{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.018306636155606407,
  "eval_steps": 50,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.8306636155606407e-05,
      "grad_norm": 6.084787368774414,
      "learning_rate": 0.0,
      "loss": 0.6392,
      "step": 1
    },
    {
      "epoch": 3.6613272311212814e-05,
      "grad_norm": 8.000160217285156,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.8375,
      "step": 2
    },
    {
      "epoch": 5.491990846681922e-05,
      "grad_norm": 9.977635383605957,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.4734,
      "step": 3
    },
    {
      "epoch": 7.322654462242563e-05,
      "grad_norm": 6.60877799987793,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.7562,
      "step": 4
    },
    {
      "epoch": 9.153318077803204e-05,
      "grad_norm": 10.706132888793945,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1.0019,
      "step": 5
    },
    {
      "epoch": 0.00010983981693363844,
      "grad_norm": 10.783683776855469,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.9186,
      "step": 6
    },
    {
      "epoch": 0.00012814645308924485,
      "grad_norm": 9.012995719909668,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.029,
      "step": 7
    },
    {
      "epoch": 0.00014645308924485126,
      "grad_norm": 2.700565814971924,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.4689,
      "step": 8
    },
    {
      "epoch": 0.00016475972540045766,
      "grad_norm": 7.6247029304504395,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.6074,
      "step": 9
    },
    {
      "epoch": 0.00018306636155606407,
      "grad_norm": 6.389184951782227,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.8861,
      "step": 10
    },
    {
      "epoch": 0.00020137299771167048,
      "grad_norm": 6.685202121734619,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.5527,
      "step": 11
    },
    {
      "epoch": 0.00021967963386727689,
      "grad_norm": 6.647484302520752,
      "learning_rate": 2.2e-06,
      "loss": 0.5459,
      "step": 12
    },
    {
      "epoch": 0.0002379862700228833,
      "grad_norm": 7.842504978179932,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.7269,
      "step": 13
    },
    {
      "epoch": 0.0002562929061784897,
      "grad_norm": 9.710944175720215,
      "learning_rate": 2.6e-06,
      "loss": 0.7054,
      "step": 14
    },
    {
      "epoch": 0.00027459954233409613,
      "grad_norm": 10.485865592956543,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.1884,
      "step": 15
    },
    {
      "epoch": 0.0002929061784897025,
      "grad_norm": 3.510791301727295,
      "learning_rate": 3e-06,
      "loss": 0.5241,
      "step": 16
    },
    {
      "epoch": 0.00031121281464530895,
      "grad_norm": 11.97901439666748,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.0192,
      "step": 17
    },
    {
      "epoch": 0.00032951945080091533,
      "grad_norm": 8.022102355957031,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.5731,
      "step": 18
    },
    {
      "epoch": 0.00034782608695652176,
      "grad_norm": 7.800467014312744,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.5588,
      "step": 19
    },
    {
      "epoch": 0.00036613272311212814,
      "grad_norm": 5.90647029876709,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.4773,
      "step": 20
    },
    {
      "epoch": 0.0003844393592677346,
      "grad_norm": 10.05663776397705,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.8649,
      "step": 21
    },
    {
      "epoch": 0.00040274599542334096,
      "grad_norm": 3.0191540718078613,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.2516,
      "step": 22
    },
    {
      "epoch": 0.0004210526315789474,
      "grad_norm": 8.017966270446777,
      "learning_rate": 4.4e-06,
      "loss": 0.7442,
      "step": 23
    },
    {
      "epoch": 0.00043935926773455377,
      "grad_norm": 6.663602352142334,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.6345,
      "step": 24
    },
    {
      "epoch": 0.0004576659038901602,
      "grad_norm": 8.327048301696777,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6079,
      "step": 25
    },
    {
      "epoch": 0.0004759725400457666,
      "grad_norm": 15.23943042755127,
      "learning_rate": 5e-06,
      "loss": 0.9292,
      "step": 26
    },
    {
      "epoch": 0.000494279176201373,
      "grad_norm": 11.373315811157227,
      "learning_rate": 5.2e-06,
      "loss": 0.589,
      "step": 27
    },
    {
      "epoch": 0.0005125858123569794,
      "grad_norm": 9.903393745422363,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.4872,
      "step": 28
    },
    {
      "epoch": 0.0005308924485125858,
      "grad_norm": 17.918140411376953,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.1316,
      "step": 29
    },
    {
      "epoch": 0.0005491990846681923,
      "grad_norm": 7.308037281036377,
      "learning_rate": 5.8e-06,
      "loss": 0.389,
      "step": 30
    },
    {
      "epoch": 0.0005675057208237986,
      "grad_norm": 8.060295104980469,
      "learning_rate": 6e-06,
      "loss": 0.582,
      "step": 31
    },
    {
      "epoch": 0.000585812356979405,
      "grad_norm": 8.332834243774414,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.1902,
      "step": 32
    },
    {
      "epoch": 0.0006041189931350115,
      "grad_norm": 8.524596214294434,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.3101,
      "step": 33
    },
    {
      "epoch": 0.0006224256292906179,
      "grad_norm": 10.255339622497559,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.5759,
      "step": 34
    },
    {
      "epoch": 0.0006407322654462242,
      "grad_norm": 12.656609535217285,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.0338,
      "step": 35
    },
    {
      "epoch": 0.0006590389016018307,
      "grad_norm": 6.707568645477295,
      "learning_rate": 7e-06,
      "loss": 0.3607,
      "step": 36
    },
    {
      "epoch": 0.0006773455377574371,
      "grad_norm": 6.263491153717041,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.4405,
      "step": 37
    },
    {
      "epoch": 0.0006956521739130435,
      "grad_norm": 5.70526647567749,
      "learning_rate": 7.4e-06,
      "loss": 0.5682,
      "step": 38
    },
    {
      "epoch": 0.0007139588100686499,
      "grad_norm": 4.024957180023193,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.5816,
      "step": 39
    },
    {
      "epoch": 0.0007322654462242563,
      "grad_norm": 8.361418724060059,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.8628,
      "step": 40
    },
    {
      "epoch": 0.0007505720823798627,
      "grad_norm": 4.543780326843262,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.3442,
      "step": 41
    },
    {
      "epoch": 0.0007688787185354692,
      "grad_norm": 8.164483070373535,
      "learning_rate": 8.2e-06,
      "loss": 0.5429,
      "step": 42
    },
    {
      "epoch": 0.0007871853546910755,
      "grad_norm": 7.965429306030273,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.7367,
      "step": 43
    },
    {
      "epoch": 0.0008054919908466819,
      "grad_norm": 16.735258102416992,
      "learning_rate": 8.6e-06,
      "loss": 0.9958,
      "step": 44
    },
    {
      "epoch": 0.0008237986270022883,
      "grad_norm": 8.915084838867188,
      "learning_rate": 8.8e-06,
      "loss": 0.4859,
      "step": 45
    },
    {
      "epoch": 0.0008421052631578948,
      "grad_norm": 9.113476753234863,
      "learning_rate": 9e-06,
      "loss": 1.1163,
      "step": 46
    },
    {
      "epoch": 0.0008604118993135011,
      "grad_norm": 11.264951705932617,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.7696,
      "step": 47
    },
    {
      "epoch": 0.0008787185354691075,
      "grad_norm": 11.313631057739258,
      "learning_rate": 9.4e-06,
      "loss": 1.0535,
      "step": 48
    },
    {
      "epoch": 0.000897025171624714,
      "grad_norm": 2.9443583488464355,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.2185,
      "step": 49
    },
    {
      "epoch": 0.0009153318077803204,
      "grad_norm": 5.448572635650635,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.3536,
      "step": 50
    },
    {
      "epoch": 0.0009336384439359267,
      "grad_norm": 7.4222731590271,
      "learning_rate": 1e-05,
      "loss": 0.5806,
      "step": 51
    },
    {
      "epoch": 0.0009519450800915332,
      "grad_norm": 6.155364990234375,
      "learning_rate": 1.02e-05,
      "loss": 0.8657,
      "step": 52
    },
    {
      "epoch": 0.0009702517162471396,
      "grad_norm": 7.7926249504089355,
      "learning_rate": 1.04e-05,
      "loss": 0.8655,
      "step": 53
    },
    {
      "epoch": 0.000988558352402746,
      "grad_norm": 7.443905830383301,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.701,
      "step": 54
    },
    {
      "epoch": 0.0010068649885583525,
      "grad_norm": 5.242746353149414,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.3073,
      "step": 55
    },
    {
      "epoch": 0.0010251716247139588,
      "grad_norm": 3.663163661956787,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.4435,
      "step": 56
    },
    {
      "epoch": 0.0010434782608695651,
      "grad_norm": 6.215541839599609,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.4789,
      "step": 57
    },
    {
      "epoch": 0.0010617848970251717,
      "grad_norm": 8.457674026489258,
      "learning_rate": 1.14e-05,
      "loss": 0.9246,
      "step": 58
    },
    {
      "epoch": 0.001080091533180778,
      "grad_norm": 5.378246307373047,
      "learning_rate": 1.16e-05,
      "loss": 0.3759,
      "step": 59
    },
    {
      "epoch": 0.0010983981693363845,
      "grad_norm": 10.745940208435059,
      "learning_rate": 1.18e-05,
      "loss": 1.2856,
      "step": 60
    },
    {
      "epoch": 0.0011167048054919909,
      "grad_norm": 2.361469030380249,
      "learning_rate": 1.2e-05,
      "loss": 0.2053,
      "step": 61
    },
    {
      "epoch": 0.0011350114416475972,
      "grad_norm": 7.341982841491699,
      "learning_rate": 1.22e-05,
      "loss": 0.5777,
      "step": 62
    },
    {
      "epoch": 0.0011533180778032037,
      "grad_norm": 1.9843751192092896,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.1818,
      "step": 63
    },
    {
      "epoch": 0.00117162471395881,
      "grad_norm": 7.640516757965088,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 1.0094,
      "step": 64
    },
    {
      "epoch": 0.0011899313501144164,
      "grad_norm": 15.899582862854004,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 1.5402,
      "step": 65
    },
    {
      "epoch": 0.001208237986270023,
      "grad_norm": 6.037946701049805,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.5179,
      "step": 66
    },
    {
      "epoch": 0.0012265446224256293,
      "grad_norm": 1.6879180669784546,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.2435,
      "step": 67
    },
    {
      "epoch": 0.0012448512585812358,
      "grad_norm": 3.539372682571411,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.2219,
      "step": 68
    },
    {
      "epoch": 0.0012631578947368421,
      "grad_norm": 9.577892303466797,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 1.0921,
      "step": 69
    },
    {
      "epoch": 0.0012814645308924484,
      "grad_norm": 7.512180328369141,
      "learning_rate": 1.38e-05,
      "loss": 1.1851,
      "step": 70
    },
    {
      "epoch": 0.001299771167048055,
      "grad_norm": 6.8162841796875,
      "learning_rate": 1.4e-05,
      "loss": 0.5342,
      "step": 71
    },
    {
      "epoch": 0.0013180778032036613,
      "grad_norm": 13.964672088623047,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.8404,
      "step": 72
    },
    {
      "epoch": 0.0013363844393592676,
      "grad_norm": 10.541035652160645,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.9331,
      "step": 73
    },
    {
      "epoch": 0.0013546910755148742,
      "grad_norm": 5.94256067276001,
      "learning_rate": 1.46e-05,
      "loss": 0.7509,
      "step": 74
    },
    {
      "epoch": 0.0013729977116704805,
      "grad_norm": 6.224687099456787,
      "learning_rate": 1.48e-05,
      "loss": 0.5732,
      "step": 75
    },
    {
      "epoch": 0.001391304347826087,
      "grad_norm": 6.311083793640137,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.6101,
      "step": 76
    },
    {
      "epoch": 0.0014096109839816934,
      "grad_norm": 5.685784339904785,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.6369,
      "step": 77
    },
    {
      "epoch": 0.0014279176201372997,
      "grad_norm": 4.5122504234313965,
      "learning_rate": 1.54e-05,
      "loss": 0.5075,
      "step": 78
    },
    {
      "epoch": 0.0014462242562929062,
      "grad_norm": 4.641362190246582,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.3565,
      "step": 79
    },
    {
      "epoch": 0.0014645308924485126,
      "grad_norm": 6.5127177238464355,
      "learning_rate": 1.58e-05,
      "loss": 0.9192,
      "step": 80
    },
    {
      "epoch": 0.001482837528604119,
      "grad_norm": 4.107303619384766,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.458,
      "step": 81
    },
    {
      "epoch": 0.0015011441647597254,
      "grad_norm": 5.691149711608887,
      "learning_rate": 1.62e-05,
      "loss": 0.4345,
      "step": 82
    },
    {
      "epoch": 0.0015194508009153318,
      "grad_norm": 4.126266002655029,
      "learning_rate": 1.64e-05,
      "loss": 0.3686,
      "step": 83
    },
    {
      "epoch": 0.0015377574370709383,
      "grad_norm": 5.336889266967773,
      "learning_rate": 1.66e-05,
      "loss": 0.3863,
      "step": 84
    },
    {
      "epoch": 0.0015560640732265446,
      "grad_norm": 9.846256256103516,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.8348,
      "step": 85
    },
    {
      "epoch": 0.001574370709382151,
      "grad_norm": 3.1803081035614014,
      "learning_rate": 1.7e-05,
      "loss": 0.5672,
      "step": 86
    },
    {
      "epoch": 0.0015926773455377575,
      "grad_norm": 4.788447380065918,
      "learning_rate": 1.72e-05,
      "loss": 0.3226,
      "step": 87
    },
    {
      "epoch": 0.0016109839816933638,
      "grad_norm": 4.775529861450195,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.7328,
      "step": 88
    },
    {
      "epoch": 0.0016292906178489702,
      "grad_norm": 4.884747505187988,
      "learning_rate": 1.76e-05,
      "loss": 0.5481,
      "step": 89
    },
    {
      "epoch": 0.0016475972540045767,
      "grad_norm": 11.710714340209961,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.7645,
      "step": 90
    },
    {
      "epoch": 0.001665903890160183,
      "grad_norm": 7.733964443206787,
      "learning_rate": 1.8e-05,
      "loss": 0.6914,
      "step": 91
    },
    {
      "epoch": 0.0016842105263157896,
      "grad_norm": 3.1426684856414795,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.214,
      "step": 92
    },
    {
      "epoch": 0.001702517162471396,
      "grad_norm": 8.66888427734375,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.7952,
      "step": 93
    },
    {
      "epoch": 0.0017208237986270022,
      "grad_norm": 4.978145599365234,
      "learning_rate": 1.86e-05,
      "loss": 0.7386,
      "step": 94
    },
    {
      "epoch": 0.0017391304347826088,
      "grad_norm": 9.449872970581055,
      "learning_rate": 1.88e-05,
      "loss": 0.7716,
      "step": 95
    },
    {
      "epoch": 0.001757437070938215,
      "grad_norm": 5.055934429168701,
      "learning_rate": 1.9e-05,
      "loss": 0.6527,
      "step": 96
    },
    {
      "epoch": 0.0017757437070938214,
      "grad_norm": 3.6194591522216797,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.424,
      "step": 97
    },
    {
      "epoch": 0.001794050343249428,
      "grad_norm": 5.86330509185791,
      "learning_rate": 1.94e-05,
      "loss": 0.4895,
      "step": 98
    },
    {
      "epoch": 0.0018123569794050343,
      "grad_norm": 5.012547492980957,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.3257,
      "step": 99
    },
    {
      "epoch": 0.0018306636155606408,
      "grad_norm": 6.429784297943115,
      "learning_rate": 1.98e-05,
      "loss": 0.4644,
      "step": 100
    },
    {
      "epoch": 0.0018489702517162471,
      "grad_norm": 7.005829811096191,
      "learning_rate": 2e-05,
      "loss": 0.3813,
      "step": 101
    },
    {
      "epoch": 0.0018672768878718535,
      "grad_norm": 5.1893415451049805,
      "learning_rate": 1.9999939076577906e-05,
      "loss": 0.484,
      "step": 102
    },
    {
      "epoch": 0.00188558352402746,
      "grad_norm": 5.120594501495361,
      "learning_rate": 1.9999756307053947e-05,
      "loss": 0.4228,
      "step": 103
    },
    {
      "epoch": 0.0019038901601830663,
      "grad_norm": 4.1509904861450195,
      "learning_rate": 1.9999451693655125e-05,
      "loss": 0.3841,
      "step": 104
    },
    {
      "epoch": 0.0019221967963386727,
      "grad_norm": 6.078530788421631,
      "learning_rate": 1.9999025240093045e-05,
      "loss": 0.509,
      "step": 105
    },
    {
      "epoch": 0.0019405034324942792,
      "grad_norm": 4.366968631744385,
      "learning_rate": 1.9998476951563914e-05,
      "loss": 0.4813,
      "step": 106
    },
    {
      "epoch": 0.0019588100686498858,
      "grad_norm": 3.9347426891326904,
      "learning_rate": 1.9997806834748455e-05,
      "loss": 0.3545,
      "step": 107
    },
    {
      "epoch": 0.001977116704805492,
      "grad_norm": 8.536993026733398,
      "learning_rate": 1.9997014897811834e-05,
      "loss": 0.7704,
      "step": 108
    },
    {
      "epoch": 0.0019954233409610984,
      "grad_norm": 4.600913047790527,
      "learning_rate": 1.9996101150403543e-05,
      "loss": 0.4502,
      "step": 109
    },
    {
      "epoch": 0.002013729977116705,
      "grad_norm": 10.658208847045898,
      "learning_rate": 1.9995065603657317e-05,
      "loss": 1.076,
      "step": 110
    },
    {
      "epoch": 0.002032036613272311,
      "grad_norm": 5.670320510864258,
      "learning_rate": 1.999390827019096e-05,
      "loss": 0.7519,
      "step": 111
    },
    {
      "epoch": 0.0020503432494279176,
      "grad_norm": 7.06428337097168,
      "learning_rate": 1.999262916410621e-05,
      "loss": 0.4423,
      "step": 112
    },
    {
      "epoch": 0.002068649885583524,
      "grad_norm": 5.1104254722595215,
      "learning_rate": 1.9991228300988586e-05,
      "loss": 0.4446,
      "step": 113
    },
    {
      "epoch": 0.0020869565217391303,
      "grad_norm": 3.8655929565429688,
      "learning_rate": 1.998970569790715e-05,
      "loss": 0.3284,
      "step": 114
    },
    {
      "epoch": 0.002105263157894737,
      "grad_norm": 2.9013712406158447,
      "learning_rate": 1.9988061373414342e-05,
      "loss": 0.3442,
      "step": 115
    },
    {
      "epoch": 0.0021235697940503433,
      "grad_norm": 13.76830768585205,
      "learning_rate": 1.9986295347545738e-05,
      "loss": 1.2942,
      "step": 116
    },
    {
      "epoch": 0.0021418764302059494,
      "grad_norm": 2.8040542602539062,
      "learning_rate": 1.9984407641819812e-05,
      "loss": 0.1902,
      "step": 117
    },
    {
      "epoch": 0.002160183066361556,
      "grad_norm": 5.416018009185791,
      "learning_rate": 1.9982398279237657e-05,
      "loss": 0.5331,
      "step": 118
    },
    {
      "epoch": 0.0021784897025171625,
      "grad_norm": 4.10576057434082,
      "learning_rate": 1.9980267284282718e-05,
      "loss": 0.352,
      "step": 119
    },
    {
      "epoch": 0.002196796338672769,
      "grad_norm": 3.7827014923095703,
      "learning_rate": 1.9978014682920503e-05,
      "loss": 0.3332,
      "step": 120
    },
    {
      "epoch": 0.002215102974828375,
      "grad_norm": 5.295076847076416,
      "learning_rate": 1.9975640502598243e-05,
      "loss": 0.2993,
      "step": 121
    },
    {
      "epoch": 0.0022334096109839817,
      "grad_norm": 5.779493808746338,
      "learning_rate": 1.997314477224458e-05,
      "loss": 0.8626,
      "step": 122
    },
    {
      "epoch": 0.0022517162471395883,
      "grad_norm": 6.701866626739502,
      "learning_rate": 1.9970527522269204e-05,
      "loss": 0.4652,
      "step": 123
    },
    {
      "epoch": 0.0022700228832951944,
      "grad_norm": 11.795001983642578,
      "learning_rate": 1.9967788784562474e-05,
      "loss": 0.4393,
      "step": 124
    },
    {
      "epoch": 0.002288329519450801,
      "grad_norm": 3.3595681190490723,
      "learning_rate": 1.9964928592495046e-05,
      "loss": 0.2178,
      "step": 125
    },
    {
      "epoch": 0.0023066361556064075,
      "grad_norm": 5.417965412139893,
      "learning_rate": 1.9961946980917457e-05,
      "loss": 0.6563,
      "step": 126
    },
    {
      "epoch": 0.0023249427917620136,
      "grad_norm": 9.861664772033691,
      "learning_rate": 1.9958843986159705e-05,
      "loss": 0.9203,
      "step": 127
    },
    {
      "epoch": 0.00234324942791762,
      "grad_norm": 6.221032619476318,
      "learning_rate": 1.99556196460308e-05,
      "loss": 0.7037,
      "step": 128
    },
    {
      "epoch": 0.0023615560640732267,
      "grad_norm": 2.6676557064056396,
      "learning_rate": 1.9952273999818312e-05,
      "loss": 0.1846,
      "step": 129
    },
    {
      "epoch": 0.0023798627002288328,
      "grad_norm": 4.812030792236328,
      "learning_rate": 1.9948807088287884e-05,
      "loss": 0.3774,
      "step": 130
    },
    {
      "epoch": 0.0023981693363844393,
      "grad_norm": 9.71884822845459,
      "learning_rate": 1.9945218953682736e-05,
      "loss": 0.4416,
      "step": 131
    },
    {
      "epoch": 0.002416475972540046,
      "grad_norm": 5.498316287994385,
      "learning_rate": 1.9941509639723155e-05,
      "loss": 0.7989,
      "step": 132
    },
    {
      "epoch": 0.002434782608695652,
      "grad_norm": 2.3858511447906494,
      "learning_rate": 1.9937679191605964e-05,
      "loss": 0.1593,
      "step": 133
    },
    {
      "epoch": 0.0024530892448512585,
      "grad_norm": 11.50888442993164,
      "learning_rate": 1.9933727656003964e-05,
      "loss": 0.5981,
      "step": 134
    },
    {
      "epoch": 0.002471395881006865,
      "grad_norm": 6.000802516937256,
      "learning_rate": 1.992965508106537e-05,
      "loss": 0.5936,
      "step": 135
    },
    {
      "epoch": 0.0024897025171624716,
      "grad_norm": 5.783041477203369,
      "learning_rate": 1.9925461516413224e-05,
      "loss": 0.5644,
      "step": 136
    },
    {
      "epoch": 0.0025080091533180777,
      "grad_norm": 4.8185954093933105,
      "learning_rate": 1.9921147013144782e-05,
      "loss": 0.5223,
      "step": 137
    },
    {
      "epoch": 0.0025263157894736842,
      "grad_norm": 5.774040222167969,
      "learning_rate": 1.9916711623830904e-05,
      "loss": 0.4565,
      "step": 138
    },
    {
      "epoch": 0.0025446224256292908,
      "grad_norm": 7.319972515106201,
      "learning_rate": 1.991215540251542e-05,
      "loss": 0.6701,
      "step": 139
    },
    {
      "epoch": 0.002562929061784897,
      "grad_norm": 7.840810775756836,
      "learning_rate": 1.9907478404714438e-05,
      "loss": 0.7881,
      "step": 140
    },
    {
      "epoch": 0.0025812356979405034,
      "grad_norm": 3.3393847942352295,
      "learning_rate": 1.9902680687415704e-05,
      "loss": 0.3798,
      "step": 141
    },
    {
      "epoch": 0.00259954233409611,
      "grad_norm": 7.432023048400879,
      "learning_rate": 1.989776230907789e-05,
      "loss": 0.4875,
      "step": 142
    },
    {
      "epoch": 0.002617848970251716,
      "grad_norm": 4.963670253753662,
      "learning_rate": 1.9892723329629885e-05,
      "loss": 0.439,
      "step": 143
    },
    {
      "epoch": 0.0026361556064073226,
      "grad_norm": 5.046631813049316,
      "learning_rate": 1.988756381047006e-05,
      "loss": 0.5672,
      "step": 144
    },
    {
      "epoch": 0.002654462242562929,
      "grad_norm": 7.650580883026123,
      "learning_rate": 1.988228381446553e-05,
      "loss": 0.7363,
      "step": 145
    },
    {
      "epoch": 0.0026727688787185353,
      "grad_norm": 7.517553806304932,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 0.707,
      "step": 146
    },
    {
      "epoch": 0.002691075514874142,
      "grad_norm": 2.3692057132720947,
      "learning_rate": 1.987136265072988e-05,
      "loss": 0.2329,
      "step": 147
    },
    {
      "epoch": 0.0027093821510297484,
      "grad_norm": 3.4414758682250977,
      "learning_rate": 1.9865721616069695e-05,
      "loss": 0.3901,
      "step": 148
    },
    {
      "epoch": 0.0027276887871853545,
      "grad_norm": 4.910306930541992,
      "learning_rate": 1.985996037070505e-05,
      "loss": 0.5497,
      "step": 149
    },
    {
      "epoch": 0.002745995423340961,
      "grad_norm": 6.384312629699707,
      "learning_rate": 1.9854078984834904e-05,
      "loss": 0.481,
      "step": 150
    },
    {
      "epoch": 0.0027643020594965676,
      "grad_norm": 3.9407761096954346,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 0.2972,
      "step": 151
    },
    {
      "epoch": 0.002782608695652174,
      "grad_norm": 5.0933122634887695,
      "learning_rate": 1.984195607969242e-05,
      "loss": 0.293,
      "step": 152
    },
    {
      "epoch": 0.00280091533180778,
      "grad_norm": 12.269235610961914,
      "learning_rate": 1.983571470813386e-05,
      "loss": 0.7898,
      "step": 153
    },
    {
      "epoch": 0.0028192219679633868,
      "grad_norm": 7.075507640838623,
      "learning_rate": 1.9829353491495545e-05,
      "loss": 0.7744,
      "step": 154
    },
    {
      "epoch": 0.0028375286041189933,
      "grad_norm": 6.747109413146973,
      "learning_rate": 1.982287250728689e-05,
      "loss": 0.8134,
      "step": 155
    },
    {
      "epoch": 0.0028558352402745994,
      "grad_norm": 6.466359615325928,
      "learning_rate": 1.9816271834476642e-05,
      "loss": 0.379,
      "step": 156
    },
    {
      "epoch": 0.002874141876430206,
      "grad_norm": 4.720534801483154,
      "learning_rate": 1.9809551553491918e-05,
      "loss": 0.3114,
      "step": 157
    },
    {
      "epoch": 0.0028924485125858125,
      "grad_norm": 6.6202921867370605,
      "learning_rate": 1.9802711746217222e-05,
      "loss": 0.4371,
      "step": 158
    },
    {
      "epoch": 0.0029107551487414186,
      "grad_norm": 1.6486483812332153,
      "learning_rate": 1.979575249599344e-05,
      "loss": 0.0489,
      "step": 159
    },
    {
      "epoch": 0.002929061784897025,
      "grad_norm": 7.626447677612305,
      "learning_rate": 1.9788673887616852e-05,
      "loss": 0.4217,
      "step": 160
    },
    {
      "epoch": 0.0029473684210526317,
      "grad_norm": 4.698169231414795,
      "learning_rate": 1.9781476007338058e-05,
      "loss": 0.264,
      "step": 161
    },
    {
      "epoch": 0.002965675057208238,
      "grad_norm": 8.18157958984375,
      "learning_rate": 1.9774158942860962e-05,
      "loss": 0.6062,
      "step": 162
    },
    {
      "epoch": 0.0029839816933638443,
      "grad_norm": 1.4313340187072754,
      "learning_rate": 1.9766722783341682e-05,
      "loss": 0.1536,
      "step": 163
    },
    {
      "epoch": 0.003002288329519451,
      "grad_norm": 5.4740705490112305,
      "learning_rate": 1.9759167619387474e-05,
      "loss": 0.4488,
      "step": 164
    },
    {
      "epoch": 0.0030205949656750574,
      "grad_norm": 8.551219940185547,
      "learning_rate": 1.9751493543055634e-05,
      "loss": 0.4957,
      "step": 165
    },
    {
      "epoch": 0.0030389016018306635,
      "grad_norm": 6.109156131744385,
      "learning_rate": 1.9743700647852356e-05,
      "loss": 0.5566,
      "step": 166
    },
    {
      "epoch": 0.00305720823798627,
      "grad_norm": 8.344240188598633,
      "learning_rate": 1.9735789028731603e-05,
      "loss": 0.8809,
      "step": 167
    },
    {
      "epoch": 0.0030755148741418766,
      "grad_norm": 7.053730010986328,
      "learning_rate": 1.972775878209397e-05,
      "loss": 0.6096,
      "step": 168
    },
    {
      "epoch": 0.0030938215102974827,
      "grad_norm": 8.59719181060791,
      "learning_rate": 1.9719610005785466e-05,
      "loss": 0.4776,
      "step": 169
    },
    {
      "epoch": 0.0031121281464530893,
      "grad_norm": 7.1130571365356445,
      "learning_rate": 1.971134279909636e-05,
      "loss": 0.8,
      "step": 170
    },
    {
      "epoch": 0.003130434782608696,
      "grad_norm": 5.579831123352051,
      "learning_rate": 1.9702957262759964e-05,
      "loss": 0.4925,
      "step": 171
    },
    {
      "epoch": 0.003148741418764302,
      "grad_norm": 3.0703647136688232,
      "learning_rate": 1.9694453498951392e-05,
      "loss": 0.2191,
      "step": 172
    },
    {
      "epoch": 0.0031670480549199085,
      "grad_norm": 2.900186538696289,
      "learning_rate": 1.9685831611286312e-05,
      "loss": 0.3114,
      "step": 173
    },
    {
      "epoch": 0.003185354691075515,
      "grad_norm": 1.6251899003982544,
      "learning_rate": 1.9677091704819714e-05,
      "loss": 0.2184,
      "step": 174
    },
    {
      "epoch": 0.003203661327231121,
      "grad_norm": 8.224297523498535,
      "learning_rate": 1.9668233886044597e-05,
      "loss": 0.6323,
      "step": 175
    },
    {
      "epoch": 0.0032219679633867277,
      "grad_norm": 7.209324836730957,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 0.6906,
      "step": 176
    },
    {
      "epoch": 0.003240274599542334,
      "grad_norm": 4.86000919342041,
      "learning_rate": 1.9650164944723116e-05,
      "loss": 0.3577,
      "step": 177
    },
    {
      "epoch": 0.0032585812356979403,
      "grad_norm": 2.0504913330078125,
      "learning_rate": 1.96409540423411e-05,
      "loss": 0.3458,
      "step": 178
    },
    {
      "epoch": 0.003276887871853547,
      "grad_norm": 3.175110101699829,
      "learning_rate": 1.9631625667976584e-05,
      "loss": 0.244,
      "step": 179
    },
    {
      "epoch": 0.0032951945080091534,
      "grad_norm": 4.946536540985107,
      "learning_rate": 1.9622179935292855e-05,
      "loss": 0.4006,
      "step": 180
    },
    {
      "epoch": 0.00331350114416476,
      "grad_norm": 5.009103298187256,
      "learning_rate": 1.961261695938319e-05,
      "loss": 0.2098,
      "step": 181
    },
    {
      "epoch": 0.003331807780320366,
      "grad_norm": 4.852262020111084,
      "learning_rate": 1.9602936856769432e-05,
      "loss": 0.3542,
      "step": 182
    },
    {
      "epoch": 0.0033501144164759726,
      "grad_norm": 3.4230353832244873,
      "learning_rate": 1.9593139745400575e-05,
      "loss": 0.3436,
      "step": 183
    },
    {
      "epoch": 0.003368421052631579,
      "grad_norm": 4.419967174530029,
      "learning_rate": 1.9583225744651334e-05,
      "loss": 0.2836,
      "step": 184
    },
    {
      "epoch": 0.0033867276887871852,
      "grad_norm": 8.700972557067871,
      "learning_rate": 1.9573194975320672e-05,
      "loss": 0.6588,
      "step": 185
    },
    {
      "epoch": 0.003405034324942792,
      "grad_norm": 8.335065841674805,
      "learning_rate": 1.9563047559630356e-05,
      "loss": 0.6679,
      "step": 186
    },
    {
      "epoch": 0.0034233409610983983,
      "grad_norm": 10.574755668640137,
      "learning_rate": 1.9552783621223437e-05,
      "loss": 0.6755,
      "step": 187
    },
    {
      "epoch": 0.0034416475972540044,
      "grad_norm": 4.3693671226501465,
      "learning_rate": 1.954240328516277e-05,
      "loss": 0.2874,
      "step": 188
    },
    {
      "epoch": 0.003459954233409611,
      "grad_norm": 3.7697315216064453,
      "learning_rate": 1.9531906677929472e-05,
      "loss": 0.2167,
      "step": 189
    },
    {
      "epoch": 0.0034782608695652175,
      "grad_norm": 9.218770027160645,
      "learning_rate": 1.9521293927421388e-05,
      "loss": 0.3588,
      "step": 190
    },
    {
      "epoch": 0.0034965675057208236,
      "grad_norm": 8.993297576904297,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.7196,
      "step": 191
    },
    {
      "epoch": 0.00351487414187643,
      "grad_norm": 5.189692974090576,
      "learning_rate": 1.9499720515246524e-05,
      "loss": 0.4028,
      "step": 192
    },
    {
      "epoch": 0.0035331807780320367,
      "grad_norm": 2.8236076831817627,
      "learning_rate": 1.9488760116444966e-05,
      "loss": 0.337,
      "step": 193
    },
    {
      "epoch": 0.003551487414187643,
      "grad_norm": 9.882134437561035,
      "learning_rate": 1.947768410009586e-05,
      "loss": 0.8133,
      "step": 194
    },
    {
      "epoch": 0.0035697940503432494,
      "grad_norm": 6.428056716918945,
      "learning_rate": 1.9466492601156964e-05,
      "loss": 0.7583,
      "step": 195
    },
    {
      "epoch": 0.003588100686498856,
      "grad_norm": 9.280491828918457,
      "learning_rate": 1.945518575599317e-05,
      "loss": 0.4404,
      "step": 196
    },
    {
      "epoch": 0.0036064073226544625,
      "grad_norm": 6.298871040344238,
      "learning_rate": 1.944376370237481e-05,
      "loss": 0.3666,
      "step": 197
    },
    {
      "epoch": 0.0036247139588100686,
      "grad_norm": 6.506961822509766,
      "learning_rate": 1.943222657947601e-05,
      "loss": 0.2982,
      "step": 198
    },
    {
      "epoch": 0.003643020594965675,
      "grad_norm": 4.006402015686035,
      "learning_rate": 1.942057452787297e-05,
      "loss": 0.3572,
      "step": 199
    },
    {
      "epoch": 0.0036613272311212816,
      "grad_norm": 3.9380829334259033,
      "learning_rate": 1.9408807689542257e-05,
      "loss": 0.2622,
      "step": 200
    },
    {
      "epoch": 0.0036796338672768878,
      "grad_norm": 3.6081972122192383,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.3038,
      "step": 201
    },
    {
      "epoch": 0.0036979405034324943,
      "grad_norm": 4.289360046386719,
      "learning_rate": 1.938493022759556e-05,
      "loss": 0.2393,
      "step": 202
    },
    {
      "epoch": 0.003716247139588101,
      "grad_norm": 5.4397358894348145,
      "learning_rate": 1.937281989491892e-05,
      "loss": 0.416,
      "step": 203
    },
    {
      "epoch": 0.003734553775743707,
      "grad_norm": 2.6657931804656982,
      "learning_rate": 1.9360595357389735e-05,
      "loss": 0.1847,
      "step": 204
    },
    {
      "epoch": 0.0037528604118993135,
      "grad_norm": 10.061616897583008,
      "learning_rate": 1.9348256763960146e-05,
      "loss": 1.0184,
      "step": 205
    },
    {
      "epoch": 0.00377116704805492,
      "grad_norm": 5.163506031036377,
      "learning_rate": 1.9335804264972018e-05,
      "loss": 0.4527,
      "step": 206
    },
    {
      "epoch": 0.003789473684210526,
      "grad_norm": 8.941624641418457,
      "learning_rate": 1.9323238012155125e-05,
      "loss": 0.4174,
      "step": 207
    },
    {
      "epoch": 0.0038077803203661327,
      "grad_norm": 10.84901237487793,
      "learning_rate": 1.9310558158625286e-05,
      "loss": 0.7763,
      "step": 208
    },
    {
      "epoch": 0.0038260869565217392,
      "grad_norm": 6.312254428863525,
      "learning_rate": 1.9297764858882516e-05,
      "loss": 0.5921,
      "step": 209
    },
    {
      "epoch": 0.0038443935926773453,
      "grad_norm": 2.949817419052124,
      "learning_rate": 1.9284858268809135e-05,
      "loss": 0.1387,
      "step": 210
    },
    {
      "epoch": 0.003862700228832952,
      "grad_norm": 8.410788536071777,
      "learning_rate": 1.9271838545667876e-05,
      "loss": 0.6722,
      "step": 211
    },
    {
      "epoch": 0.0038810068649885584,
      "grad_norm": 14.2056303024292,
      "learning_rate": 1.925870584809995e-05,
      "loss": 1.3406,
      "step": 212
    },
    {
      "epoch": 0.003899313501144165,
      "grad_norm": 6.280990123748779,
      "learning_rate": 1.9245460336123136e-05,
      "loss": 0.2529,
      "step": 213
    },
    {
      "epoch": 0.0039176201372997715,
      "grad_norm": 3.3927037715911865,
      "learning_rate": 1.923210217112981e-05,
      "loss": 0.181,
      "step": 214
    },
    {
      "epoch": 0.003935926773455377,
      "grad_norm": 4.4074811935424805,
      "learning_rate": 1.9218631515885007e-05,
      "loss": 0.3954,
      "step": 215
    },
    {
      "epoch": 0.003954233409610984,
      "grad_norm": 4.823078155517578,
      "learning_rate": 1.9205048534524405e-05,
      "loss": 0.5657,
      "step": 216
    },
    {
      "epoch": 0.00397254004576659,
      "grad_norm": 5.334933757781982,
      "learning_rate": 1.9191353392552346e-05,
      "loss": 0.4124,
      "step": 217
    },
    {
      "epoch": 0.003990846681922197,
      "grad_norm": 7.765951633453369,
      "learning_rate": 1.9177546256839814e-05,
      "loss": 0.4361,
      "step": 218
    },
    {
      "epoch": 0.004009153318077803,
      "grad_norm": 5.867908000946045,
      "learning_rate": 1.9163627295622397e-05,
      "loss": 0.3579,
      "step": 219
    },
    {
      "epoch": 0.00402745995423341,
      "grad_norm": 6.852147102355957,
      "learning_rate": 1.914959667849825e-05,
      "loss": 0.7059,
      "step": 220
    },
    {
      "epoch": 0.0040457665903890164,
      "grad_norm": 5.715799808502197,
      "learning_rate": 1.913545457642601e-05,
      "loss": 0.6624,
      "step": 221
    },
    {
      "epoch": 0.004064073226544622,
      "grad_norm": 4.819512367248535,
      "learning_rate": 1.9121201161722732e-05,
      "loss": 0.3062,
      "step": 222
    },
    {
      "epoch": 0.004082379862700229,
      "grad_norm": 5.430914878845215,
      "learning_rate": 1.910683660806177e-05,
      "loss": 0.664,
      "step": 223
    },
    {
      "epoch": 0.004100686498855835,
      "grad_norm": 16.048826217651367,
      "learning_rate": 1.9092361090470688e-05,
      "loss": 0.9425,
      "step": 224
    },
    {
      "epoch": 0.004118993135011442,
      "grad_norm": 10.175516128540039,
      "learning_rate": 1.907777478532909e-05,
      "loss": 0.4463,
      "step": 225
    },
    {
      "epoch": 0.004137299771167048,
      "grad_norm": 10.81609058380127,
      "learning_rate": 1.9063077870366504e-05,
      "loss": 0.5496,
      "step": 226
    },
    {
      "epoch": 0.004155606407322655,
      "grad_norm": 3.3909196853637695,
      "learning_rate": 1.9048270524660197e-05,
      "loss": 0.2642,
      "step": 227
    },
    {
      "epoch": 0.0041739130434782605,
      "grad_norm": 7.220968723297119,
      "learning_rate": 1.903335292863301e-05,
      "loss": 0.3606,
      "step": 228
    },
    {
      "epoch": 0.004192219679633867,
      "grad_norm": 2.8659698963165283,
      "learning_rate": 1.901832526405114e-05,
      "loss": 0.2331,
      "step": 229
    },
    {
      "epoch": 0.004210526315789474,
      "grad_norm": 8.508347511291504,
      "learning_rate": 1.9003187714021936e-05,
      "loss": 0.5547,
      "step": 230
    },
    {
      "epoch": 0.00422883295194508,
      "grad_norm": 2.9436657428741455,
      "learning_rate": 1.8987940462991673e-05,
      "loss": 0.1906,
      "step": 231
    },
    {
      "epoch": 0.004247139588100687,
      "grad_norm": 8.40645980834961,
      "learning_rate": 1.8972583696743284e-05,
      "loss": 0.5893,
      "step": 232
    },
    {
      "epoch": 0.004265446224256293,
      "grad_norm": 4.141918182373047,
      "learning_rate": 1.895711760239413e-05,
      "loss": 0.5183,
      "step": 233
    },
    {
      "epoch": 0.004283752860411899,
      "grad_norm": 7.489543914794922,
      "learning_rate": 1.8941542368393683e-05,
      "loss": 0.4938,
      "step": 234
    },
    {
      "epoch": 0.0043020594965675054,
      "grad_norm": 3.483055591583252,
      "learning_rate": 1.892585818452126e-05,
      "loss": 0.2687,
      "step": 235
    },
    {
      "epoch": 0.004320366132723112,
      "grad_norm": 5.23473596572876,
      "learning_rate": 1.891006524188368e-05,
      "loss": 0.2855,
      "step": 236
    },
    {
      "epoch": 0.0043386727688787185,
      "grad_norm": 4.449317455291748,
      "learning_rate": 1.889416373291298e-05,
      "loss": 0.3082,
      "step": 237
    },
    {
      "epoch": 0.004356979405034325,
      "grad_norm": 3.1681277751922607,
      "learning_rate": 1.8878153851364013e-05,
      "loss": 0.2741,
      "step": 238
    },
    {
      "epoch": 0.004375286041189932,
      "grad_norm": 10.63996410369873,
      "learning_rate": 1.8862035792312148e-05,
      "loss": 0.6235,
      "step": 239
    },
    {
      "epoch": 0.004393592677345538,
      "grad_norm": 5.231956958770752,
      "learning_rate": 1.884580975215084e-05,
      "loss": 0.4739,
      "step": 240
    },
    {
      "epoch": 0.004411899313501144,
      "grad_norm": 3.019294023513794,
      "learning_rate": 1.8829475928589272e-05,
      "loss": 0.4061,
      "step": 241
    },
    {
      "epoch": 0.00443020594965675,
      "grad_norm": 5.6815595626831055,
      "learning_rate": 1.8813034520649923e-05,
      "loss": 0.4274,
      "step": 242
    },
    {
      "epoch": 0.004448512585812357,
      "grad_norm": 5.3184123039245605,
      "learning_rate": 1.879648572866617e-05,
      "loss": 0.3172,
      "step": 243
    },
    {
      "epoch": 0.0044668192219679635,
      "grad_norm": 1.4854844808578491,
      "learning_rate": 1.8779829754279806e-05,
      "loss": 0.1407,
      "step": 244
    },
    {
      "epoch": 0.00448512585812357,
      "grad_norm": 9.262789726257324,
      "learning_rate": 1.8763066800438638e-05,
      "loss": 0.7686,
      "step": 245
    },
    {
      "epoch": 0.0045034324942791765,
      "grad_norm": 6.8137359619140625,
      "learning_rate": 1.874619707139396e-05,
      "loss": 0.3486,
      "step": 246
    },
    {
      "epoch": 0.004521739130434782,
      "grad_norm": 11.356719017028809,
      "learning_rate": 1.8729220772698096e-05,
      "loss": 0.4303,
      "step": 247
    },
    {
      "epoch": 0.004540045766590389,
      "grad_norm": 5.79303503036499,
      "learning_rate": 1.8712138111201898e-05,
      "loss": 0.3068,
      "step": 248
    },
    {
      "epoch": 0.004558352402745995,
      "grad_norm": 1.9887948036193848,
      "learning_rate": 1.869494929505219e-05,
      "loss": 0.1456,
      "step": 249
    },
    {
      "epoch": 0.004576659038901602,
      "grad_norm": 10.916764259338379,
      "learning_rate": 1.8677654533689287e-05,
      "loss": 1.1486,
      "step": 250
    },
    {
      "epoch": 0.004594965675057208,
      "grad_norm": 4.801982879638672,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.2377,
      "step": 251
    },
    {
      "epoch": 0.004613272311212815,
      "grad_norm": 5.9616498947143555,
      "learning_rate": 1.864274801953705e-05,
      "loss": 0.213,
      "step": 252
    },
    {
      "epoch": 0.0046315789473684215,
      "grad_norm": 4.0821099281311035,
      "learning_rate": 1.8625136692072577e-05,
      "loss": 0.2355,
      "step": 253
    },
    {
      "epoch": 0.004649885583524027,
      "grad_norm": 3.7081425189971924,
      "learning_rate": 1.860742027003944e-05,
      "loss": 0.2571,
      "step": 254
    },
    {
      "epoch": 0.004668192219679634,
      "grad_norm": 8.445603370666504,
      "learning_rate": 1.8589598969306646e-05,
      "loss": 0.6145,
      "step": 255
    },
    {
      "epoch": 0.00468649885583524,
      "grad_norm": 11.145668029785156,
      "learning_rate": 1.8571673007021124e-05,
      "loss": 0.7036,
      "step": 256
    },
    {
      "epoch": 0.004704805491990847,
      "grad_norm": 3.8045926094055176,
      "learning_rate": 1.855364260160507e-05,
      "loss": 0.3276,
      "step": 257
    },
    {
      "epoch": 0.004723112128146453,
      "grad_norm": 8.324919700622559,
      "learning_rate": 1.8535507972753275e-05,
      "loss": 0.9585,
      "step": 258
    },
    {
      "epoch": 0.00474141876430206,
      "grad_norm": 9.286887168884277,
      "learning_rate": 1.851726934143048e-05,
      "loss": 0.4821,
      "step": 259
    },
    {
      "epoch": 0.0047597254004576655,
      "grad_norm": 1.0730000734329224,
      "learning_rate": 1.849892692986864e-05,
      "loss": 0.1082,
      "step": 260
    },
    {
      "epoch": 0.004778032036613272,
      "grad_norm": 4.952276229858398,
      "learning_rate": 1.848048096156426e-05,
      "loss": 0.2672,
      "step": 261
    },
    {
      "epoch": 0.004796338672768879,
      "grad_norm": 4.847801208496094,
      "learning_rate": 1.8461931661275642e-05,
      "loss": 0.4634,
      "step": 262
    },
    {
      "epoch": 0.004814645308924485,
      "grad_norm": 2.5741231441497803,
      "learning_rate": 1.8443279255020153e-05,
      "loss": 0.209,
      "step": 263
    },
    {
      "epoch": 0.004832951945080092,
      "grad_norm": 7.57988977432251,
      "learning_rate": 1.842452397007148e-05,
      "loss": 0.4493,
      "step": 264
    },
    {
      "epoch": 0.004851258581235698,
      "grad_norm": 7.51473331451416,
      "learning_rate": 1.8405666034956842e-05,
      "loss": 0.6737,
      "step": 265
    },
    {
      "epoch": 0.004869565217391304,
      "grad_norm": 13.536398887634277,
      "learning_rate": 1.8386705679454243e-05,
      "loss": 0.8438,
      "step": 266
    },
    {
      "epoch": 0.0048878718535469105,
      "grad_norm": 3.2198519706726074,
      "learning_rate": 1.836764313458962e-05,
      "loss": 0.251,
      "step": 267
    },
    {
      "epoch": 0.004906178489702517,
      "grad_norm": 5.983921527862549,
      "learning_rate": 1.8348478632634067e-05,
      "loss": 0.3383,
      "step": 268
    },
    {
      "epoch": 0.0049244851258581235,
      "grad_norm": 10.424074172973633,
      "learning_rate": 1.8329212407100996e-05,
      "loss": 1.0936,
      "step": 269
    },
    {
      "epoch": 0.00494279176201373,
      "grad_norm": 6.141103267669678,
      "learning_rate": 1.8309844692743283e-05,
      "loss": 0.5605,
      "step": 270
    },
    {
      "epoch": 0.004961098398169337,
      "grad_norm": 2.830514430999756,
      "learning_rate": 1.8290375725550417e-05,
      "loss": 0.1908,
      "step": 271
    },
    {
      "epoch": 0.004979405034324943,
      "grad_norm": 6.214183330535889,
      "learning_rate": 1.827080574274562e-05,
      "loss": 0.6819,
      "step": 272
    },
    {
      "epoch": 0.004997711670480549,
      "grad_norm": 7.017670631408691,
      "learning_rate": 1.8251134982782952e-05,
      "loss": 0.5617,
      "step": 273
    },
    {
      "epoch": 0.005016018306636155,
      "grad_norm": 15.681363105773926,
      "learning_rate": 1.8231363685344422e-05,
      "loss": 0.8103,
      "step": 274
    },
    {
      "epoch": 0.005034324942791762,
      "grad_norm": 7.182861328125,
      "learning_rate": 1.821149209133704e-05,
      "loss": 0.4888,
      "step": 275
    },
    {
      "epoch": 0.0050526315789473685,
      "grad_norm": 10.65903377532959,
      "learning_rate": 1.819152044288992e-05,
      "loss": 0.7242,
      "step": 276
    },
    {
      "epoch": 0.005070938215102975,
      "grad_norm": 4.960131645202637,
      "learning_rate": 1.8171448983351284e-05,
      "loss": 0.4659,
      "step": 277
    },
    {
      "epoch": 0.0050892448512585816,
      "grad_norm": 2.2972419261932373,
      "learning_rate": 1.815127795728554e-05,
      "loss": 0.1826,
      "step": 278
    },
    {
      "epoch": 0.005107551487414187,
      "grad_norm": 5.904301166534424,
      "learning_rate": 1.8131007610470278e-05,
      "loss": 0.3366,
      "step": 279
    },
    {
      "epoch": 0.005125858123569794,
      "grad_norm": 5.723453998565674,
      "learning_rate": 1.8110638189893267e-05,
      "loss": 0.3315,
      "step": 280
    },
    {
      "epoch": 0.0051441647597254,
      "grad_norm": 5.037336349487305,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.3074,
      "step": 281
    },
    {
      "epoch": 0.005162471395881007,
      "grad_norm": 7.443374156951904,
      "learning_rate": 1.806960312143802e-05,
      "loss": 0.4225,
      "step": 282
    },
    {
      "epoch": 0.005180778032036613,
      "grad_norm": 4.295177459716797,
      "learning_rate": 1.804893797355914e-05,
      "loss": 0.3837,
      "step": 283
    },
    {
      "epoch": 0.00519908466819222,
      "grad_norm": 3.3948891162872314,
      "learning_rate": 1.8028174751911147e-05,
      "loss": 0.349,
      "step": 284
    },
    {
      "epoch": 0.0052173913043478265,
      "grad_norm": 6.52895975112915,
      "learning_rate": 1.8007313709487334e-05,
      "loss": 0.4161,
      "step": 285
    },
    {
      "epoch": 0.005235697940503432,
      "grad_norm": 8.563170433044434,
      "learning_rate": 1.798635510047293e-05,
      "loss": 0.5702,
      "step": 286
    },
    {
      "epoch": 0.005254004576659039,
      "grad_norm": 3.7967607975006104,
      "learning_rate": 1.7965299180241963e-05,
      "loss": 0.1922,
      "step": 287
    },
    {
      "epoch": 0.005272311212814645,
      "grad_norm": 5.493015766143799,
      "learning_rate": 1.7944146205354182e-05,
      "loss": 0.6328,
      "step": 288
    },
    {
      "epoch": 0.005290617848970252,
      "grad_norm": 10.71493148803711,
      "learning_rate": 1.792289643355191e-05,
      "loss": 0.6749,
      "step": 289
    },
    {
      "epoch": 0.005308924485125858,
      "grad_norm": 7.881803512573242,
      "learning_rate": 1.7901550123756906e-05,
      "loss": 0.4414,
      "step": 290
    },
    {
      "epoch": 0.005327231121281465,
      "grad_norm": 3.409003734588623,
      "learning_rate": 1.788010753606722e-05,
      "loss": 0.217,
      "step": 291
    },
    {
      "epoch": 0.0053455377574370706,
      "grad_norm": 5.169612884521484,
      "learning_rate": 1.785856893175402e-05,
      "loss": 0.4037,
      "step": 292
    },
    {
      "epoch": 0.005363844393592677,
      "grad_norm": 5.614249229431152,
      "learning_rate": 1.78369345732584e-05,
      "loss": 0.3583,
      "step": 293
    },
    {
      "epoch": 0.005382151029748284,
      "grad_norm": 5.413997173309326,
      "learning_rate": 1.781520472418819e-05,
      "loss": 0.4758,
      "step": 294
    },
    {
      "epoch": 0.00540045766590389,
      "grad_norm": 5.485922813415527,
      "learning_rate": 1.7793379649314743e-05,
      "loss": 0.4584,
      "step": 295
    },
    {
      "epoch": 0.005418764302059497,
      "grad_norm": 6.144253253936768,
      "learning_rate": 1.777145961456971e-05,
      "loss": 0.5489,
      "step": 296
    },
    {
      "epoch": 0.005437070938215103,
      "grad_norm": 4.56196403503418,
      "learning_rate": 1.7749444887041797e-05,
      "loss": 0.4218,
      "step": 297
    },
    {
      "epoch": 0.005455377574370709,
      "grad_norm": 12.976069450378418,
      "learning_rate": 1.7727335734973512e-05,
      "loss": 0.519,
      "step": 298
    },
    {
      "epoch": 0.0054736842105263155,
      "grad_norm": 10.657815933227539,
      "learning_rate": 1.7705132427757895e-05,
      "loss": 0.8982,
      "step": 299
    },
    {
      "epoch": 0.005491990846681922,
      "grad_norm": 3.892559766769409,
      "learning_rate": 1.7682835235935236e-05,
      "loss": 0.2004,
      "step": 300
    },
    {
      "epoch": 0.005510297482837529,
      "grad_norm": 6.767181396484375,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.5441,
      "step": 301
    },
    {
      "epoch": 0.005528604118993135,
      "grad_norm": 5.241102695465088,
      "learning_rate": 1.7637960286346423e-05,
      "loss": 0.3085,
      "step": 302
    },
    {
      "epoch": 0.005546910755148742,
      "grad_norm": 4.85935640335083,
      "learning_rate": 1.761538307536737e-05,
      "loss": 0.4702,
      "step": 303
    },
    {
      "epoch": 0.005565217391304348,
      "grad_norm": 8.737496376037598,
      "learning_rate": 1.759271307334881e-05,
      "loss": 0.7425,
      "step": 304
    },
    {
      "epoch": 0.005583524027459954,
      "grad_norm": 10.740922927856445,
      "learning_rate": 1.7569950556517566e-05,
      "loss": 0.9976,
      "step": 305
    },
    {
      "epoch": 0.00560183066361556,
      "grad_norm": 7.7551422119140625,
      "learning_rate": 1.7547095802227723e-05,
      "loss": 0.5183,
      "step": 306
    },
    {
      "epoch": 0.005620137299771167,
      "grad_norm": 6.327582836151123,
      "learning_rate": 1.7524149088957244e-05,
      "loss": 0.6134,
      "step": 307
    },
    {
      "epoch": 0.0056384439359267735,
      "grad_norm": 10.561834335327148,
      "learning_rate": 1.7501110696304598e-05,
      "loss": 0.6976,
      "step": 308
    },
    {
      "epoch": 0.00565675057208238,
      "grad_norm": 4.514830112457275,
      "learning_rate": 1.747798090498532e-05,
      "loss": 0.3118,
      "step": 309
    },
    {
      "epoch": 0.005675057208237987,
      "grad_norm": 3.9753475189208984,
      "learning_rate": 1.7454759996828622e-05,
      "loss": 0.3003,
      "step": 310
    },
    {
      "epoch": 0.005693363844393592,
      "grad_norm": 7.0595011711120605,
      "learning_rate": 1.7431448254773943e-05,
      "loss": 0.5337,
      "step": 311
    },
    {
      "epoch": 0.005711670480549199,
      "grad_norm": 4.571330547332764,
      "learning_rate": 1.74080459628675e-05,
      "loss": 0.4606,
      "step": 312
    },
    {
      "epoch": 0.005729977116704805,
      "grad_norm": 7.209574222564697,
      "learning_rate": 1.7384553406258842e-05,
      "loss": 0.4431,
      "step": 313
    },
    {
      "epoch": 0.005748283752860412,
      "grad_norm": 3.1496009826660156,
      "learning_rate": 1.7360970871197347e-05,
      "loss": 0.288,
      "step": 314
    },
    {
      "epoch": 0.0057665903890160184,
      "grad_norm": 1.1498371362686157,
      "learning_rate": 1.7337298645028764e-05,
      "loss": 0.0904,
      "step": 315
    },
    {
      "epoch": 0.005784897025171625,
      "grad_norm": 4.189731121063232,
      "learning_rate": 1.7313537016191706e-05,
      "loss": 0.2142,
      "step": 316
    },
    {
      "epoch": 0.0058032036613272315,
      "grad_norm": 3.495272636413574,
      "learning_rate": 1.7289686274214116e-05,
      "loss": 0.2809,
      "step": 317
    },
    {
      "epoch": 0.005821510297482837,
      "grad_norm": 11.465441703796387,
      "learning_rate": 1.7265746709709762e-05,
      "loss": 0.9229,
      "step": 318
    },
    {
      "epoch": 0.005839816933638444,
      "grad_norm": 3.298283338546753,
      "learning_rate": 1.7241718614374678e-05,
      "loss": 0.2105,
      "step": 319
    },
    {
      "epoch": 0.00585812356979405,
      "grad_norm": 4.148833274841309,
      "learning_rate": 1.7217602280983622e-05,
      "loss": 0.2279,
      "step": 320
    },
    {
      "epoch": 0.005876430205949657,
      "grad_norm": 2.6801838874816895,
      "learning_rate": 1.7193398003386514e-05,
      "loss": 0.2207,
      "step": 321
    },
    {
      "epoch": 0.005894736842105263,
      "grad_norm": 5.371076583862305,
      "learning_rate": 1.716910607650483e-05,
      "loss": 0.6822,
      "step": 322
    },
    {
      "epoch": 0.00591304347826087,
      "grad_norm": 5.155342102050781,
      "learning_rate": 1.7144726796328034e-05,
      "loss": 0.6256,
      "step": 323
    },
    {
      "epoch": 0.005931350114416476,
      "grad_norm": 5.462609767913818,
      "learning_rate": 1.712026045990997e-05,
      "loss": 0.3712,
      "step": 324
    },
    {
      "epoch": 0.005949656750572082,
      "grad_norm": 2.920656204223633,
      "learning_rate": 1.709570736536521e-05,
      "loss": 0.2011,
      "step": 325
    },
    {
      "epoch": 0.005967963386727689,
      "grad_norm": 6.90954065322876,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.3713,
      "step": 326
    },
    {
      "epoch": 0.005986270022883295,
      "grad_norm": 4.551551342010498,
      "learning_rate": 1.7046342099635948e-05,
      "loss": 0.3052,
      "step": 327
    },
    {
      "epoch": 0.006004576659038902,
      "grad_norm": 4.894721031188965,
      "learning_rate": 1.7021530529951627e-05,
      "loss": 0.8055,
      "step": 328
    },
    {
      "epoch": 0.006022883295194508,
      "grad_norm": 4.173544406890869,
      "learning_rate": 1.6996633405133656e-05,
      "loss": 0.4642,
      "step": 329
    },
    {
      "epoch": 0.006041189931350115,
      "grad_norm": 5.222576141357422,
      "learning_rate": 1.697165102854565e-05,
      "loss": 0.3175,
      "step": 330
    },
    {
      "epoch": 0.0060594965675057205,
      "grad_norm": 6.459454536437988,
      "learning_rate": 1.6946583704589973e-05,
      "loss": 0.5357,
      "step": 331
    },
    {
      "epoch": 0.006077803203661327,
      "grad_norm": 3.6619327068328857,
      "learning_rate": 1.692143173870407e-05,
      "loss": 0.4273,
      "step": 332
    },
    {
      "epoch": 0.006096109839816934,
      "grad_norm": 5.7568840980529785,
      "learning_rate": 1.68961954373567e-05,
      "loss": 0.4508,
      "step": 333
    },
    {
      "epoch": 0.00611441647597254,
      "grad_norm": 7.910932540893555,
      "learning_rate": 1.6870875108044233e-05,
      "loss": 0.4166,
      "step": 334
    },
    {
      "epoch": 0.006132723112128147,
      "grad_norm": 4.656468391418457,
      "learning_rate": 1.684547105928689e-05,
      "loss": 0.3099,
      "step": 335
    },
    {
      "epoch": 0.006151029748283753,
      "grad_norm": 6.832411289215088,
      "learning_rate": 1.6819983600624986e-05,
      "loss": 0.4713,
      "step": 336
    },
    {
      "epoch": 0.006169336384439359,
      "grad_norm": 3.675230026245117,
      "learning_rate": 1.6794413042615168e-05,
      "loss": 0.3749,
      "step": 337
    },
    {
      "epoch": 0.0061876430205949655,
      "grad_norm": 14.497318267822266,
      "learning_rate": 1.6768759696826608e-05,
      "loss": 1.1848,
      "step": 338
    },
    {
      "epoch": 0.006205949656750572,
      "grad_norm": 8.537349700927734,
      "learning_rate": 1.6743023875837233e-05,
      "loss": 0.4933,
      "step": 339
    },
    {
      "epoch": 0.0062242562929061785,
      "grad_norm": 4.796643257141113,
      "learning_rate": 1.6717205893229904e-05,
      "loss": 0.4819,
      "step": 340
    },
    {
      "epoch": 0.006242562929061785,
      "grad_norm": 5.899059772491455,
      "learning_rate": 1.6691306063588583e-05,
      "loss": 0.6742,
      "step": 341
    },
    {
      "epoch": 0.006260869565217392,
      "grad_norm": 8.666280746459961,
      "learning_rate": 1.6665324702494524e-05,
      "loss": 0.5037,
      "step": 342
    },
    {
      "epoch": 0.006279176201372997,
      "grad_norm": 14.649405479431152,
      "learning_rate": 1.6639262126522417e-05,
      "loss": 0.7326,
      "step": 343
    },
    {
      "epoch": 0.006297482837528604,
      "grad_norm": 6.6728034019470215,
      "learning_rate": 1.661311865323652e-05,
      "loss": 0.4329,
      "step": 344
    },
    {
      "epoch": 0.00631578947368421,
      "grad_norm": 3.9211013317108154,
      "learning_rate": 1.6586894601186804e-05,
      "loss": 0.4379,
      "step": 345
    },
    {
      "epoch": 0.006334096109839817,
      "grad_norm": 5.018696308135986,
      "learning_rate": 1.6560590289905074e-05,
      "loss": 0.3585,
      "step": 346
    },
    {
      "epoch": 0.0063524027459954235,
      "grad_norm": 10.415823936462402,
      "learning_rate": 1.6534206039901057e-05,
      "loss": 0.5712,
      "step": 347
    },
    {
      "epoch": 0.00637070938215103,
      "grad_norm": 3.3000543117523193,
      "learning_rate": 1.650774217265851e-05,
      "loss": 0.262,
      "step": 348
    },
    {
      "epoch": 0.0063890160183066366,
      "grad_norm": 5.603179454803467,
      "learning_rate": 1.6481199010631312e-05,
      "loss": 0.2821,
      "step": 349
    },
    {
      "epoch": 0.006407322654462242,
      "grad_norm": 10.997979164123535,
      "learning_rate": 1.645457687723951e-05,
      "loss": 0.8257,
      "step": 350
    },
    {
      "epoch": 0.006425629290617849,
      "grad_norm": 9.430671691894531,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 0.6242,
      "step": 351
    },
    {
      "epoch": 0.006443935926773455,
      "grad_norm": 4.974905014038086,
      "learning_rate": 1.6401096994849558e-05,
      "loss": 0.3398,
      "step": 352
    },
    {
      "epoch": 0.006462242562929062,
      "grad_norm": 6.002130031585693,
      "learning_rate": 1.63742398974869e-05,
      "loss": 0.4058,
      "step": 353
    },
    {
      "epoch": 0.006480549199084668,
      "grad_norm": 4.933552265167236,
      "learning_rate": 1.6347305132022677e-05,
      "loss": 0.4209,
      "step": 354
    },
    {
      "epoch": 0.006498855835240275,
      "grad_norm": 4.84297513961792,
      "learning_rate": 1.632029302664851e-05,
      "loss": 0.3322,
      "step": 355
    },
    {
      "epoch": 0.006517162471395881,
      "grad_norm": 5.379430294036865,
      "learning_rate": 1.6293203910498375e-05,
      "loss": 0.4546,
      "step": 356
    },
    {
      "epoch": 0.006535469107551487,
      "grad_norm": 3.9145123958587646,
      "learning_rate": 1.6266038113644605e-05,
      "loss": 0.3207,
      "step": 357
    },
    {
      "epoch": 0.006553775743707094,
      "grad_norm": 6.354018211364746,
      "learning_rate": 1.6238795967093865e-05,
      "loss": 0.4729,
      "step": 358
    },
    {
      "epoch": 0.0065720823798627,
      "grad_norm": 9.218265533447266,
      "learning_rate": 1.6211477802783105e-05,
      "loss": 0.6496,
      "step": 359
    },
    {
      "epoch": 0.006590389016018307,
      "grad_norm": 12.858467102050781,
      "learning_rate": 1.6184083953575543e-05,
      "loss": 0.4073,
      "step": 360
    },
    {
      "epoch": 0.006608695652173913,
      "grad_norm": 8.530418395996094,
      "learning_rate": 1.6156614753256583e-05,
      "loss": 0.5826,
      "step": 361
    },
    {
      "epoch": 0.00662700228832952,
      "grad_norm": 8.6444673538208,
      "learning_rate": 1.6129070536529767e-05,
      "loss": 0.6264,
      "step": 362
    },
    {
      "epoch": 0.0066453089244851255,
      "grad_norm": 9.804875373840332,
      "learning_rate": 1.610145163901268e-05,
      "loss": 0.4984,
      "step": 363
    },
    {
      "epoch": 0.006663615560640732,
      "grad_norm": 7.5225324630737305,
      "learning_rate": 1.607375839723287e-05,
      "loss": 0.4179,
      "step": 364
    },
    {
      "epoch": 0.006681922196796339,
      "grad_norm": 6.881463527679443,
      "learning_rate": 1.6045991148623752e-05,
      "loss": 0.4956,
      "step": 365
    },
    {
      "epoch": 0.006700228832951945,
      "grad_norm": 6.102224349975586,
      "learning_rate": 1.6018150231520486e-05,
      "loss": 0.4461,
      "step": 366
    },
    {
      "epoch": 0.006718535469107552,
      "grad_norm": 11.278656005859375,
      "learning_rate": 1.599023598515586e-05,
      "loss": 0.8761,
      "step": 367
    },
    {
      "epoch": 0.006736842105263158,
      "grad_norm": 5.988121032714844,
      "learning_rate": 1.5962248749656158e-05,
      "loss": 0.2178,
      "step": 368
    },
    {
      "epoch": 0.006755148741418764,
      "grad_norm": 0.9311094880104065,
      "learning_rate": 1.5934188866037017e-05,
      "loss": 0.0839,
      "step": 369
    },
    {
      "epoch": 0.0067734553775743705,
      "grad_norm": 5.130373001098633,
      "learning_rate": 1.5906056676199256e-05,
      "loss": 0.1672,
      "step": 370
    },
    {
      "epoch": 0.006791762013729977,
      "grad_norm": 5.0459418296813965,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 0.2407,
      "step": 371
    },
    {
      "epoch": 0.006810068649885584,
      "grad_norm": 2.93607497215271,
      "learning_rate": 1.584957674987216e-05,
      "loss": 0.1925,
      "step": 372
    },
    {
      "epoch": 0.00682837528604119,
      "grad_norm": 12.006101608276367,
      "learning_rate": 1.5821229701572897e-05,
      "loss": 0.8912,
      "step": 373
    },
    {
      "epoch": 0.006846681922196797,
      "grad_norm": 15.027911186218262,
      "learning_rate": 1.5792811723426787e-05,
      "loss": 0.4139,
      "step": 374
    },
    {
      "epoch": 0.006864988558352402,
      "grad_norm": 5.793044090270996,
      "learning_rate": 1.5764323161697933e-05,
      "loss": 0.2851,
      "step": 375
    },
    {
      "epoch": 0.006883295194508009,
      "grad_norm": 8.51421070098877,
      "learning_rate": 1.573576436351046e-05,
      "loss": 0.5144,
      "step": 376
    },
    {
      "epoch": 0.006901601830663615,
      "grad_norm": 9.018112182617188,
      "learning_rate": 1.570713567684432e-05,
      "loss": 0.4011,
      "step": 377
    },
    {
      "epoch": 0.006919908466819222,
      "grad_norm": 12.444863319396973,
      "learning_rate": 1.5678437450531014e-05,
      "loss": 0.8046,
      "step": 378
    },
    {
      "epoch": 0.0069382151029748285,
      "grad_norm": 7.43892765045166,
      "learning_rate": 1.564967003424938e-05,
      "loss": 0.4985,
      "step": 379
    },
    {
      "epoch": 0.006956521739130435,
      "grad_norm": 3.2682807445526123,
      "learning_rate": 1.5620833778521306e-05,
      "loss": 0.3606,
      "step": 380
    },
    {
      "epoch": 0.006974828375286042,
      "grad_norm": 6.130132675170898,
      "learning_rate": 1.5591929034707468e-05,
      "loss": 0.5154,
      "step": 381
    },
    {
      "epoch": 0.006993135011441647,
      "grad_norm": 5.139265060424805,
      "learning_rate": 1.556295615500305e-05,
      "loss": 0.3442,
      "step": 382
    },
    {
      "epoch": 0.007011441647597254,
      "grad_norm": 5.858025074005127,
      "learning_rate": 1.553391549243344e-05,
      "loss": 0.316,
      "step": 383
    },
    {
      "epoch": 0.00702974828375286,
      "grad_norm": 5.114577293395996,
      "learning_rate": 1.5504807400849957e-05,
      "loss": 0.3712,
      "step": 384
    },
    {
      "epoch": 0.007048054919908467,
      "grad_norm": 11.650129318237305,
      "learning_rate": 1.5475632234925505e-05,
      "loss": 0.2964,
      "step": 385
    },
    {
      "epoch": 0.007066361556064073,
      "grad_norm": 4.760644435882568,
      "learning_rate": 1.5446390350150272e-05,
      "loss": 0.4069,
      "step": 386
    },
    {
      "epoch": 0.00708466819221968,
      "grad_norm": 4.894204139709473,
      "learning_rate": 1.54170821028274e-05,
      "loss": 0.4782,
      "step": 387
    },
    {
      "epoch": 0.007102974828375286,
      "grad_norm": 4.8495001792907715,
      "learning_rate": 1.5387707850068633e-05,
      "loss": 0.3845,
      "step": 388
    },
    {
      "epoch": 0.007121281464530892,
      "grad_norm": 3.118293285369873,
      "learning_rate": 1.5358267949789968e-05,
      "loss": 0.17,
      "step": 389
    },
    {
      "epoch": 0.007139588100686499,
      "grad_norm": 4.566890239715576,
      "learning_rate": 1.53287627607073e-05,
      "loss": 0.207,
      "step": 390
    },
    {
      "epoch": 0.007157894736842105,
      "grad_norm": 9.8530855178833,
      "learning_rate": 1.529919264233205e-05,
      "loss": 0.5756,
      "step": 391
    },
    {
      "epoch": 0.007176201372997712,
      "grad_norm": 3.8654870986938477,
      "learning_rate": 1.5269557954966777e-05,
      "loss": 0.2833,
      "step": 392
    },
    {
      "epoch": 0.007194508009153318,
      "grad_norm": 8.076715469360352,
      "learning_rate": 1.5239859059700794e-05,
      "loss": 0.434,
      "step": 393
    },
    {
      "epoch": 0.007212814645308925,
      "grad_norm": 2.7720115184783936,
      "learning_rate": 1.5210096318405768e-05,
      "loss": 0.1749,
      "step": 394
    },
    {
      "epoch": 0.007231121281464531,
      "grad_norm": 3.135864496231079,
      "learning_rate": 1.5180270093731305e-05,
      "loss": 0.0707,
      "step": 395
    },
    {
      "epoch": 0.007249427917620137,
      "grad_norm": 7.354649543762207,
      "learning_rate": 1.5150380749100545e-05,
      "loss": 0.5681,
      "step": 396
    },
    {
      "epoch": 0.007267734553775744,
      "grad_norm": 5.514926433563232,
      "learning_rate": 1.5120428648705716e-05,
      "loss": 0.2812,
      "step": 397
    },
    {
      "epoch": 0.00728604118993135,
      "grad_norm": 4.094933986663818,
      "learning_rate": 1.5090414157503715e-05,
      "loss": 0.189,
      "step": 398
    },
    {
      "epoch": 0.007304347826086957,
      "grad_norm": 1.2968307733535767,
      "learning_rate": 1.5060337641211637e-05,
      "loss": 0.1388,
      "step": 399
    },
    {
      "epoch": 0.007322654462242563,
      "grad_norm": 9.890887260437012,
      "learning_rate": 1.5030199466302354e-05,
      "loss": 0.4099,
      "step": 400
    },
    {
      "epoch": 0.007340961098398169,
      "grad_norm": 7.428032875061035,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.5361,
      "step": 401
    },
    {
      "epoch": 0.0073592677345537755,
      "grad_norm": 6.782509803771973,
      "learning_rate": 1.4969739610275556e-05,
      "loss": 0.3347,
      "step": 402
    },
    {
      "epoch": 0.007377574370709382,
      "grad_norm": 8.611628532409668,
      "learning_rate": 1.493941866584231e-05,
      "loss": 0.6473,
      "step": 403
    },
    {
      "epoch": 0.007395881006864989,
      "grad_norm": 11.101713180541992,
      "learning_rate": 1.490903753615141e-05,
      "loss": 0.5101,
      "step": 404
    },
    {
      "epoch": 0.007414187643020595,
      "grad_norm": 2.3792383670806885,
      "learning_rate": 1.4878596591387329e-05,
      "loss": 0.2176,
      "step": 405
    },
    {
      "epoch": 0.007432494279176202,
      "grad_norm": 3.647657871246338,
      "learning_rate": 1.4848096202463373e-05,
      "loss": 0.1662,
      "step": 406
    },
    {
      "epoch": 0.007450800915331807,
      "grad_norm": 2.6126086711883545,
      "learning_rate": 1.4817536741017153e-05,
      "loss": 0.223,
      "step": 407
    },
    {
      "epoch": 0.007469107551487414,
      "grad_norm": 7.960146903991699,
      "learning_rate": 1.478691857940607e-05,
      "loss": 0.4825,
      "step": 408
    },
    {
      "epoch": 0.0074874141876430204,
      "grad_norm": 2.9891467094421387,
      "learning_rate": 1.4756242090702756e-05,
      "loss": 0.2072,
      "step": 409
    },
    {
      "epoch": 0.007505720823798627,
      "grad_norm": 7.48441743850708,
      "learning_rate": 1.4725507648690542e-05,
      "loss": 0.6046,
      "step": 410
    },
    {
      "epoch": 0.0075240274599542335,
      "grad_norm": 4.050131320953369,
      "learning_rate": 1.469471562785891e-05,
      "loss": 0.343,
      "step": 411
    },
    {
      "epoch": 0.00754233409610984,
      "grad_norm": 7.872916221618652,
      "learning_rate": 1.4663866403398915e-05,
      "loss": 0.6128,
      "step": 412
    },
    {
      "epoch": 0.007560640732265447,
      "grad_norm": 4.682326793670654,
      "learning_rate": 1.463296035119862e-05,
      "loss": 0.3317,
      "step": 413
    },
    {
      "epoch": 0.007578947368421052,
      "grad_norm": 4.932008266448975,
      "learning_rate": 1.4601997847838518e-05,
      "loss": 0.2332,
      "step": 414
    },
    {
      "epoch": 0.007597254004576659,
      "grad_norm": 7.188138008117676,
      "learning_rate": 1.4570979270586944e-05,
      "loss": 0.5127,
      "step": 415
    },
    {
      "epoch": 0.007615560640732265,
      "grad_norm": 6.72917366027832,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 0.2865,
      "step": 416
    },
    {
      "epoch": 0.007633867276887872,
      "grad_norm": 6.936765670776367,
      "learning_rate": 1.4508775406894308e-05,
      "loss": 0.4261,
      "step": 417
    },
    {
      "epoch": 0.0076521739130434785,
      "grad_norm": 6.002128601074219,
      "learning_rate": 1.4477590878387697e-05,
      "loss": 0.3089,
      "step": 418
    },
    {
      "epoch": 0.007670480549199085,
      "grad_norm": 3.699465036392212,
      "learning_rate": 1.4446351791849276e-05,
      "loss": 0.2833,
      "step": 419
    },
    {
      "epoch": 0.007688787185354691,
      "grad_norm": 5.764729022979736,
      "learning_rate": 1.4415058527917454e-05,
      "loss": 0.3462,
      "step": 420
    },
    {
      "epoch": 0.007707093821510297,
      "grad_norm": 2.2470004558563232,
      "learning_rate": 1.4383711467890776e-05,
      "loss": 0.1425,
      "step": 421
    },
    {
      "epoch": 0.007725400457665904,
      "grad_norm": 3.7865355014801025,
      "learning_rate": 1.4352310993723277e-05,
      "loss": 0.1958,
      "step": 422
    },
    {
      "epoch": 0.00774370709382151,
      "grad_norm": 5.835235595703125,
      "learning_rate": 1.4320857488019826e-05,
      "loss": 0.3167,
      "step": 423
    },
    {
      "epoch": 0.007762013729977117,
      "grad_norm": 4.612227439880371,
      "learning_rate": 1.4289351334031461e-05,
      "loss": 0.259,
      "step": 424
    },
    {
      "epoch": 0.007780320366132723,
      "grad_norm": 5.527163982391357,
      "learning_rate": 1.4257792915650728e-05,
      "loss": 0.3875,
      "step": 425
    },
    {
      "epoch": 0.00779862700228833,
      "grad_norm": 1.5776690244674683,
      "learning_rate": 1.4226182617406996e-05,
      "loss": 0.1545,
      "step": 426
    },
    {
      "epoch": 0.007816933638443936,
      "grad_norm": 5.22613525390625,
      "learning_rate": 1.4194520824461773e-05,
      "loss": 0.2102,
      "step": 427
    },
    {
      "epoch": 0.007835240274599543,
      "grad_norm": 8.434508323669434,
      "learning_rate": 1.4162807922604014e-05,
      "loss": 0.4331,
      "step": 428
    },
    {
      "epoch": 0.00785354691075515,
      "grad_norm": 6.714724063873291,
      "learning_rate": 1.413104429824542e-05,
      "loss": 0.4454,
      "step": 429
    },
    {
      "epoch": 0.007871853546910754,
      "grad_norm": 5.015308856964111,
      "learning_rate": 1.4099230338415728e-05,
      "loss": 0.1637,
      "step": 430
    },
    {
      "epoch": 0.007890160183066361,
      "grad_norm": 5.9333367347717285,
      "learning_rate": 1.4067366430758004e-05,
      "loss": 0.3313,
      "step": 431
    },
    {
      "epoch": 0.007908466819221967,
      "grad_norm": 2.7909371852874756,
      "learning_rate": 1.4035452963523903e-05,
      "loss": 0.1595,
      "step": 432
    },
    {
      "epoch": 0.007926773455377574,
      "grad_norm": 12.026054382324219,
      "learning_rate": 1.4003490325568953e-05,
      "loss": 0.5666,
      "step": 433
    },
    {
      "epoch": 0.00794508009153318,
      "grad_norm": 9.2349271774292,
      "learning_rate": 1.3971478906347806e-05,
      "loss": 0.4163,
      "step": 434
    },
    {
      "epoch": 0.007963386727688787,
      "grad_norm": 4.861546516418457,
      "learning_rate": 1.3939419095909513e-05,
      "loss": 0.2674,
      "step": 435
    },
    {
      "epoch": 0.007981693363844394,
      "grad_norm": 6.877256393432617,
      "learning_rate": 1.3907311284892737e-05,
      "loss": 0.2162,
      "step": 436
    },
    {
      "epoch": 0.008,
      "grad_norm": 9.61545467376709,
      "learning_rate": 1.3875155864521031e-05,
      "loss": 0.5273,
      "step": 437
    },
    {
      "epoch": 0.008018306636155607,
      "grad_norm": 4.861330509185791,
      "learning_rate": 1.3842953226598036e-05,
      "loss": 0.1181,
      "step": 438
    },
    {
      "epoch": 0.008036613272311213,
      "grad_norm": 12.819157600402832,
      "learning_rate": 1.3810703763502744e-05,
      "loss": 0.9099,
      "step": 439
    },
    {
      "epoch": 0.00805491990846682,
      "grad_norm": 11.029340744018555,
      "learning_rate": 1.3778407868184674e-05,
      "loss": 0.5066,
      "step": 440
    },
    {
      "epoch": 0.008073226544622426,
      "grad_norm": 12.776625633239746,
      "learning_rate": 1.3746065934159123e-05,
      "loss": 0.8391,
      "step": 441
    },
    {
      "epoch": 0.008091533180778033,
      "grad_norm": 10.414067268371582,
      "learning_rate": 1.371367835550235e-05,
      "loss": 0.579,
      "step": 442
    },
    {
      "epoch": 0.008109839816933638,
      "grad_norm": 2.4149062633514404,
      "learning_rate": 1.3681245526846782e-05,
      "loss": 0.2106,
      "step": 443
    },
    {
      "epoch": 0.008128146453089244,
      "grad_norm": 3.4909825325012207,
      "learning_rate": 1.3648767843376196e-05,
      "loss": 0.2114,
      "step": 444
    },
    {
      "epoch": 0.00814645308924485,
      "grad_norm": 2.9647722244262695,
      "learning_rate": 1.3616245700820922e-05,
      "loss": 0.2153,
      "step": 445
    },
    {
      "epoch": 0.008164759725400457,
      "grad_norm": 2.9645230770111084,
      "learning_rate": 1.3583679495453e-05,
      "loss": 0.1417,
      "step": 446
    },
    {
      "epoch": 0.008183066361556064,
      "grad_norm": 5.128696918487549,
      "learning_rate": 1.3551069624081372e-05,
      "loss": 0.2745,
      "step": 447
    },
    {
      "epoch": 0.00820137299771167,
      "grad_norm": 8.987137794494629,
      "learning_rate": 1.3518416484047018e-05,
      "loss": 0.5339,
      "step": 448
    },
    {
      "epoch": 0.008219679633867277,
      "grad_norm": 9.38960075378418,
      "learning_rate": 1.3485720473218153e-05,
      "loss": 0.4518,
      "step": 449
    },
    {
      "epoch": 0.008237986270022883,
      "grad_norm": 2.744861364364624,
      "learning_rate": 1.3452981989985347e-05,
      "loss": 0.153,
      "step": 450
    },
    {
      "epoch": 0.00825629290617849,
      "grad_norm": 6.3267998695373535,
      "learning_rate": 1.342020143325669e-05,
      "loss": 0.4077,
      "step": 451
    },
    {
      "epoch": 0.008274599542334097,
      "grad_norm": 9.530548095703125,
      "learning_rate": 1.3387379202452917e-05,
      "loss": 0.5712,
      "step": 452
    },
    {
      "epoch": 0.008292906178489703,
      "grad_norm": 1.2639689445495605,
      "learning_rate": 1.3354515697502552e-05,
      "loss": 0.0901,
      "step": 453
    },
    {
      "epoch": 0.00831121281464531,
      "grad_norm": 10.240217208862305,
      "learning_rate": 1.3321611318837033e-05,
      "loss": 0.4039,
      "step": 454
    },
    {
      "epoch": 0.008329519450800914,
      "grad_norm": 7.718742370605469,
      "learning_rate": 1.3288666467385834e-05,
      "loss": 0.5048,
      "step": 455
    },
    {
      "epoch": 0.008347826086956521,
      "grad_norm": 1.411474347114563,
      "learning_rate": 1.3255681544571568e-05,
      "loss": 0.1291,
      "step": 456
    },
    {
      "epoch": 0.008366132723112128,
      "grad_norm": 3.7180333137512207,
      "learning_rate": 1.3222656952305113e-05,
      "loss": 0.1669,
      "step": 457
    },
    {
      "epoch": 0.008384439359267734,
      "grad_norm": 4.709652423858643,
      "learning_rate": 1.3189593092980701e-05,
      "loss": 0.2531,
      "step": 458
    },
    {
      "epoch": 0.00840274599542334,
      "grad_norm": 2.9395062923431396,
      "learning_rate": 1.3156490369471026e-05,
      "loss": 0.1908,
      "step": 459
    },
    {
      "epoch": 0.008421052631578947,
      "grad_norm": 1.6285467147827148,
      "learning_rate": 1.3123349185122328e-05,
      "loss": 0.1829,
      "step": 460
    },
    {
      "epoch": 0.008439359267734554,
      "grad_norm": 2.359907627105713,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.078,
      "step": 461
    },
    {
      "epoch": 0.00845766590389016,
      "grad_norm": 5.721633434295654,
      "learning_rate": 1.3056953049631059e-05,
      "loss": 0.5493,
      "step": 462
    },
    {
      "epoch": 0.008475972540045767,
      "grad_norm": 7.441493034362793,
      "learning_rate": 1.3023698907504447e-05,
      "loss": 0.3625,
      "step": 463
    },
    {
      "epoch": 0.008494279176201373,
      "grad_norm": 8.423327445983887,
      "learning_rate": 1.2990407922560869e-05,
      "loss": 0.841,
      "step": 464
    },
    {
      "epoch": 0.00851258581235698,
      "grad_norm": 13.417081832885742,
      "learning_rate": 1.2957080500440469e-05,
      "loss": 0.3607,
      "step": 465
    },
    {
      "epoch": 0.008530892448512586,
      "grad_norm": 19.080902099609375,
      "learning_rate": 1.2923717047227368e-05,
      "loss": 0.8309,
      "step": 466
    },
    {
      "epoch": 0.008549199084668193,
      "grad_norm": 5.821935176849365,
      "learning_rate": 1.2890317969444716e-05,
      "loss": 0.1997,
      "step": 467
    },
    {
      "epoch": 0.008567505720823798,
      "grad_norm": 6.380805492401123,
      "learning_rate": 1.2856883674049736e-05,
      "loss": 0.3366,
      "step": 468
    },
    {
      "epoch": 0.008585812356979404,
      "grad_norm": 4.085707664489746,
      "learning_rate": 1.2823414568428767e-05,
      "loss": 0.2064,
      "step": 469
    },
    {
      "epoch": 0.008604118993135011,
      "grad_norm": 1.9464811086654663,
      "learning_rate": 1.2789911060392295e-05,
      "loss": 0.1908,
      "step": 470
    },
    {
      "epoch": 0.008622425629290617,
      "grad_norm": 3.138880729675293,
      "learning_rate": 1.2756373558169992e-05,
      "loss": 0.1714,
      "step": 471
    },
    {
      "epoch": 0.008640732265446224,
      "grad_norm": 4.658113479614258,
      "learning_rate": 1.2722802470405744e-05,
      "loss": 0.3097,
      "step": 472
    },
    {
      "epoch": 0.00865903890160183,
      "grad_norm": 4.06643533706665,
      "learning_rate": 1.2689198206152657e-05,
      "loss": 0.2363,
      "step": 473
    },
    {
      "epoch": 0.008677345537757437,
      "grad_norm": 7.818174839019775,
      "learning_rate": 1.265556117486809e-05,
      "loss": 0.3208,
      "step": 474
    },
    {
      "epoch": 0.008695652173913044,
      "grad_norm": 5.261242866516113,
      "learning_rate": 1.2621891786408648e-05,
      "loss": 0.3357,
      "step": 475
    },
    {
      "epoch": 0.00871395881006865,
      "grad_norm": 6.41402006149292,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 0.2269,
      "step": 476
    },
    {
      "epoch": 0.008732265446224257,
      "grad_norm": 3.909989833831787,
      "learning_rate": 1.2554457579357906e-05,
      "loss": 0.2456,
      "step": 477
    },
    {
      "epoch": 0.008750572082379863,
      "grad_norm": 6.242023944854736,
      "learning_rate": 1.252069358243114e-05,
      "loss": 0.4592,
      "step": 478
    },
    {
      "epoch": 0.00876887871853547,
      "grad_norm": 11.872090339660645,
      "learning_rate": 1.2486898871648552e-05,
      "loss": 0.6729,
      "step": 479
    },
    {
      "epoch": 0.008787185354691076,
      "grad_norm": 5.960324287414551,
      "learning_rate": 1.2453073858788027e-05,
      "loss": 0.4166,
      "step": 480
    },
    {
      "epoch": 0.008805491990846681,
      "grad_norm": 1.5860638618469238,
      "learning_rate": 1.2419218955996677e-05,
      "loss": 0.1463,
      "step": 481
    },
    {
      "epoch": 0.008823798627002288,
      "grad_norm": 2.930913209915161,
      "learning_rate": 1.238533457578581e-05,
      "loss": 0.1741,
      "step": 482
    },
    {
      "epoch": 0.008842105263157894,
      "grad_norm": 7.377974033355713,
      "learning_rate": 1.23514211310259e-05,
      "loss": 0.4284,
      "step": 483
    },
    {
      "epoch": 0.0088604118993135,
      "grad_norm": 3.70469069480896,
      "learning_rate": 1.2317479034941572e-05,
      "loss": 0.2728,
      "step": 484
    },
    {
      "epoch": 0.008878718535469107,
      "grad_norm": 5.633390426635742,
      "learning_rate": 1.2283508701106559e-05,
      "loss": 0.3421,
      "step": 485
    },
    {
      "epoch": 0.008897025171624714,
      "grad_norm": 0.9088148474693298,
      "learning_rate": 1.2249510543438652e-05,
      "loss": 0.1071,
      "step": 486
    },
    {
      "epoch": 0.00891533180778032,
      "grad_norm": 10.652883529663086,
      "learning_rate": 1.2215484976194675e-05,
      "loss": 0.4244,
      "step": 487
    },
    {
      "epoch": 0.008933638443935927,
      "grad_norm": 5.040556907653809,
      "learning_rate": 1.2181432413965428e-05,
      "loss": 0.2866,
      "step": 488
    },
    {
      "epoch": 0.008951945080091533,
      "grad_norm": 5.78645133972168,
      "learning_rate": 1.2147353271670634e-05,
      "loss": 0.2152,
      "step": 489
    },
    {
      "epoch": 0.00897025171624714,
      "grad_norm": 8.46097183227539,
      "learning_rate": 1.211324796455389e-05,
      "loss": 0.6545,
      "step": 490
    },
    {
      "epoch": 0.008988558352402747,
      "grad_norm": 5.180769443511963,
      "learning_rate": 1.2079116908177592e-05,
      "loss": 0.234,
      "step": 491
    },
    {
      "epoch": 0.009006864988558353,
      "grad_norm": 5.95151424407959,
      "learning_rate": 1.2044960518417902e-05,
      "loss": 0.2579,
      "step": 492
    },
    {
      "epoch": 0.00902517162471396,
      "grad_norm": 7.002593994140625,
      "learning_rate": 1.2010779211459649e-05,
      "loss": 0.5269,
      "step": 493
    },
    {
      "epoch": 0.009043478260869564,
      "grad_norm": 7.671480178833008,
      "learning_rate": 1.1976573403791263e-05,
      "loss": 0.4812,
      "step": 494
    },
    {
      "epoch": 0.009061784897025171,
      "grad_norm": 7.638554096221924,
      "learning_rate": 1.194234351219972e-05,
      "loss": 0.3599,
      "step": 495
    },
    {
      "epoch": 0.009080091533180778,
      "grad_norm": 6.174886226654053,
      "learning_rate": 1.190808995376545e-05,
      "loss": 0.275,
      "step": 496
    },
    {
      "epoch": 0.009098398169336384,
      "grad_norm": 3.23649525642395,
      "learning_rate": 1.187381314585725e-05,
      "loss": 0.179,
      "step": 497
    },
    {
      "epoch": 0.00911670480549199,
      "grad_norm": 6.402664661407471,
      "learning_rate": 1.1839513506127202e-05,
      "loss": 0.3912,
      "step": 498
    },
    {
      "epoch": 0.009135011441647597,
      "grad_norm": 7.799668788909912,
      "learning_rate": 1.1805191452505602e-05,
      "loss": 0.3725,
      "step": 499
    },
    {
      "epoch": 0.009153318077803204,
      "grad_norm": 13.813634872436523,
      "learning_rate": 1.1770847403195836e-05,
      "loss": 0.7032,
      "step": 500
    },
    {
      "epoch": 0.00917162471395881,
      "grad_norm": 2.5564143657684326,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.129,
      "step": 501
    },
    {
      "epoch": 0.009189931350114417,
      "grad_norm": 4.1047821044921875,
      "learning_rate": 1.1702094991660326e-05,
      "loss": 0.2844,
      "step": 502
    },
    {
      "epoch": 0.009208237986270023,
      "grad_norm": 8.63683032989502,
      "learning_rate": 1.1667687467161025e-05,
      "loss": 0.4529,
      "step": 503
    },
    {
      "epoch": 0.00922654462242563,
      "grad_norm": 5.256021976470947,
      "learning_rate": 1.1633259622416224e-05,
      "loss": 0.4106,
      "step": 504
    },
    {
      "epoch": 0.009244851258581236,
      "grad_norm": 4.8041486740112305,
      "learning_rate": 1.159881187691835e-05,
      "loss": 0.2019,
      "step": 505
    },
    {
      "epoch": 0.009263157894736843,
      "grad_norm": 5.618558406829834,
      "learning_rate": 1.156434465040231e-05,
      "loss": 0.3514,
      "step": 506
    },
    {
      "epoch": 0.009281464530892448,
      "grad_norm": 3.8470354080200195,
      "learning_rate": 1.1529858362840383e-05,
      "loss": 0.2582,
      "step": 507
    },
    {
      "epoch": 0.009299771167048054,
      "grad_norm": 4.10807991027832,
      "learning_rate": 1.1495353434437098e-05,
      "loss": 0.2775,
      "step": 508
    },
    {
      "epoch": 0.00931807780320366,
      "grad_norm": 3.2404677867889404,
      "learning_rate": 1.1460830285624119e-05,
      "loss": 0.0828,
      "step": 509
    },
    {
      "epoch": 0.009336384439359267,
      "grad_norm": 4.644530296325684,
      "learning_rate": 1.1426289337055119e-05,
      "loss": 0.2594,
      "step": 510
    },
    {
      "epoch": 0.009354691075514874,
      "grad_norm": 4.381004810333252,
      "learning_rate": 1.1391731009600655e-05,
      "loss": 0.3301,
      "step": 511
    },
    {
      "epoch": 0.00937299771167048,
      "grad_norm": 2.8095078468322754,
      "learning_rate": 1.1357155724343046e-05,
      "loss": 0.1854,
      "step": 512
    },
    {
      "epoch": 0.009391304347826087,
      "grad_norm": 4.646561622619629,
      "learning_rate": 1.1322563902571227e-05,
      "loss": 0.2856,
      "step": 513
    },
    {
      "epoch": 0.009409610983981694,
      "grad_norm": 2.435589075088501,
      "learning_rate": 1.128795596577563e-05,
      "loss": 0.158,
      "step": 514
    },
    {
      "epoch": 0.0094279176201373,
      "grad_norm": 5.535358905792236,
      "learning_rate": 1.1253332335643043e-05,
      "loss": 0.4016,
      "step": 515
    },
    {
      "epoch": 0.009446224256292907,
      "grad_norm": 2.8487491607666016,
      "learning_rate": 1.1218693434051475e-05,
      "loss": 0.1856,
      "step": 516
    },
    {
      "epoch": 0.009464530892448513,
      "grad_norm": 6.440304756164551,
      "learning_rate": 1.1184039683065014e-05,
      "loss": 0.4992,
      "step": 517
    },
    {
      "epoch": 0.00948283752860412,
      "grad_norm": 4.590803146362305,
      "learning_rate": 1.1149371504928667e-05,
      "loss": 0.3192,
      "step": 518
    },
    {
      "epoch": 0.009501144164759726,
      "grad_norm": 3.818535566329956,
      "learning_rate": 1.1114689322063255e-05,
      "loss": 0.2371,
      "step": 519
    },
    {
      "epoch": 0.009519450800915331,
      "grad_norm": 6.345479965209961,
      "learning_rate": 1.1079993557060228e-05,
      "loss": 0.3573,
      "step": 520
    },
    {
      "epoch": 0.009537757437070938,
      "grad_norm": 13.15767765045166,
      "learning_rate": 1.1045284632676535e-05,
      "loss": 0.4145,
      "step": 521
    },
    {
      "epoch": 0.009556064073226544,
      "grad_norm": 2.1148345470428467,
      "learning_rate": 1.1010562971829464e-05,
      "loss": 0.1284,
      "step": 522
    },
    {
      "epoch": 0.00957437070938215,
      "grad_norm": 5.396875858306885,
      "learning_rate": 1.0975828997591496e-05,
      "loss": 0.6029,
      "step": 523
    },
    {
      "epoch": 0.009592677345537757,
      "grad_norm": 5.095512866973877,
      "learning_rate": 1.0941083133185146e-05,
      "loss": 0.316,
      "step": 524
    },
    {
      "epoch": 0.009610983981693364,
      "grad_norm": 8.121919631958008,
      "learning_rate": 1.0906325801977804e-05,
      "loss": 0.3049,
      "step": 525
    },
    {
      "epoch": 0.00962929061784897,
      "grad_norm": 3.969857931137085,
      "learning_rate": 1.0871557427476585e-05,
      "loss": 0.1074,
      "step": 526
    },
    {
      "epoch": 0.009647597254004577,
      "grad_norm": 12.454261779785156,
      "learning_rate": 1.083677843332316e-05,
      "loss": 0.6658,
      "step": 527
    },
    {
      "epoch": 0.009665903890160183,
      "grad_norm": 4.157962799072266,
      "learning_rate": 1.0801989243288588e-05,
      "loss": 0.343,
      "step": 528
    },
    {
      "epoch": 0.00968421052631579,
      "grad_norm": 0.6788906455039978,
      "learning_rate": 1.0767190281268187e-05,
      "loss": 0.0834,
      "step": 529
    },
    {
      "epoch": 0.009702517162471396,
      "grad_norm": 7.104922294616699,
      "learning_rate": 1.0732381971276318e-05,
      "loss": 0.4511,
      "step": 530
    },
    {
      "epoch": 0.009720823798627003,
      "grad_norm": 10.119832038879395,
      "learning_rate": 1.0697564737441254e-05,
      "loss": 0.4547,
      "step": 531
    },
    {
      "epoch": 0.009739130434782608,
      "grad_norm": 1.6746511459350586,
      "learning_rate": 1.0662739004000005e-05,
      "loss": 0.1375,
      "step": 532
    },
    {
      "epoch": 0.009757437070938214,
      "grad_norm": 9.788320541381836,
      "learning_rate": 1.0627905195293135e-05,
      "loss": 0.4845,
      "step": 533
    },
    {
      "epoch": 0.009775743707093821,
      "grad_norm": 9.758012771606445,
      "learning_rate": 1.0593063735759619e-05,
      "loss": 0.4237,
      "step": 534
    },
    {
      "epoch": 0.009794050343249427,
      "grad_norm": 2.4389655590057373,
      "learning_rate": 1.055821504993164e-05,
      "loss": 0.1618,
      "step": 535
    },
    {
      "epoch": 0.009812356979405034,
      "grad_norm": 10.49396800994873,
      "learning_rate": 1.0523359562429441e-05,
      "loss": 0.5892,
      "step": 536
    },
    {
      "epoch": 0.00983066361556064,
      "grad_norm": 5.466527462005615,
      "learning_rate": 1.0488497697956134e-05,
      "loss": 0.2774,
      "step": 537
    },
    {
      "epoch": 0.009848970251716247,
      "grad_norm": 2.714876890182495,
      "learning_rate": 1.0453629881292537e-05,
      "loss": 0.1092,
      "step": 538
    },
    {
      "epoch": 0.009867276887871854,
      "grad_norm": 4.077674865722656,
      "learning_rate": 1.0418756537291996e-05,
      "loss": 0.2116,
      "step": 539
    },
    {
      "epoch": 0.00988558352402746,
      "grad_norm": 3.0669403076171875,
      "learning_rate": 1.03838780908752e-05,
      "loss": 0.1422,
      "step": 540
    },
    {
      "epoch": 0.009903890160183067,
      "grad_norm": 8.089181900024414,
      "learning_rate": 1.0348994967025012e-05,
      "loss": 0.4507,
      "step": 541
    },
    {
      "epoch": 0.009922196796338673,
      "grad_norm": 3.917191982269287,
      "learning_rate": 1.0314107590781284e-05,
      "loss": 0.2293,
      "step": 542
    },
    {
      "epoch": 0.00994050343249428,
      "grad_norm": 3.5208070278167725,
      "learning_rate": 1.0279216387235691e-05,
      "loss": 0.201,
      "step": 543
    },
    {
      "epoch": 0.009958810068649886,
      "grad_norm": 11.041540145874023,
      "learning_rate": 1.0244321781526533e-05,
      "loss": 0.4846,
      "step": 544
    },
    {
      "epoch": 0.009977116704805491,
      "grad_norm": 5.149035930633545,
      "learning_rate": 1.0209424198833571e-05,
      "loss": 0.2859,
      "step": 545
    },
    {
      "epoch": 0.009995423340961098,
      "grad_norm": 4.3648762702941895,
      "learning_rate": 1.0174524064372837e-05,
      "loss": 0.3187,
      "step": 546
    },
    {
      "epoch": 0.010013729977116704,
      "grad_norm": 14.32425594329834,
      "learning_rate": 1.0139621803391454e-05,
      "loss": 0.7906,
      "step": 547
    },
    {
      "epoch": 0.01003203661327231,
      "grad_norm": 7.4310455322265625,
      "learning_rate": 1.010471784116246e-05,
      "loss": 0.3219,
      "step": 548
    },
    {
      "epoch": 0.010050343249427917,
      "grad_norm": 4.170273780822754,
      "learning_rate": 1.0069812602979617e-05,
      "loss": 0.23,
      "step": 549
    },
    {
      "epoch": 0.010068649885583524,
      "grad_norm": 11.858202934265137,
      "learning_rate": 1.0034906514152239e-05,
      "loss": 0.4112,
      "step": 550
    },
    {
      "epoch": 0.01008695652173913,
      "grad_norm": 7.049015998840332,
      "learning_rate": 1e-05,
      "loss": 0.2449,
      "step": 551
    },
    {
      "epoch": 0.010105263157894737,
      "grad_norm": 8.323487281799316,
      "learning_rate": 9.965093485847766e-06,
      "loss": 0.3886,
      "step": 552
    },
    {
      "epoch": 0.010123569794050344,
      "grad_norm": 1.819872498512268,
      "learning_rate": 9.930187397020385e-06,
      "loss": 0.0831,
      "step": 553
    },
    {
      "epoch": 0.01014187643020595,
      "grad_norm": 8.343167304992676,
      "learning_rate": 9.895282158837545e-06,
      "loss": 0.7165,
      "step": 554
    },
    {
      "epoch": 0.010160183066361557,
      "grad_norm": 6.083837985992432,
      "learning_rate": 9.860378196608549e-06,
      "loss": 0.3716,
      "step": 555
    },
    {
      "epoch": 0.010178489702517163,
      "grad_norm": 5.233552932739258,
      "learning_rate": 9.825475935627165e-06,
      "loss": 0.3486,
      "step": 556
    },
    {
      "epoch": 0.01019679633867277,
      "grad_norm": 2.339454174041748,
      "learning_rate": 9.790575801166432e-06,
      "loss": 0.1367,
      "step": 557
    },
    {
      "epoch": 0.010215102974828374,
      "grad_norm": 5.374737739562988,
      "learning_rate": 9.75567821847347e-06,
      "loss": 0.4209,
      "step": 558
    },
    {
      "epoch": 0.010233409610983981,
      "grad_norm": 8.515985488891602,
      "learning_rate": 9.720783612764314e-06,
      "loss": 0.3302,
      "step": 559
    },
    {
      "epoch": 0.010251716247139588,
      "grad_norm": 9.269789695739746,
      "learning_rate": 9.685892409218718e-06,
      "loss": 0.2903,
      "step": 560
    },
    {
      "epoch": 0.010270022883295194,
      "grad_norm": 3.935189723968506,
      "learning_rate": 9.651005032974994e-06,
      "loss": 0.3017,
      "step": 561
    },
    {
      "epoch": 0.0102883295194508,
      "grad_norm": 10.850844383239746,
      "learning_rate": 9.616121909124801e-06,
      "loss": 0.4161,
      "step": 562
    },
    {
      "epoch": 0.010306636155606407,
      "grad_norm": 9.301300048828125,
      "learning_rate": 9.581243462708007e-06,
      "loss": 0.5199,
      "step": 563
    },
    {
      "epoch": 0.010324942791762014,
      "grad_norm": 5.9267191886901855,
      "learning_rate": 9.546370118707463e-06,
      "loss": 0.2799,
      "step": 564
    },
    {
      "epoch": 0.01034324942791762,
      "grad_norm": 9.080803871154785,
      "learning_rate": 9.511502302043867e-06,
      "loss": 0.7154,
      "step": 565
    },
    {
      "epoch": 0.010361556064073227,
      "grad_norm": 7.055878162384033,
      "learning_rate": 9.476640437570562e-06,
      "loss": 0.304,
      "step": 566
    },
    {
      "epoch": 0.010379862700228833,
      "grad_norm": 2.7219274044036865,
      "learning_rate": 9.441784950068362e-06,
      "loss": 0.1827,
      "step": 567
    },
    {
      "epoch": 0.01039816933638444,
      "grad_norm": 2.3669350147247314,
      "learning_rate": 9.406936264240386e-06,
      "loss": 0.1686,
      "step": 568
    },
    {
      "epoch": 0.010416475972540046,
      "grad_norm": 4.634760856628418,
      "learning_rate": 9.372094804706867e-06,
      "loss": 0.2163,
      "step": 569
    },
    {
      "epoch": 0.010434782608695653,
      "grad_norm": 1.3255306482315063,
      "learning_rate": 9.337260996000002e-06,
      "loss": 0.0883,
      "step": 570
    },
    {
      "epoch": 0.010453089244851258,
      "grad_norm": 2.8958237171173096,
      "learning_rate": 9.302435262558748e-06,
      "loss": 0.1593,
      "step": 571
    },
    {
      "epoch": 0.010471395881006864,
      "grad_norm": 8.792821884155273,
      "learning_rate": 9.267618028723687e-06,
      "loss": 0.4316,
      "step": 572
    },
    {
      "epoch": 0.010489702517162471,
      "grad_norm": 11.037833213806152,
      "learning_rate": 9.232809718731815e-06,
      "loss": 0.6522,
      "step": 573
    },
    {
      "epoch": 0.010508009153318077,
      "grad_norm": 2.741342067718506,
      "learning_rate": 9.198010756711413e-06,
      "loss": 0.1564,
      "step": 574
    },
    {
      "epoch": 0.010526315789473684,
      "grad_norm": 5.826968193054199,
      "learning_rate": 9.163221566676847e-06,
      "loss": 0.294,
      "step": 575
    },
    {
      "epoch": 0.01054462242562929,
      "grad_norm": 2.7347750663757324,
      "learning_rate": 9.128442572523418e-06,
      "loss": 0.1361,
      "step": 576
    },
    {
      "epoch": 0.010562929061784897,
      "grad_norm": 8.53330135345459,
      "learning_rate": 9.093674198022201e-06,
      "loss": 0.275,
      "step": 577
    },
    {
      "epoch": 0.010581235697940504,
      "grad_norm": 3.5783607959747314,
      "learning_rate": 9.058916866814857e-06,
      "loss": 0.2126,
      "step": 578
    },
    {
      "epoch": 0.01059954233409611,
      "grad_norm": 12.933378219604492,
      "learning_rate": 9.024171002408507e-06,
      "loss": 0.5151,
      "step": 579
    },
    {
      "epoch": 0.010617848970251717,
      "grad_norm": 5.769047737121582,
      "learning_rate": 8.989437028170537e-06,
      "loss": 0.2029,
      "step": 580
    },
    {
      "epoch": 0.010636155606407323,
      "grad_norm": 10.376677513122559,
      "learning_rate": 8.954715367323468e-06,
      "loss": 0.5334,
      "step": 581
    },
    {
      "epoch": 0.01065446224256293,
      "grad_norm": 11.327596664428711,
      "learning_rate": 8.920006442939772e-06,
      "loss": 0.3964,
      "step": 582
    },
    {
      "epoch": 0.010672768878718536,
      "grad_norm": 7.816294193267822,
      "learning_rate": 8.885310677936746e-06,
      "loss": 0.3265,
      "step": 583
    },
    {
      "epoch": 0.010691075514874141,
      "grad_norm": 14.427319526672363,
      "learning_rate": 8.850628495071336e-06,
      "loss": 0.6769,
      "step": 584
    },
    {
      "epoch": 0.010709382151029748,
      "grad_norm": 2.7052078247070312,
      "learning_rate": 8.815960316934991e-06,
      "loss": 0.1255,
      "step": 585
    },
    {
      "epoch": 0.010727688787185354,
      "grad_norm": 7.072470188140869,
      "learning_rate": 8.781306565948528e-06,
      "loss": 0.4344,
      "step": 586
    },
    {
      "epoch": 0.01074599542334096,
      "grad_norm": 15.953747749328613,
      "learning_rate": 8.746667664356957e-06,
      "loss": 0.6664,
      "step": 587
    },
    {
      "epoch": 0.010764302059496567,
      "grad_norm": 4.023654937744141,
      "learning_rate": 8.712044034224374e-06,
      "loss": 0.1642,
      "step": 588
    },
    {
      "epoch": 0.010782608695652174,
      "grad_norm": 9.880155563354492,
      "learning_rate": 8.677436097428775e-06,
      "loss": 0.5635,
      "step": 589
    },
    {
      "epoch": 0.01080091533180778,
      "grad_norm": 7.572893142700195,
      "learning_rate": 8.642844275656957e-06,
      "loss": 0.6194,
      "step": 590
    },
    {
      "epoch": 0.010819221967963387,
      "grad_norm": 4.445793151855469,
      "learning_rate": 8.60826899039935e-06,
      "loss": 0.2096,
      "step": 591
    },
    {
      "epoch": 0.010837528604118993,
      "grad_norm": 9.583978652954102,
      "learning_rate": 8.573710662944884e-06,
      "loss": 0.2836,
      "step": 592
    },
    {
      "epoch": 0.0108558352402746,
      "grad_norm": 8.179230690002441,
      "learning_rate": 8.539169714375885e-06,
      "loss": 0.3744,
      "step": 593
    },
    {
      "epoch": 0.010874141876430207,
      "grad_norm": 10.7511568069458,
      "learning_rate": 8.504646565562907e-06,
      "loss": 0.5062,
      "step": 594
    },
    {
      "epoch": 0.010892448512585813,
      "grad_norm": 2.8209664821624756,
      "learning_rate": 8.47014163715962e-06,
      "loss": 0.1292,
      "step": 595
    },
    {
      "epoch": 0.010910755148741418,
      "grad_norm": 5.161312580108643,
      "learning_rate": 8.43565534959769e-06,
      "loss": 0.2225,
      "step": 596
    },
    {
      "epoch": 0.010929061784897024,
      "grad_norm": 6.31545877456665,
      "learning_rate": 8.401188123081653e-06,
      "loss": 0.4825,
      "step": 597
    },
    {
      "epoch": 0.010947368421052631,
      "grad_norm": 7.428010940551758,
      "learning_rate": 8.366740377583781e-06,
      "loss": 0.28,
      "step": 598
    },
    {
      "epoch": 0.010965675057208238,
      "grad_norm": 10.632474899291992,
      "learning_rate": 8.332312532838978e-06,
      "loss": 0.4979,
      "step": 599
    },
    {
      "epoch": 0.010983981693363844,
      "grad_norm": 5.0757880210876465,
      "learning_rate": 8.297905008339677e-06,
      "loss": 0.4114,
      "step": 600
    },
    {
      "epoch": 0.01100228832951945,
      "grad_norm": 10.615399360656738,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.4028,
      "step": 601
    },
    {
      "epoch": 0.011020594965675057,
      "grad_norm": 3.5888783931732178,
      "learning_rate": 8.22915259680417e-06,
      "loss": 0.1922,
      "step": 602
    },
    {
      "epoch": 0.011038901601830664,
      "grad_norm": 7.211901664733887,
      "learning_rate": 8.194808547494401e-06,
      "loss": 0.4079,
      "step": 603
    },
    {
      "epoch": 0.01105720823798627,
      "grad_norm": 2.416304111480713,
      "learning_rate": 8.1604864938728e-06,
      "loss": 0.1201,
      "step": 604
    },
    {
      "epoch": 0.011075514874141877,
      "grad_norm": 4.646475791931152,
      "learning_rate": 8.126186854142752e-06,
      "loss": 0.2859,
      "step": 605
    },
    {
      "epoch": 0.011093821510297483,
      "grad_norm": 3.045999765396118,
      "learning_rate": 8.091910046234552e-06,
      "loss": 0.1901,
      "step": 606
    },
    {
      "epoch": 0.01111212814645309,
      "grad_norm": 0.9611418843269348,
      "learning_rate": 8.057656487800283e-06,
      "loss": 0.1063,
      "step": 607
    },
    {
      "epoch": 0.011130434782608696,
      "grad_norm": 2.64760160446167,
      "learning_rate": 8.023426596208739e-06,
      "loss": 0.1395,
      "step": 608
    },
    {
      "epoch": 0.011148741418764301,
      "grad_norm": 12.34020709991455,
      "learning_rate": 7.989220788540356e-06,
      "loss": 0.4971,
      "step": 609
    },
    {
      "epoch": 0.011167048054919908,
      "grad_norm": 5.713735103607178,
      "learning_rate": 7.955039481582098e-06,
      "loss": 0.1714,
      "step": 610
    },
    {
      "epoch": 0.011185354691075514,
      "grad_norm": 18.283546447753906,
      "learning_rate": 7.92088309182241e-06,
      "loss": 0.6558,
      "step": 611
    },
    {
      "epoch": 0.01120366132723112,
      "grad_norm": 12.52721118927002,
      "learning_rate": 7.886752035446116e-06,
      "loss": 0.5603,
      "step": 612
    },
    {
      "epoch": 0.011221967963386727,
      "grad_norm": 4.516887664794922,
      "learning_rate": 7.852646728329368e-06,
      "loss": 0.2777,
      "step": 613
    },
    {
      "epoch": 0.011240274599542334,
      "grad_norm": 12.416860580444336,
      "learning_rate": 7.818567586034578e-06,
      "loss": 0.5139,
      "step": 614
    },
    {
      "epoch": 0.01125858123569794,
      "grad_norm": 7.2105255126953125,
      "learning_rate": 7.784515023805328e-06,
      "loss": 0.3159,
      "step": 615
    },
    {
      "epoch": 0.011276887871853547,
      "grad_norm": 5.7121100425720215,
      "learning_rate": 7.750489456561351e-06,
      "loss": 0.2603,
      "step": 616
    },
    {
      "epoch": 0.011295194508009154,
      "grad_norm": 8.093098640441895,
      "learning_rate": 7.716491298893443e-06,
      "loss": 0.4499,
      "step": 617
    },
    {
      "epoch": 0.01131350114416476,
      "grad_norm": 4.29197883605957,
      "learning_rate": 7.68252096505843e-06,
      "loss": 0.1985,
      "step": 618
    },
    {
      "epoch": 0.011331807780320367,
      "grad_norm": 2.614251136779785,
      "learning_rate": 7.6485788689741e-06,
      "loss": 0.1525,
      "step": 619
    },
    {
      "epoch": 0.011350114416475973,
      "grad_norm": 7.588433742523193,
      "learning_rate": 7.6146654242141935e-06,
      "loss": 0.5013,
      "step": 620
    },
    {
      "epoch": 0.01136842105263158,
      "grad_norm": 7.058767318725586,
      "learning_rate": 7.580781044003324e-06,
      "loss": 0.2994,
      "step": 621
    },
    {
      "epoch": 0.011386727688787185,
      "grad_norm": 4.412848949432373,
      "learning_rate": 7.546926141211975e-06,
      "loss": 0.2056,
      "step": 622
    },
    {
      "epoch": 0.011405034324942791,
      "grad_norm": 7.570004463195801,
      "learning_rate": 7.513101128351454e-06,
      "loss": 0.3943,
      "step": 623
    },
    {
      "epoch": 0.011423340961098398,
      "grad_norm": 4.8941521644592285,
      "learning_rate": 7.4793064175688635e-06,
      "loss": 0.2364,
      "step": 624
    },
    {
      "epoch": 0.011441647597254004,
      "grad_norm": 7.014512538909912,
      "learning_rate": 7.445542420642097e-06,
      "loss": 0.3111,
      "step": 625
    },
    {
      "epoch": 0.01145995423340961,
      "grad_norm": 2.618640661239624,
      "learning_rate": 7.411809548974792e-06,
      "loss": 0.1614,
      "step": 626
    },
    {
      "epoch": 0.011478260869565217,
      "grad_norm": 5.4425740242004395,
      "learning_rate": 7.378108213591355e-06,
      "loss": 0.3619,
      "step": 627
    },
    {
      "epoch": 0.011496567505720824,
      "grad_norm": 3.9418482780456543,
      "learning_rate": 7.344438825131912e-06,
      "loss": 0.1967,
      "step": 628
    },
    {
      "epoch": 0.01151487414187643,
      "grad_norm": 5.235764026641846,
      "learning_rate": 7.310801793847344e-06,
      "loss": 0.2332,
      "step": 629
    },
    {
      "epoch": 0.011533180778032037,
      "grad_norm": 3.303947687149048,
      "learning_rate": 7.277197529594257e-06,
      "loss": 0.1708,
      "step": 630
    },
    {
      "epoch": 0.011551487414187643,
      "grad_norm": 13.982658386230469,
      "learning_rate": 7.243626441830009e-06,
      "loss": 0.62,
      "step": 631
    },
    {
      "epoch": 0.01156979405034325,
      "grad_norm": 5.784631252288818,
      "learning_rate": 7.210088939607709e-06,
      "loss": 0.3401,
      "step": 632
    },
    {
      "epoch": 0.011588100686498857,
      "grad_norm": 5.933520793914795,
      "learning_rate": 7.176585431571235e-06,
      "loss": 0.3589,
      "step": 633
    },
    {
      "epoch": 0.011606407322654463,
      "grad_norm": 12.28591537475586,
      "learning_rate": 7.143116325950266e-06,
      "loss": 0.6409,
      "step": 634
    },
    {
      "epoch": 0.011624713958810068,
      "grad_norm": 1.4337087869644165,
      "learning_rate": 7.109682030555283e-06,
      "loss": 0.0955,
      "step": 635
    },
    {
      "epoch": 0.011643020594965674,
      "grad_norm": 3.1845862865448,
      "learning_rate": 7.076282952772634e-06,
      "loss": 0.2039,
      "step": 636
    },
    {
      "epoch": 0.011661327231121281,
      "grad_norm": 3.719296455383301,
      "learning_rate": 7.042919499559538e-06,
      "loss": 0.1679,
      "step": 637
    },
    {
      "epoch": 0.011679633867276887,
      "grad_norm": 8.414267539978027,
      "learning_rate": 7.009592077439135e-06,
      "loss": 0.4527,
      "step": 638
    },
    {
      "epoch": 0.011697940503432494,
      "grad_norm": 4.962318420410156,
      "learning_rate": 6.976301092495556e-06,
      "loss": 0.257,
      "step": 639
    },
    {
      "epoch": 0.0117162471395881,
      "grad_norm": 5.534420967102051,
      "learning_rate": 6.943046950368944e-06,
      "loss": 0.1673,
      "step": 640
    },
    {
      "epoch": 0.011734553775743707,
      "grad_norm": 7.917168140411377,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.2011,
      "step": 641
    },
    {
      "epoch": 0.011752860411899314,
      "grad_norm": 6.127618312835693,
      "learning_rate": 6.876650814877675e-06,
      "loss": 0.2565,
      "step": 642
    },
    {
      "epoch": 0.01177116704805492,
      "grad_norm": 1.4675836563110352,
      "learning_rate": 6.843509630528977e-06,
      "loss": 0.0989,
      "step": 643
    },
    {
      "epoch": 0.011789473684210527,
      "grad_norm": 10.80344009399414,
      "learning_rate": 6.8104069070193e-06,
      "loss": 0.5205,
      "step": 644
    },
    {
      "epoch": 0.011807780320366133,
      "grad_norm": 5.06679630279541,
      "learning_rate": 6.777343047694891e-06,
      "loss": 0.2618,
      "step": 645
    },
    {
      "epoch": 0.01182608695652174,
      "grad_norm": 18.99068832397461,
      "learning_rate": 6.744318455428436e-06,
      "loss": 1.1171,
      "step": 646
    },
    {
      "epoch": 0.011844393592677346,
      "grad_norm": 4.126583576202393,
      "learning_rate": 6.711333532614168e-06,
      "loss": 0.2228,
      "step": 647
    },
    {
      "epoch": 0.011862700228832951,
      "grad_norm": 2.62414288520813,
      "learning_rate": 6.67838868116297e-06,
      "loss": 0.1465,
      "step": 648
    },
    {
      "epoch": 0.011881006864988558,
      "grad_norm": 14.556570053100586,
      "learning_rate": 6.645484302497452e-06,
      "loss": 0.4093,
      "step": 649
    },
    {
      "epoch": 0.011899313501144164,
      "grad_norm": 3.685044288635254,
      "learning_rate": 6.612620797547087e-06,
      "loss": 0.2372,
      "step": 650
    },
    {
      "epoch": 0.01191762013729977,
      "grad_norm": 10.447038650512695,
      "learning_rate": 6.579798566743314e-06,
      "loss": 0.3067,
      "step": 651
    },
    {
      "epoch": 0.011935926773455377,
      "grad_norm": 7.927669525146484,
      "learning_rate": 6.547018010014654e-06,
      "loss": 0.5193,
      "step": 652
    },
    {
      "epoch": 0.011954233409610984,
      "grad_norm": 18.05657386779785,
      "learning_rate": 6.5142795267818505e-06,
      "loss": 0.9131,
      "step": 653
    },
    {
      "epoch": 0.01197254004576659,
      "grad_norm": 3.7571918964385986,
      "learning_rate": 6.481583515952983e-06,
      "loss": 0.2054,
      "step": 654
    },
    {
      "epoch": 0.011990846681922197,
      "grad_norm": 5.364101409912109,
      "learning_rate": 6.448930375918632e-06,
      "loss": 0.1835,
      "step": 655
    },
    {
      "epoch": 0.012009153318077804,
      "grad_norm": 9.139827728271484,
      "learning_rate": 6.4163205045469975e-06,
      "loss": 0.76,
      "step": 656
    },
    {
      "epoch": 0.01202745995423341,
      "grad_norm": 1.704281210899353,
      "learning_rate": 6.383754299179079e-06,
      "loss": 0.1308,
      "step": 657
    },
    {
      "epoch": 0.012045766590389017,
      "grad_norm": 7.968606472015381,
      "learning_rate": 6.351232156623803e-06,
      "loss": 0.6407,
      "step": 658
    },
    {
      "epoch": 0.012064073226544623,
      "grad_norm": 17.709571838378906,
      "learning_rate": 6.318754473153221e-06,
      "loss": 0.9402,
      "step": 659
    },
    {
      "epoch": 0.01208237986270023,
      "grad_norm": 11.887946128845215,
      "learning_rate": 6.286321644497655e-06,
      "loss": 0.5389,
      "step": 660
    },
    {
      "epoch": 0.012100686498855834,
      "grad_norm": 4.999807357788086,
      "learning_rate": 6.25393406584088e-06,
      "loss": 0.3483,
      "step": 661
    },
    {
      "epoch": 0.012118993135011441,
      "grad_norm": 7.6661906242370605,
      "learning_rate": 6.22159213181533e-06,
      "loss": 0.4436,
      "step": 662
    },
    {
      "epoch": 0.012137299771167048,
      "grad_norm": 7.397403240203857,
      "learning_rate": 6.18929623649726e-06,
      "loss": 0.3315,
      "step": 663
    },
    {
      "epoch": 0.012155606407322654,
      "grad_norm": 8.096484184265137,
      "learning_rate": 6.157046773401964e-06,
      "loss": 0.3889,
      "step": 664
    },
    {
      "epoch": 0.01217391304347826,
      "grad_norm": 4.6776251792907715,
      "learning_rate": 6.124844135478971e-06,
      "loss": 0.2081,
      "step": 665
    },
    {
      "epoch": 0.012192219679633867,
      "grad_norm": 8.530952453613281,
      "learning_rate": 6.092688715107265e-06,
      "loss": 0.4058,
      "step": 666
    },
    {
      "epoch": 0.012210526315789474,
      "grad_norm": 9.915332794189453,
      "learning_rate": 6.06058090409049e-06,
      "loss": 0.4034,
      "step": 667
    },
    {
      "epoch": 0.01222883295194508,
      "grad_norm": 16.072465896606445,
      "learning_rate": 6.028521093652195e-06,
      "loss": 0.5987,
      "step": 668
    },
    {
      "epoch": 0.012247139588100687,
      "grad_norm": 4.9345479011535645,
      "learning_rate": 5.996509674431053e-06,
      "loss": 0.223,
      "step": 669
    },
    {
      "epoch": 0.012265446224256293,
      "grad_norm": 4.367536544799805,
      "learning_rate": 5.9645470364761e-06,
      "loss": 0.357,
      "step": 670
    },
    {
      "epoch": 0.0122837528604119,
      "grad_norm": 5.174431324005127,
      "learning_rate": 5.932633569242e-06,
      "loss": 0.3424,
      "step": 671
    },
    {
      "epoch": 0.012302059496567506,
      "grad_norm": 5.893050193786621,
      "learning_rate": 5.900769661584273e-06,
      "loss": 0.2669,
      "step": 672
    },
    {
      "epoch": 0.012320366132723111,
      "grad_norm": 10.575066566467285,
      "learning_rate": 5.868955701754584e-06,
      "loss": 0.3644,
      "step": 673
    },
    {
      "epoch": 0.012338672768878718,
      "grad_norm": 5.934246063232422,
      "learning_rate": 5.83719207739599e-06,
      "loss": 0.3487,
      "step": 674
    },
    {
      "epoch": 0.012356979405034324,
      "grad_norm": 2.32735538482666,
      "learning_rate": 5.8054791755382286e-06,
      "loss": 0.1629,
      "step": 675
    },
    {
      "epoch": 0.012375286041189931,
      "grad_norm": 10.4105224609375,
      "learning_rate": 5.773817382593008e-06,
      "loss": 0.577,
      "step": 676
    },
    {
      "epoch": 0.012393592677345537,
      "grad_norm": 8.903114318847656,
      "learning_rate": 5.742207084349274e-06,
      "loss": 0.5795,
      "step": 677
    },
    {
      "epoch": 0.012411899313501144,
      "grad_norm": 1.7812707424163818,
      "learning_rate": 5.710648665968543e-06,
      "loss": 0.0989,
      "step": 678
    },
    {
      "epoch": 0.01243020594965675,
      "grad_norm": 5.7843017578125,
      "learning_rate": 5.679142511980176e-06,
      "loss": 0.331,
      "step": 679
    },
    {
      "epoch": 0.012448512585812357,
      "grad_norm": 10.94100570678711,
      "learning_rate": 5.647689006276727e-06,
      "loss": 0.6601,
      "step": 680
    },
    {
      "epoch": 0.012466819221967964,
      "grad_norm": 3.161262273788452,
      "learning_rate": 5.616288532109225e-06,
      "loss": 0.1807,
      "step": 681
    },
    {
      "epoch": 0.01248512585812357,
      "grad_norm": 5.56771183013916,
      "learning_rate": 5.584941472082549e-06,
      "loss": 0.3165,
      "step": 682
    },
    {
      "epoch": 0.012503432494279177,
      "grad_norm": 2.321823835372925,
      "learning_rate": 5.553648208150728e-06,
      "loss": 0.0925,
      "step": 683
    },
    {
      "epoch": 0.012521739130434783,
      "grad_norm": 5.972341537475586,
      "learning_rate": 5.522409121612304e-06,
      "loss": 0.301,
      "step": 684
    },
    {
      "epoch": 0.01254004576659039,
      "grad_norm": 14.91727352142334,
      "learning_rate": 5.491224593105695e-06,
      "loss": 0.8664,
      "step": 685
    },
    {
      "epoch": 0.012558352402745995,
      "grad_norm": 4.990755558013916,
      "learning_rate": 5.460095002604533e-06,
      "loss": 0.2396,
      "step": 686
    },
    {
      "epoch": 0.012576659038901601,
      "grad_norm": 3.9700381755828857,
      "learning_rate": 5.429020729413062e-06,
      "loss": 0.2423,
      "step": 687
    },
    {
      "epoch": 0.012594965675057208,
      "grad_norm": 19.159067153930664,
      "learning_rate": 5.398002152161484e-06,
      "loss": 0.8266,
      "step": 688
    },
    {
      "epoch": 0.012613272311212814,
      "grad_norm": 6.696817398071289,
      "learning_rate": 5.367039648801386e-06,
      "loss": 0.3413,
      "step": 689
    },
    {
      "epoch": 0.01263157894736842,
      "grad_norm": 7.948276042938232,
      "learning_rate": 5.336133596601089e-06,
      "loss": 0.2789,
      "step": 690
    },
    {
      "epoch": 0.012649885583524027,
      "grad_norm": 4.675381660461426,
      "learning_rate": 5.305284372141095e-06,
      "loss": 0.2435,
      "step": 691
    },
    {
      "epoch": 0.012668192219679634,
      "grad_norm": 8.05580997467041,
      "learning_rate": 5.274492351309462e-06,
      "loss": 0.3781,
      "step": 692
    },
    {
      "epoch": 0.01268649885583524,
      "grad_norm": 8.211023330688477,
      "learning_rate": 5.243757909297247e-06,
      "loss": 0.1573,
      "step": 693
    },
    {
      "epoch": 0.012704805491990847,
      "grad_norm": 2.2971174716949463,
      "learning_rate": 5.213081420593933e-06,
      "loss": 0.1092,
      "step": 694
    },
    {
      "epoch": 0.012723112128146453,
      "grad_norm": 8.110382080078125,
      "learning_rate": 5.1824632589828465e-06,
      "loss": 0.3773,
      "step": 695
    },
    {
      "epoch": 0.01274141876430206,
      "grad_norm": 4.399291038513184,
      "learning_rate": 5.151903797536631e-06,
      "loss": 0.1347,
      "step": 696
    },
    {
      "epoch": 0.012759725400457667,
      "grad_norm": 6.359978199005127,
      "learning_rate": 5.121403408612672e-06,
      "loss": 0.1828,
      "step": 697
    },
    {
      "epoch": 0.012778032036613273,
      "grad_norm": 3.2302865982055664,
      "learning_rate": 5.090962463848592e-06,
      "loss": 0.1979,
      "step": 698
    },
    {
      "epoch": 0.012796338672768878,
      "grad_norm": 7.771028518676758,
      "learning_rate": 5.060581334157693e-06,
      "loss": 0.4509,
      "step": 699
    },
    {
      "epoch": 0.012814645308924484,
      "grad_norm": 4.159087181091309,
      "learning_rate": 5.030260389724447e-06,
      "loss": 0.1944,
      "step": 700
    },
    {
      "epoch": 0.012832951945080091,
      "grad_norm": 5.334461688995361,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.1792,
      "step": 701
    },
    {
      "epoch": 0.012851258581235698,
      "grad_norm": 2.703159809112549,
      "learning_rate": 4.96980053369765e-06,
      "loss": 0.1473,
      "step": 702
    },
    {
      "epoch": 0.012869565217391304,
      "grad_norm": 9.341408729553223,
      "learning_rate": 4.939662358788364e-06,
      "loss": 0.4148,
      "step": 703
    },
    {
      "epoch": 0.01288787185354691,
      "grad_norm": 5.660780906677246,
      "learning_rate": 4.909585842496287e-06,
      "loss": 0.4013,
      "step": 704
    },
    {
      "epoch": 0.012906178489702517,
      "grad_norm": 14.756714820861816,
      "learning_rate": 4.879571351294287e-06,
      "loss": 0.5132,
      "step": 705
    },
    {
      "epoch": 0.012924485125858124,
      "grad_norm": 3.988741159439087,
      "learning_rate": 4.849619250899458e-06,
      "loss": 0.2555,
      "step": 706
    },
    {
      "epoch": 0.01294279176201373,
      "grad_norm": 4.656280040740967,
      "learning_rate": 4.8197299062687e-06,
      "loss": 0.2938,
      "step": 707
    },
    {
      "epoch": 0.012961098398169337,
      "grad_norm": 7.464823246002197,
      "learning_rate": 4.78990368159424e-06,
      "loss": 0.4899,
      "step": 708
    },
    {
      "epoch": 0.012979405034324943,
      "grad_norm": 1.0329279899597168,
      "learning_rate": 4.76014094029921e-06,
      "loss": 0.0952,
      "step": 709
    },
    {
      "epoch": 0.01299771167048055,
      "grad_norm": 4.358473300933838,
      "learning_rate": 4.7304420450332244e-06,
      "loss": 0.2474,
      "step": 710
    },
    {
      "epoch": 0.013016018306636156,
      "grad_norm": 4.45465612411499,
      "learning_rate": 4.700807357667953e-06,
      "loss": 0.2223,
      "step": 711
    },
    {
      "epoch": 0.013034324942791761,
      "grad_norm": 2.9852383136749268,
      "learning_rate": 4.671237239292699e-06,
      "loss": 0.2014,
      "step": 712
    },
    {
      "epoch": 0.013052631578947368,
      "grad_norm": 8.365155220031738,
      "learning_rate": 4.641732050210032e-06,
      "loss": 0.4184,
      "step": 713
    },
    {
      "epoch": 0.013070938215102974,
      "grad_norm": 6.666082382202148,
      "learning_rate": 4.612292149931369e-06,
      "loss": 0.3087,
      "step": 714
    },
    {
      "epoch": 0.01308924485125858,
      "grad_norm": 3.9271891117095947,
      "learning_rate": 4.582917897172603e-06,
      "loss": 0.1685,
      "step": 715
    },
    {
      "epoch": 0.013107551487414187,
      "grad_norm": 4.308122634887695,
      "learning_rate": 4.5536096498497295e-06,
      "loss": 0.1991,
      "step": 716
    },
    {
      "epoch": 0.013125858123569794,
      "grad_norm": 4.309776306152344,
      "learning_rate": 4.524367765074499e-06,
      "loss": 0.263,
      "step": 717
    },
    {
      "epoch": 0.0131441647597254,
      "grad_norm": 3.0838520526885986,
      "learning_rate": 4.495192599150045e-06,
      "loss": 0.1774,
      "step": 718
    },
    {
      "epoch": 0.013162471395881007,
      "grad_norm": 7.4343695640563965,
      "learning_rate": 4.46608450756656e-06,
      "loss": 0.4462,
      "step": 719
    },
    {
      "epoch": 0.013180778032036614,
      "grad_norm": 1.0408004522323608,
      "learning_rate": 4.437043844996952e-06,
      "loss": 0.0999,
      "step": 720
    },
    {
      "epoch": 0.01319908466819222,
      "grad_norm": 0.9866322875022888,
      "learning_rate": 4.408070965292534e-06,
      "loss": 0.0948,
      "step": 721
    },
    {
      "epoch": 0.013217391304347827,
      "grad_norm": 1.4886623620986938,
      "learning_rate": 4.379166221478697e-06,
      "loss": 0.1075,
      "step": 722
    },
    {
      "epoch": 0.013235697940503433,
      "grad_norm": 14.501533508300781,
      "learning_rate": 4.350329965750622e-06,
      "loss": 0.4004,
      "step": 723
    },
    {
      "epoch": 0.01325400457665904,
      "grad_norm": 5.454537391662598,
      "learning_rate": 4.321562549468991e-06,
      "loss": 0.219,
      "step": 724
    },
    {
      "epoch": 0.013272311212814645,
      "grad_norm": 7.190273761749268,
      "learning_rate": 4.292864323155684e-06,
      "loss": 0.4322,
      "step": 725
    },
    {
      "epoch": 0.013290617848970251,
      "grad_norm": 13.830406188964844,
      "learning_rate": 4.264235636489542e-06,
      "loss": 0.63,
      "step": 726
    },
    {
      "epoch": 0.013308924485125858,
      "grad_norm": 2.6480069160461426,
      "learning_rate": 4.235676838302069e-06,
      "loss": 0.1278,
      "step": 727
    },
    {
      "epoch": 0.013327231121281464,
      "grad_norm": 13.472162246704102,
      "learning_rate": 4.207188276573214e-06,
      "loss": 0.5058,
      "step": 728
    },
    {
      "epoch": 0.01334553775743707,
      "grad_norm": 2.51118540763855,
      "learning_rate": 4.178770298427107e-06,
      "loss": 0.149,
      "step": 729
    },
    {
      "epoch": 0.013363844393592677,
      "grad_norm": 2.3269636631011963,
      "learning_rate": 4.150423250127846e-06,
      "loss": 0.1129,
      "step": 730
    },
    {
      "epoch": 0.013382151029748284,
      "grad_norm": 1.8809363842010498,
      "learning_rate": 4.12214747707527e-06,
      "loss": 0.1052,
      "step": 731
    },
    {
      "epoch": 0.01340045766590389,
      "grad_norm": 2.9234514236450195,
      "learning_rate": 4.093943323800746e-06,
      "loss": 0.1489,
      "step": 732
    },
    {
      "epoch": 0.013418764302059497,
      "grad_norm": 8.009096145629883,
      "learning_rate": 4.065811133962987e-06,
      "loss": 0.2956,
      "step": 733
    },
    {
      "epoch": 0.013437070938215103,
      "grad_norm": 16.991966247558594,
      "learning_rate": 4.037751250343841e-06,
      "loss": 1.0203,
      "step": 734
    },
    {
      "epoch": 0.01345537757437071,
      "grad_norm": 8.988302230834961,
      "learning_rate": 4.009764014844143e-06,
      "loss": 0.4323,
      "step": 735
    },
    {
      "epoch": 0.013473684210526317,
      "grad_norm": 2.6384384632110596,
      "learning_rate": 3.981849768479516e-06,
      "loss": 0.1374,
      "step": 736
    },
    {
      "epoch": 0.013491990846681921,
      "grad_norm": 6.374378681182861,
      "learning_rate": 3.954008851376252e-06,
      "loss": 0.2113,
      "step": 737
    },
    {
      "epoch": 0.013510297482837528,
      "grad_norm": 8.83869457244873,
      "learning_rate": 3.9262416027671354e-06,
      "loss": 0.4818,
      "step": 738
    },
    {
      "epoch": 0.013528604118993134,
      "grad_norm": 6.571460723876953,
      "learning_rate": 3.898548360987325e-06,
      "loss": 0.1552,
      "step": 739
    },
    {
      "epoch": 0.013546910755148741,
      "grad_norm": 18.608381271362305,
      "learning_rate": 3.8709294634702374e-06,
      "loss": 0.3997,
      "step": 740
    },
    {
      "epoch": 0.013565217391304348,
      "grad_norm": 9.389320373535156,
      "learning_rate": 3.8433852467434175e-06,
      "loss": 0.4528,
      "step": 741
    },
    {
      "epoch": 0.013583524027459954,
      "grad_norm": 8.535320281982422,
      "learning_rate": 3.81591604642446e-06,
      "loss": 0.4345,
      "step": 742
    },
    {
      "epoch": 0.01360183066361556,
      "grad_norm": 1.2252271175384521,
      "learning_rate": 3.7885221972168974e-06,
      "loss": 0.1047,
      "step": 743
    },
    {
      "epoch": 0.013620137299771167,
      "grad_norm": 5.069352149963379,
      "learning_rate": 3.7612040329061405e-06,
      "loss": 0.0966,
      "step": 744
    },
    {
      "epoch": 0.013638443935926774,
      "grad_norm": 9.090739250183105,
      "learning_rate": 3.7339618863553983e-06,
      "loss": 0.3199,
      "step": 745
    },
    {
      "epoch": 0.01365675057208238,
      "grad_norm": 6.387343406677246,
      "learning_rate": 3.7067960895016277e-06,
      "loss": 0.2299,
      "step": 746
    },
    {
      "epoch": 0.013675057208237987,
      "grad_norm": 7.78866720199585,
      "learning_rate": 3.679706973351491e-06,
      "loss": 0.4284,
      "step": 747
    },
    {
      "epoch": 0.013693363844393593,
      "grad_norm": 5.307522773742676,
      "learning_rate": 3.6526948679773256e-06,
      "loss": 0.1915,
      "step": 748
    },
    {
      "epoch": 0.0137116704805492,
      "grad_norm": 8.756507873535156,
      "learning_rate": 3.625760102513103e-06,
      "loss": 0.2633,
      "step": 749
    },
    {
      "epoch": 0.013729977116704805,
      "grad_norm": 3.3090696334838867,
      "learning_rate": 3.598903005150444e-06,
      "loss": 0.1686,
      "step": 750
    },
    {
      "epoch": 0.013748283752860411,
      "grad_norm": 11.242887496948242,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 0.9433,
      "step": 751
    },
    {
      "epoch": 0.013766590389016018,
      "grad_norm": 7.644121170043945,
      "learning_rate": 3.545423122760493e-06,
      "loss": 0.3955,
      "step": 752
    },
    {
      "epoch": 0.013784897025171624,
      "grad_norm": 9.132620811462402,
      "learning_rate": 3.5188009893686916e-06,
      "loss": 0.5613,
      "step": 753
    },
    {
      "epoch": 0.01380320366132723,
      "grad_norm": 8.751481056213379,
      "learning_rate": 3.492257827341492e-06,
      "loss": 0.5503,
      "step": 754
    },
    {
      "epoch": 0.013821510297482837,
      "grad_norm": 7.459012985229492,
      "learning_rate": 3.4657939600989453e-06,
      "loss": 0.2748,
      "step": 755
    },
    {
      "epoch": 0.013839816933638444,
      "grad_norm": 7.091245651245117,
      "learning_rate": 3.4394097100949286e-06,
      "loss": 0.3414,
      "step": 756
    },
    {
      "epoch": 0.01385812356979405,
      "grad_norm": 6.701774597167969,
      "learning_rate": 3.4131053988131947e-06,
      "loss": 0.2112,
      "step": 757
    },
    {
      "epoch": 0.013876430205949657,
      "grad_norm": 5.4992475509643555,
      "learning_rate": 3.3868813467634833e-06,
      "loss": 0.2098,
      "step": 758
    },
    {
      "epoch": 0.013894736842105264,
      "grad_norm": 4.821681976318359,
      "learning_rate": 3.360737873477584e-06,
      "loss": 0.2729,
      "step": 759
    },
    {
      "epoch": 0.01391304347826087,
      "grad_norm": 4.573626518249512,
      "learning_rate": 3.3346752975054763e-06,
      "loss": 0.3077,
      "step": 760
    },
    {
      "epoch": 0.013931350114416477,
      "grad_norm": 5.970454692840576,
      "learning_rate": 3.308693936411421e-06,
      "loss": 0.2764,
      "step": 761
    },
    {
      "epoch": 0.013949656750572083,
      "grad_norm": 15.272013664245605,
      "learning_rate": 3.2827941067700996e-06,
      "loss": 0.3641,
      "step": 762
    },
    {
      "epoch": 0.013967963386727688,
      "grad_norm": 7.55459451675415,
      "learning_rate": 3.2569761241627694e-06,
      "loss": 0.1848,
      "step": 763
    },
    {
      "epoch": 0.013986270022883295,
      "grad_norm": 4.483034133911133,
      "learning_rate": 3.2312403031733943e-06,
      "loss": 0.221,
      "step": 764
    },
    {
      "epoch": 0.014004576659038901,
      "grad_norm": 4.3897624015808105,
      "learning_rate": 3.2055869573848374e-06,
      "loss": 0.175,
      "step": 765
    },
    {
      "epoch": 0.014022883295194508,
      "grad_norm": 8.360304832458496,
      "learning_rate": 3.1800163993750166e-06,
      "loss": 0.3661,
      "step": 766
    },
    {
      "epoch": 0.014041189931350114,
      "grad_norm": 5.351900100708008,
      "learning_rate": 3.1545289407131128e-06,
      "loss": 0.1164,
      "step": 767
    },
    {
      "epoch": 0.01405949656750572,
      "grad_norm": 6.663723468780518,
      "learning_rate": 3.1291248919557717e-06,
      "loss": 0.1693,
      "step": 768
    },
    {
      "epoch": 0.014077803203661327,
      "grad_norm": 2.0629947185516357,
      "learning_rate": 3.103804562643302e-06,
      "loss": 0.0827,
      "step": 769
    },
    {
      "epoch": 0.014096109839816934,
      "grad_norm": 4.476418495178223,
      "learning_rate": 3.0785682612959334e-06,
      "loss": 0.238,
      "step": 770
    },
    {
      "epoch": 0.01411441647597254,
      "grad_norm": 1.5098648071289062,
      "learning_rate": 3.0534162954100264e-06,
      "loss": 0.1068,
      "step": 771
    },
    {
      "epoch": 0.014132723112128147,
      "grad_norm": 10.149811744689941,
      "learning_rate": 3.028348971454356e-06,
      "loss": 0.3284,
      "step": 772
    },
    {
      "epoch": 0.014151029748283753,
      "grad_norm": 5.156161785125732,
      "learning_rate": 3.003366594866345e-06,
      "loss": 0.2241,
      "step": 773
    },
    {
      "epoch": 0.01416933638443936,
      "grad_norm": 3.282696008682251,
      "learning_rate": 2.978469470048376e-06,
      "loss": 0.1718,
      "step": 774
    },
    {
      "epoch": 0.014187643020594966,
      "grad_norm": 7.071142196655273,
      "learning_rate": 2.953657900364053e-06,
      "loss": 0.4103,
      "step": 775
    },
    {
      "epoch": 0.014205949656750571,
      "grad_norm": 3.6657493114471436,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.1982,
      "step": 776
    },
    {
      "epoch": 0.014224256292906178,
      "grad_norm": 5.000716209411621,
      "learning_rate": 2.9042926346347932e-06,
      "loss": 0.193,
      "step": 777
    },
    {
      "epoch": 0.014242562929061784,
      "grad_norm": 2.750865936279297,
      "learning_rate": 2.8797395400900362e-06,
      "loss": 0.1396,
      "step": 778
    },
    {
      "epoch": 0.014260869565217391,
      "grad_norm": 8.415980339050293,
      "learning_rate": 2.855273203671969e-06,
      "loss": 0.5748,
      "step": 779
    },
    {
      "epoch": 0.014279176201372997,
      "grad_norm": 4.887707710266113,
      "learning_rate": 2.830893923495173e-06,
      "loss": 0.2645,
      "step": 780
    },
    {
      "epoch": 0.014297482837528604,
      "grad_norm": 4.593505382537842,
      "learning_rate": 2.8066019966134907e-06,
      "loss": 0.212,
      "step": 781
    },
    {
      "epoch": 0.01431578947368421,
      "grad_norm": 4.513943672180176,
      "learning_rate": 2.7823977190163788e-06,
      "loss": 0.2007,
      "step": 782
    },
    {
      "epoch": 0.014334096109839817,
      "grad_norm": 1.8813406229019165,
      "learning_rate": 2.7582813856253276e-06,
      "loss": 0.1123,
      "step": 783
    },
    {
      "epoch": 0.014352402745995424,
      "grad_norm": 7.425732612609863,
      "learning_rate": 2.7342532902902418e-06,
      "loss": 0.2917,
      "step": 784
    },
    {
      "epoch": 0.01437070938215103,
      "grad_norm": 3.4717419147491455,
      "learning_rate": 2.7103137257858867e-06,
      "loss": 0.1469,
      "step": 785
    },
    {
      "epoch": 0.014389016018306637,
      "grad_norm": 3.6180458068847656,
      "learning_rate": 2.6864629838082957e-06,
      "loss": 0.1825,
      "step": 786
    },
    {
      "epoch": 0.014407322654462243,
      "grad_norm": 11.820043563842773,
      "learning_rate": 2.6627013549712355e-06,
      "loss": 0.3843,
      "step": 787
    },
    {
      "epoch": 0.01442562929061785,
      "grad_norm": 1.4070264101028442,
      "learning_rate": 2.639029128802657e-06,
      "loss": 0.1179,
      "step": 788
    },
    {
      "epoch": 0.014443935926773455,
      "grad_norm": 3.2094695568084717,
      "learning_rate": 2.615446593741161e-06,
      "loss": 0.1432,
      "step": 789
    },
    {
      "epoch": 0.014462242562929061,
      "grad_norm": 11.187945365905762,
      "learning_rate": 2.5919540371325005e-06,
      "loss": 0.5913,
      "step": 790
    },
    {
      "epoch": 0.014480549199084668,
      "grad_norm": 8.291489601135254,
      "learning_rate": 2.5685517452260566e-06,
      "loss": 0.5073,
      "step": 791
    },
    {
      "epoch": 0.014498855835240274,
      "grad_norm": 3.545529365539551,
      "learning_rate": 2.5452400031713786e-06,
      "loss": 0.1465,
      "step": 792
    },
    {
      "epoch": 0.01451716247139588,
      "grad_norm": 7.315761566162109,
      "learning_rate": 2.522019095014683e-06,
      "loss": 0.624,
      "step": 793
    },
    {
      "epoch": 0.014535469107551487,
      "grad_norm": 11.311168670654297,
      "learning_rate": 2.4988893036954045e-06,
      "loss": 0.5427,
      "step": 794
    },
    {
      "epoch": 0.014553775743707094,
      "grad_norm": 7.208177089691162,
      "learning_rate": 2.4758509110427576e-06,
      "loss": 0.2099,
      "step": 795
    },
    {
      "epoch": 0.0145720823798627,
      "grad_norm": 6.952854156494141,
      "learning_rate": 2.45290419777228e-06,
      "loss": 0.2803,
      "step": 796
    },
    {
      "epoch": 0.014590389016018307,
      "grad_norm": 5.105114459991455,
      "learning_rate": 2.4300494434824373e-06,
      "loss": 0.2132,
      "step": 797
    },
    {
      "epoch": 0.014608695652173913,
      "grad_norm": 5.196943283081055,
      "learning_rate": 2.407286926651192e-06,
      "loss": 0.2229,
      "step": 798
    },
    {
      "epoch": 0.01462700228832952,
      "grad_norm": 5.220571041107178,
      "learning_rate": 2.3846169246326345e-06,
      "loss": 0.2873,
      "step": 799
    },
    {
      "epoch": 0.014645308924485127,
      "grad_norm": 2.8920493125915527,
      "learning_rate": 2.362039713653581e-06,
      "loss": 0.1196,
      "step": 800
    },
    {
      "epoch": 0.014663615560640733,
      "grad_norm": 6.959986686706543,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.2673,
      "step": 801
    },
    {
      "epoch": 0.014681922196796338,
      "grad_norm": 19.750226974487305,
      "learning_rate": 2.317164764064769e-06,
      "loss": 0.2558,
      "step": 802
    },
    {
      "epoch": 0.014700228832951944,
      "grad_norm": 4.109073638916016,
      "learning_rate": 2.2948675722421086e-06,
      "loss": 0.2048,
      "step": 803
    },
    {
      "epoch": 0.014718535469107551,
      "grad_norm": 6.572871208190918,
      "learning_rate": 2.27266426502649e-06,
      "loss": 0.5073,
      "step": 804
    },
    {
      "epoch": 0.014736842105263158,
      "grad_norm": 5.497032642364502,
      "learning_rate": 2.2505551129582047e-06,
      "loss": 0.2909,
      "step": 805
    },
    {
      "epoch": 0.014755148741418764,
      "grad_norm": 4.798393726348877,
      "learning_rate": 2.2285403854302912e-06,
      "loss": 0.1645,
      "step": 806
    },
    {
      "epoch": 0.01477345537757437,
      "grad_norm": 6.386361122131348,
      "learning_rate": 2.206620350685257e-06,
      "loss": 0.2036,
      "step": 807
    },
    {
      "epoch": 0.014791762013729977,
      "grad_norm": 15.141334533691406,
      "learning_rate": 2.1847952758118118e-06,
      "loss": 0.6469,
      "step": 808
    },
    {
      "epoch": 0.014810068649885584,
      "grad_norm": 2.4272351264953613,
      "learning_rate": 2.163065426741603e-06,
      "loss": 0.1063,
      "step": 809
    },
    {
      "epoch": 0.01482837528604119,
      "grad_norm": 9.701894760131836,
      "learning_rate": 2.1414310682459805e-06,
      "loss": 0.7668,
      "step": 810
    },
    {
      "epoch": 0.014846681922196797,
      "grad_norm": 2.3073835372924805,
      "learning_rate": 2.119892463932781e-06,
      "loss": 0.161,
      "step": 811
    },
    {
      "epoch": 0.014864988558352403,
      "grad_norm": 8.390331268310547,
      "learning_rate": 2.098449876243096e-06,
      "loss": 0.3498,
      "step": 812
    },
    {
      "epoch": 0.01488329519450801,
      "grad_norm": 5.491335868835449,
      "learning_rate": 2.0771035664480944e-06,
      "loss": 0.2787,
      "step": 813
    },
    {
      "epoch": 0.014901601830663615,
      "grad_norm": 4.279118061065674,
      "learning_rate": 2.0558537946458177e-06,
      "loss": 0.1625,
      "step": 814
    },
    {
      "epoch": 0.014919908466819221,
      "grad_norm": 7.896650791168213,
      "learning_rate": 2.0347008197580376e-06,
      "loss": 0.3681,
      "step": 815
    },
    {
      "epoch": 0.014938215102974828,
      "grad_norm": 10.665939331054688,
      "learning_rate": 2.013644899527074e-06,
      "loss": 0.4568,
      "step": 816
    },
    {
      "epoch": 0.014956521739130434,
      "grad_norm": 24.828960418701172,
      "learning_rate": 1.9926862905126663e-06,
      "loss": 1.5729,
      "step": 817
    },
    {
      "epoch": 0.014974828375286041,
      "grad_norm": 12.46844482421875,
      "learning_rate": 1.9718252480888567e-06,
      "loss": 0.587,
      "step": 818
    },
    {
      "epoch": 0.014993135011441647,
      "grad_norm": 6.524204730987549,
      "learning_rate": 1.95106202644086e-06,
      "loss": 0.2851,
      "step": 819
    },
    {
      "epoch": 0.015011441647597254,
      "grad_norm": 9.998326301574707,
      "learning_rate": 1.930396878561983e-06,
      "loss": 0.5134,
      "step": 820
    },
    {
      "epoch": 0.01502974828375286,
      "grad_norm": 15.49677848815918,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.6058,
      "step": 821
    },
    {
      "epoch": 0.015048054919908467,
      "grad_norm": 5.195704460144043,
      "learning_rate": 1.8893618101067357e-06,
      "loss": 0.2136,
      "step": 822
    },
    {
      "epoch": 0.015066361556064074,
      "grad_norm": 7.91325044631958,
      "learning_rate": 1.8689923895297247e-06,
      "loss": 0.3135,
      "step": 823
    },
    {
      "epoch": 0.01508466819221968,
      "grad_norm": 12.137787818908691,
      "learning_rate": 1.848722042714457e-06,
      "loss": 0.9225,
      "step": 824
    },
    {
      "epoch": 0.015102974828375287,
      "grad_norm": 5.33144998550415,
      "learning_rate": 1.8285510166487154e-06,
      "loss": 0.2151,
      "step": 825
    },
    {
      "epoch": 0.015121281464530893,
      "grad_norm": 6.352451324462891,
      "learning_rate": 1.808479557110081e-06,
      "loss": 0.2846,
      "step": 826
    },
    {
      "epoch": 0.015139588100686498,
      "grad_norm": 5.770841598510742,
      "learning_rate": 1.7885079086629598e-06,
      "loss": 0.258,
      "step": 827
    },
    {
      "epoch": 0.015157894736842105,
      "grad_norm": 9.67846393585205,
      "learning_rate": 1.7686363146555807e-06,
      "loss": 0.359,
      "step": 828
    },
    {
      "epoch": 0.015176201372997711,
      "grad_norm": 1.321305751800537,
      "learning_rate": 1.7488650172170496e-06,
      "loss": 0.0735,
      "step": 829
    },
    {
      "epoch": 0.015194508009153318,
      "grad_norm": 14.980170249938965,
      "learning_rate": 1.7291942572543806e-06,
      "loss": 0.3208,
      "step": 830
    },
    {
      "epoch": 0.015212814645308924,
      "grad_norm": 12.814326286315918,
      "learning_rate": 1.709624274449584e-06,
      "loss": 0.5511,
      "step": 831
    },
    {
      "epoch": 0.01523112128146453,
      "grad_norm": 9.458513259887695,
      "learning_rate": 1.6901553072567189e-06,
      "loss": 0.3438,
      "step": 832
    },
    {
      "epoch": 0.015249427917620137,
      "grad_norm": 2.2716009616851807,
      "learning_rate": 1.6707875928990059e-06,
      "loss": 0.158,
      "step": 833
    },
    {
      "epoch": 0.015267734553775744,
      "grad_norm": 4.668413162231445,
      "learning_rate": 1.651521367365936e-06,
      "loss": 0.2095,
      "step": 834
    },
    {
      "epoch": 0.01528604118993135,
      "grad_norm": 6.753487586975098,
      "learning_rate": 1.6323568654103838e-06,
      "loss": 0.3238,
      "step": 835
    },
    {
      "epoch": 0.015304347826086957,
      "grad_norm": 4.7871785163879395,
      "learning_rate": 1.6132943205457607e-06,
      "loss": 0.2238,
      "step": 836
    },
    {
      "epoch": 0.015322654462242563,
      "grad_norm": 9.5857515335083,
      "learning_rate": 1.5943339650431578e-06,
      "loss": 0.3882,
      "step": 837
    },
    {
      "epoch": 0.01534096109839817,
      "grad_norm": 5.032664775848389,
      "learning_rate": 1.5754760299285255e-06,
      "loss": 0.2316,
      "step": 838
    },
    {
      "epoch": 0.015359267734553777,
      "grad_norm": 0.9537063837051392,
      "learning_rate": 1.5567207449798517e-06,
      "loss": 0.0631,
      "step": 839
    },
    {
      "epoch": 0.015377574370709381,
      "grad_norm": 3.6486315727233887,
      "learning_rate": 1.538068338724361e-06,
      "loss": 0.1585,
      "step": 840
    },
    {
      "epoch": 0.015395881006864988,
      "grad_norm": 7.9478840827941895,
      "learning_rate": 1.5195190384357405e-06,
      "loss": 0.3083,
      "step": 841
    },
    {
      "epoch": 0.015414187643020594,
      "grad_norm": 4.013637542724609,
      "learning_rate": 1.5010730701313626e-06,
      "loss": 0.1468,
      "step": 842
    },
    {
      "epoch": 0.015432494279176201,
      "grad_norm": 13.950557708740234,
      "learning_rate": 1.4827306585695234e-06,
      "loss": 0.4453,
      "step": 843
    },
    {
      "epoch": 0.015450800915331808,
      "grad_norm": 8.558002471923828,
      "learning_rate": 1.4644920272467245e-06,
      "loss": 0.5517,
      "step": 844
    },
    {
      "epoch": 0.015469107551487414,
      "grad_norm": 4.049571514129639,
      "learning_rate": 1.446357398394934e-06,
      "loss": 0.2165,
      "step": 845
    },
    {
      "epoch": 0.01548741418764302,
      "grad_norm": 7.492833137512207,
      "learning_rate": 1.4283269929788779e-06,
      "loss": 0.3542,
      "step": 846
    },
    {
      "epoch": 0.015505720823798627,
      "grad_norm": 4.005830764770508,
      "learning_rate": 1.4104010306933558e-06,
      "loss": 0.2093,
      "step": 847
    },
    {
      "epoch": 0.015524027459954234,
      "grad_norm": 4.699175834655762,
      "learning_rate": 1.3925797299605649e-06,
      "loss": 0.1869,
      "step": 848
    },
    {
      "epoch": 0.01554233409610984,
      "grad_norm": 7.27205228805542,
      "learning_rate": 1.3748633079274254e-06,
      "loss": 0.3142,
      "step": 849
    },
    {
      "epoch": 0.015560640732265447,
      "grad_norm": 6.372773170471191,
      "learning_rate": 1.3572519804629537e-06,
      "loss": 0.2057,
      "step": 850
    },
    {
      "epoch": 0.015578947368421053,
      "grad_norm": 6.957703590393066,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.2801,
      "step": 851
    },
    {
      "epoch": 0.01559725400457666,
      "grad_norm": 7.599903106689453,
      "learning_rate": 1.322345466310717e-06,
      "loss": 0.2687,
      "step": 852
    },
    {
      "epoch": 0.015615560640732265,
      "grad_norm": 5.099765300750732,
      "learning_rate": 1.30505070494781e-06,
      "loss": 0.2243,
      "step": 853
    },
    {
      "epoch": 0.015633867276887873,
      "grad_norm": 8.895179748535156,
      "learning_rate": 1.2878618887981064e-06,
      "loss": 0.4707,
      "step": 854
    },
    {
      "epoch": 0.01565217391304348,
      "grad_norm": 2.2955360412597656,
      "learning_rate": 1.2707792273019049e-06,
      "loss": 0.0989,
      "step": 855
    },
    {
      "epoch": 0.015670480549199086,
      "grad_norm": 7.87874698638916,
      "learning_rate": 1.2538029286060428e-06,
      "loss": 0.376,
      "step": 856
    },
    {
      "epoch": 0.015688787185354693,
      "grad_norm": 9.83598518371582,
      "learning_rate": 1.2369331995613664e-06,
      "loss": 0.6476,
      "step": 857
    },
    {
      "epoch": 0.0157070938215103,
      "grad_norm": 2.0399458408355713,
      "learning_rate": 1.2201702457201948e-06,
      "loss": 0.1026,
      "step": 858
    },
    {
      "epoch": 0.015725400457665902,
      "grad_norm": 6.7959442138671875,
      "learning_rate": 1.2035142713338366e-06,
      "loss": 0.1965,
      "step": 859
    },
    {
      "epoch": 0.01574370709382151,
      "grad_norm": 9.003751754760742,
      "learning_rate": 1.1869654793500784e-06,
      "loss": 0.5292,
      "step": 860
    },
    {
      "epoch": 0.015762013729977115,
      "grad_norm": 11.155455589294434,
      "learning_rate": 1.1705240714107301e-06,
      "loss": 0.5597,
      "step": 861
    },
    {
      "epoch": 0.015780320366132722,
      "grad_norm": 9.739468574523926,
      "learning_rate": 1.1541902478491607e-06,
      "loss": 0.5616,
      "step": 862
    },
    {
      "epoch": 0.01579862700228833,
      "grad_norm": 4.332462310791016,
      "learning_rate": 1.1379642076878528e-06,
      "loss": 0.1764,
      "step": 863
    },
    {
      "epoch": 0.015816933638443935,
      "grad_norm": 4.867697715759277,
      "learning_rate": 1.1218461486359878e-06,
      "loss": 0.3082,
      "step": 864
    },
    {
      "epoch": 0.01583524027459954,
      "grad_norm": 6.073899269104004,
      "learning_rate": 1.1058362670870248e-06,
      "loss": 0.3394,
      "step": 865
    },
    {
      "epoch": 0.015853546910755148,
      "grad_norm": 15.319504737854004,
      "learning_rate": 1.0899347581163222e-06,
      "loss": 0.3632,
      "step": 866
    },
    {
      "epoch": 0.015871853546910755,
      "grad_norm": 8.388349533081055,
      "learning_rate": 1.0741418154787443e-06,
      "loss": 0.477,
      "step": 867
    },
    {
      "epoch": 0.01589016018306636,
      "grad_norm": 4.829792499542236,
      "learning_rate": 1.058457631606319e-06,
      "loss": 0.2643,
      "step": 868
    },
    {
      "epoch": 0.015908466819221968,
      "grad_norm": 10.846988677978516,
      "learning_rate": 1.042882397605871e-06,
      "loss": 0.6206,
      "step": 869
    },
    {
      "epoch": 0.015926773455377574,
      "grad_norm": 9.261481285095215,
      "learning_rate": 1.0274163032567165e-06,
      "loss": 0.3205,
      "step": 870
    },
    {
      "epoch": 0.01594508009153318,
      "grad_norm": 1.8402290344238281,
      "learning_rate": 1.012059537008332e-06,
      "loss": 0.0896,
      "step": 871
    },
    {
      "epoch": 0.015963386727688787,
      "grad_norm": 4.178070545196533,
      "learning_rate": 9.968122859780648e-07,
      "loss": 0.2115,
      "step": 872
    },
    {
      "epoch": 0.015981693363844394,
      "grad_norm": 1.212082862854004,
      "learning_rate": 9.816747359488632e-07,
      "loss": 0.0796,
      "step": 873
    },
    {
      "epoch": 0.016,
      "grad_norm": 17.429851531982422,
      "learning_rate": 9.666470713669918e-07,
      "loss": 0.6567,
      "step": 874
    },
    {
      "epoch": 0.016018306636155607,
      "grad_norm": 3.6043081283569336,
      "learning_rate": 9.517294753398066e-07,
      "loss": 0.2011,
      "step": 875
    },
    {
      "epoch": 0.016036613272311213,
      "grad_norm": 6.399504661560059,
      "learning_rate": 9.369221296335007e-07,
      "loss": 0.3302,
      "step": 876
    },
    {
      "epoch": 0.01605491990846682,
      "grad_norm": 17.06818389892578,
      "learning_rate": 9.222252146709143e-07,
      "loss": 0.853,
      "step": 877
    },
    {
      "epoch": 0.016073226544622427,
      "grad_norm": 2.9740846157073975,
      "learning_rate": 9.076389095293148e-07,
      "loss": 0.1675,
      "step": 878
    },
    {
      "epoch": 0.016091533180778033,
      "grad_norm": 3.0257978439331055,
      "learning_rate": 8.931633919382299e-07,
      "loss": 0.2347,
      "step": 879
    },
    {
      "epoch": 0.01610983981693364,
      "grad_norm": 1.34092378616333,
      "learning_rate": 8.787988382772705e-07,
      "loss": 0.118,
      "step": 880
    },
    {
      "epoch": 0.016128146453089246,
      "grad_norm": 0.5438145995140076,
      "learning_rate": 8.645454235739903e-07,
      "loss": 0.0775,
      "step": 881
    },
    {
      "epoch": 0.016146453089244853,
      "grad_norm": 6.302069187164307,
      "learning_rate": 8.504033215017527e-07,
      "loss": 0.3397,
      "step": 882
    },
    {
      "epoch": 0.01616475972540046,
      "grad_norm": 11.530533790588379,
      "learning_rate": 8.363727043776037e-07,
      "loss": 0.3956,
      "step": 883
    },
    {
      "epoch": 0.016183066361556066,
      "grad_norm": 12.510212898254395,
      "learning_rate": 8.224537431601886e-07,
      "loss": 0.2942,
      "step": 884
    },
    {
      "epoch": 0.01620137299771167,
      "grad_norm": 10.052515983581543,
      "learning_rate": 8.086466074476562e-07,
      "loss": 0.2827,
      "step": 885
    },
    {
      "epoch": 0.016219679633867275,
      "grad_norm": 2.31078839302063,
      "learning_rate": 7.949514654755963e-07,
      "loss": 0.1479,
      "step": 886
    },
    {
      "epoch": 0.016237986270022882,
      "grad_norm": 16.138065338134766,
      "learning_rate": 7.81368484114996e-07,
      "loss": 0.4199,
      "step": 887
    },
    {
      "epoch": 0.01625629290617849,
      "grad_norm": 3.326777458190918,
      "learning_rate": 7.678978288701911e-07,
      "loss": 0.1543,
      "step": 888
    },
    {
      "epoch": 0.016274599542334095,
      "grad_norm": 5.338883876800537,
      "learning_rate": 7.545396638768698e-07,
      "loss": 0.1821,
      "step": 889
    },
    {
      "epoch": 0.0162929061784897,
      "grad_norm": 7.2971014976501465,
      "learning_rate": 7.412941519000527e-07,
      "loss": 0.3375,
      "step": 890
    },
    {
      "epoch": 0.016311212814645308,
      "grad_norm": 6.16159725189209,
      "learning_rate": 7.281614543321269e-07,
      "loss": 0.2027,
      "step": 891
    },
    {
      "epoch": 0.016329519450800915,
      "grad_norm": 12.14189338684082,
      "learning_rate": 7.151417311908648e-07,
      "loss": 0.4996,
      "step": 892
    },
    {
      "epoch": 0.01634782608695652,
      "grad_norm": 9.135406494140625,
      "learning_rate": 7.022351411174866e-07,
      "loss": 0.3862,
      "step": 893
    },
    {
      "epoch": 0.016366132723112128,
      "grad_norm": 9.651875495910645,
      "learning_rate": 6.894418413747183e-07,
      "loss": 0.3808,
      "step": 894
    },
    {
      "epoch": 0.016384439359267734,
      "grad_norm": 4.33256196975708,
      "learning_rate": 6.767619878448783e-07,
      "loss": 0.2232,
      "step": 895
    },
    {
      "epoch": 0.01640274599542334,
      "grad_norm": 5.881313800811768,
      "learning_rate": 6.641957350279838e-07,
      "loss": 0.2579,
      "step": 896
    },
    {
      "epoch": 0.016421052631578947,
      "grad_norm": 2.510111093521118,
      "learning_rate": 6.517432360398556e-07,
      "loss": 0.1578,
      "step": 897
    },
    {
      "epoch": 0.016439359267734554,
      "grad_norm": 2.6253507137298584,
      "learning_rate": 6.394046426102673e-07,
      "loss": 0.1334,
      "step": 898
    },
    {
      "epoch": 0.01645766590389016,
      "grad_norm": 5.544801712036133,
      "learning_rate": 6.271801050810856e-07,
      "loss": 0.3322,
      "step": 899
    },
    {
      "epoch": 0.016475972540045767,
      "grad_norm": 5.784806251525879,
      "learning_rate": 6.150697724044407e-07,
      "loss": 0.3011,
      "step": 900
    },
    {
      "epoch": 0.016494279176201374,
      "grad_norm": 10.115022659301758,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.4213,
      "step": 901
    },
    {
      "epoch": 0.01651258581235698,
      "grad_norm": 13.211285591125488,
      "learning_rate": 5.911923104577455e-07,
      "loss": 0.2885,
      "step": 902
    },
    {
      "epoch": 0.016530892448512587,
      "grad_norm": 4.184572696685791,
      "learning_rate": 5.794254721270331e-07,
      "loss": 0.2343,
      "step": 903
    },
    {
      "epoch": 0.016549199084668193,
      "grad_norm": 6.290907859802246,
      "learning_rate": 5.677734205239904e-07,
      "loss": 0.259,
      "step": 904
    },
    {
      "epoch": 0.0165675057208238,
      "grad_norm": 4.348118305206299,
      "learning_rate": 5.562362976251901e-07,
      "loss": 0.2195,
      "step": 905
    },
    {
      "epoch": 0.016585812356979406,
      "grad_norm": 7.895175457000732,
      "learning_rate": 5.448142440068316e-07,
      "loss": 0.3582,
      "step": 906
    },
    {
      "epoch": 0.016604118993135013,
      "grad_norm": 6.373378753662109,
      "learning_rate": 5.335073988430373e-07,
      "loss": 0.4875,
      "step": 907
    },
    {
      "epoch": 0.01662242562929062,
      "grad_norm": 4.442923069000244,
      "learning_rate": 5.223158999041444e-07,
      "loss": 0.2706,
      "step": 908
    },
    {
      "epoch": 0.016640732265446226,
      "grad_norm": 5.623570442199707,
      "learning_rate": 5.112398835550348e-07,
      "loss": 0.249,
      "step": 909
    },
    {
      "epoch": 0.01665903890160183,
      "grad_norm": 5.596260070800781,
      "learning_rate": 5.002794847534765e-07,
      "loss": 0.3783,
      "step": 910
    },
    {
      "epoch": 0.016677345537757435,
      "grad_norm": 4.682590484619141,
      "learning_rate": 4.894348370484648e-07,
      "loss": 0.2253,
      "step": 911
    },
    {
      "epoch": 0.016695652173913042,
      "grad_norm": 8.983024597167969,
      "learning_rate": 4.787060725786141e-07,
      "loss": 0.3797,
      "step": 912
    },
    {
      "epoch": 0.01671395881006865,
      "grad_norm": 1.1545554399490356,
      "learning_rate": 4.6809332207053083e-07,
      "loss": 0.1166,
      "step": 913
    },
    {
      "epoch": 0.016732265446224255,
      "grad_norm": 5.555506229400635,
      "learning_rate": 4.575967148372318e-07,
      "loss": 0.2964,
      "step": 914
    },
    {
      "epoch": 0.01675057208237986,
      "grad_norm": 2.3009276390075684,
      "learning_rate": 4.4721637877656377e-07,
      "loss": 0.1593,
      "step": 915
    },
    {
      "epoch": 0.016768878718535468,
      "grad_norm": 4.247383117675781,
      "learning_rate": 4.3695244036964567e-07,
      "loss": 0.1959,
      "step": 916
    },
    {
      "epoch": 0.016787185354691075,
      "grad_norm": 8.078374862670898,
      "learning_rate": 4.268050246793276e-07,
      "loss": 0.1937,
      "step": 917
    },
    {
      "epoch": 0.01680549199084668,
      "grad_norm": 4.602707386016846,
      "learning_rate": 4.167742553486676e-07,
      "loss": 0.2737,
      "step": 918
    },
    {
      "epoch": 0.016823798627002288,
      "grad_norm": 4.030495643615723,
      "learning_rate": 4.068602545994249e-07,
      "loss": 0.1949,
      "step": 919
    },
    {
      "epoch": 0.016842105263157894,
      "grad_norm": 6.921680450439453,
      "learning_rate": 3.9706314323056936e-07,
      "loss": 0.3539,
      "step": 920
    },
    {
      "epoch": 0.0168604118993135,
      "grad_norm": 4.294434070587158,
      "learning_rate": 3.8738304061681107e-07,
      "loss": 0.1642,
      "step": 921
    },
    {
      "epoch": 0.016878718535469107,
      "grad_norm": 8.08474349975586,
      "learning_rate": 3.7782006470714614e-07,
      "loss": 0.531,
      "step": 922
    },
    {
      "epoch": 0.016897025171624714,
      "grad_norm": 5.4197540283203125,
      "learning_rate": 3.68374332023419e-07,
      "loss": 0.1086,
      "step": 923
    },
    {
      "epoch": 0.01691533180778032,
      "grad_norm": 10.48359203338623,
      "learning_rate": 3.590459576589e-07,
      "loss": 0.4211,
      "step": 924
    },
    {
      "epoch": 0.016933638443935927,
      "grad_norm": 6.148892879486084,
      "learning_rate": 3.498350552768859e-07,
      "loss": 0.2935,
      "step": 925
    },
    {
      "epoch": 0.016951945080091534,
      "grad_norm": 37.46620559692383,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 0.4102,
      "step": 926
    },
    {
      "epoch": 0.01697025171624714,
      "grad_norm": 8.583579063415527,
      "learning_rate": 3.3176611395540625e-07,
      "loss": 0.5772,
      "step": 927
    },
    {
      "epoch": 0.016988558352402747,
      "grad_norm": 7.78873872756958,
      "learning_rate": 3.2290829518028867e-07,
      "loss": 0.4475,
      "step": 928
    },
    {
      "epoch": 0.017006864988558353,
      "grad_norm": 1.913015365600586,
      "learning_rate": 3.1416838871368925e-07,
      "loss": 0.1075,
      "step": 929
    },
    {
      "epoch": 0.01702517162471396,
      "grad_norm": 19.77604103088379,
      "learning_rate": 3.0554650104861137e-07,
      "loss": 0.8314,
      "step": 930
    },
    {
      "epoch": 0.017043478260869566,
      "grad_norm": 6.767706394195557,
      "learning_rate": 2.970427372400353e-07,
      "loss": 0.2661,
      "step": 931
    },
    {
      "epoch": 0.017061784897025173,
      "grad_norm": 6.047036647796631,
      "learning_rate": 2.8865720090364037e-07,
      "loss": 0.3102,
      "step": 932
    },
    {
      "epoch": 0.01708009153318078,
      "grad_norm": 3.897958517074585,
      "learning_rate": 2.8038999421453827e-07,
      "loss": 0.2229,
      "step": 933
    },
    {
      "epoch": 0.017098398169336386,
      "grad_norm": 14.799800872802734,
      "learning_rate": 2.7224121790603517e-07,
      "loss": 0.9037,
      "step": 934
    },
    {
      "epoch": 0.017116704805491992,
      "grad_norm": 1.944534420967102,
      "learning_rate": 2.6421097126839714e-07,
      "loss": 0.0869,
      "step": 935
    },
    {
      "epoch": 0.017135011441647596,
      "grad_norm": 6.719526290893555,
      "learning_rate": 2.5629935214764866e-07,
      "loss": 0.2525,
      "step": 936
    },
    {
      "epoch": 0.017153318077803202,
      "grad_norm": 1.7623887062072754,
      "learning_rate": 2.4850645694436736e-07,
      "loss": 0.0721,
      "step": 937
    },
    {
      "epoch": 0.01717162471395881,
      "grad_norm": 2.431997776031494,
      "learning_rate": 2.4083238061252565e-07,
      "loss": 0.1643,
      "step": 938
    },
    {
      "epoch": 0.017189931350114415,
      "grad_norm": 11.728501319885254,
      "learning_rate": 2.332772166583208e-07,
      "loss": 0.234,
      "step": 939
    },
    {
      "epoch": 0.017208237986270022,
      "grad_norm": 1.5342614650726318,
      "learning_rate": 2.2584105713904126e-07,
      "loss": 0.1473,
      "step": 940
    },
    {
      "epoch": 0.01722654462242563,
      "grad_norm": 8.040061950683594,
      "learning_rate": 2.1852399266194312e-07,
      "loss": 0.3064,
      "step": 941
    },
    {
      "epoch": 0.017244851258581235,
      "grad_norm": 2.8261466026306152,
      "learning_rate": 2.1132611238315004e-07,
      "loss": 0.2135,
      "step": 942
    },
    {
      "epoch": 0.01726315789473684,
      "grad_norm": 5.933513164520264,
      "learning_rate": 2.0424750400655947e-07,
      "loss": 0.3271,
      "step": 943
    },
    {
      "epoch": 0.017281464530892448,
      "grad_norm": 3.4956777095794678,
      "learning_rate": 1.9728825378278248e-07,
      "loss": 0.1417,
      "step": 944
    },
    {
      "epoch": 0.017299771167048054,
      "grad_norm": 6.9485697746276855,
      "learning_rate": 1.9044844650808468e-07,
      "loss": 0.326,
      "step": 945
    },
    {
      "epoch": 0.01731807780320366,
      "grad_norm": 9.41317081451416,
      "learning_rate": 1.8372816552336025e-07,
      "loss": 0.4231,
      "step": 946
    },
    {
      "epoch": 0.017336384439359268,
      "grad_norm": 6.496486663818359,
      "learning_rate": 1.7712749271311392e-07,
      "loss": 0.2621,
      "step": 947
    },
    {
      "epoch": 0.017354691075514874,
      "grad_norm": 9.458244323730469,
      "learning_rate": 1.706465085044584e-07,
      "loss": 0.3791,
      "step": 948
    },
    {
      "epoch": 0.01737299771167048,
      "grad_norm": 1.7211376428604126,
      "learning_rate": 1.6428529186614195e-07,
      "loss": 0.1238,
      "step": 949
    },
    {
      "epoch": 0.017391304347826087,
      "grad_norm": 2.092992067337036,
      "learning_rate": 1.580439203075812e-07,
      "loss": 0.1183,
      "step": 950
    },
    {
      "epoch": 0.017409610983981694,
      "grad_norm": 7.563133716583252,
      "learning_rate": 1.519224698779198e-07,
      "loss": 0.2907,
      "step": 951
    },
    {
      "epoch": 0.0174279176201373,
      "grad_norm": 13.335310935974121,
      "learning_rate": 1.4592101516509916e-07,
      "loss": 0.7671,
      "step": 952
    },
    {
      "epoch": 0.017446224256292907,
      "grad_norm": 10.47811508178711,
      "learning_rate": 1.400396292949513e-07,
      "loss": 0.4623,
      "step": 953
    },
    {
      "epoch": 0.017464530892448513,
      "grad_norm": 7.000487804412842,
      "learning_rate": 1.3427838393030634e-07,
      "loss": 0.4022,
      "step": 954
    },
    {
      "epoch": 0.01748283752860412,
      "grad_norm": 6.856893062591553,
      "learning_rate": 1.2863734927012094e-07,
      "loss": 0.3277,
      "step": 955
    },
    {
      "epoch": 0.017501144164759726,
      "grad_norm": 12.255921363830566,
      "learning_rate": 1.231165940486234e-07,
      "loss": 0.502,
      "step": 956
    },
    {
      "epoch": 0.017519450800915333,
      "grad_norm": 24.51825714111328,
      "learning_rate": 1.1771618553447217e-07,
      "loss": 0.3786,
      "step": 957
    },
    {
      "epoch": 0.01753775743707094,
      "grad_norm": 2.42156982421875,
      "learning_rate": 1.1243618952994195e-07,
      "loss": 0.1061,
      "step": 958
    },
    {
      "epoch": 0.017556064073226546,
      "grad_norm": 19.74112319946289,
      "learning_rate": 1.0727667037011668e-07,
      "loss": 0.6959,
      "step": 959
    },
    {
      "epoch": 0.017574370709382153,
      "grad_norm": 2.00630784034729,
      "learning_rate": 1.0223769092211012e-07,
      "loss": 0.1106,
      "step": 960
    },
    {
      "epoch": 0.01759267734553776,
      "grad_norm": 7.67366361618042,
      "learning_rate": 9.731931258429638e-08,
      "loss": 0.173,
      "step": 961
    },
    {
      "epoch": 0.017610983981693362,
      "grad_norm": 9.130098342895508,
      "learning_rate": 9.252159528556404e-08,
      "loss": 0.4265,
      "step": 962
    },
    {
      "epoch": 0.01762929061784897,
      "grad_norm": 2.1006438732147217,
      "learning_rate": 8.784459748458318e-08,
      "loss": 0.0866,
      "step": 963
    },
    {
      "epoch": 0.017647597254004575,
      "grad_norm": 9.933135032653809,
      "learning_rate": 8.328837616909612e-08,
      "loss": 0.2907,
      "step": 964
    },
    {
      "epoch": 0.017665903890160182,
      "grad_norm": 6.022252559661865,
      "learning_rate": 7.885298685522235e-08,
      "loss": 0.3691,
      "step": 965
    },
    {
      "epoch": 0.01768421052631579,
      "grad_norm": 3.950028419494629,
      "learning_rate": 7.453848358678018e-08,
      "loss": 0.138,
      "step": 966
    },
    {
      "epoch": 0.017702517162471395,
      "grad_norm": 5.163845539093018,
      "learning_rate": 7.034491893463059e-08,
      "loss": 0.2737,
      "step": 967
    },
    {
      "epoch": 0.017720823798627,
      "grad_norm": 7.640467166900635,
      "learning_rate": 6.627234399603554e-08,
      "loss": 0.3658,
      "step": 968
    },
    {
      "epoch": 0.017739130434782608,
      "grad_norm": 6.5803985595703125,
      "learning_rate": 6.232080839403631e-08,
      "loss": 0.3093,
      "step": 969
    },
    {
      "epoch": 0.017757437070938215,
      "grad_norm": 4.211984634399414,
      "learning_rate": 5.849036027684607e-08,
      "loss": 0.1708,
      "step": 970
    },
    {
      "epoch": 0.01777574370709382,
      "grad_norm": 8.148463249206543,
      "learning_rate": 5.4781046317267103e-08,
      "loss": 0.3152,
      "step": 971
    },
    {
      "epoch": 0.017794050343249428,
      "grad_norm": 4.271562099456787,
      "learning_rate": 5.119291171211793e-08,
      "loss": 0.1852,
      "step": 972
    },
    {
      "epoch": 0.017812356979405034,
      "grad_norm": 8.320340156555176,
      "learning_rate": 4.772600018168816e-08,
      "loss": 0.3206,
      "step": 973
    },
    {
      "epoch": 0.01783066361556064,
      "grad_norm": 9.28972053527832,
      "learning_rate": 4.438035396920004e-08,
      "loss": 0.3864,
      "step": 974
    },
    {
      "epoch": 0.017848970251716247,
      "grad_norm": 12.746179580688477,
      "learning_rate": 4.115601384029666e-08,
      "loss": 0.694,
      "step": 975
    },
    {
      "epoch": 0.017867276887871854,
      "grad_norm": 12.518298149108887,
      "learning_rate": 3.805301908254455e-08,
      "loss": 0.4285,
      "step": 976
    },
    {
      "epoch": 0.01788558352402746,
      "grad_norm": 6.564645767211914,
      "learning_rate": 3.50714075049563e-08,
      "loss": 0.3002,
      "step": 977
    },
    {
      "epoch": 0.017903890160183067,
      "grad_norm": 8.353341102600098,
      "learning_rate": 3.22112154375287e-08,
      "loss": 0.3428,
      "step": 978
    },
    {
      "epoch": 0.017922196796338673,
      "grad_norm": 8.169654846191406,
      "learning_rate": 2.947247773079753e-08,
      "loss": 0.3041,
      "step": 979
    },
    {
      "epoch": 0.01794050343249428,
      "grad_norm": 3.4119813442230225,
      "learning_rate": 2.6855227755419046e-08,
      "loss": 0.1635,
      "step": 980
    },
    {
      "epoch": 0.017958810068649887,
      "grad_norm": 0.7701112627983093,
      "learning_rate": 2.4359497401758026e-08,
      "loss": 0.0799,
      "step": 981
    },
    {
      "epoch": 0.017977116704805493,
      "grad_norm": 7.056750297546387,
      "learning_rate": 2.1985317079500358e-08,
      "loss": 0.2713,
      "step": 982
    },
    {
      "epoch": 0.0179954233409611,
      "grad_norm": 13.480681419372559,
      "learning_rate": 1.973271571728441e-08,
      "loss": 0.5698,
      "step": 983
    },
    {
      "epoch": 0.018013729977116706,
      "grad_norm": 2.141348123550415,
      "learning_rate": 1.7601720762346895e-08,
      "loss": 0.1175,
      "step": 984
    },
    {
      "epoch": 0.018032036613272313,
      "grad_norm": 4.411203384399414,
      "learning_rate": 1.5592358180189782e-08,
      "loss": 0.1344,
      "step": 985
    },
    {
      "epoch": 0.01805034324942792,
      "grad_norm": 9.80256462097168,
      "learning_rate": 1.370465245426167e-08,
      "loss": 0.3765,
      "step": 986
    },
    {
      "epoch": 0.018068649885583522,
      "grad_norm": 20.578289031982422,
      "learning_rate": 1.1938626585660252e-08,
      "loss": 0.9648,
      "step": 987
    },
    {
      "epoch": 0.01808695652173913,
      "grad_norm": 6.394954681396484,
      "learning_rate": 1.0294302092853647e-08,
      "loss": 0.2917,
      "step": 988
    },
    {
      "epoch": 0.018105263157894735,
      "grad_norm": 2.968735933303833,
      "learning_rate": 8.771699011416169e-09,
      "loss": 0.1552,
      "step": 989
    },
    {
      "epoch": 0.018123569794050342,
      "grad_norm": 9.075217247009277,
      "learning_rate": 7.370835893788508e-09,
      "loss": 0.5275,
      "step": 990
    },
    {
      "epoch": 0.01814187643020595,
      "grad_norm": 0.4628898799419403,
      "learning_rate": 6.091729809042379e-09,
      "loss": 0.0548,
      "step": 991
    },
    {
      "epoch": 0.018160183066361555,
      "grad_norm": 9.862579345703125,
      "learning_rate": 4.9343963426840006e-09,
      "loss": 0.4799,
      "step": 992
    },
    {
      "epoch": 0.01817848970251716,
      "grad_norm": 0.9560385942459106,
      "learning_rate": 3.898849596456477e-09,
      "loss": 0.0928,
      "step": 993
    },
    {
      "epoch": 0.018196796338672768,
      "grad_norm": 14.042981147766113,
      "learning_rate": 2.9851021881688314e-09,
      "loss": 0.4677,
      "step": 994
    },
    {
      "epoch": 0.018215102974828375,
      "grad_norm": 1.057991862297058,
      "learning_rate": 2.193165251545004e-09,
      "loss": 0.112,
      "step": 995
    },
    {
      "epoch": 0.01823340961098398,
      "grad_norm": 2.119772434234619,
      "learning_rate": 1.5230484360873043e-09,
      "loss": 0.105,
      "step": 996
    },
    {
      "epoch": 0.018251716247139588,
      "grad_norm": 8.898747444152832,
      "learning_rate": 9.74759906957612e-10,
      "loss": 0.403,
      "step": 997
    },
    {
      "epoch": 0.018270022883295194,
      "grad_norm": 4.880192756652832,
      "learning_rate": 5.483063448785686e-10,
      "loss": 0.2703,
      "step": 998
    },
    {
      "epoch": 0.0182883295194508,
      "grad_norm": 1.7518311738967896,
      "learning_rate": 2.436929460525317e-10,
      "loss": 0.0996,
      "step": 999
    },
    {
      "epoch": 0.018306636155606407,
      "grad_norm": 4.965724945068359,
      "learning_rate": 6.092342209607083e-11,
      "loss": 0.208,
      "step": 1000
    }
  ],
  "logging_steps": 1,
  "max_steps": 1000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}