{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 461,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0021691973969631237,
      "grad_norm": 2.11081600189209,
      "learning_rate": 1e-05,
      "loss": 2.6121,
      "step": 1
    },
    {
      "epoch": 0.004338394793926247,
      "grad_norm": 1.9794774055480957,
      "learning_rate": 9.97830802603037e-06,
      "loss": 2.5875,
      "step": 2
    },
    {
      "epoch": 0.006507592190889371,
      "grad_norm": 1.8341801166534424,
      "learning_rate": 9.956616052060738e-06,
      "loss": 2.6118,
      "step": 3
    },
    {
      "epoch": 0.008676789587852495,
      "grad_norm": 1.5984541177749634,
      "learning_rate": 9.934924078091108e-06,
      "loss": 2.5194,
      "step": 4
    },
    {
      "epoch": 0.010845986984815618,
      "grad_norm": 1.3061403036117554,
      "learning_rate": 9.913232104121477e-06,
      "loss": 2.4602,
      "step": 5
    },
    {
      "epoch": 0.013015184381778741,
      "grad_norm": 1.0785109996795654,
      "learning_rate": 9.891540130151845e-06,
      "loss": 2.4176,
      "step": 6
    },
    {
      "epoch": 0.015184381778741865,
      "grad_norm": 0.8170204758644104,
      "learning_rate": 9.869848156182214e-06,
      "loss": 2.315,
      "step": 7
    },
    {
      "epoch": 0.01735357917570499,
      "grad_norm": 0.7127103209495544,
      "learning_rate": 9.848156182212582e-06,
      "loss": 2.2658,
      "step": 8
    },
    {
      "epoch": 0.019522776572668113,
      "grad_norm": 0.7081116437911987,
      "learning_rate": 9.82646420824295e-06,
      "loss": 2.281,
      "step": 9
    },
    {
      "epoch": 0.021691973969631236,
      "grad_norm": 0.768051028251648,
      "learning_rate": 9.80477223427332e-06,
      "loss": 2.2928,
      "step": 10
    },
    {
      "epoch": 0.02386117136659436,
      "grad_norm": 0.8635964393615723,
      "learning_rate": 9.783080260303689e-06,
      "loss": 2.2291,
      "step": 11
    },
    {
      "epoch": 0.026030368763557483,
      "grad_norm": 0.8809370398521423,
      "learning_rate": 9.761388286334057e-06,
      "loss": 2.2048,
      "step": 12
    },
    {
      "epoch": 0.028199566160520606,
      "grad_norm": 0.908088207244873,
      "learning_rate": 9.739696312364426e-06,
      "loss": 2.1956,
      "step": 13
    },
    {
      "epoch": 0.03036876355748373,
      "grad_norm": 1.0039963722229004,
      "learning_rate": 9.718004338394794e-06,
      "loss": 2.1148,
      "step": 14
    },
    {
      "epoch": 0.03253796095444685,
      "grad_norm": 0.907769501209259,
      "learning_rate": 9.696312364425163e-06,
      "loss": 2.0764,
      "step": 15
    },
    {
      "epoch": 0.03470715835140998,
      "grad_norm": 0.8381527066230774,
      "learning_rate": 9.674620390455533e-06,
      "loss": 2.0732,
      "step": 16
    },
    {
      "epoch": 0.0368763557483731,
      "grad_norm": 0.7855611443519592,
      "learning_rate": 9.6529284164859e-06,
      "loss": 1.9714,
      "step": 17
    },
    {
      "epoch": 0.039045553145336226,
      "grad_norm": 0.7366162538528442,
      "learning_rate": 9.63123644251627e-06,
      "loss": 1.965,
      "step": 18
    },
    {
      "epoch": 0.04121475054229935,
      "grad_norm": 0.6809542179107666,
      "learning_rate": 9.60954446854664e-06,
      "loss": 2.0019,
      "step": 19
    },
    {
      "epoch": 0.04338394793926247,
      "grad_norm": 0.6486644744873047,
      "learning_rate": 9.587852494577007e-06,
      "loss": 1.9383,
      "step": 20
    },
    {
      "epoch": 0.0455531453362256,
      "grad_norm": 0.6221187710762024,
      "learning_rate": 9.566160520607377e-06,
      "loss": 1.883,
      "step": 21
    },
    {
      "epoch": 0.04772234273318872,
      "grad_norm": 0.5827253460884094,
      "learning_rate": 9.544468546637745e-06,
      "loss": 1.8966,
      "step": 22
    },
    {
      "epoch": 0.049891540130151846,
      "grad_norm": 0.5862624049186707,
      "learning_rate": 9.522776572668114e-06,
      "loss": 1.8187,
      "step": 23
    },
    {
      "epoch": 0.052060737527114966,
      "grad_norm": 0.6064738035202026,
      "learning_rate": 9.501084598698482e-06,
      "loss": 1.8233,
      "step": 24
    },
    {
      "epoch": 0.05422993492407809,
      "grad_norm": 0.6191762685775757,
      "learning_rate": 9.479392624728851e-06,
      "loss": 1.7323,
      "step": 25
    },
    {
      "epoch": 0.05639913232104121,
      "grad_norm": 0.6317607164382935,
      "learning_rate": 9.457700650759219e-06,
      "loss": 1.7398,
      "step": 26
    },
    {
      "epoch": 0.05856832971800434,
      "grad_norm": 0.6248988509178162,
      "learning_rate": 9.436008676789589e-06,
      "loss": 1.7171,
      "step": 27
    },
    {
      "epoch": 0.06073752711496746,
      "grad_norm": 0.6342098712921143,
      "learning_rate": 9.414316702819958e-06,
      "loss": 1.6891,
      "step": 28
    },
    {
      "epoch": 0.06290672451193059,
      "grad_norm": 0.6058349609375,
      "learning_rate": 9.392624728850326e-06,
      "loss": 1.6624,
      "step": 29
    },
    {
      "epoch": 0.0650759219088937,
      "grad_norm": 0.5988097786903381,
      "learning_rate": 9.370932754880695e-06,
      "loss": 1.6438,
      "step": 30
    },
    {
      "epoch": 0.06724511930585683,
      "grad_norm": 0.6042107939720154,
      "learning_rate": 9.349240780911065e-06,
      "loss": 1.5956,
      "step": 31
    },
    {
      "epoch": 0.06941431670281996,
      "grad_norm": 0.5972896218299866,
      "learning_rate": 9.327548806941433e-06,
      "loss": 1.5142,
      "step": 32
    },
    {
      "epoch": 0.07158351409978309,
      "grad_norm": 0.5832527875900269,
      "learning_rate": 9.305856832971802e-06,
      "loss": 1.5384,
      "step": 33
    },
    {
      "epoch": 0.0737527114967462,
      "grad_norm": 0.5889501571655273,
      "learning_rate": 9.28416485900217e-06,
      "loss": 1.5183,
      "step": 34
    },
    {
      "epoch": 0.07592190889370933,
      "grad_norm": 0.622042715549469,
      "learning_rate": 9.26247288503254e-06,
      "loss": 1.4945,
      "step": 35
    },
    {
      "epoch": 0.07809110629067245,
      "grad_norm": 0.630580723285675,
      "learning_rate": 9.240780911062907e-06,
      "loss": 1.4925,
      "step": 36
    },
    {
      "epoch": 0.08026030368763558,
      "grad_norm": 0.5893872976303101,
      "learning_rate": 9.219088937093276e-06,
      "loss": 1.4306,
      "step": 37
    },
    {
      "epoch": 0.0824295010845987,
      "grad_norm": 0.5707374811172485,
      "learning_rate": 9.197396963123644e-06,
      "loss": 1.418,
      "step": 38
    },
    {
      "epoch": 0.08459869848156182,
      "grad_norm": 0.5506007075309753,
      "learning_rate": 9.175704989154014e-06,
      "loss": 1.3998,
      "step": 39
    },
    {
      "epoch": 0.08676789587852494,
      "grad_norm": 0.5373205542564392,
      "learning_rate": 9.154013015184382e-06,
      "loss": 1.3508,
      "step": 40
    },
    {
      "epoch": 0.08893709327548807,
      "grad_norm": 0.5845463871955872,
      "learning_rate": 9.132321041214751e-06,
      "loss": 1.3387,
      "step": 41
    },
    {
      "epoch": 0.0911062906724512,
      "grad_norm": 0.6560288071632385,
      "learning_rate": 9.11062906724512e-06,
      "loss": 1.3078,
      "step": 42
    },
    {
      "epoch": 0.09327548806941431,
      "grad_norm": 0.580771267414093,
      "learning_rate": 9.088937093275488e-06,
      "loss": 1.309,
      "step": 43
    },
    {
      "epoch": 0.09544468546637744,
      "grad_norm": 0.5993460416793823,
      "learning_rate": 9.067245119305858e-06,
      "loss": 1.3005,
      "step": 44
    },
    {
      "epoch": 0.09761388286334056,
      "grad_norm": 0.6303313374519348,
      "learning_rate": 9.045553145336227e-06,
      "loss": 1.2776,
      "step": 45
    },
    {
      "epoch": 0.09978308026030369,
      "grad_norm": 0.5471832156181335,
      "learning_rate": 9.023861171366595e-06,
      "loss": 1.2588,
      "step": 46
    },
    {
      "epoch": 0.1019522776572668,
      "grad_norm": 0.5459536910057068,
      "learning_rate": 9.002169197396964e-06,
      "loss": 1.2094,
      "step": 47
    },
    {
      "epoch": 0.10412147505422993,
      "grad_norm": 0.5408421754837036,
      "learning_rate": 8.980477223427332e-06,
      "loss": 1.2355,
      "step": 48
    },
    {
      "epoch": 0.10629067245119306,
      "grad_norm": 0.48613736033439636,
      "learning_rate": 8.958785249457702e-06,
      "loss": 1.1594,
      "step": 49
    },
    {
      "epoch": 0.10845986984815618,
      "grad_norm": 0.4992137849330902,
      "learning_rate": 8.93709327548807e-06,
      "loss": 1.1853,
      "step": 50
    },
    {
      "epoch": 0.11062906724511931,
      "grad_norm": 0.4883258044719696,
      "learning_rate": 8.915401301518439e-06,
      "loss": 1.1467,
      "step": 51
    },
    {
      "epoch": 0.11279826464208242,
      "grad_norm": 0.48137181997299194,
      "learning_rate": 8.893709327548807e-06,
      "loss": 1.1541,
      "step": 52
    },
    {
      "epoch": 0.11496746203904555,
      "grad_norm": 0.4401426911354065,
      "learning_rate": 8.872017353579176e-06,
      "loss": 1.1112,
      "step": 53
    },
    {
      "epoch": 0.11713665943600868,
      "grad_norm": 0.4389711320400238,
      "learning_rate": 8.850325379609546e-06,
      "loss": 1.0961,
      "step": 54
    },
    {
      "epoch": 0.1193058568329718,
      "grad_norm": 0.439217746257782,
      "learning_rate": 8.828633405639913e-06,
      "loss": 1.0838,
      "step": 55
    },
    {
      "epoch": 0.12147505422993492,
      "grad_norm": 0.521327555179596,
      "learning_rate": 8.806941431670283e-06,
      "loss": 1.0813,
      "step": 56
    },
    {
      "epoch": 0.12364425162689804,
      "grad_norm": 0.4604964256286621,
      "learning_rate": 8.785249457700652e-06,
      "loss": 1.0857,
      "step": 57
    },
    {
      "epoch": 0.12581344902386118,
      "grad_norm": 0.42932406067848206,
      "learning_rate": 8.76355748373102e-06,
      "loss": 1.0513,
      "step": 58
    },
    {
      "epoch": 0.1279826464208243,
      "grad_norm": 0.43107494711875916,
      "learning_rate": 8.74186550976139e-06,
      "loss": 1.0488,
      "step": 59
    },
    {
      "epoch": 0.1301518438177874,
      "grad_norm": 0.42236247658729553,
      "learning_rate": 8.720173535791757e-06,
      "loss": 1.028,
      "step": 60
    },
    {
      "epoch": 0.13232104121475055,
      "grad_norm": 0.43343520164489746,
      "learning_rate": 8.698481561822127e-06,
      "loss": 1.0272,
      "step": 61
    },
    {
      "epoch": 0.13449023861171366,
      "grad_norm": 0.48012059926986694,
      "learning_rate": 8.676789587852495e-06,
      "loss": 1.0099,
      "step": 62
    },
    {
      "epoch": 0.13665943600867678,
      "grad_norm": 0.42098668217658997,
      "learning_rate": 8.655097613882864e-06,
      "loss": 1.0158,
      "step": 63
    },
    {
      "epoch": 0.13882863340563992,
      "grad_norm": 0.4227031171321869,
      "learning_rate": 8.633405639913232e-06,
      "loss": 0.9815,
      "step": 64
    },
    {
      "epoch": 0.14099783080260303,
      "grad_norm": 0.41489773988723755,
      "learning_rate": 8.611713665943601e-06,
      "loss": 0.9635,
      "step": 65
    },
    {
      "epoch": 0.14316702819956617,
      "grad_norm": 0.40091952681541443,
      "learning_rate": 8.59002169197397e-06,
      "loss": 0.9528,
      "step": 66
    },
    {
      "epoch": 0.14533622559652928,
      "grad_norm": 0.4547802805900574,
      "learning_rate": 8.568329718004339e-06,
      "loss": 0.9726,
      "step": 67
    },
    {
      "epoch": 0.1475054229934924,
      "grad_norm": 0.4234392046928406,
      "learning_rate": 8.546637744034708e-06,
      "loss": 0.962,
      "step": 68
    },
    {
      "epoch": 0.14967462039045554,
      "grad_norm": 0.38503268361091614,
      "learning_rate": 8.524945770065076e-06,
      "loss": 0.9463,
      "step": 69
    },
    {
      "epoch": 0.15184381778741865,
      "grad_norm": 0.4524126350879669,
      "learning_rate": 8.503253796095445e-06,
      "loss": 0.9061,
      "step": 70
    },
    {
      "epoch": 0.1540130151843818,
      "grad_norm": 0.40082183480262756,
      "learning_rate": 8.481561822125815e-06,
      "loss": 0.9034,
      "step": 71
    },
    {
      "epoch": 0.1561822125813449,
      "grad_norm": 0.41073668003082275,
      "learning_rate": 8.459869848156183e-06,
      "loss": 0.8961,
      "step": 72
    },
    {
      "epoch": 0.15835140997830802,
      "grad_norm": 0.5651035308837891,
      "learning_rate": 8.438177874186552e-06,
      "loss": 0.8851,
      "step": 73
    },
    {
      "epoch": 0.16052060737527116,
      "grad_norm": 0.44745904207229614,
      "learning_rate": 8.416485900216922e-06,
      "loss": 0.9203,
      "step": 74
    },
    {
      "epoch": 0.16268980477223427,
      "grad_norm": 0.4688916504383087,
      "learning_rate": 8.39479392624729e-06,
      "loss": 0.9449,
      "step": 75
    },
    {
      "epoch": 0.1648590021691974,
      "grad_norm": 0.386109322309494,
      "learning_rate": 8.373101952277657e-06,
      "loss": 0.868,
      "step": 76
    },
    {
      "epoch": 0.16702819956616052,
      "grad_norm": 0.47646722197532654,
      "learning_rate": 8.351409978308027e-06,
      "loss": 0.8853,
      "step": 77
    },
    {
      "epoch": 0.16919739696312364,
      "grad_norm": 0.38544556498527527,
      "learning_rate": 8.329718004338394e-06,
      "loss": 0.8703,
      "step": 78
    },
    {
      "epoch": 0.17136659436008678,
      "grad_norm": 0.5066248178482056,
      "learning_rate": 8.308026030368764e-06,
      "loss": 0.8963,
      "step": 79
    },
    {
      "epoch": 0.1735357917570499,
      "grad_norm": 0.471044659614563,
      "learning_rate": 8.286334056399133e-06,
      "loss": 0.8547,
      "step": 80
    },
    {
      "epoch": 0.175704989154013,
      "grad_norm": 0.4496097266674042,
      "learning_rate": 8.264642082429501e-06,
      "loss": 0.8556,
      "step": 81
    },
    {
      "epoch": 0.17787418655097614,
      "grad_norm": 0.40527230501174927,
      "learning_rate": 8.24295010845987e-06,
      "loss": 0.8756,
      "step": 82
    },
    {
      "epoch": 0.18004338394793926,
      "grad_norm": 0.40289977192878723,
      "learning_rate": 8.22125813449024e-06,
      "loss": 0.8412,
      "step": 83
    },
    {
      "epoch": 0.1822125813449024,
      "grad_norm": 0.3807680606842041,
      "learning_rate": 8.199566160520608e-06,
      "loss": 0.8241,
      "step": 84
    },
    {
      "epoch": 0.1843817787418655,
      "grad_norm": 0.4011147618293762,
      "learning_rate": 8.177874186550977e-06,
      "loss": 0.8308,
      "step": 85
    },
    {
      "epoch": 0.18655097613882862,
      "grad_norm": 0.6492615938186646,
      "learning_rate": 8.156182212581345e-06,
      "loss": 0.8091,
      "step": 86
    },
    {
      "epoch": 0.18872017353579176,
      "grad_norm": 0.45752060413360596,
      "learning_rate": 8.134490238611715e-06,
      "loss": 0.86,
      "step": 87
    },
    {
      "epoch": 0.19088937093275488,
      "grad_norm": 0.39894232153892517,
      "learning_rate": 8.112798264642084e-06,
      "loss": 0.8087,
      "step": 88
    },
    {
      "epoch": 0.19305856832971802,
      "grad_norm": 0.4044681787490845,
      "learning_rate": 8.091106290672452e-06,
      "loss": 0.8287,
      "step": 89
    },
    {
      "epoch": 0.19522776572668113,
      "grad_norm": 0.5479378700256348,
      "learning_rate": 8.06941431670282e-06,
      "loss": 0.7788,
      "step": 90
    },
    {
      "epoch": 0.19739696312364424,
      "grad_norm": 0.5609095096588135,
      "learning_rate": 8.04772234273319e-06,
      "loss": 0.8196,
      "step": 91
    },
    {
      "epoch": 0.19956616052060738,
      "grad_norm": 0.3801177144050598,
      "learning_rate": 8.026030368763557e-06,
      "loss": 0.7683,
      "step": 92
    },
    {
      "epoch": 0.2017353579175705,
      "grad_norm": 0.38417235016822815,
      "learning_rate": 8.004338394793926e-06,
      "loss": 0.7955,
      "step": 93
    },
    {
      "epoch": 0.2039045553145336,
      "grad_norm": 0.6401964426040649,
      "learning_rate": 7.982646420824296e-06,
      "loss": 0.7861,
      "step": 94
    },
    {
      "epoch": 0.20607375271149675,
      "grad_norm": 0.4698239266872406,
      "learning_rate": 7.960954446854664e-06,
      "loss": 0.7872,
      "step": 95
    },
    {
      "epoch": 0.20824295010845986,
      "grad_norm": 0.45705363154411316,
      "learning_rate": 7.939262472885033e-06,
      "loss": 0.7865,
      "step": 96
    },
    {
      "epoch": 0.210412147505423,
      "grad_norm": 0.4010639488697052,
      "learning_rate": 7.917570498915403e-06,
      "loss": 0.7641,
      "step": 97
    },
    {
      "epoch": 0.21258134490238612,
      "grad_norm": 0.40273797512054443,
      "learning_rate": 7.89587852494577e-06,
      "loss": 0.7719,
      "step": 98
    },
    {
      "epoch": 0.21475054229934923,
      "grad_norm": 0.4527304470539093,
      "learning_rate": 7.87418655097614e-06,
      "loss": 0.8008,
      "step": 99
    },
    {
      "epoch": 0.21691973969631237,
      "grad_norm": 0.432674765586853,
      "learning_rate": 7.85249457700651e-06,
      "loss": 0.7426,
      "step": 100
    },
    {
      "epoch": 0.21908893709327548,
      "grad_norm": 0.4265141785144806,
      "learning_rate": 7.830802603036877e-06,
      "loss": 0.7713,
      "step": 101
    },
    {
      "epoch": 0.22125813449023862,
      "grad_norm": 0.4253065288066864,
      "learning_rate": 7.809110629067247e-06,
      "loss": 0.7484,
      "step": 102
    },
    {
      "epoch": 0.22342733188720174,
      "grad_norm": 0.44442278146743774,
      "learning_rate": 7.787418655097614e-06,
      "loss": 0.763,
      "step": 103
    },
    {
      "epoch": 0.22559652928416485,
      "grad_norm": 0.47105884552001953,
      "learning_rate": 7.765726681127982e-06,
      "loss": 0.7703,
      "step": 104
    },
    {
      "epoch": 0.227765726681128,
      "grad_norm": 0.41451483964920044,
      "learning_rate": 7.744034707158352e-06,
      "loss": 0.7558,
      "step": 105
    },
    {
      "epoch": 0.2299349240780911,
      "grad_norm": 0.4369511604309082,
      "learning_rate": 7.722342733188721e-06,
      "loss": 0.7592,
      "step": 106
    },
    {
      "epoch": 0.23210412147505424,
      "grad_norm": 0.4827646017074585,
      "learning_rate": 7.700650759219089e-06,
      "loss": 0.753,
      "step": 107
    },
    {
      "epoch": 0.23427331887201736,
      "grad_norm": 0.39963749051094055,
      "learning_rate": 7.678958785249458e-06,
      "loss": 0.7333,
      "step": 108
    },
    {
      "epoch": 0.23644251626898047,
      "grad_norm": 0.4362649917602539,
      "learning_rate": 7.657266811279828e-06,
      "loss": 0.7392,
      "step": 109
    },
    {
      "epoch": 0.2386117136659436,
      "grad_norm": 0.4753969609737396,
      "learning_rate": 7.635574837310196e-06,
      "loss": 0.7411,
      "step": 110
    },
    {
      "epoch": 0.24078091106290672,
      "grad_norm": 0.3958611488342285,
      "learning_rate": 7.613882863340565e-06,
      "loss": 0.7125,
      "step": 111
    },
    {
      "epoch": 0.24295010845986983,
      "grad_norm": 0.4026221036911011,
      "learning_rate": 7.592190889370933e-06,
      "loss": 0.7434,
      "step": 112
    },
    {
      "epoch": 0.24511930585683298,
      "grad_norm": 0.672747790813446,
      "learning_rate": 7.570498915401302e-06,
      "loss": 0.702,
      "step": 113
    },
    {
      "epoch": 0.2472885032537961,
      "grad_norm": 0.39787012338638306,
      "learning_rate": 7.548806941431671e-06,
      "loss": 0.7329,
      "step": 114
    },
    {
      "epoch": 0.24945770065075923,
      "grad_norm": 0.4380010664463043,
      "learning_rate": 7.527114967462039e-06,
      "loss": 0.7281,
      "step": 115
    },
    {
      "epoch": 0.25162689804772237,
      "grad_norm": 0.41002675890922546,
      "learning_rate": 7.505422993492408e-06,
      "loss": 0.7229,
      "step": 116
    },
    {
      "epoch": 0.25379609544468545,
      "grad_norm": 0.41404595971107483,
      "learning_rate": 7.483731019522778e-06,
      "loss": 0.7177,
      "step": 117
    },
    {
      "epoch": 0.2559652928416486,
      "grad_norm": 0.4458067715167999,
      "learning_rate": 7.4620390455531455e-06,
      "loss": 0.7226,
      "step": 118
    },
    {
      "epoch": 0.25813449023861174,
      "grad_norm": 0.3937883973121643,
      "learning_rate": 7.440347071583515e-06,
      "loss": 0.7109,
      "step": 119
    },
    {
      "epoch": 0.2603036876355748,
      "grad_norm": 0.42459914088249207,
      "learning_rate": 7.418655097613884e-06,
      "loss": 0.7292,
      "step": 120
    },
    {
      "epoch": 0.26247288503253796,
      "grad_norm": 0.47104501724243164,
      "learning_rate": 7.396963123644252e-06,
      "loss": 0.7018,
      "step": 121
    },
    {
      "epoch": 0.2646420824295011,
      "grad_norm": 0.413194477558136,
      "learning_rate": 7.375271149674621e-06,
      "loss": 0.6862,
      "step": 122
    },
    {
      "epoch": 0.2668112798264642,
      "grad_norm": 0.44947072863578796,
      "learning_rate": 7.35357917570499e-06,
      "loss": 0.7134,
      "step": 123
    },
    {
      "epoch": 0.26898047722342733,
      "grad_norm": 0.4308297038078308,
      "learning_rate": 7.331887201735358e-06,
      "loss": 0.7214,
      "step": 124
    },
    {
      "epoch": 0.27114967462039047,
      "grad_norm": 0.48938828706741333,
      "learning_rate": 7.310195227765728e-06,
      "loss": 0.7065,
      "step": 125
    },
    {
      "epoch": 0.27331887201735355,
      "grad_norm": 0.4659271538257599,
      "learning_rate": 7.288503253796096e-06,
      "loss": 0.717,
      "step": 126
    },
    {
      "epoch": 0.2754880694143167,
      "grad_norm": 0.43395182490348816,
      "learning_rate": 7.266811279826465e-06,
      "loss": 0.7137,
      "step": 127
    },
    {
      "epoch": 0.27765726681127983,
      "grad_norm": 0.4359440505504608,
      "learning_rate": 7.2451193058568335e-06,
      "loss": 0.6858,
      "step": 128
    },
    {
      "epoch": 0.279826464208243,
      "grad_norm": 0.6049463152885437,
      "learning_rate": 7.223427331887203e-06,
      "loss": 0.7262,
      "step": 129
    },
    {
      "epoch": 0.28199566160520606,
      "grad_norm": 0.9202204942703247,
      "learning_rate": 7.201735357917571e-06,
      "loss": 0.7012,
      "step": 130
    },
    {
      "epoch": 0.2841648590021692,
      "grad_norm": 0.4885154664516449,
      "learning_rate": 7.18004338394794e-06,
      "loss": 0.7016,
      "step": 131
    },
    {
      "epoch": 0.28633405639913234,
      "grad_norm": 0.5771850943565369,
      "learning_rate": 7.158351409978309e-06,
      "loss": 0.7069,
      "step": 132
    },
    {
      "epoch": 0.2885032537960954,
      "grad_norm": 0.4814640283584595,
      "learning_rate": 7.1366594360086775e-06,
      "loss": 0.7129,
      "step": 133
    },
    {
      "epoch": 0.29067245119305857,
      "grad_norm": 0.4861976206302643,
      "learning_rate": 7.114967462039046e-06,
      "loss": 0.6926,
      "step": 134
    },
    {
      "epoch": 0.2928416485900217,
      "grad_norm": 0.43924835324287415,
      "learning_rate": 7.093275488069415e-06,
      "loss": 0.6729,
      "step": 135
    },
    {
      "epoch": 0.2950108459869848,
      "grad_norm": 0.44776543974876404,
      "learning_rate": 7.071583514099783e-06,
      "loss": 0.69,
      "step": 136
    },
    {
      "epoch": 0.29718004338394793,
      "grad_norm": 0.48698529601097107,
      "learning_rate": 7.049891540130153e-06,
      "loss": 0.6707,
      "step": 137
    },
    {
      "epoch": 0.2993492407809111,
      "grad_norm": 0.49502941966056824,
      "learning_rate": 7.028199566160521e-06,
      "loss": 0.6705,
      "step": 138
    },
    {
      "epoch": 0.30151843817787416,
      "grad_norm": 0.43487516045570374,
      "learning_rate": 7.00650759219089e-06,
      "loss": 0.6468,
      "step": 139
    },
    {
      "epoch": 0.3036876355748373,
      "grad_norm": 0.45082563161849976,
      "learning_rate": 6.984815618221259e-06,
      "loss": 0.6882,
      "step": 140
    },
    {
      "epoch": 0.30585683297180044,
      "grad_norm": 0.5704577565193176,
      "learning_rate": 6.963123644251627e-06,
      "loss": 0.6546,
      "step": 141
    },
    {
      "epoch": 0.3080260303687636,
      "grad_norm": 0.5590646266937256,
      "learning_rate": 6.941431670281996e-06,
      "loss": 0.691,
      "step": 142
    },
    {
      "epoch": 0.31019522776572667,
      "grad_norm": 0.4791410565376282,
      "learning_rate": 6.9197396963123654e-06,
      "loss": 0.6327,
      "step": 143
    },
    {
      "epoch": 0.3123644251626898,
      "grad_norm": 0.435273140668869,
      "learning_rate": 6.898047722342733e-06,
      "loss": 0.675,
      "step": 144
    },
    {
      "epoch": 0.31453362255965295,
      "grad_norm": 0.5272928476333618,
      "learning_rate": 6.876355748373103e-06,
      "loss": 0.6683,
      "step": 145
    },
    {
      "epoch": 0.31670281995661603,
      "grad_norm": 0.5004708766937256,
      "learning_rate": 6.854663774403471e-06,
      "loss": 0.6518,
      "step": 146
    },
    {
      "epoch": 0.3188720173535792,
      "grad_norm": 0.45495614409446716,
      "learning_rate": 6.83297180043384e-06,
      "loss": 0.6576,
      "step": 147
    },
    {
      "epoch": 0.3210412147505423,
      "grad_norm": 0.48196735978126526,
      "learning_rate": 6.8112798264642086e-06,
      "loss": 0.6389,
      "step": 148
    },
    {
      "epoch": 0.3232104121475054,
      "grad_norm": 0.4797559082508087,
      "learning_rate": 6.789587852494578e-06,
      "loss": 0.6677,
      "step": 149
    },
    {
      "epoch": 0.32537960954446854,
      "grad_norm": 0.4930654764175415,
      "learning_rate": 6.767895878524946e-06,
      "loss": 0.6437,
      "step": 150
    },
    {
      "epoch": 0.3275488069414317,
      "grad_norm": 0.45326152443885803,
      "learning_rate": 6.746203904555315e-06,
      "loss": 0.6616,
      "step": 151
    },
    {
      "epoch": 0.3297180043383948,
      "grad_norm": 0.48853611946105957,
      "learning_rate": 6.724511930585684e-06,
      "loss": 0.6473,
      "step": 152
    },
    {
      "epoch": 0.3318872017353579,
      "grad_norm": 0.4732632040977478,
      "learning_rate": 6.7028199566160526e-06,
      "loss": 0.6396,
      "step": 153
    },
    {
      "epoch": 0.33405639913232105,
      "grad_norm": 0.4409032464027405,
      "learning_rate": 6.681127982646421e-06,
      "loss": 0.6507,
      "step": 154
    },
    {
      "epoch": 0.3362255965292842,
      "grad_norm": 0.4853736162185669,
      "learning_rate": 6.659436008676791e-06,
      "loss": 0.6495,
      "step": 155
    },
    {
      "epoch": 0.3383947939262473,
      "grad_norm": 0.5413075685501099,
      "learning_rate": 6.6377440347071584e-06,
      "loss": 0.6719,
      "step": 156
    },
    {
      "epoch": 0.3405639913232104,
      "grad_norm": 0.4533112943172455,
      "learning_rate": 6.616052060737528e-06,
      "loss": 0.6358,
      "step": 157
    },
    {
      "epoch": 0.34273318872017355,
      "grad_norm": 0.5843056440353394,
      "learning_rate": 6.594360086767897e-06,
      "loss": 0.6724,
      "step": 158
    },
    {
      "epoch": 0.34490238611713664,
      "grad_norm": 0.48672884702682495,
      "learning_rate": 6.572668112798265e-06,
      "loss": 0.6449,
      "step": 159
    },
    {
      "epoch": 0.3470715835140998,
      "grad_norm": 0.6113792657852173,
      "learning_rate": 6.550976138828634e-06,
      "loss": 0.6297,
      "step": 160
    },
    {
      "epoch": 0.3492407809110629,
      "grad_norm": 0.5860591530799866,
      "learning_rate": 6.529284164859002e-06,
      "loss": 0.6617,
      "step": 161
    },
    {
      "epoch": 0.351409978308026,
      "grad_norm": 0.5638704299926758,
      "learning_rate": 6.507592190889371e-06,
      "loss": 0.6225,
      "step": 162
    },
    {
      "epoch": 0.35357917570498915,
      "grad_norm": 0.5348539352416992,
      "learning_rate": 6.4859002169197405e-06,
      "loss": 0.6351,
      "step": 163
    },
    {
      "epoch": 0.3557483731019523,
      "grad_norm": 0.4822864830493927,
      "learning_rate": 6.464208242950108e-06,
      "loss": 0.6315,
      "step": 164
    },
    {
      "epoch": 0.3579175704989154,
      "grad_norm": 0.5151847004890442,
      "learning_rate": 6.442516268980478e-06,
      "loss": 0.6311,
      "step": 165
    },
    {
      "epoch": 0.3600867678958785,
      "grad_norm": 0.4632371962070465,
      "learning_rate": 6.420824295010846e-06,
      "loss": 0.6515,
      "step": 166
    },
    {
      "epoch": 0.36225596529284165,
      "grad_norm": 0.4700559675693512,
      "learning_rate": 6.399132321041215e-06,
      "loss": 0.6218,
      "step": 167
    },
    {
      "epoch": 0.3644251626898048,
      "grad_norm": 0.5487787127494812,
      "learning_rate": 6.377440347071584e-06,
      "loss": 0.6485,
      "step": 168
    },
    {
      "epoch": 0.3665943600867679,
      "grad_norm": 0.5952288508415222,
      "learning_rate": 6.355748373101953e-06,
      "loss": 0.6263,
      "step": 169
    },
    {
      "epoch": 0.368763557483731,
      "grad_norm": 0.4692787230014801,
      "learning_rate": 6.334056399132321e-06,
      "loss": 0.637,
      "step": 170
    },
    {
      "epoch": 0.37093275488069416,
      "grad_norm": 0.48934659361839294,
      "learning_rate": 6.31236442516269e-06,
      "loss": 0.6177,
      "step": 171
    },
    {
      "epoch": 0.37310195227765725,
      "grad_norm": 0.6284322142601013,
      "learning_rate": 6.29067245119306e-06,
      "loss": 0.6533,
      "step": 172
    },
    {
      "epoch": 0.3752711496746204,
      "grad_norm": 0.583229660987854,
      "learning_rate": 6.268980477223428e-06,
      "loss": 0.6468,
      "step": 173
    },
    {
      "epoch": 0.3774403470715835,
      "grad_norm": 0.5373253226280212,
      "learning_rate": 6.247288503253796e-06,
      "loss": 0.6219,
      "step": 174
    },
    {
      "epoch": 0.3796095444685466,
      "grad_norm": 0.5065279603004456,
      "learning_rate": 6.225596529284166e-06,
      "loss": 0.609,
      "step": 175
    },
    {
      "epoch": 0.38177874186550975,
      "grad_norm": 0.4818476438522339,
      "learning_rate": 6.2039045553145335e-06,
      "loss": 0.6203,
      "step": 176
    },
    {
      "epoch": 0.3839479392624729,
      "grad_norm": 0.47112759947776794,
      "learning_rate": 6.182212581344903e-06,
      "loss": 0.6315,
      "step": 177
    },
    {
      "epoch": 0.38611713665943603,
      "grad_norm": 0.5504185557365417,
      "learning_rate": 6.1605206073752725e-06,
      "loss": 0.6318,
      "step": 178
    },
    {
      "epoch": 0.3882863340563991,
      "grad_norm": 0.5567037463188171,
      "learning_rate": 6.13882863340564e-06,
      "loss": 0.6104,
      "step": 179
    },
    {
      "epoch": 0.39045553145336226,
      "grad_norm": 0.4654327630996704,
      "learning_rate": 6.11713665943601e-06,
      "loss": 0.619,
      "step": 180
    },
    {
      "epoch": 0.3926247288503254,
      "grad_norm": 0.47351258993148804,
      "learning_rate": 6.095444685466378e-06,
      "loss": 0.6206,
      "step": 181
    },
    {
      "epoch": 0.3947939262472885,
      "grad_norm": 0.47146716713905334,
      "learning_rate": 6.073752711496746e-06,
      "loss": 0.6111,
      "step": 182
    },
    {
      "epoch": 0.3969631236442516,
      "grad_norm": 0.49498575925827026,
      "learning_rate": 6.052060737527116e-06,
      "loss": 0.6179,
      "step": 183
    },
    {
      "epoch": 0.39913232104121477,
      "grad_norm": 0.5393364429473877,
      "learning_rate": 6.030368763557483e-06,
      "loss": 0.6331,
      "step": 184
    },
    {
      "epoch": 0.40130151843817785,
      "grad_norm": 0.5126506686210632,
      "learning_rate": 6.008676789587853e-06,
      "loss": 0.6143,
      "step": 185
    },
    {
      "epoch": 0.403470715835141,
      "grad_norm": 0.6334784626960754,
      "learning_rate": 5.986984815618222e-06,
      "loss": 0.6322,
      "step": 186
    },
    {
      "epoch": 0.40563991323210413,
      "grad_norm": 0.5737828016281128,
      "learning_rate": 5.96529284164859e-06,
      "loss": 0.6206,
      "step": 187
    },
    {
      "epoch": 0.4078091106290672,
      "grad_norm": 0.5050855875015259,
      "learning_rate": 5.943600867678959e-06,
      "loss": 0.6262,
      "step": 188
    },
    {
      "epoch": 0.40997830802603036,
      "grad_norm": 0.4655757546424866,
      "learning_rate": 5.921908893709328e-06,
      "loss": 0.6062,
      "step": 189
    },
    {
      "epoch": 0.4121475054229935,
      "grad_norm": 0.7657625675201416,
      "learning_rate": 5.900216919739696e-06,
      "loss": 0.6278,
      "step": 190
    },
    {
      "epoch": 0.41431670281995664,
      "grad_norm": 0.4740203619003296,
      "learning_rate": 5.8785249457700655e-06,
      "loss": 0.6125,
      "step": 191
    },
    {
      "epoch": 0.4164859002169197,
      "grad_norm": 0.47104138135910034,
      "learning_rate": 5.856832971800435e-06,
      "loss": 0.6181,
      "step": 192
    },
    {
      "epoch": 0.41865509761388287,
      "grad_norm": 0.6184363961219788,
      "learning_rate": 5.835140997830803e-06,
      "loss": 0.6227,
      "step": 193
    },
    {
      "epoch": 0.420824295010846,
      "grad_norm": 0.6750501990318298,
      "learning_rate": 5.813449023861172e-06,
      "loss": 0.6407,
      "step": 194
    },
    {
      "epoch": 0.4229934924078091,
      "grad_norm": 0.6018952131271362,
      "learning_rate": 5.791757049891541e-06,
      "loss": 0.5914,
      "step": 195
    },
    {
      "epoch": 0.42516268980477223,
      "grad_norm": 0.5619893670082092,
      "learning_rate": 5.770065075921909e-06,
      "loss": 0.6222,
      "step": 196
    },
    {
      "epoch": 0.42733188720173537,
      "grad_norm": 0.4996378719806671,
      "learning_rate": 5.748373101952278e-06,
      "loss": 0.6076,
      "step": 197
    },
    {
      "epoch": 0.42950108459869846,
      "grad_norm": 0.47961318492889404,
      "learning_rate": 5.7266811279826476e-06,
      "loss": 0.6102,
      "step": 198
    },
    {
      "epoch": 0.4316702819956616,
      "grad_norm": 0.6786702871322632,
      "learning_rate": 5.704989154013015e-06,
      "loss": 0.6353,
      "step": 199
    },
    {
      "epoch": 0.43383947939262474,
      "grad_norm": 0.6170738935470581,
      "learning_rate": 5.683297180043385e-06,
      "loss": 0.6038,
      "step": 200
    },
    {
      "epoch": 0.4360086767895879,
      "grad_norm": 0.5487593412399292,
      "learning_rate": 5.6616052060737535e-06,
      "loss": 0.5929,
      "step": 201
    },
    {
      "epoch": 0.43817787418655096,
      "grad_norm": 0.593626856803894,
      "learning_rate": 5.639913232104122e-06,
      "loss": 0.5977,
      "step": 202
    },
    {
      "epoch": 0.4403470715835141,
      "grad_norm": 0.7302510738372803,
      "learning_rate": 5.618221258134491e-06,
      "loss": 0.5996,
      "step": 203
    },
    {
      "epoch": 0.44251626898047725,
      "grad_norm": 0.5379276275634766,
      "learning_rate": 5.59652928416486e-06,
      "loss": 0.6119,
      "step": 204
    },
    {
      "epoch": 0.44468546637744033,
      "grad_norm": 0.6861233711242676,
      "learning_rate": 5.574837310195228e-06,
      "loss": 0.6119,
      "step": 205
    },
    {
      "epoch": 0.44685466377440347,
      "grad_norm": 0.5536229014396667,
      "learning_rate": 5.5531453362255974e-06,
      "loss": 0.5926,
      "step": 206
    },
    {
      "epoch": 0.4490238611713666,
      "grad_norm": 0.5160278677940369,
      "learning_rate": 5.531453362255966e-06,
      "loss": 0.6013,
      "step": 207
    },
    {
      "epoch": 0.4511930585683297,
      "grad_norm": 0.5844294428825378,
      "learning_rate": 5.509761388286335e-06,
      "loss": 0.6044,
      "step": 208
    },
    {
      "epoch": 0.45336225596529284,
      "grad_norm": 0.6622424721717834,
      "learning_rate": 5.488069414316703e-06,
      "loss": 0.6005,
      "step": 209
    },
    {
      "epoch": 0.455531453362256,
      "grad_norm": 0.5025097727775574,
      "learning_rate": 5.466377440347071e-06,
      "loss": 0.5802,
      "step": 210
    },
    {
      "epoch": 0.45770065075921906,
      "grad_norm": 0.5540115237236023,
      "learning_rate": 5.444685466377441e-06,
      "loss": 0.5932,
      "step": 211
    },
    {
      "epoch": 0.4598698481561822,
      "grad_norm": 0.5804458856582642,
      "learning_rate": 5.42299349240781e-06,
      "loss": 0.5987,
      "step": 212
    },
    {
      "epoch": 0.46203904555314534,
      "grad_norm": 0.5122389793395996,
      "learning_rate": 5.401301518438178e-06,
      "loss": 0.5916,
      "step": 213
    },
    {
      "epoch": 0.4642082429501085,
      "grad_norm": 0.5717864632606506,
      "learning_rate": 5.379609544468547e-06,
      "loss": 0.5948,
      "step": 214
    },
    {
      "epoch": 0.46637744034707157,
      "grad_norm": 0.6112474799156189,
      "learning_rate": 5.357917570498916e-06,
      "loss": 0.6016,
      "step": 215
    },
    {
      "epoch": 0.4685466377440347,
      "grad_norm": 0.58445805311203,
      "learning_rate": 5.3362255965292846e-06,
      "loss": 0.5943,
      "step": 216
    },
    {
      "epoch": 0.47071583514099785,
      "grad_norm": 0.5126850605010986,
      "learning_rate": 5.314533622559653e-06,
      "loss": 0.6058,
      "step": 217
    },
    {
      "epoch": 0.47288503253796094,
      "grad_norm": 0.498695969581604,
      "learning_rate": 5.292841648590023e-06,
      "loss": 0.5867,
      "step": 218
    },
    {
      "epoch": 0.4750542299349241,
      "grad_norm": 0.5060663223266602,
      "learning_rate": 5.2711496746203904e-06,
      "loss": 0.6161,
      "step": 219
    },
    {
      "epoch": 0.4772234273318872,
      "grad_norm": 0.49926185607910156,
      "learning_rate": 5.24945770065076e-06,
      "loss": 0.594,
      "step": 220
    },
    {
      "epoch": 0.4793926247288503,
      "grad_norm": 0.7231149673461914,
      "learning_rate": 5.2277657266811285e-06,
      "loss": 0.6506,
      "step": 221
    },
    {
      "epoch": 0.48156182212581344,
      "grad_norm": 0.501641571521759,
      "learning_rate": 5.206073752711497e-06,
      "loss": 0.6031,
      "step": 222
    },
    {
      "epoch": 0.4837310195227766,
      "grad_norm": 0.5351401567459106,
      "learning_rate": 5.184381778741866e-06,
      "loss": 0.5957,
      "step": 223
    },
    {
      "epoch": 0.48590021691973967,
      "grad_norm": 0.4861268997192383,
      "learning_rate": 5.162689804772235e-06,
      "loss": 0.6109,
      "step": 224
    },
    {
      "epoch": 0.4880694143167028,
      "grad_norm": 0.4920121133327484,
      "learning_rate": 5.140997830802603e-06,
      "loss": 0.5931,
      "step": 225
    },
    {
      "epoch": 0.49023861171366595,
      "grad_norm": 0.5210174322128296,
      "learning_rate": 5.1193058568329725e-06,
      "loss": 0.6122,
      "step": 226
    },
    {
      "epoch": 0.4924078091106291,
      "grad_norm": 0.7002459764480591,
      "learning_rate": 5.097613882863341e-06,
      "loss": 0.603,
      "step": 227
    },
    {
      "epoch": 0.4945770065075922,
      "grad_norm": 0.4967700242996216,
      "learning_rate": 5.07592190889371e-06,
      "loss": 0.5675,
      "step": 228
    },
    {
      "epoch": 0.4967462039045553,
      "grad_norm": 0.49998047947883606,
      "learning_rate": 5.054229934924078e-06,
      "loss": 0.5885,
      "step": 229
    },
    {
      "epoch": 0.49891540130151846,
      "grad_norm": 0.46569979190826416,
      "learning_rate": 5.032537960954448e-06,
      "loss": 0.588,
      "step": 230
    },
    {
      "epoch": 0.5010845986984815,
      "grad_norm": 0.518514096736908,
      "learning_rate": 5.010845986984816e-06,
      "loss": 0.5795,
      "step": 231
    },
    {
      "epoch": 0.5032537960954447,
      "grad_norm": 0.5044274926185608,
      "learning_rate": 4.989154013015185e-06,
      "loss": 0.5777,
      "step": 232
    },
    {
      "epoch": 0.5054229934924078,
      "grad_norm": 0.5333325862884521,
      "learning_rate": 4.967462039045554e-06,
      "loss": 0.5902,
      "step": 233
    },
    {
      "epoch": 0.5075921908893709,
      "grad_norm": 0.666131854057312,
      "learning_rate": 4.945770065075922e-06,
      "loss": 0.5735,
      "step": 234
    },
    {
      "epoch": 0.5097613882863341,
      "grad_norm": 0.5070677995681763,
      "learning_rate": 4.924078091106291e-06,
      "loss": 0.5677,
      "step": 235
    },
    {
      "epoch": 0.5119305856832972,
      "grad_norm": 0.5434484481811523,
      "learning_rate": 4.90238611713666e-06,
      "loss": 0.5984,
      "step": 236
    },
    {
      "epoch": 0.5140997830802603,
      "grad_norm": 0.5604612231254578,
      "learning_rate": 4.880694143167028e-06,
      "loss": 0.5915,
      "step": 237
    },
    {
      "epoch": 0.5162689804772235,
      "grad_norm": 0.5144227147102356,
      "learning_rate": 4.859002169197397e-06,
      "loss": 0.5765,
      "step": 238
    },
    {
      "epoch": 0.5184381778741866,
      "grad_norm": 0.4857323169708252,
      "learning_rate": 4.837310195227766e-06,
      "loss": 0.5722,
      "step": 239
    },
    {
      "epoch": 0.5206073752711496,
      "grad_norm": 0.4755032956600189,
      "learning_rate": 4.815618221258135e-06,
      "loss": 0.5749,
      "step": 240
    },
    {
      "epoch": 0.5227765726681128,
      "grad_norm": 0.587493896484375,
      "learning_rate": 4.793926247288504e-06,
      "loss": 0.5981,
      "step": 241
    },
    {
      "epoch": 0.5249457700650759,
      "grad_norm": 0.5679769515991211,
      "learning_rate": 4.772234273318872e-06,
      "loss": 0.5884,
      "step": 242
    },
    {
      "epoch": 0.527114967462039,
      "grad_norm": 0.498691588640213,
      "learning_rate": 4.750542299349241e-06,
      "loss": 0.5809,
      "step": 243
    },
    {
      "epoch": 0.5292841648590022,
      "grad_norm": 0.675168514251709,
      "learning_rate": 4.7288503253796095e-06,
      "loss": 0.5874,
      "step": 244
    },
    {
      "epoch": 0.5314533622559653,
      "grad_norm": 0.5249897241592407,
      "learning_rate": 4.707158351409979e-06,
      "loss": 0.5878,
      "step": 245
    },
    {
      "epoch": 0.5336225596529284,
      "grad_norm": 0.48491179943084717,
      "learning_rate": 4.685466377440348e-06,
      "loss": 0.5743,
      "step": 246
    },
    {
      "epoch": 0.5357917570498916,
      "grad_norm": 0.5798050761222839,
      "learning_rate": 4.663774403470716e-06,
      "loss": 0.5682,
      "step": 247
    },
    {
      "epoch": 0.5379609544468547,
      "grad_norm": 0.5193563103675842,
      "learning_rate": 4.642082429501085e-06,
      "loss": 0.5743,
      "step": 248
    },
    {
      "epoch": 0.5401301518438177,
      "grad_norm": 0.5050124526023865,
      "learning_rate": 4.6203904555314535e-06,
      "loss": 0.5823,
      "step": 249
    },
    {
      "epoch": 0.5422993492407809,
      "grad_norm": 0.5084236264228821,
      "learning_rate": 4.598698481561822e-06,
      "loss": 0.5865,
      "step": 250
    },
    {
      "epoch": 0.544468546637744,
      "grad_norm": 0.5621696710586548,
      "learning_rate": 4.577006507592191e-06,
      "loss": 0.6023,
      "step": 251
    },
    {
      "epoch": 0.5466377440347071,
      "grad_norm": 0.4698992669582367,
      "learning_rate": 4.55531453362256e-06,
      "loss": 0.5835,
      "step": 252
    },
    {
      "epoch": 0.5488069414316703,
      "grad_norm": 0.5616809129714966,
      "learning_rate": 4.533622559652929e-06,
      "loss": 0.559,
      "step": 253
    },
    {
      "epoch": 0.5509761388286334,
      "grad_norm": 0.5029488801956177,
      "learning_rate": 4.5119305856832975e-06,
      "loss": 0.5693,
      "step": 254
    },
    {
      "epoch": 0.5531453362255966,
      "grad_norm": 0.47164446115493774,
      "learning_rate": 4.490238611713666e-06,
      "loss": 0.564,
      "step": 255
    },
    {
      "epoch": 0.5553145336225597,
      "grad_norm": 0.49710217118263245,
      "learning_rate": 4.468546637744035e-06,
      "loss": 0.5826,
      "step": 256
    },
    {
      "epoch": 0.5574837310195228,
      "grad_norm": 0.48587456345558167,
      "learning_rate": 4.446854663774403e-06,
      "loss": 0.5816,
      "step": 257
    },
    {
      "epoch": 0.559652928416486,
      "grad_norm": 0.5498842597007751,
      "learning_rate": 4.425162689804773e-06,
      "loss": 0.5736,
      "step": 258
    },
    {
      "epoch": 0.561822125813449,
      "grad_norm": 0.4747694432735443,
      "learning_rate": 4.4034707158351415e-06,
      "loss": 0.574,
      "step": 259
    },
    {
      "epoch": 0.5639913232104121,
      "grad_norm": 0.9344505667686462,
      "learning_rate": 4.38177874186551e-06,
      "loss": 0.5957,
      "step": 260
    },
    {
      "epoch": 0.5661605206073753,
      "grad_norm": 0.5000290274620056,
      "learning_rate": 4.360086767895879e-06,
      "loss": 0.5774,
      "step": 261
    },
    {
      "epoch": 0.5683297180043384,
      "grad_norm": 0.5080724358558655,
      "learning_rate": 4.338394793926247e-06,
      "loss": 0.5688,
      "step": 262
    },
    {
      "epoch": 0.5704989154013015,
      "grad_norm": 0.5054919123649597,
      "learning_rate": 4.316702819956616e-06,
      "loss": 0.5834,
      "step": 263
    },
    {
      "epoch": 0.5726681127982647,
      "grad_norm": 0.5046693682670593,
      "learning_rate": 4.295010845986985e-06,
      "loss": 0.5687,
      "step": 264
    },
    {
      "epoch": 0.5748373101952278,
      "grad_norm": 0.5170451998710632,
      "learning_rate": 4.273318872017354e-06,
      "loss": 0.5497,
      "step": 265
    },
    {
      "epoch": 0.5770065075921909,
      "grad_norm": 0.5297961235046387,
      "learning_rate": 4.251626898047723e-06,
      "loss": 0.6013,
      "step": 266
    },
    {
      "epoch": 0.579175704989154,
      "grad_norm": 0.4973108172416687,
      "learning_rate": 4.229934924078091e-06,
      "loss": 0.5544,
      "step": 267
    },
    {
      "epoch": 0.5813449023861171,
      "grad_norm": 0.6242494583129883,
      "learning_rate": 4.208242950108461e-06,
      "loss": 0.5917,
      "step": 268
    },
    {
      "epoch": 0.5835140997830802,
      "grad_norm": 0.5227379202842712,
      "learning_rate": 4.186550976138829e-06,
      "loss": 0.5709,
      "step": 269
    },
    {
      "epoch": 0.5856832971800434,
      "grad_norm": 0.527258574962616,
      "learning_rate": 4.164859002169197e-06,
      "loss": 0.5746,
      "step": 270
    },
    {
      "epoch": 0.5878524945770065,
      "grad_norm": 0.5078297257423401,
      "learning_rate": 4.143167028199567e-06,
      "loss": 0.5569,
      "step": 271
    },
    {
      "epoch": 0.5900216919739696,
      "grad_norm": 0.7073683738708496,
      "learning_rate": 4.121475054229935e-06,
      "loss": 0.5806,
      "step": 272
    },
    {
      "epoch": 0.5921908893709328,
      "grad_norm": 0.6542203426361084,
      "learning_rate": 4.099783080260304e-06,
      "loss": 0.5963,
      "step": 273
    },
    {
      "epoch": 0.5943600867678959,
      "grad_norm": 0.4986463189125061,
      "learning_rate": 4.078091106290673e-06,
      "loss": 0.5786,
      "step": 274
    },
    {
      "epoch": 0.596529284164859,
      "grad_norm": 0.4862741231918335,
      "learning_rate": 4.056399132321042e-06,
      "loss": 0.5508,
      "step": 275
    },
    {
      "epoch": 0.5986984815618221,
      "grad_norm": 0.5407434105873108,
      "learning_rate": 4.03470715835141e-06,
      "loss": 0.5586,
      "step": 276
    },
    {
      "epoch": 0.6008676789587852,
      "grad_norm": 0.513647198677063,
      "learning_rate": 4.0130151843817785e-06,
      "loss": 0.5653,
      "step": 277
    },
    {
      "epoch": 0.6030368763557483,
      "grad_norm": 0.4941583573818207,
      "learning_rate": 3.991323210412148e-06,
      "loss": 0.5724,
      "step": 278
    },
    {
      "epoch": 0.6052060737527115,
      "grad_norm": 0.5022282600402832,
      "learning_rate": 3.9696312364425166e-06,
      "loss": 0.5597,
      "step": 279
    },
    {
      "epoch": 0.6073752711496746,
      "grad_norm": 0.5344776511192322,
      "learning_rate": 3.947939262472885e-06,
      "loss": 0.576,
      "step": 280
    },
    {
      "epoch": 0.6095444685466378,
      "grad_norm": 0.8711801171302795,
      "learning_rate": 3.926247288503255e-06,
      "loss": 0.5653,
      "step": 281
    },
    {
      "epoch": 0.6117136659436009,
      "grad_norm": 0.6406508684158325,
      "learning_rate": 3.904555314533623e-06,
      "loss": 0.5612,
      "step": 282
    },
    {
      "epoch": 0.613882863340564,
      "grad_norm": 0.5289674997329712,
      "learning_rate": 3.882863340563991e-06,
      "loss": 0.5553,
      "step": 283
    },
    {
      "epoch": 0.6160520607375272,
      "grad_norm": 0.49930331110954285,
      "learning_rate": 3.8611713665943606e-06,
      "loss": 0.5661,
      "step": 284
    },
    {
      "epoch": 0.6182212581344902,
      "grad_norm": 0.49601882696151733,
      "learning_rate": 3.839479392624729e-06,
      "loss": 0.5613,
      "step": 285
    },
    {
      "epoch": 0.6203904555314533,
      "grad_norm": 0.5551280975341797,
      "learning_rate": 3.817787418655098e-06,
      "loss": 0.5692,
      "step": 286
    },
    {
      "epoch": 0.6225596529284165,
      "grad_norm": 0.5268258452415466,
      "learning_rate": 3.7960954446854664e-06,
      "loss": 0.5653,
      "step": 287
    },
    {
      "epoch": 0.6247288503253796,
      "grad_norm": 0.5202245712280273,
      "learning_rate": 3.7744034707158355e-06,
      "loss": 0.5623,
      "step": 288
    },
    {
      "epoch": 0.6268980477223427,
      "grad_norm": 0.7391339540481567,
      "learning_rate": 3.752711496746204e-06,
      "loss": 0.5485,
      "step": 289
    },
    {
      "epoch": 0.6290672451193059,
      "grad_norm": 0.7942306995391846,
      "learning_rate": 3.7310195227765728e-06,
      "loss": 0.5693,
      "step": 290
    },
    {
      "epoch": 0.631236442516269,
      "grad_norm": 0.5168442130088806,
      "learning_rate": 3.709327548806942e-06,
      "loss": 0.563,
      "step": 291
    },
    {
      "epoch": 0.6334056399132321,
      "grad_norm": 0.5481964349746704,
      "learning_rate": 3.6876355748373104e-06,
      "loss": 0.5694,
      "step": 292
    },
    {
      "epoch": 0.6355748373101953,
      "grad_norm": 0.7699911594390869,
      "learning_rate": 3.665943600867679e-06,
      "loss": 0.602,
      "step": 293
    },
    {
      "epoch": 0.6377440347071583,
      "grad_norm": 0.5012184381484985,
      "learning_rate": 3.644251626898048e-06,
      "loss": 0.5566,
      "step": 294
    },
    {
      "epoch": 0.6399132321041214,
      "grad_norm": 0.5150504112243652,
      "learning_rate": 3.6225596529284167e-06,
      "loss": 0.5575,
      "step": 295
    },
    {
      "epoch": 0.6420824295010846,
      "grad_norm": 0.4825197160243988,
      "learning_rate": 3.6008676789587854e-06,
      "loss": 0.578,
      "step": 296
    },
    {
      "epoch": 0.6442516268980477,
      "grad_norm": 0.5708716511726379,
      "learning_rate": 3.5791757049891544e-06,
      "loss": 0.5748,
      "step": 297
    },
    {
      "epoch": 0.6464208242950108,
      "grad_norm": 0.49083611369132996,
      "learning_rate": 3.557483731019523e-06,
      "loss": 0.5656,
      "step": 298
    },
    {
      "epoch": 0.648590021691974,
      "grad_norm": 0.5466710329055786,
      "learning_rate": 3.5357917570498917e-06,
      "loss": 0.5808,
      "step": 299
    },
    {
      "epoch": 0.6507592190889371,
      "grad_norm": 0.6208142042160034,
      "learning_rate": 3.5140997830802603e-06,
      "loss": 0.5665,
      "step": 300
    },
    {
      "epoch": 0.6529284164859002,
      "grad_norm": 0.4832100570201874,
      "learning_rate": 3.4924078091106293e-06,
      "loss": 0.572,
      "step": 301
    },
    {
      "epoch": 0.6550976138828634,
      "grad_norm": 0.5140089392662048,
      "learning_rate": 3.470715835140998e-06,
      "loss": 0.5565,
      "step": 302
    },
    {
      "epoch": 0.6572668112798264,
      "grad_norm": 0.5033791661262512,
      "learning_rate": 3.4490238611713666e-06,
      "loss": 0.5522,
      "step": 303
    },
    {
      "epoch": 0.6594360086767896,
      "grad_norm": 0.4576267600059509,
      "learning_rate": 3.4273318872017357e-06,
      "loss": 0.5511,
      "step": 304
    },
    {
      "epoch": 0.6616052060737527,
      "grad_norm": 0.5113222002983093,
      "learning_rate": 3.4056399132321043e-06,
      "loss": 0.5736,
      "step": 305
    },
    {
      "epoch": 0.6637744034707158,
      "grad_norm": 0.528408408164978,
      "learning_rate": 3.383947939262473e-06,
      "loss": 0.5679,
      "step": 306
    },
    {
      "epoch": 0.665943600867679,
      "grad_norm": 0.4761233925819397,
      "learning_rate": 3.362255965292842e-06,
      "loss": 0.535,
      "step": 307
    },
    {
      "epoch": 0.6681127982646421,
      "grad_norm": 0.4666268527507782,
      "learning_rate": 3.3405639913232106e-06,
      "loss": 0.5731,
      "step": 308
    },
    {
      "epoch": 0.6702819956616052,
      "grad_norm": 0.5518378615379333,
      "learning_rate": 3.3188720173535792e-06,
      "loss": 0.5679,
      "step": 309
    },
    {
      "epoch": 0.6724511930585684,
      "grad_norm": 0.4833706021308899,
      "learning_rate": 3.2971800433839487e-06,
      "loss": 0.5615,
      "step": 310
    },
    {
      "epoch": 0.6746203904555315,
      "grad_norm": 0.5420160889625549,
      "learning_rate": 3.275488069414317e-06,
      "loss": 0.5529,
      "step": 311
    },
    {
      "epoch": 0.6767895878524945,
      "grad_norm": 0.6026318669319153,
      "learning_rate": 3.2537960954446855e-06,
      "loss": 0.5478,
      "step": 312
    },
    {
      "epoch": 0.6789587852494577,
      "grad_norm": 0.5486297607421875,
      "learning_rate": 3.232104121475054e-06,
      "loss": 0.5619,
      "step": 313
    },
    {
      "epoch": 0.6811279826464208,
      "grad_norm": 0.49950018525123596,
      "learning_rate": 3.210412147505423e-06,
      "loss": 0.5539,
      "step": 314
    },
    {
      "epoch": 0.6832971800433839,
      "grad_norm": 0.4778575301170349,
      "learning_rate": 3.188720173535792e-06,
      "loss": 0.546,
      "step": 315
    },
    {
      "epoch": 0.6854663774403471,
      "grad_norm": 0.4857572317123413,
      "learning_rate": 3.1670281995661605e-06,
      "loss": 0.5502,
      "step": 316
    },
    {
      "epoch": 0.6876355748373102,
      "grad_norm": 0.5437939167022705,
      "learning_rate": 3.14533622559653e-06,
      "loss": 0.5448,
      "step": 317
    },
    {
      "epoch": 0.6898047722342733,
      "grad_norm": 0.6015022397041321,
      "learning_rate": 3.123644251626898e-06,
      "loss": 0.5562,
      "step": 318
    },
    {
      "epoch": 0.6919739696312365,
      "grad_norm": 0.5425083041191101,
      "learning_rate": 3.1019522776572668e-06,
      "loss": 0.5496,
      "step": 319
    },
    {
      "epoch": 0.6941431670281996,
      "grad_norm": 0.4970496594905853,
      "learning_rate": 3.0802603036876362e-06,
      "loss": 0.5577,
      "step": 320
    },
    {
      "epoch": 0.6963123644251626,
      "grad_norm": 0.5393553972244263,
      "learning_rate": 3.058568329718005e-06,
      "loss": 0.5665,
      "step": 321
    },
    {
      "epoch": 0.6984815618221258,
      "grad_norm": 0.5788228511810303,
      "learning_rate": 3.036876355748373e-06,
      "loss": 0.5525,
      "step": 322
    },
    {
      "epoch": 0.7006507592190889,
      "grad_norm": 0.582302987575531,
      "learning_rate": 3.0151843817787417e-06,
      "loss": 0.5337,
      "step": 323
    },
    {
      "epoch": 0.702819956616052,
      "grad_norm": 0.5455271601676941,
      "learning_rate": 2.993492407809111e-06,
      "loss": 0.567,
      "step": 324
    },
    {
      "epoch": 0.7049891540130152,
      "grad_norm": 0.5106908679008484,
      "learning_rate": 2.9718004338394794e-06,
      "loss": 0.5572,
      "step": 325
    },
    {
      "epoch": 0.7071583514099783,
      "grad_norm": 0.5830085277557373,
      "learning_rate": 2.950108459869848e-06,
      "loss": 0.5799,
      "step": 326
    },
    {
      "epoch": 0.7093275488069414,
      "grad_norm": 0.543565571308136,
      "learning_rate": 2.9284164859002175e-06,
      "loss": 0.5486,
      "step": 327
    },
    {
      "epoch": 0.7114967462039046,
      "grad_norm": 0.6363551616668701,
      "learning_rate": 2.906724511930586e-06,
      "loss": 0.5589,
      "step": 328
    },
    {
      "epoch": 0.7136659436008677,
      "grad_norm": 0.5492388606071472,
      "learning_rate": 2.8850325379609543e-06,
      "loss": 0.5439,
      "step": 329
    },
    {
      "epoch": 0.7158351409978309,
      "grad_norm": 0.48855558037757874,
      "learning_rate": 2.8633405639913238e-06,
      "loss": 0.5342,
      "step": 330
    },
    {
      "epoch": 0.7180043383947939,
      "grad_norm": 0.5338431000709534,
      "learning_rate": 2.8416485900216924e-06,
      "loss": 0.5313,
      "step": 331
    },
    {
      "epoch": 0.720173535791757,
      "grad_norm": 0.5648824572563171,
      "learning_rate": 2.819956616052061e-06,
      "loss": 0.5521,
      "step": 332
    },
    {
      "epoch": 0.7223427331887202,
      "grad_norm": 0.5221781730651855,
      "learning_rate": 2.79826464208243e-06,
      "loss": 0.5445,
      "step": 333
    },
    {
      "epoch": 0.7245119305856833,
      "grad_norm": 0.4889802634716034,
      "learning_rate": 2.7765726681127987e-06,
      "loss": 0.5588,
      "step": 334
    },
    {
      "epoch": 0.7266811279826464,
      "grad_norm": 0.4981290400028229,
      "learning_rate": 2.7548806941431673e-06,
      "loss": 0.5709,
      "step": 335
    },
    {
      "epoch": 0.7288503253796096,
      "grad_norm": 0.5129498243331909,
      "learning_rate": 2.7331887201735356e-06,
      "loss": 0.5592,
      "step": 336
    },
    {
      "epoch": 0.7310195227765727,
      "grad_norm": 0.5097779631614685,
      "learning_rate": 2.711496746203905e-06,
      "loss": 0.5608,
      "step": 337
    },
    {
      "epoch": 0.7331887201735358,
      "grad_norm": 0.49256396293640137,
      "learning_rate": 2.6898047722342737e-06,
      "loss": 0.547,
      "step": 338
    },
    {
      "epoch": 0.735357917570499,
      "grad_norm": 0.49484625458717346,
      "learning_rate": 2.6681127982646423e-06,
      "loss": 0.5605,
      "step": 339
    },
    {
      "epoch": 0.737527114967462,
      "grad_norm": 0.4865284264087677,
      "learning_rate": 2.6464208242950113e-06,
      "loss": 0.552,
      "step": 340
    },
    {
      "epoch": 0.7396963123644251,
      "grad_norm": 0.5712233185768127,
      "learning_rate": 2.62472885032538e-06,
      "loss": 0.5229,
      "step": 341
    },
    {
      "epoch": 0.7418655097613883,
      "grad_norm": 0.4782271981239319,
      "learning_rate": 2.6030368763557486e-06,
      "loss": 0.5409,
      "step": 342
    },
    {
      "epoch": 0.7440347071583514,
      "grad_norm": 0.48679468035697937,
      "learning_rate": 2.5813449023861176e-06,
      "loss": 0.5502,
      "step": 343
    },
    {
      "epoch": 0.7462039045553145,
      "grad_norm": 0.49135392904281616,
      "learning_rate": 2.5596529284164863e-06,
      "loss": 0.552,
      "step": 344
    },
    {
      "epoch": 0.7483731019522777,
      "grad_norm": 0.4905093014240265,
      "learning_rate": 2.537960954446855e-06,
      "loss": 0.5388,
      "step": 345
    },
    {
      "epoch": 0.7505422993492408,
      "grad_norm": 0.4936610758304596,
      "learning_rate": 2.516268980477224e-06,
      "loss": 0.5408,
      "step": 346
    },
    {
      "epoch": 0.7527114967462039,
      "grad_norm": 0.4581211805343628,
      "learning_rate": 2.4945770065075926e-06,
      "loss": 0.5438,
      "step": 347
    },
    {
      "epoch": 0.754880694143167,
      "grad_norm": 0.9855303764343262,
      "learning_rate": 2.472885032537961e-06,
      "loss": 0.5422,
      "step": 348
    },
    {
      "epoch": 0.7570498915401301,
      "grad_norm": 0.4950404465198517,
      "learning_rate": 2.45119305856833e-06,
      "loss": 0.5488,
      "step": 349
    },
    {
      "epoch": 0.7592190889370932,
      "grad_norm": 0.5436333417892456,
      "learning_rate": 2.4295010845986985e-06,
      "loss": 0.529,
      "step": 350
    },
    {
      "epoch": 0.7613882863340564,
      "grad_norm": 0.6923372149467468,
      "learning_rate": 2.4078091106290675e-06,
      "loss": 0.566,
      "step": 351
    },
    {
      "epoch": 0.7635574837310195,
      "grad_norm": 0.4927901327610016,
      "learning_rate": 2.386117136659436e-06,
      "loss": 0.5445,
      "step": 352
    },
    {
      "epoch": 0.7657266811279827,
      "grad_norm": 0.6745809316635132,
      "learning_rate": 2.3644251626898048e-06,
      "loss": 0.5644,
      "step": 353
    },
    {
      "epoch": 0.7678958785249458,
      "grad_norm": 0.5441464185714722,
      "learning_rate": 2.342733188720174e-06,
      "loss": 0.5481,
      "step": 354
    },
    {
      "epoch": 0.7700650759219089,
      "grad_norm": 0.5015382170677185,
      "learning_rate": 2.3210412147505424e-06,
      "loss": 0.5632,
      "step": 355
    },
    {
      "epoch": 0.7722342733188721,
      "grad_norm": 0.48177745938301086,
      "learning_rate": 2.299349240780911e-06,
      "loss": 0.5435,
      "step": 356
    },
    {
      "epoch": 0.7744034707158352,
      "grad_norm": 0.6477603316307068,
      "learning_rate": 2.27765726681128e-06,
      "loss": 0.5488,
      "step": 357
    },
    {
      "epoch": 0.7765726681127982,
      "grad_norm": 0.49363404512405396,
      "learning_rate": 2.2559652928416487e-06,
      "loss": 0.5359,
      "step": 358
    },
    {
      "epoch": 0.7787418655097614,
      "grad_norm": 0.49064791202545166,
      "learning_rate": 2.2342733188720174e-06,
      "loss": 0.5708,
      "step": 359
    },
    {
      "epoch": 0.7809110629067245,
      "grad_norm": 0.5506601929664612,
      "learning_rate": 2.2125813449023864e-06,
      "loss": 0.5638,
      "step": 360
    },
    {
      "epoch": 0.7830802603036876,
      "grad_norm": 0.49600258469581604,
      "learning_rate": 2.190889370932755e-06,
      "loss": 0.5578,
      "step": 361
    },
    {
      "epoch": 0.7852494577006508,
      "grad_norm": 0.6045665740966797,
      "learning_rate": 2.1691973969631237e-06,
      "loss": 0.5387,
      "step": 362
    },
    {
      "epoch": 0.7874186550976139,
      "grad_norm": 0.4785294234752655,
      "learning_rate": 2.1475054229934923e-06,
      "loss": 0.5461,
      "step": 363
    },
    {
      "epoch": 0.789587852494577,
      "grad_norm": 0.505807101726532,
      "learning_rate": 2.1258134490238614e-06,
      "loss": 0.5421,
      "step": 364
    },
    {
      "epoch": 0.7917570498915402,
      "grad_norm": 0.5481471419334412,
      "learning_rate": 2.1041214750542304e-06,
      "loss": 0.5359,
      "step": 365
    },
    {
      "epoch": 0.7939262472885033,
      "grad_norm": 0.5260441899299622,
      "learning_rate": 2.0824295010845986e-06,
      "loss": 0.5801,
      "step": 366
    },
    {
      "epoch": 0.7960954446854663,
      "grad_norm": 0.5684512257575989,
      "learning_rate": 2.0607375271149677e-06,
      "loss": 0.5317,
      "step": 367
    },
    {
      "epoch": 0.7982646420824295,
      "grad_norm": 0.5323051810264587,
      "learning_rate": 2.0390455531453363e-06,
      "loss": 0.5669,
      "step": 368
    },
    {
      "epoch": 0.8004338394793926,
      "grad_norm": 0.6611142158508301,
      "learning_rate": 2.017353579175705e-06,
      "loss": 0.5582,
      "step": 369
    },
    {
      "epoch": 0.8026030368763557,
      "grad_norm": 0.47256988286972046,
      "learning_rate": 1.995661605206074e-06,
      "loss": 0.5276,
      "step": 370
    },
    {
      "epoch": 0.8047722342733189,
      "grad_norm": 0.6162890195846558,
      "learning_rate": 1.9739696312364426e-06,
      "loss": 0.5517,
      "step": 371
    },
    {
      "epoch": 0.806941431670282,
      "grad_norm": 0.5171302556991577,
      "learning_rate": 1.9522776572668117e-06,
      "loss": 0.5467,
      "step": 372
    },
    {
      "epoch": 0.8091106290672451,
      "grad_norm": 0.5168177485466003,
      "learning_rate": 1.9305856832971803e-06,
      "loss": 0.537,
      "step": 373
    },
    {
      "epoch": 0.8112798264642083,
      "grad_norm": 0.49433693289756775,
      "learning_rate": 1.908893709327549e-06,
      "loss": 0.5441,
      "step": 374
    },
    {
      "epoch": 0.8134490238611713,
      "grad_norm": 0.6254696249961853,
      "learning_rate": 1.8872017353579177e-06,
      "loss": 0.568,
      "step": 375
    },
    {
      "epoch": 0.8156182212581344,
      "grad_norm": 0.4846223294734955,
      "learning_rate": 1.8655097613882864e-06,
      "loss": 0.5314,
      "step": 376
    },
    {
      "epoch": 0.8177874186550976,
      "grad_norm": 0.5695744156837463,
      "learning_rate": 1.8438177874186552e-06,
      "loss": 0.5556,
      "step": 377
    },
    {
      "epoch": 0.8199566160520607,
      "grad_norm": 0.5135949850082397,
      "learning_rate": 1.822125813449024e-06,
      "loss": 0.5513,
      "step": 378
    },
    {
      "epoch": 0.8221258134490239,
      "grad_norm": 0.5039001107215881,
      "learning_rate": 1.8004338394793927e-06,
      "loss": 0.5395,
      "step": 379
    },
    {
      "epoch": 0.824295010845987,
      "grad_norm": 0.5153144598007202,
      "learning_rate": 1.7787418655097615e-06,
      "loss": 0.5499,
      "step": 380
    },
    {
      "epoch": 0.8264642082429501,
      "grad_norm": 0.6042090058326721,
      "learning_rate": 1.7570498915401301e-06,
      "loss": 0.561,
      "step": 381
    },
    {
      "epoch": 0.8286334056399133,
      "grad_norm": 0.5201963186264038,
      "learning_rate": 1.735357917570499e-06,
      "loss": 0.5514,
      "step": 382
    },
    {
      "epoch": 0.8308026030368764,
      "grad_norm": 0.7226424217224121,
      "learning_rate": 1.7136659436008678e-06,
      "loss": 0.5481,
      "step": 383
    },
    {
      "epoch": 0.8329718004338394,
      "grad_norm": 0.49008411169052124,
      "learning_rate": 1.6919739696312365e-06,
      "loss": 0.5577,
      "step": 384
    },
    {
      "epoch": 0.8351409978308026,
      "grad_norm": 0.5384876728057861,
      "learning_rate": 1.6702819956616053e-06,
      "loss": 0.5744,
      "step": 385
    },
    {
      "epoch": 0.8373101952277657,
      "grad_norm": 0.6859086751937866,
      "learning_rate": 1.6485900216919743e-06,
      "loss": 0.5495,
      "step": 386
    },
    {
      "epoch": 0.8394793926247288,
      "grad_norm": 0.5295915007591248,
      "learning_rate": 1.6268980477223428e-06,
      "loss": 0.5555,
      "step": 387
    },
    {
      "epoch": 0.841648590021692,
      "grad_norm": 0.49512577056884766,
      "learning_rate": 1.6052060737527116e-06,
      "loss": 0.5336,
      "step": 388
    },
    {
      "epoch": 0.8438177874186551,
      "grad_norm": 0.5301417112350464,
      "learning_rate": 1.5835140997830802e-06,
      "loss": 0.5366,
      "step": 389
    },
    {
      "epoch": 0.8459869848156182,
      "grad_norm": 0.5286414623260498,
      "learning_rate": 1.561822125813449e-06,
      "loss": 0.5301,
      "step": 390
    },
    {
      "epoch": 0.8481561822125814,
      "grad_norm": 0.4798189699649811,
      "learning_rate": 1.5401301518438181e-06,
      "loss": 0.5529,
      "step": 391
    },
    {
      "epoch": 0.8503253796095445,
      "grad_norm": 0.5175362825393677,
      "learning_rate": 1.5184381778741865e-06,
      "loss": 0.5418,
      "step": 392
    },
    {
      "epoch": 0.8524945770065075,
      "grad_norm": 0.5054755210876465,
      "learning_rate": 1.4967462039045556e-06,
      "loss": 0.5471,
      "step": 393
    },
    {
      "epoch": 0.8546637744034707,
      "grad_norm": 0.48426923155784607,
      "learning_rate": 1.475054229934924e-06,
      "loss": 0.543,
      "step": 394
    },
    {
      "epoch": 0.8568329718004338,
      "grad_norm": 0.49632081389427185,
      "learning_rate": 1.453362255965293e-06,
      "loss": 0.5336,
      "step": 395
    },
    {
      "epoch": 0.8590021691973969,
      "grad_norm": 0.5040681958198547,
      "learning_rate": 1.4316702819956619e-06,
      "loss": 0.5334,
      "step": 396
    },
    {
      "epoch": 0.8611713665943601,
      "grad_norm": 0.5700734853744507,
      "learning_rate": 1.4099783080260305e-06,
      "loss": 0.5365,
      "step": 397
    },
    {
      "epoch": 0.8633405639913232,
      "grad_norm": 0.5232958197593689,
      "learning_rate": 1.3882863340563994e-06,
      "loss": 0.5372,
      "step": 398
    },
    {
      "epoch": 0.8655097613882863,
      "grad_norm": 0.5089038610458374,
      "learning_rate": 1.3665943600867678e-06,
      "loss": 0.5431,
      "step": 399
    },
    {
      "epoch": 0.8676789587852495,
      "grad_norm": 0.5343663692474365,
      "learning_rate": 1.3449023861171368e-06,
      "loss": 0.5505,
      "step": 400
    },
    {
      "epoch": 0.8698481561822126,
      "grad_norm": 0.7061499953269958,
      "learning_rate": 1.3232104121475057e-06,
      "loss": 0.5394,
      "step": 401
    },
    {
      "epoch": 0.8720173535791758,
      "grad_norm": 0.5129175186157227,
      "learning_rate": 1.3015184381778743e-06,
      "loss": 0.5515,
      "step": 402
    },
    {
      "epoch": 0.8741865509761388,
      "grad_norm": 0.6809354424476624,
      "learning_rate": 1.2798264642082431e-06,
      "loss": 0.5453,
      "step": 403
    },
    {
      "epoch": 0.8763557483731019,
      "grad_norm": 0.5095650553703308,
      "learning_rate": 1.258134490238612e-06,
      "loss": 0.5559,
      "step": 404
    },
    {
      "epoch": 0.8785249457700651,
      "grad_norm": 0.707577645778656,
      "learning_rate": 1.2364425162689806e-06,
      "loss": 0.5318,
      "step": 405
    },
    {
      "epoch": 0.8806941431670282,
      "grad_norm": 0.5874403119087219,
      "learning_rate": 1.2147505422993492e-06,
      "loss": 0.5481,
      "step": 406
    },
    {
      "epoch": 0.8828633405639913,
      "grad_norm": 0.4959114193916321,
      "learning_rate": 1.193058568329718e-06,
      "loss": 0.5394,
      "step": 407
    },
    {
      "epoch": 0.8850325379609545,
      "grad_norm": 0.5033082365989685,
      "learning_rate": 1.171366594360087e-06,
      "loss": 0.5644,
      "step": 408
    },
    {
      "epoch": 0.8872017353579176,
      "grad_norm": 0.5417726635932922,
      "learning_rate": 1.1496746203904555e-06,
      "loss": 0.5209,
      "step": 409
    },
    {
      "epoch": 0.8893709327548807,
      "grad_norm": 0.488228440284729,
      "learning_rate": 1.1279826464208244e-06,
      "loss": 0.5406,
      "step": 410
    },
    {
      "epoch": 0.8915401301518439,
      "grad_norm": 0.9152227640151978,
      "learning_rate": 1.1062906724511932e-06,
      "loss": 0.532,
      "step": 411
    },
    {
      "epoch": 0.8937093275488069,
      "grad_norm": 0.5113642811775208,
      "learning_rate": 1.0845986984815618e-06,
      "loss": 0.5476,
      "step": 412
    },
    {
      "epoch": 0.89587852494577,
      "grad_norm": 0.5312101244926453,
      "learning_rate": 1.0629067245119307e-06,
      "loss": 0.5368,
      "step": 413
    },
    {
      "epoch": 0.8980477223427332,
      "grad_norm": 0.589953601360321,
      "learning_rate": 1.0412147505422993e-06,
      "loss": 0.5381,
      "step": 414
    },
    {
      "epoch": 0.9002169197396963,
      "grad_norm": 0.5381547212600708,
      "learning_rate": 1.0195227765726681e-06,
      "loss": 0.5492,
      "step": 415
    },
    {
      "epoch": 0.9023861171366594,
      "grad_norm": 0.5344692468643188,
      "learning_rate": 9.97830802603037e-07,
      "loss": 0.5507,
      "step": 416
    },
    {
      "epoch": 0.9045553145336226,
      "grad_norm": 0.48827818036079407,
      "learning_rate": 9.761388286334058e-07,
      "loss": 0.5365,
      "step": 417
    },
    {
      "epoch": 0.9067245119305857,
      "grad_norm": 0.5503534078598022,
      "learning_rate": 9.544468546637745e-07,
      "loss": 0.531,
      "step": 418
    },
    {
      "epoch": 0.9088937093275488,
      "grad_norm": 0.5214491486549377,
      "learning_rate": 9.327548806941432e-07,
      "loss": 0.5492,
      "step": 419
    },
    {
      "epoch": 0.911062906724512,
      "grad_norm": 0.5597646236419678,
      "learning_rate": 9.11062906724512e-07,
      "loss": 0.5273,
      "step": 420
    },
    {
      "epoch": 0.913232104121475,
      "grad_norm": 0.4814266264438629,
      "learning_rate": 8.893709327548808e-07,
      "loss": 0.5467,
      "step": 421
    },
    {
      "epoch": 0.9154013015184381,
      "grad_norm": 0.5103883147239685,
      "learning_rate": 8.676789587852495e-07,
      "loss": 0.5514,
      "step": 422
    },
    {
      "epoch": 0.9175704989154013,
      "grad_norm": 0.4807824492454529,
      "learning_rate": 8.459869848156182e-07,
      "loss": 0.5377,
      "step": 423
    },
    {
      "epoch": 0.9197396963123644,
      "grad_norm": 0.49137192964553833,
      "learning_rate": 8.242950108459872e-07,
      "loss": 0.5384,
      "step": 424
    },
    {
      "epoch": 0.9219088937093276,
      "grad_norm": 0.47830575704574585,
      "learning_rate": 8.026030368763558e-07,
      "loss": 0.5458,
      "step": 425
    },
    {
      "epoch": 0.9240780911062907,
      "grad_norm": 0.5331500768661499,
      "learning_rate": 7.809110629067245e-07,
      "loss": 0.514,
      "step": 426
    },
    {
      "epoch": 0.9262472885032538,
      "grad_norm": 0.4643237888813019,
      "learning_rate": 7.592190889370933e-07,
      "loss": 0.5404,
      "step": 427
    },
    {
      "epoch": 0.928416485900217,
      "grad_norm": 0.49474000930786133,
      "learning_rate": 7.37527114967462e-07,
      "loss": 0.5687,
      "step": 428
    },
    {
      "epoch": 0.93058568329718,
      "grad_norm": 0.551901638507843,
      "learning_rate": 7.158351409978309e-07,
      "loss": 0.5485,
      "step": 429
    },
    {
      "epoch": 0.9327548806941431,
      "grad_norm": 0.5091038346290588,
      "learning_rate": 6.941431670281997e-07,
      "loss": 0.545,
      "step": 430
    },
    {
      "epoch": 0.9349240780911063,
      "grad_norm": 0.4894218444824219,
      "learning_rate": 6.724511930585684e-07,
      "loss": 0.5391,
      "step": 431
    },
    {
      "epoch": 0.9370932754880694,
      "grad_norm": 0.5032333731651306,
      "learning_rate": 6.507592190889371e-07,
      "loss": 0.5404,
      "step": 432
    },
    {
      "epoch": 0.9392624728850325,
      "grad_norm": 0.5365532040596008,
      "learning_rate": 6.29067245119306e-07,
      "loss": 0.524,
      "step": 433
    },
    {
      "epoch": 0.9414316702819957,
      "grad_norm": 0.47491738200187683,
      "learning_rate": 6.073752711496746e-07,
      "loss": 0.5451,
      "step": 434
    },
    {
      "epoch": 0.9436008676789588,
      "grad_norm": 0.4964900314807892,
      "learning_rate": 5.856832971800435e-07,
      "loss": 0.5465,
      "step": 435
    },
    {
      "epoch": 0.9457700650759219,
      "grad_norm": 0.4745555520057678,
      "learning_rate": 5.639913232104122e-07,
      "loss": 0.5432,
      "step": 436
    },
    {
      "epoch": 0.9479392624728851,
      "grad_norm": 0.6818556785583496,
      "learning_rate": 5.422993492407809e-07,
      "loss": 0.5382,
      "step": 437
    },
    {
      "epoch": 0.9501084598698482,
      "grad_norm": 0.5324418544769287,
      "learning_rate": 5.206073752711497e-07,
      "loss": 0.5733,
      "step": 438
    },
    {
      "epoch": 0.9522776572668112,
      "grad_norm": 0.5001943111419678,
      "learning_rate": 4.989154013015185e-07,
      "loss": 0.5331,
      "step": 439
    },
    {
      "epoch": 0.9544468546637744,
      "grad_norm": 0.5757445096969604,
      "learning_rate": 4.772234273318872e-07,
      "loss": 0.5383,
      "step": 440
    },
    {
      "epoch": 0.9566160520607375,
      "grad_norm": 0.6072180271148682,
      "learning_rate": 4.55531453362256e-07,
      "loss": 0.5509,
      "step": 441
    },
    {
      "epoch": 0.9587852494577006,
      "grad_norm": 0.5225549936294556,
      "learning_rate": 4.3383947939262475e-07,
      "loss": 0.5279,
      "step": 442
    },
    {
      "epoch": 0.9609544468546638,
      "grad_norm": 0.4668637216091156,
      "learning_rate": 4.121475054229936e-07,
      "loss": 0.5488,
      "step": 443
    },
    {
      "epoch": 0.9631236442516269,
      "grad_norm": 0.46520882844924927,
      "learning_rate": 3.9045553145336227e-07,
      "loss": 0.5272,
      "step": 444
    },
    {
      "epoch": 0.96529284164859,
      "grad_norm": 0.45933347940444946,
      "learning_rate": 3.68763557483731e-07,
      "loss": 0.5526,
      "step": 445
    },
    {
      "epoch": 0.9674620390455532,
      "grad_norm": 0.6567378044128418,
      "learning_rate": 3.4707158351409984e-07,
      "loss": 0.5387,
      "step": 446
    },
    {
      "epoch": 0.9696312364425163,
      "grad_norm": 0.49450600147247314,
      "learning_rate": 3.2537960954446857e-07,
      "loss": 0.5329,
      "step": 447
    },
    {
      "epoch": 0.9718004338394793,
      "grad_norm": 0.5677459239959717,
      "learning_rate": 3.036876355748373e-07,
      "loss": 0.5459,
      "step": 448
    },
    {
      "epoch": 0.9739696312364425,
      "grad_norm": 0.4713030755519867,
      "learning_rate": 2.819956616052061e-07,
      "loss": 0.5394,
      "step": 449
    },
    {
      "epoch": 0.9761388286334056,
      "grad_norm": 0.5460206866264343,
      "learning_rate": 2.6030368763557483e-07,
      "loss": 0.5318,
      "step": 450
    },
    {
      "epoch": 0.9783080260303688,
      "grad_norm": 0.4780767560005188,
      "learning_rate": 2.386117136659436e-07,
      "loss": 0.5407,
      "step": 451
    },
    {
      "epoch": 0.9804772234273319,
      "grad_norm": 0.4803195297718048,
      "learning_rate": 2.1691973969631237e-07,
      "loss": 0.5199,
      "step": 452
    },
    {
      "epoch": 0.982646420824295,
      "grad_norm": 0.5350447297096252,
      "learning_rate": 1.9522776572668113e-07,
      "loss": 0.5434,
      "step": 453
    },
    {
      "epoch": 0.9848156182212582,
      "grad_norm": 0.5030990839004517,
      "learning_rate": 1.7353579175704992e-07,
      "loss": 0.5368,
      "step": 454
    },
    {
      "epoch": 0.9869848156182213,
      "grad_norm": 0.4834834933280945,
      "learning_rate": 1.5184381778741865e-07,
      "loss": 0.5386,
      "step": 455
    },
    {
      "epoch": 0.9891540130151844,
      "grad_norm": 0.5929272174835205,
      "learning_rate": 1.3015184381778741e-07,
      "loss": 0.5452,
      "step": 456
    },
    {
      "epoch": 0.9913232104121475,
      "grad_norm": 0.4731636047363281,
      "learning_rate": 1.0845986984815619e-07,
      "loss": 0.5294,
      "step": 457
    },
    {
      "epoch": 0.9934924078091106,
      "grad_norm": 0.48876383900642395,
      "learning_rate": 8.676789587852496e-08,
      "loss": 0.539,
      "step": 458
    },
    {
      "epoch": 0.9956616052060737,
      "grad_norm": 0.4793936610221863,
      "learning_rate": 6.507592190889371e-08,
      "loss": 0.5496,
      "step": 459
    },
    {
      "epoch": 0.9978308026030369,
      "grad_norm": 0.46681129932403564,
      "learning_rate": 4.338394793926248e-08,
      "loss": 0.5283,
      "step": 460
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5591140389442444,
      "learning_rate": 2.169197396963124e-08,
      "loss": 0.5435,
      "step": 461
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 461,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.281141529490227e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}