning_rate": 5.480010185892539e-05,
      "loss": 2.0646,
      "step": 1778
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2769188648549516,
      "learning_rate": 5.477463712757831e-05,
      "loss": 1.9808,
      "step": 1779
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2503357212615843,
      "learning_rate": 5.4749172396231216e-05,
      "loss": 1.9653,
      "step": 1780
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2303122134316886,
      "learning_rate": 5.472370766488414e-05,
      "loss": 1.4754,
      "step": 1781
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20188287439135372,
      "learning_rate": 5.469824293353706e-05,
      "loss": 1.7278,
      "step": 1782
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2503709425309386,
      "learning_rate": 5.4672778202189966e-05,
      "loss": 1.8916,
      "step": 1783
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.25742335282915996,
      "learning_rate": 5.4647313470842884e-05,
      "loss": 1.9465,
      "step": 1784
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24559692914006714,
      "learning_rate": 5.4621848739495796e-05,
      "loss": 2.2193,
      "step": 1785
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23153806188074785,
      "learning_rate": 5.4596384008148715e-05,
      "loss": 1.8863,
      "step": 1786
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26187621546533785,
      "learning_rate": 5.4570919276801634e-05,
      "loss": 1.7773,
      "step": 1787
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24259248835353403,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 1.8412,
      "step": 1788
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24390738217557198,
      "learning_rate": 5.4519989814107465e-05,
      "loss": 1.9268,
      "step": 1789
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26336608391576083,
      "learning_rate": 5.4494525082760384e-05,
      "loss": 1.903,
      "step": 1790
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26897674687117284,
      "learning_rate": 5.446906035141329e-05,
      "loss": 1.9287,
      "step": 1791
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2731898821489289,
      "learning_rate": 5.4443595620066215e-05,
      "loss": 1.879,
      "step": 1792
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2379223434399539,
      "learning_rate": 5.4418130888719134e-05,
      "loss": 1.8735,
      "step": 1793
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25522130654246694,
      "learning_rate": 5.439266615737204e-05,
      "loss": 1.7318,
      "step": 1794
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2856176119126413,
      "learning_rate": 5.436720142602496e-05,
      "loss": 2.0151,
      "step": 1795
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23155621796811385,
      "learning_rate": 5.434173669467787e-05,
      "loss": 1.7329,
      "step": 1796
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26792895852977083,
      "learning_rate": 5.431627196333079e-05,
      "loss": 1.9689,
      "step": 1797
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2530283201692764,
      "learning_rate": 5.429080723198371e-05,
      "loss": 1.9712,
      "step": 1798
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27743368973922417,
      "learning_rate": 5.426534250063662e-05,
      "loss": 1.7018,
      "step": 1799
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2177459459956652,
      "learning_rate": 5.423987776928954e-05,
      "loss": 1.63,
      "step": 1800
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2668439859821881,
      "learning_rate": 5.421441303794246e-05,
      "loss": 2.0334,
      "step": 1801
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.245502397208931,
      "learning_rate": 5.418894830659537e-05,
      "loss": 1.9641,
      "step": 1802
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2550140861285763,
      "learning_rate": 5.416348357524829e-05,
      "loss": 2.1039,
      "step": 1803
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26067228487509947,
      "learning_rate": 5.4138018843901194e-05,
      "loss": 1.9323,
      "step": 1804
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27090074391060487,
      "learning_rate": 5.411255411255411e-05,
      "loss": 1.7093,
      "step": 1805
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21757379284420492,
      "learning_rate": 5.408708938120703e-05,
      "loss": 1.5382,
      "step": 1806
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.339007192047313,
      "learning_rate": 5.4061624649859943e-05,
      "loss": 2.0086,
      "step": 1807
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20279521947870238,
      "learning_rate": 5.403615991851286e-05,
      "loss": 1.6432,
      "step": 1808
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28961860158386293,
      "learning_rate": 5.401069518716578e-05,
      "loss": 1.9225,
      "step": 1809
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3011554396878597,
      "learning_rate": 5.398523045581869e-05,
      "loss": 1.9699,
      "step": 1810
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23162855467459323,
      "learning_rate": 5.395976572447161e-05,
      "loss": 1.5771,
      "step": 1811
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24026675897680333,
      "learning_rate": 5.393430099312452e-05,
      "loss": 1.8187,
      "step": 1812
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23333424918528609,
      "learning_rate": 5.390883626177744e-05,
      "loss": 1.9024,
      "step": 1813
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.232109597333444,
      "learning_rate": 5.388337153043036e-05,
      "loss": 1.7673,
      "step": 1814
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25877377446919725,
      "learning_rate": 5.385790679908327e-05,
      "loss": 1.9995,
      "step": 1815
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2717296535197353,
      "learning_rate": 5.3832442067736186e-05,
      "loss": 1.9133,
      "step": 1816
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2588506805190288,
      "learning_rate": 5.380697733638911e-05,
      "loss": 1.9533,
      "step": 1817
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22576180740646443,
      "learning_rate": 5.378151260504202e-05,
      "loss": 1.7461,
      "step": 1818
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3066987268108326,
      "learning_rate": 5.3756047873694936e-05,
      "loss": 2.0566,
      "step": 1819
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2844623326773415,
      "learning_rate": 5.373058314234785e-05,
      "loss": 1.8276,
      "step": 1820
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.29458131137047866,
      "learning_rate": 5.370511841100077e-05,
      "loss": 2.1424,
      "step": 1821
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36238895063597765,
      "learning_rate": 5.3679653679653686e-05,
      "loss": 2.0555,
      "step": 1822
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24611408664295198,
      "learning_rate": 5.365418894830659e-05,
      "loss": 1.7803,
      "step": 1823
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3008588467863008,
      "learning_rate": 5.3628724216959516e-05,
      "loss": 1.9502,
      "step": 1824
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27450625136992457,
      "learning_rate": 5.3603259485612435e-05,
      "loss": 2.0259,
      "step": 1825
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2666855843857169,
      "learning_rate": 5.357779475426534e-05,
      "loss": 1.8695,
      "step": 1826
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28164166094500825,
      "learning_rate": 5.355233002291826e-05,
      "loss": 2.0724,
      "step": 1827
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23709120390140725,
      "learning_rate": 5.352686529157117e-05,
      "loss": 1.9491,
      "step": 1828
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23017238700218814,
      "learning_rate": 5.350140056022409e-05,
      "loss": 1.743,
      "step": 1829
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.271085734619902,
      "learning_rate": 5.347593582887701e-05,
      "loss": 1.9874,
      "step": 1830
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23922151562707258,
      "learning_rate": 5.345047109752992e-05,
      "loss": 1.9307,
      "step": 1831
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2518359954393692,
      "learning_rate": 5.342500636618284e-05,
      "loss": 1.8122,
      "step": 1832
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24450502285449374,
      "learning_rate": 5.339954163483576e-05,
      "loss": 1.6499,
      "step": 1833
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2874697544939995,
      "learning_rate": 5.337407690348867e-05,
      "loss": 2.2582,
      "step": 1834
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2691866980594534,
      "learning_rate": 5.334861217214159e-05,
      "loss": 1.8869,
      "step": 1835
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2238018696217045,
      "learning_rate": 5.332314744079451e-05,
      "loss": 1.9598,
      "step": 1836
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23242987290413744,
      "learning_rate": 5.3297682709447414e-05,
      "loss": 1.8713,
      "step": 1837
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.27281451782756266,
      "learning_rate": 5.327221797810033e-05,
      "loss": 1.9389,
      "step": 1838
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2587967493607052,
      "learning_rate": 5.3246753246753245e-05,
      "loss": 2.1464,
      "step": 1839
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23820474052944823,
      "learning_rate": 5.3221288515406164e-05,
      "loss": 1.8616,
      "step": 1840
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3003991908326661,
      "learning_rate": 5.319582378405908e-05,
      "loss": 1.8697,
      "step": 1841
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24640073248782213,
      "learning_rate": 5.3170359052711995e-05,
      "loss": 1.7793,
      "step": 1842
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.25041861057849335,
      "learning_rate": 5.3144894321364914e-05,
      "loss": 1.876,
      "step": 1843
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.25193421344605454,
      "learning_rate": 5.311942959001783e-05,
      "loss": 1.8664,
      "step": 1844
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2665414008704307,
      "learning_rate": 5.3093964858670745e-05,
      "loss": 1.8365,
      "step": 1845
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22118430726236193,
      "learning_rate": 5.3068500127323663e-05,
      "loss": 1.6602,
      "step": 1846
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.21641116153913692,
      "learning_rate": 5.304303539597657e-05,
      "loss": 1.7391,
      "step": 1847
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22607660807259428,
      "learning_rate": 5.301757066462949e-05,
      "loss": 1.7965,
      "step": 1848
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.29830176926819324,
      "learning_rate": 5.299210593328241e-05,
      "loss": 1.6232,
      "step": 1849
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2354726123936021,
      "learning_rate": 5.296664120193532e-05,
      "loss": 1.777,
      "step": 1850
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.257301401939976,
      "learning_rate": 5.294117647058824e-05,
      "loss": 1.7204,
      "step": 1851
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36778423842474345,
      "learning_rate": 5.2915711739241156e-05,
      "loss": 1.949,
      "step": 1852
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.25158026555515783,
      "learning_rate": 5.289024700789407e-05,
      "loss": 1.6875,
      "step": 1853
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20881459657209697,
      "learning_rate": 5.286478227654699e-05,
      "loss": 1.7929,
      "step": 1854
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.256358090142942,
      "learning_rate": 5.283931754519989e-05,
      "loss": 1.6186,
      "step": 1855
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2272036130351749,
      "learning_rate": 5.281385281385282e-05,
      "loss": 1.7572,
      "step": 1856
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35342186796256736,
      "learning_rate": 5.278838808250574e-05,
      "loss": 2.3487,
      "step": 1857
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2707479354519031,
      "learning_rate": 5.276292335115864e-05,
      "loss": 1.7802,
      "step": 1858
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23101653447518822,
      "learning_rate": 5.273745861981156e-05,
      "loss": 1.7922,
      "step": 1859
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2595925292186933,
      "learning_rate": 5.271199388846449e-05,
      "loss": 1.7631,
      "step": 1860
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.30190522873815023,
      "learning_rate": 5.268652915711739e-05,
      "loss": 2.0733,
      "step": 1861
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.26821170058681754,
      "learning_rate": 5.266106442577031e-05,
      "loss": 1.724,
      "step": 1862
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2664055016023298,
      "learning_rate": 5.263559969442322e-05,
      "loss": 2.0899,
      "step": 1863
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.29562273179891857,
      "learning_rate": 5.261013496307614e-05,
      "loss": 2.1661,
      "step": 1864
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24865126380269392,
      "learning_rate": 5.258467023172906e-05,
      "loss": 1.7643,
      "step": 1865
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2373386719783883,
      "learning_rate": 5.255920550038197e-05,
      "loss": 1.8237,
      "step": 1866
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2553863812569826,
      "learning_rate": 5.253374076903489e-05,
      "loss": 1.8489,
      "step": 1867
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26032612498186936,
      "learning_rate": 5.250827603768781e-05,
      "loss": 1.6922,
      "step": 1868
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.22680692806164088,
      "learning_rate": 5.2482811306340716e-05,
      "loss": 1.8473,
      "step": 1869
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24847787672899788,
      "learning_rate": 5.2457346574993635e-05,
      "loss": 1.6559,
      "step": 1870
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2480300987758116,
      "learning_rate": 5.243188184364656e-05,
      "loss": 1.9783,
      "step": 1871
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23003737440119815,
      "learning_rate": 5.2406417112299466e-05,
      "loss": 1.7818,
      "step": 1872
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26150164922956376,
      "learning_rate": 5.2380952380952384e-05,
      "loss": 1.8185,
      "step": 1873
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2949240981264513,
      "learning_rate": 5.2355487649605297e-05,
      "loss": 1.8645,
      "step": 1874
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26624205387149846,
      "learning_rate": 5.2330022918258215e-05,
      "loss": 1.8419,
      "step": 1875
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2664165203989785,
      "learning_rate": 5.2304558186911134e-05,
      "loss": 2.0172,
      "step": 1876
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26079884460045555,
      "learning_rate": 5.2279093455564046e-05,
      "loss": 2.0923,
      "step": 1877
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2581785955045546,
      "learning_rate": 5.2253628724216965e-05,
      "loss": 1.7634,
      "step": 1878
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23690102235485708,
      "learning_rate": 5.2228163992869884e-05,
      "loss": 1.8505,
      "step": 1879
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24509423745002412,
      "learning_rate": 5.220269926152279e-05,
      "loss": 1.7104,
      "step": 1880
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2330443277708947,
      "learning_rate": 5.217723453017571e-05,
      "loss": 1.6868,
      "step": 1881
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26653664885071376,
      "learning_rate": 5.215176979882862e-05,
      "loss": 2.1852,
      "step": 1882
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23328075873903847,
      "learning_rate": 5.212630506748154e-05,
      "loss": 1.6535,
      "step": 1883
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26751635875771723,
      "learning_rate": 5.210084033613446e-05,
      "loss": 1.8256,
      "step": 1884
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2456733351995663,
      "learning_rate": 5.207537560478737e-05,
      "loss": 2.0157,
      "step": 1885
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2917524265614974,
      "learning_rate": 5.204991087344029e-05,
      "loss": 1.7418,
      "step": 1886
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23558486360113196,
      "learning_rate": 5.202444614209321e-05,
      "loss": 1.563,
      "step": 1887
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23351339278521402,
      "learning_rate": 5.199898141074612e-05,
      "loss": 1.9542,
      "step": 1888
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24245373643445367,
      "learning_rate": 5.197351667939904e-05,
      "loss": 1.6765,
      "step": 1889
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26288320068940174,
      "learning_rate": 5.1948051948051944e-05,
      "loss": 1.9594,
      "step": 1890
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27609982717547316,
      "learning_rate": 5.192258721670486e-05,
      "loss": 1.8476,
      "step": 1891
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24672020543921191,
      "learning_rate": 5.189712248535779e-05,
      "loss": 2.0675,
      "step": 1892
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.22504124793150443,
      "learning_rate": 5.1871657754010694e-05,
      "loss": 1.8047,
      "step": 1893
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2836563430101798,
      "learning_rate": 5.184619302266361e-05,
      "loss": 2.2403,
      "step": 1894
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.25871308848958324,
      "learning_rate": 5.182072829131653e-05,
      "loss": 1.6891,
      "step": 1895
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2565945263108578,
      "learning_rate": 5.1795263559969444e-05,
      "loss": 1.8522,
      "step": 1896
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30505080448764227,
      "learning_rate": 5.176979882862236e-05,
      "loss": 1.9909,
      "step": 1897
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21616474130190202,
      "learning_rate": 5.174433409727527e-05,
      "loss": 1.9069,
      "step": 1898
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3571246636388768,
      "learning_rate": 5.171886936592819e-05,
      "loss": 2.0079,
      "step": 1899
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2444983341378719,
      "learning_rate": 5.169340463458111e-05,
      "loss": 1.9935,
      "step": 1900
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.286823805394391,
      "learning_rate": 5.166793990323402e-05,
      "loss": 1.9886,
      "step": 1901
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27263026291263703,
      "learning_rate": 5.1642475171886936e-05,
      "loss": 2.0349,
      "step": 1902
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19583473006386165,
      "learning_rate": 5.161701044053986e-05,
      "loss": 1.5778,
      "step": 1903
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30509354360695173,
      "learning_rate": 5.159154570919277e-05,
      "loss": 1.9218,
      "step": 1904
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23055331246706456,
      "learning_rate": 5.1566080977845686e-05,
      "loss": 1.8517,
      "step": 1905
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25433625708835156,
      "learning_rate": 5.15406162464986e-05,
      "loss": 1.8759,
      "step": 1906
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2331749632299269,
      "learning_rate": 5.151515151515152e-05,
      "loss": 1.8752,
      "step": 1907
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.27945951248133655,
      "learning_rate": 5.1489686783804436e-05,
      "loss": 1.7539,
      "step": 1908
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.266353240010889,
      "learning_rate": 5.146422205245735e-05,
      "loss": 1.672,
      "step": 1909
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2553481753624911,
      "learning_rate": 5.143875732111027e-05,
      "loss": 1.7656,
      "step": 1910
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26020543432542415,
      "learning_rate": 5.1413292589763186e-05,
      "loss": 1.8238,
      "step": 1911
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26424780218388744,
      "learning_rate": 5.138782785841609e-05,
      "loss": 1.8434,
      "step": 1912
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2711295134510327,
      "learning_rate": 5.136236312706901e-05,
      "loss": 1.9735,
      "step": 1913
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2157097097218708,
      "learning_rate": 5.1336898395721935e-05,
      "loss": 1.626,
      "step": 1914
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32121250434724363,
      "learning_rate": 5.131143366437484e-05,
      "loss": 2.1219,
      "step": 1915
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2880472706853323,
      "learning_rate": 5.128596893302776e-05,
      "loss": 2.1724,
      "step": 1916
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24202615224002272,
      "learning_rate": 5.126050420168067e-05,
      "loss": 2.102,
      "step": 1917
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20535616419096947,
      "learning_rate": 5.123503947033359e-05,
      "loss": 1.5963,
      "step": 1918
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24119444261746867,
      "learning_rate": 5.120957473898651e-05,
      "loss": 1.7542,
      "step": 1919
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.27009931768636947,
      "learning_rate": 5.118411000763942e-05,
      "loss": 1.8226,
      "step": 1920
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2702193253961999,
      "learning_rate": 5.115864527629234e-05,
      "loss": 1.9364,
      "step": 1921
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2631773372890249,
      "learning_rate": 5.113318054494526e-05,
      "loss": 1.9965,
      "step": 1922
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26041596253617777,
      "learning_rate": 5.1107715813598164e-05,
      "loss": 1.6373,
      "step": 1923
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3008801929182643,
      "learning_rate": 5.108225108225109e-05,
      "loss": 2.1005,
      "step": 1924
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2817868698243369,
      "learning_rate": 5.1056786350903995e-05,
      "loss": 1.9543,
      "step": 1925
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23752655834477365,
      "learning_rate": 5.1031321619556914e-05,
      "loss": 1.7105,
      "step": 1926
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29655186735574096,
      "learning_rate": 5.100585688820983e-05,
      "loss": 1.9634,
      "step": 1927
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.20010623345672113,
      "learning_rate": 5.0980392156862745e-05,
      "loss": 1.6691,
      "step": 1928
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28829755607740787,
      "learning_rate": 5.0954927425515664e-05,
      "loss": 1.9682,
      "step": 1929
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28987426442953185,
      "learning_rate": 5.092946269416858e-05,
      "loss": 1.9083,
      "step": 1930
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2243905203666095,
      "learning_rate": 5.0903997962821495e-05,
      "loss": 1.6901,
      "step": 1931
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2535684686258705,
      "learning_rate": 5.0878533231474414e-05,
      "loss": 2.0169,
      "step": 1932
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21911503783525632,
      "learning_rate": 5.085306850012732e-05,
      "loss": 1.6033,
      "step": 1933
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25088447335950004,
      "learning_rate": 5.082760376878024e-05,
      "loss": 1.9437,
      "step": 1934
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2449779799112241,
      "learning_rate": 5.0802139037433164e-05,
      "loss": 1.8041,
      "step": 1935
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2579811440487833,
      "learning_rate": 5.077667430608607e-05,
      "loss": 1.7179,
      "step": 1936
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25607906257541085,
      "learning_rate": 5.075120957473899e-05,
      "loss": 1.8907,
      "step": 1937
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26921371054512905,
      "learning_rate": 5.0725744843391907e-05,
      "loss": 1.9184,
      "step": 1938
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2804322806107065,
      "learning_rate": 5.070028011204482e-05,
      "loss": 1.9893,
      "step": 1939
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2983559012579139,
      "learning_rate": 5.067481538069774e-05,
      "loss": 2.0935,
      "step": 1940
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22076470857762523,
      "learning_rate": 5.064935064935065e-05,
      "loss": 1.6101,
      "step": 1941
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22326575894872008,
      "learning_rate": 5.062388591800357e-05,
      "loss": 1.692,
      "step": 1942
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24204786951620316,
      "learning_rate": 5.059842118665649e-05,
      "loss": 1.9055,
      "step": 1943
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25870577353603014,
      "learning_rate": 5.057295645530939e-05,
      "loss": 1.9083,
      "step": 1944
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3110877551004597,
      "learning_rate": 5.054749172396231e-05,
      "loss": 2.0132,
      "step": 1945
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2437482274431188,
      "learning_rate": 5.052202699261524e-05,
      "loss": 1.6122,
      "step": 1946
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2994332264479995,
      "learning_rate": 5.049656226126814e-05,
      "loss": 1.8626,
      "step": 1947
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24268224389050214,
      "learning_rate": 5.047109752992106e-05,
      "loss": 1.7933,
      "step": 1948
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2883783053021446,
      "learning_rate": 5.044563279857398e-05,
      "loss": 1.8863,
      "step": 1949
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22717169546645896,
      "learning_rate": 5.042016806722689e-05,
      "loss": 1.8561,
      "step": 1950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23447344619764077,
      "learning_rate": 5.039470333587981e-05,
      "loss": 1.849,
      "step": 1951
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24362514244631286,
      "learning_rate": 5.036923860453272e-05,
      "loss": 1.8075,
      "step": 1952
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26878464996614415,
      "learning_rate": 5.034377387318564e-05,
      "loss": 2.2207,
      "step": 1953
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.27428559523738466,
      "learning_rate": 5.031830914183856e-05,
      "loss": 1.9848,
      "step": 1954
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23696908649902115,
      "learning_rate": 5.0292844410491466e-05,
      "loss": 1.9916,
      "step": 1955
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2436881946799392,
      "learning_rate": 5.026737967914439e-05,
      "loss": 1.8764,
      "step": 1956
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26938643227677544,
      "learning_rate": 5.024191494779731e-05,
      "loss": 1.7523,
      "step": 1957
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25242396281496726,
      "learning_rate": 5.0216450216450216e-05,
      "loss": 1.9699,
      "step": 1958
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21741305241176564,
      "learning_rate": 5.0190985485103135e-05,
      "loss": 1.6423,
      "step": 1959
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3028444643737473,
      "learning_rate": 5.016552075375605e-05,
      "loss": 1.8337,
      "step": 1960
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25914168894988954,
      "learning_rate": 5.0140056022408966e-05,
      "loss": 2.0263,
      "step": 1961
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23670939751206618,
      "learning_rate": 5.0114591291061884e-05,
      "loss": 1.7366,
      "step": 1962
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24498226814427296,
      "learning_rate": 5.0089126559714797e-05,
      "loss": 1.8961,
      "step": 1963
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25565121460317763,
      "learning_rate": 5.0063661828367715e-05,
      "loss": 1.6696,
      "step": 1964
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.27791356933350964,
      "learning_rate": 5.0038197097020634e-05,
      "loss": 1.9723,
      "step": 1965
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29225794713005676,
      "learning_rate": 5.001273236567354e-05,
      "loss": 1.938,
      "step": 1966
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21643323730040626,
      "learning_rate": 4.9987267634326465e-05,
      "loss": 1.8585,
      "step": 1967
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23078934951650226,
      "learning_rate": 4.996180290297938e-05,
      "loss": 1.7416,
      "step": 1968
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22861821569749852,
      "learning_rate": 4.993633817163229e-05,
      "loss": 1.908,
      "step": 1969
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24688323103788157,
      "learning_rate": 4.991087344028521e-05,
      "loss": 1.7285,
      "step": 1970
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21804553938715954,
      "learning_rate": 4.988540870893813e-05,
      "loss": 1.8245,
      "step": 1971
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2532061682179545,
      "learning_rate": 4.985994397759104e-05,
      "loss": 1.9595,
      "step": 1972
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23886437277956296,
      "learning_rate": 4.983447924624395e-05,
      "loss": 1.735,
      "step": 1973
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2811014392346057,
      "learning_rate": 4.980901451489687e-05,
      "loss": 1.9136,
      "step": 1974
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2607547742684514,
      "learning_rate": 4.978354978354979e-05,
      "loss": 1.7918,
      "step": 1975
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26233559126035727,
      "learning_rate": 4.97580850522027e-05,
      "loss": 1.9276,
      "step": 1976
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24177929100967677,
      "learning_rate": 4.973262032085561e-05,
      "loss": 2.0133,
      "step": 1977
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2969375594630229,
      "learning_rate": 4.970715558950853e-05,
      "loss": 1.8052,
      "step": 1978
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24271014946897948,
      "learning_rate": 4.968169085816145e-05,
      "loss": 1.8089,
      "step": 1979
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22367153915182864,
      "learning_rate": 4.965622612681436e-05,
      "loss": 1.5761,
      "step": 1980
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23488780982079455,
      "learning_rate": 4.963076139546728e-05,
      "loss": 1.9722,
      "step": 1981
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22492626889472872,
      "learning_rate": 4.9605296664120194e-05,
      "loss": 1.7289,
      "step": 1982
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.22267668613544087,
      "learning_rate": 4.957983193277311e-05,
      "loss": 1.8544,
      "step": 1983
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2807021633506502,
      "learning_rate": 4.9554367201426025e-05,
      "loss": 1.9977,
      "step": 1984
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.26550099339272387,
      "learning_rate": 4.9528902470078944e-05,
      "loss": 1.9452,
      "step": 1985
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20578276966029152,
      "learning_rate": 4.950343773873186e-05,
      "loss": 1.6141,
      "step": 1986
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2470067285434968,
      "learning_rate": 4.9477973007384775e-05,
      "loss": 1.8066,
      "step": 1987
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22999601048399182,
      "learning_rate": 4.945250827603769e-05,
      "loss": 1.8548,
      "step": 1988
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2474929150617297,
      "learning_rate": 4.9427043544690605e-05,
      "loss": 1.8841,
      "step": 1989
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2333409865151707,
      "learning_rate": 4.9401578813343524e-05,
      "loss": 1.59,
      "step": 1990
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22675493716602751,
      "learning_rate": 4.9376114081996436e-05,
      "loss": 1.6562,
      "step": 1991
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2524665067431181,
      "learning_rate": 4.9350649350649355e-05,
      "loss": 1.8689,
      "step": 1992
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2626296603049425,
      "learning_rate": 4.932518461930227e-05,
      "loss": 1.8856,
      "step": 1993
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2791287776633305,
      "learning_rate": 4.9299719887955186e-05,
      "loss": 2.0856,
      "step": 1994
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22179263762228624,
      "learning_rate": 4.92742551566081e-05,
      "loss": 1.8493,
      "step": 1995
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2725309329893309,
      "learning_rate": 4.924879042526102e-05,
      "loss": 1.9911,
      "step": 1996
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2964582781055548,
      "learning_rate": 4.922332569391393e-05,
      "loss": 1.9537,
      "step": 1997
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2498195712358318,
      "learning_rate": 4.919786096256685e-05,
      "loss": 1.9309,
      "step": 1998
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2215946387263853,
      "learning_rate": 4.917239623121977e-05,
      "loss": 1.8822,
      "step": 1999
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23079992196823038,
      "learning_rate": 4.914693149987268e-05,
      "loss": 1.8082,
      "step": 2000
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3929,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}