{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.05,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1e-05,
      "grad_norm": 1.6317715205474108,
      "learning_rate": 3e-06,
      "loss": 10.867,
      "step": 1
    },
    {
      "epoch": 2e-05,
      "grad_norm": 1.6173147870740345,
      "learning_rate": 6e-06,
      "loss": 10.8657,
      "step": 2
    },
    {
      "epoch": 3e-05,
      "grad_norm": 1.6387509359885835,
      "learning_rate": 9e-06,
      "loss": 10.8658,
      "step": 3
    },
    {
      "epoch": 4e-05,
      "grad_norm": 1.597552357815991,
      "learning_rate": 1.2e-05,
      "loss": 10.865,
      "step": 4
    },
    {
      "epoch": 5e-05,
      "grad_norm": 1.6454471252189307,
      "learning_rate": 1.5e-05,
      "loss": 10.8617,
      "step": 5
    },
    {
      "epoch": 6e-05,
      "grad_norm": 1.6407925722175996,
      "learning_rate": 1.8e-05,
      "loss": 10.8593,
      "step": 6
    },
    {
      "epoch": 7e-05,
      "grad_norm": 1.6096088910322361,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 10.8456,
      "step": 7
    },
    {
      "epoch": 8e-05,
      "grad_norm": 1.4682698515009915,
      "learning_rate": 2.4e-05,
      "loss": 10.8184,
      "step": 8
    },
    {
      "epoch": 9e-05,
      "grad_norm": 1.3934246427009196,
      "learning_rate": 2.7e-05,
      "loss": 10.8113,
      "step": 9
    },
    {
      "epoch": 0.0001,
      "grad_norm": 1.3326087040550991,
      "learning_rate": 3e-05,
      "loss": 10.7969,
      "step": 10
    },
    {
      "epoch": 0.00011,
      "grad_norm": 1.2173090118888668,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 10.7801,
      "step": 11
    },
    {
      "epoch": 0.00012,
      "grad_norm": 1.176457724285593,
      "learning_rate": 3.6e-05,
      "loss": 10.7688,
      "step": 12
    },
    {
      "epoch": 0.00013,
      "grad_norm": 1.1304424318539597,
      "learning_rate": 3.9e-05,
      "loss": 10.7498,
      "step": 13
    },
    {
      "epoch": 0.00014,
      "grad_norm": 1.1158244568462428,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 10.739,
      "step": 14
    },
    {
      "epoch": 0.00015,
      "grad_norm": 1.10654756415174,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 10.7299,
      "step": 15
    },
    {
      "epoch": 0.00016,
      "grad_norm": 1.0751770608444569,
      "learning_rate": 4.8e-05,
      "loss": 10.7126,
      "step": 16
    },
    {
      "epoch": 0.00017,
      "grad_norm": 1.046143788290158,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 10.6968,
      "step": 17
    },
    {
      "epoch": 0.00018,
      "grad_norm": 1.0230391412556632,
      "learning_rate": 5.4e-05,
      "loss": 10.6806,
      "step": 18
    },
    {
      "epoch": 0.00019,
      "grad_norm": 0.9869982733638543,
      "learning_rate": 5.7e-05,
      "loss": 10.6649,
      "step": 19
    },
    {
      "epoch": 0.0002,
      "grad_norm": 0.9728818553338922,
      "learning_rate": 6e-05,
      "loss": 10.6526,
      "step": 20
    },
    {
      "epoch": 0.00021,
      "grad_norm": 0.9439994749998407,
      "learning_rate": 6.3e-05,
      "loss": 10.639,
      "step": 21
    },
    {
      "epoch": 0.00022,
      "grad_norm": 0.9279528910342688,
      "learning_rate": 6.599999999999999e-05,
      "loss": 10.6244,
      "step": 22
    },
    {
      "epoch": 0.00023,
      "grad_norm": 0.9208663519602571,
      "learning_rate": 6.9e-05,
      "loss": 10.6103,
      "step": 23
    },
    {
      "epoch": 0.00024,
      "grad_norm": 0.9178551557561957,
      "learning_rate": 7.2e-05,
      "loss": 10.5993,
      "step": 24
    },
    {
      "epoch": 0.00025,
      "grad_norm": 0.9193923250060233,
      "learning_rate": 7.500000000000001e-05,
      "loss": 10.5847,
      "step": 25
    },
    {
      "epoch": 0.00026,
      "grad_norm": 0.9190901609677985,
      "learning_rate": 7.8e-05,
      "loss": 10.5717,
      "step": 26
    },
    {
      "epoch": 0.00027,
      "grad_norm": 0.913753327244254,
      "learning_rate": 8.1e-05,
      "loss": 10.5597,
      "step": 27
    },
    {
      "epoch": 0.00028,
      "grad_norm": 0.9119625217070774,
      "learning_rate": 8.400000000000001e-05,
      "loss": 10.5467,
      "step": 28
    },
    {
      "epoch": 0.00029,
      "grad_norm": 0.9131038863398008,
      "learning_rate": 8.7e-05,
      "loss": 10.5323,
      "step": 29
    },
    {
      "epoch": 0.0003,
      "grad_norm": 0.9186172829723749,
      "learning_rate": 8.999999999999999e-05,
      "loss": 10.517,
      "step": 30
    },
    {
      "epoch": 0.00031,
      "grad_norm": 0.9155453723962563,
      "learning_rate": 9.3e-05,
      "loss": 10.5024,
      "step": 31
    },
    {
      "epoch": 0.00032,
      "grad_norm": 0.909575589137279,
      "learning_rate": 9.6e-05,
      "loss": 10.4882,
      "step": 32
    },
    {
      "epoch": 0.00033,
      "grad_norm": 0.9102175928717151,
      "learning_rate": 9.900000000000001e-05,
      "loss": 10.4717,
      "step": 33
    },
    {
      "epoch": 0.00034,
      "grad_norm": 0.9103991171564593,
      "learning_rate": 0.00010200000000000001,
      "loss": 10.4554,
      "step": 34
    },
    {
      "epoch": 0.00035,
      "grad_norm": 0.9164468459870767,
      "learning_rate": 0.00010500000000000002,
      "loss": 10.4388,
      "step": 35
    },
    {
      "epoch": 0.00036,
      "grad_norm": 0.9129850089149896,
      "learning_rate": 0.000108,
      "loss": 10.4213,
      "step": 36
    },
    {
      "epoch": 0.00037,
      "grad_norm": 0.9029808327462479,
      "learning_rate": 0.000111,
      "loss": 10.4047,
      "step": 37
    },
    {
      "epoch": 0.00038,
      "grad_norm": 0.9074583652458479,
      "learning_rate": 0.000114,
      "loss": 10.3859,
      "step": 38
    },
    {
      "epoch": 0.00039,
      "grad_norm": 0.9101621604618185,
      "learning_rate": 0.000117,
      "loss": 10.3665,
      "step": 39
    },
    {
      "epoch": 0.0004,
      "grad_norm": 0.9144345472354501,
      "learning_rate": 0.00012,
      "loss": 10.3443,
      "step": 40
    },
    {
      "epoch": 0.00041,
      "grad_norm": 0.9045315909874942,
      "learning_rate": 0.000123,
      "loss": 10.3264,
      "step": 41
    },
    {
      "epoch": 0.00042,
      "grad_norm": 0.9101518170592343,
      "learning_rate": 0.000126,
      "loss": 10.3059,
      "step": 42
    },
    {
      "epoch": 0.00043,
      "grad_norm": 0.9147491310031046,
      "learning_rate": 0.000129,
      "loss": 10.2831,
      "step": 43
    },
    {
      "epoch": 0.00044,
      "grad_norm": 0.9138166723808987,
      "learning_rate": 0.00013199999999999998,
      "loss": 10.2617,
      "step": 44
    },
    {
      "epoch": 0.00045,
      "grad_norm": 0.9166634386783579,
      "learning_rate": 0.000135,
      "loss": 10.2388,
      "step": 45
    },
    {
      "epoch": 0.00046,
      "grad_norm": 0.9061319871747918,
      "learning_rate": 0.000138,
      "loss": 10.2171,
      "step": 46
    },
    {
      "epoch": 0.00047,
      "grad_norm": 0.9134144094551727,
      "learning_rate": 0.000141,
      "loss": 10.1906,
      "step": 47
    },
    {
      "epoch": 0.00048,
      "grad_norm": 0.9111015850262806,
      "learning_rate": 0.000144,
      "loss": 10.1669,
      "step": 48
    },
    {
      "epoch": 0.00049,
      "grad_norm": 0.9106086112424904,
      "learning_rate": 0.000147,
      "loss": 10.1439,
      "step": 49
    },
    {
      "epoch": 0.0005,
      "grad_norm": 0.9135108001899231,
      "learning_rate": 0.00015000000000000001,
      "loss": 10.1178,
      "step": 50
    },
    {
      "epoch": 0.00051,
      "grad_norm": 0.9152031108656089,
      "learning_rate": 0.000153,
      "loss": 10.0918,
      "step": 51
    },
    {
      "epoch": 0.00052,
      "grad_norm": 0.9133043896843657,
      "learning_rate": 0.000156,
      "loss": 10.0658,
      "step": 52
    },
    {
      "epoch": 0.00053,
      "grad_norm": 0.9039676544194273,
      "learning_rate": 0.000159,
      "loss": 10.0419,
      "step": 53
    },
    {
      "epoch": 0.00054,
      "grad_norm": 0.9245050218484777,
      "learning_rate": 0.000162,
      "loss": 10.0099,
      "step": 54
    },
    {
      "epoch": 0.00055,
      "grad_norm": 0.915653013423474,
      "learning_rate": 0.000165,
      "loss": 9.9858,
      "step": 55
    },
    {
      "epoch": 0.00056,
      "grad_norm": 0.90743999026624,
      "learning_rate": 0.00016800000000000002,
      "loss": 9.9567,
      "step": 56
    },
    {
      "epoch": 0.00057,
      "grad_norm": 0.9125740935300273,
      "learning_rate": 0.000171,
      "loss": 9.9292,
      "step": 57
    },
    {
      "epoch": 0.00058,
      "grad_norm": 0.9103134473221595,
      "learning_rate": 0.000174,
      "loss": 9.9046,
      "step": 58
    },
    {
      "epoch": 0.00059,
      "grad_norm": 0.91502039796166,
      "learning_rate": 0.000177,
      "loss": 9.8727,
      "step": 59
    },
    {
      "epoch": 0.0006,
      "grad_norm": 0.902549468432534,
      "learning_rate": 0.00017999999999999998,
      "loss": 9.8467,
      "step": 60
    },
    {
      "epoch": 0.00061,
      "grad_norm": 0.9011753674575653,
      "learning_rate": 0.000183,
      "loss": 9.8184,
      "step": 61
    },
    {
      "epoch": 0.00062,
      "grad_norm": 0.9073829944096542,
      "learning_rate": 0.000186,
      "loss": 9.7865,
      "step": 62
    },
    {
      "epoch": 0.00063,
      "grad_norm": 0.900277782228101,
      "learning_rate": 0.000189,
      "loss": 9.7594,
      "step": 63
    },
    {
      "epoch": 0.00064,
      "grad_norm": 0.8964836174343672,
      "learning_rate": 0.000192,
      "loss": 9.7292,
      "step": 64
    },
    {
      "epoch": 0.00065,
      "grad_norm": 0.9040253232587138,
      "learning_rate": 0.00019500000000000002,
      "loss": 9.6969,
      "step": 65
    },
    {
      "epoch": 0.00066,
      "grad_norm": 0.8969700410935529,
      "learning_rate": 0.00019800000000000002,
      "loss": 9.6728,
      "step": 66
    },
    {
      "epoch": 0.00067,
      "grad_norm": 0.8981910064021,
      "learning_rate": 0.000201,
      "loss": 9.643,
      "step": 67
    },
    {
      "epoch": 0.00068,
      "grad_norm": 0.9049113812894196,
      "learning_rate": 0.00020400000000000003,
      "loss": 9.61,
      "step": 68
    },
    {
      "epoch": 0.00069,
      "grad_norm": 0.8903730719674341,
      "learning_rate": 0.00020700000000000002,
      "loss": 9.5824,
      "step": 69
    },
    {
      "epoch": 0.0007,
      "grad_norm": 0.9034192157313848,
      "learning_rate": 0.00021000000000000004,
      "loss": 9.5456,
      "step": 70
    },
    {
      "epoch": 0.00071,
      "grad_norm": 0.8933343163190056,
      "learning_rate": 0.00021299999999999997,
      "loss": 9.5189,
      "step": 71
    },
    {
      "epoch": 0.00072,
      "grad_norm": 0.8968593008835964,
      "learning_rate": 0.000216,
      "loss": 9.4914,
      "step": 72
    },
    {
      "epoch": 0.00073,
      "grad_norm": 0.8960167287531013,
      "learning_rate": 0.00021899999999999998,
      "loss": 9.4574,
      "step": 73
    },
    {
      "epoch": 0.00074,
      "grad_norm": 0.8965354004907367,
      "learning_rate": 0.000222,
      "loss": 9.4304,
      "step": 74
    },
    {
      "epoch": 0.00075,
      "grad_norm": 0.8923958382533757,
      "learning_rate": 0.000225,
      "loss": 9.3946,
      "step": 75
    },
    {
      "epoch": 0.00076,
      "grad_norm": 0.9011054851249423,
      "learning_rate": 0.000228,
      "loss": 9.3667,
      "step": 76
    },
    {
      "epoch": 0.00077,
      "grad_norm": 0.8959973023769764,
      "learning_rate": 0.000231,
      "loss": 9.3374,
      "step": 77
    },
    {
      "epoch": 0.00078,
      "grad_norm": 0.8901362352309407,
      "learning_rate": 0.000234,
      "loss": 9.3055,
      "step": 78
    },
    {
      "epoch": 0.00079,
      "grad_norm": 0.8931237127830396,
      "learning_rate": 0.00023700000000000001,
      "loss": 9.2694,
      "step": 79
    },
    {
      "epoch": 0.0008,
      "grad_norm": 0.8937972757596185,
      "learning_rate": 0.00024,
      "loss": 9.2355,
      "step": 80
    },
    {
      "epoch": 0.00081,
      "grad_norm": 0.8984186266430717,
      "learning_rate": 0.00024300000000000002,
      "loss": 9.2033,
      "step": 81
    },
    {
      "epoch": 0.00082,
      "grad_norm": 0.9060711515558022,
      "learning_rate": 0.000246,
      "loss": 9.1789,
      "step": 82
    },
    {
      "epoch": 0.00083,
      "grad_norm": 0.8964035250431631,
      "learning_rate": 0.00024900000000000004,
      "loss": 9.1434,
      "step": 83
    },
    {
      "epoch": 0.00084,
      "grad_norm": 0.8986377804082708,
      "learning_rate": 0.000252,
      "loss": 9.1119,
      "step": 84
    },
    {
      "epoch": 0.00085,
      "grad_norm": 0.8933390971667627,
      "learning_rate": 0.000255,
      "loss": 9.0885,
      "step": 85
    },
    {
      "epoch": 0.00086,
      "grad_norm": 0.8953786211031305,
      "learning_rate": 0.000258,
      "loss": 9.0521,
      "step": 86
    },
    {
      "epoch": 0.00087,
      "grad_norm": 0.8876773666743288,
      "learning_rate": 0.000261,
      "loss": 9.0302,
      "step": 87
    },
    {
      "epoch": 0.00088,
      "grad_norm": 0.8929152197347487,
      "learning_rate": 0.00026399999999999997,
      "loss": 8.9927,
      "step": 88
    },
    {
      "epoch": 0.00089,
      "grad_norm": 0.8872457913370299,
      "learning_rate": 0.000267,
      "loss": 8.9669,
      "step": 89
    },
    {
      "epoch": 0.0009,
      "grad_norm": 0.8862509419810688,
      "learning_rate": 0.00027,
      "loss": 8.9393,
      "step": 90
    },
    {
      "epoch": 0.00091,
      "grad_norm": 0.8807880266520192,
      "learning_rate": 0.000273,
      "loss": 8.9089,
      "step": 91
    },
    {
      "epoch": 0.00092,
      "grad_norm": 0.8831207589668301,
      "learning_rate": 0.000276,
      "loss": 8.876,
      "step": 92
    },
    {
      "epoch": 0.00093,
      "grad_norm": 0.883866988348631,
      "learning_rate": 0.000279,
      "loss": 8.8462,
      "step": 93
    },
    {
      "epoch": 0.00094,
      "grad_norm": 0.8824853161719922,
      "learning_rate": 0.000282,
      "loss": 8.8199,
      "step": 94
    },
    {
      "epoch": 0.00095,
      "grad_norm": 0.8794572899807177,
      "learning_rate": 0.000285,
      "loss": 8.789,
      "step": 95
    },
    {
      "epoch": 0.00096,
      "grad_norm": 0.8879520784944948,
      "learning_rate": 0.000288,
      "loss": 8.7571,
      "step": 96
    },
    {
      "epoch": 0.00097,
      "grad_norm": 0.8798042444972031,
      "learning_rate": 0.000291,
      "loss": 8.7321,
      "step": 97
    },
    {
      "epoch": 0.00098,
      "grad_norm": 0.8802597894834375,
      "learning_rate": 0.000294,
      "loss": 8.6954,
      "step": 98
    },
    {
      "epoch": 0.00099,
      "grad_norm": 0.8806466323910314,
      "learning_rate": 0.000297,
      "loss": 8.6749,
      "step": 99
    },
    {
      "epoch": 0.001,
      "grad_norm": 0.8777097574069823,
      "learning_rate": 0.00030000000000000003,
      "loss": 8.6485,
      "step": 100
    },
    {
      "epoch": 0.00101,
      "grad_norm": 0.8786755147609817,
      "learning_rate": 0.00030300000000000005,
      "loss": 8.614,
      "step": 101
    },
    {
      "epoch": 0.00102,
      "grad_norm": 0.8680143868447665,
      "learning_rate": 0.000306,
      "loss": 8.5949,
      "step": 102
    },
    {
      "epoch": 0.00103,
      "grad_norm": 0.8748449452730288,
      "learning_rate": 0.000309,
      "loss": 8.5706,
      "step": 103
    },
    {
      "epoch": 0.00104,
      "grad_norm": 0.8670215859388973,
      "learning_rate": 0.000312,
      "loss": 8.5498,
      "step": 104
    },
    {
      "epoch": 0.00105,
      "grad_norm": 0.8687292008731472,
      "learning_rate": 0.000315,
      "loss": 8.5231,
      "step": 105
    },
    {
      "epoch": 0.00106,
      "grad_norm": 0.8611486197845404,
      "learning_rate": 0.000318,
      "loss": 8.4945,
      "step": 106
    },
    {
      "epoch": 0.00107,
      "grad_norm": 0.8521924348129856,
      "learning_rate": 0.000321,
      "loss": 8.4693,
      "step": 107
    },
    {
      "epoch": 0.00108,
      "grad_norm": 0.8581933475380797,
      "learning_rate": 0.000324,
      "loss": 8.4407,
      "step": 108
    },
    {
      "epoch": 0.00109,
      "grad_norm": 0.8524146875206363,
      "learning_rate": 0.000327,
      "loss": 8.421,
      "step": 109
    },
    {
      "epoch": 0.0011,
      "grad_norm": 0.8682563584613229,
      "learning_rate": 0.00033,
      "loss": 8.3983,
      "step": 110
    },
    {
      "epoch": 0.00111,
      "grad_norm": 0.8871469803064874,
      "learning_rate": 0.000333,
      "loss": 8.3685,
      "step": 111
    },
    {
      "epoch": 0.00112,
      "grad_norm": 0.9236879668329372,
      "learning_rate": 0.00033600000000000004,
      "loss": 8.3463,
      "step": 112
    },
    {
      "epoch": 0.00113,
      "grad_norm": 0.9129864456504505,
      "learning_rate": 0.000339,
      "loss": 8.3063,
      "step": 113
    },
    {
      "epoch": 0.00114,
      "grad_norm": 0.8352263501003522,
      "learning_rate": 0.000342,
      "loss": 8.2966,
      "step": 114
    },
    {
      "epoch": 0.00115,
      "grad_norm": 0.8592375580090957,
      "learning_rate": 0.00034500000000000004,
      "loss": 8.2718,
      "step": 115
    },
    {
      "epoch": 0.00116,
      "grad_norm": 0.8674294753896091,
      "learning_rate": 0.000348,
      "loss": 8.2506,
      "step": 116
    },
    {
      "epoch": 0.00117,
      "grad_norm": 0.8276917493567776,
      "learning_rate": 0.000351,
      "loss": 8.2188,
      "step": 117
    },
    {
      "epoch": 0.00118,
      "grad_norm": 0.8476744963131545,
      "learning_rate": 0.000354,
      "loss": 8.2045,
      "step": 118
    },
    {
      "epoch": 0.00119,
      "grad_norm": 0.844358071644388,
      "learning_rate": 0.000357,
      "loss": 8.1926,
      "step": 119
    },
    {
      "epoch": 0.0012,
      "grad_norm": 0.8166594269287538,
      "learning_rate": 0.00035999999999999997,
      "loss": 8.1658,
      "step": 120
    },
    {
      "epoch": 0.00121,
      "grad_norm": 0.8239930081024902,
      "learning_rate": 0.000363,
      "loss": 8.1389,
      "step": 121
    },
    {
      "epoch": 0.00122,
      "grad_norm": 0.8099951959348987,
      "learning_rate": 0.000366,
      "loss": 8.1225,
      "step": 122
    },
    {
      "epoch": 0.00123,
      "grad_norm": 0.830800320388625,
      "learning_rate": 0.000369,
      "loss": 8.1005,
      "step": 123
    },
    {
      "epoch": 0.00124,
      "grad_norm": 0.8139169053139192,
      "learning_rate": 0.000372,
      "loss": 8.0791,
      "step": 124
    },
    {
      "epoch": 0.00125,
      "grad_norm": 0.8112246790149765,
      "learning_rate": 0.000375,
      "loss": 8.0547,
      "step": 125
    },
    {
      "epoch": 0.00126,
      "grad_norm": 0.7922278873371895,
      "learning_rate": 0.000378,
      "loss": 8.0424,
      "step": 126
    },
    {
      "epoch": 0.00127,
      "grad_norm": 0.7755075943975184,
      "learning_rate": 0.000381,
      "loss": 8.0182,
      "step": 127
    },
    {
      "epoch": 0.00128,
      "grad_norm": 0.8028212720713388,
      "learning_rate": 0.000384,
      "loss": 8.0,
      "step": 128
    },
    {
      "epoch": 0.00129,
      "grad_norm": 0.8765129391436198,
      "learning_rate": 0.00038700000000000003,
      "loss": 7.9904,
      "step": 129
    },
    {
      "epoch": 0.0013,
      "grad_norm": 1.1486399983200042,
      "learning_rate": 0.00039000000000000005,
      "loss": 7.9724,
      "step": 130
    },
    {
      "epoch": 0.00131,
      "grad_norm": 1.0507777578095836,
      "learning_rate": 0.000393,
      "loss": 7.9382,
      "step": 131
    },
    {
      "epoch": 0.00132,
      "grad_norm": 0.7511305165281239,
      "learning_rate": 0.00039600000000000003,
      "loss": 7.9248,
      "step": 132
    },
    {
      "epoch": 0.00133,
      "grad_norm": 0.8108077692812641,
      "learning_rate": 0.00039900000000000005,
      "loss": 7.906,
      "step": 133
    },
    {
      "epoch": 0.00134,
      "grad_norm": 0.8301313874662418,
      "learning_rate": 0.000402,
      "loss": 7.8952,
      "step": 134
    },
    {
      "epoch": 0.00135,
      "grad_norm": 0.7123568807247732,
      "learning_rate": 0.00040500000000000003,
      "loss": 7.8651,
      "step": 135
    },
    {
      "epoch": 0.00136,
      "grad_norm": 0.7697556904537746,
      "learning_rate": 0.00040800000000000005,
      "loss": 7.8515,
      "step": 136
    },
    {
      "epoch": 0.00137,
      "grad_norm": 0.7190977621725152,
      "learning_rate": 0.000411,
      "loss": 7.8299,
      "step": 137
    },
    {
      "epoch": 0.00138,
      "grad_norm": 0.7147305335216294,
      "learning_rate": 0.00041400000000000003,
      "loss": 7.8066,
      "step": 138
    },
    {
      "epoch": 0.00139,
      "grad_norm": 0.747529428120578,
      "learning_rate": 0.00041700000000000005,
      "loss": 7.7954,
      "step": 139
    },
    {
      "epoch": 0.0014,
      "grad_norm": 0.6748198408281931,
      "learning_rate": 0.00042000000000000007,
      "loss": 7.7774,
      "step": 140
    },
    {
      "epoch": 0.00141,
      "grad_norm": 0.662142968172009,
      "learning_rate": 0.000423,
      "loss": 7.7644,
      "step": 141
    },
    {
      "epoch": 0.00142,
      "grad_norm": 0.6859796391897652,
      "learning_rate": 0.00042599999999999995,
      "loss": 7.7534,
      "step": 142
    },
    {
      "epoch": 0.00143,
      "grad_norm": 0.6828514310354903,
      "learning_rate": 0.00042899999999999997,
      "loss": 7.7255,
      "step": 143
    },
    {
      "epoch": 0.00144,
      "grad_norm": 0.6490687938821236,
      "learning_rate": 0.000432,
      "loss": 7.7078,
      "step": 144
    },
    {
      "epoch": 0.00145,
      "grad_norm": 0.6453156151137228,
      "learning_rate": 0.000435,
      "loss": 7.7035,
      "step": 145
    },
    {
      "epoch": 0.00146,
      "grad_norm": 0.7166050341593803,
      "learning_rate": 0.00043799999999999997,
      "loss": 7.6672,
      "step": 146
    },
    {
      "epoch": 0.00147,
      "grad_norm": 0.7443563124698165,
      "learning_rate": 0.000441,
      "loss": 7.6627,
      "step": 147
    },
    {
      "epoch": 0.00148,
      "grad_norm": 0.7124453009383569,
      "learning_rate": 0.000444,
      "loss": 7.648,
      "step": 148
    },
    {
      "epoch": 0.00149,
      "grad_norm": 0.6775306354557482,
      "learning_rate": 0.00044699999999999997,
      "loss": 7.6419,
      "step": 149
    },
    {
      "epoch": 0.0015,
      "grad_norm": 0.8906380813028638,
      "learning_rate": 0.00045,
      "loss": 7.6135,
      "step": 150
    },
    {
      "epoch": 0.00151,
      "grad_norm": 1.2892607147030477,
      "learning_rate": 0.000453,
      "loss": 7.605,
      "step": 151
    },
    {
      "epoch": 0.00152,
      "grad_norm": 0.6492280537852009,
      "learning_rate": 0.000456,
      "loss": 7.5875,
      "step": 152
    },
    {
      "epoch": 0.00153,
      "grad_norm": 0.8226748559002907,
      "learning_rate": 0.000459,
      "loss": 7.5783,
      "step": 153
    },
    {
      "epoch": 0.00154,
      "grad_norm": 1.073498340899344,
      "learning_rate": 0.000462,
      "loss": 7.5662,
      "step": 154
    },
    {
      "epoch": 0.00155,
      "grad_norm": 0.772416311968319,
      "learning_rate": 0.000465,
      "loss": 7.5481,
      "step": 155
    },
    {
      "epoch": 0.00156,
      "grad_norm": 0.7999364848474875,
      "learning_rate": 0.000468,
      "loss": 7.5209,
      "step": 156
    },
    {
      "epoch": 0.00157,
      "grad_norm": 0.9435570004081977,
      "learning_rate": 0.000471,
      "loss": 7.5196,
      "step": 157
    },
    {
      "epoch": 0.00158,
      "grad_norm": 0.8364296006508578,
      "learning_rate": 0.00047400000000000003,
      "loss": 7.4997,
      "step": 158
    },
    {
      "epoch": 0.00159,
      "grad_norm": 0.5596750978008624,
      "learning_rate": 0.000477,
      "loss": 7.4825,
      "step": 159
    },
    {
      "epoch": 0.0016,
      "grad_norm": 0.5748856794890025,
      "learning_rate": 0.00048,
      "loss": 7.4616,
      "step": 160
    },
    {
      "epoch": 0.00161,
      "grad_norm": 0.5986411025981033,
      "learning_rate": 0.00048300000000000003,
      "loss": 7.4419,
      "step": 161
    },
    {
      "epoch": 0.00162,
      "grad_norm": 0.5117747222458712,
      "learning_rate": 0.00048600000000000005,
      "loss": 7.4429,
      "step": 162
    },
    {
      "epoch": 0.00163,
      "grad_norm": 0.5509815158764758,
      "learning_rate": 0.0004890000000000001,
      "loss": 7.4259,
      "step": 163
    },
    {
      "epoch": 0.00164,
      "grad_norm": 0.46393808675226217,
      "learning_rate": 0.000492,
      "loss": 7.4236,
      "step": 164
    },
    {
      "epoch": 0.00165,
      "grad_norm": 0.5390679177469344,
      "learning_rate": 0.000495,
      "loss": 7.4006,
      "step": 165
    },
    {
      "epoch": 0.00166,
      "grad_norm": 0.645354949444588,
      "learning_rate": 0.0004980000000000001,
      "loss": 7.3773,
      "step": 166
    },
    {
      "epoch": 0.00167,
      "grad_norm": 1.0877276724000633,
      "learning_rate": 0.000501,
      "loss": 7.3741,
      "step": 167
    },
    {
      "epoch": 0.00168,
      "grad_norm": 1.0909247587015876,
      "learning_rate": 0.000504,
      "loss": 7.3697,
      "step": 168
    },
    {
      "epoch": 0.00169,
      "grad_norm": 0.5084368948335112,
      "learning_rate": 0.0005070000000000001,
      "loss": 7.3418,
      "step": 169
    },
    {
      "epoch": 0.0017,
      "grad_norm": 1.5053615947271437,
      "learning_rate": 0.00051,
      "loss": 7.3492,
      "step": 170
    },
    {
      "epoch": 0.00171,
      "grad_norm": 0.6120595065498764,
      "learning_rate": 0.000513,
      "loss": 7.3094,
      "step": 171
    },
    {
      "epoch": 0.00172,
      "grad_norm": 0.9401489103136018,
      "learning_rate": 0.000516,
      "loss": 7.3189,
      "step": 172
    },
    {
      "epoch": 0.00173,
      "grad_norm": 0.5558816443971099,
      "learning_rate": 0.0005189999999999999,
      "loss": 7.2906,
      "step": 173
    },
    {
      "epoch": 0.00174,
      "grad_norm": 0.7810658223347703,
      "learning_rate": 0.000522,
      "loss": 7.314,
      "step": 174
    },
    {
      "epoch": 0.00175,
      "grad_norm": 0.5409187817835382,
      "learning_rate": 0.000525,
      "loss": 7.2727,
      "step": 175
    },
    {
      "epoch": 0.00176,
      "grad_norm": 0.630636606756854,
      "learning_rate": 0.0005279999999999999,
      "loss": 7.2484,
      "step": 176
    },
    {
      "epoch": 0.00177,
      "grad_norm": 0.5206138592215499,
      "learning_rate": 0.000531,
      "loss": 7.2507,
      "step": 177
    },
    {
      "epoch": 0.00178,
      "grad_norm": 0.5577166707239251,
      "learning_rate": 0.000534,
      "loss": 7.2428,
      "step": 178
    },
    {
      "epoch": 0.00179,
      "grad_norm": 0.4980094016420717,
      "learning_rate": 0.000537,
      "loss": 7.2333,
      "step": 179
    },
    {
      "epoch": 0.0018,
      "grad_norm": 0.5436132024149352,
      "learning_rate": 0.00054,
      "loss": 7.2053,
      "step": 180
    },
    {
      "epoch": 0.00181,
      "grad_norm": 0.4850436660416719,
      "learning_rate": 0.000543,
      "loss": 7.1915,
      "step": 181
    },
    {
      "epoch": 0.00182,
      "grad_norm": 0.686701132410735,
      "learning_rate": 0.000546,
      "loss": 7.1909,
      "step": 182
    },
    {
      "epoch": 0.00183,
      "grad_norm": 0.68400997444651,
      "learning_rate": 0.000549,
      "loss": 7.1802,
      "step": 183
    },
    {
      "epoch": 0.00184,
      "grad_norm": 0.8396027257351396,
      "learning_rate": 0.000552,
      "loss": 7.171,
      "step": 184
    },
    {
      "epoch": 0.00185,
      "grad_norm": 0.7665365037410753,
      "learning_rate": 0.000555,
      "loss": 7.1476,
      "step": 185
    },
    {
      "epoch": 0.00186,
      "grad_norm": 0.5359175667027454,
      "learning_rate": 0.000558,
      "loss": 7.1506,
      "step": 186
    },
    {
      "epoch": 0.00187,
      "grad_norm": 0.5513800735908173,
      "learning_rate": 0.000561,
      "loss": 7.1261,
      "step": 187
    },
    {
      "epoch": 0.00188,
      "grad_norm": 0.5352093004255375,
      "learning_rate": 0.000564,
      "loss": 7.1044,
      "step": 188
    },
    {
      "epoch": 0.00189,
      "grad_norm": 0.5938457818726526,
      "learning_rate": 0.000567,
      "loss": 7.1004,
      "step": 189
    },
    {
      "epoch": 0.0019,
      "grad_norm": 0.5273842405001533,
      "learning_rate": 0.00057,
      "loss": 7.0834,
      "step": 190
    },
    {
      "epoch": 0.00191,
      "grad_norm": 0.47487115515279366,
      "learning_rate": 0.000573,
      "loss": 7.0721,
      "step": 191
    },
    {
      "epoch": 0.00192,
      "grad_norm": 0.6036837698051599,
      "learning_rate": 0.000576,
      "loss": 7.0655,
      "step": 192
    },
    {
      "epoch": 0.00193,
      "grad_norm": 0.3951184212196986,
      "learning_rate": 0.000579,
      "loss": 7.061,
      "step": 193
    },
    {
      "epoch": 0.00194,
      "grad_norm": 0.442083567688087,
      "learning_rate": 0.000582,
      "loss": 7.0548,
      "step": 194
    },
    {
      "epoch": 0.00195,
      "grad_norm": 0.439546882682468,
      "learning_rate": 0.000585,
      "loss": 7.0348,
      "step": 195
    },
    {
      "epoch": 0.00196,
      "grad_norm": 0.46247531692771043,
      "learning_rate": 0.000588,
      "loss": 7.0228,
      "step": 196
    },
    {
      "epoch": 0.00197,
      "grad_norm": 0.4140335072217301,
      "learning_rate": 0.000591,
      "loss": 7.0171,
      "step": 197
    },
    {
      "epoch": 0.00198,
      "grad_norm": 0.3685986320410548,
      "learning_rate": 0.000594,
      "loss": 7.0081,
      "step": 198
    },
    {
      "epoch": 0.00199,
      "grad_norm": 0.4020373564129086,
      "learning_rate": 0.0005970000000000001,
      "loss": 6.9898,
      "step": 199
    },
    {
      "epoch": 0.002,
      "grad_norm": 0.37126410475941546,
      "learning_rate": 0.0006000000000000001,
      "loss": 6.9867,
      "step": 200
    },
    {
      "epoch": 0.00201,
      "grad_norm": 0.3773154493828028,
      "learning_rate": 0.000603,
      "loss": 6.9617,
      "step": 201
    },
    {
      "epoch": 0.00202,
      "grad_norm": 0.3540017416986532,
      "learning_rate": 0.0006060000000000001,
      "loss": 6.9491,
      "step": 202
    },
    {
      "epoch": 0.00203,
      "grad_norm": 0.403279648640721,
      "learning_rate": 0.0006090000000000001,
      "loss": 6.9534,
      "step": 203
    },
    {
      "epoch": 0.00204,
      "grad_norm": 0.5112949618253247,
      "learning_rate": 0.000612,
      "loss": 6.9385,
      "step": 204
    },
    {
      "epoch": 0.00205,
      "grad_norm": 0.7200998739972175,
      "learning_rate": 0.000615,
      "loss": 6.931,
      "step": 205
    },
    {
      "epoch": 0.00206,
      "grad_norm": 1.209379808685074,
      "learning_rate": 0.000618,
      "loss": 6.9351,
      "step": 206
    },
    {
      "epoch": 0.00207,
      "grad_norm": 1.0040442357645134,
      "learning_rate": 0.000621,
      "loss": 6.9279,
      "step": 207
    },
    {
      "epoch": 0.00208,
      "grad_norm": 0.7878464521205251,
      "learning_rate": 0.000624,
      "loss": 6.9129,
      "step": 208
    },
    {
      "epoch": 0.00209,
      "grad_norm": 1.3096135062236434,
      "learning_rate": 0.000627,
      "loss": 6.9067,
      "step": 209
    },
    {
      "epoch": 0.0021,
      "grad_norm": 0.910504669978176,
      "learning_rate": 0.00063,
      "loss": 6.8878,
      "step": 210
    },
    {
      "epoch": 0.00211,
      "grad_norm": 1.3015817966038044,
      "learning_rate": 0.000633,
      "loss": 6.8987,
      "step": 211
    },
    {
      "epoch": 0.00212,
      "grad_norm": 0.5587575104994011,
      "learning_rate": 0.000636,
      "loss": 6.8696,
      "step": 212
    },
    {
      "epoch": 0.00213,
      "grad_norm": 0.8081412049208773,
      "learning_rate": 0.000639,
      "loss": 6.8608,
      "step": 213
    },
    {
      "epoch": 0.00214,
      "grad_norm": 0.6397075273457759,
      "learning_rate": 0.000642,
      "loss": 6.8511,
      "step": 214
    },
    {
      "epoch": 0.00215,
      "grad_norm": 0.5875043250740225,
      "learning_rate": 0.000645,
      "loss": 6.8482,
      "step": 215
    },
    {
      "epoch": 0.00216,
      "grad_norm": 0.6060774535893669,
      "learning_rate": 0.000648,
      "loss": 6.8413,
      "step": 216
    },
    {
      "epoch": 0.00217,
      "grad_norm": 0.5183751970166313,
      "learning_rate": 0.000651,
      "loss": 6.8119,
      "step": 217
    },
    {
      "epoch": 0.00218,
      "grad_norm": 0.5539011900924167,
      "learning_rate": 0.000654,
      "loss": 6.8171,
      "step": 218
    },
    {
      "epoch": 0.00219,
      "grad_norm": 0.5004312163301685,
      "learning_rate": 0.000657,
      "loss": 6.7882,
      "step": 219
    },
    {
      "epoch": 0.0022,
      "grad_norm": 0.4762494747244133,
      "learning_rate": 0.00066,
      "loss": 6.8062,
      "step": 220
    },
    {
      "epoch": 0.00221,
      "grad_norm": 0.353452768224107,
      "learning_rate": 0.0006630000000000001,
      "loss": 6.7814,
      "step": 221
    },
    {
      "epoch": 0.00222,
      "grad_norm": 0.3879096289808107,
      "learning_rate": 0.000666,
      "loss": 6.7696,
      "step": 222
    },
    {
      "epoch": 0.00223,
      "grad_norm": 0.4253518811476648,
      "learning_rate": 0.000669,
      "loss": 6.7658,
      "step": 223
    },
    {
      "epoch": 0.00224,
      "grad_norm": 0.4076338977034843,
      "learning_rate": 0.0006720000000000001,
      "loss": 6.7609,
      "step": 224
    },
    {
      "epoch": 0.00225,
      "grad_norm": 0.4491881195721512,
      "learning_rate": 0.000675,
      "loss": 6.7489,
      "step": 225
    },
    {
      "epoch": 0.00226,
      "grad_norm": 0.4195112098951784,
      "learning_rate": 0.000678,
      "loss": 6.7444,
      "step": 226
    },
    {
      "epoch": 0.00227,
      "grad_norm": 0.371663908330708,
      "learning_rate": 0.0006810000000000001,
      "loss": 6.7174,
      "step": 227
    },
    {
      "epoch": 0.00228,
      "grad_norm": 0.3462952066328263,
      "learning_rate": 0.000684,
      "loss": 6.7197,
      "step": 228
    },
    {
      "epoch": 0.00229,
      "grad_norm": 0.5288966810878937,
      "learning_rate": 0.000687,
      "loss": 6.7178,
      "step": 229
    },
    {
      "epoch": 0.0023,
      "grad_norm": 0.7623592789505088,
      "learning_rate": 0.0006900000000000001,
      "loss": 6.6993,
      "step": 230
    },
    {
      "epoch": 0.00231,
      "grad_norm": 0.9880780315432149,
      "learning_rate": 0.000693,
      "loss": 6.6923,
      "step": 231
    },
    {
      "epoch": 0.00232,
      "grad_norm": 1.028691756937799,
      "learning_rate": 0.000696,
      "loss": 6.7142,
      "step": 232
    },
    {
      "epoch": 0.00233,
      "grad_norm": 0.9501761132787978,
      "learning_rate": 0.0006990000000000001,
      "loss": 6.6946,
      "step": 233
    },
    {
      "epoch": 0.00234,
      "grad_norm": 0.9999298841530961,
      "learning_rate": 0.000702,
      "loss": 6.6948,
      "step": 234
    },
    {
      "epoch": 0.00235,
      "grad_norm": 0.6945930996245869,
      "learning_rate": 0.000705,
      "loss": 6.6675,
      "step": 235
    },
    {
      "epoch": 0.00236,
      "grad_norm": 0.6422608041910052,
      "learning_rate": 0.000708,
      "loss": 6.6513,
      "step": 236
    },
    {
      "epoch": 0.00237,
      "grad_norm": 0.47985797682797676,
      "learning_rate": 0.0007109999999999999,
      "loss": 6.657,
      "step": 237
    },
    {
      "epoch": 0.00238,
      "grad_norm": 0.6191094472441181,
      "learning_rate": 0.000714,
      "loss": 6.6482,
      "step": 238
    },
    {
      "epoch": 0.00239,
      "grad_norm": 0.5181523072026278,
      "learning_rate": 0.000717,
      "loss": 6.628,
      "step": 239
    },
    {
      "epoch": 0.0024,
      "grad_norm": 0.5894783578801835,
      "learning_rate": 0.0007199999999999999,
      "loss": 6.645,
      "step": 240
    },
    {
      "epoch": 0.00241,
      "grad_norm": 0.5639698176641863,
      "learning_rate": 0.000723,
      "loss": 6.6279,
      "step": 241
    },
    {
      "epoch": 0.00242,
      "grad_norm": 0.9095170652542525,
      "learning_rate": 0.000726,
      "loss": 6.6087,
      "step": 242
    },
    {
      "epoch": 0.00243,
      "grad_norm": 1.3373514416355459,
      "learning_rate": 0.000729,
      "loss": 6.6191,
      "step": 243
    },
    {
      "epoch": 0.00244,
      "grad_norm": 0.5837452319331187,
      "learning_rate": 0.000732,
      "loss": 6.5991,
      "step": 244
    },
    {
      "epoch": 0.00245,
      "grad_norm": 1.0261084178764917,
      "learning_rate": 0.000735,
      "loss": 6.6035,
      "step": 245
    },
    {
      "epoch": 0.00246,
      "grad_norm": 0.5663979442820745,
      "learning_rate": 0.000738,
      "loss": 6.574,
      "step": 246
    },
    {
      "epoch": 0.00247,
      "grad_norm": 0.5896636858778472,
      "learning_rate": 0.000741,
      "loss": 6.5719,
      "step": 247
    },
    {
      "epoch": 0.00248,
      "grad_norm": 0.47326080430149503,
      "learning_rate": 0.000744,
      "loss": 6.5757,
      "step": 248
    },
    {
      "epoch": 0.00249,
      "grad_norm": 0.6248164732961499,
      "learning_rate": 0.000747,
      "loss": 6.5582,
      "step": 249
    },
    {
      "epoch": 0.0025,
      "grad_norm": 0.48362025373458484,
      "learning_rate": 0.00075,
      "loss": 6.5567,
      "step": 250
    },
    {
      "epoch": 0.00251,
      "grad_norm": 0.4858269118610639,
      "learning_rate": 0.000753,
      "loss": 6.5227,
      "step": 251
    },
    {
      "epoch": 0.00252,
      "grad_norm": 0.41862369731289734,
      "learning_rate": 0.000756,
      "loss": 6.5302,
      "step": 252
    },
    {
      "epoch": 0.00253,
      "grad_norm": 0.47200180330590646,
      "learning_rate": 0.000759,
      "loss": 6.522,
      "step": 253
    },
    {
      "epoch": 0.00254,
      "grad_norm": 0.3651546358356223,
      "learning_rate": 0.000762,
      "loss": 6.5136,
      "step": 254
    },
    {
      "epoch": 0.00255,
      "grad_norm": 0.4285192093499382,
      "learning_rate": 0.0007650000000000001,
      "loss": 6.5149,
      "step": 255
    },
    {
      "epoch": 0.00256,
      "grad_norm": 0.42767490819993825,
      "learning_rate": 0.000768,
      "loss": 6.4917,
      "step": 256
    },
    {
      "epoch": 0.00257,
      "grad_norm": 0.48143237273687123,
      "learning_rate": 0.000771,
      "loss": 6.4974,
      "step": 257
    },
    {
      "epoch": 0.00258,
      "grad_norm": 0.5467090762573645,
      "learning_rate": 0.0007740000000000001,
      "loss": 6.4744,
      "step": 258
    },
    {
      "epoch": 0.00259,
      "grad_norm": 0.6750367098607939,
      "learning_rate": 0.000777,
      "loss": 6.4781,
      "step": 259
    },
    {
      "epoch": 0.0026,
      "grad_norm": 0.8320211245129605,
      "learning_rate": 0.0007800000000000001,
      "loss": 6.4681,
      "step": 260
    },
    {
      "epoch": 0.00261,
      "grad_norm": 0.905841412497731,
      "learning_rate": 0.0007830000000000001,
      "loss": 6.4809,
      "step": 261
    },
    {
      "epoch": 0.00262,
      "grad_norm": 1.0687360869781928,
      "learning_rate": 0.000786,
      "loss": 6.4647,
      "step": 262
    },
    {
      "epoch": 0.00263,
      "grad_norm": 1.2051408681263374,
      "learning_rate": 0.0007890000000000001,
      "loss": 6.4874,
      "step": 263
    },
    {
      "epoch": 0.00264,
      "grad_norm": 0.8690142485653533,
      "learning_rate": 0.0007920000000000001,
      "loss": 6.455,
      "step": 264
    },
    {
      "epoch": 0.00265,
      "grad_norm": 1.2774066489819682,
      "learning_rate": 0.000795,
      "loss": 6.4641,
      "step": 265
    },
    {
      "epoch": 0.00266,
      "grad_norm": 0.7288978979341997,
      "learning_rate": 0.0007980000000000001,
      "loss": 6.4454,
      "step": 266
    },
    {
      "epoch": 0.00267,
      "grad_norm": 0.6506287971604123,
      "learning_rate": 0.0008010000000000001,
      "loss": 6.4371,
      "step": 267
    },
    {
      "epoch": 0.00268,
      "grad_norm": 0.6866545943797145,
      "learning_rate": 0.000804,
      "loss": 6.4338,
      "step": 268
    },
    {
      "epoch": 0.00269,
      "grad_norm": 1.0440114151727509,
      "learning_rate": 0.0008070000000000001,
      "loss": 6.4227,
      "step": 269
    },
    {
      "epoch": 0.0027,
      "grad_norm": 1.170351969791303,
      "learning_rate": 0.0008100000000000001,
      "loss": 6.4362,
      "step": 270
    },
    {
      "epoch": 0.00271,
      "grad_norm": 0.6832222674646221,
      "learning_rate": 0.000813,
      "loss": 6.4253,
      "step": 271
    },
    {
      "epoch": 0.00272,
      "grad_norm": 0.617936452008115,
      "learning_rate": 0.0008160000000000001,
      "loss": 6.4063,
      "step": 272
    },
    {
      "epoch": 0.00273,
      "grad_norm": 0.6750066600068259,
      "learning_rate": 0.0008190000000000001,
      "loss": 6.4008,
      "step": 273
    },
    {
      "epoch": 0.00274,
      "grad_norm": 0.6315560378177079,
      "learning_rate": 0.000822,
      "loss": 6.379,
      "step": 274
    },
    {
      "epoch": 0.00275,
      "grad_norm": 0.6774266517760174,
      "learning_rate": 0.0008250000000000001,
      "loss": 6.3852,
      "step": 275
    },
    {
      "epoch": 0.00276,
      "grad_norm": 0.7272895822836024,
      "learning_rate": 0.0008280000000000001,
      "loss": 6.3794,
      "step": 276
    },
    {
      "epoch": 0.00277,
      "grad_norm": 0.6752471037637485,
      "learning_rate": 0.0008310000000000001,
      "loss": 6.3735,
      "step": 277
    },
    {
      "epoch": 0.00278,
      "grad_norm": 0.5678457826039285,
      "learning_rate": 0.0008340000000000001,
      "loss": 6.354,
      "step": 278
    },
    {
      "epoch": 0.00279,
      "grad_norm": 0.4611700189072147,
      "learning_rate": 0.0008370000000000001,
      "loss": 6.3529,
      "step": 279
    },
    {
      "epoch": 0.0028,
      "grad_norm": 0.343285643042232,
      "learning_rate": 0.0008400000000000001,
      "loss": 6.3329,
      "step": 280
    },
    {
      "epoch": 0.00281,
      "grad_norm": 0.4519631747446028,
      "learning_rate": 0.0008430000000000001,
      "loss": 6.3253,
      "step": 281
    },
    {
      "epoch": 0.00282,
      "grad_norm": 0.3255189118052276,
      "learning_rate": 0.000846,
      "loss": 6.3232,
      "step": 282
    },
    {
      "epoch": 0.00283,
      "grad_norm": 0.4297016476682907,
      "learning_rate": 0.0008489999999999999,
      "loss": 6.3149,
      "step": 283
    },
    {
      "epoch": 0.00284,
      "grad_norm": 0.37515936714697207,
      "learning_rate": 0.0008519999999999999,
      "loss": 6.3058,
      "step": 284
    },
    {
      "epoch": 0.00285,
      "grad_norm": 0.3458870104505622,
      "learning_rate": 0.000855,
      "loss": 6.3075,
      "step": 285
    },
    {
      "epoch": 0.00286,
      "grad_norm": 0.39943359732325345,
      "learning_rate": 0.0008579999999999999,
      "loss": 6.2857,
      "step": 286
    },
    {
      "epoch": 0.00287,
      "grad_norm": 0.3631751365570726,
      "learning_rate": 0.000861,
      "loss": 6.2875,
      "step": 287
    },
    {
      "epoch": 0.00288,
      "grad_norm": 0.552280032213235,
      "learning_rate": 0.000864,
      "loss": 6.2707,
      "step": 288
    },
    {
      "epoch": 0.00289,
      "grad_norm": 0.9396362484724781,
      "learning_rate": 0.0008669999999999999,
      "loss": 6.2923,
      "step": 289
    },
    {
      "epoch": 0.0029,
      "grad_norm": 1.3959756273937387,
      "learning_rate": 0.00087,
      "loss": 6.2945,
      "step": 290
    },
    {
      "epoch": 0.00291,
      "grad_norm": 0.6791923155853262,
      "learning_rate": 0.000873,
      "loss": 6.2663,
      "step": 291
    },
    {
      "epoch": 0.00292,
      "grad_norm": 0.9879732835250642,
      "learning_rate": 0.0008759999999999999,
      "loss": 6.2794,
      "step": 292
    },
    {
      "epoch": 0.00293,
      "grad_norm": 1.0140384817226566,
      "learning_rate": 0.000879,
      "loss": 6.2652,
      "step": 293
    },
    {
      "epoch": 0.00294,
      "grad_norm": 0.9889225551341856,
      "learning_rate": 0.000882,
      "loss": 6.2822,
      "step": 294
    },
    {
      "epoch": 0.00295,
      "grad_norm": 1.0429477871581094,
      "learning_rate": 0.0008849999999999999,
      "loss": 6.2468,
      "step": 295
    },
    {
      "epoch": 0.00296,
      "grad_norm": 1.1774929945794055,
      "learning_rate": 0.000888,
      "loss": 6.2705,
      "step": 296
    },
    {
      "epoch": 0.00297,
      "grad_norm": 0.7364156162079134,
      "learning_rate": 0.000891,
      "loss": 6.2278,
      "step": 297
    },
    {
      "epoch": 0.00298,
      "grad_norm": 0.8424638476384282,
      "learning_rate": 0.0008939999999999999,
      "loss": 6.2455,
      "step": 298
    },
    {
      "epoch": 0.00299,
      "grad_norm": 0.8668489286879963,
      "learning_rate": 0.000897,
      "loss": 6.2225,
      "step": 299
    },
    {
      "epoch": 0.003,
      "grad_norm": 0.9039057952602142,
      "learning_rate": 0.0009,
      "loss": 6.2236,
      "step": 300
    },
    {
      "epoch": 0.00301,
      "grad_norm": 1.0344858724084711,
      "learning_rate": 0.0009029999999999999,
      "loss": 6.2222,
      "step": 301
    },
    {
      "epoch": 0.00302,
      "grad_norm": 0.801399966246171,
      "learning_rate": 0.000906,
      "loss": 6.2007,
      "step": 302
    },
    {
      "epoch": 0.00303,
      "grad_norm": 0.7276255151675343,
      "learning_rate": 0.000909,
      "loss": 6.2106,
      "step": 303
    },
    {
      "epoch": 0.00304,
      "grad_norm": 0.8306162070729353,
      "learning_rate": 0.000912,
      "loss": 6.205,
      "step": 304
    },
    {
      "epoch": 0.00305,
      "grad_norm": 0.7650178489304597,
      "learning_rate": 0.000915,
      "loss": 6.2045,
      "step": 305
    },
    {
      "epoch": 0.00306,
      "grad_norm": 0.7024963687074245,
      "learning_rate": 0.000918,
      "loss": 6.1878,
      "step": 306
    },
    {
      "epoch": 0.00307,
      "grad_norm": 0.5687961336654864,
      "learning_rate": 0.000921,
      "loss": 6.1778,
      "step": 307
    },
    {
      "epoch": 0.00308,
      "grad_norm": 0.4515866134049927,
      "learning_rate": 0.000924,
      "loss": 6.1586,
      "step": 308
    },
    {
      "epoch": 0.00309,
      "grad_norm": 0.5454081565882548,
      "learning_rate": 0.000927,
      "loss": 6.1763,
      "step": 309
    },
    {
      "epoch": 0.0031,
      "grad_norm": 0.4033959865123679,
      "learning_rate": 0.00093,
      "loss": 6.1549,
      "step": 310
    },
    {
      "epoch": 0.00311,
      "grad_norm": 0.5004494164305024,
      "learning_rate": 0.000933,
      "loss": 6.1478,
      "step": 311
    },
    {
      "epoch": 0.00312,
      "grad_norm": 0.470361361901935,
      "learning_rate": 0.000936,
      "loss": 6.1275,
      "step": 312
    },
    {
      "epoch": 0.00313,
      "grad_norm": 0.4973667290148138,
      "learning_rate": 0.0009390000000000001,
      "loss": 6.1399,
      "step": 313
    },
    {
      "epoch": 0.00314,
      "grad_norm": 0.47398584855820086,
      "learning_rate": 0.000942,
      "loss": 6.1453,
      "step": 314
    },
    {
      "epoch": 0.00315,
      "grad_norm": 0.39081338977861474,
      "learning_rate": 0.000945,
      "loss": 6.1206,
      "step": 315
    },
    {
      "epoch": 0.00316,
      "grad_norm": 0.308452368547838,
      "learning_rate": 0.0009480000000000001,
      "loss": 6.0925,
      "step": 316
    },
    {
      "epoch": 0.00317,
      "grad_norm": 0.33735261718955184,
      "learning_rate": 0.000951,
      "loss": 6.1112,
      "step": 317
    },
    {
      "epoch": 0.00318,
      "grad_norm": 0.3843792856632324,
      "learning_rate": 0.000954,
      "loss": 6.1055,
      "step": 318
    },
    {
      "epoch": 0.00319,
      "grad_norm": 0.45015697169720664,
      "learning_rate": 0.0009570000000000001,
      "loss": 6.0951,
      "step": 319
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.6347836654615971,
      "learning_rate": 0.00096,
      "loss": 6.097,
      "step": 320
    },
    {
      "epoch": 0.00321,
      "grad_norm": 0.9264517947757075,
      "learning_rate": 0.000963,
      "loss": 6.0955,
      "step": 321
    },
    {
      "epoch": 0.00322,
      "grad_norm": 1.3311662164937155,
      "learning_rate": 0.0009660000000000001,
      "loss": 6.0984,
      "step": 322
    },
    {
      "epoch": 0.00323,
      "grad_norm": 0.7192768486088142,
      "learning_rate": 0.000969,
      "loss": 6.081,
      "step": 323
    },
    {
      "epoch": 0.00324,
      "grad_norm": 0.8815686637890671,
      "learning_rate": 0.0009720000000000001,
      "loss": 6.0921,
      "step": 324
    },
    {
      "epoch": 0.00325,
      "grad_norm": 0.9399377971403509,
      "learning_rate": 0.0009750000000000001,
      "loss": 6.0559,
      "step": 325
    },
    {
      "epoch": 0.00326,
      "grad_norm": 1.1161084705724094,
      "learning_rate": 0.0009780000000000001,
      "loss": 6.0866,
      "step": 326
    },
    {
      "epoch": 0.00327,
      "grad_norm": 0.9798577225908143,
      "learning_rate": 0.000981,
      "loss": 6.08,
      "step": 327
    },
    {
      "epoch": 0.00328,
      "grad_norm": 1.4594200076446016,
      "learning_rate": 0.000984,
      "loss": 6.0949,
      "step": 328
    },
    {
      "epoch": 0.00329,
      "grad_norm": 0.7530414187502621,
      "learning_rate": 0.000987,
      "loss": 6.0592,
      "step": 329
    },
    {
      "epoch": 0.0033,
      "grad_norm": 0.9983029089238832,
      "learning_rate": 0.00099,
      "loss": 6.0707,
      "step": 330
    },
    {
      "epoch": 0.00331,
      "grad_norm": 0.835334238631822,
      "learning_rate": 0.0009930000000000002,
      "loss": 6.0609,
      "step": 331
    },
    {
      "epoch": 0.00332,
      "grad_norm": 1.2633352311024129,
      "learning_rate": 0.0009960000000000001,
      "loss": 6.0632,
      "step": 332
    },
    {
      "epoch": 0.00333,
      "grad_norm": 0.9415989578137678,
      "learning_rate": 0.000999,
      "loss": 6.0388,
      "step": 333
    },
    {
      "epoch": 0.00334,
      "grad_norm": 0.752057040577036,
      "learning_rate": 0.001002,
      "loss": 6.0434,
      "step": 334
    },
    {
      "epoch": 0.00335,
      "grad_norm": 0.6368712711427076,
      "learning_rate": 0.001005,
      "loss": 6.039,
      "step": 335
    },
    {
      "epoch": 0.00336,
      "grad_norm": 0.5277321252477392,
      "learning_rate": 0.001008,
      "loss": 6.0255,
      "step": 336
    },
    {
      "epoch": 0.00337,
      "grad_norm": 0.545640031716998,
      "learning_rate": 0.0010110000000000002,
      "loss": 6.0051,
      "step": 337
    },
    {
      "epoch": 0.00338,
      "grad_norm": 0.5605817392374447,
      "learning_rate": 0.0010140000000000001,
      "loss": 6.0183,
      "step": 338
    },
    {
      "epoch": 0.00339,
      "grad_norm": 0.4963166785854256,
      "learning_rate": 0.0010170000000000001,
      "loss": 6.0214,
      "step": 339
    },
    {
      "epoch": 0.0034,
      "grad_norm": 0.496225629593103,
      "learning_rate": 0.00102,
      "loss": 5.9915,
      "step": 340
    },
    {
      "epoch": 0.00341,
      "grad_norm": 0.45013704339915594,
      "learning_rate": 0.001023,
      "loss": 5.9719,
      "step": 341
    },
    {
      "epoch": 0.00342,
      "grad_norm": 0.37899658405778774,
      "learning_rate": 0.001026,
      "loss": 5.9765,
      "step": 342
    },
    {
      "epoch": 0.00343,
      "grad_norm": 0.4216072327604619,
      "learning_rate": 0.0010290000000000002,
      "loss": 5.9773,
      "step": 343
    },
    {
      "epoch": 0.00344,
      "grad_norm": 0.4964437378733662,
      "learning_rate": 0.001032,
      "loss": 5.9678,
      "step": 344
    },
    {
      "epoch": 0.00345,
      "grad_norm": 0.6031142308234815,
      "learning_rate": 0.001035,
      "loss": 5.9605,
      "step": 345
    },
    {
      "epoch": 0.00346,
      "grad_norm": 0.6349778009023861,
      "learning_rate": 0.0010379999999999999,
      "loss": 5.9675,
      "step": 346
    },
    {
      "epoch": 0.00347,
      "grad_norm": 0.6213366174245898,
      "learning_rate": 0.001041,
      "loss": 5.9393,
      "step": 347
    },
    {
      "epoch": 0.00348,
      "grad_norm": 0.684857817555668,
      "learning_rate": 0.001044,
      "loss": 5.9499,
      "step": 348
    },
    {
      "epoch": 0.00349,
      "grad_norm": 0.8421661368559449,
      "learning_rate": 0.001047,
      "loss": 5.9398,
      "step": 349
    },
    {
      "epoch": 0.0035,
      "grad_norm": 0.8062802732253019,
      "learning_rate": 0.00105,
      "loss": 5.9568,
      "step": 350
    },
    {
      "epoch": 0.00351,
      "grad_norm": 0.7669273491138234,
      "learning_rate": 0.001053,
      "loss": 5.9549,
      "step": 351
    },
    {
      "epoch": 0.00352,
      "grad_norm": 1.0673384228730578,
      "learning_rate": 0.0010559999999999999,
      "loss": 5.9491,
      "step": 352
    },
    {
      "epoch": 0.00353,
      "grad_norm": 1.313057737158452,
      "learning_rate": 0.001059,
      "loss": 5.9596,
      "step": 353
    },
    {
      "epoch": 0.00354,
      "grad_norm": 0.8748293544965601,
      "learning_rate": 0.001062,
      "loss": 5.9228,
      "step": 354
    },
    {
      "epoch": 0.00355,
      "grad_norm": 0.7617258072087358,
      "learning_rate": 0.001065,
      "loss": 5.9348,
      "step": 355
    },
    {
      "epoch": 0.00356,
      "grad_norm": 0.9561923051718775,
      "learning_rate": 0.001068,
      "loss": 5.9306,
      "step": 356
    },
    {
      "epoch": 0.00357,
      "grad_norm": 0.8631216371911231,
      "learning_rate": 0.001071,
      "loss": 5.9199,
      "step": 357
    },
    {
      "epoch": 0.00358,
      "grad_norm": 0.9525953351992685,
      "learning_rate": 0.001074,
      "loss": 5.9184,
      "step": 358
    },
    {
      "epoch": 0.00359,
      "grad_norm": 0.6858307664974747,
      "learning_rate": 0.001077,
      "loss": 5.8973,
      "step": 359
    },
    {
      "epoch": 0.0036,
      "grad_norm": 0.7167037627907079,
      "learning_rate": 0.00108,
      "loss": 5.9065,
      "step": 360
    },
    {
      "epoch": 0.00361,
      "grad_norm": 0.7969047134484166,
      "learning_rate": 0.001083,
      "loss": 5.8986,
      "step": 361
    },
    {
      "epoch": 0.00362,
      "grad_norm": 0.8186204145394074,
      "learning_rate": 0.001086,
      "loss": 5.888,
      "step": 362
    },
    {
      "epoch": 0.00363,
      "grad_norm": 0.7436182013517663,
      "learning_rate": 0.001089,
      "loss": 5.8766,
      "step": 363
    },
    {
      "epoch": 0.00364,
      "grad_norm": 0.6658023261534547,
      "learning_rate": 0.001092,
      "loss": 5.8872,
      "step": 364
    },
    {
      "epoch": 0.00365,
      "grad_norm": 0.7206668214896482,
      "learning_rate": 0.001095,
      "loss": 5.8872,
      "step": 365
    },
    {
      "epoch": 0.00366,
      "grad_norm": 0.6345413922647961,
      "learning_rate": 0.001098,
      "loss": 5.8617,
      "step": 366
    },
    {
      "epoch": 0.00367,
      "grad_norm": 0.6094924231218852,
      "learning_rate": 0.001101,
      "loss": 5.8618,
      "step": 367
    },
    {
      "epoch": 0.00368,
      "grad_norm": 0.7732260654201254,
      "learning_rate": 0.001104,
      "loss": 5.8653,
      "step": 368
    },
    {
      "epoch": 0.00369,
      "grad_norm": 0.9451132874875877,
      "learning_rate": 0.001107,
      "loss": 5.8779,
      "step": 369
    },
    {
      "epoch": 0.0037,
      "grad_norm": 1.047387946938651,
      "learning_rate": 0.00111,
      "loss": 5.8603,
      "step": 370
    },
    {
      "epoch": 0.00371,
      "grad_norm": 0.8709497379832931,
      "learning_rate": 0.001113,
      "loss": 5.8262,
      "step": 371
    },
    {
      "epoch": 0.00372,
      "grad_norm": 0.6993621521276565,
      "learning_rate": 0.001116,
      "loss": 5.8434,
      "step": 372
    },
    {
      "epoch": 0.00373,
      "grad_norm": 0.6889019297226876,
      "learning_rate": 0.001119,
      "loss": 5.8344,
      "step": 373
    },
    {
      "epoch": 0.00374,
      "grad_norm": 0.8187808414499578,
      "learning_rate": 0.001122,
      "loss": 5.8424,
      "step": 374
    },
    {
      "epoch": 0.00375,
      "grad_norm": 0.7395556739972736,
      "learning_rate": 0.0011250000000000001,
      "loss": 5.8404,
      "step": 375
    },
    {
      "epoch": 0.00376,
      "grad_norm": 0.5027446756427529,
      "learning_rate": 0.001128,
      "loss": 5.8296,
      "step": 376
    },
    {
      "epoch": 0.00377,
      "grad_norm": 0.45297103786338255,
      "learning_rate": 0.001131,
      "loss": 5.8239,
      "step": 377
    },
    {
      "epoch": 0.00378,
      "grad_norm": 0.5150298222384522,
      "learning_rate": 0.001134,
      "loss": 5.82,
      "step": 378
    },
    {
      "epoch": 0.00379,
      "grad_norm": 0.4216428009711753,
      "learning_rate": 0.001137,
      "loss": 5.8036,
      "step": 379
    },
    {
      "epoch": 0.0038,
      "grad_norm": 0.43574801532624385,
      "learning_rate": 0.00114,
      "loss": 5.8234,
      "step": 380
    },
    {
      "epoch": 0.00381,
      "grad_norm": 0.4737821231317218,
      "learning_rate": 0.0011430000000000001,
      "loss": 5.7908,
      "step": 381
    },
    {
      "epoch": 0.00382,
      "grad_norm": 0.5236457098681065,
      "learning_rate": 0.001146,
      "loss": 5.7778,
      "step": 382
    },
    {
      "epoch": 0.00383,
      "grad_norm": 0.5006340354259897,
      "learning_rate": 0.001149,
      "loss": 5.787,
      "step": 383
    },
    {
      "epoch": 0.00384,
      "grad_norm": 0.5762330042414852,
      "learning_rate": 0.001152,
      "loss": 5.7915,
      "step": 384
    },
    {
      "epoch": 0.00385,
      "grad_norm": 0.6623935085819848,
      "learning_rate": 0.001155,
      "loss": 5.7808,
      "step": 385
    },
    {
      "epoch": 0.00386,
      "grad_norm": 0.6780931079980513,
      "learning_rate": 0.001158,
      "loss": 5.7864,
      "step": 386
    },
    {
      "epoch": 0.00387,
      "grad_norm": 0.6283557380392781,
      "learning_rate": 0.0011610000000000001,
      "loss": 5.76,
      "step": 387
    },
    {
      "epoch": 0.00388,
      "grad_norm": 0.8661947460887196,
      "learning_rate": 0.001164,
      "loss": 5.7642,
      "step": 388
    },
    {
      "epoch": 0.00389,
      "grad_norm": 1.1877567285678448,
      "learning_rate": 0.001167,
      "loss": 5.7898,
      "step": 389
    },
    {
      "epoch": 0.0039,
      "grad_norm": 0.5285139722189788,
      "learning_rate": 0.00117,
      "loss": 5.7491,
      "step": 390
    },
    {
      "epoch": 0.00391,
      "grad_norm": 0.6430118692881639,
      "learning_rate": 0.001173,
      "loss": 5.7622,
      "step": 391
    },
    {
      "epoch": 0.00392,
      "grad_norm": 0.6748100403552353,
      "learning_rate": 0.001176,
      "loss": 5.7713,
      "step": 392
    },
    {
      "epoch": 0.00393,
      "grad_norm": 0.7344199345621275,
      "learning_rate": 0.0011790000000000001,
      "loss": 5.7448,
      "step": 393
    },
    {
      "epoch": 0.00394,
      "grad_norm": 0.8611653440542995,
      "learning_rate": 0.001182,
      "loss": 5.7377,
      "step": 394
    },
    {
      "epoch": 0.00395,
      "grad_norm": 1.0200403565527223,
      "learning_rate": 0.001185,
      "loss": 5.7297,
      "step": 395
    },
    {
      "epoch": 0.00396,
      "grad_norm": 1.3219082304761296,
      "learning_rate": 0.001188,
      "loss": 5.7571,
      "step": 396
    },
    {
      "epoch": 0.00397,
      "grad_norm": 1.0648841185793536,
      "learning_rate": 0.001191,
      "loss": 5.7452,
      "step": 397
    },
    {
      "epoch": 0.00398,
      "grad_norm": 0.8818852137031271,
      "learning_rate": 0.0011940000000000002,
      "loss": 5.7476,
      "step": 398
    },
    {
      "epoch": 0.00399,
      "grad_norm": 0.7229082271352473,
      "learning_rate": 0.0011970000000000001,
      "loss": 5.7455,
      "step": 399
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.7464785543636239,
      "learning_rate": 0.0012000000000000001,
      "loss": 5.7335,
      "step": 400
    },
    {
      "epoch": 0.00401,
      "grad_norm": 0.788008045277313,
      "learning_rate": 0.001203,
      "loss": 5.7147,
      "step": 401
    },
    {
      "epoch": 0.00402,
      "grad_norm": 0.8844811042429299,
      "learning_rate": 0.001206,
      "loss": 5.7255,
      "step": 402
    },
    {
      "epoch": 0.00403,
      "grad_norm": 0.8786909754931423,
      "learning_rate": 0.001209,
      "loss": 5.7201,
      "step": 403
    },
    {
      "epoch": 0.00404,
      "grad_norm": 0.9763167280044874,
      "learning_rate": 0.0012120000000000002,
      "loss": 5.7188,
      "step": 404
    },
    {
      "epoch": 0.00405,
      "grad_norm": 0.9708490732732808,
      "learning_rate": 0.0012150000000000002,
      "loss": 5.7216,
      "step": 405
    },
    {
      "epoch": 0.00406,
      "grad_norm": 0.7953769652219763,
      "learning_rate": 0.0012180000000000001,
      "loss": 5.6987,
      "step": 406
    },
    {
      "epoch": 0.00407,
      "grad_norm": 0.7231086032909518,
      "learning_rate": 0.0012209999999999999,
      "loss": 5.6945,
      "step": 407
    },
    {
      "epoch": 0.00408,
      "grad_norm": 0.6539281657127057,
      "learning_rate": 0.001224,
      "loss": 5.7108,
      "step": 408
    },
    {
      "epoch": 0.00409,
      "grad_norm": 0.48725126694443294,
      "learning_rate": 0.001227,
      "loss": 5.6827,
      "step": 409
    },
    {
      "epoch": 0.0041,
      "grad_norm": 0.5935989194477762,
      "learning_rate": 0.00123,
      "loss": 5.6771,
      "step": 410
    },
    {
      "epoch": 0.00411,
      "grad_norm": 0.6524672694061662,
      "learning_rate": 0.001233,
      "loss": 5.6945,
      "step": 411
    },
    {
      "epoch": 0.00412,
      "grad_norm": 0.5686252583506046,
      "learning_rate": 0.001236,
      "loss": 5.6796,
      "step": 412
    },
    {
      "epoch": 0.00413,
      "grad_norm": 0.5206794027062916,
      "learning_rate": 0.0012389999999999999,
      "loss": 5.676,
      "step": 413
    },
    {
      "epoch": 0.00414,
      "grad_norm": 0.5100681259570212,
      "learning_rate": 0.001242,
      "loss": 5.6561,
      "step": 414
    },
    {
      "epoch": 0.00415,
      "grad_norm": 0.5323494958966052,
      "learning_rate": 0.001245,
      "loss": 5.6572,
      "step": 415
    },
    {
      "epoch": 0.00416,
      "grad_norm": 0.593833682795931,
      "learning_rate": 0.001248,
      "loss": 5.6419,
      "step": 416
    },
    {
      "epoch": 0.00417,
      "grad_norm": 0.5881502982050868,
      "learning_rate": 0.001251,
      "loss": 5.6711,
      "step": 417
    },
    {
      "epoch": 0.00418,
      "grad_norm": 0.6571826832460801,
      "learning_rate": 0.001254,
      "loss": 5.6451,
      "step": 418
    },
    {
      "epoch": 0.00419,
      "grad_norm": 0.8242389183582979,
      "learning_rate": 0.0012569999999999999,
      "loss": 5.6393,
      "step": 419
    },
    {
      "epoch": 0.0042,
      "grad_norm": 0.7322278119135938,
      "learning_rate": 0.00126,
      "loss": 5.6531,
      "step": 420
    },
    {
      "epoch": 0.00421,
      "grad_norm": 0.6080486142741918,
      "learning_rate": 0.001263,
      "loss": 5.6415,
      "step": 421
    },
    {
      "epoch": 0.00422,
      "grad_norm": 0.6271805214461816,
      "learning_rate": 0.001266,
      "loss": 5.6327,
      "step": 422
    },
    {
      "epoch": 0.00423,
      "grad_norm": 0.5934268893913589,
      "learning_rate": 0.001269,
      "loss": 5.6286,
      "step": 423
    },
    {
      "epoch": 0.00424,
      "grad_norm": 0.6795296945522938,
      "learning_rate": 0.001272,
      "loss": 5.6299,
      "step": 424
    },
    {
      "epoch": 0.00425,
      "grad_norm": 0.7845276448161947,
      "learning_rate": 0.001275,
      "loss": 5.6344,
      "step": 425
    },
    {
      "epoch": 0.00426,
      "grad_norm": 0.9176562117983922,
      "learning_rate": 0.001278,
      "loss": 5.6402,
      "step": 426
    },
    {
      "epoch": 0.00427,
      "grad_norm": 1.0306205830362438,
      "learning_rate": 0.001281,
      "loss": 5.6293,
      "step": 427
    },
    {
      "epoch": 0.00428,
      "grad_norm": 1.0252381178312269,
      "learning_rate": 0.001284,
      "loss": 5.6086,
      "step": 428
    },
    {
      "epoch": 0.00429,
      "grad_norm": 1.3332162612637855,
      "learning_rate": 0.001287,
      "loss": 5.6393,
      "step": 429
    },
    {
      "epoch": 0.0043,
      "grad_norm": 0.8174291079939794,
      "learning_rate": 0.00129,
      "loss": 5.6202,
      "step": 430
    },
    {
      "epoch": 0.00431,
      "grad_norm": 0.821028463418781,
      "learning_rate": 0.001293,
      "loss": 5.6102,
      "step": 431
    },
    {
      "epoch": 0.00432,
      "grad_norm": 0.7475137161763143,
      "learning_rate": 0.001296,
      "loss": 5.6163,
      "step": 432
    },
    {
      "epoch": 0.00433,
      "grad_norm": 0.7571870052992741,
      "learning_rate": 0.001299,
      "loss": 5.604,
      "step": 433
    },
    {
      "epoch": 0.00434,
      "grad_norm": 1.0543560255015263,
      "learning_rate": 0.001302,
      "loss": 5.6184,
      "step": 434
    },
    {
      "epoch": 0.00435,
      "grad_norm": 0.8758758304266553,
      "learning_rate": 0.001305,
      "loss": 5.6066,
      "step": 435
    },
    {
      "epoch": 0.00436,
      "grad_norm": 0.9137062421440546,
      "learning_rate": 0.001308,
      "loss": 5.5859,
      "step": 436
    },
    {
      "epoch": 0.00437,
      "grad_norm": 1.301736875083812,
      "learning_rate": 0.001311,
      "loss": 5.6173,
      "step": 437
    },
    {
      "epoch": 0.00438,
      "grad_norm": 0.9356953917037294,
      "learning_rate": 0.001314,
      "loss": 5.5919,
      "step": 438
    },
    {
      "epoch": 0.00439,
      "grad_norm": 0.8522821992819578,
      "learning_rate": 0.001317,
      "loss": 5.6158,
      "step": 439
    },
    {
      "epoch": 0.0044,
      "grad_norm": 0.7182761753028103,
      "learning_rate": 0.00132,
      "loss": 5.5821,
      "step": 440
    },
    {
      "epoch": 0.00441,
      "grad_norm": 0.6293266892726601,
      "learning_rate": 0.001323,
      "loss": 5.577,
      "step": 441
    },
    {
      "epoch": 0.00442,
      "grad_norm": 0.8189921873128356,
      "learning_rate": 0.0013260000000000001,
      "loss": 5.5762,
      "step": 442
    },
    {
      "epoch": 0.00443,
      "grad_norm": 1.0212422821054057,
      "learning_rate": 0.001329,
      "loss": 5.5904,
      "step": 443
    },
    {
      "epoch": 0.00444,
      "grad_norm": 0.8951785018901781,
      "learning_rate": 0.001332,
      "loss": 5.5851,
      "step": 444
    },
    {
      "epoch": 0.00445,
      "grad_norm": 0.7868769140150608,
      "learning_rate": 0.001335,
      "loss": 5.5661,
      "step": 445
    },
    {
      "epoch": 0.00446,
      "grad_norm": 0.790102612629763,
      "learning_rate": 0.001338,
      "loss": 5.5718,
      "step": 446
    },
    {
      "epoch": 0.00447,
      "grad_norm": 0.8396194874372788,
      "learning_rate": 0.001341,
      "loss": 5.5716,
      "step": 447
    },
    {
      "epoch": 0.00448,
      "grad_norm": 0.9120841535821665,
      "learning_rate": 0.0013440000000000001,
      "loss": 5.5589,
      "step": 448
    },
    {
      "epoch": 0.00449,
      "grad_norm": 0.8573073152890212,
      "learning_rate": 0.001347,
      "loss": 5.5603,
      "step": 449
    },
    {
      "epoch": 0.0045,
      "grad_norm": 0.7420732830276576,
      "learning_rate": 0.00135,
      "loss": 5.5551,
      "step": 450
    },
    {
      "epoch": 0.00451,
      "grad_norm": 0.7054051762730813,
      "learning_rate": 0.001353,
      "loss": 5.5451,
      "step": 451
    },
    {
      "epoch": 0.00452,
      "grad_norm": 0.5383194985068459,
      "learning_rate": 0.001356,
      "loss": 5.5229,
      "step": 452
    },
    {
      "epoch": 0.00453,
      "grad_norm": 0.5845224529530345,
      "learning_rate": 0.001359,
      "loss": 5.5144,
      "step": 453
    },
    {
      "epoch": 0.00454,
      "grad_norm": 0.5197277965757966,
      "learning_rate": 0.0013620000000000001,
      "loss": 5.5269,
      "step": 454
    },
    {
      "epoch": 0.00455,
      "grad_norm": 0.44901215490386587,
      "learning_rate": 0.0013650000000000001,
      "loss": 5.5227,
      "step": 455
    },
    {
      "epoch": 0.00456,
      "grad_norm": 0.40823042735319937,
      "learning_rate": 0.001368,
      "loss": 5.5155,
      "step": 456
    },
    {
      "epoch": 0.00457,
      "grad_norm": 0.4077054091161063,
      "learning_rate": 0.001371,
      "loss": 5.5045,
      "step": 457
    },
    {
      "epoch": 0.00458,
      "grad_norm": 0.4051721280659754,
      "learning_rate": 0.001374,
      "loss": 5.5115,
      "step": 458
    },
    {
      "epoch": 0.00459,
      "grad_norm": 0.3819222938644956,
      "learning_rate": 0.0013770000000000002,
      "loss": 5.4986,
      "step": 459
    },
    {
      "epoch": 0.0046,
      "grad_norm": 0.4096791477469412,
      "learning_rate": 0.0013800000000000002,
      "loss": 5.4851,
      "step": 460
    },
    {
      "epoch": 0.00461,
      "grad_norm": 0.44167018915391465,
      "learning_rate": 0.0013830000000000001,
      "loss": 5.4805,
      "step": 461
    },
    {
      "epoch": 0.00462,
      "grad_norm": 0.49346826910496583,
      "learning_rate": 0.001386,
      "loss": 5.4928,
      "step": 462
    },
    {
      "epoch": 0.00463,
      "grad_norm": 0.463489633996114,
      "learning_rate": 0.001389,
      "loss": 5.4652,
      "step": 463
    },
    {
      "epoch": 0.00464,
      "grad_norm": 0.507205937400668,
      "learning_rate": 0.001392,
      "loss": 5.4859,
      "step": 464
    },
    {
      "epoch": 0.00465,
      "grad_norm": 0.5318639728571777,
      "learning_rate": 0.0013950000000000002,
      "loss": 5.4572,
      "step": 465
    },
    {
      "epoch": 0.00466,
      "grad_norm": 0.6489252257104292,
      "learning_rate": 0.0013980000000000002,
      "loss": 5.4792,
      "step": 466
    },
    {
      "epoch": 0.00467,
      "grad_norm": 0.7957789059873086,
      "learning_rate": 0.0014010000000000001,
      "loss": 5.4727,
      "step": 467
    },
    {
      "epoch": 0.00468,
      "grad_norm": 0.7621042481808248,
      "learning_rate": 0.001404,
      "loss": 5.462,
      "step": 468
    },
    {
      "epoch": 0.00469,
      "grad_norm": 0.573822393217689,
      "learning_rate": 0.001407,
      "loss": 5.457,
      "step": 469
    },
    {
      "epoch": 0.0047,
      "grad_norm": 0.7543746323219513,
      "learning_rate": 0.00141,
      "loss": 5.4709,
      "step": 470
    },
    {
      "epoch": 0.00471,
      "grad_norm": 0.8641651762100534,
      "learning_rate": 0.001413,
      "loss": 5.4567,
      "step": 471
    },
    {
      "epoch": 0.00472,
      "grad_norm": 0.9026674726231502,
      "learning_rate": 0.001416,
      "loss": 5.4699,
      "step": 472
    },
    {
      "epoch": 0.00473,
      "grad_norm": 0.8212885101559565,
      "learning_rate": 0.001419,
      "loss": 5.4778,
      "step": 473
    },
    {
      "epoch": 0.00474,
      "grad_norm": 0.8914030740906659,
      "learning_rate": 0.0014219999999999999,
      "loss": 5.461,
      "step": 474
    },
    {
      "epoch": 0.00475,
      "grad_norm": 1.0570094425693455,
      "learning_rate": 0.001425,
      "loss": 5.4652,
      "step": 475
    },
    {
      "epoch": 0.00476,
      "grad_norm": 0.9736444976311589,
      "learning_rate": 0.001428,
      "loss": 5.4875,
      "step": 476
    },
    {
      "epoch": 0.00477,
      "grad_norm": 1.1550380737092787,
      "learning_rate": 0.001431,
      "loss": 5.4568,
      "step": 477
    },
    {
      "epoch": 0.00478,
      "grad_norm": 0.848321570803796,
      "learning_rate": 0.001434,
      "loss": 5.4695,
      "step": 478
    },
    {
      "epoch": 0.00479,
      "grad_norm": 0.9517827225501269,
      "learning_rate": 0.001437,
      "loss": 5.4501,
      "step": 479
    },
    {
      "epoch": 0.0048,
      "grad_norm": 1.0883787540754652,
      "learning_rate": 0.0014399999999999999,
      "loss": 5.4562,
      "step": 480
    },
    {
      "epoch": 0.00481,
      "grad_norm": 0.9422991164230814,
      "learning_rate": 0.001443,
      "loss": 5.4516,
      "step": 481
    },
    {
      "epoch": 0.00482,
      "grad_norm": 0.9159499791385636,
      "learning_rate": 0.001446,
      "loss": 5.4273,
      "step": 482
    },
    {
      "epoch": 0.00483,
      "grad_norm": 0.9688645055474768,
      "learning_rate": 0.001449,
      "loss": 5.44,
      "step": 483
    },
    {
      "epoch": 0.00484,
      "grad_norm": 1.1114303023214132,
      "learning_rate": 0.001452,
      "loss": 5.4474,
      "step": 484
    },
    {
      "epoch": 0.00485,
      "grad_norm": 0.9051569573634253,
      "learning_rate": 0.001455,
      "loss": 5.4468,
      "step": 485
    },
    {
      "epoch": 0.00486,
      "grad_norm": 0.9247454458721566,
      "learning_rate": 0.001458,
      "loss": 5.4217,
      "step": 486
    },
    {
      "epoch": 0.00487,
      "grad_norm": 0.852928162562673,
      "learning_rate": 0.001461,
      "loss": 5.4339,
      "step": 487
    },
    {
      "epoch": 0.00488,
      "grad_norm": 0.8142513932978471,
      "learning_rate": 0.001464,
      "loss": 5.41,
      "step": 488
    },
    {
      "epoch": 0.00489,
      "grad_norm": 0.9329231917883894,
      "learning_rate": 0.001467,
      "loss": 5.403,
      "step": 489
    },
    {
      "epoch": 0.0049,
      "grad_norm": 0.9855383918059676,
      "learning_rate": 0.00147,
      "loss": 5.4217,
      "step": 490
    },
    {
      "epoch": 0.00491,
      "grad_norm": 0.7649189111405409,
      "learning_rate": 0.001473,
      "loss": 5.4131,
      "step": 491
    },
    {
      "epoch": 0.00492,
      "grad_norm": 0.7909135941762935,
      "learning_rate": 0.001476,
      "loss": 5.4088,
      "step": 492
    },
    {
      "epoch": 0.00493,
      "grad_norm": 0.9895574066894314,
      "learning_rate": 0.001479,
      "loss": 5.412,
      "step": 493
    },
    {
      "epoch": 0.00494,
      "grad_norm": 1.132991334341666,
      "learning_rate": 0.001482,
      "loss": 5.429,
      "step": 494
    },
    {
      "epoch": 0.00495,
      "grad_norm": 0.761507737933228,
      "learning_rate": 0.001485,
      "loss": 5.4134,
      "step": 495
    },
    {
      "epoch": 0.00496,
      "grad_norm": 0.9726159326361932,
      "learning_rate": 0.001488,
      "loss": 5.4067,
      "step": 496
    },
    {
      "epoch": 0.00497,
      "grad_norm": 1.1482864163713484,
      "learning_rate": 0.001491,
      "loss": 5.3997,
      "step": 497
    },
    {
      "epoch": 0.00498,
      "grad_norm": 0.8327332046897746,
      "learning_rate": 0.001494,
      "loss": 5.4147,
      "step": 498
    },
    {
      "epoch": 0.00499,
      "grad_norm": 0.725916611519047,
      "learning_rate": 0.001497,
      "loss": 5.3792,
      "step": 499
    },
    {
      "epoch": 0.005,
      "grad_norm": 0.6719782846245154,
      "learning_rate": 0.0015,
      "loss": 5.3842,
      "step": 500
    },
    {
      "epoch": 0.00501,
      "grad_norm": 0.7463076928465905,
      "learning_rate": 0.001503,
      "loss": 5.3796,
      "step": 501
    },
    {
      "epoch": 0.00502,
      "grad_norm": 0.8408104186601356,
      "learning_rate": 0.001506,
      "loss": 5.389,
      "step": 502
    },
    {
      "epoch": 0.00503,
      "grad_norm": 0.860909611402231,
      "learning_rate": 0.0015090000000000001,
      "loss": 5.3926,
      "step": 503
    },
    {
      "epoch": 0.00504,
      "grad_norm": 0.7026490274993983,
      "learning_rate": 0.001512,
      "loss": 5.3646,
      "step": 504
    },
    {
      "epoch": 0.00505,
      "grad_norm": 0.7184807991697565,
      "learning_rate": 0.001515,
      "loss": 5.3547,
      "step": 505
    },
    {
      "epoch": 0.00506,
      "grad_norm": 0.7839475253802514,
      "learning_rate": 0.001518,
      "loss": 5.3611,
      "step": 506
    },
    {
      "epoch": 0.00507,
      "grad_norm": 0.7039499110993044,
      "learning_rate": 0.001521,
      "loss": 5.3522,
      "step": 507
    },
    {
      "epoch": 0.00508,
      "grad_norm": 0.5587706859739108,
      "learning_rate": 0.001524,
      "loss": 5.3512,
      "step": 508
    },
    {
      "epoch": 0.00509,
      "grad_norm": 0.4952941325228141,
      "learning_rate": 0.0015270000000000001,
      "loss": 5.326,
      "step": 509
    },
    {
      "epoch": 0.0051,
      "grad_norm": 0.5131490664852795,
      "learning_rate": 0.0015300000000000001,
      "loss": 5.3428,
      "step": 510
    },
    {
      "epoch": 0.00511,
      "grad_norm": 0.5539213487194597,
      "learning_rate": 0.001533,
      "loss": 5.3196,
      "step": 511
    },
    {
      "epoch": 0.00512,
      "grad_norm": 0.5937876471208409,
      "learning_rate": 0.001536,
      "loss": 5.3158,
      "step": 512
    },
    {
      "epoch": 0.00513,
      "grad_norm": 0.5441672327389838,
      "learning_rate": 0.001539,
      "loss": 5.327,
      "step": 513
    },
    {
      "epoch": 0.00514,
      "grad_norm": 0.5371789671410057,
      "learning_rate": 0.001542,
      "loss": 5.3038,
      "step": 514
    },
    {
      "epoch": 0.00515,
      "grad_norm": 0.5194765862771661,
      "learning_rate": 0.0015450000000000001,
      "loss": 5.3109,
      "step": 515
    },
    {
      "epoch": 0.00516,
      "grad_norm": 0.5575198815834714,
      "learning_rate": 0.0015480000000000001,
      "loss": 5.3083,
      "step": 516
    },
    {
      "epoch": 0.00517,
      "grad_norm": 0.5237583962443445,
      "learning_rate": 0.001551,
      "loss": 5.2962,
      "step": 517
    },
    {
      "epoch": 0.00518,
      "grad_norm": 0.5237181167506534,
      "learning_rate": 0.001554,
      "loss": 5.2786,
      "step": 518
    },
    {
      "epoch": 0.00519,
      "grad_norm": 0.6652146969315359,
      "learning_rate": 0.001557,
      "loss": 5.2847,
      "step": 519
    },
    {
      "epoch": 0.0052,
      "grad_norm": 0.8817258231902962,
      "learning_rate": 0.0015600000000000002,
      "loss": 5.2966,
      "step": 520
    },
    {
      "epoch": 0.00521,
      "grad_norm": 0.9047592596397305,
      "learning_rate": 0.0015630000000000002,
      "loss": 5.2732,
      "step": 521
    },
    {
      "epoch": 0.00522,
      "grad_norm": 0.6139926424217688,
      "learning_rate": 0.0015660000000000001,
      "loss": 5.2701,
      "step": 522
    },
    {
      "epoch": 0.00523,
      "grad_norm": 0.6292464658556638,
      "learning_rate": 0.001569,
      "loss": 5.2826,
      "step": 523
    },
    {
      "epoch": 0.00524,
      "grad_norm": 0.752030715547053,
      "learning_rate": 0.001572,
      "loss": 5.2856,
      "step": 524
    },
    {
      "epoch": 0.00525,
      "grad_norm": 0.929770536798091,
      "learning_rate": 0.001575,
      "loss": 5.287,
      "step": 525
    },
    {
      "epoch": 0.00526,
      "grad_norm": 0.887370520628206,
      "learning_rate": 0.0015780000000000002,
      "loss": 5.2635,
      "step": 526
    },
    {
      "epoch": 0.00527,
      "grad_norm": 0.7819104471156305,
      "learning_rate": 0.0015810000000000002,
      "loss": 5.2728,
      "step": 527
    },
    {
      "epoch": 0.00528,
      "grad_norm": 0.9038037239389326,
      "learning_rate": 0.0015840000000000001,
      "loss": 5.2534,
      "step": 528
    },
    {
      "epoch": 0.00529,
      "grad_norm": 0.8898068580069259,
      "learning_rate": 0.001587,
      "loss": 5.265,
      "step": 529
    },
    {
      "epoch": 0.0053,
      "grad_norm": 1.0010848916209774,
      "learning_rate": 0.00159,
      "loss": 5.2764,
      "step": 530
    },
    {
      "epoch": 0.00531,
      "grad_norm": 1.0010984765594055,
      "learning_rate": 0.001593,
      "loss": 5.2677,
      "step": 531
    },
    {
      "epoch": 0.00532,
      "grad_norm": 1.026224264336229,
      "learning_rate": 0.0015960000000000002,
      "loss": 5.2779,
      "step": 532
    },
    {
      "epoch": 0.00533,
      "grad_norm": 0.9759289805508353,
      "learning_rate": 0.0015990000000000002,
      "loss": 5.2678,
      "step": 533
    },
    {
      "epoch": 0.00534,
      "grad_norm": 1.1376558518204782,
      "learning_rate": 0.0016020000000000001,
      "loss": 5.2612,
      "step": 534
    },
    {
      "epoch": 0.00535,
      "grad_norm": 1.0517455067238486,
      "learning_rate": 0.001605,
      "loss": 5.2672,
      "step": 535
    },
    {
      "epoch": 0.00536,
      "grad_norm": 0.9398243649272562,
      "learning_rate": 0.001608,
      "loss": 5.2627,
      "step": 536
    },
    {
      "epoch": 0.00537,
      "grad_norm": 0.9512995727424398,
      "learning_rate": 0.0016110000000000002,
      "loss": 5.2558,
      "step": 537
    },
    {
      "epoch": 0.00538,
      "grad_norm": 1.2746657535312511,
      "learning_rate": 0.0016140000000000002,
      "loss": 5.2451,
      "step": 538
    },
    {
      "epoch": 0.00539,
      "grad_norm": 1.067518036012326,
      "learning_rate": 0.0016170000000000002,
      "loss": 5.2415,
      "step": 539
    },
    {
      "epoch": 0.0054,
      "grad_norm": 1.1859628541053806,
      "learning_rate": 0.0016200000000000001,
      "loss": 5.2618,
      "step": 540
    },
    {
      "epoch": 0.00541,
      "grad_norm": 0.8882936824028492,
      "learning_rate": 0.001623,
      "loss": 5.2308,
      "step": 541
    },
    {
      "epoch": 0.00542,
      "grad_norm": 0.8517075205384302,
      "learning_rate": 0.001626,
      "loss": 5.2545,
      "step": 542
    },
    {
      "epoch": 0.00543,
      "grad_norm": 0.8283552605034004,
      "learning_rate": 0.0016290000000000002,
      "loss": 5.2098,
      "step": 543
    },
    {
      "epoch": 0.00544,
      "grad_norm": 0.9087829134911761,
      "learning_rate": 0.0016320000000000002,
      "loss": 5.2265,
      "step": 544
    },
    {
      "epoch": 0.00545,
      "grad_norm": 0.8034144620978907,
      "learning_rate": 0.0016350000000000002,
      "loss": 5.2348,
      "step": 545
    },
    {
      "epoch": 0.00546,
      "grad_norm": 0.7091235133563132,
      "learning_rate": 0.0016380000000000001,
      "loss": 5.2004,
      "step": 546
    },
    {
      "epoch": 0.00547,
      "grad_norm": 0.6683331586466694,
      "learning_rate": 0.001641,
      "loss": 5.1915,
      "step": 547
    },
    {
      "epoch": 0.00548,
      "grad_norm": 0.5441552662279447,
      "learning_rate": 0.001644,
      "loss": 5.1843,
      "step": 548
    },
    {
      "epoch": 0.00549,
      "grad_norm": 0.5462993858197037,
      "learning_rate": 0.0016470000000000002,
      "loss": 5.1806,
      "step": 549
    },
    {
      "epoch": 0.0055,
      "grad_norm": 0.5691406737163984,
      "learning_rate": 0.0016500000000000002,
      "loss": 5.1914,
      "step": 550
    },
    {
      "epoch": 0.00551,
      "grad_norm": 0.57998143312047,
      "learning_rate": 0.0016530000000000002,
      "loss": 5.1789,
      "step": 551
    },
    {
      "epoch": 0.00552,
      "grad_norm": 0.5286482487653069,
      "learning_rate": 0.0016560000000000001,
      "loss": 5.158,
      "step": 552
    },
    {
      "epoch": 0.00553,
      "grad_norm": 0.48759014943874474,
      "learning_rate": 0.001659,
      "loss": 5.1465,
      "step": 553
    },
    {
      "epoch": 0.00554,
      "grad_norm": 0.4792524526847805,
      "learning_rate": 0.0016620000000000003,
      "loss": 5.1537,
      "step": 554
    },
    {
      "epoch": 0.00555,
      "grad_norm": 0.5054837577144806,
      "learning_rate": 0.0016650000000000002,
      "loss": 5.1496,
      "step": 555
    },
    {
      "epoch": 0.00556,
      "grad_norm": 0.5148667639200912,
      "learning_rate": 0.0016680000000000002,
      "loss": 5.1288,
      "step": 556
    },
    {
      "epoch": 0.00557,
      "grad_norm": 0.5122706189056161,
      "learning_rate": 0.0016710000000000002,
      "loss": 5.1153,
      "step": 557
    },
    {
      "epoch": 0.00558,
      "grad_norm": 0.6046224327964763,
      "learning_rate": 0.0016740000000000001,
      "loss": 5.1358,
      "step": 558
    },
    {
      "epoch": 0.00559,
      "grad_norm": 0.6851636176151574,
      "learning_rate": 0.001677,
      "loss": 5.1268,
      "step": 559
    },
    {
      "epoch": 0.0056,
      "grad_norm": 0.6922547745331437,
      "learning_rate": 0.0016800000000000003,
      "loss": 5.096,
      "step": 560
    },
    {
      "epoch": 0.00561,
      "grad_norm": 0.7411075733344746,
      "learning_rate": 0.0016830000000000003,
      "loss": 5.1176,
      "step": 561
    },
    {
      "epoch": 0.00562,
      "grad_norm": 0.8981627852593407,
      "learning_rate": 0.0016860000000000002,
      "loss": 5.1206,
      "step": 562
    },
    {
      "epoch": 0.00563,
      "grad_norm": 1.1260200882948381,
      "learning_rate": 0.001689,
      "loss": 5.1239,
      "step": 563
    },
    {
      "epoch": 0.00564,
      "grad_norm": 1.1027210513289374,
      "learning_rate": 0.001692,
      "loss": 5.113,
      "step": 564
    },
    {
      "epoch": 0.00565,
      "grad_norm": 0.890213024695838,
      "learning_rate": 0.001695,
      "loss": 5.1071,
      "step": 565
    },
    {
      "epoch": 0.00566,
      "grad_norm": 0.8627979775394023,
      "learning_rate": 0.0016979999999999999,
      "loss": 5.0976,
      "step": 566
    },
    {
      "epoch": 0.00567,
      "grad_norm": 0.9823945407746334,
      "learning_rate": 0.0017009999999999998,
      "loss": 5.101,
      "step": 567
    },
    {
      "epoch": 0.00568,
      "grad_norm": 0.9715849000742567,
      "learning_rate": 0.0017039999999999998,
      "loss": 5.1139,
      "step": 568
    },
    {
      "epoch": 0.00569,
      "grad_norm": 0.9989913016974431,
      "learning_rate": 0.001707,
      "loss": 5.0861,
      "step": 569
    },
    {
      "epoch": 0.0057,
      "grad_norm": 1.0927877338999235,
      "learning_rate": 0.00171,
      "loss": 5.1262,
      "step": 570
    },
    {
      "epoch": 0.00571,
      "grad_norm": 1.0328737042176641,
      "learning_rate": 0.001713,
      "loss": 5.1192,
      "step": 571
    },
    {
      "epoch": 0.00572,
      "grad_norm": 1.3722388350456287,
      "learning_rate": 0.0017159999999999999,
      "loss": 5.1049,
      "step": 572
    },
    {
      "epoch": 0.00573,
      "grad_norm": 0.9525979203379623,
      "learning_rate": 0.0017189999999999998,
      "loss": 5.1081,
      "step": 573
    },
    {
      "epoch": 0.00574,
      "grad_norm": 1.0626128534442882,
      "learning_rate": 0.001722,
      "loss": 5.1048,
      "step": 574
    },
    {
      "epoch": 0.00575,
      "grad_norm": 0.9331527599734185,
      "learning_rate": 0.001725,
      "loss": 5.074,
      "step": 575
    },
    {
      "epoch": 0.00576,
      "grad_norm": 0.9277735357052385,
      "learning_rate": 0.001728,
      "loss": 5.0843,
      "step": 576
    },
    {
      "epoch": 0.00577,
      "grad_norm": 0.9070967517565243,
      "learning_rate": 0.001731,
      "loss": 5.0908,
      "step": 577
    },
    {
      "epoch": 0.00578,
      "grad_norm": 0.8451551366430134,
      "learning_rate": 0.0017339999999999999,
      "loss": 5.0704,
      "step": 578
    },
    {
      "epoch": 0.00579,
      "grad_norm": 0.7590562073625285,
      "learning_rate": 0.0017369999999999998,
      "loss": 5.058,
      "step": 579
    },
    {
      "epoch": 0.0058,
      "grad_norm": 0.6385326977156373,
      "learning_rate": 0.00174,
      "loss": 5.0662,
      "step": 580
    },
    {
      "epoch": 0.00581,
      "grad_norm": 0.5982129206576257,
      "learning_rate": 0.001743,
      "loss": 5.0624,
      "step": 581
    },
    {
      "epoch": 0.00582,
      "grad_norm": 0.7358321954275717,
      "learning_rate": 0.001746,
      "loss": 5.0371,
      "step": 582
    },
    {
      "epoch": 0.00583,
      "grad_norm": 0.790704997209863,
      "learning_rate": 0.001749,
      "loss": 5.0641,
      "step": 583
    },
    {
      "epoch": 0.00584,
      "grad_norm": 0.7351931257233056,
      "learning_rate": 0.0017519999999999999,
      "loss": 5.0549,
      "step": 584
    },
    {
      "epoch": 0.00585,
      "grad_norm": 0.6167540927458872,
      "learning_rate": 0.0017549999999999998,
      "loss": 5.0362,
      "step": 585
    },
    {
      "epoch": 0.00586,
      "grad_norm": 0.6143004439139178,
      "learning_rate": 0.001758,
      "loss": 5.0316,
      "step": 586
    },
    {
      "epoch": 0.00587,
      "grad_norm": 0.619830997492515,
      "learning_rate": 0.001761,
      "loss": 5.018,
      "step": 587
    },
    {
      "epoch": 0.00588,
      "grad_norm": 0.676754014852622,
      "learning_rate": 0.001764,
      "loss": 5.0113,
      "step": 588
    },
    {
      "epoch": 0.00589,
      "grad_norm": 0.8152367183756798,
      "learning_rate": 0.001767,
      "loss": 5.0063,
      "step": 589
    },
    {
      "epoch": 0.0059,
      "grad_norm": 0.8067730505459064,
      "learning_rate": 0.0017699999999999999,
      "loss": 5.0009,
      "step": 590
    },
    {
      "epoch": 0.00591,
      "grad_norm": 0.7857115694134444,
      "learning_rate": 0.001773,
      "loss": 5.0014,
      "step": 591
    },
    {
      "epoch": 0.00592,
      "grad_norm": 1.0305248086573016,
      "learning_rate": 0.001776,
      "loss": 5.0384,
      "step": 592
    },
    {
      "epoch": 0.00593,
      "grad_norm": 1.1034322728795254,
      "learning_rate": 0.001779,
      "loss": 5.0387,
      "step": 593
    },
    {
      "epoch": 0.00594,
      "grad_norm": 0.796782800390064,
      "learning_rate": 0.001782,
      "loss": 5.0138,
      "step": 594
    },
    {
      "epoch": 0.00595,
      "grad_norm": 0.947951394101208,
      "learning_rate": 0.001785,
      "loss": 5.0288,
      "step": 595
    },
    {
      "epoch": 0.00596,
      "grad_norm": 0.948001541672118,
      "learning_rate": 0.0017879999999999999,
      "loss": 5.0236,
      "step": 596
    },
    {
      "epoch": 0.00597,
      "grad_norm": 0.8614311642183788,
      "learning_rate": 0.001791,
      "loss": 4.983,
      "step": 597
    },
    {
      "epoch": 0.00598,
      "grad_norm": 0.7822502963641336,
      "learning_rate": 0.001794,
      "loss": 5.0032,
      "step": 598
    },
    {
      "epoch": 0.00599,
      "grad_norm": 0.662502314672726,
      "learning_rate": 0.001797,
      "loss": 4.9824,
      "step": 599
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.612691797581944,
      "learning_rate": 0.0018,
      "loss": 4.9865,
      "step": 600
    },
    {
      "epoch": 0.00601,
      "grad_norm": 0.6872791409552744,
      "learning_rate": 0.001803,
      "loss": 4.9768,
      "step": 601
    },
    {
      "epoch": 0.00602,
      "grad_norm": 0.6805787880992017,
      "learning_rate": 0.0018059999999999999,
      "loss": 4.9611,
      "step": 602
    },
    {
      "epoch": 0.00603,
      "grad_norm": 0.6726518476757681,
      "learning_rate": 0.001809,
      "loss": 4.9863,
      "step": 603
    },
    {
      "epoch": 0.00604,
      "grad_norm": 0.6569683723636671,
      "learning_rate": 0.001812,
      "loss": 4.9594,
      "step": 604
    },
    {
      "epoch": 0.00605,
      "grad_norm": 0.7635898698708745,
      "learning_rate": 0.001815,
      "loss": 4.9733,
      "step": 605
    },
    {
      "epoch": 0.00606,
      "grad_norm": 0.7455554600622951,
      "learning_rate": 0.001818,
      "loss": 4.9743,
      "step": 606
    },
    {
      "epoch": 0.00607,
      "grad_norm": 0.7493825131119356,
      "learning_rate": 0.001821,
      "loss": 4.968,
      "step": 607
    },
    {
      "epoch": 0.00608,
      "grad_norm": 0.7957828614510537,
      "learning_rate": 0.001824,
      "loss": 4.9536,
      "step": 608
    },
    {
      "epoch": 0.00609,
      "grad_norm": 0.8304899179729943,
      "learning_rate": 0.001827,
      "loss": 4.9564,
      "step": 609
    },
    {
      "epoch": 0.0061,
      "grad_norm": 0.8109281284885322,
      "learning_rate": 0.00183,
      "loss": 4.9141,
      "step": 610
    },
    {
      "epoch": 0.00611,
      "grad_norm": 0.7476158062529945,
      "learning_rate": 0.001833,
      "loss": 4.936,
      "step": 611
    },
    {
      "epoch": 0.00612,
      "grad_norm": 0.7704555466663776,
      "learning_rate": 0.001836,
      "loss": 4.9262,
      "step": 612
    },
    {
      "epoch": 0.00613,
      "grad_norm": 0.7179188189545482,
      "learning_rate": 0.001839,
      "loss": 4.9285,
      "step": 613
    },
    {
      "epoch": 0.00614,
      "grad_norm": 0.6487628040476678,
      "learning_rate": 0.001842,
      "loss": 4.9483,
      "step": 614
    },
    {
      "epoch": 0.00615,
      "grad_norm": 0.6346341426047015,
      "learning_rate": 0.001845,
      "loss": 4.9336,
      "step": 615
    },
    {
      "epoch": 0.00616,
      "grad_norm": 0.6919482296512011,
      "learning_rate": 0.001848,
      "loss": 4.92,
      "step": 616
    },
    {
      "epoch": 0.00617,
      "grad_norm": 0.8470324326487232,
      "learning_rate": 0.001851,
      "loss": 4.937,
      "step": 617
    },
    {
      "epoch": 0.00618,
      "grad_norm": 0.9780451420741652,
      "learning_rate": 0.001854,
      "loss": 4.918,
      "step": 618
    },
    {
      "epoch": 0.00619,
      "grad_norm": 1.0393940937639112,
      "learning_rate": 0.001857,
      "loss": 4.9393,
      "step": 619
    },
    {
      "epoch": 0.0062,
      "grad_norm": 1.0864888594503201,
      "learning_rate": 0.00186,
      "loss": 4.9315,
      "step": 620
    },
    {
      "epoch": 0.00621,
      "grad_norm": 1.1243573748169302,
      "learning_rate": 0.001863,
      "loss": 4.931,
      "step": 621
    },
    {
      "epoch": 0.00622,
      "grad_norm": 1.2701399731469665,
      "learning_rate": 0.001866,
      "loss": 4.9502,
      "step": 622
    },
    {
      "epoch": 0.00623,
      "grad_norm": 0.8273605340089976,
      "learning_rate": 0.001869,
      "loss": 4.9139,
      "step": 623
    },
    {
      "epoch": 0.00624,
      "grad_norm": 0.9643311427939416,
      "learning_rate": 0.001872,
      "loss": 4.9253,
      "step": 624
    },
    {
      "epoch": 0.00625,
      "grad_norm": 1.2310098040523034,
      "learning_rate": 0.001875,
      "loss": 4.9539,
      "step": 625
    },
    {
      "epoch": 0.00626,
      "grad_norm": 0.8268848993203812,
      "learning_rate": 0.0018780000000000001,
      "loss": 4.9199,
      "step": 626
    },
    {
      "epoch": 0.00627,
      "grad_norm": 0.9575639973355906,
      "learning_rate": 0.001881,
      "loss": 4.909,
      "step": 627
    },
    {
      "epoch": 0.00628,
      "grad_norm": 0.9391450767019313,
      "learning_rate": 0.001884,
      "loss": 4.9487,
      "step": 628
    },
    {
      "epoch": 0.00629,
      "grad_norm": 0.8387975784827085,
      "learning_rate": 0.001887,
      "loss": 4.8976,
      "step": 629
    },
    {
      "epoch": 0.0063,
      "grad_norm": 0.7745062985579545,
      "learning_rate": 0.00189,
      "loss": 4.9062,
      "step": 630
    },
    {
      "epoch": 0.00631,
      "grad_norm": 0.8213366097020988,
      "learning_rate": 0.0018930000000000002,
      "loss": 4.9006,
      "step": 631
    },
    {
      "epoch": 0.00632,
      "grad_norm": 0.8670608316656828,
      "learning_rate": 0.0018960000000000001,
      "loss": 4.9226,
      "step": 632
    },
    {
      "epoch": 0.00633,
      "grad_norm": 0.9183940724690223,
      "learning_rate": 0.001899,
      "loss": 4.9055,
      "step": 633
    },
    {
      "epoch": 0.00634,
      "grad_norm": 0.8953336214105924,
      "learning_rate": 0.001902,
      "loss": 4.8931,
      "step": 634
    },
    {
      "epoch": 0.00635,
      "grad_norm": 0.8869843611928877,
      "learning_rate": 0.001905,
      "loss": 4.8851,
      "step": 635
    },
    {
      "epoch": 0.00636,
      "grad_norm": 1.014617544478078,
      "learning_rate": 0.001908,
      "loss": 4.911,
      "step": 636
    },
    {
      "epoch": 0.00637,
      "grad_norm": 0.8681767489690893,
      "learning_rate": 0.0019110000000000002,
      "loss": 4.909,
      "step": 637
    },
    {
      "epoch": 0.00638,
      "grad_norm": 0.9792692499883113,
      "learning_rate": 0.0019140000000000001,
      "loss": 4.9073,
      "step": 638
    },
    {
      "epoch": 0.00639,
      "grad_norm": 0.8702384519183142,
      "learning_rate": 0.001917,
      "loss": 4.8724,
      "step": 639
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.7656979903996737,
      "learning_rate": 0.00192,
      "loss": 4.8809,
      "step": 640
    },
    {
      "epoch": 0.00641,
      "grad_norm": 0.7895438458212233,
      "learning_rate": 0.001923,
      "loss": 4.8762,
      "step": 641
    },
    {
      "epoch": 0.00642,
      "grad_norm": 0.8281699495124627,
      "learning_rate": 0.001926,
      "loss": 4.8749,
      "step": 642
    },
    {
      "epoch": 0.00643,
      "grad_norm": 0.7845360432833325,
      "learning_rate": 0.0019290000000000002,
      "loss": 4.8856,
      "step": 643
    },
    {
      "epoch": 0.00644,
      "grad_norm": 0.6857666802048429,
      "learning_rate": 0.0019320000000000001,
      "loss": 4.8583,
      "step": 644
    },
    {
      "epoch": 0.00645,
      "grad_norm": 0.5056941566313082,
      "learning_rate": 0.001935,
      "loss": 4.8522,
      "step": 645
    },
    {
      "epoch": 0.00646,
      "grad_norm": 0.47964630717910517,
      "learning_rate": 0.001938,
      "loss": 4.8628,
      "step": 646
    },
    {
      "epoch": 0.00647,
      "grad_norm": 0.519046406898008,
      "learning_rate": 0.001941,
      "loss": 4.8552,
      "step": 647
    },
    {
      "epoch": 0.00648,
      "grad_norm": 0.4935185080219269,
      "learning_rate": 0.0019440000000000002,
      "loss": 4.8488,
      "step": 648
    },
    {
      "epoch": 0.00649,
      "grad_norm": 0.5205599594018984,
      "learning_rate": 0.0019470000000000002,
      "loss": 4.8333,
      "step": 649
    },
    {
      "epoch": 0.0065,
      "grad_norm": 0.45295352715192905,
      "learning_rate": 0.0019500000000000001,
      "loss": 4.8617,
      "step": 650
    },
    {
      "epoch": 0.00651,
      "grad_norm": 0.4378817965282719,
      "learning_rate": 0.001953,
      "loss": 4.84,
      "step": 651
    },
    {
      "epoch": 0.00652,
      "grad_norm": 0.48395025642257355,
      "learning_rate": 0.0019560000000000003,
      "loss": 4.8266,
      "step": 652
    },
    {
      "epoch": 0.00653,
      "grad_norm": 0.5194845024092328,
      "learning_rate": 0.0019590000000000002,
      "loss": 4.8294,
      "step": 653
    },
    {
      "epoch": 0.00654,
      "grad_norm": 0.5707068593609262,
      "learning_rate": 0.001962,
      "loss": 4.8026,
      "step": 654
    },
    {
      "epoch": 0.00655,
      "grad_norm": 0.6725172209521785,
      "learning_rate": 0.001965,
      "loss": 4.834,
      "step": 655
    },
    {
      "epoch": 0.00656,
      "grad_norm": 0.7729568432311581,
      "learning_rate": 0.001968,
      "loss": 4.8398,
      "step": 656
    },
    {
      "epoch": 0.00657,
      "grad_norm": 0.9182317536069756,
      "learning_rate": 0.001971,
      "loss": 4.8175,
      "step": 657
    },
    {
      "epoch": 0.00658,
      "grad_norm": 0.9241834469814327,
      "learning_rate": 0.001974,
      "loss": 4.8482,
      "step": 658
    },
    {
      "epoch": 0.00659,
      "grad_norm": 0.7554289117841486,
      "learning_rate": 0.001977,
      "loss": 4.8199,
      "step": 659
    },
    {
      "epoch": 0.0066,
      "grad_norm": 0.786222644002383,
      "learning_rate": 0.00198,
      "loss": 4.8336,
      "step": 660
    },
    {
      "epoch": 0.00661,
      "grad_norm": 0.7486956414167568,
      "learning_rate": 0.001983,
      "loss": 4.8035,
      "step": 661
    },
    {
      "epoch": 0.00662,
      "grad_norm": 0.8224936329580866,
      "learning_rate": 0.0019860000000000004,
      "loss": 4.8038,
      "step": 662
    },
    {
      "epoch": 0.00663,
      "grad_norm": 1.0333891900276588,
      "learning_rate": 0.0019890000000000003,
      "loss": 4.8216,
      "step": 663
    },
    {
      "epoch": 0.00664,
      "grad_norm": 1.0454184895583787,
      "learning_rate": 0.0019920000000000003,
      "loss": 4.826,
      "step": 664
    },
    {
      "epoch": 0.00665,
      "grad_norm": 0.9684168839706867,
      "learning_rate": 0.0019950000000000002,
      "loss": 4.7952,
      "step": 665
    },
    {
      "epoch": 0.00666,
      "grad_norm": 1.0353944208678731,
      "learning_rate": 0.001998,
      "loss": 4.8213,
      "step": 666
    },
    {
      "epoch": 0.00667,
      "grad_norm": 0.9360230286599108,
      "learning_rate": 0.002001,
      "loss": 4.8166,
      "step": 667
    },
    {
      "epoch": 0.00668,
      "grad_norm": 0.7256915823313501,
      "learning_rate": 0.002004,
      "loss": 4.83,
      "step": 668
    },
    {
      "epoch": 0.00669,
      "grad_norm": 0.8394069732991961,
      "learning_rate": 0.002007,
      "loss": 4.8494,
      "step": 669
    },
    {
      "epoch": 0.0067,
      "grad_norm": 1.077577977883947,
      "learning_rate": 0.00201,
      "loss": 4.8532,
      "step": 670
    },
    {
      "epoch": 0.00671,
      "grad_norm": 0.944781077919628,
      "learning_rate": 0.002013,
      "loss": 4.8236,
      "step": 671
    },
    {
      "epoch": 0.00672,
      "grad_norm": 0.7024086289087288,
      "learning_rate": 0.002016,
      "loss": 4.7966,
      "step": 672
    },
    {
      "epoch": 0.00673,
      "grad_norm": 0.6931174948166909,
      "learning_rate": 0.002019,
      "loss": 4.809,
      "step": 673
    },
    {
      "epoch": 0.00674,
      "grad_norm": 0.6108100930625565,
      "learning_rate": 0.0020220000000000004,
      "loss": 4.8005,
      "step": 674
    },
    {
      "epoch": 0.00675,
      "grad_norm": 0.5498762136107717,
      "learning_rate": 0.0020250000000000003,
      "loss": 4.8171,
      "step": 675
    },
    {
      "epoch": 0.00676,
      "grad_norm": 0.6244881777878202,
      "learning_rate": 0.0020280000000000003,
      "loss": 4.7951,
      "step": 676
    },
    {
      "epoch": 0.00677,
      "grad_norm": 0.5815667573949113,
      "learning_rate": 0.0020310000000000003,
      "loss": 4.7913,
      "step": 677
    },
    {
      "epoch": 0.00678,
      "grad_norm": 0.521491074875552,
      "learning_rate": 0.0020340000000000002,
      "loss": 4.7959,
      "step": 678
    },
    {
      "epoch": 0.00679,
      "grad_norm": 0.5801192243287052,
      "learning_rate": 0.002037,
      "loss": 4.7864,
      "step": 679
    },
    {
      "epoch": 0.0068,
      "grad_norm": 0.6398788453868361,
      "learning_rate": 0.00204,
      "loss": 4.7875,
      "step": 680
    },
    {
      "epoch": 0.00681,
      "grad_norm": 0.6456567912241706,
      "learning_rate": 0.002043,
      "loss": 4.7807,
      "step": 681
    },
    {
      "epoch": 0.00682,
      "grad_norm": 0.6889535888719373,
      "learning_rate": 0.002046,
      "loss": 4.7776,
      "step": 682
    },
    {
      "epoch": 0.00683,
      "grad_norm": 0.68132819276425,
      "learning_rate": 0.002049,
      "loss": 4.7612,
      "step": 683
    },
    {
      "epoch": 0.00684,
      "grad_norm": 0.6724100911918068,
      "learning_rate": 0.002052,
      "loss": 4.7814,
      "step": 684
    },
    {
      "epoch": 0.00685,
      "grad_norm": 0.815928121483196,
      "learning_rate": 0.0020550000000000004,
      "loss": 4.7913,
      "step": 685
    },
    {
      "epoch": 0.00686,
      "grad_norm": 0.7537150884203632,
      "learning_rate": 0.0020580000000000004,
      "loss": 4.7584,
      "step": 686
    },
    {
      "epoch": 0.00687,
      "grad_norm": 0.7326770278926477,
      "learning_rate": 0.0020610000000000003,
      "loss": 4.7655,
      "step": 687
    },
    {
      "epoch": 0.00688,
      "grad_norm": 0.7145131916015852,
      "learning_rate": 0.002064,
      "loss": 4.7631,
      "step": 688
    },
    {
      "epoch": 0.00689,
      "grad_norm": 0.694828893634318,
      "learning_rate": 0.002067,
      "loss": 4.7685,
      "step": 689
    },
    {
      "epoch": 0.0069,
      "grad_norm": 0.7402628986076095,
      "learning_rate": 0.00207,
      "loss": 4.7629,
      "step": 690
    },
    {
      "epoch": 0.00691,
      "grad_norm": 0.7809038473743037,
      "learning_rate": 0.0020729999999999998,
      "loss": 4.734,
      "step": 691
    },
    {
      "epoch": 0.00692,
      "grad_norm": 0.9321737974219415,
      "learning_rate": 0.0020759999999999997,
      "loss": 4.7623,
      "step": 692
    },
    {
      "epoch": 0.00693,
      "grad_norm": 1.0940875671633443,
      "learning_rate": 0.0020789999999999997,
      "loss": 4.788,
      "step": 693
    },
    {
      "epoch": 0.00694,
      "grad_norm": 0.9030072391970948,
      "learning_rate": 0.002082,
      "loss": 4.7582,
      "step": 694
    },
    {
      "epoch": 0.00695,
      "grad_norm": 0.8857004379777869,
      "learning_rate": 0.002085,
      "loss": 4.7649,
      "step": 695
    },
    {
      "epoch": 0.00696,
      "grad_norm": 0.8991843577664433,
      "learning_rate": 0.002088,
      "loss": 4.7712,
      "step": 696
    },
    {
      "epoch": 0.00697,
      "grad_norm": 1.0094726318481275,
      "learning_rate": 0.002091,
      "loss": 4.7869,
      "step": 697
    },
    {
      "epoch": 0.00698,
      "grad_norm": 1.09376996979269,
      "learning_rate": 0.002094,
      "loss": 4.7773,
      "step": 698
    },
    {
      "epoch": 0.00699,
      "grad_norm": 0.8766228601427865,
      "learning_rate": 0.002097,
      "loss": 4.7548,
      "step": 699
    },
    {
      "epoch": 0.007,
      "grad_norm": 0.8580408760742435,
      "learning_rate": 0.0021,
      "loss": 4.7811,
      "step": 700
    },
    {
      "epoch": 0.00701,
      "grad_norm": 0.7398805390269545,
      "learning_rate": 0.002103,
      "loss": 4.7496,
      "step": 701
    },
    {
      "epoch": 0.00702,
      "grad_norm": 0.8329034135969645,
      "learning_rate": 0.002106,
      "loss": 4.762,
      "step": 702
    },
    {
      "epoch": 0.00703,
      "grad_norm": 0.8505784581176613,
      "learning_rate": 0.0021089999999999998,
      "loss": 4.7427,
      "step": 703
    },
    {
      "epoch": 0.00704,
      "grad_norm": 0.8003313240374758,
      "learning_rate": 0.0021119999999999997,
      "loss": 4.7516,
      "step": 704
    },
    {
      "epoch": 0.00705,
      "grad_norm": 0.7109629500256387,
      "learning_rate": 0.002115,
      "loss": 4.7635,
      "step": 705
    },
    {
      "epoch": 0.00706,
      "grad_norm": 0.6289559048526784,
      "learning_rate": 0.002118,
      "loss": 4.7473,
      "step": 706
    },
    {
      "epoch": 0.00707,
      "grad_norm": 0.5482382237711074,
      "learning_rate": 0.002121,
      "loss": 4.7193,
      "step": 707
    },
    {
      "epoch": 0.00708,
      "grad_norm": 0.5598390325768088,
      "learning_rate": 0.002124,
      "loss": 4.7015,
      "step": 708
    },
    {
      "epoch": 0.00709,
      "grad_norm": 0.5529551760214286,
      "learning_rate": 0.002127,
      "loss": 4.7331,
      "step": 709
    },
    {
      "epoch": 0.0071,
      "grad_norm": 0.6671084870548735,
      "learning_rate": 0.00213,
      "loss": 4.7271,
      "step": 710
    },
    {
      "epoch": 0.00711,
      "grad_norm": 1.1336515667563218,
      "learning_rate": 0.002133,
      "loss": 4.7225,
      "step": 711
    },
    {
      "epoch": 0.00712,
      "grad_norm": 0.8878745395415352,
      "learning_rate": 0.002136,
      "loss": 4.761,
      "step": 712
    },
    {
      "epoch": 0.00713,
      "grad_norm": 0.6767692274074788,
      "learning_rate": 0.002139,
      "loss": 4.7547,
      "step": 713
    },
    {
      "epoch": 0.00714,
      "grad_norm": 0.7698487468076475,
      "learning_rate": 0.002142,
      "loss": 4.7544,
      "step": 714
    },
    {
      "epoch": 0.00715,
      "grad_norm": 0.6189384208707256,
      "learning_rate": 0.0021449999999999998,
      "loss": 4.7077,
      "step": 715
    },
    {
      "epoch": 0.00716,
      "grad_norm": 0.5968344999809513,
      "learning_rate": 0.002148,
      "loss": 4.7064,
      "step": 716
    },
    {
      "epoch": 0.00717,
      "grad_norm": 0.5117629580972962,
      "learning_rate": 0.002151,
      "loss": 4.7111,
      "step": 717
    },
    {
      "epoch": 0.00718,
      "grad_norm": 0.5586203984201494,
      "learning_rate": 0.002154,
      "loss": 4.7079,
      "step": 718
    },
    {
      "epoch": 0.00719,
      "grad_norm": 0.5075479633130432,
      "learning_rate": 0.002157,
      "loss": 4.7121,
      "step": 719
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.4840284295584723,
      "learning_rate": 0.00216,
      "loss": 4.7045,
      "step": 720
    },
    {
      "epoch": 0.00721,
      "grad_norm": 0.4971167150526394,
      "learning_rate": 0.002163,
      "loss": 4.7056,
      "step": 721
    },
    {
      "epoch": 0.00722,
      "grad_norm": 0.5347204422021664,
      "learning_rate": 0.002166,
      "loss": 4.6807,
      "step": 722
    },
    {
      "epoch": 0.00723,
      "grad_norm": 0.5424206330484346,
      "learning_rate": 0.002169,
      "loss": 4.6942,
      "step": 723
    },
    {
      "epoch": 0.00724,
      "grad_norm": 0.5219918511718452,
      "learning_rate": 0.002172,
      "loss": 4.6912,
      "step": 724
    },
    {
      "epoch": 0.00725,
      "grad_norm": 0.4975444213381179,
      "learning_rate": 0.002175,
      "loss": 4.6834,
      "step": 725
    },
    {
      "epoch": 0.00726,
      "grad_norm": 0.5840922524665152,
      "learning_rate": 0.002178,
      "loss": 4.7008,
      "step": 726
    },
    {
      "epoch": 0.00727,
      "grad_norm": 0.7633698368434719,
      "learning_rate": 0.0021809999999999998,
      "loss": 4.6777,
      "step": 727
    },
    {
      "epoch": 0.00728,
      "grad_norm": 0.9052599955462425,
      "learning_rate": 0.002184,
      "loss": 4.7084,
      "step": 728
    },
    {
      "epoch": 0.00729,
      "grad_norm": 0.9052862109699649,
      "learning_rate": 0.002187,
      "loss": 4.7065,
      "step": 729
    },
    {
      "epoch": 0.0073,
      "grad_norm": 0.9788098741089467,
      "learning_rate": 0.00219,
      "loss": 4.7163,
      "step": 730
    },
    {
      "epoch": 0.00731,
      "grad_norm": 0.9218858714566311,
      "learning_rate": 0.002193,
      "loss": 4.7193,
      "step": 731
    },
    {
      "epoch": 0.00732,
      "grad_norm": 1.1412145043722297,
      "learning_rate": 0.002196,
      "loss": 4.7513,
      "step": 732
    },
    {
      "epoch": 0.00733,
      "grad_norm": 1.0851884555194036,
      "learning_rate": 0.002199,
      "loss": 4.7135,
      "step": 733
    },
    {
      "epoch": 0.00734,
      "grad_norm": 0.9295060612046938,
      "learning_rate": 0.002202,
      "loss": 4.745,
      "step": 734
    },
    {
      "epoch": 0.00735,
      "grad_norm": 1.1744848346988581,
      "learning_rate": 0.002205,
      "loss": 4.766,
      "step": 735
    },
    {
      "epoch": 0.00736,
      "grad_norm": 0.8278931321658332,
      "learning_rate": 0.002208,
      "loss": 4.7413,
      "step": 736
    },
    {
      "epoch": 0.00737,
      "grad_norm": 0.8170748474782394,
      "learning_rate": 0.002211,
      "loss": 4.701,
      "step": 737
    },
    {
      "epoch": 0.00738,
      "grad_norm": 0.772640277783885,
      "learning_rate": 0.002214,
      "loss": 4.7083,
      "step": 738
    },
    {
      "epoch": 0.00739,
      "grad_norm": 0.6328057525928463,
      "learning_rate": 0.0022170000000000002,
      "loss": 4.6886,
      "step": 739
    },
    {
      "epoch": 0.0074,
      "grad_norm": 0.6456777453177256,
      "learning_rate": 0.00222,
      "loss": 4.7001,
      "step": 740
    },
    {
      "epoch": 0.00741,
      "grad_norm": 0.5912580097221507,
      "learning_rate": 0.002223,
      "loss": 4.6652,
      "step": 741
    },
    {
      "epoch": 0.00742,
      "grad_norm": 0.4622238730689768,
      "learning_rate": 0.002226,
      "loss": 4.6972,
      "step": 742
    },
    {
      "epoch": 0.00743,
      "grad_norm": 0.4904966133460914,
      "learning_rate": 0.002229,
      "loss": 4.6585,
      "step": 743
    },
    {
      "epoch": 0.00744,
      "grad_norm": 0.5611948954376664,
      "learning_rate": 0.002232,
      "loss": 4.6643,
      "step": 744
    },
    {
      "epoch": 0.00745,
      "grad_norm": 0.6222946629538655,
      "learning_rate": 0.002235,
      "loss": 4.6777,
      "step": 745
    },
    {
      "epoch": 0.00746,
      "grad_norm": 0.6678221674623648,
      "learning_rate": 0.002238,
      "loss": 4.6735,
      "step": 746
    },
    {
      "epoch": 0.00747,
      "grad_norm": 0.7945455347777014,
      "learning_rate": 0.002241,
      "loss": 4.6637,
      "step": 747
    },
    {
      "epoch": 0.00748,
      "grad_norm": 0.9027593299924651,
      "learning_rate": 0.002244,
      "loss": 4.683,
      "step": 748
    },
    {
      "epoch": 0.00749,
      "grad_norm": 0.8224517051186598,
      "learning_rate": 0.002247,
      "loss": 4.6612,
      "step": 749
    },
    {
      "epoch": 0.0075,
      "grad_norm": 0.6737194715747743,
      "learning_rate": 0.0022500000000000003,
      "loss": 4.6955,
      "step": 750
    },
    {
      "epoch": 0.00751,
      "grad_norm": 0.7232822830562855,
      "learning_rate": 0.0022530000000000002,
      "loss": 4.6893,
      "step": 751
    },
    {
      "epoch": 0.00752,
      "grad_norm": 0.6449225627821611,
      "learning_rate": 0.002256,
      "loss": 4.637,
      "step": 752
    },
    {
      "epoch": 0.00753,
      "grad_norm": 0.6558342053271724,
      "learning_rate": 0.002259,
      "loss": 4.6782,
      "step": 753
    },
    {
      "epoch": 0.00754,
      "grad_norm": 0.7056102529212358,
      "learning_rate": 0.002262,
      "loss": 4.6906,
      "step": 754
    },
    {
      "epoch": 0.00755,
      "grad_norm": 0.9043185481447295,
      "learning_rate": 0.002265,
      "loss": 4.6606,
      "step": 755
    },
    {
      "epoch": 0.00756,
      "grad_norm": 0.8770375990036015,
      "learning_rate": 0.002268,
      "loss": 4.6551,
      "step": 756
    },
    {
      "epoch": 0.00757,
      "grad_norm": 0.784655485815756,
      "learning_rate": 0.002271,
      "loss": 4.6655,
      "step": 757
    },
    {
      "epoch": 0.00758,
      "grad_norm": 0.6912079950422929,
      "learning_rate": 0.002274,
      "loss": 4.655,
      "step": 758
    },
    {
      "epoch": 0.00759,
      "grad_norm": 0.6281371231708264,
      "learning_rate": 0.002277,
      "loss": 4.6629,
      "step": 759
    },
    {
      "epoch": 0.0076,
      "grad_norm": 0.5992557531443571,
      "learning_rate": 0.00228,
      "loss": 4.6463,
      "step": 760
    },
    {
      "epoch": 0.00761,
      "grad_norm": 0.6034804931159479,
      "learning_rate": 0.002283,
      "loss": 4.6146,
      "step": 761
    },
    {
      "epoch": 0.00762,
      "grad_norm": 0.6060668832918386,
      "learning_rate": 0.0022860000000000003,
      "loss": 4.5953,
      "step": 762
    },
    {
      "epoch": 0.00763,
      "grad_norm": 0.5816892837581542,
      "learning_rate": 0.0022890000000000002,
      "loss": 4.6459,
      "step": 763
    },
    {
      "epoch": 0.00764,
      "grad_norm": 0.599786845424844,
      "learning_rate": 0.002292,
      "loss": 4.6554,
      "step": 764
    },
    {
      "epoch": 0.00765,
      "grad_norm": 0.6640243443171511,
      "learning_rate": 0.002295,
      "loss": 4.6451,
      "step": 765
    },
    {
      "epoch": 0.00766,
      "grad_norm": 0.6663176934335011,
      "learning_rate": 0.002298,
      "loss": 4.6438,
      "step": 766
    },
    {
      "epoch": 0.00767,
      "grad_norm": 0.6384162747773746,
      "learning_rate": 0.002301,
      "loss": 4.6392,
      "step": 767
    },
    {
      "epoch": 0.00768,
      "grad_norm": 0.8537658256313299,
      "learning_rate": 0.002304,
      "loss": 4.6457,
      "step": 768
    },
    {
      "epoch": 0.00769,
      "grad_norm": 1.0098120726516953,
      "learning_rate": 0.002307,
      "loss": 4.6694,
      "step": 769
    },
    {
      "epoch": 0.0077,
      "grad_norm": 0.7423003820485509,
      "learning_rate": 0.00231,
      "loss": 4.6223,
      "step": 770
    },
    {
      "epoch": 0.00771,
      "grad_norm": 0.6691351211965346,
      "learning_rate": 0.002313,
      "loss": 4.6569,
      "step": 771
    },
    {
      "epoch": 0.00772,
      "grad_norm": 0.8185609860643415,
      "learning_rate": 0.002316,
      "loss": 4.6531,
      "step": 772
    },
    {
      "epoch": 0.00773,
      "grad_norm": 1.0039613061235502,
      "learning_rate": 0.0023190000000000003,
      "loss": 4.6664,
      "step": 773
    },
    {
      "epoch": 0.00774,
      "grad_norm": 0.8500607234716588,
      "learning_rate": 0.0023220000000000003,
      "loss": 4.6845,
      "step": 774
    },
    {
      "epoch": 0.00775,
      "grad_norm": 0.7057192127032131,
      "learning_rate": 0.0023250000000000002,
      "loss": 4.6688,
      "step": 775
    },
    {
      "epoch": 0.00776,
      "grad_norm": 0.8527617729361273,
      "learning_rate": 0.002328,
      "loss": 4.6732,
      "step": 776
    },
    {
      "epoch": 0.00777,
      "grad_norm": 0.7987783133918631,
      "learning_rate": 0.002331,
      "loss": 4.6469,
      "step": 777
    },
    {
      "epoch": 0.00778,
      "grad_norm": 0.8221364860163118,
      "learning_rate": 0.002334,
      "loss": 4.6645,
      "step": 778
    },
    {
      "epoch": 0.00779,
      "grad_norm": 0.8901185821524535,
      "learning_rate": 0.002337,
      "loss": 4.6243,
      "step": 779
    },
    {
      "epoch": 0.0078,
      "grad_norm": 0.889956887452623,
      "learning_rate": 0.00234,
      "loss": 4.646,
      "step": 780
    },
    {
      "epoch": 0.00781,
      "grad_norm": 0.763195026878423,
      "learning_rate": 0.002343,
      "loss": 4.6465,
      "step": 781
    },
    {
      "epoch": 0.00782,
      "grad_norm": 0.6941360203492353,
      "learning_rate": 0.002346,
      "loss": 4.6378,
      "step": 782
    },
    {
      "epoch": 0.00783,
      "grad_norm": 0.6504570243922513,
      "learning_rate": 0.002349,
      "loss": 4.6195,
      "step": 783
    },
    {
      "epoch": 0.00784,
      "grad_norm": 0.5612286515783087,
      "learning_rate": 0.002352,
      "loss": 4.6293,
      "step": 784
    },
    {
      "epoch": 0.00785,
      "grad_norm": 0.6491841258536649,
      "learning_rate": 0.0023550000000000003,
      "loss": 4.6258,
      "step": 785
    },
    {
      "epoch": 0.00786,
      "grad_norm": 0.7618604140557245,
      "learning_rate": 0.0023580000000000003,
      "loss": 4.6353,
      "step": 786
    },
    {
      "epoch": 0.00787,
      "grad_norm": 0.9289251225559676,
      "learning_rate": 0.0023610000000000003,
      "loss": 4.6478,
      "step": 787
    },
    {
      "epoch": 0.00788,
      "grad_norm": 1.0160410825626025,
      "learning_rate": 0.002364,
      "loss": 4.6569,
      "step": 788
    },
    {
      "epoch": 0.00789,
      "grad_norm": 0.9904632547103471,
      "learning_rate": 0.002367,
      "loss": 4.6841,
      "step": 789
    },
    {
      "epoch": 0.0079,
      "grad_norm": 0.978549262142993,
      "learning_rate": 0.00237,
      "loss": 4.6525,
      "step": 790
    },
    {
      "epoch": 0.00791,
      "grad_norm": 0.7626000247604355,
      "learning_rate": 0.002373,
      "loss": 4.6008,
      "step": 791
    },
    {
      "epoch": 0.00792,
      "grad_norm": 0.7032433100729657,
      "learning_rate": 0.002376,
      "loss": 4.6361,
      "step": 792
    },
    {
      "epoch": 0.00793,
      "grad_norm": 0.6164393933780811,
      "learning_rate": 0.002379,
      "loss": 4.6263,
      "step": 793
    },
    {
      "epoch": 0.00794,
      "grad_norm": 0.6849838631141278,
      "learning_rate": 0.002382,
      "loss": 4.6402,
      "step": 794
    },
    {
      "epoch": 0.00795,
      "grad_norm": 0.7541453624973187,
      "learning_rate": 0.002385,
      "loss": 4.6181,
      "step": 795
    },
    {
      "epoch": 0.00796,
      "grad_norm": 0.7900314117496404,
      "learning_rate": 0.0023880000000000004,
      "loss": 4.6154,
      "step": 796
    },
    {
      "epoch": 0.00797,
      "grad_norm": 0.8032302751261208,
      "learning_rate": 0.0023910000000000003,
      "loss": 4.6598,
      "step": 797
    },
    {
      "epoch": 0.00798,
      "grad_norm": 0.7037506066431396,
      "learning_rate": 0.0023940000000000003,
      "loss": 4.6206,
      "step": 798
    },
    {
      "epoch": 0.00799,
      "grad_norm": 0.6873143294249761,
      "learning_rate": 0.0023970000000000003,
      "loss": 4.6565,
      "step": 799
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.5611968184575689,
      "learning_rate": 0.0024000000000000002,
      "loss": 4.6002,
      "step": 800
    },
    {
      "epoch": 0.00801,
      "grad_norm": 0.6431524925156966,
      "learning_rate": 0.002403,
      "loss": 4.6326,
      "step": 801
    },
    {
      "epoch": 0.00802,
      "grad_norm": 0.664361851854149,
      "learning_rate": 0.002406,
      "loss": 4.624,
      "step": 802
    },
    {
      "epoch": 0.00803,
      "grad_norm": 0.7385532765716779,
      "learning_rate": 0.002409,
      "loss": 4.6034,
      "step": 803
    },
    {
      "epoch": 0.00804,
      "grad_norm": 0.6763586866376322,
      "learning_rate": 0.002412,
      "loss": 4.609,
      "step": 804
    },
    {
      "epoch": 0.00805,
      "grad_norm": 0.4984516329675046,
      "learning_rate": 0.002415,
      "loss": 4.5958,
      "step": 805
    },
    {
      "epoch": 0.00806,
      "grad_norm": 0.6422151194456196,
      "learning_rate": 0.002418,
      "loss": 4.6202,
      "step": 806
    },
    {
      "epoch": 0.00807,
      "grad_norm": 0.6520347813599764,
      "learning_rate": 0.0024210000000000004,
      "loss": 4.6111,
      "step": 807
    },
    {
      "epoch": 0.00808,
      "grad_norm": 0.6466117231395049,
      "learning_rate": 0.0024240000000000004,
      "loss": 4.5938,
      "step": 808
    },
    {
      "epoch": 0.00809,
      "grad_norm": 0.7173763198404414,
      "learning_rate": 0.0024270000000000003,
      "loss": 4.5875,
      "step": 809
    },
    {
      "epoch": 0.0081,
      "grad_norm": 0.7630708316428362,
      "learning_rate": 0.0024300000000000003,
      "loss": 4.6257,
      "step": 810
    },
    {
      "epoch": 0.00811,
      "grad_norm": 0.6597258634486322,
      "learning_rate": 0.0024330000000000003,
      "loss": 4.5916,
      "step": 811
    },
    {
      "epoch": 0.00812,
      "grad_norm": 0.5596135689444884,
      "learning_rate": 0.0024360000000000002,
      "loss": 4.5639,
      "step": 812
    },
    {
      "epoch": 0.00813,
      "grad_norm": 0.5455291997717118,
      "learning_rate": 0.0024389999999999998,
      "loss": 4.6039,
      "step": 813
    },
    {
      "epoch": 0.00814,
      "grad_norm": 0.643394229982758,
      "learning_rate": 0.0024419999999999997,
      "loss": 4.6125,
      "step": 814
    },
    {
      "epoch": 0.00815,
      "grad_norm": 0.7154153020141445,
      "learning_rate": 0.0024449999999999997,
      "loss": 4.6204,
      "step": 815
    },
    {
      "epoch": 0.00816,
      "grad_norm": 0.7061775094598466,
      "learning_rate": 0.002448,
      "loss": 4.6005,
      "step": 816
    },
    {
      "epoch": 0.00817,
      "grad_norm": 0.7497949470997279,
      "learning_rate": 0.002451,
      "loss": 4.6054,
      "step": 817
    },
    {
      "epoch": 0.00818,
      "grad_norm": 0.8303821582765404,
      "learning_rate": 0.002454,
      "loss": 4.6047,
      "step": 818
    },
    {
      "epoch": 0.00819,
      "grad_norm": 0.7870772660210968,
      "learning_rate": 0.002457,
      "loss": 4.6129,
      "step": 819
    },
    {
      "epoch": 0.0082,
      "grad_norm": 0.8045799279983024,
      "learning_rate": 0.00246,
      "loss": 4.5959,
      "step": 820
    },
    {
      "epoch": 0.00821,
      "grad_norm": 0.6847814476141592,
      "learning_rate": 0.002463,
      "loss": 4.5753,
      "step": 821
    },
    {
      "epoch": 0.00822,
      "grad_norm": 0.6767342952639601,
      "learning_rate": 0.002466,
      "loss": 4.5743,
      "step": 822
    },
    {
      "epoch": 0.00823,
      "grad_norm": 0.7814972414925468,
      "learning_rate": 0.002469,
      "loss": 4.5892,
      "step": 823
    },
    {
      "epoch": 0.00824,
      "grad_norm": 0.7970591329920511,
      "learning_rate": 0.002472,
      "loss": 4.6124,
      "step": 824
    },
    {
      "epoch": 0.00825,
      "grad_norm": 0.7237059608391628,
      "learning_rate": 0.0024749999999999998,
      "loss": 4.6158,
      "step": 825
    },
    {
      "epoch": 0.00826,
      "grad_norm": 0.8825063151832095,
      "learning_rate": 0.0024779999999999997,
      "loss": 4.5838,
      "step": 826
    },
    {
      "epoch": 0.00827,
      "grad_norm": 0.895197450200386,
      "learning_rate": 0.002481,
      "loss": 4.6224,
      "step": 827
    },
    {
      "epoch": 0.00828,
      "grad_norm": 0.7760924275661016,
      "learning_rate": 0.002484,
      "loss": 4.6087,
      "step": 828
    },
    {
      "epoch": 0.00829,
      "grad_norm": 0.8835354563964355,
      "learning_rate": 0.002487,
      "loss": 4.6157,
      "step": 829
    },
    {
      "epoch": 0.0083,
      "grad_norm": 0.8984013262238549,
      "learning_rate": 0.00249,
      "loss": 4.6232,
      "step": 830
    },
    {
      "epoch": 0.00831,
      "grad_norm": 0.8653834946214679,
      "learning_rate": 0.002493,
      "loss": 4.6038,
      "step": 831
    },
    {
      "epoch": 0.00832,
      "grad_norm": 0.8641563564291244,
      "learning_rate": 0.002496,
      "loss": 4.6089,
      "step": 832
    },
    {
      "epoch": 0.00833,
      "grad_norm": 0.8849940351412556,
      "learning_rate": 0.002499,
      "loss": 4.6069,
      "step": 833
    },
    {
      "epoch": 0.00834,
      "grad_norm": 0.9333438912141526,
      "learning_rate": 0.002502,
      "loss": 4.6165,
      "step": 834
    },
    {
      "epoch": 0.00835,
      "grad_norm": 0.8853225829488065,
      "learning_rate": 0.002505,
      "loss": 4.6108,
      "step": 835
    },
    {
      "epoch": 0.00836,
      "grad_norm": 0.9042165265638229,
      "learning_rate": 0.002508,
      "loss": 4.6433,
      "step": 836
    },
    {
      "epoch": 0.00837,
      "grad_norm": 0.8034248166459536,
      "learning_rate": 0.0025109999999999998,
      "loss": 4.5745,
      "step": 837
    },
    {
      "epoch": 0.00838,
      "grad_norm": 0.8086998635281932,
      "learning_rate": 0.0025139999999999997,
      "loss": 4.6114,
      "step": 838
    },
    {
      "epoch": 0.00839,
      "grad_norm": 0.587375456181757,
      "learning_rate": 0.002517,
      "loss": 4.5954,
      "step": 839
    },
    {
      "epoch": 0.0084,
      "grad_norm": 0.6561997865962053,
      "learning_rate": 0.00252,
      "loss": 4.5818,
      "step": 840
    },
    {
      "epoch": 0.00841,
      "grad_norm": 0.6860624256233325,
      "learning_rate": 0.002523,
      "loss": 4.5813,
      "step": 841
    },
    {
      "epoch": 0.00842,
      "grad_norm": 0.6384378914075728,
      "learning_rate": 0.002526,
      "loss": 4.5783,
      "step": 842
    },
    {
      "epoch": 0.00843,
      "grad_norm": 0.5397302637540236,
      "learning_rate": 0.002529,
      "loss": 4.5728,
      "step": 843
    },
    {
      "epoch": 0.00844,
      "grad_norm": 0.5753282450552862,
      "learning_rate": 0.002532,
      "loss": 4.5912,
      "step": 844
    },
    {
      "epoch": 0.00845,
      "grad_norm": 0.605188801053789,
      "learning_rate": 0.002535,
      "loss": 4.5588,
      "step": 845
    },
    {
      "epoch": 0.00846,
      "grad_norm": 0.523231421872222,
      "learning_rate": 0.002538,
      "loss": 4.5771,
      "step": 846
    },
    {
      "epoch": 0.00847,
      "grad_norm": 0.4442194357732046,
      "learning_rate": 0.002541,
      "loss": 4.5539,
      "step": 847
    },
    {
      "epoch": 0.00848,
      "grad_norm": 0.4340595323225716,
      "learning_rate": 0.002544,
      "loss": 4.5562,
      "step": 848
    },
    {
      "epoch": 0.00849,
      "grad_norm": 0.40987686572717924,
      "learning_rate": 0.002547,
      "loss": 4.5471,
      "step": 849
    },
    {
      "epoch": 0.0085,
      "grad_norm": 0.38823479672015115,
      "learning_rate": 0.00255,
      "loss": 4.5547,
      "step": 850
    },
    {
      "epoch": 0.00851,
      "grad_norm": 0.42997075207568086,
      "learning_rate": 0.002553,
      "loss": 4.5733,
      "step": 851
    },
    {
      "epoch": 0.00852,
      "grad_norm": 0.5192269912554525,
      "learning_rate": 0.002556,
      "loss": 4.5624,
      "step": 852
    },
    {
      "epoch": 0.00853,
      "grad_norm": 0.6626364019198889,
      "learning_rate": 0.002559,
      "loss": 4.5465,
      "step": 853
    },
    {
      "epoch": 0.00854,
      "grad_norm": 0.9475750496728129,
      "learning_rate": 0.002562,
      "loss": 4.5544,
      "step": 854
    },
    {
      "epoch": 0.00855,
      "grad_norm": 0.9196861875673602,
      "learning_rate": 0.002565,
      "loss": 4.5554,
      "step": 855
    },
    {
      "epoch": 0.00856,
      "grad_norm": 0.6972970172901616,
      "learning_rate": 0.002568,
      "loss": 4.5747,
      "step": 856
    },
    {
      "epoch": 0.00857,
      "grad_norm": 0.7425487517589463,
      "learning_rate": 0.002571,
      "loss": 4.5717,
      "step": 857
    },
    {
      "epoch": 0.00858,
      "grad_norm": 0.678911102843075,
      "learning_rate": 0.002574,
      "loss": 4.5576,
      "step": 858
    },
    {
      "epoch": 0.00859,
      "grad_norm": 0.6680695283315139,
      "learning_rate": 0.002577,
      "loss": 4.5494,
      "step": 859
    },
    {
      "epoch": 0.0086,
      "grad_norm": 0.7323389776740598,
      "learning_rate": 0.00258,
      "loss": 4.5797,
      "step": 860
    },
    {
      "epoch": 0.00861,
      "grad_norm": 0.6564756555111457,
      "learning_rate": 0.0025830000000000002,
      "loss": 4.5928,
      "step": 861
    },
    {
      "epoch": 0.00862,
      "grad_norm": 0.6458920384469554,
      "learning_rate": 0.002586,
      "loss": 4.5402,
      "step": 862
    },
    {
      "epoch": 0.00863,
      "grad_norm": 0.56307833723133,
      "learning_rate": 0.002589,
      "loss": 4.5659,
      "step": 863
    },
    {
      "epoch": 0.00864,
      "grad_norm": 0.6301527121328465,
      "learning_rate": 0.002592,
      "loss": 4.5169,
      "step": 864
    },
    {
      "epoch": 0.00865,
      "grad_norm": 0.5741592962502369,
      "learning_rate": 0.002595,
      "loss": 4.5342,
      "step": 865
    },
    {
      "epoch": 0.00866,
      "grad_norm": 0.6185631950272453,
      "learning_rate": 0.002598,
      "loss": 4.5624,
      "step": 866
    },
    {
      "epoch": 0.00867,
      "grad_norm": 0.7226342011827854,
      "learning_rate": 0.002601,
      "loss": 4.5538,
      "step": 867
    },
    {
      "epoch": 0.00868,
      "grad_norm": 0.6871063796143311,
      "learning_rate": 0.002604,
      "loss": 4.5354,
      "step": 868
    },
    {
      "epoch": 0.00869,
      "grad_norm": 0.6146544871717051,
      "learning_rate": 0.002607,
      "loss": 4.5411,
      "step": 869
    },
    {
      "epoch": 0.0087,
      "grad_norm": 0.5850681754527672,
      "learning_rate": 0.00261,
      "loss": 4.5279,
      "step": 870
    },
    {
      "epoch": 0.00871,
      "grad_norm": 0.6619050810997609,
      "learning_rate": 0.002613,
      "loss": 4.5496,
      "step": 871
    },
    {
      "epoch": 0.00872,
      "grad_norm": 0.6309510333250544,
      "learning_rate": 0.002616,
      "loss": 4.5312,
      "step": 872
    },
    {
      "epoch": 0.00873,
      "grad_norm": 0.620458951387438,
      "learning_rate": 0.0026190000000000002,
      "loss": 4.5409,
      "step": 873
    },
    {
      "epoch": 0.00874,
      "grad_norm": 0.7675582961358233,
      "learning_rate": 0.002622,
      "loss": 4.5679,
      "step": 874
    },
    {
      "epoch": 0.00875,
      "grad_norm": 0.8413199287183839,
      "learning_rate": 0.002625,
      "loss": 4.5217,
      "step": 875
    },
    {
      "epoch": 0.00876,
      "grad_norm": 0.6439305106538762,
      "learning_rate": 0.002628,
      "loss": 4.5435,
      "step": 876
    },
    {
      "epoch": 0.00877,
      "grad_norm": 0.6155973146282673,
      "learning_rate": 0.002631,
      "loss": 4.5607,
      "step": 877
    },
    {
      "epoch": 0.00878,
      "grad_norm": 0.6572664983016872,
      "learning_rate": 0.002634,
      "loss": 4.4959,
      "step": 878
    },
    {
      "epoch": 0.00879,
      "grad_norm": 0.678308084092591,
      "learning_rate": 0.002637,
      "loss": 4.526,
      "step": 879
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.682644138366955,
      "learning_rate": 0.00264,
      "loss": 4.5397,
      "step": 880
    },
    {
      "epoch": 0.00881,
      "grad_norm": 0.6916960221367301,
      "learning_rate": 0.002643,
      "loss": 4.5424,
      "step": 881
    },
    {
      "epoch": 0.00882,
      "grad_norm": 0.8625989079873627,
      "learning_rate": 0.002646,
      "loss": 4.5523,
      "step": 882
    },
    {
      "epoch": 0.00883,
      "grad_norm": 0.9115056000231379,
      "learning_rate": 0.002649,
      "loss": 4.5538,
      "step": 883
    },
    {
      "epoch": 0.00884,
      "grad_norm": 1.1386535317654105,
      "learning_rate": 0.0026520000000000003,
      "loss": 4.5528,
      "step": 884
    },
    {
      "epoch": 0.00885,
      "grad_norm": 0.9401738564991132,
      "learning_rate": 0.0026550000000000002,
      "loss": 4.5478,
      "step": 885
    },
    {
      "epoch": 0.00886,
      "grad_norm": 1.0261918984073035,
      "learning_rate": 0.002658,
      "loss": 4.5785,
      "step": 886
    },
    {
      "epoch": 0.00887,
      "grad_norm": 1.2427377192417122,
      "learning_rate": 0.002661,
      "loss": 4.6,
      "step": 887
    },
    {
      "epoch": 0.00888,
      "grad_norm": 1.061546131863703,
      "learning_rate": 0.002664,
      "loss": 4.5885,
      "step": 888
    },
    {
      "epoch": 0.00889,
      "grad_norm": 0.907593068920597,
      "learning_rate": 0.002667,
      "loss": 4.535,
      "step": 889
    },
    {
      "epoch": 0.0089,
      "grad_norm": 0.8129313603021157,
      "learning_rate": 0.00267,
      "loss": 4.5855,
      "step": 890
    },
    {
      "epoch": 0.00891,
      "grad_norm": 0.8985699151758613,
      "learning_rate": 0.002673,
      "loss": 4.5762,
      "step": 891
    },
    {
      "epoch": 0.00892,
      "grad_norm": 0.9328845387585828,
      "learning_rate": 0.002676,
      "loss": 4.5962,
      "step": 892
    },
    {
      "epoch": 0.00893,
      "grad_norm": 0.9658499469408992,
      "learning_rate": 0.002679,
      "loss": 4.5743,
      "step": 893
    },
    {
      "epoch": 0.00894,
      "grad_norm": 0.9262060176852961,
      "learning_rate": 0.002682,
      "loss": 4.5624,
      "step": 894
    },
    {
      "epoch": 0.00895,
      "grad_norm": 1.0340643070320608,
      "learning_rate": 0.0026850000000000003,
      "loss": 4.5798,
      "step": 895
    },
    {
      "epoch": 0.00896,
      "grad_norm": 0.961568519948311,
      "learning_rate": 0.0026880000000000003,
      "loss": 4.5724,
      "step": 896
    },
    {
      "epoch": 0.00897,
      "grad_norm": 0.9522979094666822,
      "learning_rate": 0.0026910000000000002,
      "loss": 4.5901,
      "step": 897
    },
    {
      "epoch": 0.00898,
      "grad_norm": 0.90558868323236,
      "learning_rate": 0.002694,
      "loss": 4.5802,
      "step": 898
    },
    {
      "epoch": 0.00899,
      "grad_norm": 0.7752291080346148,
      "learning_rate": 0.002697,
      "loss": 4.5848,
      "step": 899
    },
    {
      "epoch": 0.009,
      "grad_norm": 0.7198757797557334,
      "learning_rate": 0.0027,
      "loss": 4.5998,
      "step": 900
    },
    {
      "epoch": 0.00901,
      "grad_norm": 0.7328950523365488,
      "learning_rate": 0.002703,
      "loss": 4.5552,
      "step": 901
    },
    {
      "epoch": 0.00902,
      "grad_norm": 0.5974121750232518,
      "learning_rate": 0.002706,
      "loss": 4.558,
      "step": 902
    },
    {
      "epoch": 0.00903,
      "grad_norm": 0.5753314236304967,
      "learning_rate": 0.002709,
      "loss": 4.5534,
      "step": 903
    },
    {
      "epoch": 0.00904,
      "grad_norm": 0.6128643778394687,
      "learning_rate": 0.002712,
      "loss": 4.5329,
      "step": 904
    },
    {
      "epoch": 0.00905,
      "grad_norm": 0.5691916638541509,
      "learning_rate": 0.002715,
      "loss": 4.5446,
      "step": 905
    },
    {
      "epoch": 0.00906,
      "grad_norm": 0.5122473095411203,
      "learning_rate": 0.002718,
      "loss": 4.5538,
      "step": 906
    },
    {
      "epoch": 0.00907,
      "grad_norm": 0.4479956748150508,
      "learning_rate": 0.0027210000000000003,
      "loss": 4.5603,
      "step": 907
    },
    {
      "epoch": 0.00908,
      "grad_norm": 0.4387217017050332,
      "learning_rate": 0.0027240000000000003,
      "loss": 4.5304,
      "step": 908
    },
    {
      "epoch": 0.00909,
      "grad_norm": 0.4358600325368994,
      "learning_rate": 0.0027270000000000003,
      "loss": 4.5454,
      "step": 909
    },
    {
      "epoch": 0.0091,
      "grad_norm": 0.3849373702796595,
      "learning_rate": 0.0027300000000000002,
      "loss": 4.5477,
      "step": 910
    },
    {
      "epoch": 0.00911,
      "grad_norm": 0.4574591880095763,
      "learning_rate": 0.002733,
      "loss": 4.5038,
      "step": 911
    },
    {
      "epoch": 0.00912,
      "grad_norm": 0.6121326743360266,
      "learning_rate": 0.002736,
      "loss": 4.5131,
      "step": 912
    },
    {
      "epoch": 0.00913,
      "grad_norm": 0.6845161679319078,
      "learning_rate": 0.002739,
      "loss": 4.513,
      "step": 913
    },
    {
      "epoch": 0.00914,
      "grad_norm": 0.7333722528690678,
      "learning_rate": 0.002742,
      "loss": 4.5432,
      "step": 914
    },
    {
      "epoch": 0.00915,
      "grad_norm": 0.6714442153045657,
      "learning_rate": 0.002745,
      "loss": 4.5048,
      "step": 915
    },
    {
      "epoch": 0.00916,
      "grad_norm": 0.5861682792098877,
      "learning_rate": 0.002748,
      "loss": 4.4768,
      "step": 916
    },
    {
      "epoch": 0.00917,
      "grad_norm": 0.7568015257717708,
      "learning_rate": 0.002751,
      "loss": 4.4985,
      "step": 917
    },
    {
      "epoch": 0.00918,
      "grad_norm": 0.67845085346177,
      "learning_rate": 0.0027540000000000004,
      "loss": 4.5185,
      "step": 918
    },
    {
      "epoch": 0.00919,
      "grad_norm": 0.6376249584521229,
      "learning_rate": 0.0027570000000000003,
      "loss": 4.4976,
      "step": 919
    },
    {
      "epoch": 0.0092,
      "grad_norm": 0.5953295490629925,
      "learning_rate": 0.0027600000000000003,
      "loss": 4.5104,
      "step": 920
    },
    {
      "epoch": 0.00921,
      "grad_norm": 0.5470113372090761,
      "learning_rate": 0.0027630000000000003,
      "loss": 4.4892,
      "step": 921
    },
    {
      "epoch": 0.00922,
      "grad_norm": 0.5586775976924615,
      "learning_rate": 0.0027660000000000002,
      "loss": 4.5007,
      "step": 922
    },
    {
      "epoch": 0.00923,
      "grad_norm": 0.5602976125001545,
      "learning_rate": 0.002769,
      "loss": 4.4767,
      "step": 923
    },
    {
      "epoch": 0.00924,
      "grad_norm": 0.5512257756150963,
      "learning_rate": 0.002772,
      "loss": 4.4992,
      "step": 924
    },
    {
      "epoch": 0.00925,
      "grad_norm": 0.6104014154605217,
      "learning_rate": 0.002775,
      "loss": 4.4804,
      "step": 925
    },
    {
      "epoch": 0.00926,
      "grad_norm": 0.627882987139063,
      "learning_rate": 0.002778,
      "loss": 4.4994,
      "step": 926
    },
    {
      "epoch": 0.00927,
      "grad_norm": 0.5637389395797978,
      "learning_rate": 0.002781,
      "loss": 4.4582,
      "step": 927
    },
    {
      "epoch": 0.00928,
      "grad_norm": 0.5793012970445044,
      "learning_rate": 0.002784,
      "loss": 4.505,
      "step": 928
    },
    {
      "epoch": 0.00929,
      "grad_norm": 0.5538600481661593,
      "learning_rate": 0.0027870000000000004,
      "loss": 4.4867,
      "step": 929
    },
    {
      "epoch": 0.0093,
      "grad_norm": 0.4915384406563099,
      "learning_rate": 0.0027900000000000004,
      "loss": 4.4652,
      "step": 930
    },
    {
      "epoch": 0.00931,
      "grad_norm": 0.5291025545236148,
      "learning_rate": 0.0027930000000000003,
      "loss": 4.4796,
      "step": 931
    },
    {
      "epoch": 0.00932,
      "grad_norm": 0.5189072536981689,
      "learning_rate": 0.0027960000000000003,
      "loss": 4.4881,
      "step": 932
    },
    {
      "epoch": 0.00933,
      "grad_norm": 0.4391692113712953,
      "learning_rate": 0.0027990000000000003,
      "loss": 4.4495,
      "step": 933
    },
    {
      "epoch": 0.00934,
      "grad_norm": 0.5067956527040228,
      "learning_rate": 0.0028020000000000002,
      "loss": 4.4811,
      "step": 934
    },
    {
      "epoch": 0.00935,
      "grad_norm": 0.5752082887405049,
      "learning_rate": 0.002805,
      "loss": 4.4893,
      "step": 935
    },
    {
      "epoch": 0.00936,
      "grad_norm": 0.8597610157431208,
      "learning_rate": 0.002808,
      "loss": 4.5012,
      "step": 936
    },
    {
      "epoch": 0.00937,
      "grad_norm": 1.0923006968336397,
      "learning_rate": 0.002811,
      "loss": 4.5462,
      "step": 937
    },
    {
      "epoch": 0.00938,
      "grad_norm": 0.8211742236965874,
      "learning_rate": 0.002814,
      "loss": 4.5052,
      "step": 938
    },
    {
      "epoch": 0.00939,
      "grad_norm": 0.699189062953733,
      "learning_rate": 0.002817,
      "loss": 4.4989,
      "step": 939
    },
    {
      "epoch": 0.0094,
      "grad_norm": 0.6251265662758575,
      "learning_rate": 0.00282,
      "loss": 4.495,
      "step": 940
    },
    {
      "epoch": 0.00941,
      "grad_norm": 0.8614570213641495,
      "learning_rate": 0.002823,
      "loss": 4.4823,
      "step": 941
    },
    {
      "epoch": 0.00942,
      "grad_norm": 0.9710835908546839,
      "learning_rate": 0.002826,
      "loss": 4.5282,
      "step": 942
    },
    {
      "epoch": 0.00943,
      "grad_norm": 0.8255156288963836,
      "learning_rate": 0.002829,
      "loss": 4.5175,
      "step": 943
    },
    {
      "epoch": 0.00944,
      "grad_norm": 0.8529303551550831,
      "learning_rate": 0.002832,
      "loss": 4.5233,
      "step": 944
    },
    {
      "epoch": 0.00945,
      "grad_norm": 0.8805636588256177,
      "learning_rate": 0.002835,
      "loss": 4.5307,
      "step": 945
    },
    {
      "epoch": 0.00946,
      "grad_norm": 0.8815394682869104,
      "learning_rate": 0.002838,
      "loss": 4.545,
      "step": 946
    },
    {
      "epoch": 0.00947,
      "grad_norm": 0.9792537736888025,
      "learning_rate": 0.0028409999999999998,
      "loss": 4.5259,
      "step": 947
    },
    {
      "epoch": 0.00948,
      "grad_norm": 1.16480627354709,
      "learning_rate": 0.0028439999999999997,
      "loss": 4.5258,
      "step": 948
    },
    {
      "epoch": 0.00949,
      "grad_norm": 0.9311558570406092,
      "learning_rate": 0.002847,
      "loss": 4.5678,
      "step": 949
    },
    {
      "epoch": 0.0095,
      "grad_norm": 0.9151500308490546,
      "learning_rate": 0.00285,
      "loss": 4.5566,
      "step": 950
    },
    {
      "epoch": 0.00951,
      "grad_norm": 0.8788038193040383,
      "learning_rate": 0.002853,
      "loss": 4.539,
      "step": 951
    },
    {
      "epoch": 0.00952,
      "grad_norm": 0.6883343465241535,
      "learning_rate": 0.002856,
      "loss": 4.5492,
      "step": 952
    },
    {
      "epoch": 0.00953,
      "grad_norm": 0.7428316161957196,
      "learning_rate": 0.002859,
      "loss": 4.4912,
      "step": 953
    },
    {
      "epoch": 0.00954,
      "grad_norm": 0.6589075049765231,
      "learning_rate": 0.002862,
      "loss": 4.5478,
      "step": 954
    },
    {
      "epoch": 0.00955,
      "grad_norm": 0.6374906159512967,
      "learning_rate": 0.002865,
      "loss": 4.5305,
      "step": 955
    },
    {
      "epoch": 0.00956,
      "grad_norm": 0.7037593758453927,
      "learning_rate": 0.002868,
      "loss": 4.5123,
      "step": 956
    },
    {
      "epoch": 0.00957,
      "grad_norm": 0.7575837270625414,
      "learning_rate": 0.002871,
      "loss": 4.5364,
      "step": 957
    },
    {
      "epoch": 0.00958,
      "grad_norm": 0.8136537306721928,
      "learning_rate": 0.002874,
      "loss": 4.5221,
      "step": 958
    },
    {
      "epoch": 0.00959,
      "grad_norm": 0.7529303499273822,
      "learning_rate": 0.002877,
      "loss": 4.5134,
      "step": 959
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.7723176518474248,
      "learning_rate": 0.0028799999999999997,
      "loss": 4.5128,
      "step": 960
    },
    {
      "epoch": 0.00961,
      "grad_norm": 0.6826701565045171,
      "learning_rate": 0.002883,
      "loss": 4.4802,
      "step": 961
    },
    {
      "epoch": 0.00962,
      "grad_norm": 0.6022941634910601,
      "learning_rate": 0.002886,
      "loss": 4.5086,
      "step": 962
    },
    {
      "epoch": 0.00963,
      "grad_norm": 0.5423609760386227,
      "learning_rate": 0.002889,
      "loss": 4.5025,
      "step": 963
    },
    {
      "epoch": 0.00964,
      "grad_norm": 0.5430502092899187,
      "learning_rate": 0.002892,
      "loss": 4.493,
      "step": 964
    },
    {
      "epoch": 0.00965,
      "grad_norm": 0.5046734063237629,
      "learning_rate": 0.002895,
      "loss": 4.4643,
      "step": 965
    },
    {
      "epoch": 0.00966,
      "grad_norm": 0.4625107724617308,
      "learning_rate": 0.002898,
      "loss": 4.4758,
      "step": 966
    },
    {
      "epoch": 0.00967,
      "grad_norm": 0.4403845150464526,
      "learning_rate": 0.002901,
      "loss": 4.5117,
      "step": 967
    },
    {
      "epoch": 0.00968,
      "grad_norm": 0.40739936008519134,
      "learning_rate": 0.002904,
      "loss": 4.4717,
      "step": 968
    },
    {
      "epoch": 0.00969,
      "grad_norm": 0.41620271754109195,
      "learning_rate": 0.002907,
      "loss": 4.5059,
      "step": 969
    },
    {
      "epoch": 0.0097,
      "grad_norm": 0.3884080596123629,
      "learning_rate": 0.00291,
      "loss": 4.4561,
      "step": 970
    },
    {
      "epoch": 0.00971,
      "grad_norm": 0.40963591049792675,
      "learning_rate": 0.002913,
      "loss": 4.4624,
      "step": 971
    },
    {
      "epoch": 0.00972,
      "grad_norm": 0.427860611994679,
      "learning_rate": 0.002916,
      "loss": 4.4693,
      "step": 972
    },
    {
      "epoch": 0.00973,
      "grad_norm": 0.6266830722962007,
      "learning_rate": 0.002919,
      "loss": 4.4689,
      "step": 973
    },
    {
      "epoch": 0.00974,
      "grad_norm": 0.7795241673951117,
      "learning_rate": 0.002922,
      "loss": 4.4734,
      "step": 974
    },
    {
      "epoch": 0.00975,
      "grad_norm": 0.7392088002205494,
      "learning_rate": 0.002925,
      "loss": 4.4915,
      "step": 975
    },
    {
      "epoch": 0.00976,
      "grad_norm": 0.6090919745791438,
      "learning_rate": 0.002928,
      "loss": 4.4688,
      "step": 976
    },
    {
      "epoch": 0.00977,
      "grad_norm": 0.6721667952819499,
      "learning_rate": 0.002931,
      "loss": 4.5114,
      "step": 977
    },
    {
      "epoch": 0.00978,
      "grad_norm": 0.7157395676490952,
      "learning_rate": 0.002934,
      "loss": 4.4828,
      "step": 978
    },
    {
      "epoch": 0.00979,
      "grad_norm": 0.6673604586868628,
      "learning_rate": 0.002937,
      "loss": 4.4724,
      "step": 979
    },
    {
      "epoch": 0.0098,
      "grad_norm": 0.6241827666303977,
      "learning_rate": 0.00294,
      "loss": 4.4641,
      "step": 980
    },
    {
      "epoch": 0.00981,
      "grad_norm": 0.6477735028588306,
      "learning_rate": 0.002943,
      "loss": 4.479,
      "step": 981
    },
    {
      "epoch": 0.00982,
      "grad_norm": 0.5903229373466107,
      "learning_rate": 0.002946,
      "loss": 4.4717,
      "step": 982
    },
    {
      "epoch": 0.00983,
      "grad_norm": 0.5367081021491611,
      "learning_rate": 0.0029490000000000002,
      "loss": 4.4684,
      "step": 983
    },
    {
      "epoch": 0.00984,
      "grad_norm": 0.55855952891459,
      "learning_rate": 0.002952,
      "loss": 4.455,
      "step": 984
    },
    {
      "epoch": 0.00985,
      "grad_norm": 0.6566800838357816,
      "learning_rate": 0.002955,
      "loss": 4.4789,
      "step": 985
    },
    {
      "epoch": 0.00986,
      "grad_norm": 0.7602203626787503,
      "learning_rate": 0.002958,
      "loss": 4.4461,
      "step": 986
    },
    {
      "epoch": 0.00987,
      "grad_norm": 0.7515002206199092,
      "learning_rate": 0.002961,
      "loss": 4.4656,
      "step": 987
    },
    {
      "epoch": 0.00988,
      "grad_norm": 0.7510028045825011,
      "learning_rate": 0.002964,
      "loss": 4.497,
      "step": 988
    },
    {
      "epoch": 0.00989,
      "grad_norm": 0.6686524266681442,
      "learning_rate": 0.002967,
      "loss": 4.438,
      "step": 989
    },
    {
      "epoch": 0.0099,
      "grad_norm": 0.7110181433203292,
      "learning_rate": 0.00297,
      "loss": 4.4759,
      "step": 990
    },
    {
      "epoch": 0.00991,
      "grad_norm": 0.9487700940206629,
      "learning_rate": 0.002973,
      "loss": 4.4909,
      "step": 991
    },
    {
      "epoch": 0.00992,
      "grad_norm": 1.0128664486589338,
      "learning_rate": 0.002976,
      "loss": 4.4936,
      "step": 992
    },
    {
      "epoch": 0.00993,
      "grad_norm": 0.776315556008045,
      "learning_rate": 0.002979,
      "loss": 4.4927,
      "step": 993
    },
    {
      "epoch": 0.00994,
      "grad_norm": 0.6954881784010936,
      "learning_rate": 0.002982,
      "loss": 4.4894,
      "step": 994
    },
    {
      "epoch": 0.00995,
      "grad_norm": 0.8343170542469196,
      "learning_rate": 0.0029850000000000002,
      "loss": 4.4966,
      "step": 995
    },
    {
      "epoch": 0.00996,
      "grad_norm": 0.8064256733311401,
      "learning_rate": 0.002988,
      "loss": 4.4974,
      "step": 996
    },
    {
      "epoch": 0.00997,
      "grad_norm": 0.7600837252115415,
      "learning_rate": 0.002991,
      "loss": 4.4859,
      "step": 997
    },
    {
      "epoch": 0.00998,
      "grad_norm": 0.7413851219310601,
      "learning_rate": 0.002994,
      "loss": 4.4878,
      "step": 998
    },
    {
      "epoch": 0.00999,
      "grad_norm": 0.7789142980889687,
      "learning_rate": 0.002997,
      "loss": 4.4774,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5973093879700552,
      "learning_rate": 0.003,
      "loss": 4.4903,
      "step": 1000
    },
    {
      "epoch": 0.01001,
      "grad_norm": 0.5861183186896064,
      "learning_rate": 0.003,
      "loss": 4.4874,
      "step": 1001
    },
    {
      "epoch": 0.01002,
      "grad_norm": 0.6212400302116947,
      "learning_rate": 0.003,
      "loss": 4.4557,
      "step": 1002
    },
    {
      "epoch": 0.01003,
      "grad_norm": 0.6083357500633214,
      "learning_rate": 0.003,
      "loss": 4.4779,
      "step": 1003
    },
    {
      "epoch": 0.01004,
      "grad_norm": 0.5830404519486425,
      "learning_rate": 0.003,
      "loss": 4.4653,
      "step": 1004
    },
    {
      "epoch": 0.01005,
      "grad_norm": 0.7477664572876674,
      "learning_rate": 0.003,
      "loss": 4.4465,
      "step": 1005
    },
    {
      "epoch": 0.01006,
      "grad_norm": 0.7841384646860249,
      "learning_rate": 0.003,
      "loss": 4.449,
      "step": 1006
    },
    {
      "epoch": 0.01007,
      "grad_norm": 0.7235135066769743,
      "learning_rate": 0.003,
      "loss": 4.4927,
      "step": 1007
    },
    {
      "epoch": 0.01008,
      "grad_norm": 0.6474532468386683,
      "learning_rate": 0.003,
      "loss": 4.4993,
      "step": 1008
    },
    {
      "epoch": 0.01009,
      "grad_norm": 0.5980408215414467,
      "learning_rate": 0.003,
      "loss": 4.4651,
      "step": 1009
    },
    {
      "epoch": 0.0101,
      "grad_norm": 0.5537282779479755,
      "learning_rate": 0.003,
      "loss": 4.4693,
      "step": 1010
    },
    {
      "epoch": 0.01011,
      "grad_norm": 0.5708819324924145,
      "learning_rate": 0.003,
      "loss": 4.4785,
      "step": 1011
    },
    {
      "epoch": 0.01012,
      "grad_norm": 0.4974883261450977,
      "learning_rate": 0.003,
      "loss": 4.4576,
      "step": 1012
    },
    {
      "epoch": 0.01013,
      "grad_norm": 0.4665247551568978,
      "learning_rate": 0.003,
      "loss": 4.4798,
      "step": 1013
    },
    {
      "epoch": 0.01014,
      "grad_norm": 0.42165803422689113,
      "learning_rate": 0.003,
      "loss": 4.4182,
      "step": 1014
    },
    {
      "epoch": 0.01015,
      "grad_norm": 0.4067134475964037,
      "learning_rate": 0.003,
      "loss": 4.4608,
      "step": 1015
    },
    {
      "epoch": 0.01016,
      "grad_norm": 0.5123147411688123,
      "learning_rate": 0.003,
      "loss": 4.4714,
      "step": 1016
    },
    {
      "epoch": 0.01017,
      "grad_norm": 0.6275035755668664,
      "learning_rate": 0.003,
      "loss": 4.4159,
      "step": 1017
    },
    {
      "epoch": 0.01018,
      "grad_norm": 0.8581692725167155,
      "learning_rate": 0.003,
      "loss": 4.4817,
      "step": 1018
    },
    {
      "epoch": 0.01019,
      "grad_norm": 0.8393947155453974,
      "learning_rate": 0.003,
      "loss": 4.4755,
      "step": 1019
    },
    {
      "epoch": 0.0102,
      "grad_norm": 0.6933428051970444,
      "learning_rate": 0.003,
      "loss": 4.4592,
      "step": 1020
    },
    {
      "epoch": 0.01021,
      "grad_norm": 0.7990025577528214,
      "learning_rate": 0.003,
      "loss": 4.4512,
      "step": 1021
    },
    {
      "epoch": 0.01022,
      "grad_norm": 0.8110623623420127,
      "learning_rate": 0.003,
      "loss": 4.4643,
      "step": 1022
    },
    {
      "epoch": 0.01023,
      "grad_norm": 0.7434736091722897,
      "learning_rate": 0.003,
      "loss": 4.4609,
      "step": 1023
    },
    {
      "epoch": 0.01024,
      "grad_norm": 0.7213203816141717,
      "learning_rate": 0.003,
      "loss": 4.417,
      "step": 1024
    },
    {
      "epoch": 0.01025,
      "grad_norm": 0.7010589428906292,
      "learning_rate": 0.003,
      "loss": 4.4518,
      "step": 1025
    },
    {
      "epoch": 0.01026,
      "grad_norm": 0.6280712957780686,
      "learning_rate": 0.003,
      "loss": 4.44,
      "step": 1026
    },
    {
      "epoch": 0.01027,
      "grad_norm": 0.5410145302803399,
      "learning_rate": 0.003,
      "loss": 4.4579,
      "step": 1027
    },
    {
      "epoch": 0.01028,
      "grad_norm": 0.4828232064846382,
      "learning_rate": 0.003,
      "loss": 4.4353,
      "step": 1028
    },
    {
      "epoch": 0.01029,
      "grad_norm": 0.4739043118914256,
      "learning_rate": 0.003,
      "loss": 4.412,
      "step": 1029
    },
    {
      "epoch": 0.0103,
      "grad_norm": 0.45225551629540334,
      "learning_rate": 0.003,
      "loss": 4.4408,
      "step": 1030
    },
    {
      "epoch": 0.01031,
      "grad_norm": 0.40901639056794215,
      "learning_rate": 0.003,
      "loss": 4.4521,
      "step": 1031
    },
    {
      "epoch": 0.01032,
      "grad_norm": 0.4266157306452721,
      "learning_rate": 0.003,
      "loss": 4.4241,
      "step": 1032
    },
    {
      "epoch": 0.01033,
      "grad_norm": 0.476318572377966,
      "learning_rate": 0.003,
      "loss": 4.4623,
      "step": 1033
    },
    {
      "epoch": 0.01034,
      "grad_norm": 0.4976367752405493,
      "learning_rate": 0.003,
      "loss": 4.4286,
      "step": 1034
    },
    {
      "epoch": 0.01035,
      "grad_norm": 0.48706616162522565,
      "learning_rate": 0.003,
      "loss": 4.427,
      "step": 1035
    },
    {
      "epoch": 0.01036,
      "grad_norm": 0.5540613528135478,
      "learning_rate": 0.003,
      "loss": 4.4185,
      "step": 1036
    },
    {
      "epoch": 0.01037,
      "grad_norm": 0.5585932570414466,
      "learning_rate": 0.003,
      "loss": 4.4458,
      "step": 1037
    },
    {
      "epoch": 0.01038,
      "grad_norm": 0.5014561937587317,
      "learning_rate": 0.003,
      "loss": 4.44,
      "step": 1038
    },
    {
      "epoch": 0.01039,
      "grad_norm": 0.5070723079813054,
      "learning_rate": 0.003,
      "loss": 4.4296,
      "step": 1039
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.5609823517547058,
      "learning_rate": 0.003,
      "loss": 4.4462,
      "step": 1040
    },
    {
      "epoch": 0.01041,
      "grad_norm": 0.664341991841828,
      "learning_rate": 0.003,
      "loss": 4.4332,
      "step": 1041
    },
    {
      "epoch": 0.01042,
      "grad_norm": 0.631896992045698,
      "learning_rate": 0.003,
      "loss": 4.458,
      "step": 1042
    },
    {
      "epoch": 0.01043,
      "grad_norm": 0.8655430264592057,
      "learning_rate": 0.003,
      "loss": 4.4103,
      "step": 1043
    },
    {
      "epoch": 0.01044,
      "grad_norm": 1.0843670364695495,
      "learning_rate": 0.003,
      "loss": 4.4702,
      "step": 1044
    },
    {
      "epoch": 0.01045,
      "grad_norm": 0.9544830822145469,
      "learning_rate": 0.003,
      "loss": 4.4986,
      "step": 1045
    },
    {
      "epoch": 0.01046,
      "grad_norm": 0.9579295900139632,
      "learning_rate": 0.003,
      "loss": 4.4718,
      "step": 1046
    },
    {
      "epoch": 0.01047,
      "grad_norm": 0.9275123969574217,
      "learning_rate": 0.003,
      "loss": 4.4903,
      "step": 1047
    },
    {
      "epoch": 0.01048,
      "grad_norm": 0.8417384302831535,
      "learning_rate": 0.003,
      "loss": 4.4694,
      "step": 1048
    },
    {
      "epoch": 0.01049,
      "grad_norm": 0.7711730353324007,
      "learning_rate": 0.003,
      "loss": 4.449,
      "step": 1049
    },
    {
      "epoch": 0.0105,
      "grad_norm": 0.747892326408584,
      "learning_rate": 0.003,
      "loss": 4.4932,
      "step": 1050
    },
    {
      "epoch": 0.01051,
      "grad_norm": 0.8100969209243005,
      "learning_rate": 0.003,
      "loss": 4.4724,
      "step": 1051
    },
    {
      "epoch": 0.01052,
      "grad_norm": 0.7211696952483453,
      "learning_rate": 0.003,
      "loss": 4.4859,
      "step": 1052
    },
    {
      "epoch": 0.01053,
      "grad_norm": 0.761937042029771,
      "learning_rate": 0.003,
      "loss": 4.4907,
      "step": 1053
    },
    {
      "epoch": 0.01054,
      "grad_norm": 0.791018085034848,
      "learning_rate": 0.003,
      "loss": 4.4849,
      "step": 1054
    },
    {
      "epoch": 0.01055,
      "grad_norm": 0.76513637315221,
      "learning_rate": 0.003,
      "loss": 4.464,
      "step": 1055
    },
    {
      "epoch": 0.01056,
      "grad_norm": 0.844999763652046,
      "learning_rate": 0.003,
      "loss": 4.4686,
      "step": 1056
    },
    {
      "epoch": 0.01057,
      "grad_norm": 0.7706046759321704,
      "learning_rate": 0.003,
      "loss": 4.476,
      "step": 1057
    },
    {
      "epoch": 0.01058,
      "grad_norm": 0.6536852085679028,
      "learning_rate": 0.003,
      "loss": 4.4607,
      "step": 1058
    },
    {
      "epoch": 0.01059,
      "grad_norm": 0.7239862733740466,
      "learning_rate": 0.003,
      "loss": 4.4656,
      "step": 1059
    },
    {
      "epoch": 0.0106,
      "grad_norm": 0.8167528488457484,
      "learning_rate": 0.003,
      "loss": 4.4597,
      "step": 1060
    },
    {
      "epoch": 0.01061,
      "grad_norm": 0.6791440431932436,
      "learning_rate": 0.003,
      "loss": 4.4551,
      "step": 1061
    },
    {
      "epoch": 0.01062,
      "grad_norm": 0.7024673414145854,
      "learning_rate": 0.003,
      "loss": 4.4488,
      "step": 1062
    },
    {
      "epoch": 0.01063,
      "grad_norm": 0.5950427080817663,
      "learning_rate": 0.003,
      "loss": 4.4376,
      "step": 1063
    },
    {
      "epoch": 0.01064,
      "grad_norm": 0.4616517916448735,
      "learning_rate": 0.003,
      "loss": 4.4487,
      "step": 1064
    },
    {
      "epoch": 0.01065,
      "grad_norm": 0.43744796242075984,
      "learning_rate": 0.003,
      "loss": 4.4449,
      "step": 1065
    },
    {
      "epoch": 0.01066,
      "grad_norm": 0.3954250450266796,
      "learning_rate": 0.003,
      "loss": 4.4441,
      "step": 1066
    },
    {
      "epoch": 0.01067,
      "grad_norm": 0.44911980407136426,
      "learning_rate": 0.003,
      "loss": 4.4535,
      "step": 1067
    },
    {
      "epoch": 0.01068,
      "grad_norm": 0.6121110106038351,
      "learning_rate": 0.003,
      "loss": 4.4381,
      "step": 1068
    },
    {
      "epoch": 0.01069,
      "grad_norm": 0.8069569048435549,
      "learning_rate": 0.003,
      "loss": 4.4677,
      "step": 1069
    },
    {
      "epoch": 0.0107,
      "grad_norm": 0.9567755776549173,
      "learning_rate": 0.003,
      "loss": 4.5257,
      "step": 1070
    },
    {
      "epoch": 0.01071,
      "grad_norm": 0.7710306613088521,
      "learning_rate": 0.003,
      "loss": 4.4413,
      "step": 1071
    },
    {
      "epoch": 0.01072,
      "grad_norm": 0.6442072811444085,
      "learning_rate": 0.003,
      "loss": 4.4742,
      "step": 1072
    },
    {
      "epoch": 0.01073,
      "grad_norm": 0.6426873609153096,
      "learning_rate": 0.003,
      "loss": 4.464,
      "step": 1073
    },
    {
      "epoch": 0.01074,
      "grad_norm": 0.5296718591337678,
      "learning_rate": 0.003,
      "loss": 4.4244,
      "step": 1074
    },
    {
      "epoch": 0.01075,
      "grad_norm": 0.4598820284284391,
      "learning_rate": 0.003,
      "loss": 4.4589,
      "step": 1075
    },
    {
      "epoch": 0.01076,
      "grad_norm": 0.4671476178414801,
      "learning_rate": 0.003,
      "loss": 4.4514,
      "step": 1076
    },
    {
      "epoch": 0.01077,
      "grad_norm": 0.5378166305898745,
      "learning_rate": 0.003,
      "loss": 4.4483,
      "step": 1077
    },
    {
      "epoch": 0.01078,
      "grad_norm": 0.4897385837475064,
      "learning_rate": 0.003,
      "loss": 4.3993,
      "step": 1078
    },
    {
      "epoch": 0.01079,
      "grad_norm": 0.40845103276318817,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1079
    },
    {
      "epoch": 0.0108,
      "grad_norm": 0.36771124599267696,
      "learning_rate": 0.003,
      "loss": 4.4028,
      "step": 1080
    },
    {
      "epoch": 0.01081,
      "grad_norm": 0.4088680521129332,
      "learning_rate": 0.003,
      "loss": 4.41,
      "step": 1081
    },
    {
      "epoch": 0.01082,
      "grad_norm": 0.41501548740150246,
      "learning_rate": 0.003,
      "loss": 4.4279,
      "step": 1082
    },
    {
      "epoch": 0.01083,
      "grad_norm": 0.3846754433134812,
      "learning_rate": 0.003,
      "loss": 4.4131,
      "step": 1083
    },
    {
      "epoch": 0.01084,
      "grad_norm": 0.3759434314813712,
      "learning_rate": 0.003,
      "loss": 4.4174,
      "step": 1084
    },
    {
      "epoch": 0.01085,
      "grad_norm": 0.38199974093520966,
      "learning_rate": 0.003,
      "loss": 4.4048,
      "step": 1085
    },
    {
      "epoch": 0.01086,
      "grad_norm": 0.417068769159195,
      "learning_rate": 0.003,
      "loss": 4.3806,
      "step": 1086
    },
    {
      "epoch": 0.01087,
      "grad_norm": 0.4600653555422747,
      "learning_rate": 0.003,
      "loss": 4.3971,
      "step": 1087
    },
    {
      "epoch": 0.01088,
      "grad_norm": 0.46093681942570613,
      "learning_rate": 0.003,
      "loss": 4.4326,
      "step": 1088
    },
    {
      "epoch": 0.01089,
      "grad_norm": 0.3917178547896987,
      "learning_rate": 0.003,
      "loss": 4.3982,
      "step": 1089
    },
    {
      "epoch": 0.0109,
      "grad_norm": 0.40807888361788897,
      "learning_rate": 0.003,
      "loss": 4.4241,
      "step": 1090
    },
    {
      "epoch": 0.01091,
      "grad_norm": 0.48077039467346944,
      "learning_rate": 0.003,
      "loss": 4.3944,
      "step": 1091
    },
    {
      "epoch": 0.01092,
      "grad_norm": 0.5412841894896883,
      "learning_rate": 0.003,
      "loss": 4.3919,
      "step": 1092
    },
    {
      "epoch": 0.01093,
      "grad_norm": 0.6050990959157666,
      "learning_rate": 0.003,
      "loss": 4.4164,
      "step": 1093
    },
    {
      "epoch": 0.01094,
      "grad_norm": 0.7349015923043832,
      "learning_rate": 0.003,
      "loss": 4.4107,
      "step": 1094
    },
    {
      "epoch": 0.01095,
      "grad_norm": 0.8311905656610761,
      "learning_rate": 0.003,
      "loss": 4.4503,
      "step": 1095
    },
    {
      "epoch": 0.01096,
      "grad_norm": 0.7514639212089553,
      "learning_rate": 0.003,
      "loss": 4.4205,
      "step": 1096
    },
    {
      "epoch": 0.01097,
      "grad_norm": 0.7811392154893482,
      "learning_rate": 0.003,
      "loss": 4.4143,
      "step": 1097
    },
    {
      "epoch": 0.01098,
      "grad_norm": 0.784974458960817,
      "learning_rate": 0.003,
      "loss": 4.4358,
      "step": 1098
    },
    {
      "epoch": 0.01099,
      "grad_norm": 0.8366574206737059,
      "learning_rate": 0.003,
      "loss": 4.4105,
      "step": 1099
    },
    {
      "epoch": 0.011,
      "grad_norm": 0.8308538935944765,
      "learning_rate": 0.003,
      "loss": 4.4197,
      "step": 1100
    },
    {
      "epoch": 0.01101,
      "grad_norm": 0.7744395626168694,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1101
    },
    {
      "epoch": 0.01102,
      "grad_norm": 0.790296389339303,
      "learning_rate": 0.003,
      "loss": 4.4394,
      "step": 1102
    },
    {
      "epoch": 0.01103,
      "grad_norm": 0.888052985760201,
      "learning_rate": 0.003,
      "loss": 4.4459,
      "step": 1103
    },
    {
      "epoch": 0.01104,
      "grad_norm": 0.7149547126246749,
      "learning_rate": 0.003,
      "loss": 4.429,
      "step": 1104
    },
    {
      "epoch": 0.01105,
      "grad_norm": 0.6372550706190871,
      "learning_rate": 0.003,
      "loss": 4.4469,
      "step": 1105
    },
    {
      "epoch": 0.01106,
      "grad_norm": 0.6914402582964255,
      "learning_rate": 0.003,
      "loss": 4.4208,
      "step": 1106
    },
    {
      "epoch": 0.01107,
      "grad_norm": 0.6242116385034774,
      "learning_rate": 0.003,
      "loss": 4.4302,
      "step": 1107
    },
    {
      "epoch": 0.01108,
      "grad_norm": 0.5046709947552265,
      "learning_rate": 0.003,
      "loss": 4.3916,
      "step": 1108
    },
    {
      "epoch": 0.01109,
      "grad_norm": 0.44296219090623196,
      "learning_rate": 0.003,
      "loss": 4.4076,
      "step": 1109
    },
    {
      "epoch": 0.0111,
      "grad_norm": 0.42911746175989035,
      "learning_rate": 0.003,
      "loss": 4.4253,
      "step": 1110
    },
    {
      "epoch": 0.01111,
      "grad_norm": 0.4399471864730289,
      "learning_rate": 0.003,
      "loss": 4.401,
      "step": 1111
    },
    {
      "epoch": 0.01112,
      "grad_norm": 0.4608476260695391,
      "learning_rate": 0.003,
      "loss": 4.4197,
      "step": 1112
    },
    {
      "epoch": 0.01113,
      "grad_norm": 0.48813138086989816,
      "learning_rate": 0.003,
      "loss": 4.388,
      "step": 1113
    },
    {
      "epoch": 0.01114,
      "grad_norm": 0.5565547407761932,
      "learning_rate": 0.003,
      "loss": 4.4024,
      "step": 1114
    },
    {
      "epoch": 0.01115,
      "grad_norm": 0.6052109602996041,
      "learning_rate": 0.003,
      "loss": 4.4297,
      "step": 1115
    },
    {
      "epoch": 0.01116,
      "grad_norm": 0.6641979249916196,
      "learning_rate": 0.003,
      "loss": 4.4207,
      "step": 1116
    },
    {
      "epoch": 0.01117,
      "grad_norm": 0.6083566793907378,
      "learning_rate": 0.003,
      "loss": 4.4056,
      "step": 1117
    },
    {
      "epoch": 0.01118,
      "grad_norm": 0.5238007850801012,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1118
    },
    {
      "epoch": 0.01119,
      "grad_norm": 0.5793839332710249,
      "learning_rate": 0.003,
      "loss": 4.4112,
      "step": 1119
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.5969906941980586,
      "learning_rate": 0.003,
      "loss": 4.3958,
      "step": 1120
    },
    {
      "epoch": 0.01121,
      "grad_norm": 0.6030781006503133,
      "learning_rate": 0.003,
      "loss": 4.4075,
      "step": 1121
    },
    {
      "epoch": 0.01122,
      "grad_norm": 0.5551009392133712,
      "learning_rate": 0.003,
      "loss": 4.382,
      "step": 1122
    },
    {
      "epoch": 0.01123,
      "grad_norm": 0.5109047416253846,
      "learning_rate": 0.003,
      "loss": 4.3689,
      "step": 1123
    },
    {
      "epoch": 0.01124,
      "grad_norm": 0.5407283441136803,
      "learning_rate": 0.003,
      "loss": 4.4164,
      "step": 1124
    },
    {
      "epoch": 0.01125,
      "grad_norm": 0.538491909735447,
      "learning_rate": 0.003,
      "loss": 4.3877,
      "step": 1125
    },
    {
      "epoch": 0.01126,
      "grad_norm": 0.4826610539173529,
      "learning_rate": 0.003,
      "loss": 4.392,
      "step": 1126
    },
    {
      "epoch": 0.01127,
      "grad_norm": 0.41526400192632873,
      "learning_rate": 0.003,
      "loss": 4.421,
      "step": 1127
    },
    {
      "epoch": 0.01128,
      "grad_norm": 0.4898630750070376,
      "learning_rate": 0.003,
      "loss": 4.4001,
      "step": 1128
    },
    {
      "epoch": 0.01129,
      "grad_norm": 0.6900647874251847,
      "learning_rate": 0.003,
      "loss": 4.3988,
      "step": 1129
    },
    {
      "epoch": 0.0113,
      "grad_norm": 0.6715183668037129,
      "learning_rate": 0.003,
      "loss": 4.3803,
      "step": 1130
    },
    {
      "epoch": 0.01131,
      "grad_norm": 0.5369210919725328,
      "learning_rate": 0.003,
      "loss": 4.3895,
      "step": 1131
    },
    {
      "epoch": 0.01132,
      "grad_norm": 0.7090208352393726,
      "learning_rate": 0.003,
      "loss": 4.4216,
      "step": 1132
    },
    {
      "epoch": 0.01133,
      "grad_norm": 0.8026720884346313,
      "learning_rate": 0.003,
      "loss": 4.3951,
      "step": 1133
    },
    {
      "epoch": 0.01134,
      "grad_norm": 0.7235729334025307,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1134
    },
    {
      "epoch": 0.01135,
      "grad_norm": 0.5789287370870784,
      "learning_rate": 0.003,
      "loss": 4.4255,
      "step": 1135
    },
    {
      "epoch": 0.01136,
      "grad_norm": 0.5872111107371502,
      "learning_rate": 0.003,
      "loss": 4.4037,
      "step": 1136
    },
    {
      "epoch": 0.01137,
      "grad_norm": 0.5973670624507018,
      "learning_rate": 0.003,
      "loss": 4.4236,
      "step": 1137
    },
    {
      "epoch": 0.01138,
      "grad_norm": 0.5814805523631724,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1138
    },
    {
      "epoch": 0.01139,
      "grad_norm": 0.5749927487336652,
      "learning_rate": 0.003,
      "loss": 4.389,
      "step": 1139
    },
    {
      "epoch": 0.0114,
      "grad_norm": 0.5177155580469588,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1140
    },
    {
      "epoch": 0.01141,
      "grad_norm": 0.5098128826528755,
      "learning_rate": 0.003,
      "loss": 4.3956,
      "step": 1141
    },
    {
      "epoch": 0.01142,
      "grad_norm": 0.5805509550413942,
      "learning_rate": 0.003,
      "loss": 4.3764,
      "step": 1142
    },
    {
      "epoch": 0.01143,
      "grad_norm": 0.5868263165469564,
      "learning_rate": 0.003,
      "loss": 4.3885,
      "step": 1143
    },
    {
      "epoch": 0.01144,
      "grad_norm": 0.6024984248815738,
      "learning_rate": 0.003,
      "loss": 4.3595,
      "step": 1144
    },
    {
      "epoch": 0.01145,
      "grad_norm": 0.6011422862217553,
      "learning_rate": 0.003,
      "loss": 4.396,
      "step": 1145
    },
    {
      "epoch": 0.01146,
      "grad_norm": 0.6291389752386858,
      "learning_rate": 0.003,
      "loss": 4.3845,
      "step": 1146
    },
    {
      "epoch": 0.01147,
      "grad_norm": 0.5858383738783732,
      "learning_rate": 0.003,
      "loss": 4.3734,
      "step": 1147
    },
    {
      "epoch": 0.01148,
      "grad_norm": 0.5690631266524724,
      "learning_rate": 0.003,
      "loss": 4.3732,
      "step": 1148
    },
    {
      "epoch": 0.01149,
      "grad_norm": 0.47603449917616164,
      "learning_rate": 0.003,
      "loss": 4.3879,
      "step": 1149
    },
    {
      "epoch": 0.0115,
      "grad_norm": 0.5321591521167512,
      "learning_rate": 0.003,
      "loss": 4.3815,
      "step": 1150
    },
    {
      "epoch": 0.01151,
      "grad_norm": 0.5902114027403508,
      "learning_rate": 0.003,
      "loss": 4.3804,
      "step": 1151
    },
    {
      "epoch": 0.01152,
      "grad_norm": 0.720836087411827,
      "learning_rate": 0.003,
      "loss": 4.3994,
      "step": 1152
    },
    {
      "epoch": 0.01153,
      "grad_norm": 0.858014811127527,
      "learning_rate": 0.003,
      "loss": 4.4233,
      "step": 1153
    },
    {
      "epoch": 0.01154,
      "grad_norm": 0.8276250061884273,
      "learning_rate": 0.003,
      "loss": 4.4035,
      "step": 1154
    },
    {
      "epoch": 0.01155,
      "grad_norm": 0.9316719148571102,
      "learning_rate": 0.003,
      "loss": 4.4003,
      "step": 1155
    },
    {
      "epoch": 0.01156,
      "grad_norm": 0.9492720138035956,
      "learning_rate": 0.003,
      "loss": 4.4137,
      "step": 1156
    },
    {
      "epoch": 0.01157,
      "grad_norm": 1.1562300430530366,
      "learning_rate": 0.003,
      "loss": 4.471,
      "step": 1157
    },
    {
      "epoch": 0.01158,
      "grad_norm": 0.8082778562361257,
      "learning_rate": 0.003,
      "loss": 4.4212,
      "step": 1158
    },
    {
      "epoch": 0.01159,
      "grad_norm": 0.8296228637182822,
      "learning_rate": 0.003,
      "loss": 4.4254,
      "step": 1159
    },
    {
      "epoch": 0.0116,
      "grad_norm": 0.9729894065836546,
      "learning_rate": 0.003,
      "loss": 4.4435,
      "step": 1160
    },
    {
      "epoch": 0.01161,
      "grad_norm": 0.986208959900233,
      "learning_rate": 0.003,
      "loss": 4.4396,
      "step": 1161
    },
    {
      "epoch": 0.01162,
      "grad_norm": 0.9917874948238292,
      "learning_rate": 0.003,
      "loss": 4.4412,
      "step": 1162
    },
    {
      "epoch": 0.01163,
      "grad_norm": 1.022093528985829,
      "learning_rate": 0.003,
      "loss": 4.4772,
      "step": 1163
    },
    {
      "epoch": 0.01164,
      "grad_norm": 0.8274710934079806,
      "learning_rate": 0.003,
      "loss": 4.4588,
      "step": 1164
    },
    {
      "epoch": 0.01165,
      "grad_norm": 0.8544376883862705,
      "learning_rate": 0.003,
      "loss": 4.4518,
      "step": 1165
    },
    {
      "epoch": 0.01166,
      "grad_norm": 0.9826443233479919,
      "learning_rate": 0.003,
      "loss": 4.4596,
      "step": 1166
    },
    {
      "epoch": 0.01167,
      "grad_norm": 0.8638656323292462,
      "learning_rate": 0.003,
      "loss": 4.427,
      "step": 1167
    },
    {
      "epoch": 0.01168,
      "grad_norm": 0.7965515530800152,
      "learning_rate": 0.003,
      "loss": 4.4285,
      "step": 1168
    },
    {
      "epoch": 0.01169,
      "grad_norm": 0.621289788302785,
      "learning_rate": 0.003,
      "loss": 4.4631,
      "step": 1169
    },
    {
      "epoch": 0.0117,
      "grad_norm": 0.5251543844281131,
      "learning_rate": 0.003,
      "loss": 4.441,
      "step": 1170
    },
    {
      "epoch": 0.01171,
      "grad_norm": 0.6074391771886244,
      "learning_rate": 0.003,
      "loss": 4.423,
      "step": 1171
    },
    {
      "epoch": 0.01172,
      "grad_norm": 0.5839489329030457,
      "learning_rate": 0.003,
      "loss": 4.4084,
      "step": 1172
    },
    {
      "epoch": 0.01173,
      "grad_norm": 0.561004783856832,
      "learning_rate": 0.003,
      "loss": 4.4404,
      "step": 1173
    },
    {
      "epoch": 0.01174,
      "grad_norm": 0.5821159441803843,
      "learning_rate": 0.003,
      "loss": 4.4125,
      "step": 1174
    },
    {
      "epoch": 0.01175,
      "grad_norm": 0.5941266163389948,
      "learning_rate": 0.003,
      "loss": 4.3949,
      "step": 1175
    },
    {
      "epoch": 0.01176,
      "grad_norm": 0.6502157975630524,
      "learning_rate": 0.003,
      "loss": 4.432,
      "step": 1176
    },
    {
      "epoch": 0.01177,
      "grad_norm": 0.8335241720178251,
      "learning_rate": 0.003,
      "loss": 4.4041,
      "step": 1177
    },
    {
      "epoch": 0.01178,
      "grad_norm": 1.0000155108324966,
      "learning_rate": 0.003,
      "loss": 4.4462,
      "step": 1178
    },
    {
      "epoch": 0.01179,
      "grad_norm": 0.9223184266441969,
      "learning_rate": 0.003,
      "loss": 4.4309,
      "step": 1179
    },
    {
      "epoch": 0.0118,
      "grad_norm": 0.6248971968290589,
      "learning_rate": 0.003,
      "loss": 4.4722,
      "step": 1180
    },
    {
      "epoch": 0.01181,
      "grad_norm": 0.6327502868796897,
      "learning_rate": 0.003,
      "loss": 4.4606,
      "step": 1181
    },
    {
      "epoch": 0.01182,
      "grad_norm": 0.6032057741631047,
      "learning_rate": 0.003,
      "loss": 4.4179,
      "step": 1182
    },
    {
      "epoch": 0.01183,
      "grad_norm": 0.6177419756766389,
      "learning_rate": 0.003,
      "loss": 4.4068,
      "step": 1183
    },
    {
      "epoch": 0.01184,
      "grad_norm": 0.4603977151305031,
      "learning_rate": 0.003,
      "loss": 4.4388,
      "step": 1184
    },
    {
      "epoch": 0.01185,
      "grad_norm": 0.39775931274104387,
      "learning_rate": 0.003,
      "loss": 4.4039,
      "step": 1185
    },
    {
      "epoch": 0.01186,
      "grad_norm": 0.43457323577233337,
      "learning_rate": 0.003,
      "loss": 4.4263,
      "step": 1186
    },
    {
      "epoch": 0.01187,
      "grad_norm": 0.3445355147627072,
      "learning_rate": 0.003,
      "loss": 4.3841,
      "step": 1187
    },
    {
      "epoch": 0.01188,
      "grad_norm": 0.31570813535234143,
      "learning_rate": 0.003,
      "loss": 4.3952,
      "step": 1188
    },
    {
      "epoch": 0.01189,
      "grad_norm": 0.3234807855930626,
      "learning_rate": 0.003,
      "loss": 4.3874,
      "step": 1189
    },
    {
      "epoch": 0.0119,
      "grad_norm": 0.3007248052516205,
      "learning_rate": 0.003,
      "loss": 4.3694,
      "step": 1190
    },
    {
      "epoch": 0.01191,
      "grad_norm": 0.289692411083719,
      "learning_rate": 0.003,
      "loss": 4.3476,
      "step": 1191
    },
    {
      "epoch": 0.01192,
      "grad_norm": 0.2772752758842908,
      "learning_rate": 0.003,
      "loss": 4.3779,
      "step": 1192
    },
    {
      "epoch": 0.01193,
      "grad_norm": 0.2637988935747097,
      "learning_rate": 0.003,
      "loss": 4.3787,
      "step": 1193
    },
    {
      "epoch": 0.01194,
      "grad_norm": 0.2480894935347364,
      "learning_rate": 0.003,
      "loss": 4.3791,
      "step": 1194
    },
    {
      "epoch": 0.01195,
      "grad_norm": 0.2744828686479978,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1195
    },
    {
      "epoch": 0.01196,
      "grad_norm": 0.2694264980010376,
      "learning_rate": 0.003,
      "loss": 4.3608,
      "step": 1196
    },
    {
      "epoch": 0.01197,
      "grad_norm": 0.2911995649935099,
      "learning_rate": 0.003,
      "loss": 4.3243,
      "step": 1197
    },
    {
      "epoch": 0.01198,
      "grad_norm": 0.3049686361769555,
      "learning_rate": 0.003,
      "loss": 4.3978,
      "step": 1198
    },
    {
      "epoch": 0.01199,
      "grad_norm": 0.4091366608598886,
      "learning_rate": 0.003,
      "loss": 4.3744,
      "step": 1199
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.5260409537460498,
      "learning_rate": 0.003,
      "loss": 4.3673,
      "step": 1200
    },
    {
      "epoch": 0.01201,
      "grad_norm": 0.6358041499118771,
      "learning_rate": 0.003,
      "loss": 4.4018,
      "step": 1201
    },
    {
      "epoch": 0.01202,
      "grad_norm": 0.5923034972204151,
      "learning_rate": 0.003,
      "loss": 4.3715,
      "step": 1202
    },
    {
      "epoch": 0.01203,
      "grad_norm": 0.49351792103508907,
      "learning_rate": 0.003,
      "loss": 4.3737,
      "step": 1203
    },
    {
      "epoch": 0.01204,
      "grad_norm": 0.6405227463764683,
      "learning_rate": 0.003,
      "loss": 4.3772,
      "step": 1204
    },
    {
      "epoch": 0.01205,
      "grad_norm": 0.7035300072883678,
      "learning_rate": 0.003,
      "loss": 4.3759,
      "step": 1205
    },
    {
      "epoch": 0.01206,
      "grad_norm": 0.6623021094722211,
      "learning_rate": 0.003,
      "loss": 4.4176,
      "step": 1206
    },
    {
      "epoch": 0.01207,
      "grad_norm": 0.6906429556810797,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1207
    },
    {
      "epoch": 0.01208,
      "grad_norm": 0.7416830674542866,
      "learning_rate": 0.003,
      "loss": 4.4101,
      "step": 1208
    },
    {
      "epoch": 0.01209,
      "grad_norm": 0.7431220077552587,
      "learning_rate": 0.003,
      "loss": 4.3922,
      "step": 1209
    },
    {
      "epoch": 0.0121,
      "grad_norm": 0.7626206113161277,
      "learning_rate": 0.003,
      "loss": 4.397,
      "step": 1210
    },
    {
      "epoch": 0.01211,
      "grad_norm": 0.7273957054715338,
      "learning_rate": 0.003,
      "loss": 4.381,
      "step": 1211
    },
    {
      "epoch": 0.01212,
      "grad_norm": 0.5840029568104971,
      "learning_rate": 0.003,
      "loss": 4.4002,
      "step": 1212
    },
    {
      "epoch": 0.01213,
      "grad_norm": 0.5534819591366041,
      "learning_rate": 0.003,
      "loss": 4.3747,
      "step": 1213
    },
    {
      "epoch": 0.01214,
      "grad_norm": 0.5479838122682379,
      "learning_rate": 0.003,
      "loss": 4.3596,
      "step": 1214
    },
    {
      "epoch": 0.01215,
      "grad_norm": 0.6541197102272803,
      "learning_rate": 0.003,
      "loss": 4.3736,
      "step": 1215
    },
    {
      "epoch": 0.01216,
      "grad_norm": 0.6998905075239406,
      "learning_rate": 0.003,
      "loss": 4.369,
      "step": 1216
    },
    {
      "epoch": 0.01217,
      "grad_norm": 0.6272982074936522,
      "learning_rate": 0.003,
      "loss": 4.3725,
      "step": 1217
    },
    {
      "epoch": 0.01218,
      "grad_norm": 0.6037550089398132,
      "learning_rate": 0.003,
      "loss": 4.3687,
      "step": 1218
    },
    {
      "epoch": 0.01219,
      "grad_norm": 0.5872809556780119,
      "learning_rate": 0.003,
      "loss": 4.3682,
      "step": 1219
    },
    {
      "epoch": 0.0122,
      "grad_norm": 0.45035452365127926,
      "learning_rate": 0.003,
      "loss": 4.3537,
      "step": 1220
    },
    {
      "epoch": 0.01221,
      "grad_norm": 0.41160354363285073,
      "learning_rate": 0.003,
      "loss": 4.3666,
      "step": 1221
    },
    {
      "epoch": 0.01222,
      "grad_norm": 0.5075830142490442,
      "learning_rate": 0.003,
      "loss": 4.3511,
      "step": 1222
    },
    {
      "epoch": 0.01223,
      "grad_norm": 0.5662251248767171,
      "learning_rate": 0.003,
      "loss": 4.3541,
      "step": 1223
    },
    {
      "epoch": 0.01224,
      "grad_norm": 0.690794501870828,
      "learning_rate": 0.003,
      "loss": 4.3444,
      "step": 1224
    },
    {
      "epoch": 0.01225,
      "grad_norm": 0.675909010637242,
      "learning_rate": 0.003,
      "loss": 4.3698,
      "step": 1225
    },
    {
      "epoch": 0.01226,
      "grad_norm": 0.6653173755559965,
      "learning_rate": 0.003,
      "loss": 4.36,
      "step": 1226
    },
    {
      "epoch": 0.01227,
      "grad_norm": 0.8124014201072856,
      "learning_rate": 0.003,
      "loss": 4.4022,
      "step": 1227
    },
    {
      "epoch": 0.01228,
      "grad_norm": 0.8601590167217709,
      "learning_rate": 0.003,
      "loss": 4.3869,
      "step": 1228
    },
    {
      "epoch": 0.01229,
      "grad_norm": 0.7816229118417856,
      "learning_rate": 0.003,
      "loss": 4.3791,
      "step": 1229
    },
    {
      "epoch": 0.0123,
      "grad_norm": 0.7167415795127342,
      "learning_rate": 0.003,
      "loss": 4.4104,
      "step": 1230
    },
    {
      "epoch": 0.01231,
      "grad_norm": 0.7227237760755019,
      "learning_rate": 0.003,
      "loss": 4.3789,
      "step": 1231
    },
    {
      "epoch": 0.01232,
      "grad_norm": 0.6338303976227229,
      "learning_rate": 0.003,
      "loss": 4.3516,
      "step": 1232
    },
    {
      "epoch": 0.01233,
      "grad_norm": 0.5121760575897172,
      "learning_rate": 0.003,
      "loss": 4.3548,
      "step": 1233
    },
    {
      "epoch": 0.01234,
      "grad_norm": 0.562103643632394,
      "learning_rate": 0.003,
      "loss": 4.3735,
      "step": 1234
    },
    {
      "epoch": 0.01235,
      "grad_norm": 0.5094370853466236,
      "learning_rate": 0.003,
      "loss": 4.3603,
      "step": 1235
    },
    {
      "epoch": 0.01236,
      "grad_norm": 0.42756368516441545,
      "learning_rate": 0.003,
      "loss": 4.3622,
      "step": 1236
    },
    {
      "epoch": 0.01237,
      "grad_norm": 0.41715621646232526,
      "learning_rate": 0.003,
      "loss": 4.3707,
      "step": 1237
    },
    {
      "epoch": 0.01238,
      "grad_norm": 0.41900290536826335,
      "learning_rate": 0.003,
      "loss": 4.3664,
      "step": 1238
    },
    {
      "epoch": 0.01239,
      "grad_norm": 0.4057196989115332,
      "learning_rate": 0.003,
      "loss": 4.3485,
      "step": 1239
    },
    {
      "epoch": 0.0124,
      "grad_norm": 0.41828480066605506,
      "learning_rate": 0.003,
      "loss": 4.331,
      "step": 1240
    },
    {
      "epoch": 0.01241,
      "grad_norm": 0.4114484855240935,
      "learning_rate": 0.003,
      "loss": 4.3417,
      "step": 1241
    },
    {
      "epoch": 0.01242,
      "grad_norm": 0.4335820493661237,
      "learning_rate": 0.003,
      "loss": 4.348,
      "step": 1242
    },
    {
      "epoch": 0.01243,
      "grad_norm": 0.5022671727550078,
      "learning_rate": 0.003,
      "loss": 4.3316,
      "step": 1243
    },
    {
      "epoch": 0.01244,
      "grad_norm": 0.5275944783156026,
      "learning_rate": 0.003,
      "loss": 4.3445,
      "step": 1244
    },
    {
      "epoch": 0.01245,
      "grad_norm": 0.5485058077885037,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1245
    },
    {
      "epoch": 0.01246,
      "grad_norm": 0.5308550603908434,
      "learning_rate": 0.003,
      "loss": 4.312,
      "step": 1246
    },
    {
      "epoch": 0.01247,
      "grad_norm": 0.5603455205230277,
      "learning_rate": 0.003,
      "loss": 4.373,
      "step": 1247
    },
    {
      "epoch": 0.01248,
      "grad_norm": 0.6616614283266387,
      "learning_rate": 0.003,
      "loss": 4.3622,
      "step": 1248
    },
    {
      "epoch": 0.01249,
      "grad_norm": 0.6950989703790047,
      "learning_rate": 0.003,
      "loss": 4.3579,
      "step": 1249
    },
    {
      "epoch": 0.0125,
      "grad_norm": 0.6807657788060711,
      "learning_rate": 0.003,
      "loss": 4.3266,
      "step": 1250
    },
    {
      "epoch": 0.01251,
      "grad_norm": 0.7044512298181651,
      "learning_rate": 0.003,
      "loss": 4.3715,
      "step": 1251
    },
    {
      "epoch": 0.01252,
      "grad_norm": 0.7187283948109732,
      "learning_rate": 0.003,
      "loss": 4.3491,
      "step": 1252
    },
    {
      "epoch": 0.01253,
      "grad_norm": 0.6168851566630172,
      "learning_rate": 0.003,
      "loss": 4.3238,
      "step": 1253
    },
    {
      "epoch": 0.01254,
      "grad_norm": 0.5482483685536774,
      "learning_rate": 0.003,
      "loss": 4.3671,
      "step": 1254
    },
    {
      "epoch": 0.01255,
      "grad_norm": 0.5158716427473324,
      "learning_rate": 0.003,
      "loss": 4.3154,
      "step": 1255
    },
    {
      "epoch": 0.01256,
      "grad_norm": 0.46371154610018284,
      "learning_rate": 0.003,
      "loss": 4.3421,
      "step": 1256
    },
    {
      "epoch": 0.01257,
      "grad_norm": 0.4261160552146568,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1257
    },
    {
      "epoch": 0.01258,
      "grad_norm": 0.507565461332445,
      "learning_rate": 0.003,
      "loss": 4.3375,
      "step": 1258
    },
    {
      "epoch": 0.01259,
      "grad_norm": 0.7023530299208391,
      "learning_rate": 0.003,
      "loss": 4.3461,
      "step": 1259
    },
    {
      "epoch": 0.0126,
      "grad_norm": 0.8862638702569765,
      "learning_rate": 0.003,
      "loss": 4.3945,
      "step": 1260
    },
    {
      "epoch": 0.01261,
      "grad_norm": 0.92528404873397,
      "learning_rate": 0.003,
      "loss": 4.3876,
      "step": 1261
    },
    {
      "epoch": 0.01262,
      "grad_norm": 0.9810348621626278,
      "learning_rate": 0.003,
      "loss": 4.3924,
      "step": 1262
    },
    {
      "epoch": 0.01263,
      "grad_norm": 0.7811680454044853,
      "learning_rate": 0.003,
      "loss": 4.3835,
      "step": 1263
    },
    {
      "epoch": 0.01264,
      "grad_norm": 0.6458262464633842,
      "learning_rate": 0.003,
      "loss": 4.365,
      "step": 1264
    },
    {
      "epoch": 0.01265,
      "grad_norm": 0.6726321620447105,
      "learning_rate": 0.003,
      "loss": 4.3777,
      "step": 1265
    },
    {
      "epoch": 0.01266,
      "grad_norm": 0.7462208985516781,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1266
    },
    {
      "epoch": 0.01267,
      "grad_norm": 0.8334608532491081,
      "learning_rate": 0.003,
      "loss": 4.4053,
      "step": 1267
    },
    {
      "epoch": 0.01268,
      "grad_norm": 0.722253192550455,
      "learning_rate": 0.003,
      "loss": 4.3914,
      "step": 1268
    },
    {
      "epoch": 0.01269,
      "grad_norm": 0.5434607483157148,
      "learning_rate": 0.003,
      "loss": 4.3468,
      "step": 1269
    },
    {
      "epoch": 0.0127,
      "grad_norm": 0.4389640950286333,
      "learning_rate": 0.003,
      "loss": 4.3757,
      "step": 1270
    },
    {
      "epoch": 0.01271,
      "grad_norm": 0.4493097096002282,
      "learning_rate": 0.003,
      "loss": 4.357,
      "step": 1271
    },
    {
      "epoch": 0.01272,
      "grad_norm": 0.4602878891635523,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1272
    },
    {
      "epoch": 0.01273,
      "grad_norm": 0.4816441128313513,
      "learning_rate": 0.003,
      "loss": 4.3755,
      "step": 1273
    },
    {
      "epoch": 0.01274,
      "grad_norm": 0.5275730605295648,
      "learning_rate": 0.003,
      "loss": 4.3705,
      "step": 1274
    },
    {
      "epoch": 0.01275,
      "grad_norm": 0.5285253035410874,
      "learning_rate": 0.003,
      "loss": 4.3384,
      "step": 1275
    },
    {
      "epoch": 0.01276,
      "grad_norm": 0.42608260843601453,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1276
    },
    {
      "epoch": 0.01277,
      "grad_norm": 0.394113210363875,
      "learning_rate": 0.003,
      "loss": 4.3444,
      "step": 1277
    },
    {
      "epoch": 0.01278,
      "grad_norm": 0.485427057016395,
      "learning_rate": 0.003,
      "loss": 4.3534,
      "step": 1278
    },
    {
      "epoch": 0.01279,
      "grad_norm": 0.4912898619415999,
      "learning_rate": 0.003,
      "loss": 4.3362,
      "step": 1279
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.4970947767829747,
      "learning_rate": 0.003,
      "loss": 4.3331,
      "step": 1280
    },
    {
      "epoch": 0.01281,
      "grad_norm": 0.45661885236926897,
      "learning_rate": 0.003,
      "loss": 4.3167,
      "step": 1281
    },
    {
      "epoch": 0.01282,
      "grad_norm": 0.3777414491844073,
      "learning_rate": 0.003,
      "loss": 4.3562,
      "step": 1282
    },
    {
      "epoch": 0.01283,
      "grad_norm": 0.4194385613191725,
      "learning_rate": 0.003,
      "loss": 4.3264,
      "step": 1283
    },
    {
      "epoch": 0.01284,
      "grad_norm": 0.455784660804564,
      "learning_rate": 0.003,
      "loss": 4.328,
      "step": 1284
    },
    {
      "epoch": 0.01285,
      "grad_norm": 0.5164075234519105,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1285
    },
    {
      "epoch": 0.01286,
      "grad_norm": 0.6148100521816782,
      "learning_rate": 0.003,
      "loss": 4.3568,
      "step": 1286
    },
    {
      "epoch": 0.01287,
      "grad_norm": 0.8435250802638579,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1287
    },
    {
      "epoch": 0.01288,
      "grad_norm": 0.8948772567277533,
      "learning_rate": 0.003,
      "loss": 4.3616,
      "step": 1288
    },
    {
      "epoch": 0.01289,
      "grad_norm": 0.5768385697936498,
      "learning_rate": 0.003,
      "loss": 4.3412,
      "step": 1289
    },
    {
      "epoch": 0.0129,
      "grad_norm": 0.48744560346189664,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1290
    },
    {
      "epoch": 0.01291,
      "grad_norm": 0.5888239794628021,
      "learning_rate": 0.003,
      "loss": 4.3458,
      "step": 1291
    },
    {
      "epoch": 0.01292,
      "grad_norm": 0.5667412086864034,
      "learning_rate": 0.003,
      "loss": 4.3773,
      "step": 1292
    },
    {
      "epoch": 0.01293,
      "grad_norm": 0.6066487822204776,
      "learning_rate": 0.003,
      "loss": 4.3648,
      "step": 1293
    },
    {
      "epoch": 0.01294,
      "grad_norm": 0.7065627866796411,
      "learning_rate": 0.003,
      "loss": 4.3371,
      "step": 1294
    },
    {
      "epoch": 0.01295,
      "grad_norm": 0.7389325253179186,
      "learning_rate": 0.003,
      "loss": 4.3679,
      "step": 1295
    },
    {
      "epoch": 0.01296,
      "grad_norm": 0.7660662423328507,
      "learning_rate": 0.003,
      "loss": 4.3399,
      "step": 1296
    },
    {
      "epoch": 0.01297,
      "grad_norm": 0.7847737463247922,
      "learning_rate": 0.003,
      "loss": 4.3573,
      "step": 1297
    },
    {
      "epoch": 0.01298,
      "grad_norm": 0.8150967050528192,
      "learning_rate": 0.003,
      "loss": 4.3384,
      "step": 1298
    },
    {
      "epoch": 0.01299,
      "grad_norm": 0.7237792681987708,
      "learning_rate": 0.003,
      "loss": 4.3533,
      "step": 1299
    },
    {
      "epoch": 0.013,
      "grad_norm": 0.6452098996944176,
      "learning_rate": 0.003,
      "loss": 4.3355,
      "step": 1300
    },
    {
      "epoch": 0.01301,
      "grad_norm": 0.6428524108659182,
      "learning_rate": 0.003,
      "loss": 4.3495,
      "step": 1301
    },
    {
      "epoch": 0.01302,
      "grad_norm": 0.7475054826389519,
      "learning_rate": 0.003,
      "loss": 4.4013,
      "step": 1302
    },
    {
      "epoch": 0.01303,
      "grad_norm": 0.9085755519721904,
      "learning_rate": 0.003,
      "loss": 4.3576,
      "step": 1303
    },
    {
      "epoch": 0.01304,
      "grad_norm": 0.8565196075884696,
      "learning_rate": 0.003,
      "loss": 4.3934,
      "step": 1304
    },
    {
      "epoch": 0.01305,
      "grad_norm": 0.8916073679840805,
      "learning_rate": 0.003,
      "loss": 4.3789,
      "step": 1305
    },
    {
      "epoch": 0.01306,
      "grad_norm": 0.7910970693035262,
      "learning_rate": 0.003,
      "loss": 4.3681,
      "step": 1306
    },
    {
      "epoch": 0.01307,
      "grad_norm": 0.7398667792205602,
      "learning_rate": 0.003,
      "loss": 4.3875,
      "step": 1307
    },
    {
      "epoch": 0.01308,
      "grad_norm": 0.6800690310649735,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1308
    },
    {
      "epoch": 0.01309,
      "grad_norm": 0.610129543183007,
      "learning_rate": 0.003,
      "loss": 4.3978,
      "step": 1309
    },
    {
      "epoch": 0.0131,
      "grad_norm": 0.6342445693889617,
      "learning_rate": 0.003,
      "loss": 4.3435,
      "step": 1310
    },
    {
      "epoch": 0.01311,
      "grad_norm": 0.5612187898652411,
      "learning_rate": 0.003,
      "loss": 4.3787,
      "step": 1311
    },
    {
      "epoch": 0.01312,
      "grad_norm": 0.6173335576080043,
      "learning_rate": 0.003,
      "loss": 4.3472,
      "step": 1312
    },
    {
      "epoch": 0.01313,
      "grad_norm": 0.5958431337675327,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1313
    },
    {
      "epoch": 0.01314,
      "grad_norm": 0.6624135563680444,
      "learning_rate": 0.003,
      "loss": 4.3367,
      "step": 1314
    },
    {
      "epoch": 0.01315,
      "grad_norm": 0.6490907209314275,
      "learning_rate": 0.003,
      "loss": 4.368,
      "step": 1315
    },
    {
      "epoch": 0.01316,
      "grad_norm": 0.5500794253188088,
      "learning_rate": 0.003,
      "loss": 4.3457,
      "step": 1316
    },
    {
      "epoch": 0.01317,
      "grad_norm": 0.6275958010748339,
      "learning_rate": 0.003,
      "loss": 4.3746,
      "step": 1317
    },
    {
      "epoch": 0.01318,
      "grad_norm": 0.5648473326010665,
      "learning_rate": 0.003,
      "loss": 4.3513,
      "step": 1318
    },
    {
      "epoch": 0.01319,
      "grad_norm": 0.45510587840319355,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1319
    },
    {
      "epoch": 0.0132,
      "grad_norm": 0.44264387716031117,
      "learning_rate": 0.003,
      "loss": 4.3472,
      "step": 1320
    },
    {
      "epoch": 0.01321,
      "grad_norm": 0.40376640077570203,
      "learning_rate": 0.003,
      "loss": 4.322,
      "step": 1321
    },
    {
      "epoch": 0.01322,
      "grad_norm": 0.3877836192326668,
      "learning_rate": 0.003,
      "loss": 4.3502,
      "step": 1322
    },
    {
      "epoch": 0.01323,
      "grad_norm": 0.3937183906523392,
      "learning_rate": 0.003,
      "loss": 4.341,
      "step": 1323
    },
    {
      "epoch": 0.01324,
      "grad_norm": 0.4100022537274817,
      "learning_rate": 0.003,
      "loss": 4.3255,
      "step": 1324
    },
    {
      "epoch": 0.01325,
      "grad_norm": 0.46279809510243225,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1325
    },
    {
      "epoch": 0.01326,
      "grad_norm": 0.4767280607611473,
      "learning_rate": 0.003,
      "loss": 4.3173,
      "step": 1326
    },
    {
      "epoch": 0.01327,
      "grad_norm": 0.5123257836661763,
      "learning_rate": 0.003,
      "loss": 4.3366,
      "step": 1327
    },
    {
      "epoch": 0.01328,
      "grad_norm": 0.5434476080816805,
      "learning_rate": 0.003,
      "loss": 4.3286,
      "step": 1328
    },
    {
      "epoch": 0.01329,
      "grad_norm": 0.48979784037018337,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1329
    },
    {
      "epoch": 0.0133,
      "grad_norm": 0.42591921766495494,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1330
    },
    {
      "epoch": 0.01331,
      "grad_norm": 0.48229010910172454,
      "learning_rate": 0.003,
      "loss": 4.318,
      "step": 1331
    },
    {
      "epoch": 0.01332,
      "grad_norm": 0.514388044498456,
      "learning_rate": 0.003,
      "loss": 4.3386,
      "step": 1332
    },
    {
      "epoch": 0.01333,
      "grad_norm": 0.5443479600217087,
      "learning_rate": 0.003,
      "loss": 4.3145,
      "step": 1333
    },
    {
      "epoch": 0.01334,
      "grad_norm": 0.5537269490906184,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1334
    },
    {
      "epoch": 0.01335,
      "grad_norm": 0.5158317522709531,
      "learning_rate": 0.003,
      "loss": 4.3253,
      "step": 1335
    },
    {
      "epoch": 0.01336,
      "grad_norm": 0.5108167456714484,
      "learning_rate": 0.003,
      "loss": 4.3223,
      "step": 1336
    },
    {
      "epoch": 0.01337,
      "grad_norm": 0.6228714832065689,
      "learning_rate": 0.003,
      "loss": 4.3418,
      "step": 1337
    },
    {
      "epoch": 0.01338,
      "grad_norm": 0.7769527340926119,
      "learning_rate": 0.003,
      "loss": 4.3169,
      "step": 1338
    },
    {
      "epoch": 0.01339,
      "grad_norm": 1.0163715747866278,
      "learning_rate": 0.003,
      "loss": 4.3527,
      "step": 1339
    },
    {
      "epoch": 0.0134,
      "grad_norm": 0.8875484689453775,
      "learning_rate": 0.003,
      "loss": 4.3448,
      "step": 1340
    },
    {
      "epoch": 0.01341,
      "grad_norm": 0.7593448848803411,
      "learning_rate": 0.003,
      "loss": 4.3557,
      "step": 1341
    },
    {
      "epoch": 0.01342,
      "grad_norm": 0.7038826124517171,
      "learning_rate": 0.003,
      "loss": 4.3272,
      "step": 1342
    },
    {
      "epoch": 0.01343,
      "grad_norm": 0.6570836329978591,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1343
    },
    {
      "epoch": 0.01344,
      "grad_norm": 0.7840667137315036,
      "learning_rate": 0.003,
      "loss": 4.3338,
      "step": 1344
    },
    {
      "epoch": 0.01345,
      "grad_norm": 0.8057220270427362,
      "learning_rate": 0.003,
      "loss": 4.3808,
      "step": 1345
    },
    {
      "epoch": 0.01346,
      "grad_norm": 0.7768442613998496,
      "learning_rate": 0.003,
      "loss": 4.3555,
      "step": 1346
    },
    {
      "epoch": 0.01347,
      "grad_norm": 0.7491448291061306,
      "learning_rate": 0.003,
      "loss": 4.3463,
      "step": 1347
    },
    {
      "epoch": 0.01348,
      "grad_norm": 0.7233252261262297,
      "learning_rate": 0.003,
      "loss": 4.3476,
      "step": 1348
    },
    {
      "epoch": 0.01349,
      "grad_norm": 0.7147253457500066,
      "learning_rate": 0.003,
      "loss": 4.3662,
      "step": 1349
    },
    {
      "epoch": 0.0135,
      "grad_norm": 0.8413435804466313,
      "learning_rate": 0.003,
      "loss": 4.3673,
      "step": 1350
    },
    {
      "epoch": 0.01351,
      "grad_norm": 0.9169233695908002,
      "learning_rate": 0.003,
      "loss": 4.406,
      "step": 1351
    },
    {
      "epoch": 0.01352,
      "grad_norm": 0.7136998484165762,
      "learning_rate": 0.003,
      "loss": 4.371,
      "step": 1352
    },
    {
      "epoch": 0.01353,
      "grad_norm": 0.7015996798304995,
      "learning_rate": 0.003,
      "loss": 4.3575,
      "step": 1353
    },
    {
      "epoch": 0.01354,
      "grad_norm": 0.7349461226651943,
      "learning_rate": 0.003,
      "loss": 4.3427,
      "step": 1354
    },
    {
      "epoch": 0.01355,
      "grad_norm": 0.7908101359678147,
      "learning_rate": 0.003,
      "loss": 4.3615,
      "step": 1355
    },
    {
      "epoch": 0.01356,
      "grad_norm": 0.6826667144170498,
      "learning_rate": 0.003,
      "loss": 4.3532,
      "step": 1356
    },
    {
      "epoch": 0.01357,
      "grad_norm": 0.5539967821197357,
      "learning_rate": 0.003,
      "loss": 4.3308,
      "step": 1357
    },
    {
      "epoch": 0.01358,
      "grad_norm": 0.6087032954279707,
      "learning_rate": 0.003,
      "loss": 4.3295,
      "step": 1358
    },
    {
      "epoch": 0.01359,
      "grad_norm": 0.7945356614677406,
      "learning_rate": 0.003,
      "loss": 4.3753,
      "step": 1359
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.7371837851861581,
      "learning_rate": 0.003,
      "loss": 4.3479,
      "step": 1360
    },
    {
      "epoch": 0.01361,
      "grad_norm": 0.5619004170613455,
      "learning_rate": 0.003,
      "loss": 4.3527,
      "step": 1361
    },
    {
      "epoch": 0.01362,
      "grad_norm": 0.5285167943277829,
      "learning_rate": 0.003,
      "loss": 4.3621,
      "step": 1362
    },
    {
      "epoch": 0.01363,
      "grad_norm": 0.5076817025344985,
      "learning_rate": 0.003,
      "loss": 4.3625,
      "step": 1363
    },
    {
      "epoch": 0.01364,
      "grad_norm": 0.45959981115038123,
      "learning_rate": 0.003,
      "loss": 4.3487,
      "step": 1364
    },
    {
      "epoch": 0.01365,
      "grad_norm": 0.4790054557157507,
      "learning_rate": 0.003,
      "loss": 4.3663,
      "step": 1365
    },
    {
      "epoch": 0.01366,
      "grad_norm": 0.47982525149920063,
      "learning_rate": 0.003,
      "loss": 4.3186,
      "step": 1366
    },
    {
      "epoch": 0.01367,
      "grad_norm": 0.4158841510813793,
      "learning_rate": 0.003,
      "loss": 4.3634,
      "step": 1367
    },
    {
      "epoch": 0.01368,
      "grad_norm": 0.40516006301845847,
      "learning_rate": 0.003,
      "loss": 4.3337,
      "step": 1368
    },
    {
      "epoch": 0.01369,
      "grad_norm": 0.4130828938355455,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1369
    },
    {
      "epoch": 0.0137,
      "grad_norm": 0.4124457908402306,
      "learning_rate": 0.003,
      "loss": 4.3486,
      "step": 1370
    },
    {
      "epoch": 0.01371,
      "grad_norm": 0.3816075969714427,
      "learning_rate": 0.003,
      "loss": 4.3376,
      "step": 1371
    },
    {
      "epoch": 0.01372,
      "grad_norm": 0.360968745189086,
      "learning_rate": 0.003,
      "loss": 4.2843,
      "step": 1372
    },
    {
      "epoch": 0.01373,
      "grad_norm": 0.3546042076579899,
      "learning_rate": 0.003,
      "loss": 4.3232,
      "step": 1373
    },
    {
      "epoch": 0.01374,
      "grad_norm": 0.35618421664945743,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1374
    },
    {
      "epoch": 0.01375,
      "grad_norm": 0.3353008720630168,
      "learning_rate": 0.003,
      "loss": 4.2881,
      "step": 1375
    },
    {
      "epoch": 0.01376,
      "grad_norm": 0.3199771170841034,
      "learning_rate": 0.003,
      "loss": 4.3127,
      "step": 1376
    },
    {
      "epoch": 0.01377,
      "grad_norm": 0.3779659083446569,
      "learning_rate": 0.003,
      "loss": 4.3119,
      "step": 1377
    },
    {
      "epoch": 0.01378,
      "grad_norm": 0.5745139158210405,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1378
    },
    {
      "epoch": 0.01379,
      "grad_norm": 0.8747330770544977,
      "learning_rate": 0.003,
      "loss": 4.3327,
      "step": 1379
    },
    {
      "epoch": 0.0138,
      "grad_norm": 1.1013712635151454,
      "learning_rate": 0.003,
      "loss": 4.3554,
      "step": 1380
    },
    {
      "epoch": 0.01381,
      "grad_norm": 0.6702950145107445,
      "learning_rate": 0.003,
      "loss": 4.3457,
      "step": 1381
    },
    {
      "epoch": 0.01382,
      "grad_norm": 0.674972596014113,
      "learning_rate": 0.003,
      "loss": 4.3437,
      "step": 1382
    },
    {
      "epoch": 0.01383,
      "grad_norm": 0.6997537624642522,
      "learning_rate": 0.003,
      "loss": 4.3369,
      "step": 1383
    },
    {
      "epoch": 0.01384,
      "grad_norm": 0.6469176723359975,
      "learning_rate": 0.003,
      "loss": 4.3638,
      "step": 1384
    },
    {
      "epoch": 0.01385,
      "grad_norm": 0.5530980119483301,
      "learning_rate": 0.003,
      "loss": 4.3456,
      "step": 1385
    },
    {
      "epoch": 0.01386,
      "grad_norm": 0.6135404549370512,
      "learning_rate": 0.003,
      "loss": 4.3261,
      "step": 1386
    },
    {
      "epoch": 0.01387,
      "grad_norm": 0.552332815368935,
      "learning_rate": 0.003,
      "loss": 4.3116,
      "step": 1387
    },
    {
      "epoch": 0.01388,
      "grad_norm": 0.49081872927843356,
      "learning_rate": 0.003,
      "loss": 4.3469,
      "step": 1388
    },
    {
      "epoch": 0.01389,
      "grad_norm": 0.495957727461612,
      "learning_rate": 0.003,
      "loss": 4.3269,
      "step": 1389
    },
    {
      "epoch": 0.0139,
      "grad_norm": 0.4896458918623629,
      "learning_rate": 0.003,
      "loss": 4.3267,
      "step": 1390
    },
    {
      "epoch": 0.01391,
      "grad_norm": 0.49070092673030247,
      "learning_rate": 0.003,
      "loss": 4.3285,
      "step": 1391
    },
    {
      "epoch": 0.01392,
      "grad_norm": 0.475140973661053,
      "learning_rate": 0.003,
      "loss": 4.3381,
      "step": 1392
    },
    {
      "epoch": 0.01393,
      "grad_norm": 0.45720050629487835,
      "learning_rate": 0.003,
      "loss": 4.3042,
      "step": 1393
    },
    {
      "epoch": 0.01394,
      "grad_norm": 0.5119120302085031,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1394
    },
    {
      "epoch": 0.01395,
      "grad_norm": 0.6190916799083034,
      "learning_rate": 0.003,
      "loss": 4.3279,
      "step": 1395
    },
    {
      "epoch": 0.01396,
      "grad_norm": 0.6508817551577758,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1396
    },
    {
      "epoch": 0.01397,
      "grad_norm": 0.6075070397922983,
      "learning_rate": 0.003,
      "loss": 4.3359,
      "step": 1397
    },
    {
      "epoch": 0.01398,
      "grad_norm": 0.6196238264215417,
      "learning_rate": 0.003,
      "loss": 4.3165,
      "step": 1398
    },
    {
      "epoch": 0.01399,
      "grad_norm": 0.6249218455396569,
      "learning_rate": 0.003,
      "loss": 4.3396,
      "step": 1399
    },
    {
      "epoch": 0.014,
      "grad_norm": 0.7148437815408427,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1400
    },
    {
      "epoch": 0.01401,
      "grad_norm": 0.6723818092453638,
      "learning_rate": 0.003,
      "loss": 4.3228,
      "step": 1401
    },
    {
      "epoch": 0.01402,
      "grad_norm": 0.7455518751558462,
      "learning_rate": 0.003,
      "loss": 4.3392,
      "step": 1402
    },
    {
      "epoch": 0.01403,
      "grad_norm": 0.6611806665670097,
      "learning_rate": 0.003,
      "loss": 4.3114,
      "step": 1403
    },
    {
      "epoch": 0.01404,
      "grad_norm": 0.6150645931793568,
      "learning_rate": 0.003,
      "loss": 4.3339,
      "step": 1404
    },
    {
      "epoch": 0.01405,
      "grad_norm": 0.591029520159785,
      "learning_rate": 0.003,
      "loss": 4.3493,
      "step": 1405
    },
    {
      "epoch": 0.01406,
      "grad_norm": 0.6587405723247819,
      "learning_rate": 0.003,
      "loss": 4.3328,
      "step": 1406
    },
    {
      "epoch": 0.01407,
      "grad_norm": 0.7112635318635802,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1407
    },
    {
      "epoch": 0.01408,
      "grad_norm": 0.7760050777299251,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1408
    },
    {
      "epoch": 0.01409,
      "grad_norm": 0.7313606609426694,
      "learning_rate": 0.003,
      "loss": 4.3489,
      "step": 1409
    },
    {
      "epoch": 0.0141,
      "grad_norm": 0.7213404812022338,
      "learning_rate": 0.003,
      "loss": 4.326,
      "step": 1410
    },
    {
      "epoch": 0.01411,
      "grad_norm": 0.6616036396606474,
      "learning_rate": 0.003,
      "loss": 4.3016,
      "step": 1411
    },
    {
      "epoch": 0.01412,
      "grad_norm": 0.6008069011437791,
      "learning_rate": 0.003,
      "loss": 4.3378,
      "step": 1412
    },
    {
      "epoch": 0.01413,
      "grad_norm": 0.7020614194036953,
      "learning_rate": 0.003,
      "loss": 4.3356,
      "step": 1413
    },
    {
      "epoch": 0.01414,
      "grad_norm": 0.6966655821964342,
      "learning_rate": 0.003,
      "loss": 4.322,
      "step": 1414
    },
    {
      "epoch": 0.01415,
      "grad_norm": 0.7878825992848615,
      "learning_rate": 0.003,
      "loss": 4.3182,
      "step": 1415
    },
    {
      "epoch": 0.01416,
      "grad_norm": 0.9133447782042373,
      "learning_rate": 0.003,
      "loss": 4.3606,
      "step": 1416
    },
    {
      "epoch": 0.01417,
      "grad_norm": 1.0833856715197905,
      "learning_rate": 0.003,
      "loss": 4.3655,
      "step": 1417
    },
    {
      "epoch": 0.01418,
      "grad_norm": 0.8776183907401212,
      "learning_rate": 0.003,
      "loss": 4.3779,
      "step": 1418
    },
    {
      "epoch": 0.01419,
      "grad_norm": 0.8716355229415133,
      "learning_rate": 0.003,
      "loss": 4.3745,
      "step": 1419
    },
    {
      "epoch": 0.0142,
      "grad_norm": 0.6900477550364122,
      "learning_rate": 0.003,
      "loss": 4.3678,
      "step": 1420
    },
    {
      "epoch": 0.01421,
      "grad_norm": 0.6518139426774641,
      "learning_rate": 0.003,
      "loss": 4.3347,
      "step": 1421
    },
    {
      "epoch": 0.01422,
      "grad_norm": 0.6549157122857109,
      "learning_rate": 0.003,
      "loss": 4.3382,
      "step": 1422
    },
    {
      "epoch": 0.01423,
      "grad_norm": 0.6092791265519953,
      "learning_rate": 0.003,
      "loss": 4.3402,
      "step": 1423
    },
    {
      "epoch": 0.01424,
      "grad_norm": 0.5943123649374407,
      "learning_rate": 0.003,
      "loss": 4.3537,
      "step": 1424
    },
    {
      "epoch": 0.01425,
      "grad_norm": 0.5086164834561862,
      "learning_rate": 0.003,
      "loss": 4.3528,
      "step": 1425
    },
    {
      "epoch": 0.01426,
      "grad_norm": 0.45109246822140664,
      "learning_rate": 0.003,
      "loss": 4.3278,
      "step": 1426
    },
    {
      "epoch": 0.01427,
      "grad_norm": 0.43801724179324136,
      "learning_rate": 0.003,
      "loss": 4.3209,
      "step": 1427
    },
    {
      "epoch": 0.01428,
      "grad_norm": 0.5018488931475374,
      "learning_rate": 0.003,
      "loss": 4.3159,
      "step": 1428
    },
    {
      "epoch": 0.01429,
      "grad_norm": 0.47066094753828186,
      "learning_rate": 0.003,
      "loss": 4.3234,
      "step": 1429
    },
    {
      "epoch": 0.0143,
      "grad_norm": 0.4403238802934058,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1430
    },
    {
      "epoch": 0.01431,
      "grad_norm": 0.44039117208650785,
      "learning_rate": 0.003,
      "loss": 4.3183,
      "step": 1431
    },
    {
      "epoch": 0.01432,
      "grad_norm": 0.477019489570791,
      "learning_rate": 0.003,
      "loss": 4.3074,
      "step": 1432
    },
    {
      "epoch": 0.01433,
      "grad_norm": 0.4264410276196829,
      "learning_rate": 0.003,
      "loss": 4.3017,
      "step": 1433
    },
    {
      "epoch": 0.01434,
      "grad_norm": 0.3944694636285646,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1434
    },
    {
      "epoch": 0.01435,
      "grad_norm": 0.3924802801708403,
      "learning_rate": 0.003,
      "loss": 4.2942,
      "step": 1435
    },
    {
      "epoch": 0.01436,
      "grad_norm": 0.4172733802501552,
      "learning_rate": 0.003,
      "loss": 4.3128,
      "step": 1436
    },
    {
      "epoch": 0.01437,
      "grad_norm": 0.3851734831243834,
      "learning_rate": 0.003,
      "loss": 4.3237,
      "step": 1437
    },
    {
      "epoch": 0.01438,
      "grad_norm": 0.35404107026104975,
      "learning_rate": 0.003,
      "loss": 4.3163,
      "step": 1438
    },
    {
      "epoch": 0.01439,
      "grad_norm": 0.3935862503777362,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1439
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.48537387134962556,
      "learning_rate": 0.003,
      "loss": 4.2941,
      "step": 1440
    },
    {
      "epoch": 0.01441,
      "grad_norm": 0.6205869970599557,
      "learning_rate": 0.003,
      "loss": 4.3081,
      "step": 1441
    },
    {
      "epoch": 0.01442,
      "grad_norm": 0.733625820231546,
      "learning_rate": 0.003,
      "loss": 4.2954,
      "step": 1442
    },
    {
      "epoch": 0.01443,
      "grad_norm": 0.6539626450528481,
      "learning_rate": 0.003,
      "loss": 4.3597,
      "step": 1443
    },
    {
      "epoch": 0.01444,
      "grad_norm": 0.4715932337839517,
      "learning_rate": 0.003,
      "loss": 4.3144,
      "step": 1444
    },
    {
      "epoch": 0.01445,
      "grad_norm": 0.5734177290330968,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 1445
    },
    {
      "epoch": 0.01446,
      "grad_norm": 0.6345599077429956,
      "learning_rate": 0.003,
      "loss": 4.3,
      "step": 1446
    },
    {
      "epoch": 0.01447,
      "grad_norm": 0.7159211854064349,
      "learning_rate": 0.003,
      "loss": 4.3124,
      "step": 1447
    },
    {
      "epoch": 0.01448,
      "grad_norm": 0.847805509409058,
      "learning_rate": 0.003,
      "loss": 4.3346,
      "step": 1448
    },
    {
      "epoch": 0.01449,
      "grad_norm": 0.8334367502615841,
      "learning_rate": 0.003,
      "loss": 4.3316,
      "step": 1449
    },
    {
      "epoch": 0.0145,
      "grad_norm": 0.8387485961034247,
      "learning_rate": 0.003,
      "loss": 4.3521,
      "step": 1450
    },
    {
      "epoch": 0.01451,
      "grad_norm": 0.7069735242725316,
      "learning_rate": 0.003,
      "loss": 4.3254,
      "step": 1451
    },
    {
      "epoch": 0.01452,
      "grad_norm": 0.7128306367496979,
      "learning_rate": 0.003,
      "loss": 4.3315,
      "step": 1452
    },
    {
      "epoch": 0.01453,
      "grad_norm": 0.7607386324594154,
      "learning_rate": 0.003,
      "loss": 4.3465,
      "step": 1453
    },
    {
      "epoch": 0.01454,
      "grad_norm": 0.8531443855411419,
      "learning_rate": 0.003,
      "loss": 4.3152,
      "step": 1454
    },
    {
      "epoch": 0.01455,
      "grad_norm": 0.9072653937920705,
      "learning_rate": 0.003,
      "loss": 4.3602,
      "step": 1455
    },
    {
      "epoch": 0.01456,
      "grad_norm": 0.9678032867174141,
      "learning_rate": 0.003,
      "loss": 4.3448,
      "step": 1456
    },
    {
      "epoch": 0.01457,
      "grad_norm": 0.8263837469314969,
      "learning_rate": 0.003,
      "loss": 4.3358,
      "step": 1457
    },
    {
      "epoch": 0.01458,
      "grad_norm": 0.7586740850669937,
      "learning_rate": 0.003,
      "loss": 4.3274,
      "step": 1458
    },
    {
      "epoch": 0.01459,
      "grad_norm": 0.7104303096772606,
      "learning_rate": 0.003,
      "loss": 4.3506,
      "step": 1459
    },
    {
      "epoch": 0.0146,
      "grad_norm": 0.7105635489785329,
      "learning_rate": 0.003,
      "loss": 4.358,
      "step": 1460
    },
    {
      "epoch": 0.01461,
      "grad_norm": 0.6051123991598398,
      "learning_rate": 0.003,
      "loss": 4.3255,
      "step": 1461
    },
    {
      "epoch": 0.01462,
      "grad_norm": 0.49818465497872577,
      "learning_rate": 0.003,
      "loss": 4.3277,
      "step": 1462
    },
    {
      "epoch": 0.01463,
      "grad_norm": 0.4705387398623177,
      "learning_rate": 0.003,
      "loss": 4.3234,
      "step": 1463
    },
    {
      "epoch": 0.01464,
      "grad_norm": 0.43420725887330847,
      "learning_rate": 0.003,
      "loss": 4.2947,
      "step": 1464
    },
    {
      "epoch": 0.01465,
      "grad_norm": 0.416394344592158,
      "learning_rate": 0.003,
      "loss": 4.3376,
      "step": 1465
    },
    {
      "epoch": 0.01466,
      "grad_norm": 0.46274333085308,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1466
    },
    {
      "epoch": 0.01467,
      "grad_norm": 0.412473133729069,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1467
    },
    {
      "epoch": 0.01468,
      "grad_norm": 0.4311218055740771,
      "learning_rate": 0.003,
      "loss": 4.3146,
      "step": 1468
    },
    {
      "epoch": 0.01469,
      "grad_norm": 0.4074958191436831,
      "learning_rate": 0.003,
      "loss": 4.312,
      "step": 1469
    },
    {
      "epoch": 0.0147,
      "grad_norm": 0.43399248954304653,
      "learning_rate": 0.003,
      "loss": 4.3033,
      "step": 1470
    },
    {
      "epoch": 0.01471,
      "grad_norm": 0.41361534167776814,
      "learning_rate": 0.003,
      "loss": 4.3141,
      "step": 1471
    },
    {
      "epoch": 0.01472,
      "grad_norm": 0.4291933422709367,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1472
    },
    {
      "epoch": 0.01473,
      "grad_norm": 0.4534239206252403,
      "learning_rate": 0.003,
      "loss": 4.3122,
      "step": 1473
    },
    {
      "epoch": 0.01474,
      "grad_norm": 0.5901119500480289,
      "learning_rate": 0.003,
      "loss": 4.3004,
      "step": 1474
    },
    {
      "epoch": 0.01475,
      "grad_norm": 0.893255157155039,
      "learning_rate": 0.003,
      "loss": 4.3326,
      "step": 1475
    },
    {
      "epoch": 0.01476,
      "grad_norm": 1.4183061783370785,
      "learning_rate": 0.003,
      "loss": 4.3574,
      "step": 1476
    },
    {
      "epoch": 0.01477,
      "grad_norm": 0.6731343966041683,
      "learning_rate": 0.003,
      "loss": 4.3345,
      "step": 1477
    },
    {
      "epoch": 0.01478,
      "grad_norm": 0.6759295392700941,
      "learning_rate": 0.003,
      "loss": 4.3246,
      "step": 1478
    },
    {
      "epoch": 0.01479,
      "grad_norm": 0.5808036667947172,
      "learning_rate": 0.003,
      "loss": 4.3486,
      "step": 1479
    },
    {
      "epoch": 0.0148,
      "grad_norm": 0.5015217636867718,
      "learning_rate": 0.003,
      "loss": 4.3142,
      "step": 1480
    },
    {
      "epoch": 0.01481,
      "grad_norm": 0.4567088802488701,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1481
    },
    {
      "epoch": 0.01482,
      "grad_norm": 0.4183399036149268,
      "learning_rate": 0.003,
      "loss": 4.2956,
      "step": 1482
    },
    {
      "epoch": 0.01483,
      "grad_norm": 0.4450691330955631,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1483
    },
    {
      "epoch": 0.01484,
      "grad_norm": 0.3765600896203587,
      "learning_rate": 0.003,
      "loss": 4.3398,
      "step": 1484
    },
    {
      "epoch": 0.01485,
      "grad_norm": 0.3992681409647294,
      "learning_rate": 0.003,
      "loss": 4.3029,
      "step": 1485
    },
    {
      "epoch": 0.01486,
      "grad_norm": 0.3969999829277036,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1486
    },
    {
      "epoch": 0.01487,
      "grad_norm": 0.3308465500623018,
      "learning_rate": 0.003,
      "loss": 4.2984,
      "step": 1487
    },
    {
      "epoch": 0.01488,
      "grad_norm": 0.2814090123328206,
      "learning_rate": 0.003,
      "loss": 4.3052,
      "step": 1488
    },
    {
      "epoch": 0.01489,
      "grad_norm": 0.30005415710473554,
      "learning_rate": 0.003,
      "loss": 4.2986,
      "step": 1489
    },
    {
      "epoch": 0.0149,
      "grad_norm": 0.32421623348009687,
      "learning_rate": 0.003,
      "loss": 4.2912,
      "step": 1490
    },
    {
      "epoch": 0.01491,
      "grad_norm": 0.35115426496092417,
      "learning_rate": 0.003,
      "loss": 4.3043,
      "step": 1491
    },
    {
      "epoch": 0.01492,
      "grad_norm": 0.37018314484045245,
      "learning_rate": 0.003,
      "loss": 4.2686,
      "step": 1492
    },
    {
      "epoch": 0.01493,
      "grad_norm": 0.4075858968364619,
      "learning_rate": 0.003,
      "loss": 4.3015,
      "step": 1493
    },
    {
      "epoch": 0.01494,
      "grad_norm": 0.4912830051602376,
      "learning_rate": 0.003,
      "loss": 4.2818,
      "step": 1494
    },
    {
      "epoch": 0.01495,
      "grad_norm": 0.5565886808339906,
      "learning_rate": 0.003,
      "loss": 4.2955,
      "step": 1495
    },
    {
      "epoch": 0.01496,
      "grad_norm": 0.6104515927182645,
      "learning_rate": 0.003,
      "loss": 4.2731,
      "step": 1496
    },
    {
      "epoch": 0.01497,
      "grad_norm": 0.6049970308529968,
      "learning_rate": 0.003,
      "loss": 4.2492,
      "step": 1497
    },
    {
      "epoch": 0.01498,
      "grad_norm": 0.5392848754302815,
      "learning_rate": 0.003,
      "loss": 4.304,
      "step": 1498
    },
    {
      "epoch": 0.01499,
      "grad_norm": 0.544325685587062,
      "learning_rate": 0.003,
      "loss": 4.3252,
      "step": 1499
    },
    {
      "epoch": 0.015,
      "grad_norm": 0.5744687962127912,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1500
    },
    {
      "epoch": 0.01501,
      "grad_norm": 0.6650575718497268,
      "learning_rate": 0.003,
      "loss": 4.2836,
      "step": 1501
    },
    {
      "epoch": 0.01502,
      "grad_norm": 0.762494815469482,
      "learning_rate": 0.003,
      "loss": 4.3344,
      "step": 1502
    },
    {
      "epoch": 0.01503,
      "grad_norm": 0.8042887596944258,
      "learning_rate": 0.003,
      "loss": 4.3203,
      "step": 1503
    },
    {
      "epoch": 0.01504,
      "grad_norm": 0.8888233604685658,
      "learning_rate": 0.003,
      "loss": 4.3247,
      "step": 1504
    },
    {
      "epoch": 0.01505,
      "grad_norm": 0.9445171252403785,
      "learning_rate": 0.003,
      "loss": 4.3231,
      "step": 1505
    },
    {
      "epoch": 0.01506,
      "grad_norm": 1.0444284185336012,
      "learning_rate": 0.003,
      "loss": 4.3641,
      "step": 1506
    },
    {
      "epoch": 0.01507,
      "grad_norm": 1.1261023785455877,
      "learning_rate": 0.003,
      "loss": 4.3931,
      "step": 1507
    },
    {
      "epoch": 0.01508,
      "grad_norm": 1.234913329930944,
      "learning_rate": 0.003,
      "loss": 4.3763,
      "step": 1508
    },
    {
      "epoch": 0.01509,
      "grad_norm": 0.8497166266198067,
      "learning_rate": 0.003,
      "loss": 4.3805,
      "step": 1509
    },
    {
      "epoch": 0.0151,
      "grad_norm": 0.8376711335758371,
      "learning_rate": 0.003,
      "loss": 4.3542,
      "step": 1510
    },
    {
      "epoch": 0.01511,
      "grad_norm": 0.724530376603094,
      "learning_rate": 0.003,
      "loss": 4.3387,
      "step": 1511
    },
    {
      "epoch": 0.01512,
      "grad_norm": 0.6831424127546084,
      "learning_rate": 0.003,
      "loss": 4.3429,
      "step": 1512
    },
    {
      "epoch": 0.01513,
      "grad_norm": 0.5603527644059662,
      "learning_rate": 0.003,
      "loss": 4.3405,
      "step": 1513
    },
    {
      "epoch": 0.01514,
      "grad_norm": 0.5930589812891277,
      "learning_rate": 0.003,
      "loss": 4.3465,
      "step": 1514
    },
    {
      "epoch": 0.01515,
      "grad_norm": 0.7232891849880219,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1515
    },
    {
      "epoch": 0.01516,
      "grad_norm": 0.7759860995114962,
      "learning_rate": 0.003,
      "loss": 4.3508,
      "step": 1516
    },
    {
      "epoch": 0.01517,
      "grad_norm": 0.6404044980036384,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1517
    },
    {
      "epoch": 0.01518,
      "grad_norm": 0.640364805854496,
      "learning_rate": 0.003,
      "loss": 4.3263,
      "step": 1518
    },
    {
      "epoch": 0.01519,
      "grad_norm": 0.7392318945229942,
      "learning_rate": 0.003,
      "loss": 4.3661,
      "step": 1519
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.9450167317883038,
      "learning_rate": 0.003,
      "loss": 4.3284,
      "step": 1520
    },
    {
      "epoch": 0.01521,
      "grad_norm": 0.9107576371118652,
      "learning_rate": 0.003,
      "loss": 4.3495,
      "step": 1521
    },
    {
      "epoch": 0.01522,
      "grad_norm": 0.7612821914439016,
      "learning_rate": 0.003,
      "loss": 4.3394,
      "step": 1522
    },
    {
      "epoch": 0.01523,
      "grad_norm": 0.6613113244902528,
      "learning_rate": 0.003,
      "loss": 4.3287,
      "step": 1523
    },
    {
      "epoch": 0.01524,
      "grad_norm": 0.5975485724244859,
      "learning_rate": 0.003,
      "loss": 4.3558,
      "step": 1524
    },
    {
      "epoch": 0.01525,
      "grad_norm": 0.49953522198986544,
      "learning_rate": 0.003,
      "loss": 4.349,
      "step": 1525
    },
    {
      "epoch": 0.01526,
      "grad_norm": 0.5020786027972391,
      "learning_rate": 0.003,
      "loss": 4.339,
      "step": 1526
    },
    {
      "epoch": 0.01527,
      "grad_norm": 0.4272044641011338,
      "learning_rate": 0.003,
      "loss": 4.301,
      "step": 1527
    },
    {
      "epoch": 0.01528,
      "grad_norm": 0.38333031076180046,
      "learning_rate": 0.003,
      "loss": 4.3172,
      "step": 1528
    },
    {
      "epoch": 0.01529,
      "grad_norm": 0.37710300633649013,
      "learning_rate": 0.003,
      "loss": 4.3078,
      "step": 1529
    },
    {
      "epoch": 0.0153,
      "grad_norm": 0.3539206796857484,
      "learning_rate": 0.003,
      "loss": 4.3136,
      "step": 1530
    },
    {
      "epoch": 0.01531,
      "grad_norm": 0.31579499408524375,
      "learning_rate": 0.003,
      "loss": 4.3044,
      "step": 1531
    },
    {
      "epoch": 0.01532,
      "grad_norm": 0.28740282558130736,
      "learning_rate": 0.003,
      "loss": 4.2793,
      "step": 1532
    },
    {
      "epoch": 0.01533,
      "grad_norm": 0.32298125914434667,
      "learning_rate": 0.003,
      "loss": 4.3053,
      "step": 1533
    },
    {
      "epoch": 0.01534,
      "grad_norm": 0.34152544816126584,
      "learning_rate": 0.003,
      "loss": 4.302,
      "step": 1534
    },
    {
      "epoch": 0.01535,
      "grad_norm": 0.40345568055827374,
      "learning_rate": 0.003,
      "loss": 4.3133,
      "step": 1535
    },
    {
      "epoch": 0.01536,
      "grad_norm": 0.47716276358652043,
      "learning_rate": 0.003,
      "loss": 4.2948,
      "step": 1536
    },
    {
      "epoch": 0.01537,
      "grad_norm": 0.5673450395929736,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1537
    },
    {
      "epoch": 0.01538,
      "grad_norm": 0.5662986079723249,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1538
    },
    {
      "epoch": 0.01539,
      "grad_norm": 0.5250656872774493,
      "learning_rate": 0.003,
      "loss": 4.2993,
      "step": 1539
    },
    {
      "epoch": 0.0154,
      "grad_norm": 0.4759070450942551,
      "learning_rate": 0.003,
      "loss": 4.3011,
      "step": 1540
    },
    {
      "epoch": 0.01541,
      "grad_norm": 0.525208601732405,
      "learning_rate": 0.003,
      "loss": 4.2841,
      "step": 1541
    },
    {
      "epoch": 0.01542,
      "grad_norm": 0.6195424842603703,
      "learning_rate": 0.003,
      "loss": 4.3158,
      "step": 1542
    },
    {
      "epoch": 0.01543,
      "grad_norm": 0.7066044834019704,
      "learning_rate": 0.003,
      "loss": 4.3015,
      "step": 1543
    },
    {
      "epoch": 0.01544,
      "grad_norm": 0.6094492242185264,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1544
    },
    {
      "epoch": 0.01545,
      "grad_norm": 0.6542596992859665,
      "learning_rate": 0.003,
      "loss": 4.3554,
      "step": 1545
    },
    {
      "epoch": 0.01546,
      "grad_norm": 0.7470834345800471,
      "learning_rate": 0.003,
      "loss": 4.3329,
      "step": 1546
    },
    {
      "epoch": 0.01547,
      "grad_norm": 0.7200211733418024,
      "learning_rate": 0.003,
      "loss": 4.3307,
      "step": 1547
    },
    {
      "epoch": 0.01548,
      "grad_norm": 0.6295943931814879,
      "learning_rate": 0.003,
      "loss": 4.2934,
      "step": 1548
    },
    {
      "epoch": 0.01549,
      "grad_norm": 0.54945766010764,
      "learning_rate": 0.003,
      "loss": 4.2766,
      "step": 1549
    },
    {
      "epoch": 0.0155,
      "grad_norm": 0.5950909013773433,
      "learning_rate": 0.003,
      "loss": 4.297,
      "step": 1550
    },
    {
      "epoch": 0.01551,
      "grad_norm": 0.6224936273193388,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 1551
    },
    {
      "epoch": 0.01552,
      "grad_norm": 0.6624794177801069,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1552
    },
    {
      "epoch": 0.01553,
      "grad_norm": 0.7863145855853837,
      "learning_rate": 0.003,
      "loss": 4.335,
      "step": 1553
    },
    {
      "epoch": 0.01554,
      "grad_norm": 0.7964168301162362,
      "learning_rate": 0.003,
      "loss": 4.3213,
      "step": 1554
    },
    {
      "epoch": 0.01555,
      "grad_norm": 0.6686376199306979,
      "learning_rate": 0.003,
      "loss": 4.292,
      "step": 1555
    },
    {
      "epoch": 0.01556,
      "grad_norm": 0.6587445991969533,
      "learning_rate": 0.003,
      "loss": 4.327,
      "step": 1556
    },
    {
      "epoch": 0.01557,
      "grad_norm": 0.8168668532513341,
      "learning_rate": 0.003,
      "loss": 4.3101,
      "step": 1557
    },
    {
      "epoch": 0.01558,
      "grad_norm": 0.6341929215834538,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 1558
    },
    {
      "epoch": 0.01559,
      "grad_norm": 0.5639732491475488,
      "learning_rate": 0.003,
      "loss": 4.3265,
      "step": 1559
    },
    {
      "epoch": 0.0156,
      "grad_norm": 0.6344606998016725,
      "learning_rate": 0.003,
      "loss": 4.3189,
      "step": 1560
    },
    {
      "epoch": 0.01561,
      "grad_norm": 0.6020928854417142,
      "learning_rate": 0.003,
      "loss": 4.301,
      "step": 1561
    },
    {
      "epoch": 0.01562,
      "grad_norm": 0.6447473041378132,
      "learning_rate": 0.003,
      "loss": 4.2957,
      "step": 1562
    },
    {
      "epoch": 0.01563,
      "grad_norm": 0.637527615370107,
      "learning_rate": 0.003,
      "loss": 4.3012,
      "step": 1563
    },
    {
      "epoch": 0.01564,
      "grad_norm": 0.6065116381126623,
      "learning_rate": 0.003,
      "loss": 4.3275,
      "step": 1564
    },
    {
      "epoch": 0.01565,
      "grad_norm": 0.5530540273129289,
      "learning_rate": 0.003,
      "loss": 4.3008,
      "step": 1565
    },
    {
      "epoch": 0.01566,
      "grad_norm": 0.523767221886022,
      "learning_rate": 0.003,
      "loss": 4.3185,
      "step": 1566
    },
    {
      "epoch": 0.01567,
      "grad_norm": 0.44666292802850066,
      "learning_rate": 0.003,
      "loss": 4.3073,
      "step": 1567
    },
    {
      "epoch": 0.01568,
      "grad_norm": 0.4825649782380437,
      "learning_rate": 0.003,
      "loss": 4.2912,
      "step": 1568
    },
    {
      "epoch": 0.01569,
      "grad_norm": 0.4768074953920202,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1569
    },
    {
      "epoch": 0.0157,
      "grad_norm": 0.4756489341732161,
      "learning_rate": 0.003,
      "loss": 4.3112,
      "step": 1570
    },
    {
      "epoch": 0.01571,
      "grad_norm": 0.5079994555205132,
      "learning_rate": 0.003,
      "loss": 4.2965,
      "step": 1571
    },
    {
      "epoch": 0.01572,
      "grad_norm": 0.5289839099051613,
      "learning_rate": 0.003,
      "loss": 4.3054,
      "step": 1572
    },
    {
      "epoch": 0.01573,
      "grad_norm": 0.5222081857718813,
      "learning_rate": 0.003,
      "loss": 4.2898,
      "step": 1573
    },
    {
      "epoch": 0.01574,
      "grad_norm": 0.6451289760332761,
      "learning_rate": 0.003,
      "loss": 4.3076,
      "step": 1574
    },
    {
      "epoch": 0.01575,
      "grad_norm": 0.7877974193239016,
      "learning_rate": 0.003,
      "loss": 4.2813,
      "step": 1575
    },
    {
      "epoch": 0.01576,
      "grad_norm": 0.8607804040917739,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1576
    },
    {
      "epoch": 0.01577,
      "grad_norm": 0.8142185085462708,
      "learning_rate": 0.003,
      "loss": 4.2736,
      "step": 1577
    },
    {
      "epoch": 0.01578,
      "grad_norm": 0.7720878068888671,
      "learning_rate": 0.003,
      "loss": 4.3026,
      "step": 1578
    },
    {
      "epoch": 0.01579,
      "grad_norm": 0.7192241887142894,
      "learning_rate": 0.003,
      "loss": 4.2996,
      "step": 1579
    },
    {
      "epoch": 0.0158,
      "grad_norm": 0.606159564012465,
      "learning_rate": 0.003,
      "loss": 4.3013,
      "step": 1580
    },
    {
      "epoch": 0.01581,
      "grad_norm": 0.4880751313224698,
      "learning_rate": 0.003,
      "loss": 4.3071,
      "step": 1581
    },
    {
      "epoch": 0.01582,
      "grad_norm": 0.5304647614551631,
      "learning_rate": 0.003,
      "loss": 4.2891,
      "step": 1582
    },
    {
      "epoch": 0.01583,
      "grad_norm": 0.46382514166554145,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1583
    },
    {
      "epoch": 0.01584,
      "grad_norm": 0.47578841605412814,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1584
    },
    {
      "epoch": 0.01585,
      "grad_norm": 0.4985689869447637,
      "learning_rate": 0.003,
      "loss": 4.2972,
      "step": 1585
    },
    {
      "epoch": 0.01586,
      "grad_norm": 0.4793261849715971,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 1586
    },
    {
      "epoch": 0.01587,
      "grad_norm": 0.4867805801070396,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1587
    },
    {
      "epoch": 0.01588,
      "grad_norm": 0.49512882436423494,
      "learning_rate": 0.003,
      "loss": 4.3139,
      "step": 1588
    },
    {
      "epoch": 0.01589,
      "grad_norm": 0.5289005946834403,
      "learning_rate": 0.003,
      "loss": 4.306,
      "step": 1589
    },
    {
      "epoch": 0.0159,
      "grad_norm": 0.6096773535511837,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1590
    },
    {
      "epoch": 0.01591,
      "grad_norm": 0.6229667469755624,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1591
    },
    {
      "epoch": 0.01592,
      "grad_norm": 0.4936084502041978,
      "learning_rate": 0.003,
      "loss": 4.2967,
      "step": 1592
    },
    {
      "epoch": 0.01593,
      "grad_norm": 0.42446523481816184,
      "learning_rate": 0.003,
      "loss": 4.2946,
      "step": 1593
    },
    {
      "epoch": 0.01594,
      "grad_norm": 0.40613386955274233,
      "learning_rate": 0.003,
      "loss": 4.2814,
      "step": 1594
    },
    {
      "epoch": 0.01595,
      "grad_norm": 0.4303780684833319,
      "learning_rate": 0.003,
      "loss": 4.308,
      "step": 1595
    },
    {
      "epoch": 0.01596,
      "grad_norm": 0.4038673300031157,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1596
    },
    {
      "epoch": 0.01597,
      "grad_norm": 0.42824275182777244,
      "learning_rate": 0.003,
      "loss": 4.2611,
      "step": 1597
    },
    {
      "epoch": 0.01598,
      "grad_norm": 0.4949313860176736,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1598
    },
    {
      "epoch": 0.01599,
      "grad_norm": 0.5005123390361799,
      "learning_rate": 0.003,
      "loss": 4.2803,
      "step": 1599
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.5929838548199676,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 1600
    },
    {
      "epoch": 0.01601,
      "grad_norm": 0.8301504230585989,
      "learning_rate": 0.003,
      "loss": 4.315,
      "step": 1601
    },
    {
      "epoch": 0.01602,
      "grad_norm": 0.9853750506753596,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1602
    },
    {
      "epoch": 0.01603,
      "grad_norm": 1.003124318602637,
      "learning_rate": 0.003,
      "loss": 4.3258,
      "step": 1603
    },
    {
      "epoch": 0.01604,
      "grad_norm": 0.912578535311941,
      "learning_rate": 0.003,
      "loss": 4.327,
      "step": 1604
    },
    {
      "epoch": 0.01605,
      "grad_norm": 0.8301067621469277,
      "learning_rate": 0.003,
      "loss": 4.3002,
      "step": 1605
    },
    {
      "epoch": 0.01606,
      "grad_norm": 0.7646428413230717,
      "learning_rate": 0.003,
      "loss": 4.3034,
      "step": 1606
    },
    {
      "epoch": 0.01607,
      "grad_norm": 0.8210614746463125,
      "learning_rate": 0.003,
      "loss": 4.3396,
      "step": 1607
    },
    {
      "epoch": 0.01608,
      "grad_norm": 0.8162978830688478,
      "learning_rate": 0.003,
      "loss": 4.3084,
      "step": 1608
    },
    {
      "epoch": 0.01609,
      "grad_norm": 0.7940470937374723,
      "learning_rate": 0.003,
      "loss": 4.3225,
      "step": 1609
    },
    {
      "epoch": 0.0161,
      "grad_norm": 0.7532673019192272,
      "learning_rate": 0.003,
      "loss": 4.2882,
      "step": 1610
    },
    {
      "epoch": 0.01611,
      "grad_norm": 0.77086180161133,
      "learning_rate": 0.003,
      "loss": 4.3117,
      "step": 1611
    },
    {
      "epoch": 0.01612,
      "grad_norm": 0.6754537663457288,
      "learning_rate": 0.003,
      "loss": 4.2961,
      "step": 1612
    },
    {
      "epoch": 0.01613,
      "grad_norm": 0.599175162495873,
      "learning_rate": 0.003,
      "loss": 4.3386,
      "step": 1613
    },
    {
      "epoch": 0.01614,
      "grad_norm": 0.5411062109149707,
      "learning_rate": 0.003,
      "loss": 4.3566,
      "step": 1614
    },
    {
      "epoch": 0.01615,
      "grad_norm": 0.512685672879493,
      "learning_rate": 0.003,
      "loss": 4.3203,
      "step": 1615
    },
    {
      "epoch": 0.01616,
      "grad_norm": 0.49714926242353863,
      "learning_rate": 0.003,
      "loss": 4.3009,
      "step": 1616
    },
    {
      "epoch": 0.01617,
      "grad_norm": 0.4964011411644138,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1617
    },
    {
      "epoch": 0.01618,
      "grad_norm": 0.5061608293930215,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1618
    },
    {
      "epoch": 0.01619,
      "grad_norm": 0.5494277838056353,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1619
    },
    {
      "epoch": 0.0162,
      "grad_norm": 0.6308404917903175,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 1620
    },
    {
      "epoch": 0.01621,
      "grad_norm": 0.731307523044738,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1621
    },
    {
      "epoch": 0.01622,
      "grad_norm": 0.9046402251963843,
      "learning_rate": 0.003,
      "loss": 4.3346,
      "step": 1622
    },
    {
      "epoch": 0.01623,
      "grad_norm": 0.7735409772436379,
      "learning_rate": 0.003,
      "loss": 4.3147,
      "step": 1623
    },
    {
      "epoch": 0.01624,
      "grad_norm": 0.672645275853148,
      "learning_rate": 0.003,
      "loss": 4.3229,
      "step": 1624
    },
    {
      "epoch": 0.01625,
      "grad_norm": 0.6719787877984026,
      "learning_rate": 0.003,
      "loss": 4.2983,
      "step": 1625
    },
    {
      "epoch": 0.01626,
      "grad_norm": 0.6708647627350693,
      "learning_rate": 0.003,
      "loss": 4.3168,
      "step": 1626
    },
    {
      "epoch": 0.01627,
      "grad_norm": 0.649063452794678,
      "learning_rate": 0.003,
      "loss": 4.2886,
      "step": 1627
    },
    {
      "epoch": 0.01628,
      "grad_norm": 0.6161891488716968,
      "learning_rate": 0.003,
      "loss": 4.2901,
      "step": 1628
    },
    {
      "epoch": 0.01629,
      "grad_norm": 0.5565589154413988,
      "learning_rate": 0.003,
      "loss": 4.2903,
      "step": 1629
    },
    {
      "epoch": 0.0163,
      "grad_norm": 0.5030071975984001,
      "learning_rate": 0.003,
      "loss": 4.2755,
      "step": 1630
    },
    {
      "epoch": 0.01631,
      "grad_norm": 0.48590911922639535,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1631
    },
    {
      "epoch": 0.01632,
      "grad_norm": 0.5725399795731132,
      "learning_rate": 0.003,
      "loss": 4.3061,
      "step": 1632
    },
    {
      "epoch": 0.01633,
      "grad_norm": 0.7370492556479813,
      "learning_rate": 0.003,
      "loss": 4.2846,
      "step": 1633
    },
    {
      "epoch": 0.01634,
      "grad_norm": 0.8641758267856038,
      "learning_rate": 0.003,
      "loss": 4.3131,
      "step": 1634
    },
    {
      "epoch": 0.01635,
      "grad_norm": 0.7344051689167792,
      "learning_rate": 0.003,
      "loss": 4.2907,
      "step": 1635
    },
    {
      "epoch": 0.01636,
      "grad_norm": 0.5215072843138687,
      "learning_rate": 0.003,
      "loss": 4.2982,
      "step": 1636
    },
    {
      "epoch": 0.01637,
      "grad_norm": 0.4749653803767528,
      "learning_rate": 0.003,
      "loss": 4.28,
      "step": 1637
    },
    {
      "epoch": 0.01638,
      "grad_norm": 0.4605003028840984,
      "learning_rate": 0.003,
      "loss": 4.2856,
      "step": 1638
    },
    {
      "epoch": 0.01639,
      "grad_norm": 0.42552013547580053,
      "learning_rate": 0.003,
      "loss": 4.3073,
      "step": 1639
    },
    {
      "epoch": 0.0164,
      "grad_norm": 0.4220228883023831,
      "learning_rate": 0.003,
      "loss": 4.2853,
      "step": 1640
    },
    {
      "epoch": 0.01641,
      "grad_norm": 0.48264096364388565,
      "learning_rate": 0.003,
      "loss": 4.2601,
      "step": 1641
    },
    {
      "epoch": 0.01642,
      "grad_norm": 0.5522814057439278,
      "learning_rate": 0.003,
      "loss": 4.3056,
      "step": 1642
    },
    {
      "epoch": 0.01643,
      "grad_norm": 0.5448290909435038,
      "learning_rate": 0.003,
      "loss": 4.2899,
      "step": 1643
    },
    {
      "epoch": 0.01644,
      "grad_norm": 0.4693252715517831,
      "learning_rate": 0.003,
      "loss": 4.3048,
      "step": 1644
    },
    {
      "epoch": 0.01645,
      "grad_norm": 0.38397009629291196,
      "learning_rate": 0.003,
      "loss": 4.2478,
      "step": 1645
    },
    {
      "epoch": 0.01646,
      "grad_norm": 0.47479208422272845,
      "learning_rate": 0.003,
      "loss": 4.2841,
      "step": 1646
    },
    {
      "epoch": 0.01647,
      "grad_norm": 0.5095257415638307,
      "learning_rate": 0.003,
      "loss": 4.2814,
      "step": 1647
    },
    {
      "epoch": 0.01648,
      "grad_norm": 0.5223808020316233,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1648
    },
    {
      "epoch": 0.01649,
      "grad_norm": 0.4909186651871344,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1649
    },
    {
      "epoch": 0.0165,
      "grad_norm": 0.4445354228997141,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 1650
    },
    {
      "epoch": 0.01651,
      "grad_norm": 0.47462201592081604,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1651
    },
    {
      "epoch": 0.01652,
      "grad_norm": 0.5712842642113961,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1652
    },
    {
      "epoch": 0.01653,
      "grad_norm": 0.6633586551393839,
      "learning_rate": 0.003,
      "loss": 4.2872,
      "step": 1653
    },
    {
      "epoch": 0.01654,
      "grad_norm": 0.6882337845469141,
      "learning_rate": 0.003,
      "loss": 4.2797,
      "step": 1654
    },
    {
      "epoch": 0.01655,
      "grad_norm": 0.5883903260155589,
      "learning_rate": 0.003,
      "loss": 4.2855,
      "step": 1655
    },
    {
      "epoch": 0.01656,
      "grad_norm": 0.692175129507891,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 1656
    },
    {
      "epoch": 0.01657,
      "grad_norm": 0.7208024346780962,
      "learning_rate": 0.003,
      "loss": 4.285,
      "step": 1657
    },
    {
      "epoch": 0.01658,
      "grad_norm": 0.7405524865262961,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1658
    },
    {
      "epoch": 0.01659,
      "grad_norm": 0.6893828101785304,
      "learning_rate": 0.003,
      "loss": 4.2781,
      "step": 1659
    },
    {
      "epoch": 0.0166,
      "grad_norm": 0.6598372867680238,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1660
    },
    {
      "epoch": 0.01661,
      "grad_norm": 0.47529948561931934,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1661
    },
    {
      "epoch": 0.01662,
      "grad_norm": 0.46023265898975174,
      "learning_rate": 0.003,
      "loss": 4.2646,
      "step": 1662
    },
    {
      "epoch": 0.01663,
      "grad_norm": 0.4507377258313933,
      "learning_rate": 0.003,
      "loss": 4.2842,
      "step": 1663
    },
    {
      "epoch": 0.01664,
      "grad_norm": 0.4200561869900329,
      "learning_rate": 0.003,
      "loss": 4.2999,
      "step": 1664
    },
    {
      "epoch": 0.01665,
      "grad_norm": 0.4244798072607477,
      "learning_rate": 0.003,
      "loss": 4.2616,
      "step": 1665
    },
    {
      "epoch": 0.01666,
      "grad_norm": 0.5427314965460833,
      "learning_rate": 0.003,
      "loss": 4.2661,
      "step": 1666
    },
    {
      "epoch": 0.01667,
      "grad_norm": 0.5950166710073447,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1667
    },
    {
      "epoch": 0.01668,
      "grad_norm": 0.7033617268618768,
      "learning_rate": 0.003,
      "loss": 4.2937,
      "step": 1668
    },
    {
      "epoch": 0.01669,
      "grad_norm": 0.7337795007439258,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1669
    },
    {
      "epoch": 0.0167,
      "grad_norm": 0.7228924020186139,
      "learning_rate": 0.003,
      "loss": 4.2759,
      "step": 1670
    },
    {
      "epoch": 0.01671,
      "grad_norm": 0.9293602473006644,
      "learning_rate": 0.003,
      "loss": 4.2785,
      "step": 1671
    },
    {
      "epoch": 0.01672,
      "grad_norm": 1.029746366393395,
      "learning_rate": 0.003,
      "loss": 4.3064,
      "step": 1672
    },
    {
      "epoch": 0.01673,
      "grad_norm": 0.9727787844526897,
      "learning_rate": 0.003,
      "loss": 4.3102,
      "step": 1673
    },
    {
      "epoch": 0.01674,
      "grad_norm": 0.9554074542733949,
      "learning_rate": 0.003,
      "loss": 4.3312,
      "step": 1674
    },
    {
      "epoch": 0.01675,
      "grad_norm": 0.9706031573401521,
      "learning_rate": 0.003,
      "loss": 4.3194,
      "step": 1675
    },
    {
      "epoch": 0.01676,
      "grad_norm": 0.9822377528346978,
      "learning_rate": 0.003,
      "loss": 4.3319,
      "step": 1676
    },
    {
      "epoch": 0.01677,
      "grad_norm": 0.9324715324675055,
      "learning_rate": 0.003,
      "loss": 4.335,
      "step": 1677
    },
    {
      "epoch": 0.01678,
      "grad_norm": 0.8555988190467146,
      "learning_rate": 0.003,
      "loss": 4.3399,
      "step": 1678
    },
    {
      "epoch": 0.01679,
      "grad_norm": 0.9906267827738694,
      "learning_rate": 0.003,
      "loss": 4.3344,
      "step": 1679
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.9961431022000035,
      "learning_rate": 0.003,
      "loss": 4.3626,
      "step": 1680
    },
    {
      "epoch": 0.01681,
      "grad_norm": 0.8880584596730197,
      "learning_rate": 0.003,
      "loss": 4.362,
      "step": 1681
    },
    {
      "epoch": 0.01682,
      "grad_norm": 0.864035612954157,
      "learning_rate": 0.003,
      "loss": 4.3519,
      "step": 1682
    },
    {
      "epoch": 0.01683,
      "grad_norm": 0.830453292981266,
      "learning_rate": 0.003,
      "loss": 4.3671,
      "step": 1683
    },
    {
      "epoch": 0.01684,
      "grad_norm": 0.8911963046678297,
      "learning_rate": 0.003,
      "loss": 4.3454,
      "step": 1684
    },
    {
      "epoch": 0.01685,
      "grad_norm": 0.8195919434333439,
      "learning_rate": 0.003,
      "loss": 4.3594,
      "step": 1685
    },
    {
      "epoch": 0.01686,
      "grad_norm": 0.8629708344208772,
      "learning_rate": 0.003,
      "loss": 4.3459,
      "step": 1686
    },
    {
      "epoch": 0.01687,
      "grad_norm": 0.7564331005412712,
      "learning_rate": 0.003,
      "loss": 4.3417,
      "step": 1687
    },
    {
      "epoch": 0.01688,
      "grad_norm": 0.5556779425751791,
      "learning_rate": 0.003,
      "loss": 4.3178,
      "step": 1688
    },
    {
      "epoch": 0.01689,
      "grad_norm": 0.48730261859047036,
      "learning_rate": 0.003,
      "loss": 4.3272,
      "step": 1689
    },
    {
      "epoch": 0.0169,
      "grad_norm": 0.44730185947220713,
      "learning_rate": 0.003,
      "loss": 4.3108,
      "step": 1690
    },
    {
      "epoch": 0.01691,
      "grad_norm": 0.4517455858191217,
      "learning_rate": 0.003,
      "loss": 4.3185,
      "step": 1691
    },
    {
      "epoch": 0.01692,
      "grad_norm": 0.3859623506260394,
      "learning_rate": 0.003,
      "loss": 4.3134,
      "step": 1692
    },
    {
      "epoch": 0.01693,
      "grad_norm": 0.4280684753336336,
      "learning_rate": 0.003,
      "loss": 4.3374,
      "step": 1693
    },
    {
      "epoch": 0.01694,
      "grad_norm": 0.430827914877835,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1694
    },
    {
      "epoch": 0.01695,
      "grad_norm": 0.43375392594165474,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1695
    },
    {
      "epoch": 0.01696,
      "grad_norm": 0.38718080857219184,
      "learning_rate": 0.003,
      "loss": 4.2992,
      "step": 1696
    },
    {
      "epoch": 0.01697,
      "grad_norm": 0.30363972361127406,
      "learning_rate": 0.003,
      "loss": 4.272,
      "step": 1697
    },
    {
      "epoch": 0.01698,
      "grad_norm": 0.31558190433180666,
      "learning_rate": 0.003,
      "loss": 4.2652,
      "step": 1698
    },
    {
      "epoch": 0.01699,
      "grad_norm": 0.3101672448572644,
      "learning_rate": 0.003,
      "loss": 4.2732,
      "step": 1699
    },
    {
      "epoch": 0.017,
      "grad_norm": 0.28571559628804677,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1700
    },
    {
      "epoch": 0.01701,
      "grad_norm": 0.31584250090208377,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 1701
    },
    {
      "epoch": 0.01702,
      "grad_norm": 0.36960398708639597,
      "learning_rate": 0.003,
      "loss": 4.2921,
      "step": 1702
    },
    {
      "epoch": 0.01703,
      "grad_norm": 0.4412523037970288,
      "learning_rate": 0.003,
      "loss": 4.277,
      "step": 1703
    },
    {
      "epoch": 0.01704,
      "grad_norm": 0.6716654424542439,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 1704
    },
    {
      "epoch": 0.01705,
      "grad_norm": 1.1072937714524183,
      "learning_rate": 0.003,
      "loss": 4.2932,
      "step": 1705
    },
    {
      "epoch": 0.01706,
      "grad_norm": 0.8287243662947943,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1706
    },
    {
      "epoch": 0.01707,
      "grad_norm": 0.6387856390419926,
      "learning_rate": 0.003,
      "loss": 4.3079,
      "step": 1707
    },
    {
      "epoch": 0.01708,
      "grad_norm": 0.7665689860690651,
      "learning_rate": 0.003,
      "loss": 4.2925,
      "step": 1708
    },
    {
      "epoch": 0.01709,
      "grad_norm": 0.6842592484083714,
      "learning_rate": 0.003,
      "loss": 4.3099,
      "step": 1709
    },
    {
      "epoch": 0.0171,
      "grad_norm": 0.5640128407150151,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 1710
    },
    {
      "epoch": 0.01711,
      "grad_norm": 0.5706999470034969,
      "learning_rate": 0.003,
      "loss": 4.2824,
      "step": 1711
    },
    {
      "epoch": 0.01712,
      "grad_norm": 0.4795902281641074,
      "learning_rate": 0.003,
      "loss": 4.2884,
      "step": 1712
    },
    {
      "epoch": 0.01713,
      "grad_norm": 0.425849272235493,
      "learning_rate": 0.003,
      "loss": 4.2609,
      "step": 1713
    },
    {
      "epoch": 0.01714,
      "grad_norm": 0.38297971865347935,
      "learning_rate": 0.003,
      "loss": 4.2951,
      "step": 1714
    },
    {
      "epoch": 0.01715,
      "grad_norm": 0.3866137138438359,
      "learning_rate": 0.003,
      "loss": 4.3021,
      "step": 1715
    },
    {
      "epoch": 0.01716,
      "grad_norm": 0.35314982927049987,
      "learning_rate": 0.003,
      "loss": 4.2748,
      "step": 1716
    },
    {
      "epoch": 0.01717,
      "grad_norm": 0.31552154478040295,
      "learning_rate": 0.003,
      "loss": 4.3044,
      "step": 1717
    },
    {
      "epoch": 0.01718,
      "grad_norm": 0.30638930868708425,
      "learning_rate": 0.003,
      "loss": 4.2709,
      "step": 1718
    },
    {
      "epoch": 0.01719,
      "grad_norm": 0.3127248594754382,
      "learning_rate": 0.003,
      "loss": 4.2905,
      "step": 1719
    },
    {
      "epoch": 0.0172,
      "grad_norm": 0.33132313581376,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 1720
    },
    {
      "epoch": 0.01721,
      "grad_norm": 0.3400058323027828,
      "learning_rate": 0.003,
      "loss": 4.2635,
      "step": 1721
    },
    {
      "epoch": 0.01722,
      "grad_norm": 0.3468025060877176,
      "learning_rate": 0.003,
      "loss": 4.2559,
      "step": 1722
    },
    {
      "epoch": 0.01723,
      "grad_norm": 0.35695313771503173,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 1723
    },
    {
      "epoch": 0.01724,
      "grad_norm": 0.34758213513255676,
      "learning_rate": 0.003,
      "loss": 4.2648,
      "step": 1724
    },
    {
      "epoch": 0.01725,
      "grad_norm": 0.40361756493758844,
      "learning_rate": 0.003,
      "loss": 4.2508,
      "step": 1725
    },
    {
      "epoch": 0.01726,
      "grad_norm": 0.5053798954640991,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 1726
    },
    {
      "epoch": 0.01727,
      "grad_norm": 0.6257283402319211,
      "learning_rate": 0.003,
      "loss": 4.2558,
      "step": 1727
    },
    {
      "epoch": 0.01728,
      "grad_norm": 0.5898179215522085,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1728
    },
    {
      "epoch": 0.01729,
      "grad_norm": 0.5313524252553607,
      "learning_rate": 0.003,
      "loss": 4.2643,
      "step": 1729
    },
    {
      "epoch": 0.0173,
      "grad_norm": 0.6103904325056944,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 1730
    },
    {
      "epoch": 0.01731,
      "grad_norm": 0.498773172096663,
      "learning_rate": 0.003,
      "loss": 4.2592,
      "step": 1731
    },
    {
      "epoch": 0.01732,
      "grad_norm": 0.6373537928005123,
      "learning_rate": 0.003,
      "loss": 4.2725,
      "step": 1732
    },
    {
      "epoch": 0.01733,
      "grad_norm": 0.7241473038492131,
      "learning_rate": 0.003,
      "loss": 4.257,
      "step": 1733
    },
    {
      "epoch": 0.01734,
      "grad_norm": 0.8759884069698398,
      "learning_rate": 0.003,
      "loss": 4.2776,
      "step": 1734
    },
    {
      "epoch": 0.01735,
      "grad_norm": 0.8320108473030087,
      "learning_rate": 0.003,
      "loss": 4.2551,
      "step": 1735
    },
    {
      "epoch": 0.01736,
      "grad_norm": 0.6879588721711866,
      "learning_rate": 0.003,
      "loss": 4.2513,
      "step": 1736
    },
    {
      "epoch": 0.01737,
      "grad_norm": 0.6180790162276626,
      "learning_rate": 0.003,
      "loss": 4.2782,
      "step": 1737
    },
    {
      "epoch": 0.01738,
      "grad_norm": 0.6570007295297866,
      "learning_rate": 0.003,
      "loss": 4.2689,
      "step": 1738
    },
    {
      "epoch": 0.01739,
      "grad_norm": 0.7234192244460489,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1739
    },
    {
      "epoch": 0.0174,
      "grad_norm": 0.646718379715682,
      "learning_rate": 0.003,
      "loss": 4.2832,
      "step": 1740
    },
    {
      "epoch": 0.01741,
      "grad_norm": 0.7667295546998241,
      "learning_rate": 0.003,
      "loss": 4.2989,
      "step": 1741
    },
    {
      "epoch": 0.01742,
      "grad_norm": 0.6671673777635381,
      "learning_rate": 0.003,
      "loss": 4.2839,
      "step": 1742
    },
    {
      "epoch": 0.01743,
      "grad_norm": 0.6506146600686363,
      "learning_rate": 0.003,
      "loss": 4.2938,
      "step": 1743
    },
    {
      "epoch": 0.01744,
      "grad_norm": 0.6573560282994302,
      "learning_rate": 0.003,
      "loss": 4.2796,
      "step": 1744
    },
    {
      "epoch": 0.01745,
      "grad_norm": 0.640684434992135,
      "learning_rate": 0.003,
      "loss": 4.2436,
      "step": 1745
    },
    {
      "epoch": 0.01746,
      "grad_norm": 0.657793650803233,
      "learning_rate": 0.003,
      "loss": 4.2671,
      "step": 1746
    },
    {
      "epoch": 0.01747,
      "grad_norm": 0.7709605353943576,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1747
    },
    {
      "epoch": 0.01748,
      "grad_norm": 0.8278007400279545,
      "learning_rate": 0.003,
      "loss": 4.2678,
      "step": 1748
    },
    {
      "epoch": 0.01749,
      "grad_norm": 0.7466841943172032,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1749
    },
    {
      "epoch": 0.0175,
      "grad_norm": 0.763703983777172,
      "learning_rate": 0.003,
      "loss": 4.2821,
      "step": 1750
    },
    {
      "epoch": 0.01751,
      "grad_norm": 0.7380887125320881,
      "learning_rate": 0.003,
      "loss": 4.2741,
      "step": 1751
    },
    {
      "epoch": 0.01752,
      "grad_norm": 0.6415685853882744,
      "learning_rate": 0.003,
      "loss": 4.314,
      "step": 1752
    },
    {
      "epoch": 0.01753,
      "grad_norm": 0.6180937585023022,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 1753
    },
    {
      "epoch": 0.01754,
      "grad_norm": 0.6258159102980642,
      "learning_rate": 0.003,
      "loss": 4.291,
      "step": 1754
    },
    {
      "epoch": 0.01755,
      "grad_norm": 0.5258294344568889,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 1755
    },
    {
      "epoch": 0.01756,
      "grad_norm": 0.4224710169236117,
      "learning_rate": 0.003,
      "loss": 4.3014,
      "step": 1756
    },
    {
      "epoch": 0.01757,
      "grad_norm": 0.4609603622580685,
      "learning_rate": 0.003,
      "loss": 4.2799,
      "step": 1757
    },
    {
      "epoch": 0.01758,
      "grad_norm": 0.5088646105086777,
      "learning_rate": 0.003,
      "loss": 4.2854,
      "step": 1758
    },
    {
      "epoch": 0.01759,
      "grad_norm": 0.6116749982391423,
      "learning_rate": 0.003,
      "loss": 4.2559,
      "step": 1759
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.6421141472355831,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1760
    },
    {
      "epoch": 0.01761,
      "grad_norm": 0.683100042759788,
      "learning_rate": 0.003,
      "loss": 4.2664,
      "step": 1761
    },
    {
      "epoch": 0.01762,
      "grad_norm": 0.6507618810640869,
      "learning_rate": 0.003,
      "loss": 4.2868,
      "step": 1762
    },
    {
      "epoch": 0.01763,
      "grad_norm": 0.6268432713329561,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1763
    },
    {
      "epoch": 0.01764,
      "grad_norm": 0.5913417648906798,
      "learning_rate": 0.003,
      "loss": 4.2433,
      "step": 1764
    },
    {
      "epoch": 0.01765,
      "grad_norm": 0.6076223877479996,
      "learning_rate": 0.003,
      "loss": 4.2712,
      "step": 1765
    },
    {
      "epoch": 0.01766,
      "grad_norm": 0.7266026161134381,
      "learning_rate": 0.003,
      "loss": 4.2888,
      "step": 1766
    },
    {
      "epoch": 0.01767,
      "grad_norm": 0.7022487568636244,
      "learning_rate": 0.003,
      "loss": 4.2548,
      "step": 1767
    },
    {
      "epoch": 0.01768,
      "grad_norm": 0.5934615837163861,
      "learning_rate": 0.003,
      "loss": 4.2563,
      "step": 1768
    },
    {
      "epoch": 0.01769,
      "grad_norm": 0.4361100704507323,
      "learning_rate": 0.003,
      "loss": 4.2511,
      "step": 1769
    },
    {
      "epoch": 0.0177,
      "grad_norm": 0.48611120007740843,
      "learning_rate": 0.003,
      "loss": 4.2382,
      "step": 1770
    },
    {
      "epoch": 0.01771,
      "grad_norm": 0.45731585839886957,
      "learning_rate": 0.003,
      "loss": 4.266,
      "step": 1771
    },
    {
      "epoch": 0.01772,
      "grad_norm": 0.4581498365355611,
      "learning_rate": 0.003,
      "loss": 4.2403,
      "step": 1772
    },
    {
      "epoch": 0.01773,
      "grad_norm": 0.3835918309901609,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 1773
    },
    {
      "epoch": 0.01774,
      "grad_norm": 0.4161002686293629,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 1774
    },
    {
      "epoch": 0.01775,
      "grad_norm": 0.48314560217147945,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 1775
    },
    {
      "epoch": 0.01776,
      "grad_norm": 0.550180805967614,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 1776
    },
    {
      "epoch": 0.01777,
      "grad_norm": 0.570528755486745,
      "learning_rate": 0.003,
      "loss": 4.2588,
      "step": 1777
    },
    {
      "epoch": 0.01778,
      "grad_norm": 0.6830277922868914,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1778
    },
    {
      "epoch": 0.01779,
      "grad_norm": 0.8262794189445563,
      "learning_rate": 0.003,
      "loss": 4.281,
      "step": 1779
    },
    {
      "epoch": 0.0178,
      "grad_norm": 0.85849663326723,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1780
    },
    {
      "epoch": 0.01781,
      "grad_norm": 0.7027313657081399,
      "learning_rate": 0.003,
      "loss": 4.2685,
      "step": 1781
    },
    {
      "epoch": 0.01782,
      "grad_norm": 0.5620584115129139,
      "learning_rate": 0.003,
      "loss": 4.2878,
      "step": 1782
    },
    {
      "epoch": 0.01783,
      "grad_norm": 0.5096347371759039,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 1783
    },
    {
      "epoch": 0.01784,
      "grad_norm": 0.5297569056185621,
      "learning_rate": 0.003,
      "loss": 4.2633,
      "step": 1784
    },
    {
      "epoch": 0.01785,
      "grad_norm": 0.6034694109958456,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 1785
    },
    {
      "epoch": 0.01786,
      "grad_norm": 0.6309039198553906,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 1786
    },
    {
      "epoch": 0.01787,
      "grad_norm": 0.6879917544738386,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1787
    },
    {
      "epoch": 0.01788,
      "grad_norm": 0.6818187529366616,
      "learning_rate": 0.003,
      "loss": 4.2789,
      "step": 1788
    },
    {
      "epoch": 0.01789,
      "grad_norm": 0.5930877660332807,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 1789
    },
    {
      "epoch": 0.0179,
      "grad_norm": 0.5104382089041954,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 1790
    },
    {
      "epoch": 0.01791,
      "grad_norm": 0.48343390667073793,
      "learning_rate": 0.003,
      "loss": 4.2706,
      "step": 1791
    },
    {
      "epoch": 0.01792,
      "grad_norm": 0.5502398127361193,
      "learning_rate": 0.003,
      "loss": 4.2742,
      "step": 1792
    },
    {
      "epoch": 0.01793,
      "grad_norm": 0.514623976410524,
      "learning_rate": 0.003,
      "loss": 4.235,
      "step": 1793
    },
    {
      "epoch": 0.01794,
      "grad_norm": 0.5328860529823664,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 1794
    },
    {
      "epoch": 0.01795,
      "grad_norm": 0.5324866930035196,
      "learning_rate": 0.003,
      "loss": 4.2511,
      "step": 1795
    },
    {
      "epoch": 0.01796,
      "grad_norm": 0.4431327306726926,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1796
    },
    {
      "epoch": 0.01797,
      "grad_norm": 0.4163239838007479,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 1797
    },
    {
      "epoch": 0.01798,
      "grad_norm": 0.4680057225647005,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 1798
    },
    {
      "epoch": 0.01799,
      "grad_norm": 0.5257929933308816,
      "learning_rate": 0.003,
      "loss": 4.2569,
      "step": 1799
    },
    {
      "epoch": 0.018,
      "grad_norm": 0.6818533479906342,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1800
    },
    {
      "epoch": 0.01801,
      "grad_norm": 0.8347036722912681,
      "learning_rate": 0.003,
      "loss": 4.2531,
      "step": 1801
    },
    {
      "epoch": 0.01802,
      "grad_norm": 0.8911873404291202,
      "learning_rate": 0.003,
      "loss": 4.2851,
      "step": 1802
    },
    {
      "epoch": 0.01803,
      "grad_norm": 0.8674521299479792,
      "learning_rate": 0.003,
      "loss": 4.2905,
      "step": 1803
    },
    {
      "epoch": 0.01804,
      "grad_norm": 0.8602615225653094,
      "learning_rate": 0.003,
      "loss": 4.2682,
      "step": 1804
    },
    {
      "epoch": 0.01805,
      "grad_norm": 0.8301928044098125,
      "learning_rate": 0.003,
      "loss": 4.2953,
      "step": 1805
    },
    {
      "epoch": 0.01806,
      "grad_norm": 0.7981063012004259,
      "learning_rate": 0.003,
      "loss": 4.2503,
      "step": 1806
    },
    {
      "epoch": 0.01807,
      "grad_norm": 0.8614656914400735,
      "learning_rate": 0.003,
      "loss": 4.2917,
      "step": 1807
    },
    {
      "epoch": 0.01808,
      "grad_norm": 0.7448619096800406,
      "learning_rate": 0.003,
      "loss": 4.2867,
      "step": 1808
    },
    {
      "epoch": 0.01809,
      "grad_norm": 0.6456900848105672,
      "learning_rate": 0.003,
      "loss": 4.2963,
      "step": 1809
    },
    {
      "epoch": 0.0181,
      "grad_norm": 0.5841374638529157,
      "learning_rate": 0.003,
      "loss": 4.2907,
      "step": 1810
    },
    {
      "epoch": 0.01811,
      "grad_norm": 0.5584017956424331,
      "learning_rate": 0.003,
      "loss": 4.2718,
      "step": 1811
    },
    {
      "epoch": 0.01812,
      "grad_norm": 0.5780683839817958,
      "learning_rate": 0.003,
      "loss": 4.2546,
      "step": 1812
    },
    {
      "epoch": 0.01813,
      "grad_norm": 0.5210766747257082,
      "learning_rate": 0.003,
      "loss": 4.298,
      "step": 1813
    },
    {
      "epoch": 0.01814,
      "grad_norm": 0.5061687855504291,
      "learning_rate": 0.003,
      "loss": 4.2617,
      "step": 1814
    },
    {
      "epoch": 0.01815,
      "grad_norm": 0.5646752857396716,
      "learning_rate": 0.003,
      "loss": 4.2497,
      "step": 1815
    },
    {
      "epoch": 0.01816,
      "grad_norm": 0.6329676204068563,
      "learning_rate": 0.003,
      "loss": 4.2779,
      "step": 1816
    },
    {
      "epoch": 0.01817,
      "grad_norm": 0.576517926530056,
      "learning_rate": 0.003,
      "loss": 4.2664,
      "step": 1817
    },
    {
      "epoch": 0.01818,
      "grad_norm": 0.4950065288657199,
      "learning_rate": 0.003,
      "loss": 4.2615,
      "step": 1818
    },
    {
      "epoch": 0.01819,
      "grad_norm": 0.5161367831332595,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 1819
    },
    {
      "epoch": 0.0182,
      "grad_norm": 0.5814618735674658,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 1820
    },
    {
      "epoch": 0.01821,
      "grad_norm": 0.556938739734867,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 1821
    },
    {
      "epoch": 0.01822,
      "grad_norm": 0.5708870221750882,
      "learning_rate": 0.003,
      "loss": 4.2636,
      "step": 1822
    },
    {
      "epoch": 0.01823,
      "grad_norm": 0.5398260260215829,
      "learning_rate": 0.003,
      "loss": 4.2764,
      "step": 1823
    },
    {
      "epoch": 0.01824,
      "grad_norm": 0.5454416785979428,
      "learning_rate": 0.003,
      "loss": 4.2571,
      "step": 1824
    },
    {
      "epoch": 0.01825,
      "grad_norm": 0.6656252297688491,
      "learning_rate": 0.003,
      "loss": 4.2668,
      "step": 1825
    },
    {
      "epoch": 0.01826,
      "grad_norm": 0.9269512928572629,
      "learning_rate": 0.003,
      "loss": 4.289,
      "step": 1826
    },
    {
      "epoch": 0.01827,
      "grad_norm": 1.076217745090645,
      "learning_rate": 0.003,
      "loss": 4.2819,
      "step": 1827
    },
    {
      "epoch": 0.01828,
      "grad_norm": 0.650029561792675,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1828
    },
    {
      "epoch": 0.01829,
      "grad_norm": 0.5962218186152778,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1829
    },
    {
      "epoch": 0.0183,
      "grad_norm": 0.6672932732108263,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 1830
    },
    {
      "epoch": 0.01831,
      "grad_norm": 0.6051971108876394,
      "learning_rate": 0.003,
      "loss": 4.2699,
      "step": 1831
    },
    {
      "epoch": 0.01832,
      "grad_norm": 0.5439007581448924,
      "learning_rate": 0.003,
      "loss": 4.2598,
      "step": 1832
    },
    {
      "epoch": 0.01833,
      "grad_norm": 0.55902144104146,
      "learning_rate": 0.003,
      "loss": 4.2938,
      "step": 1833
    },
    {
      "epoch": 0.01834,
      "grad_norm": 0.5728509198658349,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1834
    },
    {
      "epoch": 0.01835,
      "grad_norm": 0.5340971547409412,
      "learning_rate": 0.003,
      "loss": 4.2588,
      "step": 1835
    },
    {
      "epoch": 0.01836,
      "grad_norm": 0.47778636042203493,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 1836
    },
    {
      "epoch": 0.01837,
      "grad_norm": 0.4158033301420617,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 1837
    },
    {
      "epoch": 0.01838,
      "grad_norm": 0.5117944575544118,
      "learning_rate": 0.003,
      "loss": 4.2549,
      "step": 1838
    },
    {
      "epoch": 0.01839,
      "grad_norm": 0.6067550068103876,
      "learning_rate": 0.003,
      "loss": 4.2791,
      "step": 1839
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.6159872555117356,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1840
    },
    {
      "epoch": 0.01841,
      "grad_norm": 0.5609724495757069,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 1841
    },
    {
      "epoch": 0.01842,
      "grad_norm": 0.5051094597061737,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 1842
    },
    {
      "epoch": 0.01843,
      "grad_norm": 0.5100819064264481,
      "learning_rate": 0.003,
      "loss": 4.228,
      "step": 1843
    },
    {
      "epoch": 0.01844,
      "grad_norm": 0.5582900536993392,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 1844
    },
    {
      "epoch": 0.01845,
      "grad_norm": 0.6111866156562342,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 1845
    },
    {
      "epoch": 0.01846,
      "grad_norm": 0.6619002433974149,
      "learning_rate": 0.003,
      "loss": 4.2608,
      "step": 1846
    },
    {
      "epoch": 0.01847,
      "grad_norm": 1.0200537693834115,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 1847
    },
    {
      "epoch": 0.01848,
      "grad_norm": 1.0513050095404686,
      "learning_rate": 0.003,
      "loss": 4.3125,
      "step": 1848
    },
    {
      "epoch": 0.01849,
      "grad_norm": 0.8927900194204563,
      "learning_rate": 0.003,
      "loss": 4.3063,
      "step": 1849
    },
    {
      "epoch": 0.0185,
      "grad_norm": 0.9502466432757984,
      "learning_rate": 0.003,
      "loss": 4.316,
      "step": 1850
    },
    {
      "epoch": 0.01851,
      "grad_norm": 0.9099370930692962,
      "learning_rate": 0.003,
      "loss": 4.2922,
      "step": 1851
    },
    {
      "epoch": 0.01852,
      "grad_norm": 0.7734254610813497,
      "learning_rate": 0.003,
      "loss": 4.293,
      "step": 1852
    },
    {
      "epoch": 0.01853,
      "grad_norm": 0.7300466041466253,
      "learning_rate": 0.003,
      "loss": 4.2883,
      "step": 1853
    },
    {
      "epoch": 0.01854,
      "grad_norm": 0.7064556599969221,
      "learning_rate": 0.003,
      "loss": 4.319,
      "step": 1854
    },
    {
      "epoch": 0.01855,
      "grad_norm": 0.5941104488816943,
      "learning_rate": 0.003,
      "loss": 4.2971,
      "step": 1855
    },
    {
      "epoch": 0.01856,
      "grad_norm": 0.5639252296572004,
      "learning_rate": 0.003,
      "loss": 4.2765,
      "step": 1856
    },
    {
      "epoch": 0.01857,
      "grad_norm": 0.5497157641923732,
      "learning_rate": 0.003,
      "loss": 4.261,
      "step": 1857
    },
    {
      "epoch": 0.01858,
      "grad_norm": 0.5999940363832145,
      "learning_rate": 0.003,
      "loss": 4.2675,
      "step": 1858
    },
    {
      "epoch": 0.01859,
      "grad_norm": 0.6524431060775852,
      "learning_rate": 0.003,
      "loss": 4.2673,
      "step": 1859
    },
    {
      "epoch": 0.0186,
      "grad_norm": 0.5786076880789475,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 1860
    },
    {
      "epoch": 0.01861,
      "grad_norm": 0.5737732368480609,
      "learning_rate": 0.003,
      "loss": 4.2866,
      "step": 1861
    },
    {
      "epoch": 0.01862,
      "grad_norm": 0.6266977494361998,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 1862
    },
    {
      "epoch": 0.01863,
      "grad_norm": 0.5490443064279302,
      "learning_rate": 0.003,
      "loss": 4.2575,
      "step": 1863
    },
    {
      "epoch": 0.01864,
      "grad_norm": 0.4951546054862347,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1864
    },
    {
      "epoch": 0.01865,
      "grad_norm": 0.4870782718763887,
      "learning_rate": 0.003,
      "loss": 4.2626,
      "step": 1865
    },
    {
      "epoch": 0.01866,
      "grad_norm": 0.46735035399196706,
      "learning_rate": 0.003,
      "loss": 4.2603,
      "step": 1866
    },
    {
      "epoch": 0.01867,
      "grad_norm": 0.4716752041768468,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 1867
    },
    {
      "epoch": 0.01868,
      "grad_norm": 0.44705830429737475,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 1868
    },
    {
      "epoch": 0.01869,
      "grad_norm": 0.4991544420243406,
      "learning_rate": 0.003,
      "loss": 4.2625,
      "step": 1869
    },
    {
      "epoch": 0.0187,
      "grad_norm": 0.4942905986952634,
      "learning_rate": 0.003,
      "loss": 4.2547,
      "step": 1870
    },
    {
      "epoch": 0.01871,
      "grad_norm": 0.5121508182199742,
      "learning_rate": 0.003,
      "loss": 4.2287,
      "step": 1871
    },
    {
      "epoch": 0.01872,
      "grad_norm": 0.5132249094210571,
      "learning_rate": 0.003,
      "loss": 4.2747,
      "step": 1872
    },
    {
      "epoch": 0.01873,
      "grad_norm": 0.45833884531988217,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 1873
    },
    {
      "epoch": 0.01874,
      "grad_norm": 0.4075899028314144,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1874
    },
    {
      "epoch": 0.01875,
      "grad_norm": 0.4250746194260601,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 1875
    },
    {
      "epoch": 0.01876,
      "grad_norm": 0.4231054620615194,
      "learning_rate": 0.003,
      "loss": 4.2577,
      "step": 1876
    },
    {
      "epoch": 0.01877,
      "grad_norm": 0.39967365013750716,
      "learning_rate": 0.003,
      "loss": 4.2584,
      "step": 1877
    },
    {
      "epoch": 0.01878,
      "grad_norm": 0.4152490558819513,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 1878
    },
    {
      "epoch": 0.01879,
      "grad_norm": 0.42538683613055095,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 1879
    },
    {
      "epoch": 0.0188,
      "grad_norm": 0.44463665765346116,
      "learning_rate": 0.003,
      "loss": 4.2318,
      "step": 1880
    },
    {
      "epoch": 0.01881,
      "grad_norm": 0.4351959985696958,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 1881
    },
    {
      "epoch": 0.01882,
      "grad_norm": 0.5024559892307489,
      "learning_rate": 0.003,
      "loss": 4.2542,
      "step": 1882
    },
    {
      "epoch": 0.01883,
      "grad_norm": 0.6063994875754019,
      "learning_rate": 0.003,
      "loss": 4.2323,
      "step": 1883
    },
    {
      "epoch": 0.01884,
      "grad_norm": 0.7335676135390363,
      "learning_rate": 0.003,
      "loss": 4.2381,
      "step": 1884
    },
    {
      "epoch": 0.01885,
      "grad_norm": 0.9057741844151627,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 1885
    },
    {
      "epoch": 0.01886,
      "grad_norm": 1.0977669981210454,
      "learning_rate": 0.003,
      "loss": 4.2863,
      "step": 1886
    },
    {
      "epoch": 0.01887,
      "grad_norm": 0.6506964260225092,
      "learning_rate": 0.003,
      "loss": 4.2622,
      "step": 1887
    },
    {
      "epoch": 0.01888,
      "grad_norm": 0.7018734791941831,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 1888
    },
    {
      "epoch": 0.01889,
      "grad_norm": 0.6726610800126339,
      "learning_rate": 0.003,
      "loss": 4.2655,
      "step": 1889
    },
    {
      "epoch": 0.0189,
      "grad_norm": 0.7428766906603239,
      "learning_rate": 0.003,
      "loss": 4.2512,
      "step": 1890
    },
    {
      "epoch": 0.01891,
      "grad_norm": 0.684421659513949,
      "learning_rate": 0.003,
      "loss": 4.2969,
      "step": 1891
    },
    {
      "epoch": 0.01892,
      "grad_norm": 0.6909566019627704,
      "learning_rate": 0.003,
      "loss": 4.2744,
      "step": 1892
    },
    {
      "epoch": 0.01893,
      "grad_norm": 0.6520197616765968,
      "learning_rate": 0.003,
      "loss": 4.2532,
      "step": 1893
    },
    {
      "epoch": 0.01894,
      "grad_norm": 0.6739359420697347,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 1894
    },
    {
      "epoch": 0.01895,
      "grad_norm": 0.697458296569538,
      "learning_rate": 0.003,
      "loss": 4.2498,
      "step": 1895
    },
    {
      "epoch": 0.01896,
      "grad_norm": 0.5936568619706635,
      "learning_rate": 0.003,
      "loss": 4.2688,
      "step": 1896
    },
    {
      "epoch": 0.01897,
      "grad_norm": 0.5494055764115975,
      "learning_rate": 0.003,
      "loss": 4.2647,
      "step": 1897
    },
    {
      "epoch": 0.01898,
      "grad_norm": 0.5676106882870927,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 1898
    },
    {
      "epoch": 0.01899,
      "grad_norm": 0.6101424369862174,
      "learning_rate": 0.003,
      "loss": 4.2482,
      "step": 1899
    },
    {
      "epoch": 0.019,
      "grad_norm": 0.6545357483725045,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 1900
    },
    {
      "epoch": 0.01901,
      "grad_norm": 0.6221145886969454,
      "learning_rate": 0.003,
      "loss": 4.29,
      "step": 1901
    },
    {
      "epoch": 0.01902,
      "grad_norm": 0.47771631002661874,
      "learning_rate": 0.003,
      "loss": 4.2605,
      "step": 1902
    },
    {
      "epoch": 0.01903,
      "grad_norm": 0.503444994520635,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 1903
    },
    {
      "epoch": 0.01904,
      "grad_norm": 0.5071441864936951,
      "learning_rate": 0.003,
      "loss": 4.2711,
      "step": 1904
    },
    {
      "epoch": 0.01905,
      "grad_norm": 0.5754503645956631,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1905
    },
    {
      "epoch": 0.01906,
      "grad_norm": 0.6654408282834325,
      "learning_rate": 0.003,
      "loss": 4.2694,
      "step": 1906
    },
    {
      "epoch": 0.01907,
      "grad_norm": 0.7906613760805888,
      "learning_rate": 0.003,
      "loss": 4.2662,
      "step": 1907
    },
    {
      "epoch": 0.01908,
      "grad_norm": 0.8850433480932498,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 1908
    },
    {
      "epoch": 0.01909,
      "grad_norm": 0.806456676126049,
      "learning_rate": 0.003,
      "loss": 4.2573,
      "step": 1909
    },
    {
      "epoch": 0.0191,
      "grad_norm": 0.5544806973924623,
      "learning_rate": 0.003,
      "loss": 4.256,
      "step": 1910
    },
    {
      "epoch": 0.01911,
      "grad_norm": 0.6152192065162949,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 1911
    },
    {
      "epoch": 0.01912,
      "grad_norm": 0.5774523882886098,
      "learning_rate": 0.003,
      "loss": 4.2416,
      "step": 1912
    },
    {
      "epoch": 0.01913,
      "grad_norm": 0.5526940825090116,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 1913
    },
    {
      "epoch": 0.01914,
      "grad_norm": 0.5202691625530954,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 1914
    },
    {
      "epoch": 0.01915,
      "grad_norm": 0.4828726686790076,
      "learning_rate": 0.003,
      "loss": 4.257,
      "step": 1915
    },
    {
      "epoch": 0.01916,
      "grad_norm": 0.4684476049653884,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1916
    },
    {
      "epoch": 0.01917,
      "grad_norm": 0.47660708071333646,
      "learning_rate": 0.003,
      "loss": 4.2552,
      "step": 1917
    },
    {
      "epoch": 0.01918,
      "grad_norm": 0.4979902159298128,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 1918
    },
    {
      "epoch": 0.01919,
      "grad_norm": 0.4677465287869337,
      "learning_rate": 0.003,
      "loss": 4.2412,
      "step": 1919
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.42528920687498767,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 1920
    },
    {
      "epoch": 0.01921,
      "grad_norm": 0.4717244787464016,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 1921
    },
    {
      "epoch": 0.01922,
      "grad_norm": 0.4420806610942072,
      "learning_rate": 0.003,
      "loss": 4.2448,
      "step": 1922
    },
    {
      "epoch": 0.01923,
      "grad_norm": 0.42058067854047654,
      "learning_rate": 0.003,
      "loss": 4.2491,
      "step": 1923
    },
    {
      "epoch": 0.01924,
      "grad_norm": 0.3467314160541135,
      "learning_rate": 0.003,
      "loss": 4.2469,
      "step": 1924
    },
    {
      "epoch": 0.01925,
      "grad_norm": 0.3293466925660612,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 1925
    },
    {
      "epoch": 0.01926,
      "grad_norm": 0.41102145828379955,
      "learning_rate": 0.003,
      "loss": 4.2416,
      "step": 1926
    },
    {
      "epoch": 0.01927,
      "grad_norm": 0.720656359870687,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1927
    },
    {
      "epoch": 0.01928,
      "grad_norm": 1.1844212503280394,
      "learning_rate": 0.003,
      "loss": 4.249,
      "step": 1928
    },
    {
      "epoch": 0.01929,
      "grad_norm": 0.9144474201054079,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 1929
    },
    {
      "epoch": 0.0193,
      "grad_norm": 0.8894910892570072,
      "learning_rate": 0.003,
      "loss": 4.2586,
      "step": 1930
    },
    {
      "epoch": 0.01931,
      "grad_norm": 1.0125767880809944,
      "learning_rate": 0.003,
      "loss": 4.2809,
      "step": 1931
    },
    {
      "epoch": 0.01932,
      "grad_norm": 0.6957841268471371,
      "learning_rate": 0.003,
      "loss": 4.2689,
      "step": 1932
    },
    {
      "epoch": 0.01933,
      "grad_norm": 0.967902005641644,
      "learning_rate": 0.003,
      "loss": 4.2719,
      "step": 1933
    },
    {
      "epoch": 0.01934,
      "grad_norm": 0.9288023682490801,
      "learning_rate": 0.003,
      "loss": 4.2872,
      "step": 1934
    },
    {
      "epoch": 0.01935,
      "grad_norm": 0.9414019337954325,
      "learning_rate": 0.003,
      "loss": 4.3063,
      "step": 1935
    },
    {
      "epoch": 0.01936,
      "grad_norm": 0.9694221827159286,
      "learning_rate": 0.003,
      "loss": 4.2962,
      "step": 1936
    },
    {
      "epoch": 0.01937,
      "grad_norm": 1.001938596118208,
      "learning_rate": 0.003,
      "loss": 4.3121,
      "step": 1937
    },
    {
      "epoch": 0.01938,
      "grad_norm": 1.0140385438058088,
      "learning_rate": 0.003,
      "loss": 4.3241,
      "step": 1938
    },
    {
      "epoch": 0.01939,
      "grad_norm": 1.2235452961348985,
      "learning_rate": 0.003,
      "loss": 4.3646,
      "step": 1939
    },
    {
      "epoch": 0.0194,
      "grad_norm": 0.8846695796998231,
      "learning_rate": 0.003,
      "loss": 4.3248,
      "step": 1940
    },
    {
      "epoch": 0.01941,
      "grad_norm": 0.8286594640280003,
      "learning_rate": 0.003,
      "loss": 4.3066,
      "step": 1941
    },
    {
      "epoch": 0.01942,
      "grad_norm": 0.7123881249011045,
      "learning_rate": 0.003,
      "loss": 4.3226,
      "step": 1942
    },
    {
      "epoch": 0.01943,
      "grad_norm": 0.6089638608494183,
      "learning_rate": 0.003,
      "loss": 4.288,
      "step": 1943
    },
    {
      "epoch": 0.01944,
      "grad_norm": 0.5712540744847615,
      "learning_rate": 0.003,
      "loss": 4.3258,
      "step": 1944
    },
    {
      "epoch": 0.01945,
      "grad_norm": 0.5597808479773261,
      "learning_rate": 0.003,
      "loss": 4.2885,
      "step": 1945
    },
    {
      "epoch": 0.01946,
      "grad_norm": 0.5954089244059145,
      "learning_rate": 0.003,
      "loss": 4.2798,
      "step": 1946
    },
    {
      "epoch": 0.01947,
      "grad_norm": 0.5594072113001303,
      "learning_rate": 0.003,
      "loss": 4.2869,
      "step": 1947
    },
    {
      "epoch": 0.01948,
      "grad_norm": 0.47470802243769694,
      "learning_rate": 0.003,
      "loss": 4.3012,
      "step": 1948
    },
    {
      "epoch": 0.01949,
      "grad_norm": 0.5123315729723936,
      "learning_rate": 0.003,
      "loss": 4.2904,
      "step": 1949
    },
    {
      "epoch": 0.0195,
      "grad_norm": 0.5068647256374985,
      "learning_rate": 0.003,
      "loss": 4.2768,
      "step": 1950
    },
    {
      "epoch": 0.01951,
      "grad_norm": 0.4942238546854525,
      "learning_rate": 0.003,
      "loss": 4.2687,
      "step": 1951
    },
    {
      "epoch": 0.01952,
      "grad_norm": 0.4526189118950263,
      "learning_rate": 0.003,
      "loss": 4.271,
      "step": 1952
    },
    {
      "epoch": 0.01953,
      "grad_norm": 0.4370221049383596,
      "learning_rate": 0.003,
      "loss": 4.275,
      "step": 1953
    },
    {
      "epoch": 0.01954,
      "grad_norm": 0.46324265238758683,
      "learning_rate": 0.003,
      "loss": 4.2733,
      "step": 1954
    },
    {
      "epoch": 0.01955,
      "grad_norm": 0.5071666518202474,
      "learning_rate": 0.003,
      "loss": 4.2654,
      "step": 1955
    },
    {
      "epoch": 0.01956,
      "grad_norm": 0.6181690576282608,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1956
    },
    {
      "epoch": 0.01957,
      "grad_norm": 0.5796028080408675,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 1957
    },
    {
      "epoch": 0.01958,
      "grad_norm": 0.41315504837285055,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 1958
    },
    {
      "epoch": 0.01959,
      "grad_norm": 0.3341067438397588,
      "learning_rate": 0.003,
      "loss": 4.2387,
      "step": 1959
    },
    {
      "epoch": 0.0196,
      "grad_norm": 0.3891099796372788,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1960
    },
    {
      "epoch": 0.01961,
      "grad_norm": 0.43000625586768587,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1961
    },
    {
      "epoch": 0.01962,
      "grad_norm": 0.43925709381263844,
      "learning_rate": 0.003,
      "loss": 4.232,
      "step": 1962
    },
    {
      "epoch": 0.01963,
      "grad_norm": 0.42981173506186865,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 1963
    },
    {
      "epoch": 0.01964,
      "grad_norm": 0.4097996230629943,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 1964
    },
    {
      "epoch": 0.01965,
      "grad_norm": 0.3678222300164623,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 1965
    },
    {
      "epoch": 0.01966,
      "grad_norm": 0.3826622250643601,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 1966
    },
    {
      "epoch": 0.01967,
      "grad_norm": 0.40760983332927614,
      "learning_rate": 0.003,
      "loss": 4.2115,
      "step": 1967
    },
    {
      "epoch": 0.01968,
      "grad_norm": 0.47242767064200136,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 1968
    },
    {
      "epoch": 0.01969,
      "grad_norm": 0.588742886193704,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 1969
    },
    {
      "epoch": 0.0197,
      "grad_norm": 0.7240833755896967,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 1970
    },
    {
      "epoch": 0.01971,
      "grad_norm": 0.7681447765560055,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 1971
    },
    {
      "epoch": 0.01972,
      "grad_norm": 0.6596049230522953,
      "learning_rate": 0.003,
      "loss": 4.2329,
      "step": 1972
    },
    {
      "epoch": 0.01973,
      "grad_norm": 0.5971490272128623,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 1973
    },
    {
      "epoch": 0.01974,
      "grad_norm": 0.5788210723943882,
      "learning_rate": 0.003,
      "loss": 4.2527,
      "step": 1974
    },
    {
      "epoch": 0.01975,
      "grad_norm": 0.5716517631988168,
      "learning_rate": 0.003,
      "loss": 4.2434,
      "step": 1975
    },
    {
      "epoch": 0.01976,
      "grad_norm": 0.5714647695261786,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 1976
    },
    {
      "epoch": 0.01977,
      "grad_norm": 0.4672772924879757,
      "learning_rate": 0.003,
      "loss": 4.2626,
      "step": 1977
    },
    {
      "epoch": 0.01978,
      "grad_norm": 0.4919465044055173,
      "learning_rate": 0.003,
      "loss": 4.2624,
      "step": 1978
    },
    {
      "epoch": 0.01979,
      "grad_norm": 0.5361240818371313,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 1979
    },
    {
      "epoch": 0.0198,
      "grad_norm": 0.6323418711078611,
      "learning_rate": 0.003,
      "loss": 4.2334,
      "step": 1980
    },
    {
      "epoch": 0.01981,
      "grad_norm": 0.6568538387899918,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 1981
    },
    {
      "epoch": 0.01982,
      "grad_norm": 0.5202437330212393,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 1982
    },
    {
      "epoch": 0.01983,
      "grad_norm": 0.50161503607477,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 1983
    },
    {
      "epoch": 0.01984,
      "grad_norm": 0.6088039267450153,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 1984
    },
    {
      "epoch": 0.01985,
      "grad_norm": 0.8039703319398841,
      "learning_rate": 0.003,
      "loss": 4.247,
      "step": 1985
    },
    {
      "epoch": 0.01986,
      "grad_norm": 1.0279100079228762,
      "learning_rate": 0.003,
      "loss": 4.2896,
      "step": 1986
    },
    {
      "epoch": 0.01987,
      "grad_norm": 0.8414880380409095,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 1987
    },
    {
      "epoch": 0.01988,
      "grad_norm": 0.8182097035427517,
      "learning_rate": 0.003,
      "loss": 4.2565,
      "step": 1988
    },
    {
      "epoch": 0.01989,
      "grad_norm": 0.7584478804000142,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 1989
    },
    {
      "epoch": 0.0199,
      "grad_norm": 0.6298670596846694,
      "learning_rate": 0.003,
      "loss": 4.2542,
      "step": 1990
    },
    {
      "epoch": 0.01991,
      "grad_norm": 0.5918495849768016,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 1991
    },
    {
      "epoch": 0.01992,
      "grad_norm": 0.6341092964743213,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1992
    },
    {
      "epoch": 0.01993,
      "grad_norm": 0.690814419492087,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 1993
    },
    {
      "epoch": 0.01994,
      "grad_norm": 0.738327319887457,
      "learning_rate": 0.003,
      "loss": 4.2804,
      "step": 1994
    },
    {
      "epoch": 0.01995,
      "grad_norm": 0.9314757846548688,
      "learning_rate": 0.003,
      "loss": 4.284,
      "step": 1995
    },
    {
      "epoch": 0.01996,
      "grad_norm": 0.9849485592924706,
      "learning_rate": 0.003,
      "loss": 4.2595,
      "step": 1996
    },
    {
      "epoch": 0.01997,
      "grad_norm": 0.8581365774503161,
      "learning_rate": 0.003,
      "loss": 4.2889,
      "step": 1997
    },
    {
      "epoch": 0.01998,
      "grad_norm": 0.824906277718389,
      "learning_rate": 0.003,
      "loss": 4.2717,
      "step": 1998
    },
    {
      "epoch": 0.01999,
      "grad_norm": 0.8642378679098298,
      "learning_rate": 0.003,
      "loss": 4.274,
      "step": 1999
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7257405855967545,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 2000
    },
    {
      "epoch": 0.02001,
      "grad_norm": 0.7164159344618267,
      "learning_rate": 0.003,
      "loss": 4.2892,
      "step": 2001
    },
    {
      "epoch": 0.02002,
      "grad_norm": 0.7067957991027757,
      "learning_rate": 0.003,
      "loss": 4.2811,
      "step": 2002
    },
    {
      "epoch": 0.02003,
      "grad_norm": 0.7144493226672485,
      "learning_rate": 0.003,
      "loss": 4.2802,
      "step": 2003
    },
    {
      "epoch": 0.02004,
      "grad_norm": 0.8392479210866416,
      "learning_rate": 0.003,
      "loss": 4.2749,
      "step": 2004
    },
    {
      "epoch": 0.02005,
      "grad_norm": 0.7896031315305061,
      "learning_rate": 0.003,
      "loss": 4.2792,
      "step": 2005
    },
    {
      "epoch": 0.02006,
      "grad_norm": 0.7812166363163485,
      "learning_rate": 0.003,
      "loss": 4.2791,
      "step": 2006
    },
    {
      "epoch": 0.02007,
      "grad_norm": 0.720616320355632,
      "learning_rate": 0.003,
      "loss": 4.2906,
      "step": 2007
    },
    {
      "epoch": 0.02008,
      "grad_norm": 0.7275037377798674,
      "learning_rate": 0.003,
      "loss": 4.2831,
      "step": 2008
    },
    {
      "epoch": 0.02009,
      "grad_norm": 0.8040540578216816,
      "learning_rate": 0.003,
      "loss": 4.2757,
      "step": 2009
    },
    {
      "epoch": 0.0201,
      "grad_norm": 0.8028591997627386,
      "learning_rate": 0.003,
      "loss": 4.2928,
      "step": 2010
    },
    {
      "epoch": 0.02011,
      "grad_norm": 0.7203630829267829,
      "learning_rate": 0.003,
      "loss": 4.2738,
      "step": 2011
    },
    {
      "epoch": 0.02012,
      "grad_norm": 0.6305312780193182,
      "learning_rate": 0.003,
      "loss": 4.2707,
      "step": 2012
    },
    {
      "epoch": 0.02013,
      "grad_norm": 0.4989290466950007,
      "learning_rate": 0.003,
      "loss": 4.2708,
      "step": 2013
    },
    {
      "epoch": 0.02014,
      "grad_norm": 0.5235253503408636,
      "learning_rate": 0.003,
      "loss": 4.2549,
      "step": 2014
    },
    {
      "epoch": 0.02015,
      "grad_norm": 0.49148432915624135,
      "learning_rate": 0.003,
      "loss": 4.2456,
      "step": 2015
    },
    {
      "epoch": 0.02016,
      "grad_norm": 0.5171965972543717,
      "learning_rate": 0.003,
      "loss": 4.2537,
      "step": 2016
    },
    {
      "epoch": 0.02017,
      "grad_norm": 0.48975410612827497,
      "learning_rate": 0.003,
      "loss": 4.2428,
      "step": 2017
    },
    {
      "epoch": 0.02018,
      "grad_norm": 0.532740553989744,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 2018
    },
    {
      "epoch": 0.02019,
      "grad_norm": 0.48750173027142363,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 2019
    },
    {
      "epoch": 0.0202,
      "grad_norm": 0.4089724033791691,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2020
    },
    {
      "epoch": 0.02021,
      "grad_norm": 0.4207487504513156,
      "learning_rate": 0.003,
      "loss": 4.265,
      "step": 2021
    },
    {
      "epoch": 0.02022,
      "grad_norm": 0.4057461757528402,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2022
    },
    {
      "epoch": 0.02023,
      "grad_norm": 0.3580229208354821,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2023
    },
    {
      "epoch": 0.02024,
      "grad_norm": 0.3567498277820771,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2024
    },
    {
      "epoch": 0.02025,
      "grad_norm": 0.43295354825834576,
      "learning_rate": 0.003,
      "loss": 4.2663,
      "step": 2025
    },
    {
      "epoch": 0.02026,
      "grad_norm": 0.4300502575103401,
      "learning_rate": 0.003,
      "loss": 4.2452,
      "step": 2026
    },
    {
      "epoch": 0.02027,
      "grad_norm": 0.5507801143322275,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 2027
    },
    {
      "epoch": 0.02028,
      "grad_norm": 0.6603955599139062,
      "learning_rate": 0.003,
      "loss": 4.2319,
      "step": 2028
    },
    {
      "epoch": 0.02029,
      "grad_norm": 0.7462999560296353,
      "learning_rate": 0.003,
      "loss": 4.2391,
      "step": 2029
    },
    {
      "epoch": 0.0203,
      "grad_norm": 0.6312368220072189,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2030
    },
    {
      "epoch": 0.02031,
      "grad_norm": 0.5002880523213,
      "learning_rate": 0.003,
      "loss": 4.2308,
      "step": 2031
    },
    {
      "epoch": 0.02032,
      "grad_norm": 0.7101695088498386,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2032
    },
    {
      "epoch": 0.02033,
      "grad_norm": 0.7026230897172011,
      "learning_rate": 0.003,
      "loss": 4.2596,
      "step": 2033
    },
    {
      "epoch": 0.02034,
      "grad_norm": 0.4672671459141678,
      "learning_rate": 0.003,
      "loss": 4.2396,
      "step": 2034
    },
    {
      "epoch": 0.02035,
      "grad_norm": 0.5167562590378834,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2035
    },
    {
      "epoch": 0.02036,
      "grad_norm": 0.5204324719963792,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2036
    },
    {
      "epoch": 0.02037,
      "grad_norm": 0.44171452211516693,
      "learning_rate": 0.003,
      "loss": 4.2379,
      "step": 2037
    },
    {
      "epoch": 0.02038,
      "grad_norm": 0.443929190404615,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2038
    },
    {
      "epoch": 0.02039,
      "grad_norm": 0.48540831581824334,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2039
    },
    {
      "epoch": 0.0204,
      "grad_norm": 0.5410457621233177,
      "learning_rate": 0.003,
      "loss": 4.236,
      "step": 2040
    },
    {
      "epoch": 0.02041,
      "grad_norm": 0.5544368623280378,
      "learning_rate": 0.003,
      "loss": 4.1991,
      "step": 2041
    },
    {
      "epoch": 0.02042,
      "grad_norm": 0.5894517852974022,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2042
    },
    {
      "epoch": 0.02043,
      "grad_norm": 0.6318863235101739,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 2043
    },
    {
      "epoch": 0.02044,
      "grad_norm": 0.6044374925441985,
      "learning_rate": 0.003,
      "loss": 4.2207,
      "step": 2044
    },
    {
      "epoch": 0.02045,
      "grad_norm": 0.5638648488316749,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 2045
    },
    {
      "epoch": 0.02046,
      "grad_norm": 0.514843677649273,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2046
    },
    {
      "epoch": 0.02047,
      "grad_norm": 0.4850628116632027,
      "learning_rate": 0.003,
      "loss": 4.2466,
      "step": 2047
    },
    {
      "epoch": 0.02048,
      "grad_norm": 0.5067500543661038,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2048
    },
    {
      "epoch": 0.02049,
      "grad_norm": 0.5005632559023494,
      "learning_rate": 0.003,
      "loss": 4.2517,
      "step": 2049
    },
    {
      "epoch": 0.0205,
      "grad_norm": 0.5338842758485439,
      "learning_rate": 0.003,
      "loss": 4.2016,
      "step": 2050
    },
    {
      "epoch": 0.02051,
      "grad_norm": 0.5673366342029961,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 2051
    },
    {
      "epoch": 0.02052,
      "grad_norm": 0.6208513233991159,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 2052
    },
    {
      "epoch": 0.02053,
      "grad_norm": 0.7925288751630323,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2053
    },
    {
      "epoch": 0.02054,
      "grad_norm": 0.9689478763760375,
      "learning_rate": 0.003,
      "loss": 4.2703,
      "step": 2054
    },
    {
      "epoch": 0.02055,
      "grad_norm": 0.9174532624746199,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2055
    },
    {
      "epoch": 0.02056,
      "grad_norm": 0.7759474219779441,
      "learning_rate": 0.003,
      "loss": 4.2289,
      "step": 2056
    },
    {
      "epoch": 0.02057,
      "grad_norm": 0.6855719086766776,
      "learning_rate": 0.003,
      "loss": 4.2628,
      "step": 2057
    },
    {
      "epoch": 0.02058,
      "grad_norm": 0.664344253711328,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2058
    },
    {
      "epoch": 0.02059,
      "grad_norm": 0.7126024175545171,
      "learning_rate": 0.003,
      "loss": 4.2205,
      "step": 2059
    },
    {
      "epoch": 0.0206,
      "grad_norm": 0.8304297596763678,
      "learning_rate": 0.003,
      "loss": 4.2739,
      "step": 2060
    },
    {
      "epoch": 0.02061,
      "grad_norm": 0.8055338565488124,
      "learning_rate": 0.003,
      "loss": 4.2502,
      "step": 2061
    },
    {
      "epoch": 0.02062,
      "grad_norm": 0.7656045427907143,
      "learning_rate": 0.003,
      "loss": 4.2474,
      "step": 2062
    },
    {
      "epoch": 0.02063,
      "grad_norm": 0.7423249019964357,
      "learning_rate": 0.003,
      "loss": 4.2458,
      "step": 2063
    },
    {
      "epoch": 0.02064,
      "grad_norm": 0.7236715566519402,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2064
    },
    {
      "epoch": 0.02065,
      "grad_norm": 0.7531415944255598,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2065
    },
    {
      "epoch": 0.02066,
      "grad_norm": 0.6867969137582566,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2066
    },
    {
      "epoch": 0.02067,
      "grad_norm": 0.5938985343647097,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2067
    },
    {
      "epoch": 0.02068,
      "grad_norm": 0.4863087277096077,
      "learning_rate": 0.003,
      "loss": 4.2209,
      "step": 2068
    },
    {
      "epoch": 0.02069,
      "grad_norm": 0.4635605849081945,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2069
    },
    {
      "epoch": 0.0207,
      "grad_norm": 0.5457323036329618,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2070
    },
    {
      "epoch": 0.02071,
      "grad_norm": 0.5432685692905308,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2071
    },
    {
      "epoch": 0.02072,
      "grad_norm": 0.5685190700350947,
      "learning_rate": 0.003,
      "loss": 4.2519,
      "step": 2072
    },
    {
      "epoch": 0.02073,
      "grad_norm": 0.535137018993922,
      "learning_rate": 0.003,
      "loss": 4.2291,
      "step": 2073
    },
    {
      "epoch": 0.02074,
      "grad_norm": 0.5467639426283901,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2074
    },
    {
      "epoch": 0.02075,
      "grad_norm": 0.5451184946505037,
      "learning_rate": 0.003,
      "loss": 4.2644,
      "step": 2075
    },
    {
      "epoch": 0.02076,
      "grad_norm": 0.4600936162495081,
      "learning_rate": 0.003,
      "loss": 4.2518,
      "step": 2076
    },
    {
      "epoch": 0.02077,
      "grad_norm": 0.465650503497137,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2077
    },
    {
      "epoch": 0.02078,
      "grad_norm": 0.5411332311026555,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2078
    },
    {
      "epoch": 0.02079,
      "grad_norm": 0.5765467214387223,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 2079
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.5738914259466256,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2080
    },
    {
      "epoch": 0.02081,
      "grad_norm": 0.5810191862607845,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2081
    },
    {
      "epoch": 0.02082,
      "grad_norm": 0.5527464904961061,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2082
    },
    {
      "epoch": 0.02083,
      "grad_norm": 0.5331974469414995,
      "learning_rate": 0.003,
      "loss": 4.2351,
      "step": 2083
    },
    {
      "epoch": 0.02084,
      "grad_norm": 0.5880130787094308,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2084
    },
    {
      "epoch": 0.02085,
      "grad_norm": 0.6630695440600266,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2085
    },
    {
      "epoch": 0.02086,
      "grad_norm": 0.6095683840332357,
      "learning_rate": 0.003,
      "loss": 4.2148,
      "step": 2086
    },
    {
      "epoch": 0.02087,
      "grad_norm": 0.6289181823836659,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2087
    },
    {
      "epoch": 0.02088,
      "grad_norm": 0.6691073473732546,
      "learning_rate": 0.003,
      "loss": 4.2514,
      "step": 2088
    },
    {
      "epoch": 0.02089,
      "grad_norm": 0.757943428461077,
      "learning_rate": 0.003,
      "loss": 4.2252,
      "step": 2089
    },
    {
      "epoch": 0.0209,
      "grad_norm": 0.6942072774820578,
      "learning_rate": 0.003,
      "loss": 4.2464,
      "step": 2090
    },
    {
      "epoch": 0.02091,
      "grad_norm": 0.5265412592049787,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 2091
    },
    {
      "epoch": 0.02092,
      "grad_norm": 0.6385734634683451,
      "learning_rate": 0.003,
      "loss": 4.2259,
      "step": 2092
    },
    {
      "epoch": 0.02093,
      "grad_norm": 0.7053440248117775,
      "learning_rate": 0.003,
      "loss": 4.1968,
      "step": 2093
    },
    {
      "epoch": 0.02094,
      "grad_norm": 0.696978044878893,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2094
    },
    {
      "epoch": 0.02095,
      "grad_norm": 0.6272978652521859,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2095
    },
    {
      "epoch": 0.02096,
      "grad_norm": 0.643764393951801,
      "learning_rate": 0.003,
      "loss": 4.2395,
      "step": 2096
    },
    {
      "epoch": 0.02097,
      "grad_norm": 0.6586650506136257,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2097
    },
    {
      "epoch": 0.02098,
      "grad_norm": 0.5374026111379894,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 2098
    },
    {
      "epoch": 0.02099,
      "grad_norm": 0.463394138236571,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 2099
    },
    {
      "epoch": 0.021,
      "grad_norm": 0.43247948811237297,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 2100
    },
    {
      "epoch": 0.02101,
      "grad_norm": 0.4029126824700864,
      "learning_rate": 0.003,
      "loss": 4.2421,
      "step": 2101
    },
    {
      "epoch": 0.02102,
      "grad_norm": 0.4076539308302002,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2102
    },
    {
      "epoch": 0.02103,
      "grad_norm": 0.48572239313264476,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2103
    },
    {
      "epoch": 0.02104,
      "grad_norm": 0.6011024498535701,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2104
    },
    {
      "epoch": 0.02105,
      "grad_norm": 0.7452127300170479,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2105
    },
    {
      "epoch": 0.02106,
      "grad_norm": 0.8682450979969631,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2106
    },
    {
      "epoch": 0.02107,
      "grad_norm": 0.8835439612331999,
      "learning_rate": 0.003,
      "loss": 4.2496,
      "step": 2107
    },
    {
      "epoch": 0.02108,
      "grad_norm": 0.7854943005617361,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2108
    },
    {
      "epoch": 0.02109,
      "grad_norm": 0.7880515723278634,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2109
    },
    {
      "epoch": 0.0211,
      "grad_norm": 0.7374506931044654,
      "learning_rate": 0.003,
      "loss": 4.234,
      "step": 2110
    },
    {
      "epoch": 0.02111,
      "grad_norm": 0.8090504896236194,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2111
    },
    {
      "epoch": 0.02112,
      "grad_norm": 0.7918025257999607,
      "learning_rate": 0.003,
      "loss": 4.2488,
      "step": 2112
    },
    {
      "epoch": 0.02113,
      "grad_norm": 0.8131758042516601,
      "learning_rate": 0.003,
      "loss": 4.2484,
      "step": 2113
    },
    {
      "epoch": 0.02114,
      "grad_norm": 0.7527860897593042,
      "learning_rate": 0.003,
      "loss": 4.2365,
      "step": 2114
    },
    {
      "epoch": 0.02115,
      "grad_norm": 0.7359409139929027,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 2115
    },
    {
      "epoch": 0.02116,
      "grad_norm": 0.6808675729987482,
      "learning_rate": 0.003,
      "loss": 4.2544,
      "step": 2116
    },
    {
      "epoch": 0.02117,
      "grad_norm": 0.5755888828726318,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2117
    },
    {
      "epoch": 0.02118,
      "grad_norm": 0.5921868695179046,
      "learning_rate": 0.003,
      "loss": 4.2397,
      "step": 2118
    },
    {
      "epoch": 0.02119,
      "grad_norm": 0.5455603705053529,
      "learning_rate": 0.003,
      "loss": 4.2459,
      "step": 2119
    },
    {
      "epoch": 0.0212,
      "grad_norm": 0.627867188843003,
      "learning_rate": 0.003,
      "loss": 4.2226,
      "step": 2120
    },
    {
      "epoch": 0.02121,
      "grad_norm": 0.611945450027123,
      "learning_rate": 0.003,
      "loss": 4.2171,
      "step": 2121
    },
    {
      "epoch": 0.02122,
      "grad_norm": 0.568436827457235,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2122
    },
    {
      "epoch": 0.02123,
      "grad_norm": 0.5350506674860982,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2123
    },
    {
      "epoch": 0.02124,
      "grad_norm": 0.5325059896638067,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 2124
    },
    {
      "epoch": 0.02125,
      "grad_norm": 0.55145580695339,
      "learning_rate": 0.003,
      "loss": 4.2384,
      "step": 2125
    },
    {
      "epoch": 0.02126,
      "grad_norm": 0.6190295521827949,
      "learning_rate": 0.003,
      "loss": 4.2377,
      "step": 2126
    },
    {
      "epoch": 0.02127,
      "grad_norm": 0.5815814384311068,
      "learning_rate": 0.003,
      "loss": 4.244,
      "step": 2127
    },
    {
      "epoch": 0.02128,
      "grad_norm": 0.624207657173836,
      "learning_rate": 0.003,
      "loss": 4.2362,
      "step": 2128
    },
    {
      "epoch": 0.02129,
      "grad_norm": 0.637928608874603,
      "learning_rate": 0.003,
      "loss": 4.2364,
      "step": 2129
    },
    {
      "epoch": 0.0213,
      "grad_norm": 0.5372631019114236,
      "learning_rate": 0.003,
      "loss": 4.2069,
      "step": 2130
    },
    {
      "epoch": 0.02131,
      "grad_norm": 0.5117120142897393,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2131
    },
    {
      "epoch": 0.02132,
      "grad_norm": 0.5272503917214815,
      "learning_rate": 0.003,
      "loss": 4.2429,
      "step": 2132
    },
    {
      "epoch": 0.02133,
      "grad_norm": 0.4922684782001162,
      "learning_rate": 0.003,
      "loss": 4.2337,
      "step": 2133
    },
    {
      "epoch": 0.02134,
      "grad_norm": 0.4711594178288037,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2134
    },
    {
      "epoch": 0.02135,
      "grad_norm": 0.5354798117430423,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2135
    },
    {
      "epoch": 0.02136,
      "grad_norm": 0.5972846763679789,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2136
    },
    {
      "epoch": 0.02137,
      "grad_norm": 0.6875217013113815,
      "learning_rate": 0.003,
      "loss": 4.2468,
      "step": 2137
    },
    {
      "epoch": 0.02138,
      "grad_norm": 0.9636076155805002,
      "learning_rate": 0.003,
      "loss": 4.273,
      "step": 2138
    },
    {
      "epoch": 0.02139,
      "grad_norm": 1.148302508200364,
      "learning_rate": 0.003,
      "loss": 4.2769,
      "step": 2139
    },
    {
      "epoch": 0.0214,
      "grad_norm": 0.7025550992151242,
      "learning_rate": 0.003,
      "loss": 4.2515,
      "step": 2140
    },
    {
      "epoch": 0.02141,
      "grad_norm": 0.7381063691718428,
      "learning_rate": 0.003,
      "loss": 4.2408,
      "step": 2141
    },
    {
      "epoch": 0.02142,
      "grad_norm": 0.6164956740364308,
      "learning_rate": 0.003,
      "loss": 4.2223,
      "step": 2142
    },
    {
      "epoch": 0.02143,
      "grad_norm": 0.6846147414300063,
      "learning_rate": 0.003,
      "loss": 4.2414,
      "step": 2143
    },
    {
      "epoch": 0.02144,
      "grad_norm": 0.7043565504624957,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2144
    },
    {
      "epoch": 0.02145,
      "grad_norm": 0.6306040325927637,
      "learning_rate": 0.003,
      "loss": 4.2262,
      "step": 2145
    },
    {
      "epoch": 0.02146,
      "grad_norm": 0.5782503991731893,
      "learning_rate": 0.003,
      "loss": 4.2372,
      "step": 2146
    },
    {
      "epoch": 0.02147,
      "grad_norm": 0.5354606170184357,
      "learning_rate": 0.003,
      "loss": 4.2548,
      "step": 2147
    },
    {
      "epoch": 0.02148,
      "grad_norm": 0.4707102689450845,
      "learning_rate": 0.003,
      "loss": 4.2629,
      "step": 2148
    },
    {
      "epoch": 0.02149,
      "grad_norm": 0.5098358773254507,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2149
    },
    {
      "epoch": 0.0215,
      "grad_norm": 0.5420555838700287,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 2150
    },
    {
      "epoch": 0.02151,
      "grad_norm": 0.5090710234825387,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2151
    },
    {
      "epoch": 0.02152,
      "grad_norm": 0.5210838533104809,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2152
    },
    {
      "epoch": 0.02153,
      "grad_norm": 0.4582952267704145,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2153
    },
    {
      "epoch": 0.02154,
      "grad_norm": 0.5288143265950994,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2154
    },
    {
      "epoch": 0.02155,
      "grad_norm": 0.568233364630795,
      "learning_rate": 0.003,
      "loss": 4.2465,
      "step": 2155
    },
    {
      "epoch": 0.02156,
      "grad_norm": 0.6709041798606031,
      "learning_rate": 0.003,
      "loss": 4.2356,
      "step": 2156
    },
    {
      "epoch": 0.02157,
      "grad_norm": 0.837609334642183,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2157
    },
    {
      "epoch": 0.02158,
      "grad_norm": 0.9025483102003374,
      "learning_rate": 0.003,
      "loss": 4.2406,
      "step": 2158
    },
    {
      "epoch": 0.02159,
      "grad_norm": 0.6757453813747746,
      "learning_rate": 0.003,
      "loss": 4.2475,
      "step": 2159
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.6195925236969745,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2160
    },
    {
      "epoch": 0.02161,
      "grad_norm": 0.6541705747885649,
      "learning_rate": 0.003,
      "loss": 4.2346,
      "step": 2161
    },
    {
      "epoch": 0.02162,
      "grad_norm": 0.713506704868693,
      "learning_rate": 0.003,
      "loss": 4.2445,
      "step": 2162
    },
    {
      "epoch": 0.02163,
      "grad_norm": 0.7618153376621027,
      "learning_rate": 0.003,
      "loss": 4.2619,
      "step": 2163
    },
    {
      "epoch": 0.02164,
      "grad_norm": 0.7193438764197988,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2164
    },
    {
      "epoch": 0.02165,
      "grad_norm": 0.7614872081882262,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2165
    },
    {
      "epoch": 0.02166,
      "grad_norm": 0.7583750488187075,
      "learning_rate": 0.003,
      "loss": 4.2616,
      "step": 2166
    },
    {
      "epoch": 0.02167,
      "grad_norm": 0.5886983647176827,
      "learning_rate": 0.003,
      "loss": 4.2327,
      "step": 2167
    },
    {
      "epoch": 0.02168,
      "grad_norm": 0.5673249719944714,
      "learning_rate": 0.003,
      "loss": 4.229,
      "step": 2168
    },
    {
      "epoch": 0.02169,
      "grad_norm": 0.539350720737554,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2169
    },
    {
      "epoch": 0.0217,
      "grad_norm": 0.505175388578022,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2170
    },
    {
      "epoch": 0.02171,
      "grad_norm": 0.523933575801358,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2171
    },
    {
      "epoch": 0.02172,
      "grad_norm": 0.4755293785333703,
      "learning_rate": 0.003,
      "loss": 4.2128,
      "step": 2172
    },
    {
      "epoch": 0.02173,
      "grad_norm": 0.5360790560792152,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2173
    },
    {
      "epoch": 0.02174,
      "grad_norm": 0.5613507175665455,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2174
    },
    {
      "epoch": 0.02175,
      "grad_norm": 0.7273309266731635,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2175
    },
    {
      "epoch": 0.02176,
      "grad_norm": 0.8055224057820037,
      "learning_rate": 0.003,
      "loss": 4.2357,
      "step": 2176
    },
    {
      "epoch": 0.02177,
      "grad_norm": 0.7934658981354097,
      "learning_rate": 0.003,
      "loss": 4.2405,
      "step": 2177
    },
    {
      "epoch": 0.02178,
      "grad_norm": 0.680464890242631,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2178
    },
    {
      "epoch": 0.02179,
      "grad_norm": 0.6249756747331623,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2179
    },
    {
      "epoch": 0.0218,
      "grad_norm": 0.6384620261389609,
      "learning_rate": 0.003,
      "loss": 4.245,
      "step": 2180
    },
    {
      "epoch": 0.02181,
      "grad_norm": 0.6338413006913513,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2181
    },
    {
      "epoch": 0.02182,
      "grad_norm": 0.602201817766222,
      "learning_rate": 0.003,
      "loss": 4.2311,
      "step": 2182
    },
    {
      "epoch": 0.02183,
      "grad_norm": 0.6159529942047347,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2183
    },
    {
      "epoch": 0.02184,
      "grad_norm": 0.654749250757439,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2184
    },
    {
      "epoch": 0.02185,
      "grad_norm": 0.798418167230075,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2185
    },
    {
      "epoch": 0.02186,
      "grad_norm": 0.9118929371333987,
      "learning_rate": 0.003,
      "loss": 4.2486,
      "step": 2186
    },
    {
      "epoch": 0.02187,
      "grad_norm": 0.8282661348302262,
      "learning_rate": 0.003,
      "loss": 4.2454,
      "step": 2187
    },
    {
      "epoch": 0.02188,
      "grad_norm": 0.6832425004401769,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2188
    },
    {
      "epoch": 0.02189,
      "grad_norm": 0.6446063652482951,
      "learning_rate": 0.003,
      "loss": 4.2521,
      "step": 2189
    },
    {
      "epoch": 0.0219,
      "grad_norm": 0.5887684645304796,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2190
    },
    {
      "epoch": 0.02191,
      "grad_norm": 0.5977971628739048,
      "learning_rate": 0.003,
      "loss": 4.2321,
      "step": 2191
    },
    {
      "epoch": 0.02192,
      "grad_norm": 0.4949212999134859,
      "learning_rate": 0.003,
      "loss": 4.2523,
      "step": 2192
    },
    {
      "epoch": 0.02193,
      "grad_norm": 0.5003189031862624,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2193
    },
    {
      "epoch": 0.02194,
      "grad_norm": 0.48226309241618065,
      "learning_rate": 0.003,
      "loss": 4.2678,
      "step": 2194
    },
    {
      "epoch": 0.02195,
      "grad_norm": 0.544272590519486,
      "learning_rate": 0.003,
      "loss": 4.2332,
      "step": 2195
    },
    {
      "epoch": 0.02196,
      "grad_norm": 0.6071662200973025,
      "learning_rate": 0.003,
      "loss": 4.2417,
      "step": 2196
    },
    {
      "epoch": 0.02197,
      "grad_norm": 0.7366010094026804,
      "learning_rate": 0.003,
      "loss": 4.2374,
      "step": 2197
    },
    {
      "epoch": 0.02198,
      "grad_norm": 0.7102801228364903,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2198
    },
    {
      "epoch": 0.02199,
      "grad_norm": 0.6007866089006326,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2199
    },
    {
      "epoch": 0.022,
      "grad_norm": 0.5670114788301772,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2200
    },
    {
      "epoch": 0.02201,
      "grad_norm": 0.4763785124586338,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2201
    },
    {
      "epoch": 0.02202,
      "grad_norm": 0.42903876892090725,
      "learning_rate": 0.003,
      "loss": 4.2163,
      "step": 2202
    },
    {
      "epoch": 0.02203,
      "grad_norm": 0.4156329769757173,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 2203
    },
    {
      "epoch": 0.02204,
      "grad_norm": 0.4681197264286167,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2204
    },
    {
      "epoch": 0.02205,
      "grad_norm": 0.5059629089964722,
      "learning_rate": 0.003,
      "loss": 4.2388,
      "step": 2205
    },
    {
      "epoch": 0.02206,
      "grad_norm": 0.6309485655238869,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2206
    },
    {
      "epoch": 0.02207,
      "grad_norm": 0.6824022074223641,
      "learning_rate": 0.003,
      "loss": 4.2243,
      "step": 2207
    },
    {
      "epoch": 0.02208,
      "grad_norm": 0.6950242305361404,
      "learning_rate": 0.003,
      "loss": 4.2081,
      "step": 2208
    },
    {
      "epoch": 0.02209,
      "grad_norm": 0.8264786459594594,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2209
    },
    {
      "epoch": 0.0221,
      "grad_norm": 0.8486387875780823,
      "learning_rate": 0.003,
      "loss": 4.2174,
      "step": 2210
    },
    {
      "epoch": 0.02211,
      "grad_norm": 0.7761571300506404,
      "learning_rate": 0.003,
      "loss": 4.2269,
      "step": 2211
    },
    {
      "epoch": 0.02212,
      "grad_norm": 0.6379539901749441,
      "learning_rate": 0.003,
      "loss": 4.2233,
      "step": 2212
    },
    {
      "epoch": 0.02213,
      "grad_norm": 0.6089238962453585,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2213
    },
    {
      "epoch": 0.02214,
      "grad_norm": 0.5796568814987876,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2214
    },
    {
      "epoch": 0.02215,
      "grad_norm": 0.5660272621079528,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2215
    },
    {
      "epoch": 0.02216,
      "grad_norm": 0.5885926435515716,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2216
    },
    {
      "epoch": 0.02217,
      "grad_norm": 0.5475660813804676,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2217
    },
    {
      "epoch": 0.02218,
      "grad_norm": 0.5260715025238017,
      "learning_rate": 0.003,
      "loss": 4.2304,
      "step": 2218
    },
    {
      "epoch": 0.02219,
      "grad_norm": 0.5716509549906074,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2219
    },
    {
      "epoch": 0.0222,
      "grad_norm": 0.5917539957492576,
      "learning_rate": 0.003,
      "loss": 4.2195,
      "step": 2220
    },
    {
      "epoch": 0.02221,
      "grad_norm": 0.6044205892693139,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 2221
    },
    {
      "epoch": 0.02222,
      "grad_norm": 0.567760157124041,
      "learning_rate": 0.003,
      "loss": 4.2063,
      "step": 2222
    },
    {
      "epoch": 0.02223,
      "grad_norm": 0.577201891352313,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2223
    },
    {
      "epoch": 0.02224,
      "grad_norm": 0.5589539299230697,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2224
    },
    {
      "epoch": 0.02225,
      "grad_norm": 0.578872867820463,
      "learning_rate": 0.003,
      "loss": 4.206,
      "step": 2225
    },
    {
      "epoch": 0.02226,
      "grad_norm": 0.6018889202924489,
      "learning_rate": 0.003,
      "loss": 4.2223,
      "step": 2226
    },
    {
      "epoch": 0.02227,
      "grad_norm": 0.6659111341486127,
      "learning_rate": 0.003,
      "loss": 4.2192,
      "step": 2227
    },
    {
      "epoch": 0.02228,
      "grad_norm": 0.6307543898445819,
      "learning_rate": 0.003,
      "loss": 4.2071,
      "step": 2228
    },
    {
      "epoch": 0.02229,
      "grad_norm": 0.5993087464100946,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 2229
    },
    {
      "epoch": 0.0223,
      "grad_norm": 0.6286376964291409,
      "learning_rate": 0.003,
      "loss": 4.226,
      "step": 2230
    },
    {
      "epoch": 0.02231,
      "grad_norm": 0.872328678246551,
      "learning_rate": 0.003,
      "loss": 4.2394,
      "step": 2231
    },
    {
      "epoch": 0.02232,
      "grad_norm": 1.0461803323605487,
      "learning_rate": 0.003,
      "loss": 4.2634,
      "step": 2232
    },
    {
      "epoch": 0.02233,
      "grad_norm": 0.9329588220097066,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 2233
    },
    {
      "epoch": 0.02234,
      "grad_norm": 0.6760145633983702,
      "learning_rate": 0.003,
      "loss": 4.2457,
      "step": 2234
    },
    {
      "epoch": 0.02235,
      "grad_norm": 0.7832849984021282,
      "learning_rate": 0.003,
      "loss": 4.281,
      "step": 2235
    },
    {
      "epoch": 0.02236,
      "grad_norm": 0.6196461184379554,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2236
    },
    {
      "epoch": 0.02237,
      "grad_norm": 0.5630794415245316,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2237
    },
    {
      "epoch": 0.02238,
      "grad_norm": 0.4338854754415404,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2238
    },
    {
      "epoch": 0.02239,
      "grad_norm": 0.43304313838731295,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2239
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.3849146263035166,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 2240
    },
    {
      "epoch": 0.02241,
      "grad_norm": 0.3700093297043444,
      "learning_rate": 0.003,
      "loss": 4.221,
      "step": 2241
    },
    {
      "epoch": 0.02242,
      "grad_norm": 0.3483834690743904,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2242
    },
    {
      "epoch": 0.02243,
      "grad_norm": 0.38069058438024445,
      "learning_rate": 0.003,
      "loss": 4.2335,
      "step": 2243
    },
    {
      "epoch": 0.02244,
      "grad_norm": 0.4503598461041079,
      "learning_rate": 0.003,
      "loss": 4.21,
      "step": 2244
    },
    {
      "epoch": 0.02245,
      "grad_norm": 0.5173040984777282,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2245
    },
    {
      "epoch": 0.02246,
      "grad_norm": 0.5743673128739117,
      "learning_rate": 0.003,
      "loss": 4.2315,
      "step": 2246
    },
    {
      "epoch": 0.02247,
      "grad_norm": 0.6564841670845484,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2247
    },
    {
      "epoch": 0.02248,
      "grad_norm": 0.7271803573068377,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2248
    },
    {
      "epoch": 0.02249,
      "grad_norm": 0.6357921170928627,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2249
    },
    {
      "epoch": 0.0225,
      "grad_norm": 0.5333359806181992,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2250
    },
    {
      "epoch": 0.02251,
      "grad_norm": 0.5481515163618663,
      "learning_rate": 0.003,
      "loss": 4.2423,
      "step": 2251
    },
    {
      "epoch": 0.02252,
      "grad_norm": 0.5849718573683582,
      "learning_rate": 0.003,
      "loss": 4.2326,
      "step": 2252
    },
    {
      "epoch": 0.02253,
      "grad_norm": 0.692221940613648,
      "learning_rate": 0.003,
      "loss": 4.2188,
      "step": 2253
    },
    {
      "epoch": 0.02254,
      "grad_norm": 0.8350743358894537,
      "learning_rate": 0.003,
      "loss": 4.2328,
      "step": 2254
    },
    {
      "epoch": 0.02255,
      "grad_norm": 0.9700169658063796,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2255
    },
    {
      "epoch": 0.02256,
      "grad_norm": 1.1849964912746291,
      "learning_rate": 0.003,
      "loss": 4.2446,
      "step": 2256
    },
    {
      "epoch": 0.02257,
      "grad_norm": 0.8019987212669306,
      "learning_rate": 0.003,
      "loss": 4.2432,
      "step": 2257
    },
    {
      "epoch": 0.02258,
      "grad_norm": 0.7178719029324007,
      "learning_rate": 0.003,
      "loss": 4.2513,
      "step": 2258
    },
    {
      "epoch": 0.02259,
      "grad_norm": 0.8182127992878601,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 2259
    },
    {
      "epoch": 0.0226,
      "grad_norm": 0.7294977278785716,
      "learning_rate": 0.003,
      "loss": 4.2695,
      "step": 2260
    },
    {
      "epoch": 0.02261,
      "grad_norm": 0.6869467420486108,
      "learning_rate": 0.003,
      "loss": 4.2439,
      "step": 2261
    },
    {
      "epoch": 0.02262,
      "grad_norm": 0.7258556366490507,
      "learning_rate": 0.003,
      "loss": 4.2451,
      "step": 2262
    },
    {
      "epoch": 0.02263,
      "grad_norm": 0.6949107967496511,
      "learning_rate": 0.003,
      "loss": 4.2554,
      "step": 2263
    },
    {
      "epoch": 0.02264,
      "grad_norm": 0.6840843296451264,
      "learning_rate": 0.003,
      "loss": 4.2468,
      "step": 2264
    },
    {
      "epoch": 0.02265,
      "grad_norm": 0.6101088207573817,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 2265
    },
    {
      "epoch": 0.02266,
      "grad_norm": 0.6149463166497737,
      "learning_rate": 0.003,
      "loss": 4.2182,
      "step": 2266
    },
    {
      "epoch": 0.02267,
      "grad_norm": 0.566032425336374,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 2267
    },
    {
      "epoch": 0.02268,
      "grad_norm": 0.5712592856836044,
      "learning_rate": 0.003,
      "loss": 4.2156,
      "step": 2268
    },
    {
      "epoch": 0.02269,
      "grad_norm": 0.5122859378101323,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2269
    },
    {
      "epoch": 0.0227,
      "grad_norm": 0.4651490853981594,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2270
    },
    {
      "epoch": 0.02271,
      "grad_norm": 0.39776686394339583,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2271
    },
    {
      "epoch": 0.02272,
      "grad_norm": 0.4302685557347805,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 2272
    },
    {
      "epoch": 0.02273,
      "grad_norm": 0.467593917531494,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2273
    },
    {
      "epoch": 0.02274,
      "grad_norm": 0.632433536625551,
      "learning_rate": 0.003,
      "loss": 4.2141,
      "step": 2274
    },
    {
      "epoch": 0.02275,
      "grad_norm": 0.8580139662336462,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2275
    },
    {
      "epoch": 0.02276,
      "grad_norm": 0.9516059646785548,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2276
    },
    {
      "epoch": 0.02277,
      "grad_norm": 0.7436294288419117,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2277
    },
    {
      "epoch": 0.02278,
      "grad_norm": 0.6418671737236206,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2278
    },
    {
      "epoch": 0.02279,
      "grad_norm": 0.7656816707126055,
      "learning_rate": 0.003,
      "loss": 4.2435,
      "step": 2279
    },
    {
      "epoch": 0.0228,
      "grad_norm": 0.6177672920210129,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2280
    },
    {
      "epoch": 0.02281,
      "grad_norm": 0.5895210519635931,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2281
    },
    {
      "epoch": 0.02282,
      "grad_norm": 0.5868753488307538,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2282
    },
    {
      "epoch": 0.02283,
      "grad_norm": 0.6052691283316047,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2283
    },
    {
      "epoch": 0.02284,
      "grad_norm": 0.5832141161476754,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2284
    },
    {
      "epoch": 0.02285,
      "grad_norm": 0.5796401382132979,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2285
    },
    {
      "epoch": 0.02286,
      "grad_norm": 0.5270223366823183,
      "learning_rate": 0.003,
      "loss": 4.2443,
      "step": 2286
    },
    {
      "epoch": 0.02287,
      "grad_norm": 0.4022883813311557,
      "learning_rate": 0.003,
      "loss": 4.2176,
      "step": 2287
    },
    {
      "epoch": 0.02288,
      "grad_norm": 0.4249187639605795,
      "learning_rate": 0.003,
      "loss": 4.2409,
      "step": 2288
    },
    {
      "epoch": 0.02289,
      "grad_norm": 0.39026571046663244,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2289
    },
    {
      "epoch": 0.0229,
      "grad_norm": 0.4142451551190753,
      "learning_rate": 0.003,
      "loss": 4.2147,
      "step": 2290
    },
    {
      "epoch": 0.02291,
      "grad_norm": 0.48765765895038643,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2291
    },
    {
      "epoch": 0.02292,
      "grad_norm": 0.5671990063966691,
      "learning_rate": 0.003,
      "loss": 4.1916,
      "step": 2292
    },
    {
      "epoch": 0.02293,
      "grad_norm": 0.5656764518255547,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2293
    },
    {
      "epoch": 0.02294,
      "grad_norm": 0.5275578505406908,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2294
    },
    {
      "epoch": 0.02295,
      "grad_norm": 0.48221866990126555,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 2295
    },
    {
      "epoch": 0.02296,
      "grad_norm": 0.5188277549634542,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2296
    },
    {
      "epoch": 0.02297,
      "grad_norm": 0.6417395019786397,
      "learning_rate": 0.003,
      "loss": 4.199,
      "step": 2297
    },
    {
      "epoch": 0.02298,
      "grad_norm": 0.7094858197213939,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2298
    },
    {
      "epoch": 0.02299,
      "grad_norm": 0.7715781893149263,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2299
    },
    {
      "epoch": 0.023,
      "grad_norm": 0.8183493893838668,
      "learning_rate": 0.003,
      "loss": 4.2495,
      "step": 2300
    },
    {
      "epoch": 0.02301,
      "grad_norm": 0.7727005595462422,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2301
    },
    {
      "epoch": 0.02302,
      "grad_norm": 0.789250865027278,
      "learning_rate": 0.003,
      "loss": 4.2244,
      "step": 2302
    },
    {
      "epoch": 0.02303,
      "grad_norm": 0.8561833032989024,
      "learning_rate": 0.003,
      "loss": 4.2229,
      "step": 2303
    },
    {
      "epoch": 0.02304,
      "grad_norm": 0.7983476489527453,
      "learning_rate": 0.003,
      "loss": 4.2526,
      "step": 2304
    },
    {
      "epoch": 0.02305,
      "grad_norm": 0.792691144527778,
      "learning_rate": 0.003,
      "loss": 4.2307,
      "step": 2305
    },
    {
      "epoch": 0.02306,
      "grad_norm": 0.8333614803540381,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2306
    },
    {
      "epoch": 0.02307,
      "grad_norm": 0.8709733257410114,
      "learning_rate": 0.003,
      "loss": 4.248,
      "step": 2307
    },
    {
      "epoch": 0.02308,
      "grad_norm": 0.7840461913162227,
      "learning_rate": 0.003,
      "loss": 4.2376,
      "step": 2308
    },
    {
      "epoch": 0.02309,
      "grad_norm": 0.7160902424813279,
      "learning_rate": 0.003,
      "loss": 4.263,
      "step": 2309
    },
    {
      "epoch": 0.0231,
      "grad_norm": 0.7757918757353481,
      "learning_rate": 0.003,
      "loss": 4.2432,
      "step": 2310
    },
    {
      "epoch": 0.02311,
      "grad_norm": 0.8687865089693275,
      "learning_rate": 0.003,
      "loss": 4.2573,
      "step": 2311
    },
    {
      "epoch": 0.02312,
      "grad_norm": 0.7846302126542865,
      "learning_rate": 0.003,
      "loss": 4.2298,
      "step": 2312
    },
    {
      "epoch": 0.02313,
      "grad_norm": 0.7297339955221905,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2313
    },
    {
      "epoch": 0.02314,
      "grad_norm": 0.6956047853880688,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2314
    },
    {
      "epoch": 0.02315,
      "grad_norm": 0.6937368414181051,
      "learning_rate": 0.003,
      "loss": 4.2413,
      "step": 2315
    },
    {
      "epoch": 0.02316,
      "grad_norm": 0.6709916211621512,
      "learning_rate": 0.003,
      "loss": 4.2522,
      "step": 2316
    },
    {
      "epoch": 0.02317,
      "grad_norm": 0.6213877076356003,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2317
    },
    {
      "epoch": 0.02318,
      "grad_norm": 0.6088693827583501,
      "learning_rate": 0.003,
      "loss": 4.2184,
      "step": 2318
    },
    {
      "epoch": 0.02319,
      "grad_norm": 0.5989733797969562,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2319
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.55615386704494,
      "learning_rate": 0.003,
      "loss": 4.2463,
      "step": 2320
    },
    {
      "epoch": 0.02321,
      "grad_norm": 0.5520477296154882,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2321
    },
    {
      "epoch": 0.02322,
      "grad_norm": 0.5223023663129024,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 2322
    },
    {
      "epoch": 0.02323,
      "grad_norm": 0.5716320780098701,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2323
    },
    {
      "epoch": 0.02324,
      "grad_norm": 0.6956733472532048,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2324
    },
    {
      "epoch": 0.02325,
      "grad_norm": 0.6924168616855679,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2325
    },
    {
      "epoch": 0.02326,
      "grad_norm": 0.6921696722241616,
      "learning_rate": 0.003,
      "loss": 4.2476,
      "step": 2326
    },
    {
      "epoch": 0.02327,
      "grad_norm": 0.6351653773148184,
      "learning_rate": 0.003,
      "loss": 4.2319,
      "step": 2327
    },
    {
      "epoch": 0.02328,
      "grad_norm": 0.5655372914501792,
      "learning_rate": 0.003,
      "loss": 4.2261,
      "step": 2328
    },
    {
      "epoch": 0.02329,
      "grad_norm": 0.6342553615471473,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 2329
    },
    {
      "epoch": 0.0233,
      "grad_norm": 0.7154577148277532,
      "learning_rate": 0.003,
      "loss": 4.2152,
      "step": 2330
    },
    {
      "epoch": 0.02331,
      "grad_norm": 0.73894366097317,
      "learning_rate": 0.003,
      "loss": 4.2587,
      "step": 2331
    },
    {
      "epoch": 0.02332,
      "grad_norm": 0.7851778924150568,
      "learning_rate": 0.003,
      "loss": 4.2242,
      "step": 2332
    },
    {
      "epoch": 0.02333,
      "grad_norm": 0.7287502366417434,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2333
    },
    {
      "epoch": 0.02334,
      "grad_norm": 0.69094548866664,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 2334
    },
    {
      "epoch": 0.02335,
      "grad_norm": 0.6431810869476228,
      "learning_rate": 0.003,
      "loss": 4.2336,
      "step": 2335
    },
    {
      "epoch": 0.02336,
      "grad_norm": 0.6507673975963663,
      "learning_rate": 0.003,
      "loss": 4.2276,
      "step": 2336
    },
    {
      "epoch": 0.02337,
      "grad_norm": 0.7087163474933083,
      "learning_rate": 0.003,
      "loss": 4.2296,
      "step": 2337
    },
    {
      "epoch": 0.02338,
      "grad_norm": 0.6659743716890278,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2338
    },
    {
      "epoch": 0.02339,
      "grad_norm": 0.5625472175847084,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2339
    },
    {
      "epoch": 0.0234,
      "grad_norm": 0.536202676220793,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2340
    },
    {
      "epoch": 0.02341,
      "grad_norm": 0.4892866338203838,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2341
    },
    {
      "epoch": 0.02342,
      "grad_norm": 0.4796875190367091,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2342
    },
    {
      "epoch": 0.02343,
      "grad_norm": 0.43636850083944123,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 2343
    },
    {
      "epoch": 0.02344,
      "grad_norm": 0.4075473684449614,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2344
    },
    {
      "epoch": 0.02345,
      "grad_norm": 0.4126961162680678,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2345
    },
    {
      "epoch": 0.02346,
      "grad_norm": 0.4484646299481725,
      "learning_rate": 0.003,
      "loss": 4.1936,
      "step": 2346
    },
    {
      "epoch": 0.02347,
      "grad_norm": 0.4884767738301177,
      "learning_rate": 0.003,
      "loss": 4.2318,
      "step": 2347
    },
    {
      "epoch": 0.02348,
      "grad_norm": 0.4734903472149947,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2348
    },
    {
      "epoch": 0.02349,
      "grad_norm": 0.4515057387841677,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2349
    },
    {
      "epoch": 0.0235,
      "grad_norm": 0.43769490231507474,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2350
    },
    {
      "epoch": 0.02351,
      "grad_norm": 0.3875892706722824,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2351
    },
    {
      "epoch": 0.02352,
      "grad_norm": 0.42012415359844946,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 2352
    },
    {
      "epoch": 0.02353,
      "grad_norm": 0.5800085038244271,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2353
    },
    {
      "epoch": 0.02354,
      "grad_norm": 0.9008192298721557,
      "learning_rate": 0.003,
      "loss": 4.2263,
      "step": 2354
    },
    {
      "epoch": 0.02355,
      "grad_norm": 1.174355012660842,
      "learning_rate": 0.003,
      "loss": 4.2398,
      "step": 2355
    },
    {
      "epoch": 0.02356,
      "grad_norm": 0.6757502362960481,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 2356
    },
    {
      "epoch": 0.02357,
      "grad_norm": 0.7784416987559685,
      "learning_rate": 0.003,
      "loss": 4.2185,
      "step": 2357
    },
    {
      "epoch": 0.02358,
      "grad_norm": 0.8207171439845993,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2358
    },
    {
      "epoch": 0.02359,
      "grad_norm": 0.6974291909709516,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 2359
    },
    {
      "epoch": 0.0236,
      "grad_norm": 0.6185472104370391,
      "learning_rate": 0.003,
      "loss": 4.2471,
      "step": 2360
    },
    {
      "epoch": 0.02361,
      "grad_norm": 0.5773489611025617,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2361
    },
    {
      "epoch": 0.02362,
      "grad_norm": 0.5560165828585443,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 2362
    },
    {
      "epoch": 0.02363,
      "grad_norm": 0.5892532979179639,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2363
    },
    {
      "epoch": 0.02364,
      "grad_norm": 0.6748161886878429,
      "learning_rate": 0.003,
      "loss": 4.2153,
      "step": 2364
    },
    {
      "epoch": 0.02365,
      "grad_norm": 0.6598699309472811,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2365
    },
    {
      "epoch": 0.02366,
      "grad_norm": 0.6235317744814385,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2366
    },
    {
      "epoch": 0.02367,
      "grad_norm": 0.6532173738874076,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 2367
    },
    {
      "epoch": 0.02368,
      "grad_norm": 0.6882158454966492,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2368
    },
    {
      "epoch": 0.02369,
      "grad_norm": 0.7139133860202842,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2369
    },
    {
      "epoch": 0.0237,
      "grad_norm": 0.6776525799328785,
      "learning_rate": 0.003,
      "loss": 4.2355,
      "step": 2370
    },
    {
      "epoch": 0.02371,
      "grad_norm": 0.7002833727021253,
      "learning_rate": 0.003,
      "loss": 4.2219,
      "step": 2371
    },
    {
      "epoch": 0.02372,
      "grad_norm": 0.7223141017308237,
      "learning_rate": 0.003,
      "loss": 4.2127,
      "step": 2372
    },
    {
      "epoch": 0.02373,
      "grad_norm": 0.7004144412962903,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 2373
    },
    {
      "epoch": 0.02374,
      "grad_norm": 0.6806743664027574,
      "learning_rate": 0.003,
      "loss": 4.2074,
      "step": 2374
    },
    {
      "epoch": 0.02375,
      "grad_norm": 0.6705132166888766,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2375
    },
    {
      "epoch": 0.02376,
      "grad_norm": 0.5005122224104306,
      "learning_rate": 0.003,
      "loss": 4.2267,
      "step": 2376
    },
    {
      "epoch": 0.02377,
      "grad_norm": 0.4877832394535443,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2377
    },
    {
      "epoch": 0.02378,
      "grad_norm": 0.5211637717940496,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2378
    },
    {
      "epoch": 0.02379,
      "grad_norm": 0.6384313580109909,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2379
    },
    {
      "epoch": 0.0238,
      "grad_norm": 0.7131026886322172,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 2380
    },
    {
      "epoch": 0.02381,
      "grad_norm": 0.6937779462786806,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2381
    },
    {
      "epoch": 0.02382,
      "grad_norm": 0.6337775600473754,
      "learning_rate": 0.003,
      "loss": 4.2401,
      "step": 2382
    },
    {
      "epoch": 0.02383,
      "grad_norm": 0.5646754547957066,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2383
    },
    {
      "epoch": 0.02384,
      "grad_norm": 0.5471514877116583,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 2384
    },
    {
      "epoch": 0.02385,
      "grad_norm": 0.5791601386868904,
      "learning_rate": 0.003,
      "loss": 4.2135,
      "step": 2385
    },
    {
      "epoch": 0.02386,
      "grad_norm": 0.5976046883190522,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2386
    },
    {
      "epoch": 0.02387,
      "grad_norm": 0.6655482933999028,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2387
    },
    {
      "epoch": 0.02388,
      "grad_norm": 0.7334421734543064,
      "learning_rate": 0.003,
      "loss": 4.2504,
      "step": 2388
    },
    {
      "epoch": 0.02389,
      "grad_norm": 0.7475937838490452,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2389
    },
    {
      "epoch": 0.0239,
      "grad_norm": 0.7091495046860523,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2390
    },
    {
      "epoch": 0.02391,
      "grad_norm": 0.6788986600308993,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2391
    },
    {
      "epoch": 0.02392,
      "grad_norm": 0.6156974740652746,
      "learning_rate": 0.003,
      "loss": 4.2308,
      "step": 2392
    },
    {
      "epoch": 0.02393,
      "grad_norm": 0.5941793885935496,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 2393
    },
    {
      "epoch": 0.02394,
      "grad_norm": 0.5863170046009283,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2394
    },
    {
      "epoch": 0.02395,
      "grad_norm": 0.7380191806299783,
      "learning_rate": 0.003,
      "loss": 4.2191,
      "step": 2395
    },
    {
      "epoch": 0.02396,
      "grad_norm": 0.8712169644578591,
      "learning_rate": 0.003,
      "loss": 4.2425,
      "step": 2396
    },
    {
      "epoch": 0.02397,
      "grad_norm": 0.8536583911176343,
      "learning_rate": 0.003,
      "loss": 4.2305,
      "step": 2397
    },
    {
      "epoch": 0.02398,
      "grad_norm": 0.7082106435478536,
      "learning_rate": 0.003,
      "loss": 4.2343,
      "step": 2398
    },
    {
      "epoch": 0.02399,
      "grad_norm": 0.6784314246730624,
      "learning_rate": 0.003,
      "loss": 4.2124,
      "step": 2399
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.7409520225246221,
      "learning_rate": 0.003,
      "loss": 4.2253,
      "step": 2400
    },
    {
      "epoch": 0.02401,
      "grad_norm": 0.7471077758325675,
      "learning_rate": 0.003,
      "loss": 4.2007,
      "step": 2401
    },
    {
      "epoch": 0.02402,
      "grad_norm": 0.7167664651670812,
      "learning_rate": 0.003,
      "loss": 4.2353,
      "step": 2402
    },
    {
      "epoch": 0.02403,
      "grad_norm": 0.7133428609804063,
      "learning_rate": 0.003,
      "loss": 4.2011,
      "step": 2403
    },
    {
      "epoch": 0.02404,
      "grad_norm": 0.6914213864059761,
      "learning_rate": 0.003,
      "loss": 4.2426,
      "step": 2404
    },
    {
      "epoch": 0.02405,
      "grad_norm": 0.6223491766257139,
      "learning_rate": 0.003,
      "loss": 4.2449,
      "step": 2405
    },
    {
      "epoch": 0.02406,
      "grad_norm": 0.6402462299592815,
      "learning_rate": 0.003,
      "loss": 4.2265,
      "step": 2406
    },
    {
      "epoch": 0.02407,
      "grad_norm": 0.5600794609103489,
      "learning_rate": 0.003,
      "loss": 4.2013,
      "step": 2407
    },
    {
      "epoch": 0.02408,
      "grad_norm": 0.4916021474517548,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 2408
    },
    {
      "epoch": 0.02409,
      "grad_norm": 0.46461844041782707,
      "learning_rate": 0.003,
      "loss": 4.2116,
      "step": 2409
    },
    {
      "epoch": 0.0241,
      "grad_norm": 0.5180903118600274,
      "learning_rate": 0.003,
      "loss": 4.2239,
      "step": 2410
    },
    {
      "epoch": 0.02411,
      "grad_norm": 0.6289211485922821,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2411
    },
    {
      "epoch": 0.02412,
      "grad_norm": 0.625292309840426,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2412
    },
    {
      "epoch": 0.02413,
      "grad_norm": 0.5316162511007585,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 2413
    },
    {
      "epoch": 0.02414,
      "grad_norm": 0.569080239332035,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2414
    },
    {
      "epoch": 0.02415,
      "grad_norm": 0.6830500278720256,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2415
    },
    {
      "epoch": 0.02416,
      "grad_norm": 0.8230987152330054,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 2416
    },
    {
      "epoch": 0.02417,
      "grad_norm": 0.9992750696467811,
      "learning_rate": 0.003,
      "loss": 4.241,
      "step": 2417
    },
    {
      "epoch": 0.02418,
      "grad_norm": 0.7172338223207045,
      "learning_rate": 0.003,
      "loss": 4.2271,
      "step": 2418
    },
    {
      "epoch": 0.02419,
      "grad_norm": 0.6066239591362272,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2419
    },
    {
      "epoch": 0.0242,
      "grad_norm": 0.6615008268071535,
      "learning_rate": 0.003,
      "loss": 4.2339,
      "step": 2420
    },
    {
      "epoch": 0.02421,
      "grad_norm": 0.601398245236051,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2421
    },
    {
      "epoch": 0.02422,
      "grad_norm": 0.5164536803360787,
      "learning_rate": 0.003,
      "loss": 4.2444,
      "step": 2422
    },
    {
      "epoch": 0.02423,
      "grad_norm": 0.47425853655922334,
      "learning_rate": 0.003,
      "loss": 4.2323,
      "step": 2423
    },
    {
      "epoch": 0.02424,
      "grad_norm": 0.45557836325278056,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 2424
    },
    {
      "epoch": 0.02425,
      "grad_norm": 0.37992471673130124,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2425
    },
    {
      "epoch": 0.02426,
      "grad_norm": 0.4388918700897719,
      "learning_rate": 0.003,
      "loss": 4.2146,
      "step": 2426
    },
    {
      "epoch": 0.02427,
      "grad_norm": 0.47592644536510775,
      "learning_rate": 0.003,
      "loss": 4.2395,
      "step": 2427
    },
    {
      "epoch": 0.02428,
      "grad_norm": 0.41079892436338244,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2428
    },
    {
      "epoch": 0.02429,
      "grad_norm": 0.48266596555203656,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2429
    },
    {
      "epoch": 0.0243,
      "grad_norm": 0.58630499435022,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2430
    },
    {
      "epoch": 0.02431,
      "grad_norm": 0.6362348679805433,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2431
    },
    {
      "epoch": 0.02432,
      "grad_norm": 0.6409895194011197,
      "learning_rate": 0.003,
      "loss": 4.2237,
      "step": 2432
    },
    {
      "epoch": 0.02433,
      "grad_norm": 0.7045349857959661,
      "learning_rate": 0.003,
      "loss": 4.239,
      "step": 2433
    },
    {
      "epoch": 0.02434,
      "grad_norm": 0.6987807589267505,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2434
    },
    {
      "epoch": 0.02435,
      "grad_norm": 0.5878135110602212,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2435
    },
    {
      "epoch": 0.02436,
      "grad_norm": 0.5898177674736178,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2436
    },
    {
      "epoch": 0.02437,
      "grad_norm": 0.5929083773979196,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 2437
    },
    {
      "epoch": 0.02438,
      "grad_norm": 0.6434362451996457,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2438
    },
    {
      "epoch": 0.02439,
      "grad_norm": 0.6611917410629563,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2439
    },
    {
      "epoch": 0.0244,
      "grad_norm": 0.7743759672777403,
      "learning_rate": 0.003,
      "loss": 4.2424,
      "step": 2440
    },
    {
      "epoch": 0.02441,
      "grad_norm": 0.7462269728851785,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2441
    },
    {
      "epoch": 0.02442,
      "grad_norm": 0.7608157833129767,
      "learning_rate": 0.003,
      "loss": 4.2431,
      "step": 2442
    },
    {
      "epoch": 0.02443,
      "grad_norm": 0.79170341480151,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 2443
    },
    {
      "epoch": 0.02444,
      "grad_norm": 0.7939192302282532,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2444
    },
    {
      "epoch": 0.02445,
      "grad_norm": 0.7598676209330011,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2445
    },
    {
      "epoch": 0.02446,
      "grad_norm": 0.657657458880643,
      "learning_rate": 0.003,
      "loss": 4.2177,
      "step": 2446
    },
    {
      "epoch": 0.02447,
      "grad_norm": 0.636985404351659,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2447
    },
    {
      "epoch": 0.02448,
      "grad_norm": 0.6540065013234154,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 2448
    },
    {
      "epoch": 0.02449,
      "grad_norm": 0.6430617212308102,
      "learning_rate": 0.003,
      "loss": 4.218,
      "step": 2449
    },
    {
      "epoch": 0.0245,
      "grad_norm": 0.6859449218709958,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2450
    },
    {
      "epoch": 0.02451,
      "grad_norm": 0.6585394183553759,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2451
    },
    {
      "epoch": 0.02452,
      "grad_norm": 0.5944042022079538,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2452
    },
    {
      "epoch": 0.02453,
      "grad_norm": 0.6245280633667185,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 2453
    },
    {
      "epoch": 0.02454,
      "grad_norm": 0.6285302652257769,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2454
    },
    {
      "epoch": 0.02455,
      "grad_norm": 0.630536065900029,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 2455
    },
    {
      "epoch": 0.02456,
      "grad_norm": 0.6392454947407117,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 2456
    },
    {
      "epoch": 0.02457,
      "grad_norm": 0.732191714374005,
      "learning_rate": 0.003,
      "loss": 4.2215,
      "step": 2457
    },
    {
      "epoch": 0.02458,
      "grad_norm": 0.8168261785432145,
      "learning_rate": 0.003,
      "loss": 4.1988,
      "step": 2458
    },
    {
      "epoch": 0.02459,
      "grad_norm": 0.8118895866121905,
      "learning_rate": 0.003,
      "loss": 4.2313,
      "step": 2459
    },
    {
      "epoch": 0.0246,
      "grad_norm": 0.7597126257704254,
      "learning_rate": 0.003,
      "loss": 4.2347,
      "step": 2460
    },
    {
      "epoch": 0.02461,
      "grad_norm": 0.7096048570719959,
      "learning_rate": 0.003,
      "loss": 4.2076,
      "step": 2461
    },
    {
      "epoch": 0.02462,
      "grad_norm": 0.6412516044054245,
      "learning_rate": 0.003,
      "loss": 4.2333,
      "step": 2462
    },
    {
      "epoch": 0.02463,
      "grad_norm": 0.6158305280803775,
      "learning_rate": 0.003,
      "loss": 4.2316,
      "step": 2463
    },
    {
      "epoch": 0.02464,
      "grad_norm": 0.5837936979418801,
      "learning_rate": 0.003,
      "loss": 4.2078,
      "step": 2464
    },
    {
      "epoch": 0.02465,
      "grad_norm": 0.5652391806413423,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2465
    },
    {
      "epoch": 0.02466,
      "grad_norm": 0.5314994263522052,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 2466
    },
    {
      "epoch": 0.02467,
      "grad_norm": 0.5492980443942803,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 2467
    },
    {
      "epoch": 0.02468,
      "grad_norm": 0.5476949286484541,
      "learning_rate": 0.003,
      "loss": 4.2248,
      "step": 2468
    },
    {
      "epoch": 0.02469,
      "grad_norm": 0.5739182126857109,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 2469
    },
    {
      "epoch": 0.0247,
      "grad_norm": 0.6455625601186472,
      "learning_rate": 0.003,
      "loss": 4.2003,
      "step": 2470
    },
    {
      "epoch": 0.02471,
      "grad_norm": 0.7500801989374446,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2471
    },
    {
      "epoch": 0.02472,
      "grad_norm": 0.8015350177741498,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 2472
    },
    {
      "epoch": 0.02473,
      "grad_norm": 0.7700030742386493,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 2473
    },
    {
      "epoch": 0.02474,
      "grad_norm": 0.6591968552189625,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 2474
    },
    {
      "epoch": 0.02475,
      "grad_norm": 0.5913092934763802,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2475
    },
    {
      "epoch": 0.02476,
      "grad_norm": 0.6828419322627938,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2476
    },
    {
      "epoch": 0.02477,
      "grad_norm": 0.7429757949041498,
      "learning_rate": 0.003,
      "loss": 4.2273,
      "step": 2477
    },
    {
      "epoch": 0.02478,
      "grad_norm": 0.7533856527859732,
      "learning_rate": 0.003,
      "loss": 4.2029,
      "step": 2478
    },
    {
      "epoch": 0.02479,
      "grad_norm": 0.6271813598380395,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2479
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.7048163325838352,
      "learning_rate": 0.003,
      "loss": 4.2341,
      "step": 2480
    },
    {
      "epoch": 0.02481,
      "grad_norm": 0.7059455660386844,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2481
    },
    {
      "epoch": 0.02482,
      "grad_norm": 0.6095979307747832,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 2482
    },
    {
      "epoch": 0.02483,
      "grad_norm": 0.589007344890445,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 2483
    },
    {
      "epoch": 0.02484,
      "grad_norm": 0.5795798273023001,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2484
    },
    {
      "epoch": 0.02485,
      "grad_norm": 0.6558357821242041,
      "learning_rate": 0.003,
      "loss": 4.2264,
      "step": 2485
    },
    {
      "epoch": 0.02486,
      "grad_norm": 0.6694443853300054,
      "learning_rate": 0.003,
      "loss": 4.2506,
      "step": 2486
    },
    {
      "epoch": 0.02487,
      "grad_norm": 0.7548583562826651,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2487
    },
    {
      "epoch": 0.02488,
      "grad_norm": 0.6086868759932925,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 2488
    },
    {
      "epoch": 0.02489,
      "grad_norm": 0.5543068870038478,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 2489
    },
    {
      "epoch": 0.0249,
      "grad_norm": 0.4695404599404173,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 2490
    },
    {
      "epoch": 0.02491,
      "grad_norm": 0.5098980691467636,
      "learning_rate": 0.003,
      "loss": 4.2001,
      "step": 2491
    },
    {
      "epoch": 0.02492,
      "grad_norm": 0.5029414147729624,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2492
    },
    {
      "epoch": 0.02493,
      "grad_norm": 0.5644399016032551,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 2493
    },
    {
      "epoch": 0.02494,
      "grad_norm": 0.6367872115761151,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2494
    },
    {
      "epoch": 0.02495,
      "grad_norm": 0.6767787025658096,
      "learning_rate": 0.003,
      "loss": 4.2079,
      "step": 2495
    },
    {
      "epoch": 0.02496,
      "grad_norm": 0.696192347480972,
      "learning_rate": 0.003,
      "loss": 4.1973,
      "step": 2496
    },
    {
      "epoch": 0.02497,
      "grad_norm": 0.6551341280875559,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2497
    },
    {
      "epoch": 0.02498,
      "grad_norm": 0.6087269213019051,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 2498
    },
    {
      "epoch": 0.02499,
      "grad_norm": 0.6686291413156635,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2499
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.6959767014574844,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 2500
    },
    {
      "epoch": 0.02501,
      "grad_norm": 0.6984509411978452,
      "learning_rate": 0.003,
      "loss": 4.2249,
      "step": 2501
    },
    {
      "epoch": 0.02502,
      "grad_norm": 0.6296748558389454,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2502
    },
    {
      "epoch": 0.02503,
      "grad_norm": 0.7588933964884115,
      "learning_rate": 0.003,
      "loss": 4.2176,
      "step": 2503
    },
    {
      "epoch": 0.02504,
      "grad_norm": 0.7608780929496015,
      "learning_rate": 0.003,
      "loss": 4.2023,
      "step": 2504
    },
    {
      "epoch": 0.02505,
      "grad_norm": 0.7626036518863434,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2505
    },
    {
      "epoch": 0.02506,
      "grad_norm": 0.8721628417499054,
      "learning_rate": 0.003,
      "loss": 4.2222,
      "step": 2506
    },
    {
      "epoch": 0.02507,
      "grad_norm": 0.77178561857626,
      "learning_rate": 0.003,
      "loss": 4.2111,
      "step": 2507
    },
    {
      "epoch": 0.02508,
      "grad_norm": 0.728729150203243,
      "learning_rate": 0.003,
      "loss": 4.2073,
      "step": 2508
    },
    {
      "epoch": 0.02509,
      "grad_norm": 0.6476865147041562,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2509
    },
    {
      "epoch": 0.0251,
      "grad_norm": 0.6920620401373876,
      "learning_rate": 0.003,
      "loss": 4.2245,
      "step": 2510
    },
    {
      "epoch": 0.02511,
      "grad_norm": 0.7129030622439039,
      "learning_rate": 0.003,
      "loss": 4.2288,
      "step": 2511
    },
    {
      "epoch": 0.02512,
      "grad_norm": 0.7903195113153619,
      "learning_rate": 0.003,
      "loss": 4.2189,
      "step": 2512
    },
    {
      "epoch": 0.02513,
      "grad_norm": 0.7730609450390953,
      "learning_rate": 0.003,
      "loss": 4.2181,
      "step": 2513
    },
    {
      "epoch": 0.02514,
      "grad_norm": 0.656826259281822,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2514
    },
    {
      "epoch": 0.02515,
      "grad_norm": 0.6482134555064485,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2515
    },
    {
      "epoch": 0.02516,
      "grad_norm": 0.5967395863996815,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 2516
    },
    {
      "epoch": 0.02517,
      "grad_norm": 0.5007565780061344,
      "learning_rate": 0.003,
      "loss": 4.2015,
      "step": 2517
    },
    {
      "epoch": 0.02518,
      "grad_norm": 0.5217525966473281,
      "learning_rate": 0.003,
      "loss": 4.2114,
      "step": 2518
    },
    {
      "epoch": 0.02519,
      "grad_norm": 0.4784613905987089,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 2519
    },
    {
      "epoch": 0.0252,
      "grad_norm": 0.4743563932590768,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2520
    },
    {
      "epoch": 0.02521,
      "grad_norm": 0.487720505759532,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 2521
    },
    {
      "epoch": 0.02522,
      "grad_norm": 0.5322961687406275,
      "learning_rate": 0.003,
      "loss": 4.1897,
      "step": 2522
    },
    {
      "epoch": 0.02523,
      "grad_norm": 0.5314708118979667,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 2523
    },
    {
      "epoch": 0.02524,
      "grad_norm": 0.565643482510681,
      "learning_rate": 0.003,
      "loss": 4.2052,
      "step": 2524
    },
    {
      "epoch": 0.02525,
      "grad_norm": 0.5980154086501233,
      "learning_rate": 0.003,
      "loss": 4.2204,
      "step": 2525
    },
    {
      "epoch": 0.02526,
      "grad_norm": 0.668292503382125,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 2526
    },
    {
      "epoch": 0.02527,
      "grad_norm": 0.8972899326232674,
      "learning_rate": 0.003,
      "loss": 4.2125,
      "step": 2527
    },
    {
      "epoch": 0.02528,
      "grad_norm": 1.071209575597214,
      "learning_rate": 0.003,
      "loss": 4.2507,
      "step": 2528
    },
    {
      "epoch": 0.02529,
      "grad_norm": 0.7116312192642624,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2529
    },
    {
      "epoch": 0.0253,
      "grad_norm": 0.6832964815647001,
      "learning_rate": 0.003,
      "loss": 4.2324,
      "step": 2530
    },
    {
      "epoch": 0.02531,
      "grad_norm": 0.625455406806653,
      "learning_rate": 0.003,
      "loss": 4.2283,
      "step": 2531
    },
    {
      "epoch": 0.02532,
      "grad_norm": 0.5481673077994964,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2532
    },
    {
      "epoch": 0.02533,
      "grad_norm": 0.5897257835444514,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2533
    },
    {
      "epoch": 0.02534,
      "grad_norm": 0.5146548286556276,
      "learning_rate": 0.003,
      "loss": 4.2083,
      "step": 2534
    },
    {
      "epoch": 0.02535,
      "grad_norm": 0.5177859418119969,
      "learning_rate": 0.003,
      "loss": 4.2022,
      "step": 2535
    },
    {
      "epoch": 0.02536,
      "grad_norm": 0.5298155181993902,
      "learning_rate": 0.003,
      "loss": 4.1967,
      "step": 2536
    },
    {
      "epoch": 0.02537,
      "grad_norm": 0.4935521410250174,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2537
    },
    {
      "epoch": 0.02538,
      "grad_norm": 0.46468573263998625,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 2538
    },
    {
      "epoch": 0.02539,
      "grad_norm": 0.517571911001853,
      "learning_rate": 0.003,
      "loss": 4.2155,
      "step": 2539
    },
    {
      "epoch": 0.0254,
      "grad_norm": 0.5342760221554671,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2540
    },
    {
      "epoch": 0.02541,
      "grad_norm": 0.54814214731398,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 2541
    },
    {
      "epoch": 0.02542,
      "grad_norm": 0.6094140763884728,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 2542
    },
    {
      "epoch": 0.02543,
      "grad_norm": 0.7209045415511117,
      "learning_rate": 0.003,
      "loss": 4.2097,
      "step": 2543
    },
    {
      "epoch": 0.02544,
      "grad_norm": 0.8126054331647753,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 2544
    },
    {
      "epoch": 0.02545,
      "grad_norm": 0.7645479610935821,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2545
    },
    {
      "epoch": 0.02546,
      "grad_norm": 0.6438407031724117,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2546
    },
    {
      "epoch": 0.02547,
      "grad_norm": 0.6119848106712312,
      "learning_rate": 0.003,
      "loss": 4.2309,
      "step": 2547
    },
    {
      "epoch": 0.02548,
      "grad_norm": 0.5927448337585458,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2548
    },
    {
      "epoch": 0.02549,
      "grad_norm": 0.8336977020901954,
      "learning_rate": 0.003,
      "loss": 4.2269,
      "step": 2549
    },
    {
      "epoch": 0.0255,
      "grad_norm": 1.0008812022371014,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2550
    },
    {
      "epoch": 0.02551,
      "grad_norm": 1.047497616869583,
      "learning_rate": 0.003,
      "loss": 4.2455,
      "step": 2551
    },
    {
      "epoch": 0.02552,
      "grad_norm": 0.9125531296176879,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2552
    },
    {
      "epoch": 0.02553,
      "grad_norm": 1.0094140787791404,
      "learning_rate": 0.003,
      "loss": 4.2917,
      "step": 2553
    },
    {
      "epoch": 0.02554,
      "grad_norm": 0.9401220726777249,
      "learning_rate": 0.003,
      "loss": 4.2404,
      "step": 2554
    },
    {
      "epoch": 0.02555,
      "grad_norm": 0.8810312726451306,
      "learning_rate": 0.003,
      "loss": 4.233,
      "step": 2555
    },
    {
      "epoch": 0.02556,
      "grad_norm": 0.8398872404556826,
      "learning_rate": 0.003,
      "loss": 4.242,
      "step": 2556
    },
    {
      "epoch": 0.02557,
      "grad_norm": 0.8674958106494866,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2557
    },
    {
      "epoch": 0.02558,
      "grad_norm": 0.9359611753863145,
      "learning_rate": 0.003,
      "loss": 4.2763,
      "step": 2558
    },
    {
      "epoch": 0.02559,
      "grad_norm": 0.919728435180346,
      "learning_rate": 0.003,
      "loss": 4.2634,
      "step": 2559
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.9704240601933475,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2560
    },
    {
      "epoch": 0.02561,
      "grad_norm": 0.9645281717104208,
      "learning_rate": 0.003,
      "loss": 4.2533,
      "step": 2561
    },
    {
      "epoch": 0.02562,
      "grad_norm": 0.9553791325261367,
      "learning_rate": 0.003,
      "loss": 4.2754,
      "step": 2562
    },
    {
      "epoch": 0.02563,
      "grad_norm": 0.7877299682148988,
      "learning_rate": 0.003,
      "loss": 4.2515,
      "step": 2563
    },
    {
      "epoch": 0.02564,
      "grad_norm": 0.6960170044487899,
      "learning_rate": 0.003,
      "loss": 4.264,
      "step": 2564
    },
    {
      "epoch": 0.02565,
      "grad_norm": 0.5853819163555847,
      "learning_rate": 0.003,
      "loss": 4.2346,
      "step": 2565
    },
    {
      "epoch": 0.02566,
      "grad_norm": 0.56684551943134,
      "learning_rate": 0.003,
      "loss": 4.2543,
      "step": 2566
    },
    {
      "epoch": 0.02567,
      "grad_norm": 0.5518127543434307,
      "learning_rate": 0.003,
      "loss": 4.251,
      "step": 2567
    },
    {
      "epoch": 0.02568,
      "grad_norm": 0.5196560828829251,
      "learning_rate": 0.003,
      "loss": 4.2402,
      "step": 2568
    },
    {
      "epoch": 0.02569,
      "grad_norm": 0.595356322869423,
      "learning_rate": 0.003,
      "loss": 4.2224,
      "step": 2569
    },
    {
      "epoch": 0.0257,
      "grad_norm": 0.7758881983100933,
      "learning_rate": 0.003,
      "loss": 4.2302,
      "step": 2570
    },
    {
      "epoch": 0.02571,
      "grad_norm": 0.9784633915798607,
      "learning_rate": 0.003,
      "loss": 4.2591,
      "step": 2571
    },
    {
      "epoch": 0.02572,
      "grad_norm": 0.9890948659534822,
      "learning_rate": 0.003,
      "loss": 4.2413,
      "step": 2572
    },
    {
      "epoch": 0.02573,
      "grad_norm": 0.7618440411400482,
      "learning_rate": 0.003,
      "loss": 4.2415,
      "step": 2573
    },
    {
      "epoch": 0.02574,
      "grad_norm": 0.5829959121875294,
      "learning_rate": 0.003,
      "loss": 4.2662,
      "step": 2574
    },
    {
      "epoch": 0.02575,
      "grad_norm": 0.5754964401101073,
      "learning_rate": 0.003,
      "loss": 4.1946,
      "step": 2575
    },
    {
      "epoch": 0.02576,
      "grad_norm": 0.5023263337601672,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2576
    },
    {
      "epoch": 0.02577,
      "grad_norm": 0.5094113404385299,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 2577
    },
    {
      "epoch": 0.02578,
      "grad_norm": 0.45458875378333236,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2578
    },
    {
      "epoch": 0.02579,
      "grad_norm": 0.376766997653558,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 2579
    },
    {
      "epoch": 0.0258,
      "grad_norm": 0.3615714006582123,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 2580
    },
    {
      "epoch": 0.02581,
      "grad_norm": 0.34139864457507546,
      "learning_rate": 0.003,
      "loss": 4.2314,
      "step": 2581
    },
    {
      "epoch": 0.02582,
      "grad_norm": 0.31484861485882354,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 2582
    },
    {
      "epoch": 0.02583,
      "grad_norm": 0.27908396977569033,
      "learning_rate": 0.003,
      "loss": 4.1944,
      "step": 2583
    },
    {
      "epoch": 0.02584,
      "grad_norm": 0.28343226190675325,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2584
    },
    {
      "epoch": 0.02585,
      "grad_norm": 0.31066810836348263,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 2585
    },
    {
      "epoch": 0.02586,
      "grad_norm": 0.29167242806584853,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2586
    },
    {
      "epoch": 0.02587,
      "grad_norm": 0.3309461372612561,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 2587
    },
    {
      "epoch": 0.02588,
      "grad_norm": 0.3812190342327403,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 2588
    },
    {
      "epoch": 0.02589,
      "grad_norm": 0.48608400124636386,
      "learning_rate": 0.003,
      "loss": 4.1958,
      "step": 2589
    },
    {
      "epoch": 0.0259,
      "grad_norm": 0.6708650835599492,
      "learning_rate": 0.003,
      "loss": 4.2112,
      "step": 2590
    },
    {
      "epoch": 0.02591,
      "grad_norm": 0.7815243168293263,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 2591
    },
    {
      "epoch": 0.02592,
      "grad_norm": 0.6070733396030551,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2592
    },
    {
      "epoch": 0.02593,
      "grad_norm": 0.36755746327659883,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2593
    },
    {
      "epoch": 0.02594,
      "grad_norm": 0.5461931648813682,
      "learning_rate": 0.003,
      "loss": 4.2113,
      "step": 2594
    },
    {
      "epoch": 0.02595,
      "grad_norm": 0.5548085387163001,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 2595
    },
    {
      "epoch": 0.02596,
      "grad_norm": 0.6279120258670815,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2596
    },
    {
      "epoch": 0.02597,
      "grad_norm": 0.6833879500904073,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 2597
    },
    {
      "epoch": 0.02598,
      "grad_norm": 0.7079000881755875,
      "learning_rate": 0.003,
      "loss": 4.2216,
      "step": 2598
    },
    {
      "epoch": 0.02599,
      "grad_norm": 0.6366436124583713,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2599
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.5901740968221895,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2600
    },
    {
      "epoch": 0.02601,
      "grad_norm": 0.5956938888246321,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2601
    },
    {
      "epoch": 0.02602,
      "grad_norm": 0.5493627512392831,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2602
    },
    {
      "epoch": 0.02603,
      "grad_norm": 0.44261728447277765,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2603
    },
    {
      "epoch": 0.02604,
      "grad_norm": 0.47710609408550947,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 2604
    },
    {
      "epoch": 0.02605,
      "grad_norm": 0.4918626291452229,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 2605
    },
    {
      "epoch": 0.02606,
      "grad_norm": 0.46913897415532124,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2606
    },
    {
      "epoch": 0.02607,
      "grad_norm": 0.48033538034271883,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 2607
    },
    {
      "epoch": 0.02608,
      "grad_norm": 0.5430081805575081,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 2608
    },
    {
      "epoch": 0.02609,
      "grad_norm": 0.7745086701981942,
      "learning_rate": 0.003,
      "loss": 4.2301,
      "step": 2609
    },
    {
      "epoch": 0.0261,
      "grad_norm": 0.9079228266743159,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2610
    },
    {
      "epoch": 0.02611,
      "grad_norm": 0.8915671065001584,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 2611
    },
    {
      "epoch": 0.02612,
      "grad_norm": 0.8970335727218054,
      "learning_rate": 0.003,
      "loss": 4.2272,
      "step": 2612
    },
    {
      "epoch": 0.02613,
      "grad_norm": 0.7184841949590459,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 2613
    },
    {
      "epoch": 0.02614,
      "grad_norm": 0.6361221793660993,
      "learning_rate": 0.003,
      "loss": 4.1896,
      "step": 2614
    },
    {
      "epoch": 0.02615,
      "grad_norm": 0.6008095298009088,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 2615
    },
    {
      "epoch": 0.02616,
      "grad_norm": 0.5894242905465984,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2616
    },
    {
      "epoch": 0.02617,
      "grad_norm": 0.6436011542740302,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 2617
    },
    {
      "epoch": 0.02618,
      "grad_norm": 0.7283160674464684,
      "learning_rate": 0.003,
      "loss": 4.2126,
      "step": 2618
    },
    {
      "epoch": 0.02619,
      "grad_norm": 0.7948866659668812,
      "learning_rate": 0.003,
      "loss": 4.2279,
      "step": 2619
    },
    {
      "epoch": 0.0262,
      "grad_norm": 0.7483085908476613,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2620
    },
    {
      "epoch": 0.02621,
      "grad_norm": 0.6674824981271792,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 2621
    },
    {
      "epoch": 0.02622,
      "grad_norm": 0.6038002286952583,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2622
    },
    {
      "epoch": 0.02623,
      "grad_norm": 0.6014737867483161,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2623
    },
    {
      "epoch": 0.02624,
      "grad_norm": 0.5903601826725474,
      "learning_rate": 0.003,
      "loss": 4.2493,
      "step": 2624
    },
    {
      "epoch": 0.02625,
      "grad_norm": 0.6662689883238633,
      "learning_rate": 0.003,
      "loss": 4.2487,
      "step": 2625
    },
    {
      "epoch": 0.02626,
      "grad_norm": 0.6773686120638865,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2626
    },
    {
      "epoch": 0.02627,
      "grad_norm": 0.6189409649870486,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 2627
    },
    {
      "epoch": 0.02628,
      "grad_norm": 0.6566341590011362,
      "learning_rate": 0.003,
      "loss": 4.2282,
      "step": 2628
    },
    {
      "epoch": 0.02629,
      "grad_norm": 0.7499843369890263,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 2629
    },
    {
      "epoch": 0.0263,
      "grad_norm": 0.7612666434746392,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2630
    },
    {
      "epoch": 0.02631,
      "grad_norm": 0.6535861726328742,
      "learning_rate": 0.003,
      "loss": 4.2211,
      "step": 2631
    },
    {
      "epoch": 0.02632,
      "grad_norm": 0.6394295181872582,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2632
    },
    {
      "epoch": 0.02633,
      "grad_norm": 0.6261623421196629,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 2633
    },
    {
      "epoch": 0.02634,
      "grad_norm": 0.574244319715974,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2634
    },
    {
      "epoch": 0.02635,
      "grad_norm": 0.4907934619440069,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2635
    },
    {
      "epoch": 0.02636,
      "grad_norm": 0.47213288352020044,
      "learning_rate": 0.003,
      "loss": 4.1862,
      "step": 2636
    },
    {
      "epoch": 0.02637,
      "grad_norm": 0.4457276996768415,
      "learning_rate": 0.003,
      "loss": 4.2168,
      "step": 2637
    },
    {
      "epoch": 0.02638,
      "grad_norm": 0.45345762795109046,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2638
    },
    {
      "epoch": 0.02639,
      "grad_norm": 0.49627655516291913,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 2639
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.5785078427721256,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 2640
    },
    {
      "epoch": 0.02641,
      "grad_norm": 0.7855264178492048,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 2641
    },
    {
      "epoch": 0.02642,
      "grad_norm": 0.9564797199743369,
      "learning_rate": 0.003,
      "loss": 4.2106,
      "step": 2642
    },
    {
      "epoch": 0.02643,
      "grad_norm": 0.9009180286527748,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 2643
    },
    {
      "epoch": 0.02644,
      "grad_norm": 0.6806094030785705,
      "learning_rate": 0.003,
      "loss": 4.2066,
      "step": 2644
    },
    {
      "epoch": 0.02645,
      "grad_norm": 0.6950198239199242,
      "learning_rate": 0.003,
      "loss": 4.2213,
      "step": 2645
    },
    {
      "epoch": 0.02646,
      "grad_norm": 0.6983266391512944,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2646
    },
    {
      "epoch": 0.02647,
      "grad_norm": 0.6601491893781873,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 2647
    },
    {
      "epoch": 0.02648,
      "grad_norm": 0.6355139962402977,
      "learning_rate": 0.003,
      "loss": 4.2257,
      "step": 2648
    },
    {
      "epoch": 0.02649,
      "grad_norm": 0.5820480039604559,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2649
    },
    {
      "epoch": 0.0265,
      "grad_norm": 0.5775761818558494,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 2650
    },
    {
      "epoch": 0.02651,
      "grad_norm": 0.5277676161314399,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2651
    },
    {
      "epoch": 0.02652,
      "grad_norm": 0.5297857391988389,
      "learning_rate": 0.003,
      "loss": 4.2161,
      "step": 2652
    },
    {
      "epoch": 0.02653,
      "grad_norm": 0.5897074985274656,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 2653
    },
    {
      "epoch": 0.02654,
      "grad_norm": 0.6546317550221565,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 2654
    },
    {
      "epoch": 0.02655,
      "grad_norm": 0.6571917314744374,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 2655
    },
    {
      "epoch": 0.02656,
      "grad_norm": 0.6300742703920978,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 2656
    },
    {
      "epoch": 0.02657,
      "grad_norm": 0.6437986671993883,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 2657
    },
    {
      "epoch": 0.02658,
      "grad_norm": 0.6540624785623048,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 2658
    },
    {
      "epoch": 0.02659,
      "grad_norm": 0.6833039431896774,
      "learning_rate": 0.003,
      "loss": 4.2004,
      "step": 2659
    },
    {
      "epoch": 0.0266,
      "grad_norm": 0.6560030968945707,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 2660
    },
    {
      "epoch": 0.02661,
      "grad_norm": 0.6045077056358651,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2661
    },
    {
      "epoch": 0.02662,
      "grad_norm": 0.5982743265464419,
      "learning_rate": 0.003,
      "loss": 4.1919,
      "step": 2662
    },
    {
      "epoch": 0.02663,
      "grad_norm": 0.6142468679426575,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 2663
    },
    {
      "epoch": 0.02664,
      "grad_norm": 0.5662553451652035,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 2664
    },
    {
      "epoch": 0.02665,
      "grad_norm": 0.5761075041061567,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2665
    },
    {
      "epoch": 0.02666,
      "grad_norm": 0.6870114427260798,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2666
    },
    {
      "epoch": 0.02667,
      "grad_norm": 0.8311453279031553,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 2667
    },
    {
      "epoch": 0.02668,
      "grad_norm": 0.9940869003642662,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 2668
    },
    {
      "epoch": 0.02669,
      "grad_norm": 0.9365809099626523,
      "learning_rate": 0.003,
      "loss": 4.2092,
      "step": 2669
    },
    {
      "epoch": 0.0267,
      "grad_norm": 0.7814988387932911,
      "learning_rate": 0.003,
      "loss": 4.2044,
      "step": 2670
    },
    {
      "epoch": 0.02671,
      "grad_norm": 0.7536334880750852,
      "learning_rate": 0.003,
      "loss": 4.2232,
      "step": 2671
    },
    {
      "epoch": 0.02672,
      "grad_norm": 0.6570195531624667,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2672
    },
    {
      "epoch": 0.02673,
      "grad_norm": 0.5512466402663184,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 2673
    },
    {
      "epoch": 0.02674,
      "grad_norm": 0.61723011024151,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 2674
    },
    {
      "epoch": 0.02675,
      "grad_norm": 0.6694888220820355,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 2675
    },
    {
      "epoch": 0.02676,
      "grad_norm": 0.684903042237445,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2676
    },
    {
      "epoch": 0.02677,
      "grad_norm": 0.6916594775917959,
      "learning_rate": 0.003,
      "loss": 4.2254,
      "step": 2677
    },
    {
      "epoch": 0.02678,
      "grad_norm": 0.5888995948139836,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 2678
    },
    {
      "epoch": 0.02679,
      "grad_norm": 0.46001488738668683,
      "learning_rate": 0.003,
      "loss": 4.1818,
      "step": 2679
    },
    {
      "epoch": 0.0268,
      "grad_norm": 0.48559488053080857,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 2680
    },
    {
      "epoch": 0.02681,
      "grad_norm": 0.5328577853550642,
      "learning_rate": 0.003,
      "loss": 4.2162,
      "step": 2681
    },
    {
      "epoch": 0.02682,
      "grad_norm": 0.5478209027693539,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 2682
    },
    {
      "epoch": 0.02683,
      "grad_norm": 0.5231582728945893,
      "learning_rate": 0.003,
      "loss": 4.214,
      "step": 2683
    },
    {
      "epoch": 0.02684,
      "grad_norm": 0.5004578011633336,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2684
    },
    {
      "epoch": 0.02685,
      "grad_norm": 0.5339999052577957,
      "learning_rate": 0.003,
      "loss": 4.1934,
      "step": 2685
    },
    {
      "epoch": 0.02686,
      "grad_norm": 0.5808031106775001,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2686
    },
    {
      "epoch": 0.02687,
      "grad_norm": 0.6474970996641412,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2687
    },
    {
      "epoch": 0.02688,
      "grad_norm": 0.6071304394655341,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 2688
    },
    {
      "epoch": 0.02689,
      "grad_norm": 0.6303081168202617,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 2689
    },
    {
      "epoch": 0.0269,
      "grad_norm": 0.6875418523901222,
      "learning_rate": 0.003,
      "loss": 4.2036,
      "step": 2690
    },
    {
      "epoch": 0.02691,
      "grad_norm": 0.8385284997095487,
      "learning_rate": 0.003,
      "loss": 4.2047,
      "step": 2691
    },
    {
      "epoch": 0.02692,
      "grad_norm": 0.9480846054518673,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2692
    },
    {
      "epoch": 0.02693,
      "grad_norm": 0.8071009083907058,
      "learning_rate": 0.003,
      "loss": 4.2239,
      "step": 2693
    },
    {
      "epoch": 0.02694,
      "grad_norm": 0.726325622879689,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 2694
    },
    {
      "epoch": 0.02695,
      "grad_norm": 0.6357501756622079,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 2695
    },
    {
      "epoch": 0.02696,
      "grad_norm": 0.5914031503189017,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2696
    },
    {
      "epoch": 0.02697,
      "grad_norm": 0.6523831217047013,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2697
    },
    {
      "epoch": 0.02698,
      "grad_norm": 0.6028229475417466,
      "learning_rate": 0.003,
      "loss": 4.1939,
      "step": 2698
    },
    {
      "epoch": 0.02699,
      "grad_norm": 0.5476388167406291,
      "learning_rate": 0.003,
      "loss": 4.2094,
      "step": 2699
    },
    {
      "epoch": 0.027,
      "grad_norm": 0.4742225811072004,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2700
    },
    {
      "epoch": 0.02701,
      "grad_norm": 0.4202221652460036,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 2701
    },
    {
      "epoch": 0.02702,
      "grad_norm": 0.43365355703010106,
      "learning_rate": 0.003,
      "loss": 4.2144,
      "step": 2702
    },
    {
      "epoch": 0.02703,
      "grad_norm": 0.4838782470885569,
      "learning_rate": 0.003,
      "loss": 4.2107,
      "step": 2703
    },
    {
      "epoch": 0.02704,
      "grad_norm": 0.5383020066295563,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 2704
    },
    {
      "epoch": 0.02705,
      "grad_norm": 0.6989516882897651,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2705
    },
    {
      "epoch": 0.02706,
      "grad_norm": 0.7486144178237021,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 2706
    },
    {
      "epoch": 0.02707,
      "grad_norm": 0.7126817427692109,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 2707
    },
    {
      "epoch": 0.02708,
      "grad_norm": 0.6105439188174862,
      "learning_rate": 0.003,
      "loss": 4.2058,
      "step": 2708
    },
    {
      "epoch": 0.02709,
      "grad_norm": 0.5700397686001112,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 2709
    },
    {
      "epoch": 0.0271,
      "grad_norm": 0.6208775729436938,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 2710
    },
    {
      "epoch": 0.02711,
      "grad_norm": 0.6014094255252589,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 2711
    },
    {
      "epoch": 0.02712,
      "grad_norm": 0.5261948655268764,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 2712
    },
    {
      "epoch": 0.02713,
      "grad_norm": 0.5550123847377345,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 2713
    },
    {
      "epoch": 0.02714,
      "grad_norm": 0.56736840403282,
      "learning_rate": 0.003,
      "loss": 4.1717,
      "step": 2714
    },
    {
      "epoch": 0.02715,
      "grad_norm": 0.6628029062916901,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2715
    },
    {
      "epoch": 0.02716,
      "grad_norm": 0.768979209194684,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2716
    },
    {
      "epoch": 0.02717,
      "grad_norm": 0.7482884136468829,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 2717
    },
    {
      "epoch": 0.02718,
      "grad_norm": 0.8846409333789726,
      "learning_rate": 0.003,
      "loss": 4.219,
      "step": 2718
    },
    {
      "epoch": 0.02719,
      "grad_norm": 0.9429345703060683,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2719
    },
    {
      "epoch": 0.0272,
      "grad_norm": 1.002947468833307,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2720
    },
    {
      "epoch": 0.02721,
      "grad_norm": 1.0695835852166422,
      "learning_rate": 0.003,
      "loss": 4.2286,
      "step": 2721
    },
    {
      "epoch": 0.02722,
      "grad_norm": 1.1056646889932085,
      "learning_rate": 0.003,
      "loss": 4.2273,
      "step": 2722
    },
    {
      "epoch": 0.02723,
      "grad_norm": 0.8267701861491207,
      "learning_rate": 0.003,
      "loss": 4.2297,
      "step": 2723
    },
    {
      "epoch": 0.02724,
      "grad_norm": 0.7367716189571131,
      "learning_rate": 0.003,
      "loss": 4.2292,
      "step": 2724
    },
    {
      "epoch": 0.02725,
      "grad_norm": 0.6044896083683174,
      "learning_rate": 0.003,
      "loss": 4.2154,
      "step": 2725
    },
    {
      "epoch": 0.02726,
      "grad_norm": 0.6956030827952617,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2726
    },
    {
      "epoch": 0.02727,
      "grad_norm": 0.7192323257722564,
      "learning_rate": 0.003,
      "loss": 4.2197,
      "step": 2727
    },
    {
      "epoch": 0.02728,
      "grad_norm": 0.7783208693346495,
      "learning_rate": 0.003,
      "loss": 4.2088,
      "step": 2728
    },
    {
      "epoch": 0.02729,
      "grad_norm": 0.8318571752999154,
      "learning_rate": 0.003,
      "loss": 4.2427,
      "step": 2729
    },
    {
      "epoch": 0.0273,
      "grad_norm": 0.6686152556603595,
      "learning_rate": 0.003,
      "loss": 4.2064,
      "step": 2730
    },
    {
      "epoch": 0.02731,
      "grad_norm": 0.6832226230920126,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 2731
    },
    {
      "epoch": 0.02732,
      "grad_norm": 0.6331150968875209,
      "learning_rate": 0.003,
      "loss": 4.1987,
      "step": 2732
    },
    {
      "epoch": 0.02733,
      "grad_norm": 0.5278467341667037,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2733
    },
    {
      "epoch": 0.02734,
      "grad_norm": 0.6138764744804129,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 2734
    },
    {
      "epoch": 0.02735,
      "grad_norm": 0.561739809597342,
      "learning_rate": 0.003,
      "loss": 4.2193,
      "step": 2735
    },
    {
      "epoch": 0.02736,
      "grad_norm": 0.5908346453733819,
      "learning_rate": 0.003,
      "loss": 4.1928,
      "step": 2736
    },
    {
      "epoch": 0.02737,
      "grad_norm": 0.6511959635659452,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 2737
    },
    {
      "epoch": 0.02738,
      "grad_norm": 0.758698109151286,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 2738
    },
    {
      "epoch": 0.02739,
      "grad_norm": 0.6956474809291019,
      "learning_rate": 0.003,
      "loss": 4.2037,
      "step": 2739
    },
    {
      "epoch": 0.0274,
      "grad_norm": 0.5058919640608095,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2740
    },
    {
      "epoch": 0.02741,
      "grad_norm": 0.5563232333292056,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2741
    },
    {
      "epoch": 0.02742,
      "grad_norm": 0.5987655354363043,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 2742
    },
    {
      "epoch": 0.02743,
      "grad_norm": 0.5914696396737046,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 2743
    },
    {
      "epoch": 0.02744,
      "grad_norm": 0.5630038713206789,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2744
    },
    {
      "epoch": 0.02745,
      "grad_norm": 0.587542894365965,
      "learning_rate": 0.003,
      "loss": 4.2027,
      "step": 2745
    },
    {
      "epoch": 0.02746,
      "grad_norm": 0.5934042861360144,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 2746
    },
    {
      "epoch": 0.02747,
      "grad_norm": 0.6474787518036716,
      "learning_rate": 0.003,
      "loss": 4.212,
      "step": 2747
    },
    {
      "epoch": 0.02748,
      "grad_norm": 0.8113625213803506,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 2748
    },
    {
      "epoch": 0.02749,
      "grad_norm": 0.8229694069124209,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 2749
    },
    {
      "epoch": 0.0275,
      "grad_norm": 0.6793478072709392,
      "learning_rate": 0.003,
      "loss": 4.2077,
      "step": 2750
    },
    {
      "epoch": 0.02751,
      "grad_norm": 0.5656831593340759,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 2751
    },
    {
      "epoch": 0.02752,
      "grad_norm": 0.5267924527455705,
      "learning_rate": 0.003,
      "loss": 4.2017,
      "step": 2752
    },
    {
      "epoch": 0.02753,
      "grad_norm": 0.502928202841188,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 2753
    },
    {
      "epoch": 0.02754,
      "grad_norm": 0.5044833881071361,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 2754
    },
    {
      "epoch": 0.02755,
      "grad_norm": 0.5267594677995416,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 2755
    },
    {
      "epoch": 0.02756,
      "grad_norm": 0.456921415827541,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 2756
    },
    {
      "epoch": 0.02757,
      "grad_norm": 0.3836455619810196,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 2757
    },
    {
      "epoch": 0.02758,
      "grad_norm": 0.34559498426529056,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 2758
    },
    {
      "epoch": 0.02759,
      "grad_norm": 0.34478709094038296,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 2759
    },
    {
      "epoch": 0.0276,
      "grad_norm": 0.4284232121155303,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 2760
    },
    {
      "epoch": 0.02761,
      "grad_norm": 0.5071377792443562,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2761
    },
    {
      "epoch": 0.02762,
      "grad_norm": 0.6171334423886827,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 2762
    },
    {
      "epoch": 0.02763,
      "grad_norm": 0.7563311828472695,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2763
    },
    {
      "epoch": 0.02764,
      "grad_norm": 0.8185996809603724,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 2764
    },
    {
      "epoch": 0.02765,
      "grad_norm": 0.7325003643978839,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 2765
    },
    {
      "epoch": 0.02766,
      "grad_norm": 0.6977141939643235,
      "learning_rate": 0.003,
      "loss": 4.161,
      "step": 2766
    },
    {
      "epoch": 0.02767,
      "grad_norm": 0.7178457591084105,
      "learning_rate": 0.003,
      "loss": 4.2198,
      "step": 2767
    },
    {
      "epoch": 0.02768,
      "grad_norm": 0.6603543944468468,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 2768
    },
    {
      "epoch": 0.02769,
      "grad_norm": 0.5908808640128104,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2769
    },
    {
      "epoch": 0.0277,
      "grad_norm": 0.5584733753440203,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 2770
    },
    {
      "epoch": 0.02771,
      "grad_norm": 0.5513412132170109,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 2771
    },
    {
      "epoch": 0.02772,
      "grad_norm": 0.601007172067716,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 2772
    },
    {
      "epoch": 0.02773,
      "grad_norm": 0.6222859555759426,
      "learning_rate": 0.003,
      "loss": 4.1902,
      "step": 2773
    },
    {
      "epoch": 0.02774,
      "grad_norm": 0.6298455082538394,
      "learning_rate": 0.003,
      "loss": 4.2142,
      "step": 2774
    },
    {
      "epoch": 0.02775,
      "grad_norm": 0.6384186594957891,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 2775
    },
    {
      "epoch": 0.02776,
      "grad_norm": 0.7183697808219275,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 2776
    },
    {
      "epoch": 0.02777,
      "grad_norm": 0.6932591793310324,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 2777
    },
    {
      "epoch": 0.02778,
      "grad_norm": 0.6667113422946584,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 2778
    },
    {
      "epoch": 0.02779,
      "grad_norm": 0.7407795606572514,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 2779
    },
    {
      "epoch": 0.0278,
      "grad_norm": 0.7295794698337943,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 2780
    },
    {
      "epoch": 0.02781,
      "grad_norm": 0.6893914165082023,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 2781
    },
    {
      "epoch": 0.02782,
      "grad_norm": 0.6620017500188085,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 2782
    },
    {
      "epoch": 0.02783,
      "grad_norm": 0.6131468932443346,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 2783
    },
    {
      "epoch": 0.02784,
      "grad_norm": 0.5579371518365059,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 2784
    },
    {
      "epoch": 0.02785,
      "grad_norm": 0.49723848657236286,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 2785
    },
    {
      "epoch": 0.02786,
      "grad_norm": 0.5125002182675953,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 2786
    },
    {
      "epoch": 0.02787,
      "grad_norm": 0.4914565242864978,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 2787
    },
    {
      "epoch": 0.02788,
      "grad_norm": 0.526484975930702,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 2788
    },
    {
      "epoch": 0.02789,
      "grad_norm": 0.5321807051700945,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2789
    },
    {
      "epoch": 0.0279,
      "grad_norm": 0.5840495108099473,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 2790
    },
    {
      "epoch": 0.02791,
      "grad_norm": 0.9202391536169947,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 2791
    },
    {
      "epoch": 0.02792,
      "grad_norm": 1.2841478556987087,
      "learning_rate": 0.003,
      "loss": 4.2354,
      "step": 2792
    },
    {
      "epoch": 0.02793,
      "grad_norm": 0.6986972433822171,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 2793
    },
    {
      "epoch": 0.02794,
      "grad_norm": 0.7400669972097395,
      "learning_rate": 0.003,
      "loss": 4.2072,
      "step": 2794
    },
    {
      "epoch": 0.02795,
      "grad_norm": 0.7145362265408897,
      "learning_rate": 0.003,
      "loss": 4.2139,
      "step": 2795
    },
    {
      "epoch": 0.02796,
      "grad_norm": 0.7481134935341638,
      "learning_rate": 0.003,
      "loss": 4.2118,
      "step": 2796
    },
    {
      "epoch": 0.02797,
      "grad_norm": 0.8256336591556148,
      "learning_rate": 0.003,
      "loss": 4.2281,
      "step": 2797
    },
    {
      "epoch": 0.02798,
      "grad_norm": 0.7297049193325166,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 2798
    },
    {
      "epoch": 0.02799,
      "grad_norm": 0.7233108511585719,
      "learning_rate": 0.003,
      "loss": 4.2063,
      "step": 2799
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.6743448137124989,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2800
    },
    {
      "epoch": 0.02801,
      "grad_norm": 0.5867417428513181,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2801
    },
    {
      "epoch": 0.02802,
      "grad_norm": 0.557083005494101,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 2802
    },
    {
      "epoch": 0.02803,
      "grad_norm": 0.5239629190819096,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2803
    },
    {
      "epoch": 0.02804,
      "grad_norm": 0.6122062528440474,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 2804
    },
    {
      "epoch": 0.02805,
      "grad_norm": 0.6123404104585692,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2805
    },
    {
      "epoch": 0.02806,
      "grad_norm": 0.5813003741795452,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 2806
    },
    {
      "epoch": 0.02807,
      "grad_norm": 0.555658293190141,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 2807
    },
    {
      "epoch": 0.02808,
      "grad_norm": 0.4856792434285324,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 2808
    },
    {
      "epoch": 0.02809,
      "grad_norm": 0.468227745280838,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2809
    },
    {
      "epoch": 0.0281,
      "grad_norm": 0.5953915884658988,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 2810
    },
    {
      "epoch": 0.02811,
      "grad_norm": 0.9133221862875325,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 2811
    },
    {
      "epoch": 0.02812,
      "grad_norm": 1.275925834959494,
      "learning_rate": 0.003,
      "loss": 4.2304,
      "step": 2812
    },
    {
      "epoch": 0.02813,
      "grad_norm": 0.8483651283891129,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2813
    },
    {
      "epoch": 0.02814,
      "grad_norm": 0.7605354432635538,
      "learning_rate": 0.003,
      "loss": 4.202,
      "step": 2814
    },
    {
      "epoch": 0.02815,
      "grad_norm": 0.7567043943015741,
      "learning_rate": 0.003,
      "loss": 4.2385,
      "step": 2815
    },
    {
      "epoch": 0.02816,
      "grad_norm": 0.7448943831936015,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2816
    },
    {
      "epoch": 0.02817,
      "grad_norm": 0.7400121216625585,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2817
    },
    {
      "epoch": 0.02818,
      "grad_norm": 0.7612379755026737,
      "learning_rate": 0.003,
      "loss": 4.1838,
      "step": 2818
    },
    {
      "epoch": 0.02819,
      "grad_norm": 0.8916358831666468,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2819
    },
    {
      "epoch": 0.0282,
      "grad_norm": 0.9386399567045983,
      "learning_rate": 0.003,
      "loss": 4.2191,
      "step": 2820
    },
    {
      "epoch": 0.02821,
      "grad_norm": 0.7968339375438817,
      "learning_rate": 0.003,
      "loss": 4.2369,
      "step": 2821
    },
    {
      "epoch": 0.02822,
      "grad_norm": 0.6707514048211725,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2822
    },
    {
      "epoch": 0.02823,
      "grad_norm": 0.5937548176700269,
      "learning_rate": 0.003,
      "loss": 4.2229,
      "step": 2823
    },
    {
      "epoch": 0.02824,
      "grad_norm": 0.6160995840345511,
      "learning_rate": 0.003,
      "loss": 4.2078,
      "step": 2824
    },
    {
      "epoch": 0.02825,
      "grad_norm": 0.5897186283971302,
      "learning_rate": 0.003,
      "loss": 4.2084,
      "step": 2825
    },
    {
      "epoch": 0.02826,
      "grad_norm": 0.5674030456447229,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 2826
    },
    {
      "epoch": 0.02827,
      "grad_norm": 0.5509162115210677,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 2827
    },
    {
      "epoch": 0.02828,
      "grad_norm": 0.5443211688206793,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 2828
    },
    {
      "epoch": 0.02829,
      "grad_norm": 0.5032331740089974,
      "learning_rate": 0.003,
      "loss": 4.227,
      "step": 2829
    },
    {
      "epoch": 0.0283,
      "grad_norm": 0.5225637902550168,
      "learning_rate": 0.003,
      "loss": 4.1889,
      "step": 2830
    },
    {
      "epoch": 0.02831,
      "grad_norm": 0.5311323075927029,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 2831
    },
    {
      "epoch": 0.02832,
      "grad_norm": 0.5001952573238827,
      "learning_rate": 0.003,
      "loss": 4.1993,
      "step": 2832
    },
    {
      "epoch": 0.02833,
      "grad_norm": 0.466662882413672,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 2833
    },
    {
      "epoch": 0.02834,
      "grad_norm": 0.4772870203488476,
      "learning_rate": 0.003,
      "loss": 4.1575,
      "step": 2834
    },
    {
      "epoch": 0.02835,
      "grad_norm": 0.5535871706800527,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 2835
    },
    {
      "epoch": 0.02836,
      "grad_norm": 0.7239867170041147,
      "learning_rate": 0.003,
      "loss": 4.2004,
      "step": 2836
    },
    {
      "epoch": 0.02837,
      "grad_norm": 0.7597045607127227,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 2837
    },
    {
      "epoch": 0.02838,
      "grad_norm": 0.7252587205241631,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 2838
    },
    {
      "epoch": 0.02839,
      "grad_norm": 0.8402252149672428,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 2839
    },
    {
      "epoch": 0.0284,
      "grad_norm": 0.8241726365065422,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 2840
    },
    {
      "epoch": 0.02841,
      "grad_norm": 0.8096094003742889,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 2841
    },
    {
      "epoch": 0.02842,
      "grad_norm": 0.7156086401419698,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 2842
    },
    {
      "epoch": 0.02843,
      "grad_norm": 0.6040106819890795,
      "learning_rate": 0.003,
      "loss": 4.2061,
      "step": 2843
    },
    {
      "epoch": 0.02844,
      "grad_norm": 0.5060645820894467,
      "learning_rate": 0.003,
      "loss": 4.194,
      "step": 2844
    },
    {
      "epoch": 0.02845,
      "grad_norm": 0.5041938926420962,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 2845
    },
    {
      "epoch": 0.02846,
      "grad_norm": 0.5025147774215044,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 2846
    },
    {
      "epoch": 0.02847,
      "grad_norm": 0.5625355648833802,
      "learning_rate": 0.003,
      "loss": 4.2236,
      "step": 2847
    },
    {
      "epoch": 0.02848,
      "grad_norm": 0.6048155911479377,
      "learning_rate": 0.003,
      "loss": 4.1823,
      "step": 2848
    },
    {
      "epoch": 0.02849,
      "grad_norm": 0.593616598386133,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 2849
    },
    {
      "epoch": 0.0285,
      "grad_norm": 0.5421469421547838,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 2850
    },
    {
      "epoch": 0.02851,
      "grad_norm": 0.5145785466987678,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 2851
    },
    {
      "epoch": 0.02852,
      "grad_norm": 0.5364202493290657,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 2852
    },
    {
      "epoch": 0.02853,
      "grad_norm": 0.5092167845332397,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 2853
    },
    {
      "epoch": 0.02854,
      "grad_norm": 0.505959030155662,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 2854
    },
    {
      "epoch": 0.02855,
      "grad_norm": 0.48727747027775575,
      "learning_rate": 0.003,
      "loss": 4.1836,
      "step": 2855
    },
    {
      "epoch": 0.02856,
      "grad_norm": 0.5114663663985488,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 2856
    },
    {
      "epoch": 0.02857,
      "grad_norm": 0.6029175042652142,
      "learning_rate": 0.003,
      "loss": 4.1703,
      "step": 2857
    },
    {
      "epoch": 0.02858,
      "grad_norm": 0.811258279718228,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 2858
    },
    {
      "epoch": 0.02859,
      "grad_norm": 1.000140796502162,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 2859
    },
    {
      "epoch": 0.0286,
      "grad_norm": 1.2449496760884864,
      "learning_rate": 0.003,
      "loss": 4.2208,
      "step": 2860
    },
    {
      "epoch": 0.02861,
      "grad_norm": 0.7665309921680512,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 2861
    },
    {
      "epoch": 0.02862,
      "grad_norm": 0.7010737782221613,
      "learning_rate": 0.003,
      "loss": 4.1959,
      "step": 2862
    },
    {
      "epoch": 0.02863,
      "grad_norm": 0.6652925807236452,
      "learning_rate": 0.003,
      "loss": 4.216,
      "step": 2863
    },
    {
      "epoch": 0.02864,
      "grad_norm": 0.6660638892873599,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 2864
    },
    {
      "epoch": 0.02865,
      "grad_norm": 0.7030991193559302,
      "learning_rate": 0.003,
      "loss": 4.2231,
      "step": 2865
    },
    {
      "epoch": 0.02866,
      "grad_norm": 0.6597702895413415,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2866
    },
    {
      "epoch": 0.02867,
      "grad_norm": 0.582862490658125,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 2867
    },
    {
      "epoch": 0.02868,
      "grad_norm": 0.5208311845539698,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 2868
    },
    {
      "epoch": 0.02869,
      "grad_norm": 0.5269361881811672,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2869
    },
    {
      "epoch": 0.0287,
      "grad_norm": 0.49588966178431665,
      "learning_rate": 0.003,
      "loss": 4.1982,
      "step": 2870
    },
    {
      "epoch": 0.02871,
      "grad_norm": 0.4427797615796878,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 2871
    },
    {
      "epoch": 0.02872,
      "grad_norm": 0.46354755259563285,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2872
    },
    {
      "epoch": 0.02873,
      "grad_norm": 0.5003897365609394,
      "learning_rate": 0.003,
      "loss": 4.1828,
      "step": 2873
    },
    {
      "epoch": 0.02874,
      "grad_norm": 0.5095674852487324,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2874
    },
    {
      "epoch": 0.02875,
      "grad_norm": 0.5767552596221609,
      "learning_rate": 0.003,
      "loss": 4.2025,
      "step": 2875
    },
    {
      "epoch": 0.02876,
      "grad_norm": 0.7553686836410436,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 2876
    },
    {
      "epoch": 0.02877,
      "grad_norm": 0.927267465496802,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 2877
    },
    {
      "epoch": 0.02878,
      "grad_norm": 0.8275286861154829,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 2878
    },
    {
      "epoch": 0.02879,
      "grad_norm": 0.6254369579498115,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 2879
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.6608202030691483,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2880
    },
    {
      "epoch": 0.02881,
      "grad_norm": 0.6264026993054118,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2881
    },
    {
      "epoch": 0.02882,
      "grad_norm": 0.6456049764969926,
      "learning_rate": 0.003,
      "loss": 4.2046,
      "step": 2882
    },
    {
      "epoch": 0.02883,
      "grad_norm": 0.631140439713875,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2883
    },
    {
      "epoch": 0.02884,
      "grad_norm": 0.6167258778841849,
      "learning_rate": 0.003,
      "loss": 4.2212,
      "step": 2884
    },
    {
      "epoch": 0.02885,
      "grad_norm": 0.6237538943365489,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 2885
    },
    {
      "epoch": 0.02886,
      "grad_norm": 0.577915341916395,
      "learning_rate": 0.003,
      "loss": 4.1922,
      "step": 2886
    },
    {
      "epoch": 0.02887,
      "grad_norm": 0.5869716498143316,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2887
    },
    {
      "epoch": 0.02888,
      "grad_norm": 0.6367834629014674,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 2888
    },
    {
      "epoch": 0.02889,
      "grad_norm": 0.6424828882281004,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 2889
    },
    {
      "epoch": 0.0289,
      "grad_norm": 0.7456946160039144,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 2890
    },
    {
      "epoch": 0.02891,
      "grad_norm": 0.8611032962542791,
      "learning_rate": 0.003,
      "loss": 4.2268,
      "step": 2891
    },
    {
      "epoch": 0.02892,
      "grad_norm": 1.0298003752084743,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 2892
    },
    {
      "epoch": 0.02893,
      "grad_norm": 1.0054991154141122,
      "learning_rate": 0.003,
      "loss": 4.1976,
      "step": 2893
    },
    {
      "epoch": 0.02894,
      "grad_norm": 0.8964526887398983,
      "learning_rate": 0.003,
      "loss": 4.2137,
      "step": 2894
    },
    {
      "epoch": 0.02895,
      "grad_norm": 0.786662499637371,
      "learning_rate": 0.003,
      "loss": 4.2494,
      "step": 2895
    },
    {
      "epoch": 0.02896,
      "grad_norm": 0.8924645045032006,
      "learning_rate": 0.003,
      "loss": 4.2136,
      "step": 2896
    },
    {
      "epoch": 0.02897,
      "grad_norm": 0.9444560121857192,
      "learning_rate": 0.003,
      "loss": 4.1979,
      "step": 2897
    },
    {
      "epoch": 0.02898,
      "grad_norm": 0.9041547595653124,
      "learning_rate": 0.003,
      "loss": 4.2475,
      "step": 2898
    },
    {
      "epoch": 0.02899,
      "grad_norm": 0.8479058583034529,
      "learning_rate": 0.003,
      "loss": 4.2306,
      "step": 2899
    },
    {
      "epoch": 0.029,
      "grad_norm": 0.8224886244859558,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2900
    },
    {
      "epoch": 0.02901,
      "grad_norm": 0.8698134304184088,
      "learning_rate": 0.003,
      "loss": 4.2375,
      "step": 2901
    },
    {
      "epoch": 0.02902,
      "grad_norm": 0.8486827254456236,
      "learning_rate": 0.003,
      "loss": 4.2524,
      "step": 2902
    },
    {
      "epoch": 0.02903,
      "grad_norm": 0.7415128492690074,
      "learning_rate": 0.003,
      "loss": 4.2059,
      "step": 2903
    },
    {
      "epoch": 0.02904,
      "grad_norm": 0.7083318227885205,
      "learning_rate": 0.003,
      "loss": 4.2241,
      "step": 2904
    },
    {
      "epoch": 0.02905,
      "grad_norm": 0.7648353967865369,
      "learning_rate": 0.003,
      "loss": 4.2383,
      "step": 2905
    },
    {
      "epoch": 0.02906,
      "grad_norm": 0.7593644185333667,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 2906
    },
    {
      "epoch": 0.02907,
      "grad_norm": 0.7189838848454702,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 2907
    },
    {
      "epoch": 0.02908,
      "grad_norm": 0.7101023301849051,
      "learning_rate": 0.003,
      "loss": 4.2105,
      "step": 2908
    },
    {
      "epoch": 0.02909,
      "grad_norm": 0.6273709094950615,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 2909
    },
    {
      "epoch": 0.0291,
      "grad_norm": 0.5304669195124173,
      "learning_rate": 0.003,
      "loss": 4.2082,
      "step": 2910
    },
    {
      "epoch": 0.02911,
      "grad_norm": 0.509577660549234,
      "learning_rate": 0.003,
      "loss": 4.2021,
      "step": 2911
    },
    {
      "epoch": 0.02912,
      "grad_norm": 0.4255206741887875,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 2912
    },
    {
      "epoch": 0.02913,
      "grad_norm": 0.4561469328827923,
      "learning_rate": 0.003,
      "loss": 4.1998,
      "step": 2913
    },
    {
      "epoch": 0.02914,
      "grad_norm": 0.42117739519724395,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 2914
    },
    {
      "epoch": 0.02915,
      "grad_norm": 0.4011573145594706,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 2915
    },
    {
      "epoch": 0.02916,
      "grad_norm": 0.42970374324262517,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 2916
    },
    {
      "epoch": 0.02917,
      "grad_norm": 0.33514332187348916,
      "learning_rate": 0.003,
      "loss": 4.2087,
      "step": 2917
    },
    {
      "epoch": 0.02918,
      "grad_norm": 0.34222241717451907,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 2918
    },
    {
      "epoch": 0.02919,
      "grad_norm": 0.36583286994162584,
      "learning_rate": 0.003,
      "loss": 4.2045,
      "step": 2919
    },
    {
      "epoch": 0.0292,
      "grad_norm": 0.4446133039291897,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 2920
    },
    {
      "epoch": 0.02921,
      "grad_norm": 0.6109713288069767,
      "learning_rate": 0.003,
      "loss": 4.1869,
      "step": 2921
    },
    {
      "epoch": 0.02922,
      "grad_norm": 0.9167859634758244,
      "learning_rate": 0.003,
      "loss": 4.1961,
      "step": 2922
    },
    {
      "epoch": 0.02923,
      "grad_norm": 1.009808447907654,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 2923
    },
    {
      "epoch": 0.02924,
      "grad_norm": 0.8524826722350543,
      "learning_rate": 0.003,
      "loss": 4.2159,
      "step": 2924
    },
    {
      "epoch": 0.02925,
      "grad_norm": 0.8032284343679224,
      "learning_rate": 0.003,
      "loss": 4.2199,
      "step": 2925
    },
    {
      "epoch": 0.02926,
      "grad_norm": 0.749692494912656,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2926
    },
    {
      "epoch": 0.02927,
      "grad_norm": 0.7088429283617566,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 2927
    },
    {
      "epoch": 0.02928,
      "grad_norm": 0.6042074975753539,
      "learning_rate": 0.003,
      "loss": 4.2151,
      "step": 2928
    },
    {
      "epoch": 0.02929,
      "grad_norm": 0.563152730036225,
      "learning_rate": 0.003,
      "loss": 4.2,
      "step": 2929
    },
    {
      "epoch": 0.0293,
      "grad_norm": 0.44275110512836524,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 2930
    },
    {
      "epoch": 0.02931,
      "grad_norm": 0.4579944440403804,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 2931
    },
    {
      "epoch": 0.02932,
      "grad_norm": 0.4515235240773822,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 2932
    },
    {
      "epoch": 0.02933,
      "grad_norm": 0.46401883936201904,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2933
    },
    {
      "epoch": 0.02934,
      "grad_norm": 0.4813031783733538,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 2934
    },
    {
      "epoch": 0.02935,
      "grad_norm": 0.4967851984899721,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 2935
    },
    {
      "epoch": 0.02936,
      "grad_norm": 0.5320015979752634,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 2936
    },
    {
      "epoch": 0.02937,
      "grad_norm": 0.5208542547284235,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 2937
    },
    {
      "epoch": 0.02938,
      "grad_norm": 0.4869610046256959,
      "learning_rate": 0.003,
      "loss": 4.1841,
      "step": 2938
    },
    {
      "epoch": 0.02939,
      "grad_norm": 0.43982195879012664,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 2939
    },
    {
      "epoch": 0.0294,
      "grad_norm": 0.43165943569371323,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2940
    },
    {
      "epoch": 0.02941,
      "grad_norm": 0.42317517860985626,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2941
    },
    {
      "epoch": 0.02942,
      "grad_norm": 0.46023364048218307,
      "learning_rate": 0.003,
      "loss": 4.1756,
      "step": 2942
    },
    {
      "epoch": 0.02943,
      "grad_norm": 0.5385981811058449,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 2943
    },
    {
      "epoch": 0.02944,
      "grad_norm": 0.7649280046904252,
      "learning_rate": 0.003,
      "loss": 4.1965,
      "step": 2944
    },
    {
      "epoch": 0.02945,
      "grad_norm": 1.1539036490870935,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 2945
    },
    {
      "epoch": 0.02946,
      "grad_norm": 0.9395984374200569,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 2946
    },
    {
      "epoch": 0.02947,
      "grad_norm": 0.8080599605170256,
      "learning_rate": 0.003,
      "loss": 4.224,
      "step": 2947
    },
    {
      "epoch": 0.02948,
      "grad_norm": 0.7434332174406536,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 2948
    },
    {
      "epoch": 0.02949,
      "grad_norm": 0.7928578212224346,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 2949
    },
    {
      "epoch": 0.0295,
      "grad_norm": 0.8333301095683224,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 2950
    },
    {
      "epoch": 0.02951,
      "grad_norm": 0.9482608719829599,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 2951
    },
    {
      "epoch": 0.02952,
      "grad_norm": 0.8328172653762101,
      "learning_rate": 0.003,
      "loss": 4.1962,
      "step": 2952
    },
    {
      "epoch": 0.02953,
      "grad_norm": 0.8348300580576069,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 2953
    },
    {
      "epoch": 0.02954,
      "grad_norm": 0.8191965718315765,
      "learning_rate": 0.003,
      "loss": 4.2499,
      "step": 2954
    },
    {
      "epoch": 0.02955,
      "grad_norm": 0.8592579474886709,
      "learning_rate": 0.003,
      "loss": 4.2234,
      "step": 2955
    },
    {
      "epoch": 0.02956,
      "grad_norm": 0.768153733637731,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 2956
    },
    {
      "epoch": 0.02957,
      "grad_norm": 0.619447070815721,
      "learning_rate": 0.003,
      "loss": 4.2157,
      "step": 2957
    },
    {
      "epoch": 0.02958,
      "grad_norm": 0.6117541520949576,
      "learning_rate": 0.003,
      "loss": 4.2284,
      "step": 2958
    },
    {
      "epoch": 0.02959,
      "grad_norm": 0.5765518020385798,
      "learning_rate": 0.003,
      "loss": 4.2146,
      "step": 2959
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.5472369901466232,
      "learning_rate": 0.003,
      "loss": 4.2178,
      "step": 2960
    },
    {
      "epoch": 0.02961,
      "grad_norm": 0.5697675686509917,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 2961
    },
    {
      "epoch": 0.02962,
      "grad_norm": 0.5322939405778847,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 2962
    },
    {
      "epoch": 0.02963,
      "grad_norm": 0.4995062550237162,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 2963
    },
    {
      "epoch": 0.02964,
      "grad_norm": 0.48408735290928806,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 2964
    },
    {
      "epoch": 0.02965,
      "grad_norm": 0.49796931779139914,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 2965
    },
    {
      "epoch": 0.02966,
      "grad_norm": 0.49309681218010465,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 2966
    },
    {
      "epoch": 0.02967,
      "grad_norm": 0.46885663161103275,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 2967
    },
    {
      "epoch": 0.02968,
      "grad_norm": 0.4307711779753192,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 2968
    },
    {
      "epoch": 0.02969,
      "grad_norm": 0.38466337051238414,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 2969
    },
    {
      "epoch": 0.0297,
      "grad_norm": 0.46985591464325377,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 2970
    },
    {
      "epoch": 0.02971,
      "grad_norm": 0.6319423638686542,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 2971
    },
    {
      "epoch": 0.02972,
      "grad_norm": 0.8754437886538992,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 2972
    },
    {
      "epoch": 0.02973,
      "grad_norm": 1.020941685436833,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 2973
    },
    {
      "epoch": 0.02974,
      "grad_norm": 0.6824462354167666,
      "learning_rate": 0.003,
      "loss": 4.193,
      "step": 2974
    },
    {
      "epoch": 0.02975,
      "grad_norm": 0.5456820041365307,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 2975
    },
    {
      "epoch": 0.02976,
      "grad_norm": 0.7254587394041584,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 2976
    },
    {
      "epoch": 0.02977,
      "grad_norm": 0.6896012196641018,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 2977
    },
    {
      "epoch": 0.02978,
      "grad_norm": 0.6446956558129,
      "learning_rate": 0.003,
      "loss": 4.1956,
      "step": 2978
    },
    {
      "epoch": 0.02979,
      "grad_norm": 0.5613291446841128,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 2979
    },
    {
      "epoch": 0.0298,
      "grad_norm": 0.578998122498601,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 2980
    },
    {
      "epoch": 0.02981,
      "grad_norm": 0.6195176572378017,
      "learning_rate": 0.003,
      "loss": 4.1881,
      "step": 2981
    },
    {
      "epoch": 0.02982,
      "grad_norm": 0.6240562576963727,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 2982
    },
    {
      "epoch": 0.02983,
      "grad_norm": 0.5968266153844792,
      "learning_rate": 0.003,
      "loss": 4.1978,
      "step": 2983
    },
    {
      "epoch": 0.02984,
      "grad_norm": 0.5306795307358292,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 2984
    },
    {
      "epoch": 0.02985,
      "grad_norm": 0.5463323846035709,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 2985
    },
    {
      "epoch": 0.02986,
      "grad_norm": 0.5421781401610759,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 2986
    },
    {
      "epoch": 0.02987,
      "grad_norm": 0.5293611287733503,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 2987
    },
    {
      "epoch": 0.02988,
      "grad_norm": 0.6408997854011281,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 2988
    },
    {
      "epoch": 0.02989,
      "grad_norm": 0.7426268122761913,
      "learning_rate": 0.003,
      "loss": 4.1915,
      "step": 2989
    },
    {
      "epoch": 0.0299,
      "grad_norm": 0.7556899506663288,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 2990
    },
    {
      "epoch": 0.02991,
      "grad_norm": 0.6415270525884926,
      "learning_rate": 0.003,
      "loss": 4.1985,
      "step": 2991
    },
    {
      "epoch": 0.02992,
      "grad_norm": 0.6368634341681172,
      "learning_rate": 0.003,
      "loss": 4.1951,
      "step": 2992
    },
    {
      "epoch": 0.02993,
      "grad_norm": 0.6410538934602643,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 2993
    },
    {
      "epoch": 0.02994,
      "grad_norm": 0.6332716830387952,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 2994
    },
    {
      "epoch": 0.02995,
      "grad_norm": 0.7865491161110946,
      "learning_rate": 0.003,
      "loss": 4.2175,
      "step": 2995
    },
    {
      "epoch": 0.02996,
      "grad_norm": 0.9230963051423904,
      "learning_rate": 0.003,
      "loss": 4.2133,
      "step": 2996
    },
    {
      "epoch": 0.02997,
      "grad_norm": 1.1769719664750078,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 2997
    },
    {
      "epoch": 0.02998,
      "grad_norm": 0.965735521679244,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 2998
    },
    {
      "epoch": 0.02999,
      "grad_norm": 0.9025433565090871,
      "learning_rate": 0.003,
      "loss": 4.1874,
      "step": 2999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9697259711524333,
      "learning_rate": 0.003,
      "loss": 4.2399,
      "step": 3000
    },
    {
      "epoch": 0.03001,
      "grad_norm": 0.9254213834179852,
      "learning_rate": 0.003,
      "loss": 4.2091,
      "step": 3001
    },
    {
      "epoch": 0.03002,
      "grad_norm": 0.8893198223091714,
      "learning_rate": 0.003,
      "loss": 4.225,
      "step": 3002
    },
    {
      "epoch": 0.03003,
      "grad_norm": 0.7192875473487559,
      "learning_rate": 0.003,
      "loss": 4.197,
      "step": 3003
    },
    {
      "epoch": 0.03004,
      "grad_norm": 0.7886569353807684,
      "learning_rate": 0.003,
      "loss": 4.1778,
      "step": 3004
    },
    {
      "epoch": 0.03005,
      "grad_norm": 0.8011423250434415,
      "learning_rate": 0.003,
      "loss": 4.2294,
      "step": 3005
    },
    {
      "epoch": 0.03006,
      "grad_norm": 0.7535404952002084,
      "learning_rate": 0.003,
      "loss": 4.2201,
      "step": 3006
    },
    {
      "epoch": 0.03007,
      "grad_norm": 0.7314424724357438,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3007
    },
    {
      "epoch": 0.03008,
      "grad_norm": 0.6096876125818365,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 3008
    },
    {
      "epoch": 0.03009,
      "grad_norm": 0.5575234963619043,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3009
    },
    {
      "epoch": 0.0301,
      "grad_norm": 0.5598489300006476,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 3010
    },
    {
      "epoch": 0.03011,
      "grad_norm": 0.7310511801710661,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 3011
    },
    {
      "epoch": 0.03012,
      "grad_norm": 0.8536268304907085,
      "learning_rate": 0.003,
      "loss": 4.2002,
      "step": 3012
    },
    {
      "epoch": 0.03013,
      "grad_norm": 0.9271684945658873,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3013
    },
    {
      "epoch": 0.03014,
      "grad_norm": 0.748204257953396,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3014
    },
    {
      "epoch": 0.03015,
      "grad_norm": 0.7294374875175993,
      "learning_rate": 0.003,
      "loss": 4.1879,
      "step": 3015
    },
    {
      "epoch": 0.03016,
      "grad_norm": 0.8462639497588901,
      "learning_rate": 0.003,
      "loss": 4.204,
      "step": 3016
    },
    {
      "epoch": 0.03017,
      "grad_norm": 0.6499456234415616,
      "learning_rate": 0.003,
      "loss": 4.189,
      "step": 3017
    },
    {
      "epoch": 0.03018,
      "grad_norm": 0.5875121164816828,
      "learning_rate": 0.003,
      "loss": 4.1824,
      "step": 3018
    },
    {
      "epoch": 0.03019,
      "grad_norm": 0.47508344745103015,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3019
    },
    {
      "epoch": 0.0302,
      "grad_norm": 0.5142366251911704,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3020
    },
    {
      "epoch": 0.03021,
      "grad_norm": 0.5290255986656376,
      "learning_rate": 0.003,
      "loss": 4.2069,
      "step": 3021
    },
    {
      "epoch": 0.03022,
      "grad_norm": 0.5261392358925515,
      "learning_rate": 0.003,
      "loss": 4.2026,
      "step": 3022
    },
    {
      "epoch": 0.03023,
      "grad_norm": 0.5585956938981086,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 3023
    },
    {
      "epoch": 0.03024,
      "grad_norm": 0.6114817871124547,
      "learning_rate": 0.003,
      "loss": 4.2096,
      "step": 3024
    },
    {
      "epoch": 0.03025,
      "grad_norm": 0.6824857482246931,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3025
    },
    {
      "epoch": 0.03026,
      "grad_norm": 0.7984707065148595,
      "learning_rate": 0.003,
      "loss": 4.1723,
      "step": 3026
    },
    {
      "epoch": 0.03027,
      "grad_norm": 0.7174251636707903,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3027
    },
    {
      "epoch": 0.03028,
      "grad_norm": 0.7106059576910744,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3028
    },
    {
      "epoch": 0.03029,
      "grad_norm": 0.6122719077257656,
      "learning_rate": 0.003,
      "loss": 4.1927,
      "step": 3029
    },
    {
      "epoch": 0.0303,
      "grad_norm": 0.5305954769529888,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3030
    },
    {
      "epoch": 0.03031,
      "grad_norm": 0.4997654568254282,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3031
    },
    {
      "epoch": 0.03032,
      "grad_norm": 0.4912614975804802,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3032
    },
    {
      "epoch": 0.03033,
      "grad_norm": 0.4920559211009805,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3033
    },
    {
      "epoch": 0.03034,
      "grad_norm": 0.5605701380054549,
      "learning_rate": 0.003,
      "loss": 4.1747,
      "step": 3034
    },
    {
      "epoch": 0.03035,
      "grad_norm": 0.6077583896543486,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 3035
    },
    {
      "epoch": 0.03036,
      "grad_norm": 0.7202855019986794,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3036
    },
    {
      "epoch": 0.03037,
      "grad_norm": 0.6575890137997805,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 3037
    },
    {
      "epoch": 0.03038,
      "grad_norm": 0.46568815919813755,
      "learning_rate": 0.003,
      "loss": 4.1585,
      "step": 3038
    },
    {
      "epoch": 0.03039,
      "grad_norm": 0.49641007446862917,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3039
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.559012034381986,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3040
    },
    {
      "epoch": 0.03041,
      "grad_norm": 0.5533954561872954,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3041
    },
    {
      "epoch": 0.03042,
      "grad_norm": 0.47365032322165673,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 3042
    },
    {
      "epoch": 0.03043,
      "grad_norm": 0.4920139603559505,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3043
    },
    {
      "epoch": 0.03044,
      "grad_norm": 0.48267895485751067,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 3044
    },
    {
      "epoch": 0.03045,
      "grad_norm": 0.40892249484891385,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3045
    },
    {
      "epoch": 0.03046,
      "grad_norm": 0.5179956198309833,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3046
    },
    {
      "epoch": 0.03047,
      "grad_norm": 0.6458691485842255,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3047
    },
    {
      "epoch": 0.03048,
      "grad_norm": 0.805749623290173,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3048
    },
    {
      "epoch": 0.03049,
      "grad_norm": 0.8776696886409071,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3049
    },
    {
      "epoch": 0.0305,
      "grad_norm": 0.8822405330269477,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 3050
    },
    {
      "epoch": 0.03051,
      "grad_norm": 0.9277197792519675,
      "learning_rate": 0.003,
      "loss": 4.1966,
      "step": 3051
    },
    {
      "epoch": 0.03052,
      "grad_norm": 0.9287607223949175,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3052
    },
    {
      "epoch": 0.03053,
      "grad_norm": 0.8621618117255709,
      "learning_rate": 0.003,
      "loss": 4.2095,
      "step": 3053
    },
    {
      "epoch": 0.03054,
      "grad_norm": 0.9192475292904252,
      "learning_rate": 0.003,
      "loss": 4.1952,
      "step": 3054
    },
    {
      "epoch": 0.03055,
      "grad_norm": 0.9250509957208142,
      "learning_rate": 0.003,
      "loss": 4.2062,
      "step": 3055
    },
    {
      "epoch": 0.03056,
      "grad_norm": 0.9134243567876251,
      "learning_rate": 0.003,
      "loss": 4.217,
      "step": 3056
    },
    {
      "epoch": 0.03057,
      "grad_norm": 1.0483835943699353,
      "learning_rate": 0.003,
      "loss": 4.2258,
      "step": 3057
    },
    {
      "epoch": 0.03058,
      "grad_norm": 1.0909797627134183,
      "learning_rate": 0.003,
      "loss": 4.2072,
      "step": 3058
    },
    {
      "epoch": 0.03059,
      "grad_norm": 1.0929367898362812,
      "learning_rate": 0.003,
      "loss": 4.2194,
      "step": 3059
    },
    {
      "epoch": 0.0306,
      "grad_norm": 0.9499317872652826,
      "learning_rate": 0.003,
      "loss": 4.2104,
      "step": 3060
    },
    {
      "epoch": 0.03061,
      "grad_norm": 0.7530643446613368,
      "learning_rate": 0.003,
      "loss": 4.227,
      "step": 3061
    },
    {
      "epoch": 0.03062,
      "grad_norm": 0.7822441045416462,
      "learning_rate": 0.003,
      "loss": 4.25,
      "step": 3062
    },
    {
      "epoch": 0.03063,
      "grad_norm": 0.7991934606488392,
      "learning_rate": 0.003,
      "loss": 4.1973,
      "step": 3063
    },
    {
      "epoch": 0.03064,
      "grad_norm": 0.7614903963432643,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 3064
    },
    {
      "epoch": 0.03065,
      "grad_norm": 0.6047999858157871,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3065
    },
    {
      "epoch": 0.03066,
      "grad_norm": 0.6254938393897825,
      "learning_rate": 0.003,
      "loss": 4.231,
      "step": 3066
    },
    {
      "epoch": 0.03067,
      "grad_norm": 0.5333801571885867,
      "learning_rate": 0.003,
      "loss": 4.2275,
      "step": 3067
    },
    {
      "epoch": 0.03068,
      "grad_norm": 0.47159284158763254,
      "learning_rate": 0.003,
      "loss": 4.1942,
      "step": 3068
    },
    {
      "epoch": 0.03069,
      "grad_norm": 0.4095985602684375,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3069
    },
    {
      "epoch": 0.0307,
      "grad_norm": 0.4380075531913516,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 3070
    },
    {
      "epoch": 0.03071,
      "grad_norm": 0.520842979900822,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 3071
    },
    {
      "epoch": 0.03072,
      "grad_norm": 0.6276268407397474,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3072
    },
    {
      "epoch": 0.03073,
      "grad_norm": 0.7317184572201801,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3073
    },
    {
      "epoch": 0.03074,
      "grad_norm": 0.7701795105904193,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3074
    },
    {
      "epoch": 0.03075,
      "grad_norm": 0.6957977102957552,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3075
    },
    {
      "epoch": 0.03076,
      "grad_norm": 0.6144911834647919,
      "learning_rate": 0.003,
      "loss": 4.2033,
      "step": 3076
    },
    {
      "epoch": 0.03077,
      "grad_norm": 0.692008332614251,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 3077
    },
    {
      "epoch": 0.03078,
      "grad_norm": 0.6946737312829024,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 3078
    },
    {
      "epoch": 0.03079,
      "grad_norm": 0.7552084198206422,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3079
    },
    {
      "epoch": 0.0308,
      "grad_norm": 0.7025348126305976,
      "learning_rate": 0.003,
      "loss": 4.203,
      "step": 3080
    },
    {
      "epoch": 0.03081,
      "grad_norm": 0.5947816851076059,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3081
    },
    {
      "epoch": 0.03082,
      "grad_norm": 0.6508903175127416,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3082
    },
    {
      "epoch": 0.03083,
      "grad_norm": 0.8364672171654759,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3083
    },
    {
      "epoch": 0.03084,
      "grad_norm": 0.7270205647692827,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3084
    },
    {
      "epoch": 0.03085,
      "grad_norm": 0.5251957082188036,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 3085
    },
    {
      "epoch": 0.03086,
      "grad_norm": 0.4928794972094713,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 3086
    },
    {
      "epoch": 0.03087,
      "grad_norm": 0.4574857190743631,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3087
    },
    {
      "epoch": 0.03088,
      "grad_norm": 0.48967300167720096,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3088
    },
    {
      "epoch": 0.03089,
      "grad_norm": 0.5020327161233649,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3089
    },
    {
      "epoch": 0.0309,
      "grad_norm": 0.5318564623713369,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3090
    },
    {
      "epoch": 0.03091,
      "grad_norm": 0.5321720354076895,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 3091
    },
    {
      "epoch": 0.03092,
      "grad_norm": 0.45641360749571647,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 3092
    },
    {
      "epoch": 0.03093,
      "grad_norm": 0.40991320622723604,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3093
    },
    {
      "epoch": 0.03094,
      "grad_norm": 0.4792246990497778,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3094
    },
    {
      "epoch": 0.03095,
      "grad_norm": 0.5559515338310668,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3095
    },
    {
      "epoch": 0.03096,
      "grad_norm": 0.7462971856723287,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 3096
    },
    {
      "epoch": 0.03097,
      "grad_norm": 0.9002817455009458,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3097
    },
    {
      "epoch": 0.03098,
      "grad_norm": 0.8351003992845788,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3098
    },
    {
      "epoch": 0.03099,
      "grad_norm": 0.5935585890716526,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3099
    },
    {
      "epoch": 0.031,
      "grad_norm": 0.6353446062973269,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3100
    },
    {
      "epoch": 0.03101,
      "grad_norm": 0.7592462024173295,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3101
    },
    {
      "epoch": 0.03102,
      "grad_norm": 0.8288461234391342,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3102
    },
    {
      "epoch": 0.03103,
      "grad_norm": 0.8186091723896449,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3103
    },
    {
      "epoch": 0.03104,
      "grad_norm": 0.730911001668181,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3104
    },
    {
      "epoch": 0.03105,
      "grad_norm": 0.6266459445253275,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 3105
    },
    {
      "epoch": 0.03106,
      "grad_norm": 0.6545747432475193,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3106
    },
    {
      "epoch": 0.03107,
      "grad_norm": 0.5231672698914397,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 3107
    },
    {
      "epoch": 0.03108,
      "grad_norm": 0.5734245951904136,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 3108
    },
    {
      "epoch": 0.03109,
      "grad_norm": 0.5288754437005536,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3109
    },
    {
      "epoch": 0.0311,
      "grad_norm": 0.5320691683351324,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 3110
    },
    {
      "epoch": 0.03111,
      "grad_norm": 0.5370642425053924,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3111
    },
    {
      "epoch": 0.03112,
      "grad_norm": 0.49890497665091543,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3112
    },
    {
      "epoch": 0.03113,
      "grad_norm": 0.4336869241892021,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3113
    },
    {
      "epoch": 0.03114,
      "grad_norm": 0.45513880225763054,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3114
    },
    {
      "epoch": 0.03115,
      "grad_norm": 0.5541487651615646,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3115
    },
    {
      "epoch": 0.03116,
      "grad_norm": 0.6085416142601979,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3116
    },
    {
      "epoch": 0.03117,
      "grad_norm": 0.6289436131363786,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3117
    },
    {
      "epoch": 0.03118,
      "grad_norm": 0.7648349803443671,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3118
    },
    {
      "epoch": 0.03119,
      "grad_norm": 0.9905397386097399,
      "learning_rate": 0.003,
      "loss": 4.1975,
      "step": 3119
    },
    {
      "epoch": 0.0312,
      "grad_norm": 1.002027828062209,
      "learning_rate": 0.003,
      "loss": 4.2085,
      "step": 3120
    },
    {
      "epoch": 0.03121,
      "grad_norm": 0.9945235603188975,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3121
    },
    {
      "epoch": 0.03122,
      "grad_norm": 0.9749277139432609,
      "learning_rate": 0.003,
      "loss": 4.2158,
      "step": 3122
    },
    {
      "epoch": 0.03123,
      "grad_norm": 0.9176347284109458,
      "learning_rate": 0.003,
      "loss": 4.2123,
      "step": 3123
    },
    {
      "epoch": 0.03124,
      "grad_norm": 0.8222379595933585,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3124
    },
    {
      "epoch": 0.03125,
      "grad_norm": 0.8029594745648053,
      "learning_rate": 0.003,
      "loss": 4.1886,
      "step": 3125
    },
    {
      "epoch": 0.03126,
      "grad_norm": 0.7973924670105094,
      "learning_rate": 0.003,
      "loss": 4.1935,
      "step": 3126
    },
    {
      "epoch": 0.03127,
      "grad_norm": 0.8183339868293253,
      "learning_rate": 0.003,
      "loss": 4.2012,
      "step": 3127
    },
    {
      "epoch": 0.03128,
      "grad_norm": 0.8564648965676082,
      "learning_rate": 0.003,
      "loss": 4.1971,
      "step": 3128
    },
    {
      "epoch": 0.03129,
      "grad_norm": 0.6856535056536724,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 3129
    },
    {
      "epoch": 0.0313,
      "grad_norm": 0.6641053792145878,
      "learning_rate": 0.003,
      "loss": 4.2086,
      "step": 3130
    },
    {
      "epoch": 0.03131,
      "grad_norm": 0.5479229508159585,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3131
    },
    {
      "epoch": 0.03132,
      "grad_norm": 0.5311614836536821,
      "learning_rate": 0.003,
      "loss": 4.2024,
      "step": 3132
    },
    {
      "epoch": 0.03133,
      "grad_norm": 0.5105588102398402,
      "learning_rate": 0.003,
      "loss": 4.1908,
      "step": 3133
    },
    {
      "epoch": 0.03134,
      "grad_norm": 0.5009888970015204,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3134
    },
    {
      "epoch": 0.03135,
      "grad_norm": 0.4297510332483796,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3135
    },
    {
      "epoch": 0.03136,
      "grad_norm": 0.4354586739325799,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3136
    },
    {
      "epoch": 0.03137,
      "grad_norm": 0.5083671646646589,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3137
    },
    {
      "epoch": 0.03138,
      "grad_norm": 0.6199429454274028,
      "learning_rate": 0.003,
      "loss": 4.1701,
      "step": 3138
    },
    {
      "epoch": 0.03139,
      "grad_norm": 0.7271551302730402,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3139
    },
    {
      "epoch": 0.0314,
      "grad_norm": 0.7358854576257201,
      "learning_rate": 0.003,
      "loss": 4.1788,
      "step": 3140
    },
    {
      "epoch": 0.03141,
      "grad_norm": 0.6326154627482451,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 3141
    },
    {
      "epoch": 0.03142,
      "grad_norm": 0.5588364977660311,
      "learning_rate": 0.003,
      "loss": 4.183,
      "step": 3142
    },
    {
      "epoch": 0.03143,
      "grad_norm": 0.6311591083308525,
      "learning_rate": 0.003,
      "loss": 4.1728,
      "step": 3143
    },
    {
      "epoch": 0.03144,
      "grad_norm": 0.5392546760146013,
      "learning_rate": 0.003,
      "loss": 4.1371,
      "step": 3144
    },
    {
      "epoch": 0.03145,
      "grad_norm": 0.4557888424936214,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3145
    },
    {
      "epoch": 0.03146,
      "grad_norm": 0.4216370024975091,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 3146
    },
    {
      "epoch": 0.03147,
      "grad_norm": 0.4295618903624941,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3147
    },
    {
      "epoch": 0.03148,
      "grad_norm": 0.4097670324789773,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3148
    },
    {
      "epoch": 0.03149,
      "grad_norm": 0.4802416636246548,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 3149
    },
    {
      "epoch": 0.0315,
      "grad_norm": 0.46441368907739394,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 3150
    },
    {
      "epoch": 0.03151,
      "grad_norm": 0.44059093730228843,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3151
    },
    {
      "epoch": 0.03152,
      "grad_norm": 0.4172894456146658,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3152
    },
    {
      "epoch": 0.03153,
      "grad_norm": 0.42959521977879306,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 3153
    },
    {
      "epoch": 0.03154,
      "grad_norm": 0.43398804524875734,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3154
    },
    {
      "epoch": 0.03155,
      "grad_norm": 0.4562270589856803,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3155
    },
    {
      "epoch": 0.03156,
      "grad_norm": 0.5126486371492089,
      "learning_rate": 0.003,
      "loss": 4.142,
      "step": 3156
    },
    {
      "epoch": 0.03157,
      "grad_norm": 0.6526844959193616,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3157
    },
    {
      "epoch": 0.03158,
      "grad_norm": 0.9480089169547891,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3158
    },
    {
      "epoch": 0.03159,
      "grad_norm": 1.3561889811332835,
      "learning_rate": 0.003,
      "loss": 4.1997,
      "step": 3159
    },
    {
      "epoch": 0.0316,
      "grad_norm": 0.5982846825437411,
      "learning_rate": 0.003,
      "loss": 4.1882,
      "step": 3160
    },
    {
      "epoch": 0.03161,
      "grad_norm": 0.8732729212340313,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3161
    },
    {
      "epoch": 0.03162,
      "grad_norm": 0.886847324325859,
      "learning_rate": 0.003,
      "loss": 4.2169,
      "step": 3162
    },
    {
      "epoch": 0.03163,
      "grad_norm": 0.99270442281753,
      "learning_rate": 0.003,
      "loss": 4.2108,
      "step": 3163
    },
    {
      "epoch": 0.03164,
      "grad_norm": 0.9630909842912381,
      "learning_rate": 0.003,
      "loss": 4.2102,
      "step": 3164
    },
    {
      "epoch": 0.03165,
      "grad_norm": 0.7916297793185595,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3165
    },
    {
      "epoch": 0.03166,
      "grad_norm": 0.6926990051674062,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 3166
    },
    {
      "epoch": 0.03167,
      "grad_norm": 0.681279202527116,
      "learning_rate": 0.003,
      "loss": 4.1779,
      "step": 3167
    },
    {
      "epoch": 0.03168,
      "grad_norm": 0.6889533564228932,
      "learning_rate": 0.003,
      "loss": 4.2042,
      "step": 3168
    },
    {
      "epoch": 0.03169,
      "grad_norm": 0.8396344102094035,
      "learning_rate": 0.003,
      "loss": 4.2217,
      "step": 3169
    },
    {
      "epoch": 0.0317,
      "grad_norm": 0.9428653644577029,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 3170
    },
    {
      "epoch": 0.03171,
      "grad_norm": 0.691282163039264,
      "learning_rate": 0.003,
      "loss": 4.2143,
      "step": 3171
    },
    {
      "epoch": 0.03172,
      "grad_norm": 0.665795792414971,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 3172
    },
    {
      "epoch": 0.03173,
      "grad_norm": 0.6751127918552098,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3173
    },
    {
      "epoch": 0.03174,
      "grad_norm": 0.6066754986654972,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 3174
    },
    {
      "epoch": 0.03175,
      "grad_norm": 0.5400765945883416,
      "learning_rate": 0.003,
      "loss": 4.2066,
      "step": 3175
    },
    {
      "epoch": 0.03176,
      "grad_norm": 0.5542770572580099,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3176
    },
    {
      "epoch": 0.03177,
      "grad_norm": 0.5967671493496074,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3177
    },
    {
      "epoch": 0.03178,
      "grad_norm": 0.6105572005396745,
      "learning_rate": 0.003,
      "loss": 4.222,
      "step": 3178
    },
    {
      "epoch": 0.03179,
      "grad_norm": 0.6783275770388425,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3179
    },
    {
      "epoch": 0.0318,
      "grad_norm": 0.7073200859467578,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3180
    },
    {
      "epoch": 0.03181,
      "grad_norm": 0.6783381680497546,
      "learning_rate": 0.003,
      "loss": 4.1947,
      "step": 3181
    },
    {
      "epoch": 0.03182,
      "grad_norm": 0.7230568807336094,
      "learning_rate": 0.003,
      "loss": 4.1921,
      "step": 3182
    },
    {
      "epoch": 0.03183,
      "grad_norm": 0.873578952331532,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 3183
    },
    {
      "epoch": 0.03184,
      "grad_norm": 0.9905329983068216,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 3184
    },
    {
      "epoch": 0.03185,
      "grad_norm": 1.2552538351706124,
      "learning_rate": 0.003,
      "loss": 4.2167,
      "step": 3185
    },
    {
      "epoch": 0.03186,
      "grad_norm": 0.6881365341663627,
      "learning_rate": 0.003,
      "loss": 4.1977,
      "step": 3186
    },
    {
      "epoch": 0.03187,
      "grad_norm": 0.6603777519922737,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3187
    },
    {
      "epoch": 0.03188,
      "grad_norm": 0.6114536474568962,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3188
    },
    {
      "epoch": 0.03189,
      "grad_norm": 0.7652970454563929,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3189
    },
    {
      "epoch": 0.0319,
      "grad_norm": 0.7756244045621472,
      "learning_rate": 0.003,
      "loss": 4.1785,
      "step": 3190
    },
    {
      "epoch": 0.03191,
      "grad_norm": 0.7586188953077919,
      "learning_rate": 0.003,
      "loss": 4.2132,
      "step": 3191
    },
    {
      "epoch": 0.03192,
      "grad_norm": 0.7248898112293884,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 3192
    },
    {
      "epoch": 0.03193,
      "grad_norm": 0.6819049982170161,
      "learning_rate": 0.003,
      "loss": 4.1837,
      "step": 3193
    },
    {
      "epoch": 0.03194,
      "grad_norm": 0.6384912160022395,
      "learning_rate": 0.003,
      "loss": 4.208,
      "step": 3194
    },
    {
      "epoch": 0.03195,
      "grad_norm": 0.6207563594114237,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3195
    },
    {
      "epoch": 0.03196,
      "grad_norm": 0.5716998964191464,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 3196
    },
    {
      "epoch": 0.03197,
      "grad_norm": 0.5492769888070425,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3197
    },
    {
      "epoch": 0.03198,
      "grad_norm": 0.5349886855188608,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 3198
    },
    {
      "epoch": 0.03199,
      "grad_norm": 0.4722551144791865,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3199
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.5241213036464709,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3200
    },
    {
      "epoch": 0.03201,
      "grad_norm": 0.5556232177618735,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 3201
    },
    {
      "epoch": 0.03202,
      "grad_norm": 0.5951717317344422,
      "learning_rate": 0.003,
      "loss": 4.1729,
      "step": 3202
    },
    {
      "epoch": 0.03203,
      "grad_norm": 0.5874723863194432,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 3203
    },
    {
      "epoch": 0.03204,
      "grad_norm": 0.5662458323617429,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3204
    },
    {
      "epoch": 0.03205,
      "grad_norm": 0.55236769595308,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3205
    },
    {
      "epoch": 0.03206,
      "grad_norm": 0.5082397875263351,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3206
    },
    {
      "epoch": 0.03207,
      "grad_norm": 0.4452387164637836,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 3207
    },
    {
      "epoch": 0.03208,
      "grad_norm": 0.48293819644672603,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3208
    },
    {
      "epoch": 0.03209,
      "grad_norm": 0.4498162894819178,
      "learning_rate": 0.003,
      "loss": 4.1943,
      "step": 3209
    },
    {
      "epoch": 0.0321,
      "grad_norm": 0.5708600401920843,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3210
    },
    {
      "epoch": 0.03211,
      "grad_norm": 0.6897782500954599,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3211
    },
    {
      "epoch": 0.03212,
      "grad_norm": 0.8577311234564894,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3212
    },
    {
      "epoch": 0.03213,
      "grad_norm": 0.7479816094202391,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3213
    },
    {
      "epoch": 0.03214,
      "grad_norm": 0.5575837603075381,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 3214
    },
    {
      "epoch": 0.03215,
      "grad_norm": 0.619113370679725,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 3215
    },
    {
      "epoch": 0.03216,
      "grad_norm": 0.6857581625178918,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3216
    },
    {
      "epoch": 0.03217,
      "grad_norm": 0.6339316692312089,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 3217
    },
    {
      "epoch": 0.03218,
      "grad_norm": 0.5361569097117066,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3218
    },
    {
      "epoch": 0.03219,
      "grad_norm": 0.5146188024497156,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3219
    },
    {
      "epoch": 0.0322,
      "grad_norm": 0.5580750782179962,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 3220
    },
    {
      "epoch": 0.03221,
      "grad_norm": 0.5127806683543551,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3221
    },
    {
      "epoch": 0.03222,
      "grad_norm": 0.5372160864602477,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3222
    },
    {
      "epoch": 0.03223,
      "grad_norm": 0.5923510679532299,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3223
    },
    {
      "epoch": 0.03224,
      "grad_norm": 0.6861592854273826,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 3224
    },
    {
      "epoch": 0.03225,
      "grad_norm": 0.7582363764178276,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3225
    },
    {
      "epoch": 0.03226,
      "grad_norm": 0.8207513579274669,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 3226
    },
    {
      "epoch": 0.03227,
      "grad_norm": 0.7077066475989446,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3227
    },
    {
      "epoch": 0.03228,
      "grad_norm": 0.6232151026364198,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3228
    },
    {
      "epoch": 0.03229,
      "grad_norm": 0.6228354375566014,
      "learning_rate": 0.003,
      "loss": 4.1653,
      "step": 3229
    },
    {
      "epoch": 0.0323,
      "grad_norm": 0.6011100180442941,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3230
    },
    {
      "epoch": 0.03231,
      "grad_norm": 0.6278235480602589,
      "learning_rate": 0.003,
      "loss": 4.1733,
      "step": 3231
    },
    {
      "epoch": 0.03232,
      "grad_norm": 0.5999968897563468,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3232
    },
    {
      "epoch": 0.03233,
      "grad_norm": 0.5467043906444932,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 3233
    },
    {
      "epoch": 0.03234,
      "grad_norm": 0.616637901979197,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 3234
    },
    {
      "epoch": 0.03235,
      "grad_norm": 0.7932914435950059,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3235
    },
    {
      "epoch": 0.03236,
      "grad_norm": 0.9734805326028607,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3236
    },
    {
      "epoch": 0.03237,
      "grad_norm": 0.9035113763708699,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3237
    },
    {
      "epoch": 0.03238,
      "grad_norm": 0.8124977646963972,
      "learning_rate": 0.003,
      "loss": 4.1995,
      "step": 3238
    },
    {
      "epoch": 0.03239,
      "grad_norm": 0.7169306328435582,
      "learning_rate": 0.003,
      "loss": 4.1789,
      "step": 3239
    },
    {
      "epoch": 0.0324,
      "grad_norm": 0.7507848511489641,
      "learning_rate": 0.003,
      "loss": 4.1984,
      "step": 3240
    },
    {
      "epoch": 0.03241,
      "grad_norm": 0.871055197405415,
      "learning_rate": 0.003,
      "loss": 4.1945,
      "step": 3241
    },
    {
      "epoch": 0.03242,
      "grad_norm": 0.9726257649238742,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3242
    },
    {
      "epoch": 0.03243,
      "grad_norm": 0.8829139791894937,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 3243
    },
    {
      "epoch": 0.03244,
      "grad_norm": 0.9226257984088151,
      "learning_rate": 0.003,
      "loss": 4.1964,
      "step": 3244
    },
    {
      "epoch": 0.03245,
      "grad_norm": 1.0602033368371822,
      "learning_rate": 0.003,
      "loss": 4.2203,
      "step": 3245
    },
    {
      "epoch": 0.03246,
      "grad_norm": 0.8832790843056474,
      "learning_rate": 0.003,
      "loss": 4.2041,
      "step": 3246
    },
    {
      "epoch": 0.03247,
      "grad_norm": 0.8993256456137283,
      "learning_rate": 0.003,
      "loss": 4.2129,
      "step": 3247
    },
    {
      "epoch": 0.03248,
      "grad_norm": 0.9121935412385299,
      "learning_rate": 0.003,
      "loss": 4.2361,
      "step": 3248
    },
    {
      "epoch": 0.03249,
      "grad_norm": 0.8613725910612215,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3249
    },
    {
      "epoch": 0.0325,
      "grad_norm": 0.8167292628339079,
      "learning_rate": 0.003,
      "loss": 4.2315,
      "step": 3250
    },
    {
      "epoch": 0.03251,
      "grad_norm": 0.7863361659710474,
      "learning_rate": 0.003,
      "loss": 4.2274,
      "step": 3251
    },
    {
      "epoch": 0.03252,
      "grad_norm": 0.7493875681769112,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3252
    },
    {
      "epoch": 0.03253,
      "grad_norm": 0.7679048561396902,
      "learning_rate": 0.003,
      "loss": 4.1955,
      "step": 3253
    },
    {
      "epoch": 0.03254,
      "grad_norm": 0.7654672625645944,
      "learning_rate": 0.003,
      "loss": 4.1868,
      "step": 3254
    },
    {
      "epoch": 0.03255,
      "grad_norm": 0.7464808737181244,
      "learning_rate": 0.003,
      "loss": 4.1989,
      "step": 3255
    },
    {
      "epoch": 0.03256,
      "grad_norm": 0.6526585542637006,
      "learning_rate": 0.003,
      "loss": 4.1913,
      "step": 3256
    },
    {
      "epoch": 0.03257,
      "grad_norm": 0.6001937061564249,
      "learning_rate": 0.003,
      "loss": 4.1765,
      "step": 3257
    },
    {
      "epoch": 0.03258,
      "grad_norm": 0.5472158831270236,
      "learning_rate": 0.003,
      "loss": 4.1851,
      "step": 3258
    },
    {
      "epoch": 0.03259,
      "grad_norm": 0.4785164815267161,
      "learning_rate": 0.003,
      "loss": 4.1806,
      "step": 3259
    },
    {
      "epoch": 0.0326,
      "grad_norm": 0.38861603840814574,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3260
    },
    {
      "epoch": 0.03261,
      "grad_norm": 0.41654558334327835,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3261
    },
    {
      "epoch": 0.03262,
      "grad_norm": 0.41147900352780037,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3262
    },
    {
      "epoch": 0.03263,
      "grad_norm": 0.4340386920329771,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 3263
    },
    {
      "epoch": 0.03264,
      "grad_norm": 0.47093951437519954,
      "learning_rate": 0.003,
      "loss": 4.1909,
      "step": 3264
    },
    {
      "epoch": 0.03265,
      "grad_norm": 0.5195090423674352,
      "learning_rate": 0.003,
      "loss": 4.184,
      "step": 3265
    },
    {
      "epoch": 0.03266,
      "grad_norm": 0.5846656667385064,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 3266
    },
    {
      "epoch": 0.03267,
      "grad_norm": 0.6621919529113969,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 3267
    },
    {
      "epoch": 0.03268,
      "grad_norm": 0.6864096228211959,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 3268
    },
    {
      "epoch": 0.03269,
      "grad_norm": 0.5308237484050747,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3269
    },
    {
      "epoch": 0.0327,
      "grad_norm": 0.44850448788276254,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3270
    },
    {
      "epoch": 0.03271,
      "grad_norm": 0.4693563167350624,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3271
    },
    {
      "epoch": 0.03272,
      "grad_norm": 0.5323713025056388,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3272
    },
    {
      "epoch": 0.03273,
      "grad_norm": 0.6692459000105278,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 3273
    },
    {
      "epoch": 0.03274,
      "grad_norm": 0.7924729751268105,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3274
    },
    {
      "epoch": 0.03275,
      "grad_norm": 0.8832852537291822,
      "learning_rate": 0.003,
      "loss": 4.1914,
      "step": 3275
    },
    {
      "epoch": 0.03276,
      "grad_norm": 0.9440382055401524,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3276
    },
    {
      "epoch": 0.03277,
      "grad_norm": 0.9157723514845009,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 3277
    },
    {
      "epoch": 0.03278,
      "grad_norm": 0.7757959986300896,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 3278
    },
    {
      "epoch": 0.03279,
      "grad_norm": 0.7052884741699046,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3279
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.6748931568117461,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3280
    },
    {
      "epoch": 0.03281,
      "grad_norm": 0.5834976642550973,
      "learning_rate": 0.003,
      "loss": 4.1984,
      "step": 3281
    },
    {
      "epoch": 0.03282,
      "grad_norm": 0.5830462450862388,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3282
    },
    {
      "epoch": 0.03283,
      "grad_norm": 0.6046371763595715,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3283
    },
    {
      "epoch": 0.03284,
      "grad_norm": 0.498692543866341,
      "learning_rate": 0.003,
      "loss": 4.188,
      "step": 3284
    },
    {
      "epoch": 0.03285,
      "grad_norm": 0.5300083365332573,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 3285
    },
    {
      "epoch": 0.03286,
      "grad_norm": 0.5302483285107112,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3286
    },
    {
      "epoch": 0.03287,
      "grad_norm": 0.5280212789702404,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3287
    },
    {
      "epoch": 0.03288,
      "grad_norm": 0.49328823308562514,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3288
    },
    {
      "epoch": 0.03289,
      "grad_norm": 0.5955758847359914,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3289
    },
    {
      "epoch": 0.0329,
      "grad_norm": 0.7318966219724384,
      "learning_rate": 0.003,
      "loss": 4.1954,
      "step": 3290
    },
    {
      "epoch": 0.03291,
      "grad_norm": 0.8969015331966617,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 3291
    },
    {
      "epoch": 0.03292,
      "grad_norm": 0.8171747463806796,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3292
    },
    {
      "epoch": 0.03293,
      "grad_norm": 0.6454312253277896,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3293
    },
    {
      "epoch": 0.03294,
      "grad_norm": 0.5476436730454269,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3294
    },
    {
      "epoch": 0.03295,
      "grad_norm": 0.6343237379540597,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3295
    },
    {
      "epoch": 0.03296,
      "grad_norm": 0.6111699411729721,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3296
    },
    {
      "epoch": 0.03297,
      "grad_norm": 0.5298030355791016,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3297
    },
    {
      "epoch": 0.03298,
      "grad_norm": 0.4836663999032712,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3298
    },
    {
      "epoch": 0.03299,
      "grad_norm": 0.4701496219943231,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3299
    },
    {
      "epoch": 0.033,
      "grad_norm": 0.4623026228792837,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 3300
    },
    {
      "epoch": 0.03301,
      "grad_norm": 0.43500908939154875,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 3301
    },
    {
      "epoch": 0.03302,
      "grad_norm": 0.4287201358461857,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3302
    },
    {
      "epoch": 0.03303,
      "grad_norm": 0.46641640405345786,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3303
    },
    {
      "epoch": 0.03304,
      "grad_norm": 0.588941013591916,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3304
    },
    {
      "epoch": 0.03305,
      "grad_norm": 0.7079568978470296,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3305
    },
    {
      "epoch": 0.03306,
      "grad_norm": 0.829985756720969,
      "learning_rate": 0.003,
      "loss": 4.1857,
      "step": 3306
    },
    {
      "epoch": 0.03307,
      "grad_norm": 0.8462386242609186,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3307
    },
    {
      "epoch": 0.03308,
      "grad_norm": 0.8293495815638529,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3308
    },
    {
      "epoch": 0.03309,
      "grad_norm": 0.8456212329927598,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3309
    },
    {
      "epoch": 0.0331,
      "grad_norm": 0.8565848835132709,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3310
    },
    {
      "epoch": 0.03311,
      "grad_norm": 0.9394898981807235,
      "learning_rate": 0.003,
      "loss": 4.1969,
      "step": 3311
    },
    {
      "epoch": 0.03312,
      "grad_norm": 1.0401594073450542,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3312
    },
    {
      "epoch": 0.03313,
      "grad_norm": 0.8840161283334239,
      "learning_rate": 0.003,
      "loss": 4.2028,
      "step": 3313
    },
    {
      "epoch": 0.03314,
      "grad_norm": 0.8560525559559692,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3314
    },
    {
      "epoch": 0.03315,
      "grad_norm": 0.7954589166824607,
      "learning_rate": 0.003,
      "loss": 4.2009,
      "step": 3315
    },
    {
      "epoch": 0.03316,
      "grad_norm": 0.7407877779230415,
      "learning_rate": 0.003,
      "loss": 4.2018,
      "step": 3316
    },
    {
      "epoch": 0.03317,
      "grad_norm": 0.7593343697168953,
      "learning_rate": 0.003,
      "loss": 4.1938,
      "step": 3317
    },
    {
      "epoch": 0.03318,
      "grad_norm": 0.7588615866933531,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3318
    },
    {
      "epoch": 0.03319,
      "grad_norm": 0.6406451388828956,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3319
    },
    {
      "epoch": 0.0332,
      "grad_norm": 0.6857052599814105,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3320
    },
    {
      "epoch": 0.03321,
      "grad_norm": 0.6857879313241626,
      "learning_rate": 0.003,
      "loss": 4.2019,
      "step": 3321
    },
    {
      "epoch": 0.03322,
      "grad_norm": 0.7686016779789973,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3322
    },
    {
      "epoch": 0.03323,
      "grad_norm": 0.8269613306774392,
      "learning_rate": 0.003,
      "loss": 4.2053,
      "step": 3323
    },
    {
      "epoch": 0.03324,
      "grad_norm": 0.9528423775782789,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3324
    },
    {
      "epoch": 0.03325,
      "grad_norm": 1.0794820717536844,
      "learning_rate": 0.003,
      "loss": 4.2202,
      "step": 3325
    },
    {
      "epoch": 0.03326,
      "grad_norm": 0.8445799040478842,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3326
    },
    {
      "epoch": 0.03327,
      "grad_norm": 0.6629463030927865,
      "learning_rate": 0.003,
      "loss": 4.1918,
      "step": 3327
    },
    {
      "epoch": 0.03328,
      "grad_norm": 0.6487519835718909,
      "learning_rate": 0.003,
      "loss": 4.211,
      "step": 3328
    },
    {
      "epoch": 0.03329,
      "grad_norm": 0.6298798087889773,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3329
    },
    {
      "epoch": 0.0333,
      "grad_norm": 0.6333191541087022,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3330
    },
    {
      "epoch": 0.03331,
      "grad_norm": 0.7837692756055505,
      "learning_rate": 0.003,
      "loss": 4.1711,
      "step": 3331
    },
    {
      "epoch": 0.03332,
      "grad_norm": 0.9187938715930933,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3332
    },
    {
      "epoch": 0.03333,
      "grad_norm": 0.7284609287843756,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3333
    },
    {
      "epoch": 0.03334,
      "grad_norm": 0.5950364415684378,
      "learning_rate": 0.003,
      "loss": 4.182,
      "step": 3334
    },
    {
      "epoch": 0.03335,
      "grad_norm": 0.5204758750774462,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3335
    },
    {
      "epoch": 0.03336,
      "grad_norm": 0.5061539602167332,
      "learning_rate": 0.003,
      "loss": 4.1839,
      "step": 3336
    },
    {
      "epoch": 0.03337,
      "grad_norm": 0.42913500260913456,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3337
    },
    {
      "epoch": 0.03338,
      "grad_norm": 0.4415664602358343,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3338
    },
    {
      "epoch": 0.03339,
      "grad_norm": 0.4739540929710416,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3339
    },
    {
      "epoch": 0.0334,
      "grad_norm": 0.4177466598922738,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3340
    },
    {
      "epoch": 0.03341,
      "grad_norm": 0.4450986126917478,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3341
    },
    {
      "epoch": 0.03342,
      "grad_norm": 0.4857573129812489,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 3342
    },
    {
      "epoch": 0.03343,
      "grad_norm": 0.6478603105495,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3343
    },
    {
      "epoch": 0.03344,
      "grad_norm": 0.7654176175893398,
      "learning_rate": 0.003,
      "loss": 4.1891,
      "step": 3344
    },
    {
      "epoch": 0.03345,
      "grad_norm": 0.708719161890371,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3345
    },
    {
      "epoch": 0.03346,
      "grad_norm": 0.5692838217453585,
      "learning_rate": 0.003,
      "loss": 4.1739,
      "step": 3346
    },
    {
      "epoch": 0.03347,
      "grad_norm": 0.5899959723847616,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3347
    },
    {
      "epoch": 0.03348,
      "grad_norm": 0.6486978345976652,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3348
    },
    {
      "epoch": 0.03349,
      "grad_norm": 0.5576273675617567,
      "learning_rate": 0.003,
      "loss": 4.1904,
      "step": 3349
    },
    {
      "epoch": 0.0335,
      "grad_norm": 0.537004248184913,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3350
    },
    {
      "epoch": 0.03351,
      "grad_norm": 0.5567929131713881,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 3351
    },
    {
      "epoch": 0.03352,
      "grad_norm": 0.6714613195190436,
      "learning_rate": 0.003,
      "loss": 4.171,
      "step": 3352
    },
    {
      "epoch": 0.03353,
      "grad_norm": 0.7428392470438228,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 3353
    },
    {
      "epoch": 0.03354,
      "grad_norm": 0.709523442230735,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 3354
    },
    {
      "epoch": 0.03355,
      "grad_norm": 0.5743763408378283,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3355
    },
    {
      "epoch": 0.03356,
      "grad_norm": 0.5372193296353462,
      "learning_rate": 0.003,
      "loss": 4.1833,
      "step": 3356
    },
    {
      "epoch": 0.03357,
      "grad_norm": 0.6152109328121239,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3357
    },
    {
      "epoch": 0.03358,
      "grad_norm": 0.693520738126006,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3358
    },
    {
      "epoch": 0.03359,
      "grad_norm": 0.6800317331746373,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3359
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.6462929043552973,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3360
    },
    {
      "epoch": 0.03361,
      "grad_norm": 0.6413289136062978,
      "learning_rate": 0.003,
      "loss": 4.1788,
      "step": 3361
    },
    {
      "epoch": 0.03362,
      "grad_norm": 0.6955328156949359,
      "learning_rate": 0.003,
      "loss": 4.1767,
      "step": 3362
    },
    {
      "epoch": 0.03363,
      "grad_norm": 0.6235123487755582,
      "learning_rate": 0.003,
      "loss": 4.1244,
      "step": 3363
    },
    {
      "epoch": 0.03364,
      "grad_norm": 0.602598837086171,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 3364
    },
    {
      "epoch": 0.03365,
      "grad_norm": 0.7356364448082628,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 3365
    },
    {
      "epoch": 0.03366,
      "grad_norm": 0.8244656072335483,
      "learning_rate": 0.003,
      "loss": 4.1255,
      "step": 3366
    },
    {
      "epoch": 0.03367,
      "grad_norm": 0.7703618126132621,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3367
    },
    {
      "epoch": 0.03368,
      "grad_norm": 0.6946923139107304,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3368
    },
    {
      "epoch": 0.03369,
      "grad_norm": 0.7107703988620854,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3369
    },
    {
      "epoch": 0.0337,
      "grad_norm": 0.6762732803354303,
      "learning_rate": 0.003,
      "loss": 4.1834,
      "step": 3370
    },
    {
      "epoch": 0.03371,
      "grad_norm": 0.6545021306919707,
      "learning_rate": 0.003,
      "loss": 4.117,
      "step": 3371
    },
    {
      "epoch": 0.03372,
      "grad_norm": 0.6538700683175451,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3372
    },
    {
      "epoch": 0.03373,
      "grad_norm": 0.601220154679101,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3373
    },
    {
      "epoch": 0.03374,
      "grad_norm": 0.5930617819105674,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3374
    },
    {
      "epoch": 0.03375,
      "grad_norm": 0.5268626720580842,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 3375
    },
    {
      "epoch": 0.03376,
      "grad_norm": 0.5347783123792315,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 3376
    },
    {
      "epoch": 0.03377,
      "grad_norm": 0.5216348207352511,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3377
    },
    {
      "epoch": 0.03378,
      "grad_norm": 0.6194694687851028,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3378
    },
    {
      "epoch": 0.03379,
      "grad_norm": 0.6549669142309517,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3379
    },
    {
      "epoch": 0.0338,
      "grad_norm": 0.6860555079944507,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3380
    },
    {
      "epoch": 0.03381,
      "grad_norm": 0.7654947303377718,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3381
    },
    {
      "epoch": 0.03382,
      "grad_norm": 0.8842593404054855,
      "learning_rate": 0.003,
      "loss": 4.177,
      "step": 3382
    },
    {
      "epoch": 0.03383,
      "grad_norm": 1.2792086110225658,
      "learning_rate": 0.003,
      "loss": 4.1617,
      "step": 3383
    },
    {
      "epoch": 0.03384,
      "grad_norm": 0.9270742666385557,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3384
    },
    {
      "epoch": 0.03385,
      "grad_norm": 0.7871637830554715,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3385
    },
    {
      "epoch": 0.03386,
      "grad_norm": 0.8239773523698274,
      "learning_rate": 0.003,
      "loss": 4.1807,
      "step": 3386
    },
    {
      "epoch": 0.03387,
      "grad_norm": 0.73835170938435,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3387
    },
    {
      "epoch": 0.03388,
      "grad_norm": 0.6929973646655294,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 3388
    },
    {
      "epoch": 0.03389,
      "grad_norm": 0.7227682380021057,
      "learning_rate": 0.003,
      "loss": 4.2038,
      "step": 3389
    },
    {
      "epoch": 0.0339,
      "grad_norm": 0.6143798247462846,
      "learning_rate": 0.003,
      "loss": 4.1894,
      "step": 3390
    },
    {
      "epoch": 0.03391,
      "grad_norm": 0.6734624195180631,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3391
    },
    {
      "epoch": 0.03392,
      "grad_norm": 0.6594794409943178,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3392
    },
    {
      "epoch": 0.03393,
      "grad_norm": 0.5959392487968519,
      "learning_rate": 0.003,
      "loss": 4.1822,
      "step": 3393
    },
    {
      "epoch": 0.03394,
      "grad_norm": 0.5142376929909485,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3394
    },
    {
      "epoch": 0.03395,
      "grad_norm": 0.48066315678667376,
      "learning_rate": 0.003,
      "loss": 4.1776,
      "step": 3395
    },
    {
      "epoch": 0.03396,
      "grad_norm": 0.4969135823877688,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 3396
    },
    {
      "epoch": 0.03397,
      "grad_norm": 0.4793333077073278,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 3397
    },
    {
      "epoch": 0.03398,
      "grad_norm": 0.5618846612168108,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 3398
    },
    {
      "epoch": 0.03399,
      "grad_norm": 0.598645896556551,
      "learning_rate": 0.003,
      "loss": 4.2048,
      "step": 3399
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.5909466070203917,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3400
    },
    {
      "epoch": 0.03401,
      "grad_norm": 0.5544973914148591,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3401
    },
    {
      "epoch": 0.03402,
      "grad_norm": 0.5799704954663493,
      "learning_rate": 0.003,
      "loss": 4.1801,
      "step": 3402
    },
    {
      "epoch": 0.03403,
      "grad_norm": 0.6679893543628809,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3403
    },
    {
      "epoch": 0.03404,
      "grad_norm": 0.7464230169510123,
      "learning_rate": 0.003,
      "loss": 4.1854,
      "step": 3404
    },
    {
      "epoch": 0.03405,
      "grad_norm": 0.7894447498122126,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 3405
    },
    {
      "epoch": 0.03406,
      "grad_norm": 0.7021432861635766,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 3406
    },
    {
      "epoch": 0.03407,
      "grad_norm": 0.5878306021412367,
      "learning_rate": 0.003,
      "loss": 4.1642,
      "step": 3407
    },
    {
      "epoch": 0.03408,
      "grad_norm": 0.5913015706762682,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 3408
    },
    {
      "epoch": 0.03409,
      "grad_norm": 0.6312028321034342,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 3409
    },
    {
      "epoch": 0.0341,
      "grad_norm": 0.7898832429117774,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3410
    },
    {
      "epoch": 0.03411,
      "grad_norm": 0.7714877743902544,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3411
    },
    {
      "epoch": 0.03412,
      "grad_norm": 0.7513990558104037,
      "learning_rate": 0.003,
      "loss": 4.19,
      "step": 3412
    },
    {
      "epoch": 0.03413,
      "grad_norm": 0.7461988498435763,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3413
    },
    {
      "epoch": 0.03414,
      "grad_norm": 0.7030092576304908,
      "learning_rate": 0.003,
      "loss": 4.1861,
      "step": 3414
    },
    {
      "epoch": 0.03415,
      "grad_norm": 0.7996163589913181,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 3415
    },
    {
      "epoch": 0.03416,
      "grad_norm": 0.9173276805644571,
      "learning_rate": 0.003,
      "loss": 4.1873,
      "step": 3416
    },
    {
      "epoch": 0.03417,
      "grad_norm": 1.1614777308174031,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 3417
    },
    {
      "epoch": 0.03418,
      "grad_norm": 1.0497658143246895,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3418
    },
    {
      "epoch": 0.03419,
      "grad_norm": 0.8865365686182854,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3419
    },
    {
      "epoch": 0.0342,
      "grad_norm": 0.9104853531188979,
      "learning_rate": 0.003,
      "loss": 4.2051,
      "step": 3420
    },
    {
      "epoch": 0.03421,
      "grad_norm": 0.8728209224738288,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3421
    },
    {
      "epoch": 0.03422,
      "grad_norm": 0.8151487908159633,
      "learning_rate": 0.003,
      "loss": 4.2121,
      "step": 3422
    },
    {
      "epoch": 0.03423,
      "grad_norm": 0.7760561564619395,
      "learning_rate": 0.003,
      "loss": 4.1877,
      "step": 3423
    },
    {
      "epoch": 0.03424,
      "grad_norm": 0.7902274791310419,
      "learning_rate": 0.003,
      "loss": 4.2138,
      "step": 3424
    },
    {
      "epoch": 0.03425,
      "grad_norm": 0.7840677839560138,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3425
    },
    {
      "epoch": 0.03426,
      "grad_norm": 0.7200842223622085,
      "learning_rate": 0.003,
      "loss": 4.1972,
      "step": 3426
    },
    {
      "epoch": 0.03427,
      "grad_norm": 0.6907951008112484,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3427
    },
    {
      "epoch": 0.03428,
      "grad_norm": 0.6706145072117141,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3428
    },
    {
      "epoch": 0.03429,
      "grad_norm": 0.5775498037003282,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3429
    },
    {
      "epoch": 0.0343,
      "grad_norm": 0.5280875253924556,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3430
    },
    {
      "epoch": 0.03431,
      "grad_norm": 0.5133378996898088,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3431
    },
    {
      "epoch": 0.03432,
      "grad_norm": 0.5402448334590585,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3432
    },
    {
      "epoch": 0.03433,
      "grad_norm": 0.5171387300004384,
      "learning_rate": 0.003,
      "loss": 4.1346,
      "step": 3433
    },
    {
      "epoch": 0.03434,
      "grad_norm": 0.5654372536507138,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 3434
    },
    {
      "epoch": 0.03435,
      "grad_norm": 0.7195487294501742,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3435
    },
    {
      "epoch": 0.03436,
      "grad_norm": 0.9836128063160017,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3436
    },
    {
      "epoch": 0.03437,
      "grad_norm": 1.000819383038808,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 3437
    },
    {
      "epoch": 0.03438,
      "grad_norm": 0.7833210304414927,
      "learning_rate": 0.003,
      "loss": 4.1883,
      "step": 3438
    },
    {
      "epoch": 0.03439,
      "grad_norm": 0.7155950508827565,
      "learning_rate": 0.003,
      "loss": 4.1864,
      "step": 3439
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.7239664621860699,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3440
    },
    {
      "epoch": 0.03441,
      "grad_norm": 0.6825035000283939,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3441
    },
    {
      "epoch": 0.03442,
      "grad_norm": 0.6099118062197426,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3442
    },
    {
      "epoch": 0.03443,
      "grad_norm": 0.6931617154820389,
      "learning_rate": 0.003,
      "loss": 4.1731,
      "step": 3443
    },
    {
      "epoch": 0.03444,
      "grad_norm": 0.6606283519642352,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3444
    },
    {
      "epoch": 0.03445,
      "grad_norm": 0.5738857319007685,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 3445
    },
    {
      "epoch": 0.03446,
      "grad_norm": 0.6247920672226246,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 3446
    },
    {
      "epoch": 0.03447,
      "grad_norm": 0.6141551380791748,
      "learning_rate": 0.003,
      "loss": 4.1852,
      "step": 3447
    },
    {
      "epoch": 0.03448,
      "grad_norm": 0.5719627595458622,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 3448
    },
    {
      "epoch": 0.03449,
      "grad_norm": 0.5136603871940236,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 3449
    },
    {
      "epoch": 0.0345,
      "grad_norm": 0.4556692204413723,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3450
    },
    {
      "epoch": 0.03451,
      "grad_norm": 0.43785806673883965,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 3451
    },
    {
      "epoch": 0.03452,
      "grad_norm": 0.4135943023058251,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3452
    },
    {
      "epoch": 0.03453,
      "grad_norm": 0.41685265470403127,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 3453
    },
    {
      "epoch": 0.03454,
      "grad_norm": 0.4541366407006024,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 3454
    },
    {
      "epoch": 0.03455,
      "grad_norm": 0.4820485430446041,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 3455
    },
    {
      "epoch": 0.03456,
      "grad_norm": 0.6070175235924494,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 3456
    },
    {
      "epoch": 0.03457,
      "grad_norm": 0.9155671860985577,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 3457
    },
    {
      "epoch": 0.03458,
      "grad_norm": 1.0743568532560335,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3458
    },
    {
      "epoch": 0.03459,
      "grad_norm": 0.6911554531998789,
      "learning_rate": 0.003,
      "loss": 4.1933,
      "step": 3459
    },
    {
      "epoch": 0.0346,
      "grad_norm": 0.7160032080825989,
      "learning_rate": 0.003,
      "loss": 4.1932,
      "step": 3460
    },
    {
      "epoch": 0.03461,
      "grad_norm": 0.7205146036072163,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3461
    },
    {
      "epoch": 0.03462,
      "grad_norm": 0.6026232163742171,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 3462
    },
    {
      "epoch": 0.03463,
      "grad_norm": 0.6794227213204947,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 3463
    },
    {
      "epoch": 0.03464,
      "grad_norm": 0.6456036736035401,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3464
    },
    {
      "epoch": 0.03465,
      "grad_norm": 0.7130971772716332,
      "learning_rate": 0.003,
      "loss": 4.1705,
      "step": 3465
    },
    {
      "epoch": 0.03466,
      "grad_norm": 0.6881411959714172,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3466
    },
    {
      "epoch": 0.03467,
      "grad_norm": 0.5261069185968785,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3467
    },
    {
      "epoch": 0.03468,
      "grad_norm": 0.6133848295254514,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3468
    },
    {
      "epoch": 0.03469,
      "grad_norm": 0.5746409041252338,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3469
    },
    {
      "epoch": 0.0347,
      "grad_norm": 0.509910411545196,
      "learning_rate": 0.003,
      "loss": 4.1691,
      "step": 3470
    },
    {
      "epoch": 0.03471,
      "grad_norm": 0.5549185257359567,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 3471
    },
    {
      "epoch": 0.03472,
      "grad_norm": 0.5919654155995967,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3472
    },
    {
      "epoch": 0.03473,
      "grad_norm": 0.6922934316382823,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3473
    },
    {
      "epoch": 0.03474,
      "grad_norm": 0.664824640285904,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 3474
    },
    {
      "epoch": 0.03475,
      "grad_norm": 0.690088727777654,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 3475
    },
    {
      "epoch": 0.03476,
      "grad_norm": 0.6776855408026875,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3476
    },
    {
      "epoch": 0.03477,
      "grad_norm": 0.7001953445833428,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3477
    },
    {
      "epoch": 0.03478,
      "grad_norm": 0.7620329811614004,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3478
    },
    {
      "epoch": 0.03479,
      "grad_norm": 0.8981370425674162,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 3479
    },
    {
      "epoch": 0.0348,
      "grad_norm": 0.8250170448327679,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3480
    },
    {
      "epoch": 0.03481,
      "grad_norm": 0.6849653124584689,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3481
    },
    {
      "epoch": 0.03482,
      "grad_norm": 0.6955143755202231,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3482
    },
    {
      "epoch": 0.03483,
      "grad_norm": 0.8314937740860378,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 3483
    },
    {
      "epoch": 0.03484,
      "grad_norm": 0.8591376385119152,
      "learning_rate": 0.003,
      "loss": 4.205,
      "step": 3484
    },
    {
      "epoch": 0.03485,
      "grad_norm": 0.908594692668108,
      "learning_rate": 0.003,
      "loss": 4.1994,
      "step": 3485
    },
    {
      "epoch": 0.03486,
      "grad_norm": 0.9172194797930587,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3486
    },
    {
      "epoch": 0.03487,
      "grad_norm": 0.8921568436215902,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 3487
    },
    {
      "epoch": 0.03488,
      "grad_norm": 0.7601873754760123,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3488
    },
    {
      "epoch": 0.03489,
      "grad_norm": 0.7787148687779002,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3489
    },
    {
      "epoch": 0.0349,
      "grad_norm": 0.7649701226102734,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3490
    },
    {
      "epoch": 0.03491,
      "grad_norm": 0.5441805689029632,
      "learning_rate": 0.003,
      "loss": 4.1827,
      "step": 3491
    },
    {
      "epoch": 0.03492,
      "grad_norm": 0.5662780979613814,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 3492
    },
    {
      "epoch": 0.03493,
      "grad_norm": 0.5236480373981434,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3493
    },
    {
      "epoch": 0.03494,
      "grad_norm": 0.5399778596820843,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3494
    },
    {
      "epoch": 0.03495,
      "grad_norm": 0.5540768413769496,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3495
    },
    {
      "epoch": 0.03496,
      "grad_norm": 0.5933746135701535,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3496
    },
    {
      "epoch": 0.03497,
      "grad_norm": 0.5852489998748386,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3497
    },
    {
      "epoch": 0.03498,
      "grad_norm": 0.6599448025327912,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3498
    },
    {
      "epoch": 0.03499,
      "grad_norm": 0.7172393903833442,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3499
    },
    {
      "epoch": 0.035,
      "grad_norm": 0.6405157363259029,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3500
    },
    {
      "epoch": 0.03501,
      "grad_norm": 0.5654687699259198,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3501
    },
    {
      "epoch": 0.03502,
      "grad_norm": 0.618591430417172,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3502
    },
    {
      "epoch": 0.03503,
      "grad_norm": 0.6943996447233877,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3503
    },
    {
      "epoch": 0.03504,
      "grad_norm": 0.6621001515018434,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3504
    },
    {
      "epoch": 0.03505,
      "grad_norm": 0.597184679883689,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3505
    },
    {
      "epoch": 0.03506,
      "grad_norm": 0.5659263992573133,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3506
    },
    {
      "epoch": 0.03507,
      "grad_norm": 0.5952738202818753,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 3507
    },
    {
      "epoch": 0.03508,
      "grad_norm": 0.6319710224532459,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 3508
    },
    {
      "epoch": 0.03509,
      "grad_norm": 0.7003635970154289,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 3509
    },
    {
      "epoch": 0.0351,
      "grad_norm": 0.7544632618493929,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 3510
    },
    {
      "epoch": 0.03511,
      "grad_norm": 0.8665700214441323,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3511
    },
    {
      "epoch": 0.03512,
      "grad_norm": 0.8564897161621264,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 3512
    },
    {
      "epoch": 0.03513,
      "grad_norm": 0.9197981119921014,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3513
    },
    {
      "epoch": 0.03514,
      "grad_norm": 0.8492203049153929,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3514
    },
    {
      "epoch": 0.03515,
      "grad_norm": 0.8901010567034684,
      "learning_rate": 0.003,
      "loss": 4.1863,
      "step": 3515
    },
    {
      "epoch": 0.03516,
      "grad_norm": 0.9283878054698729,
      "learning_rate": 0.003,
      "loss": 4.2172,
      "step": 3516
    },
    {
      "epoch": 0.03517,
      "grad_norm": 0.8804785262543259,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3517
    },
    {
      "epoch": 0.03518,
      "grad_norm": 0.8785403040867484,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3518
    },
    {
      "epoch": 0.03519,
      "grad_norm": 0.8144080502479469,
      "learning_rate": 0.003,
      "loss": 4.2005,
      "step": 3519
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.7871723040057216,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 3520
    },
    {
      "epoch": 0.03521,
      "grad_norm": 0.7368932966297566,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3521
    },
    {
      "epoch": 0.03522,
      "grad_norm": 0.7353338197865396,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3522
    },
    {
      "epoch": 0.03523,
      "grad_norm": 0.80714477911056,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 3523
    },
    {
      "epoch": 0.03524,
      "grad_norm": 0.8183318273894814,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3524
    },
    {
      "epoch": 0.03525,
      "grad_norm": 0.7083620836915047,
      "learning_rate": 0.003,
      "loss": 4.1538,
      "step": 3525
    },
    {
      "epoch": 0.03526,
      "grad_norm": 0.6358752706280991,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 3526
    },
    {
      "epoch": 0.03527,
      "grad_norm": 0.5882454641700064,
      "learning_rate": 0.003,
      "loss": 4.1866,
      "step": 3527
    },
    {
      "epoch": 0.03528,
      "grad_norm": 0.5418834609571274,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3528
    },
    {
      "epoch": 0.03529,
      "grad_norm": 0.4938060985233679,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3529
    },
    {
      "epoch": 0.0353,
      "grad_norm": 0.47222674272043563,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 3530
    },
    {
      "epoch": 0.03531,
      "grad_norm": 0.4494800604692719,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3531
    },
    {
      "epoch": 0.03532,
      "grad_norm": 0.46025033508749047,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3532
    },
    {
      "epoch": 0.03533,
      "grad_norm": 0.5287514582627705,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3533
    },
    {
      "epoch": 0.03534,
      "grad_norm": 0.6330388929499089,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 3534
    },
    {
      "epoch": 0.03535,
      "grad_norm": 0.6032059148264879,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 3535
    },
    {
      "epoch": 0.03536,
      "grad_norm": 0.5769078485745863,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 3536
    },
    {
      "epoch": 0.03537,
      "grad_norm": 0.7065282154758016,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3537
    },
    {
      "epoch": 0.03538,
      "grad_norm": 0.8671025316072222,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 3538
    },
    {
      "epoch": 0.03539,
      "grad_norm": 1.0158613655750308,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3539
    },
    {
      "epoch": 0.0354,
      "grad_norm": 0.8533960614991879,
      "learning_rate": 0.003,
      "loss": 4.1884,
      "step": 3540
    },
    {
      "epoch": 0.03541,
      "grad_norm": 0.6659419541207255,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3541
    },
    {
      "epoch": 0.03542,
      "grad_norm": 0.6774420684048681,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3542
    },
    {
      "epoch": 0.03543,
      "grad_norm": 0.7402573585555693,
      "learning_rate": 0.003,
      "loss": 4.1676,
      "step": 3543
    },
    {
      "epoch": 0.03544,
      "grad_norm": 0.6226751107190542,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3544
    },
    {
      "epoch": 0.03545,
      "grad_norm": 0.5532949073684433,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 3545
    },
    {
      "epoch": 0.03546,
      "grad_norm": 0.521811868333153,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3546
    },
    {
      "epoch": 0.03547,
      "grad_norm": 0.5542126267573285,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 3547
    },
    {
      "epoch": 0.03548,
      "grad_norm": 0.5294342446885085,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 3548
    },
    {
      "epoch": 0.03549,
      "grad_norm": 0.46845119607442226,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3549
    },
    {
      "epoch": 0.0355,
      "grad_norm": 0.4173574369972876,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 3550
    },
    {
      "epoch": 0.03551,
      "grad_norm": 0.5597688720149216,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3551
    },
    {
      "epoch": 0.03552,
      "grad_norm": 0.684279124843178,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 3552
    },
    {
      "epoch": 0.03553,
      "grad_norm": 0.7906598471557775,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 3553
    },
    {
      "epoch": 0.03554,
      "grad_norm": 0.8477558409372302,
      "learning_rate": 0.003,
      "loss": 4.1631,
      "step": 3554
    },
    {
      "epoch": 0.03555,
      "grad_norm": 0.9130757811703537,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3555
    },
    {
      "epoch": 0.03556,
      "grad_norm": 1.0233508386779784,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3556
    },
    {
      "epoch": 0.03557,
      "grad_norm": 0.9577495816046034,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3557
    },
    {
      "epoch": 0.03558,
      "grad_norm": 0.6971645674236855,
      "learning_rate": 0.003,
      "loss": 4.1907,
      "step": 3558
    },
    {
      "epoch": 0.03559,
      "grad_norm": 0.615773531816,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3559
    },
    {
      "epoch": 0.0356,
      "grad_norm": 0.7565427067348032,
      "learning_rate": 0.003,
      "loss": 4.179,
      "step": 3560
    },
    {
      "epoch": 0.03561,
      "grad_norm": 0.7654646543935898,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3561
    },
    {
      "epoch": 0.03562,
      "grad_norm": 0.7171787225466916,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3562
    },
    {
      "epoch": 0.03563,
      "grad_norm": 0.7202637052044578,
      "learning_rate": 0.003,
      "loss": 4.1981,
      "step": 3563
    },
    {
      "epoch": 0.03564,
      "grad_norm": 0.7813914185176697,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3564
    },
    {
      "epoch": 0.03565,
      "grad_norm": 0.9681384729900618,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3565
    },
    {
      "epoch": 0.03566,
      "grad_norm": 0.9815276969782409,
      "learning_rate": 0.003,
      "loss": 4.1871,
      "step": 3566
    },
    {
      "epoch": 0.03567,
      "grad_norm": 0.8545575756116207,
      "learning_rate": 0.003,
      "loss": 4.1799,
      "step": 3567
    },
    {
      "epoch": 0.03568,
      "grad_norm": 0.8579697593867708,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3568
    },
    {
      "epoch": 0.03569,
      "grad_norm": 0.9295006292416876,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 3569
    },
    {
      "epoch": 0.0357,
      "grad_norm": 0.7824959494204987,
      "learning_rate": 0.003,
      "loss": 4.1757,
      "step": 3570
    },
    {
      "epoch": 0.03571,
      "grad_norm": 0.715056502621953,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3571
    },
    {
      "epoch": 0.03572,
      "grad_norm": 0.7939274654308445,
      "learning_rate": 0.003,
      "loss": 4.2031,
      "step": 3572
    },
    {
      "epoch": 0.03573,
      "grad_norm": 0.8990540619234125,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 3573
    },
    {
      "epoch": 0.03574,
      "grad_norm": 0.8924167007734459,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3574
    },
    {
      "epoch": 0.03575,
      "grad_norm": 0.9633999021454877,
      "learning_rate": 0.003,
      "loss": 4.1853,
      "step": 3575
    },
    {
      "epoch": 0.03576,
      "grad_norm": 0.8337435293260566,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3576
    },
    {
      "epoch": 0.03577,
      "grad_norm": 0.7990789046598609,
      "learning_rate": 0.003,
      "loss": 4.1996,
      "step": 3577
    },
    {
      "epoch": 0.03578,
      "grad_norm": 0.740247280194808,
      "learning_rate": 0.003,
      "loss": 4.1872,
      "step": 3578
    },
    {
      "epoch": 0.03579,
      "grad_norm": 0.736535666256918,
      "learning_rate": 0.003,
      "loss": 4.2039,
      "step": 3579
    },
    {
      "epoch": 0.0358,
      "grad_norm": 0.7123148002792392,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 3580
    },
    {
      "epoch": 0.03581,
      "grad_norm": 0.6711995328778474,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3581
    },
    {
      "epoch": 0.03582,
      "grad_norm": 0.6405574599918185,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 3582
    },
    {
      "epoch": 0.03583,
      "grad_norm": 0.6115022182938825,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3583
    },
    {
      "epoch": 0.03584,
      "grad_norm": 0.5561430180920142,
      "learning_rate": 0.003,
      "loss": 4.1518,
      "step": 3584
    },
    {
      "epoch": 0.03585,
      "grad_norm": 0.5150442741371403,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 3585
    },
    {
      "epoch": 0.03586,
      "grad_norm": 0.5324475951916865,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3586
    },
    {
      "epoch": 0.03587,
      "grad_norm": 0.5865627283177408,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3587
    },
    {
      "epoch": 0.03588,
      "grad_norm": 0.6404975493032687,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 3588
    },
    {
      "epoch": 0.03589,
      "grad_norm": 0.6198477170373378,
      "learning_rate": 0.003,
      "loss": 4.1792,
      "step": 3589
    },
    {
      "epoch": 0.0359,
      "grad_norm": 0.5436091514667869,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3590
    },
    {
      "epoch": 0.03591,
      "grad_norm": 0.5153318776417137,
      "learning_rate": 0.003,
      "loss": 4.1718,
      "step": 3591
    },
    {
      "epoch": 0.03592,
      "grad_norm": 0.4789482382988687,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 3592
    },
    {
      "epoch": 0.03593,
      "grad_norm": 0.4332591255641176,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3593
    },
    {
      "epoch": 0.03594,
      "grad_norm": 0.46006005882861767,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3594
    },
    {
      "epoch": 0.03595,
      "grad_norm": 0.39445716900409733,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 3595
    },
    {
      "epoch": 0.03596,
      "grad_norm": 0.40621407877255555,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3596
    },
    {
      "epoch": 0.03597,
      "grad_norm": 0.4360669588971265,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 3597
    },
    {
      "epoch": 0.03598,
      "grad_norm": 0.5144534194928903,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3598
    },
    {
      "epoch": 0.03599,
      "grad_norm": 0.7248031737778203,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3599
    },
    {
      "epoch": 0.036,
      "grad_norm": 1.0800621032582747,
      "learning_rate": 0.003,
      "loss": 4.1892,
      "step": 3600
    },
    {
      "epoch": 0.03601,
      "grad_norm": 0.9508478687848839,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3601
    },
    {
      "epoch": 0.03602,
      "grad_norm": 0.6514600121121226,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3602
    },
    {
      "epoch": 0.03603,
      "grad_norm": 0.6278285308162193,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 3603
    },
    {
      "epoch": 0.03604,
      "grad_norm": 0.7950592611481373,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3604
    },
    {
      "epoch": 0.03605,
      "grad_norm": 0.7151902116041485,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3605
    },
    {
      "epoch": 0.03606,
      "grad_norm": 0.7086662453671662,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3606
    },
    {
      "epoch": 0.03607,
      "grad_norm": 0.7175966521512804,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3607
    },
    {
      "epoch": 0.03608,
      "grad_norm": 0.587764679138397,
      "learning_rate": 0.003,
      "loss": 4.1457,
      "step": 3608
    },
    {
      "epoch": 0.03609,
      "grad_norm": 0.5305480603629071,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3609
    },
    {
      "epoch": 0.0361,
      "grad_norm": 0.5522309324109884,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3610
    },
    {
      "epoch": 0.03611,
      "grad_norm": 0.5828315338943078,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 3611
    },
    {
      "epoch": 0.03612,
      "grad_norm": 0.6300633453025702,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 3612
    },
    {
      "epoch": 0.03613,
      "grad_norm": 0.617268364161966,
      "learning_rate": 0.003,
      "loss": 4.1666,
      "step": 3613
    },
    {
      "epoch": 0.03614,
      "grad_norm": 0.6309109479392245,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 3614
    },
    {
      "epoch": 0.03615,
      "grad_norm": 0.570782892253401,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 3615
    },
    {
      "epoch": 0.03616,
      "grad_norm": 0.5541049972487797,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3616
    },
    {
      "epoch": 0.03617,
      "grad_norm": 0.5951507982053302,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3617
    },
    {
      "epoch": 0.03618,
      "grad_norm": 0.7159646588024825,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 3618
    },
    {
      "epoch": 0.03619,
      "grad_norm": 0.8176290428711182,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 3619
    },
    {
      "epoch": 0.0362,
      "grad_norm": 0.920159490234485,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3620
    },
    {
      "epoch": 0.03621,
      "grad_norm": 0.9049277558974237,
      "learning_rate": 0.003,
      "loss": 4.173,
      "step": 3621
    },
    {
      "epoch": 0.03622,
      "grad_norm": 0.9134672000988132,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 3622
    },
    {
      "epoch": 0.03623,
      "grad_norm": 0.8632217222934753,
      "learning_rate": 0.003,
      "loss": 4.1353,
      "step": 3623
    },
    {
      "epoch": 0.03624,
      "grad_norm": 0.7748608966158578,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3624
    },
    {
      "epoch": 0.03625,
      "grad_norm": 0.6549902845991534,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3625
    },
    {
      "epoch": 0.03626,
      "grad_norm": 0.6025866731625074,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3626
    },
    {
      "epoch": 0.03627,
      "grad_norm": 0.5718452823869131,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 3627
    },
    {
      "epoch": 0.03628,
      "grad_norm": 0.5380938313028744,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3628
    },
    {
      "epoch": 0.03629,
      "grad_norm": 0.5051723060670195,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 3629
    },
    {
      "epoch": 0.0363,
      "grad_norm": 0.5404489026019382,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 3630
    },
    {
      "epoch": 0.03631,
      "grad_norm": 0.6007935793023943,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 3631
    },
    {
      "epoch": 0.03632,
      "grad_norm": 0.620386838070519,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 3632
    },
    {
      "epoch": 0.03633,
      "grad_norm": 0.7326091251339881,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3633
    },
    {
      "epoch": 0.03634,
      "grad_norm": 0.9520374375008709,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 3634
    },
    {
      "epoch": 0.03635,
      "grad_norm": 0.9435644392052507,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 3635
    },
    {
      "epoch": 0.03636,
      "grad_norm": 0.8034652176528725,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 3636
    },
    {
      "epoch": 0.03637,
      "grad_norm": 0.8215010352273984,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 3637
    },
    {
      "epoch": 0.03638,
      "grad_norm": 0.7729628364890975,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 3638
    },
    {
      "epoch": 0.03639,
      "grad_norm": 0.7939041200619633,
      "learning_rate": 0.003,
      "loss": 4.1923,
      "step": 3639
    },
    {
      "epoch": 0.0364,
      "grad_norm": 0.7624978886682702,
      "learning_rate": 0.003,
      "loss": 4.1911,
      "step": 3640
    },
    {
      "epoch": 0.03641,
      "grad_norm": 0.6972108651120591,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 3641
    },
    {
      "epoch": 0.03642,
      "grad_norm": 0.686374864648811,
      "learning_rate": 0.003,
      "loss": 4.192,
      "step": 3642
    },
    {
      "epoch": 0.03643,
      "grad_norm": 0.6994446209399746,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 3643
    },
    {
      "epoch": 0.03644,
      "grad_norm": 0.7789189969606384,
      "learning_rate": 0.003,
      "loss": 4.1819,
      "step": 3644
    },
    {
      "epoch": 0.03645,
      "grad_norm": 0.7984886139835612,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3645
    },
    {
      "epoch": 0.03646,
      "grad_norm": 0.6181112574802159,
      "learning_rate": 0.003,
      "loss": 4.1681,
      "step": 3646
    },
    {
      "epoch": 0.03647,
      "grad_norm": 0.6061654420727669,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3647
    },
    {
      "epoch": 0.03648,
      "grad_norm": 0.6696338465160281,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 3648
    },
    {
      "epoch": 0.03649,
      "grad_norm": 0.7331012822671169,
      "learning_rate": 0.003,
      "loss": 4.1808,
      "step": 3649
    },
    {
      "epoch": 0.0365,
      "grad_norm": 0.7037525814274137,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3650
    },
    {
      "epoch": 0.03651,
      "grad_norm": 0.692119661540781,
      "learning_rate": 0.003,
      "loss": 4.1791,
      "step": 3651
    },
    {
      "epoch": 0.03652,
      "grad_norm": 0.6604395418921041,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 3652
    },
    {
      "epoch": 0.03653,
      "grad_norm": 0.6346433017565251,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3653
    },
    {
      "epoch": 0.03654,
      "grad_norm": 0.6759345560485471,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3654
    },
    {
      "epoch": 0.03655,
      "grad_norm": 0.7374777892156636,
      "learning_rate": 0.003,
      "loss": 4.1719,
      "step": 3655
    },
    {
      "epoch": 0.03656,
      "grad_norm": 0.8895381978397107,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3656
    },
    {
      "epoch": 0.03657,
      "grad_norm": 1.0700577195843572,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3657
    },
    {
      "epoch": 0.03658,
      "grad_norm": 0.8922825561167493,
      "learning_rate": 0.003,
      "loss": 4.2122,
      "step": 3658
    },
    {
      "epoch": 0.03659,
      "grad_norm": 0.754621627580339,
      "learning_rate": 0.003,
      "loss": 4.1549,
      "step": 3659
    },
    {
      "epoch": 0.0366,
      "grad_norm": 0.8002619699354713,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 3660
    },
    {
      "epoch": 0.03661,
      "grad_norm": 0.7711130288745789,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 3661
    },
    {
      "epoch": 0.03662,
      "grad_norm": 0.7607029556463581,
      "learning_rate": 0.003,
      "loss": 4.162,
      "step": 3662
    },
    {
      "epoch": 0.03663,
      "grad_norm": 0.7440287835939471,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 3663
    },
    {
      "epoch": 0.03664,
      "grad_norm": 0.563780432136041,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 3664
    },
    {
      "epoch": 0.03665,
      "grad_norm": 0.4357399894939763,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3665
    },
    {
      "epoch": 0.03666,
      "grad_norm": 0.45515608243379135,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 3666
    },
    {
      "epoch": 0.03667,
      "grad_norm": 0.4706495224710868,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 3667
    },
    {
      "epoch": 0.03668,
      "grad_norm": 0.4175208514628951,
      "learning_rate": 0.003,
      "loss": 4.1477,
      "step": 3668
    },
    {
      "epoch": 0.03669,
      "grad_norm": 0.3785566363290307,
      "learning_rate": 0.003,
      "loss": 4.1216,
      "step": 3669
    },
    {
      "epoch": 0.0367,
      "grad_norm": 0.3437728850428848,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3670
    },
    {
      "epoch": 0.03671,
      "grad_norm": 0.3820679778361713,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 3671
    },
    {
      "epoch": 0.03672,
      "grad_norm": 0.44348210068257454,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 3672
    },
    {
      "epoch": 0.03673,
      "grad_norm": 0.5490971493740642,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3673
    },
    {
      "epoch": 0.03674,
      "grad_norm": 0.6223490419700332,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 3674
    },
    {
      "epoch": 0.03675,
      "grad_norm": 0.6342035142985044,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3675
    },
    {
      "epoch": 0.03676,
      "grad_norm": 0.631553220636773,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 3676
    },
    {
      "epoch": 0.03677,
      "grad_norm": 0.8208223460663321,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 3677
    },
    {
      "epoch": 0.03678,
      "grad_norm": 1.057395404894455,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 3678
    },
    {
      "epoch": 0.03679,
      "grad_norm": 1.059489837092536,
      "learning_rate": 0.003,
      "loss": 4.1768,
      "step": 3679
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.9648807335139974,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3680
    },
    {
      "epoch": 0.03681,
      "grad_norm": 0.9082948485919852,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 3681
    },
    {
      "epoch": 0.03682,
      "grad_norm": 0.7608012026223391,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3682
    },
    {
      "epoch": 0.03683,
      "grad_norm": 0.7012632382999905,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 3683
    },
    {
      "epoch": 0.03684,
      "grad_norm": 0.7244234234962599,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3684
    },
    {
      "epoch": 0.03685,
      "grad_norm": 0.6796055439571254,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 3685
    },
    {
      "epoch": 0.03686,
      "grad_norm": 0.6397461073626532,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3686
    },
    {
      "epoch": 0.03687,
      "grad_norm": 0.6837549545590554,
      "learning_rate": 0.003,
      "loss": 4.1937,
      "step": 3687
    },
    {
      "epoch": 0.03688,
      "grad_norm": 0.612454393388502,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3688
    },
    {
      "epoch": 0.03689,
      "grad_norm": 0.6808423570337984,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3689
    },
    {
      "epoch": 0.0369,
      "grad_norm": 0.6866647703326533,
      "learning_rate": 0.003,
      "loss": 4.1793,
      "step": 3690
    },
    {
      "epoch": 0.03691,
      "grad_norm": 0.7296388488733391,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3691
    },
    {
      "epoch": 0.03692,
      "grad_norm": 0.8733542369985744,
      "learning_rate": 0.003,
      "loss": 4.2056,
      "step": 3692
    },
    {
      "epoch": 0.03693,
      "grad_norm": 0.9120312999635036,
      "learning_rate": 0.003,
      "loss": 4.1787,
      "step": 3693
    },
    {
      "epoch": 0.03694,
      "grad_norm": 0.7928228320045024,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3694
    },
    {
      "epoch": 0.03695,
      "grad_norm": 0.7574597489784519,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3695
    },
    {
      "epoch": 0.03696,
      "grad_norm": 0.7264595127076895,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3696
    },
    {
      "epoch": 0.03697,
      "grad_norm": 0.797502557212387,
      "learning_rate": 0.003,
      "loss": 4.1553,
      "step": 3697
    },
    {
      "epoch": 0.03698,
      "grad_norm": 0.8741564039842601,
      "learning_rate": 0.003,
      "loss": 4.1931,
      "step": 3698
    },
    {
      "epoch": 0.03699,
      "grad_norm": 0.8855946379470947,
      "learning_rate": 0.003,
      "loss": 4.1856,
      "step": 3699
    },
    {
      "epoch": 0.037,
      "grad_norm": 0.8586025114326333,
      "learning_rate": 0.003,
      "loss": 4.1795,
      "step": 3700
    },
    {
      "epoch": 0.03701,
      "grad_norm": 0.8930614814476187,
      "learning_rate": 0.003,
      "loss": 4.1832,
      "step": 3701
    },
    {
      "epoch": 0.03702,
      "grad_norm": 0.8802482281802025,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3702
    },
    {
      "epoch": 0.03703,
      "grad_norm": 0.7549215853060027,
      "learning_rate": 0.003,
      "loss": 4.1805,
      "step": 3703
    },
    {
      "epoch": 0.03704,
      "grad_norm": 0.6295591556831198,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3704
    },
    {
      "epoch": 0.03705,
      "grad_norm": 0.5580114287621604,
      "learning_rate": 0.003,
      "loss": 4.1901,
      "step": 3705
    },
    {
      "epoch": 0.03706,
      "grad_norm": 0.493748778061454,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3706
    },
    {
      "epoch": 0.03707,
      "grad_norm": 0.501411513908697,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3707
    },
    {
      "epoch": 0.03708,
      "grad_norm": 0.46976615957109447,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3708
    },
    {
      "epoch": 0.03709,
      "grad_norm": 0.4882450257172406,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 3709
    },
    {
      "epoch": 0.0371,
      "grad_norm": 0.5609237339392169,
      "learning_rate": 0.003,
      "loss": 4.1796,
      "step": 3710
    },
    {
      "epoch": 0.03711,
      "grad_norm": 0.577744009339436,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3711
    },
    {
      "epoch": 0.03712,
      "grad_norm": 0.5617641082054192,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 3712
    },
    {
      "epoch": 0.03713,
      "grad_norm": 0.6129339279774879,
      "learning_rate": 0.003,
      "loss": 4.1383,
      "step": 3713
    },
    {
      "epoch": 0.03714,
      "grad_norm": 0.761778934710984,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 3714
    },
    {
      "epoch": 0.03715,
      "grad_norm": 0.936702433405622,
      "learning_rate": 0.003,
      "loss": 4.1692,
      "step": 3715
    },
    {
      "epoch": 0.03716,
      "grad_norm": 1.0004110813808647,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 3716
    },
    {
      "epoch": 0.03717,
      "grad_norm": 0.9246847467387567,
      "learning_rate": 0.003,
      "loss": 4.1684,
      "step": 3717
    },
    {
      "epoch": 0.03718,
      "grad_norm": 0.8130004478592973,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 3718
    },
    {
      "epoch": 0.03719,
      "grad_norm": 0.8110626454843745,
      "learning_rate": 0.003,
      "loss": 4.1701,
      "step": 3719
    },
    {
      "epoch": 0.0372,
      "grad_norm": 0.6478944328491517,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 3720
    },
    {
      "epoch": 0.03721,
      "grad_norm": 0.6036054997551548,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 3721
    },
    {
      "epoch": 0.03722,
      "grad_norm": 0.47934112751684016,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 3722
    },
    {
      "epoch": 0.03723,
      "grad_norm": 0.5465870050613278,
      "learning_rate": 0.003,
      "loss": 4.1804,
      "step": 3723
    },
    {
      "epoch": 0.03724,
      "grad_norm": 0.5794138576951311,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3724
    },
    {
      "epoch": 0.03725,
      "grad_norm": 0.5728374193198114,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3725
    },
    {
      "epoch": 0.03726,
      "grad_norm": 0.5999625053130218,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3726
    },
    {
      "epoch": 0.03727,
      "grad_norm": 0.6205981266168359,
      "learning_rate": 0.003,
      "loss": 4.1749,
      "step": 3727
    },
    {
      "epoch": 0.03728,
      "grad_norm": 0.6360363945769689,
      "learning_rate": 0.003,
      "loss": 4.1215,
      "step": 3728
    },
    {
      "epoch": 0.03729,
      "grad_norm": 0.5842605492528835,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 3729
    },
    {
      "epoch": 0.0373,
      "grad_norm": 0.5328025691146133,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 3730
    },
    {
      "epoch": 0.03731,
      "grad_norm": 0.53824919717479,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 3731
    },
    {
      "epoch": 0.03732,
      "grad_norm": 0.6816149389539496,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3732
    },
    {
      "epoch": 0.03733,
      "grad_norm": 0.7649227254747324,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3733
    },
    {
      "epoch": 0.03734,
      "grad_norm": 0.723858182349595,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 3734
    },
    {
      "epoch": 0.03735,
      "grad_norm": 0.7221215059978716,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3735
    },
    {
      "epoch": 0.03736,
      "grad_norm": 0.7400835788486658,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 3736
    },
    {
      "epoch": 0.03737,
      "grad_norm": 0.6969381819145593,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 3737
    },
    {
      "epoch": 0.03738,
      "grad_norm": 0.6127048568575166,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3738
    },
    {
      "epoch": 0.03739,
      "grad_norm": 0.6440493642792959,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 3739
    },
    {
      "epoch": 0.0374,
      "grad_norm": 0.72519391026854,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3740
    },
    {
      "epoch": 0.03741,
      "grad_norm": 0.7715264840802699,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3741
    },
    {
      "epoch": 0.03742,
      "grad_norm": 0.6234960786729838,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 3742
    },
    {
      "epoch": 0.03743,
      "grad_norm": 0.529614255393143,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 3743
    },
    {
      "epoch": 0.03744,
      "grad_norm": 0.5395949119845926,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 3744
    },
    {
      "epoch": 0.03745,
      "grad_norm": 0.5717733035088027,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3745
    },
    {
      "epoch": 0.03746,
      "grad_norm": 0.7148827950766982,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3746
    },
    {
      "epoch": 0.03747,
      "grad_norm": 0.9619906254460838,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 3747
    },
    {
      "epoch": 0.03748,
      "grad_norm": 0.9454260095382823,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 3748
    },
    {
      "epoch": 0.03749,
      "grad_norm": 0.8505540045662432,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3749
    },
    {
      "epoch": 0.0375,
      "grad_norm": 0.8001368316453729,
      "learning_rate": 0.003,
      "loss": 4.1825,
      "step": 3750
    },
    {
      "epoch": 0.03751,
      "grad_norm": 0.6787039450042418,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3751
    },
    {
      "epoch": 0.03752,
      "grad_norm": 0.6675510373317245,
      "learning_rate": 0.003,
      "loss": 4.1573,
      "step": 3752
    },
    {
      "epoch": 0.03753,
      "grad_norm": 0.659564153036602,
      "learning_rate": 0.003,
      "loss": 4.1596,
      "step": 3753
    },
    {
      "epoch": 0.03754,
      "grad_norm": 0.5879492269067329,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3754
    },
    {
      "epoch": 0.03755,
      "grad_norm": 0.5251202407537541,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3755
    },
    {
      "epoch": 0.03756,
      "grad_norm": 0.47935868138717536,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 3756
    },
    {
      "epoch": 0.03757,
      "grad_norm": 0.5841333632592701,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 3757
    },
    {
      "epoch": 0.03758,
      "grad_norm": 0.7521988739330352,
      "learning_rate": 0.003,
      "loss": 4.1101,
      "step": 3758
    },
    {
      "epoch": 0.03759,
      "grad_norm": 0.8503529306111015,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 3759
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.8653069459339167,
      "learning_rate": 0.003,
      "loss": 4.1566,
      "step": 3760
    },
    {
      "epoch": 0.03761,
      "grad_norm": 0.8195705554984235,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 3761
    },
    {
      "epoch": 0.03762,
      "grad_norm": 0.7776430458932468,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 3762
    },
    {
      "epoch": 0.03763,
      "grad_norm": 0.8090187425891137,
      "learning_rate": 0.003,
      "loss": 4.1714,
      "step": 3763
    },
    {
      "epoch": 0.03764,
      "grad_norm": 0.7299629395412605,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3764
    },
    {
      "epoch": 0.03765,
      "grad_norm": 0.7220941046527883,
      "learning_rate": 0.003,
      "loss": 4.2014,
      "step": 3765
    },
    {
      "epoch": 0.03766,
      "grad_norm": 0.7938574946684543,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3766
    },
    {
      "epoch": 0.03767,
      "grad_norm": 0.848179339225032,
      "learning_rate": 0.003,
      "loss": 4.1885,
      "step": 3767
    },
    {
      "epoch": 0.03768,
      "grad_norm": 0.8273555270559375,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3768
    },
    {
      "epoch": 0.03769,
      "grad_norm": 0.7189795085544652,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 3769
    },
    {
      "epoch": 0.0377,
      "grad_norm": 0.6877419998361242,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 3770
    },
    {
      "epoch": 0.03771,
      "grad_norm": 0.7347886642206836,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 3771
    },
    {
      "epoch": 0.03772,
      "grad_norm": 0.7369268379376811,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3772
    },
    {
      "epoch": 0.03773,
      "grad_norm": 0.7648153892095982,
      "learning_rate": 0.003,
      "loss": 4.1771,
      "step": 3773
    },
    {
      "epoch": 0.03774,
      "grad_norm": 0.689330468097358,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 3774
    },
    {
      "epoch": 0.03775,
      "grad_norm": 0.7165358486039451,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 3775
    },
    {
      "epoch": 0.03776,
      "grad_norm": 0.7489974209205834,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 3776
    },
    {
      "epoch": 0.03777,
      "grad_norm": 0.8249125931936281,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 3777
    },
    {
      "epoch": 0.03778,
      "grad_norm": 0.7978997683857884,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 3778
    },
    {
      "epoch": 0.03779,
      "grad_norm": 0.7763812864712815,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3779
    },
    {
      "epoch": 0.0378,
      "grad_norm": 0.6854145264424516,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 3780
    },
    {
      "epoch": 0.03781,
      "grad_norm": 0.7017290642180594,
      "learning_rate": 0.003,
      "loss": 4.1924,
      "step": 3781
    },
    {
      "epoch": 0.03782,
      "grad_norm": 0.7499872472851988,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3782
    },
    {
      "epoch": 0.03783,
      "grad_norm": 0.7066696741848768,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 3783
    },
    {
      "epoch": 0.03784,
      "grad_norm": 0.6293823118533233,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 3784
    },
    {
      "epoch": 0.03785,
      "grad_norm": 0.5600002308125822,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3785
    },
    {
      "epoch": 0.03786,
      "grad_norm": 0.4985251419936004,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3786
    },
    {
      "epoch": 0.03787,
      "grad_norm": 0.4682470683406906,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3787
    },
    {
      "epoch": 0.03788,
      "grad_norm": 0.507866310001821,
      "learning_rate": 0.003,
      "loss": 4.1774,
      "step": 3788
    },
    {
      "epoch": 0.03789,
      "grad_norm": 0.5463073884456529,
      "learning_rate": 0.003,
      "loss": 4.1066,
      "step": 3789
    },
    {
      "epoch": 0.0379,
      "grad_norm": 0.6172159531979761,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 3790
    },
    {
      "epoch": 0.03791,
      "grad_norm": 0.7402780853790922,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 3791
    },
    {
      "epoch": 0.03792,
      "grad_norm": 0.9096314434511164,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3792
    },
    {
      "epoch": 0.03793,
      "grad_norm": 0.8009156787875767,
      "learning_rate": 0.003,
      "loss": 4.1815,
      "step": 3793
    },
    {
      "epoch": 0.03794,
      "grad_norm": 0.6007769119418681,
      "learning_rate": 0.003,
      "loss": 4.1744,
      "step": 3794
    },
    {
      "epoch": 0.03795,
      "grad_norm": 0.6248284046940696,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3795
    },
    {
      "epoch": 0.03796,
      "grad_norm": 0.6313675517193291,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 3796
    },
    {
      "epoch": 0.03797,
      "grad_norm": 0.6260974457476847,
      "learning_rate": 0.003,
      "loss": 4.1287,
      "step": 3797
    },
    {
      "epoch": 0.03798,
      "grad_norm": 0.6329676643978727,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 3798
    },
    {
      "epoch": 0.03799,
      "grad_norm": 0.6413791763016119,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 3799
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.5746316704038654,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 3800
    },
    {
      "epoch": 0.03801,
      "grad_norm": 0.5630475643545394,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3801
    },
    {
      "epoch": 0.03802,
      "grad_norm": 0.5962762680426152,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 3802
    },
    {
      "epoch": 0.03803,
      "grad_norm": 0.7181192675393114,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 3803
    },
    {
      "epoch": 0.03804,
      "grad_norm": 0.7862836747313122,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 3804
    },
    {
      "epoch": 0.03805,
      "grad_norm": 0.8710510112491459,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3805
    },
    {
      "epoch": 0.03806,
      "grad_norm": 0.8521659658182485,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 3806
    },
    {
      "epoch": 0.03807,
      "grad_norm": 0.7446144627159563,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 3807
    },
    {
      "epoch": 0.03808,
      "grad_norm": 0.7136381828235824,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 3808
    },
    {
      "epoch": 0.03809,
      "grad_norm": 0.6617555150784667,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3809
    },
    {
      "epoch": 0.0381,
      "grad_norm": 0.5589720929569325,
      "learning_rate": 0.003,
      "loss": 4.1297,
      "step": 3810
    },
    {
      "epoch": 0.03811,
      "grad_norm": 0.513976956124665,
      "learning_rate": 0.003,
      "loss": 4.1583,
      "step": 3811
    },
    {
      "epoch": 0.03812,
      "grad_norm": 0.5341053812083173,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3812
    },
    {
      "epoch": 0.03813,
      "grad_norm": 0.5180551827098935,
      "learning_rate": 0.003,
      "loss": 4.1206,
      "step": 3813
    },
    {
      "epoch": 0.03814,
      "grad_norm": 0.5664522386981432,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 3814
    },
    {
      "epoch": 0.03815,
      "grad_norm": 0.7610361605531731,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 3815
    },
    {
      "epoch": 0.03816,
      "grad_norm": 0.9587327701936716,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 3816
    },
    {
      "epoch": 0.03817,
      "grad_norm": 0.935563437309735,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 3817
    },
    {
      "epoch": 0.03818,
      "grad_norm": 1.0060426088546508,
      "learning_rate": 0.003,
      "loss": 4.2034,
      "step": 3818
    },
    {
      "epoch": 0.03819,
      "grad_norm": 0.9721837543200701,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 3819
    },
    {
      "epoch": 0.0382,
      "grad_norm": 1.1304312225449777,
      "learning_rate": 0.003,
      "loss": 4.2256,
      "step": 3820
    },
    {
      "epoch": 0.03821,
      "grad_norm": 1.0177206332843154,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 3821
    },
    {
      "epoch": 0.03822,
      "grad_norm": 1.031220537341774,
      "learning_rate": 0.003,
      "loss": 4.1929,
      "step": 3822
    },
    {
      "epoch": 0.03823,
      "grad_norm": 1.0388061376025717,
      "learning_rate": 0.003,
      "loss": 4.1926,
      "step": 3823
    },
    {
      "epoch": 0.03824,
      "grad_norm": 0.9677858739885562,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 3824
    },
    {
      "epoch": 0.03825,
      "grad_norm": 0.937987879509307,
      "learning_rate": 0.003,
      "loss": 4.2046,
      "step": 3825
    },
    {
      "epoch": 0.03826,
      "grad_norm": 0.8230886831468718,
      "learning_rate": 0.003,
      "loss": 4.1875,
      "step": 3826
    },
    {
      "epoch": 0.03827,
      "grad_norm": 0.6935375745622523,
      "learning_rate": 0.003,
      "loss": 4.2068,
      "step": 3827
    },
    {
      "epoch": 0.03828,
      "grad_norm": 0.6856721015265079,
      "learning_rate": 0.003,
      "loss": 4.1814,
      "step": 3828
    },
    {
      "epoch": 0.03829,
      "grad_norm": 0.6394378471082416,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 3829
    },
    {
      "epoch": 0.0383,
      "grad_norm": 0.6570235063756633,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3830
    },
    {
      "epoch": 0.03831,
      "grad_norm": 0.6505218073754993,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 3831
    },
    {
      "epoch": 0.03832,
      "grad_norm": 0.5403168994861741,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 3832
    },
    {
      "epoch": 0.03833,
      "grad_norm": 0.42095483529999317,
      "learning_rate": 0.003,
      "loss": 4.1578,
      "step": 3833
    },
    {
      "epoch": 0.03834,
      "grad_norm": 0.40483366089412304,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 3834
    },
    {
      "epoch": 0.03835,
      "grad_norm": 0.4630812031805051,
      "learning_rate": 0.003,
      "loss": 4.172,
      "step": 3835
    },
    {
      "epoch": 0.03836,
      "grad_norm": 0.4841208007987568,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3836
    },
    {
      "epoch": 0.03837,
      "grad_norm": 0.536907199381203,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3837
    },
    {
      "epoch": 0.03838,
      "grad_norm": 0.6117335045256618,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3838
    },
    {
      "epoch": 0.03839,
      "grad_norm": 0.6780085079827656,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3839
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.697829286285506,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3840
    },
    {
      "epoch": 0.03841,
      "grad_norm": 0.8027464809531286,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 3841
    },
    {
      "epoch": 0.03842,
      "grad_norm": 0.8177694533436255,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 3842
    },
    {
      "epoch": 0.03843,
      "grad_norm": 0.6989170877767756,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 3843
    },
    {
      "epoch": 0.03844,
      "grad_norm": 0.6383070406944257,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3844
    },
    {
      "epoch": 0.03845,
      "grad_norm": 0.6201943361574275,
      "learning_rate": 0.003,
      "loss": 4.1557,
      "step": 3845
    },
    {
      "epoch": 0.03846,
      "grad_norm": 0.6699207633477972,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 3846
    },
    {
      "epoch": 0.03847,
      "grad_norm": 0.5790976587711703,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 3847
    },
    {
      "epoch": 0.03848,
      "grad_norm": 0.5911075701092868,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 3848
    },
    {
      "epoch": 0.03849,
      "grad_norm": 0.6969380862395563,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 3849
    },
    {
      "epoch": 0.0385,
      "grad_norm": 0.7972901632396154,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3850
    },
    {
      "epoch": 0.03851,
      "grad_norm": 0.8999501436569268,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 3851
    },
    {
      "epoch": 0.03852,
      "grad_norm": 0.9359674147442539,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 3852
    },
    {
      "epoch": 0.03853,
      "grad_norm": 0.7466141042924882,
      "learning_rate": 0.003,
      "loss": 4.1683,
      "step": 3853
    },
    {
      "epoch": 0.03854,
      "grad_norm": 0.5969895365980001,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 3854
    },
    {
      "epoch": 0.03855,
      "grad_norm": 0.5270288768727529,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 3855
    },
    {
      "epoch": 0.03856,
      "grad_norm": 0.5338061114857245,
      "learning_rate": 0.003,
      "loss": 4.1404,
      "step": 3856
    },
    {
      "epoch": 0.03857,
      "grad_norm": 0.5294295191115548,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 3857
    },
    {
      "epoch": 0.03858,
      "grad_norm": 0.5523522268399486,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 3858
    },
    {
      "epoch": 0.03859,
      "grad_norm": 0.6152187330517869,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3859
    },
    {
      "epoch": 0.0386,
      "grad_norm": 0.6117532516023865,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 3860
    },
    {
      "epoch": 0.03861,
      "grad_norm": 0.5524498434352159,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3861
    },
    {
      "epoch": 0.03862,
      "grad_norm": 0.5236517072626808,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 3862
    },
    {
      "epoch": 0.03863,
      "grad_norm": 0.6248759021675342,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 3863
    },
    {
      "epoch": 0.03864,
      "grad_norm": 0.679464652102799,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 3864
    },
    {
      "epoch": 0.03865,
      "grad_norm": 0.6453393413543939,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 3865
    },
    {
      "epoch": 0.03866,
      "grad_norm": 0.6101015046125351,
      "learning_rate": 0.003,
      "loss": 4.1319,
      "step": 3866
    },
    {
      "epoch": 0.03867,
      "grad_norm": 0.5167462630710293,
      "learning_rate": 0.003,
      "loss": 4.1116,
      "step": 3867
    },
    {
      "epoch": 0.03868,
      "grad_norm": 0.501446932546633,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 3868
    },
    {
      "epoch": 0.03869,
      "grad_norm": 0.4709811100238584,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 3869
    },
    {
      "epoch": 0.0387,
      "grad_norm": 0.4875337874718785,
      "learning_rate": 0.003,
      "loss": 4.1432,
      "step": 3870
    },
    {
      "epoch": 0.03871,
      "grad_norm": 0.5681317313929587,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 3871
    },
    {
      "epoch": 0.03872,
      "grad_norm": 0.6978195894520031,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 3872
    },
    {
      "epoch": 0.03873,
      "grad_norm": 0.9026562899987328,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 3873
    },
    {
      "epoch": 0.03874,
      "grad_norm": 1.2643110286573773,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 3874
    },
    {
      "epoch": 0.03875,
      "grad_norm": 0.8522074037692399,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 3875
    },
    {
      "epoch": 0.03876,
      "grad_norm": 0.7221351700519376,
      "learning_rate": 0.003,
      "loss": 4.1817,
      "step": 3876
    },
    {
      "epoch": 0.03877,
      "grad_norm": 0.6790217877458674,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 3877
    },
    {
      "epoch": 0.03878,
      "grad_norm": 0.6442745541470235,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3878
    },
    {
      "epoch": 0.03879,
      "grad_norm": 0.5773572680715529,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3879
    },
    {
      "epoch": 0.0388,
      "grad_norm": 0.6630378193548411,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 3880
    },
    {
      "epoch": 0.03881,
      "grad_norm": 0.6679169810032245,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 3881
    },
    {
      "epoch": 0.03882,
      "grad_norm": 0.7019796842116967,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 3882
    },
    {
      "epoch": 0.03883,
      "grad_norm": 0.6656806453760354,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 3883
    },
    {
      "epoch": 0.03884,
      "grad_norm": 0.7319329418461972,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 3884
    },
    {
      "epoch": 0.03885,
      "grad_norm": 0.9033013251391556,
      "learning_rate": 0.003,
      "loss": 4.1751,
      "step": 3885
    },
    {
      "epoch": 0.03886,
      "grad_norm": 1.0950277457821216,
      "learning_rate": 0.003,
      "loss": 4.1612,
      "step": 3886
    },
    {
      "epoch": 0.03887,
      "grad_norm": 0.7794992948946918,
      "learning_rate": 0.003,
      "loss": 4.1655,
      "step": 3887
    },
    {
      "epoch": 0.03888,
      "grad_norm": 0.6776556880519633,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 3888
    },
    {
      "epoch": 0.03889,
      "grad_norm": 0.6740634088352366,
      "learning_rate": 0.003,
      "loss": 4.1595,
      "step": 3889
    },
    {
      "epoch": 0.0389,
      "grad_norm": 0.6800323734503214,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 3890
    },
    {
      "epoch": 0.03891,
      "grad_norm": 0.6548487525930513,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 3891
    },
    {
      "epoch": 0.03892,
      "grad_norm": 0.6949410167297108,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3892
    },
    {
      "epoch": 0.03893,
      "grad_norm": 0.71600442881932,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 3893
    },
    {
      "epoch": 0.03894,
      "grad_norm": 0.6445397470488516,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3894
    },
    {
      "epoch": 0.03895,
      "grad_norm": 0.5602823203406133,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 3895
    },
    {
      "epoch": 0.03896,
      "grad_norm": 0.5429794704774388,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 3896
    },
    {
      "epoch": 0.03897,
      "grad_norm": 0.5936531994862636,
      "learning_rate": 0.003,
      "loss": 4.1226,
      "step": 3897
    },
    {
      "epoch": 0.03898,
      "grad_norm": 0.6856826306823404,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 3898
    },
    {
      "epoch": 0.03899,
      "grad_norm": 0.7274726655026192,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 3899
    },
    {
      "epoch": 0.039,
      "grad_norm": 0.8037659881555415,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 3900
    },
    {
      "epoch": 0.03901,
      "grad_norm": 0.8316671306971751,
      "learning_rate": 0.003,
      "loss": 4.1724,
      "step": 3901
    },
    {
      "epoch": 0.03902,
      "grad_norm": 0.8611987009716957,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 3902
    },
    {
      "epoch": 0.03903,
      "grad_norm": 0.8344071211818236,
      "learning_rate": 0.003,
      "loss": 4.1813,
      "step": 3903
    },
    {
      "epoch": 0.03904,
      "grad_norm": 0.7880927476353365,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 3904
    },
    {
      "epoch": 0.03905,
      "grad_norm": 0.778724013223126,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 3905
    },
    {
      "epoch": 0.03906,
      "grad_norm": 0.7715175718675301,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 3906
    },
    {
      "epoch": 0.03907,
      "grad_norm": 0.7877862833527057,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 3907
    },
    {
      "epoch": 0.03908,
      "grad_norm": 0.5993067619626092,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 3908
    },
    {
      "epoch": 0.03909,
      "grad_norm": 0.5889198426858406,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 3909
    },
    {
      "epoch": 0.0391,
      "grad_norm": 0.5523722340939393,
      "learning_rate": 0.003,
      "loss": 4.1743,
      "step": 3910
    },
    {
      "epoch": 0.03911,
      "grad_norm": 0.5127903685980498,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 3911
    },
    {
      "epoch": 0.03912,
      "grad_norm": 0.4766305366831587,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 3912
    },
    {
      "epoch": 0.03913,
      "grad_norm": 0.5086608596079191,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 3913
    },
    {
      "epoch": 0.03914,
      "grad_norm": 0.5215056781723886,
      "learning_rate": 0.003,
      "loss": 4.1622,
      "step": 3914
    },
    {
      "epoch": 0.03915,
      "grad_norm": 0.5669876961115498,
      "learning_rate": 0.003,
      "loss": 4.133,
      "step": 3915
    },
    {
      "epoch": 0.03916,
      "grad_norm": 0.5924249172540584,
      "learning_rate": 0.003,
      "loss": 4.1582,
      "step": 3916
    },
    {
      "epoch": 0.03917,
      "grad_norm": 0.6024923400756873,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3917
    },
    {
      "epoch": 0.03918,
      "grad_norm": 0.7101951940455182,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 3918
    },
    {
      "epoch": 0.03919,
      "grad_norm": 0.8265326178992785,
      "learning_rate": 0.003,
      "loss": 4.1114,
      "step": 3919
    },
    {
      "epoch": 0.0392,
      "grad_norm": 1.076222106498144,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 3920
    },
    {
      "epoch": 0.03921,
      "grad_norm": 0.9195914439109347,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 3921
    },
    {
      "epoch": 0.03922,
      "grad_norm": 0.8170632240304666,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 3922
    },
    {
      "epoch": 0.03923,
      "grad_norm": 0.8296100524137636,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 3923
    },
    {
      "epoch": 0.03924,
      "grad_norm": 0.8264396450686449,
      "learning_rate": 0.003,
      "loss": 4.1545,
      "step": 3924
    },
    {
      "epoch": 0.03925,
      "grad_norm": 0.8983164449879958,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 3925
    },
    {
      "epoch": 0.03926,
      "grad_norm": 0.8746883993886628,
      "learning_rate": 0.003,
      "loss": 4.1589,
      "step": 3926
    },
    {
      "epoch": 0.03927,
      "grad_norm": 0.8997462460768232,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3927
    },
    {
      "epoch": 0.03928,
      "grad_norm": 0.8314963835673442,
      "learning_rate": 0.003,
      "loss": 4.1865,
      "step": 3928
    },
    {
      "epoch": 0.03929,
      "grad_norm": 0.7199296155552709,
      "learning_rate": 0.003,
      "loss": 4.191,
      "step": 3929
    },
    {
      "epoch": 0.0393,
      "grad_norm": 0.6856199549997152,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 3930
    },
    {
      "epoch": 0.03931,
      "grad_norm": 0.6802907106791054,
      "learning_rate": 0.003,
      "loss": 4.1764,
      "step": 3931
    },
    {
      "epoch": 0.03932,
      "grad_norm": 0.7056782066262727,
      "learning_rate": 0.003,
      "loss": 4.1781,
      "step": 3932
    },
    {
      "epoch": 0.03933,
      "grad_norm": 0.84224228414277,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 3933
    },
    {
      "epoch": 0.03934,
      "grad_norm": 0.9417246071568263,
      "learning_rate": 0.003,
      "loss": 4.1677,
      "step": 3934
    },
    {
      "epoch": 0.03935,
      "grad_norm": 0.8605169810225617,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 3935
    },
    {
      "epoch": 0.03936,
      "grad_norm": 0.8668689820297688,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 3936
    },
    {
      "epoch": 0.03937,
      "grad_norm": 0.7302292343482224,
      "learning_rate": 0.003,
      "loss": 4.1638,
      "step": 3937
    },
    {
      "epoch": 0.03938,
      "grad_norm": 0.7852118199812202,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 3938
    },
    {
      "epoch": 0.03939,
      "grad_norm": 0.7372234256187263,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 3939
    },
    {
      "epoch": 0.0394,
      "grad_norm": 0.7135965416181232,
      "learning_rate": 0.003,
      "loss": 4.1736,
      "step": 3940
    },
    {
      "epoch": 0.03941,
      "grad_norm": 0.598564224938674,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 3941
    },
    {
      "epoch": 0.03942,
      "grad_norm": 0.5266569159659855,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 3942
    },
    {
      "epoch": 0.03943,
      "grad_norm": 0.5409156386273862,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 3943
    },
    {
      "epoch": 0.03944,
      "grad_norm": 0.604265732288973,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 3944
    },
    {
      "epoch": 0.03945,
      "grad_norm": 0.5831656289599461,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 3945
    },
    {
      "epoch": 0.03946,
      "grad_norm": 0.5502377902249579,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3946
    },
    {
      "epoch": 0.03947,
      "grad_norm": 0.48330198086471926,
      "learning_rate": 0.003,
      "loss": 4.1665,
      "step": 3947
    },
    {
      "epoch": 0.03948,
      "grad_norm": 0.45335380074512316,
      "learning_rate": 0.003,
      "loss": 4.18,
      "step": 3948
    },
    {
      "epoch": 0.03949,
      "grad_norm": 0.4365298076816215,
      "learning_rate": 0.003,
      "loss": 4.1187,
      "step": 3949
    },
    {
      "epoch": 0.0395,
      "grad_norm": 0.4330234892790934,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 3950
    },
    {
      "epoch": 0.03951,
      "grad_norm": 0.4549033232539336,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 3951
    },
    {
      "epoch": 0.03952,
      "grad_norm": 0.5088568380699875,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 3952
    },
    {
      "epoch": 0.03953,
      "grad_norm": 0.611258865255683,
      "learning_rate": 0.003,
      "loss": 4.1375,
      "step": 3953
    },
    {
      "epoch": 0.03954,
      "grad_norm": 0.6882585765157053,
      "learning_rate": 0.003,
      "loss": 4.1464,
      "step": 3954
    },
    {
      "epoch": 0.03955,
      "grad_norm": 0.7656685556557281,
      "learning_rate": 0.003,
      "loss": 4.1484,
      "step": 3955
    },
    {
      "epoch": 0.03956,
      "grad_norm": 0.8113542097479195,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 3956
    },
    {
      "epoch": 0.03957,
      "grad_norm": 0.8632342447207374,
      "learning_rate": 0.003,
      "loss": 4.1646,
      "step": 3957
    },
    {
      "epoch": 0.03958,
      "grad_norm": 1.0300805032856515,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 3958
    },
    {
      "epoch": 0.03959,
      "grad_norm": 1.4201099525549408,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 3959
    },
    {
      "epoch": 0.0396,
      "grad_norm": 0.8108000226279894,
      "learning_rate": 0.003,
      "loss": 4.1741,
      "step": 3960
    },
    {
      "epoch": 0.03961,
      "grad_norm": 0.8230830778410857,
      "learning_rate": 0.003,
      "loss": 4.1708,
      "step": 3961
    },
    {
      "epoch": 0.03962,
      "grad_norm": 0.8784449509664598,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 3962
    },
    {
      "epoch": 0.03963,
      "grad_norm": 0.8465039181944972,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 3963
    },
    {
      "epoch": 0.03964,
      "grad_norm": 0.8689576663713054,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 3964
    },
    {
      "epoch": 0.03965,
      "grad_norm": 0.8789965519139235,
      "learning_rate": 0.003,
      "loss": 4.1722,
      "step": 3965
    },
    {
      "epoch": 0.03966,
      "grad_norm": 0.7504804588349563,
      "learning_rate": 0.003,
      "loss": 4.2098,
      "step": 3966
    },
    {
      "epoch": 0.03967,
      "grad_norm": 0.7521340457571327,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 3967
    },
    {
      "epoch": 0.03968,
      "grad_norm": 0.6566639153668963,
      "learning_rate": 0.003,
      "loss": 4.1895,
      "step": 3968
    },
    {
      "epoch": 0.03969,
      "grad_norm": 0.6193517682452371,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 3969
    },
    {
      "epoch": 0.0397,
      "grad_norm": 0.618307384827639,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 3970
    },
    {
      "epoch": 0.03971,
      "grad_norm": 0.5807398202018539,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 3971
    },
    {
      "epoch": 0.03972,
      "grad_norm": 0.5003114971259446,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 3972
    },
    {
      "epoch": 0.03973,
      "grad_norm": 0.5672839085406842,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 3973
    },
    {
      "epoch": 0.03974,
      "grad_norm": 0.6148540153542106,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 3974
    },
    {
      "epoch": 0.03975,
      "grad_norm": 0.6299273778345961,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 3975
    },
    {
      "epoch": 0.03976,
      "grad_norm": 0.7124442131113389,
      "learning_rate": 0.003,
      "loss": 4.1845,
      "step": 3976
    },
    {
      "epoch": 0.03977,
      "grad_norm": 0.8025542381027738,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 3977
    },
    {
      "epoch": 0.03978,
      "grad_norm": 0.8588749028025513,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 3978
    },
    {
      "epoch": 0.03979,
      "grad_norm": 0.8573131443233668,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 3979
    },
    {
      "epoch": 0.0398,
      "grad_norm": 0.99280645435225,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 3980
    },
    {
      "epoch": 0.03981,
      "grad_norm": 1.1460073770386754,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 3981
    },
    {
      "epoch": 0.03982,
      "grad_norm": 0.9873321030261758,
      "learning_rate": 0.003,
      "loss": 4.1496,
      "step": 3982
    },
    {
      "epoch": 0.03983,
      "grad_norm": 0.8416549998631022,
      "learning_rate": 0.003,
      "loss": 4.1659,
      "step": 3983
    },
    {
      "epoch": 0.03984,
      "grad_norm": 0.7402406551683491,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 3984
    },
    {
      "epoch": 0.03985,
      "grad_norm": 0.7030279267262828,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 3985
    },
    {
      "epoch": 0.03986,
      "grad_norm": 0.6894694286447128,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 3986
    },
    {
      "epoch": 0.03987,
      "grad_norm": 0.7314973515669282,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 3987
    },
    {
      "epoch": 0.03988,
      "grad_norm": 0.8590107763209545,
      "learning_rate": 0.003,
      "loss": 4.1782,
      "step": 3988
    },
    {
      "epoch": 0.03989,
      "grad_norm": 1.074747879568787,
      "learning_rate": 0.003,
      "loss": 4.1798,
      "step": 3989
    },
    {
      "epoch": 0.0399,
      "grad_norm": 0.8434989382900636,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 3990
    },
    {
      "epoch": 0.03991,
      "grad_norm": 0.7206979857545632,
      "learning_rate": 0.003,
      "loss": 4.1859,
      "step": 3991
    },
    {
      "epoch": 0.03992,
      "grad_norm": 0.8037268570780255,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 3992
    },
    {
      "epoch": 0.03993,
      "grad_norm": 0.7620905271249203,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 3993
    },
    {
      "epoch": 0.03994,
      "grad_norm": 0.738025311420094,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 3994
    },
    {
      "epoch": 0.03995,
      "grad_norm": 0.6172466009583261,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 3995
    },
    {
      "epoch": 0.03996,
      "grad_norm": 0.452402581761887,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 3996
    },
    {
      "epoch": 0.03997,
      "grad_norm": 0.48368404250480046,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 3997
    },
    {
      "epoch": 0.03998,
      "grad_norm": 0.47207135976156783,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 3998
    },
    {
      "epoch": 0.03999,
      "grad_norm": 0.502083976437197,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 3999
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4504866573185321,
      "learning_rate": 0.003,
      "loss": 4.1697,
      "step": 4000
    },
    {
      "epoch": 0.04001,
      "grad_norm": 0.37734509945831174,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4001
    },
    {
      "epoch": 0.04002,
      "grad_norm": 0.4339763243340729,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4002
    },
    {
      "epoch": 0.04003,
      "grad_norm": 0.5726849542049914,
      "learning_rate": 0.003,
      "loss": 4.1598,
      "step": 4003
    },
    {
      "epoch": 0.04004,
      "grad_norm": 0.71841237008688,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4004
    },
    {
      "epoch": 0.04005,
      "grad_norm": 0.9930905212295545,
      "learning_rate": 0.003,
      "loss": 4.1849,
      "step": 4005
    },
    {
      "epoch": 0.04006,
      "grad_norm": 0.9991984295961828,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4006
    },
    {
      "epoch": 0.04007,
      "grad_norm": 0.7547073099000725,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4007
    },
    {
      "epoch": 0.04008,
      "grad_norm": 0.8454739420896483,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4008
    },
    {
      "epoch": 0.04009,
      "grad_norm": 0.9067851236139133,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4009
    },
    {
      "epoch": 0.0401,
      "grad_norm": 0.8062732196382164,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4010
    },
    {
      "epoch": 0.04011,
      "grad_norm": 0.6938007348622179,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4011
    },
    {
      "epoch": 0.04012,
      "grad_norm": 0.7757979776927771,
      "learning_rate": 0.003,
      "loss": 4.1668,
      "step": 4012
    },
    {
      "epoch": 0.04013,
      "grad_norm": 0.7089158793355179,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4013
    },
    {
      "epoch": 0.04014,
      "grad_norm": 0.6111952836402943,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 4014
    },
    {
      "epoch": 0.04015,
      "grad_norm": 0.5804891668774357,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4015
    },
    {
      "epoch": 0.04016,
      "grad_norm": 0.5604585117010126,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4016
    },
    {
      "epoch": 0.04017,
      "grad_norm": 0.5830753494163682,
      "learning_rate": 0.003,
      "loss": 4.1797,
      "step": 4017
    },
    {
      "epoch": 0.04018,
      "grad_norm": 0.52884054520163,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4018
    },
    {
      "epoch": 0.04019,
      "grad_norm": 0.5435284400458248,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 4019
    },
    {
      "epoch": 0.0402,
      "grad_norm": 0.46054232620433116,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4020
    },
    {
      "epoch": 0.04021,
      "grad_norm": 0.5140849815317029,
      "learning_rate": 0.003,
      "loss": 4.1291,
      "step": 4021
    },
    {
      "epoch": 0.04022,
      "grad_norm": 0.6311629918528258,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4022
    },
    {
      "epoch": 0.04023,
      "grad_norm": 0.7137869361747575,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4023
    },
    {
      "epoch": 0.04024,
      "grad_norm": 0.738420470053228,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4024
    },
    {
      "epoch": 0.04025,
      "grad_norm": 0.6521094783928104,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4025
    },
    {
      "epoch": 0.04026,
      "grad_norm": 0.5809109807556073,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4026
    },
    {
      "epoch": 0.04027,
      "grad_norm": 0.6439840472927373,
      "learning_rate": 0.003,
      "loss": 4.1593,
      "step": 4027
    },
    {
      "epoch": 0.04028,
      "grad_norm": 0.6879545278017376,
      "learning_rate": 0.003,
      "loss": 4.1395,
      "step": 4028
    },
    {
      "epoch": 0.04029,
      "grad_norm": 0.7939138849392092,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4029
    },
    {
      "epoch": 0.0403,
      "grad_norm": 0.7119398099333982,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4030
    },
    {
      "epoch": 0.04031,
      "grad_norm": 0.6966756671526879,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4031
    },
    {
      "epoch": 0.04032,
      "grad_norm": 0.6596032753240082,
      "learning_rate": 0.003,
      "loss": 4.1647,
      "step": 4032
    },
    {
      "epoch": 0.04033,
      "grad_norm": 0.6274155855378644,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4033
    },
    {
      "epoch": 0.04034,
      "grad_norm": 0.6095955743702651,
      "learning_rate": 0.003,
      "loss": 4.1253,
      "step": 4034
    },
    {
      "epoch": 0.04035,
      "grad_norm": 0.555350198090057,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4035
    },
    {
      "epoch": 0.04036,
      "grad_norm": 0.5197769667331041,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4036
    },
    {
      "epoch": 0.04037,
      "grad_norm": 0.5661546926366932,
      "learning_rate": 0.003,
      "loss": 4.1094,
      "step": 4037
    },
    {
      "epoch": 0.04038,
      "grad_norm": 0.7699720543941366,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4038
    },
    {
      "epoch": 0.04039,
      "grad_norm": 1.210094587806928,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4039
    },
    {
      "epoch": 0.0404,
      "grad_norm": 0.868075390136909,
      "learning_rate": 0.003,
      "loss": 4.1441,
      "step": 4040
    },
    {
      "epoch": 0.04041,
      "grad_norm": 0.7416181269518949,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4041
    },
    {
      "epoch": 0.04042,
      "grad_norm": 0.6648042947771692,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 4042
    },
    {
      "epoch": 0.04043,
      "grad_norm": 0.6609860437382262,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4043
    },
    {
      "epoch": 0.04044,
      "grad_norm": 0.7121985413545872,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4044
    },
    {
      "epoch": 0.04045,
      "grad_norm": 0.7259672418788464,
      "learning_rate": 0.003,
      "loss": 4.1065,
      "step": 4045
    },
    {
      "epoch": 0.04046,
      "grad_norm": 0.6676245217285697,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4046
    },
    {
      "epoch": 0.04047,
      "grad_norm": 0.6576773790552027,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4047
    },
    {
      "epoch": 0.04048,
      "grad_norm": 0.7504199241468649,
      "learning_rate": 0.003,
      "loss": 4.1392,
      "step": 4048
    },
    {
      "epoch": 0.04049,
      "grad_norm": 0.6934067855311765,
      "learning_rate": 0.003,
      "loss": 4.1769,
      "step": 4049
    },
    {
      "epoch": 0.0405,
      "grad_norm": 0.6792487502012533,
      "learning_rate": 0.003,
      "loss": 4.1721,
      "step": 4050
    },
    {
      "epoch": 0.04051,
      "grad_norm": 0.6504046520169164,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4051
    },
    {
      "epoch": 0.04052,
      "grad_norm": 0.5049122912594802,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 4052
    },
    {
      "epoch": 0.04053,
      "grad_norm": 0.5305812046390571,
      "learning_rate": 0.003,
      "loss": 4.102,
      "step": 4053
    },
    {
      "epoch": 0.04054,
      "grad_norm": 0.595657788891312,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4054
    },
    {
      "epoch": 0.04055,
      "grad_norm": 0.8174417297511951,
      "learning_rate": 0.003,
      "loss": 4.1246,
      "step": 4055
    },
    {
      "epoch": 0.04056,
      "grad_norm": 1.2164744618417813,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4056
    },
    {
      "epoch": 0.04057,
      "grad_norm": 0.7248345125453411,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4057
    },
    {
      "epoch": 0.04058,
      "grad_norm": 0.6128740383918565,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4058
    },
    {
      "epoch": 0.04059,
      "grad_norm": 0.7656551573931919,
      "learning_rate": 0.003,
      "loss": 4.1078,
      "step": 4059
    },
    {
      "epoch": 0.0406,
      "grad_norm": 0.7644654016064842,
      "learning_rate": 0.003,
      "loss": 4.1715,
      "step": 4060
    },
    {
      "epoch": 0.04061,
      "grad_norm": 0.7754515611223848,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4061
    },
    {
      "epoch": 0.04062,
      "grad_norm": 0.8608274358738601,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4062
    },
    {
      "epoch": 0.04063,
      "grad_norm": 0.9067607362608292,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4063
    },
    {
      "epoch": 0.04064,
      "grad_norm": 1.0529705240815344,
      "learning_rate": 0.003,
      "loss": 4.1734,
      "step": 4064
    },
    {
      "epoch": 0.04065,
      "grad_norm": 1.0102927624010265,
      "learning_rate": 0.003,
      "loss": 4.215,
      "step": 4065
    },
    {
      "epoch": 0.04066,
      "grad_norm": 0.9478175395069827,
      "learning_rate": 0.003,
      "loss": 4.1748,
      "step": 4066
    },
    {
      "epoch": 0.04067,
      "grad_norm": 0.8098599255455219,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4067
    },
    {
      "epoch": 0.04068,
      "grad_norm": 0.724209064256899,
      "learning_rate": 0.003,
      "loss": 4.1803,
      "step": 4068
    },
    {
      "epoch": 0.04069,
      "grad_norm": 0.6809092940169613,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4069
    },
    {
      "epoch": 0.0407,
      "grad_norm": 0.8132916778007174,
      "learning_rate": 0.003,
      "loss": 4.174,
      "step": 4070
    },
    {
      "epoch": 0.04071,
      "grad_norm": 1.041265005432495,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4071
    },
    {
      "epoch": 0.04072,
      "grad_norm": 1.0457271292201864,
      "learning_rate": 0.003,
      "loss": 4.1752,
      "step": 4072
    },
    {
      "epoch": 0.04073,
      "grad_norm": 1.0651363452233251,
      "learning_rate": 0.003,
      "loss": 4.167,
      "step": 4073
    },
    {
      "epoch": 0.04074,
      "grad_norm": 0.7177727168488233,
      "learning_rate": 0.003,
      "loss": 4.1844,
      "step": 4074
    },
    {
      "epoch": 0.04075,
      "grad_norm": 0.6005805584962535,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 4075
    },
    {
      "epoch": 0.04076,
      "grad_norm": 0.5914694996561694,
      "learning_rate": 0.003,
      "loss": 4.1674,
      "step": 4076
    },
    {
      "epoch": 0.04077,
      "grad_norm": 0.6911502018139879,
      "learning_rate": 0.003,
      "loss": 4.1809,
      "step": 4077
    },
    {
      "epoch": 0.04078,
      "grad_norm": 0.7912939945570531,
      "learning_rate": 0.003,
      "loss": 4.1605,
      "step": 4078
    },
    {
      "epoch": 0.04079,
      "grad_norm": 0.8570836696530718,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4079
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.7585471415380697,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4080
    },
    {
      "epoch": 0.04081,
      "grad_norm": 0.7520683150191194,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4081
    },
    {
      "epoch": 0.04082,
      "grad_norm": 0.6537470491867149,
      "learning_rate": 0.003,
      "loss": 4.1702,
      "step": 4082
    },
    {
      "epoch": 0.04083,
      "grad_norm": 0.6723574711008242,
      "learning_rate": 0.003,
      "loss": 4.1454,
      "step": 4083
    },
    {
      "epoch": 0.04084,
      "grad_norm": 0.6506573739916557,
      "learning_rate": 0.003,
      "loss": 4.1816,
      "step": 4084
    },
    {
      "epoch": 0.04085,
      "grad_norm": 0.7215838325103069,
      "learning_rate": 0.003,
      "loss": 4.1658,
      "step": 4085
    },
    {
      "epoch": 0.04086,
      "grad_norm": 0.7304080342645308,
      "learning_rate": 0.003,
      "loss": 4.1831,
      "step": 4086
    },
    {
      "epoch": 0.04087,
      "grad_norm": 0.736672307736761,
      "learning_rate": 0.003,
      "loss": 4.164,
      "step": 4087
    },
    {
      "epoch": 0.04088,
      "grad_norm": 0.7327696397562437,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4088
    },
    {
      "epoch": 0.04089,
      "grad_norm": 0.737838253705245,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4089
    },
    {
      "epoch": 0.0409,
      "grad_norm": 0.6618217098522069,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4090
    },
    {
      "epoch": 0.04091,
      "grad_norm": 0.6534380155154119,
      "learning_rate": 0.003,
      "loss": 4.1704,
      "step": 4091
    },
    {
      "epoch": 0.04092,
      "grad_norm": 0.6174715229139612,
      "learning_rate": 0.003,
      "loss": 4.1248,
      "step": 4092
    },
    {
      "epoch": 0.04093,
      "grad_norm": 0.6189167119848286,
      "learning_rate": 0.003,
      "loss": 4.1494,
      "step": 4093
    },
    {
      "epoch": 0.04094,
      "grad_norm": 0.6641703981374822,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4094
    },
    {
      "epoch": 0.04095,
      "grad_norm": 0.6395123879631477,
      "learning_rate": 0.003,
      "loss": 4.1173,
      "step": 4095
    },
    {
      "epoch": 0.04096,
      "grad_norm": 0.6349994704357759,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4096
    },
    {
      "epoch": 0.04097,
      "grad_norm": 0.6463455807108017,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4097
    },
    {
      "epoch": 0.04098,
      "grad_norm": 0.6932323673060099,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4098
    },
    {
      "epoch": 0.04099,
      "grad_norm": 0.8807834378700455,
      "learning_rate": 0.003,
      "loss": 4.1775,
      "step": 4099
    },
    {
      "epoch": 0.041,
      "grad_norm": 1.010325661821192,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4100
    },
    {
      "epoch": 0.04101,
      "grad_norm": 0.8888939381607318,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4101
    },
    {
      "epoch": 0.04102,
      "grad_norm": 0.7433180771066499,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4102
    },
    {
      "epoch": 0.04103,
      "grad_norm": 0.6923500471319527,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4103
    },
    {
      "epoch": 0.04104,
      "grad_norm": 0.6523846830610261,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4104
    },
    {
      "epoch": 0.04105,
      "grad_norm": 0.6124945853172781,
      "learning_rate": 0.003,
      "loss": 4.12,
      "step": 4105
    },
    {
      "epoch": 0.04106,
      "grad_norm": 0.6180742314657086,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4106
    },
    {
      "epoch": 0.04107,
      "grad_norm": 0.7675574080126237,
      "learning_rate": 0.003,
      "loss": 4.1539,
      "step": 4107
    },
    {
      "epoch": 0.04108,
      "grad_norm": 0.7972279438911031,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4108
    },
    {
      "epoch": 0.04109,
      "grad_norm": 0.7338110856400634,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4109
    },
    {
      "epoch": 0.0411,
      "grad_norm": 0.5620305965378466,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4110
    },
    {
      "epoch": 0.04111,
      "grad_norm": 0.5981690023054039,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4111
    },
    {
      "epoch": 0.04112,
      "grad_norm": 0.5672857968992558,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4112
    },
    {
      "epoch": 0.04113,
      "grad_norm": 0.6312145325378008,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 4113
    },
    {
      "epoch": 0.04114,
      "grad_norm": 0.6035265736834228,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4114
    },
    {
      "epoch": 0.04115,
      "grad_norm": 0.5962258124728763,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4115
    },
    {
      "epoch": 0.04116,
      "grad_norm": 0.5290947291389287,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4116
    },
    {
      "epoch": 0.04117,
      "grad_norm": 0.475360712634243,
      "learning_rate": 0.003,
      "loss": 4.1061,
      "step": 4117
    },
    {
      "epoch": 0.04118,
      "grad_norm": 0.48139980415634037,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4118
    },
    {
      "epoch": 0.04119,
      "grad_norm": 0.4768773023811471,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4119
    },
    {
      "epoch": 0.0412,
      "grad_norm": 0.4715918252478233,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 4120
    },
    {
      "epoch": 0.04121,
      "grad_norm": 0.4631389077981398,
      "learning_rate": 0.003,
      "loss": 4.1537,
      "step": 4121
    },
    {
      "epoch": 0.04122,
      "grad_norm": 0.5695292166099397,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4122
    },
    {
      "epoch": 0.04123,
      "grad_norm": 0.6344096447219111,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4123
    },
    {
      "epoch": 0.04124,
      "grad_norm": 0.7628425133873309,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4124
    },
    {
      "epoch": 0.04125,
      "grad_norm": 1.0834853187583475,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 4125
    },
    {
      "epoch": 0.04126,
      "grad_norm": 1.108357219795881,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4126
    },
    {
      "epoch": 0.04127,
      "grad_norm": 0.7334164794838763,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4127
    },
    {
      "epoch": 0.04128,
      "grad_norm": 0.7315856737953997,
      "learning_rate": 0.003,
      "loss": 4.1277,
      "step": 4128
    },
    {
      "epoch": 0.04129,
      "grad_norm": 0.67750855822388,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4129
    },
    {
      "epoch": 0.0413,
      "grad_norm": 0.6477790148509263,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4130
    },
    {
      "epoch": 0.04131,
      "grad_norm": 0.6976553932039135,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4131
    },
    {
      "epoch": 0.04132,
      "grad_norm": 0.7575720136801286,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4132
    },
    {
      "epoch": 0.04133,
      "grad_norm": 0.7834309464101223,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4133
    },
    {
      "epoch": 0.04134,
      "grad_norm": 0.8769385664533508,
      "learning_rate": 0.003,
      "loss": 4.1636,
      "step": 4134
    },
    {
      "epoch": 0.04135,
      "grad_norm": 0.9288719850360465,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4135
    },
    {
      "epoch": 0.04136,
      "grad_norm": 0.8613590746245083,
      "learning_rate": 0.003,
      "loss": 4.1479,
      "step": 4136
    },
    {
      "epoch": 0.04137,
      "grad_norm": 0.7914311956529176,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 4137
    },
    {
      "epoch": 0.04138,
      "grad_norm": 0.8323094727140243,
      "learning_rate": 0.003,
      "loss": 4.1727,
      "step": 4138
    },
    {
      "epoch": 0.04139,
      "grad_norm": 0.7498292112429189,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4139
    },
    {
      "epoch": 0.0414,
      "grad_norm": 0.7470547424534141,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4140
    },
    {
      "epoch": 0.04141,
      "grad_norm": 0.7086395261007737,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4141
    },
    {
      "epoch": 0.04142,
      "grad_norm": 0.7475027864042157,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4142
    },
    {
      "epoch": 0.04143,
      "grad_norm": 0.7096941472876405,
      "learning_rate": 0.003,
      "loss": 4.151,
      "step": 4143
    },
    {
      "epoch": 0.04144,
      "grad_norm": 0.6807921123663281,
      "learning_rate": 0.003,
      "loss": 4.1511,
      "step": 4144
    },
    {
      "epoch": 0.04145,
      "grad_norm": 0.6209916627342668,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4145
    },
    {
      "epoch": 0.04146,
      "grad_norm": 0.6603935131450228,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4146
    },
    {
      "epoch": 0.04147,
      "grad_norm": 0.766665555184346,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4147
    },
    {
      "epoch": 0.04148,
      "grad_norm": 0.8232977544301274,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4148
    },
    {
      "epoch": 0.04149,
      "grad_norm": 0.9384903955246353,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4149
    },
    {
      "epoch": 0.0415,
      "grad_norm": 0.8526053994314199,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4150
    },
    {
      "epoch": 0.04151,
      "grad_norm": 0.703764014227338,
      "learning_rate": 0.003,
      "loss": 4.1608,
      "step": 4151
    },
    {
      "epoch": 0.04152,
      "grad_norm": 0.5883825477373993,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4152
    },
    {
      "epoch": 0.04153,
      "grad_norm": 0.5178464710197032,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4153
    },
    {
      "epoch": 0.04154,
      "grad_norm": 0.4801954114695538,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4154
    },
    {
      "epoch": 0.04155,
      "grad_norm": 0.4883851788041572,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4155
    },
    {
      "epoch": 0.04156,
      "grad_norm": 0.5305154970213135,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4156
    },
    {
      "epoch": 0.04157,
      "grad_norm": 0.5969580348221766,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4157
    },
    {
      "epoch": 0.04158,
      "grad_norm": 0.604419085621543,
      "learning_rate": 0.003,
      "loss": 4.1654,
      "step": 4158
    },
    {
      "epoch": 0.04159,
      "grad_norm": 0.5952666616236362,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4159
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.6015748305688609,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4160
    },
    {
      "epoch": 0.04161,
      "grad_norm": 0.6509334448144796,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4161
    },
    {
      "epoch": 0.04162,
      "grad_norm": 0.8275491529133575,
      "learning_rate": 0.003,
      "loss": 4.1164,
      "step": 4162
    },
    {
      "epoch": 0.04163,
      "grad_norm": 0.9190060913193079,
      "learning_rate": 0.003,
      "loss": 4.1536,
      "step": 4163
    },
    {
      "epoch": 0.04164,
      "grad_norm": 1.0804296755917129,
      "learning_rate": 0.003,
      "loss": 4.1812,
      "step": 4164
    },
    {
      "epoch": 0.04165,
      "grad_norm": 0.8918375998135468,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4165
    },
    {
      "epoch": 0.04166,
      "grad_norm": 0.6444987754007603,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4166
    },
    {
      "epoch": 0.04167,
      "grad_norm": 0.5987188953434583,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4167
    },
    {
      "epoch": 0.04168,
      "grad_norm": 0.6931415065499815,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 4168
    },
    {
      "epoch": 0.04169,
      "grad_norm": 0.7939284298013517,
      "learning_rate": 0.003,
      "loss": 4.1706,
      "step": 4169
    },
    {
      "epoch": 0.0417,
      "grad_norm": 0.8929923791952147,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4170
    },
    {
      "epoch": 0.04171,
      "grad_norm": 0.8713733021965931,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4171
    },
    {
      "epoch": 0.04172,
      "grad_norm": 0.741654776741016,
      "learning_rate": 0.003,
      "loss": 4.1745,
      "step": 4172
    },
    {
      "epoch": 0.04173,
      "grad_norm": 0.6869389765405233,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4173
    },
    {
      "epoch": 0.04174,
      "grad_norm": 0.7104276340717113,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4174
    },
    {
      "epoch": 0.04175,
      "grad_norm": 0.7652108394533633,
      "learning_rate": 0.003,
      "loss": 4.1867,
      "step": 4175
    },
    {
      "epoch": 0.04176,
      "grad_norm": 0.6595942295349072,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4176
    },
    {
      "epoch": 0.04177,
      "grad_norm": 0.6050614217333806,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4177
    },
    {
      "epoch": 0.04178,
      "grad_norm": 0.6649317208659106,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4178
    },
    {
      "epoch": 0.04179,
      "grad_norm": 0.7497130033497713,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4179
    },
    {
      "epoch": 0.0418,
      "grad_norm": 0.8135721504617128,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4180
    },
    {
      "epoch": 0.04181,
      "grad_norm": 0.8382943348914027,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4181
    },
    {
      "epoch": 0.04182,
      "grad_norm": 0.7581608595797077,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4182
    },
    {
      "epoch": 0.04183,
      "grad_norm": 0.7367386623848073,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4183
    },
    {
      "epoch": 0.04184,
      "grad_norm": 0.8223299343686243,
      "learning_rate": 0.003,
      "loss": 4.2008,
      "step": 4184
    },
    {
      "epoch": 0.04185,
      "grad_norm": 0.8253315995103551,
      "learning_rate": 0.003,
      "loss": 4.1794,
      "step": 4185
    },
    {
      "epoch": 0.04186,
      "grad_norm": 1.021967899357659,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 4186
    },
    {
      "epoch": 0.04187,
      "grad_norm": 1.1194144390778462,
      "learning_rate": 0.003,
      "loss": 4.196,
      "step": 4187
    },
    {
      "epoch": 0.04188,
      "grad_norm": 0.8663014575987595,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4188
    },
    {
      "epoch": 0.04189,
      "grad_norm": 0.7730217042128824,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4189
    },
    {
      "epoch": 0.0419,
      "grad_norm": 0.718425867412397,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4190
    },
    {
      "epoch": 0.04191,
      "grad_norm": 0.634536414286615,
      "learning_rate": 0.003,
      "loss": 4.1592,
      "step": 4191
    },
    {
      "epoch": 0.04192,
      "grad_norm": 0.6074939161177713,
      "learning_rate": 0.003,
      "loss": 4.1386,
      "step": 4192
    },
    {
      "epoch": 0.04193,
      "grad_norm": 0.5595906619172795,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4193
    },
    {
      "epoch": 0.04194,
      "grad_norm": 0.591321468676565,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4194
    },
    {
      "epoch": 0.04195,
      "grad_norm": 0.6331925592194563,
      "learning_rate": 0.003,
      "loss": 4.1652,
      "step": 4195
    },
    {
      "epoch": 0.04196,
      "grad_norm": 0.66777685730357,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 4196
    },
    {
      "epoch": 0.04197,
      "grad_norm": 0.6581215996860487,
      "learning_rate": 0.003,
      "loss": 4.1336,
      "step": 4197
    },
    {
      "epoch": 0.04198,
      "grad_norm": 0.5775968736275588,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4198
    },
    {
      "epoch": 0.04199,
      "grad_norm": 0.5115173192319249,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4199
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.4907326781591073,
      "learning_rate": 0.003,
      "loss": 4.1414,
      "step": 4200
    },
    {
      "epoch": 0.04201,
      "grad_norm": 0.4340051119862648,
      "learning_rate": 0.003,
      "loss": 4.1156,
      "step": 4201
    },
    {
      "epoch": 0.04202,
      "grad_norm": 0.4270978633731861,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4202
    },
    {
      "epoch": 0.04203,
      "grad_norm": 0.5415535220350182,
      "learning_rate": 0.003,
      "loss": 4.1529,
      "step": 4203
    },
    {
      "epoch": 0.04204,
      "grad_norm": 0.6653136591047163,
      "learning_rate": 0.003,
      "loss": 4.1194,
      "step": 4204
    },
    {
      "epoch": 0.04205,
      "grad_norm": 0.9015195834615578,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4205
    },
    {
      "epoch": 0.04206,
      "grad_norm": 0.9511163331115945,
      "learning_rate": 0.003,
      "loss": 4.1532,
      "step": 4206
    },
    {
      "epoch": 0.04207,
      "grad_norm": 0.8183175873095769,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4207
    },
    {
      "epoch": 0.04208,
      "grad_norm": 0.7983775516444301,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4208
    },
    {
      "epoch": 0.04209,
      "grad_norm": 0.9111818683711085,
      "learning_rate": 0.003,
      "loss": 4.1635,
      "step": 4209
    },
    {
      "epoch": 0.0421,
      "grad_norm": 0.8760572861926779,
      "learning_rate": 0.003,
      "loss": 4.1198,
      "step": 4210
    },
    {
      "epoch": 0.04211,
      "grad_norm": 0.8385383847029609,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4211
    },
    {
      "epoch": 0.04212,
      "grad_norm": 0.8523572548315141,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4212
    },
    {
      "epoch": 0.04213,
      "grad_norm": 0.87521517308783,
      "learning_rate": 0.003,
      "loss": 4.1412,
      "step": 4213
    },
    {
      "epoch": 0.04214,
      "grad_norm": 0.8765783386936129,
      "learning_rate": 0.003,
      "loss": 4.1505,
      "step": 4214
    },
    {
      "epoch": 0.04215,
      "grad_norm": 0.797769330073817,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4215
    },
    {
      "epoch": 0.04216,
      "grad_norm": 0.7247822776339875,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4216
    },
    {
      "epoch": 0.04217,
      "grad_norm": 0.7261327052126046,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4217
    },
    {
      "epoch": 0.04218,
      "grad_norm": 0.7319178674334204,
      "learning_rate": 0.003,
      "loss": 4.1821,
      "step": 4218
    },
    {
      "epoch": 0.04219,
      "grad_norm": 0.7694832016892729,
      "learning_rate": 0.003,
      "loss": 4.1021,
      "step": 4219
    },
    {
      "epoch": 0.0422,
      "grad_norm": 0.757141250932244,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 4220
    },
    {
      "epoch": 0.04221,
      "grad_norm": 0.7290165220800948,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4221
    },
    {
      "epoch": 0.04222,
      "grad_norm": 0.6346702624845955,
      "learning_rate": 0.003,
      "loss": 4.1069,
      "step": 4222
    },
    {
      "epoch": 0.04223,
      "grad_norm": 0.583770301965496,
      "learning_rate": 0.003,
      "loss": 4.1682,
      "step": 4223
    },
    {
      "epoch": 0.04224,
      "grad_norm": 0.5350148270680584,
      "learning_rate": 0.003,
      "loss": 4.17,
      "step": 4224
    },
    {
      "epoch": 0.04225,
      "grad_norm": 0.5316639979135299,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 4225
    },
    {
      "epoch": 0.04226,
      "grad_norm": 0.5763902675717133,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4226
    },
    {
      "epoch": 0.04227,
      "grad_norm": 0.6468360436042053,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4227
    },
    {
      "epoch": 0.04228,
      "grad_norm": 0.7289273143868911,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4228
    },
    {
      "epoch": 0.04229,
      "grad_norm": 0.759894491275707,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4229
    },
    {
      "epoch": 0.0423,
      "grad_norm": 0.7516690067272278,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4230
    },
    {
      "epoch": 0.04231,
      "grad_norm": 0.6070127843457722,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4231
    },
    {
      "epoch": 0.04232,
      "grad_norm": 0.6987145523188838,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4232
    },
    {
      "epoch": 0.04233,
      "grad_norm": 0.9287852741458625,
      "learning_rate": 0.003,
      "loss": 4.1067,
      "step": 4233
    },
    {
      "epoch": 0.04234,
      "grad_norm": 0.9918383708905326,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4234
    },
    {
      "epoch": 0.04235,
      "grad_norm": 0.9367701682265075,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4235
    },
    {
      "epoch": 0.04236,
      "grad_norm": 0.8703490177121154,
      "learning_rate": 0.003,
      "loss": 4.1472,
      "step": 4236
    },
    {
      "epoch": 0.04237,
      "grad_norm": 0.8156234434712465,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 4237
    },
    {
      "epoch": 0.04238,
      "grad_norm": 0.6835933384976534,
      "learning_rate": 0.003,
      "loss": 4.175,
      "step": 4238
    },
    {
      "epoch": 0.04239,
      "grad_norm": 0.6391989756025438,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4239
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.6447631067586589,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4240
    },
    {
      "epoch": 0.04241,
      "grad_norm": 0.6715155999130873,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4241
    },
    {
      "epoch": 0.04242,
      "grad_norm": 0.6273748467552226,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4242
    },
    {
      "epoch": 0.04243,
      "grad_norm": 0.6467254126434867,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4243
    },
    {
      "epoch": 0.04244,
      "grad_norm": 0.6955126933633631,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4244
    },
    {
      "epoch": 0.04245,
      "grad_norm": 0.681078617510812,
      "learning_rate": 0.003,
      "loss": 4.1367,
      "step": 4245
    },
    {
      "epoch": 0.04246,
      "grad_norm": 0.69651922194907,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4246
    },
    {
      "epoch": 0.04247,
      "grad_norm": 0.7659042143848519,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4247
    },
    {
      "epoch": 0.04248,
      "grad_norm": 0.8045443211607587,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4248
    },
    {
      "epoch": 0.04249,
      "grad_norm": 0.7807725187988103,
      "learning_rate": 0.003,
      "loss": 4.1323,
      "step": 4249
    },
    {
      "epoch": 0.0425,
      "grad_norm": 0.797629821833922,
      "learning_rate": 0.003,
      "loss": 4.1528,
      "step": 4250
    },
    {
      "epoch": 0.04251,
      "grad_norm": 0.8773737541427695,
      "learning_rate": 0.003,
      "loss": 4.1179,
      "step": 4251
    },
    {
      "epoch": 0.04252,
      "grad_norm": 0.8764481339534494,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4252
    },
    {
      "epoch": 0.04253,
      "grad_norm": 0.771295115347259,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4253
    },
    {
      "epoch": 0.04254,
      "grad_norm": 0.6812529052304258,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4254
    },
    {
      "epoch": 0.04255,
      "grad_norm": 0.6528883858102312,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4255
    },
    {
      "epoch": 0.04256,
      "grad_norm": 0.6360597468005796,
      "learning_rate": 0.003,
      "loss": 4.1349,
      "step": 4256
    },
    {
      "epoch": 0.04257,
      "grad_norm": 0.6045194570542003,
      "learning_rate": 0.003,
      "loss": 4.152,
      "step": 4257
    },
    {
      "epoch": 0.04258,
      "grad_norm": 0.6237209269525407,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 4258
    },
    {
      "epoch": 0.04259,
      "grad_norm": 0.6534065774304211,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4259
    },
    {
      "epoch": 0.0426,
      "grad_norm": 0.7228160196304746,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4260
    },
    {
      "epoch": 0.04261,
      "grad_norm": 0.7225504963671482,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4261
    },
    {
      "epoch": 0.04262,
      "grad_norm": 0.6030677337778022,
      "learning_rate": 0.003,
      "loss": 4.1634,
      "step": 4262
    },
    {
      "epoch": 0.04263,
      "grad_norm": 0.5576016271704423,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4263
    },
    {
      "epoch": 0.04264,
      "grad_norm": 0.4622794122851678,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4264
    },
    {
      "epoch": 0.04265,
      "grad_norm": 0.4498062625739869,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 4265
    },
    {
      "epoch": 0.04266,
      "grad_norm": 0.4663874509947672,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4266
    },
    {
      "epoch": 0.04267,
      "grad_norm": 0.5819381640130065,
      "learning_rate": 0.003,
      "loss": 4.0999,
      "step": 4267
    },
    {
      "epoch": 0.04268,
      "grad_norm": 0.7240543464190408,
      "learning_rate": 0.003,
      "loss": 4.1466,
      "step": 4268
    },
    {
      "epoch": 0.04269,
      "grad_norm": 0.7869636361035491,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4269
    },
    {
      "epoch": 0.0427,
      "grad_norm": 0.7316564623442252,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 4270
    },
    {
      "epoch": 0.04271,
      "grad_norm": 0.6962340655904857,
      "learning_rate": 0.003,
      "loss": 4.1429,
      "step": 4271
    },
    {
      "epoch": 0.04272,
      "grad_norm": 0.7468024354946826,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4272
    },
    {
      "epoch": 0.04273,
      "grad_norm": 0.8315685932461937,
      "learning_rate": 0.003,
      "loss": 4.1228,
      "step": 4273
    },
    {
      "epoch": 0.04274,
      "grad_norm": 0.7971355124213763,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4274
    },
    {
      "epoch": 0.04275,
      "grad_norm": 0.7711074822549306,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4275
    },
    {
      "epoch": 0.04276,
      "grad_norm": 0.8849517437250696,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4276
    },
    {
      "epoch": 0.04277,
      "grad_norm": 1.058466724741216,
      "learning_rate": 0.003,
      "loss": 4.1737,
      "step": 4277
    },
    {
      "epoch": 0.04278,
      "grad_norm": 0.9806877039749324,
      "learning_rate": 0.003,
      "loss": 4.1763,
      "step": 4278
    },
    {
      "epoch": 0.04279,
      "grad_norm": 1.1506116450442327,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4279
    },
    {
      "epoch": 0.0428,
      "grad_norm": 1.0111616301127737,
      "learning_rate": 0.003,
      "loss": 4.1444,
      "step": 4280
    },
    {
      "epoch": 0.04281,
      "grad_norm": 1.1240409097888207,
      "learning_rate": 0.003,
      "loss": 4.2032,
      "step": 4281
    },
    {
      "epoch": 0.04282,
      "grad_norm": 0.9707964829866726,
      "learning_rate": 0.003,
      "loss": 4.1738,
      "step": 4282
    },
    {
      "epoch": 0.04283,
      "grad_norm": 0.8952580683408602,
      "learning_rate": 0.003,
      "loss": 4.2067,
      "step": 4283
    },
    {
      "epoch": 0.04284,
      "grad_norm": 0.8037590999213429,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4284
    },
    {
      "epoch": 0.04285,
      "grad_norm": 0.7180846206853434,
      "learning_rate": 0.003,
      "loss": 4.1689,
      "step": 4285
    },
    {
      "epoch": 0.04286,
      "grad_norm": 0.6600031955008437,
      "learning_rate": 0.003,
      "loss": 4.1846,
      "step": 4286
    },
    {
      "epoch": 0.04287,
      "grad_norm": 0.624643362166525,
      "learning_rate": 0.003,
      "loss": 4.1603,
      "step": 4287
    },
    {
      "epoch": 0.04288,
      "grad_norm": 0.7314395975343404,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4288
    },
    {
      "epoch": 0.04289,
      "grad_norm": 0.7552436651554467,
      "learning_rate": 0.003,
      "loss": 4.149,
      "step": 4289
    },
    {
      "epoch": 0.0429,
      "grad_norm": 0.7444307165904086,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4290
    },
    {
      "epoch": 0.04291,
      "grad_norm": 0.6792525045452005,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4291
    },
    {
      "epoch": 0.04292,
      "grad_norm": 0.6366634706551982,
      "learning_rate": 0.003,
      "loss": 4.1393,
      "step": 4292
    },
    {
      "epoch": 0.04293,
      "grad_norm": 0.5177321350541024,
      "learning_rate": 0.003,
      "loss": 4.1696,
      "step": 4293
    },
    {
      "epoch": 0.04294,
      "grad_norm": 0.5199647077622895,
      "learning_rate": 0.003,
      "loss": 4.1474,
      "step": 4294
    },
    {
      "epoch": 0.04295,
      "grad_norm": 0.533363434433449,
      "learning_rate": 0.003,
      "loss": 4.1196,
      "step": 4295
    },
    {
      "epoch": 0.04296,
      "grad_norm": 0.4910620913342609,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 4296
    },
    {
      "epoch": 0.04297,
      "grad_norm": 0.4528594955305011,
      "learning_rate": 0.003,
      "loss": 4.1018,
      "step": 4297
    },
    {
      "epoch": 0.04298,
      "grad_norm": 0.5508851288276945,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4298
    },
    {
      "epoch": 0.04299,
      "grad_norm": 0.6926359286804635,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4299
    },
    {
      "epoch": 0.043,
      "grad_norm": 0.776299147583224,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4300
    },
    {
      "epoch": 0.04301,
      "grad_norm": 0.7543517253530487,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4301
    },
    {
      "epoch": 0.04302,
      "grad_norm": 0.7190955364079332,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4302
    },
    {
      "epoch": 0.04303,
      "grad_norm": 0.7273212030383343,
      "learning_rate": 0.003,
      "loss": 4.1498,
      "step": 4303
    },
    {
      "epoch": 0.04304,
      "grad_norm": 0.8042177427504058,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4304
    },
    {
      "epoch": 0.04305,
      "grad_norm": 0.889845680686584,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4305
    },
    {
      "epoch": 0.04306,
      "grad_norm": 0.8019734206580067,
      "learning_rate": 0.003,
      "loss": 4.1256,
      "step": 4306
    },
    {
      "epoch": 0.04307,
      "grad_norm": 0.7599014266097025,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4307
    },
    {
      "epoch": 0.04308,
      "grad_norm": 0.6928021109826253,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4308
    },
    {
      "epoch": 0.04309,
      "grad_norm": 0.6623360644735573,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4309
    },
    {
      "epoch": 0.0431,
      "grad_norm": 0.643977742791814,
      "learning_rate": 0.003,
      "loss": 4.1542,
      "step": 4310
    },
    {
      "epoch": 0.04311,
      "grad_norm": 0.6752224622600257,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4311
    },
    {
      "epoch": 0.04312,
      "grad_norm": 0.685683319654323,
      "learning_rate": 0.003,
      "loss": 4.1286,
      "step": 4312
    },
    {
      "epoch": 0.04313,
      "grad_norm": 0.6828332575779398,
      "learning_rate": 0.003,
      "loss": 4.1407,
      "step": 4313
    },
    {
      "epoch": 0.04314,
      "grad_norm": 0.7207034799508291,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4314
    },
    {
      "epoch": 0.04315,
      "grad_norm": 0.6920628089309946,
      "learning_rate": 0.003,
      "loss": 4.1534,
      "step": 4315
    },
    {
      "epoch": 0.04316,
      "grad_norm": 0.6697894292684422,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4316
    },
    {
      "epoch": 0.04317,
      "grad_norm": 0.6529885706954907,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4317
    },
    {
      "epoch": 0.04318,
      "grad_norm": 0.6780113889358966,
      "learning_rate": 0.003,
      "loss": 4.1063,
      "step": 4318
    },
    {
      "epoch": 0.04319,
      "grad_norm": 0.6696493184651962,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 4319
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.6379331339678891,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4320
    },
    {
      "epoch": 0.04321,
      "grad_norm": 0.713977041740171,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4321
    },
    {
      "epoch": 0.04322,
      "grad_norm": 0.9061136199776986,
      "learning_rate": 0.003,
      "loss": 4.1716,
      "step": 4322
    },
    {
      "epoch": 0.04323,
      "grad_norm": 1.0087052417142888,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4323
    },
    {
      "epoch": 0.04324,
      "grad_norm": 0.9066972139100388,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4324
    },
    {
      "epoch": 0.04325,
      "grad_norm": 0.829210403933647,
      "learning_rate": 0.003,
      "loss": 4.1709,
      "step": 4325
    },
    {
      "epoch": 0.04326,
      "grad_norm": 0.766217873896832,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4326
    },
    {
      "epoch": 0.04327,
      "grad_norm": 0.66336507203185,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4327
    },
    {
      "epoch": 0.04328,
      "grad_norm": 0.6687205522581232,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4328
    },
    {
      "epoch": 0.04329,
      "grad_norm": 0.6982687394228022,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 4329
    },
    {
      "epoch": 0.0433,
      "grad_norm": 0.8226633234518121,
      "learning_rate": 0.003,
      "loss": 4.1366,
      "step": 4330
    },
    {
      "epoch": 0.04331,
      "grad_norm": 0.7328602014834829,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4331
    },
    {
      "epoch": 0.04332,
      "grad_norm": 0.7041041791657667,
      "learning_rate": 0.003,
      "loss": 4.1754,
      "step": 4332
    },
    {
      "epoch": 0.04333,
      "grad_norm": 0.7435632839058932,
      "learning_rate": 0.003,
      "loss": 4.1487,
      "step": 4333
    },
    {
      "epoch": 0.04334,
      "grad_norm": 0.8773446956801049,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4334
    },
    {
      "epoch": 0.04335,
      "grad_norm": 0.8446265063365133,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4335
    },
    {
      "epoch": 0.04336,
      "grad_norm": 0.7068555070364958,
      "learning_rate": 0.003,
      "loss": 4.1594,
      "step": 4336
    },
    {
      "epoch": 0.04337,
      "grad_norm": 0.7529497673494582,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4337
    },
    {
      "epoch": 0.04338,
      "grad_norm": 0.8822495743753999,
      "learning_rate": 0.003,
      "loss": 4.1558,
      "step": 4338
    },
    {
      "epoch": 0.04339,
      "grad_norm": 0.8362014809442191,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4339
    },
    {
      "epoch": 0.0434,
      "grad_norm": 0.8580345532471818,
      "learning_rate": 0.003,
      "loss": 4.1423,
      "step": 4340
    },
    {
      "epoch": 0.04341,
      "grad_norm": 0.8559919642238094,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4341
    },
    {
      "epoch": 0.04342,
      "grad_norm": 0.9879150305605277,
      "learning_rate": 0.003,
      "loss": 4.1501,
      "step": 4342
    },
    {
      "epoch": 0.04343,
      "grad_norm": 1.0224948991525673,
      "learning_rate": 0.003,
      "loss": 4.1953,
      "step": 4343
    },
    {
      "epoch": 0.04344,
      "grad_norm": 0.8797661268558509,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4344
    },
    {
      "epoch": 0.04345,
      "grad_norm": 0.7071974800056289,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 4345
    },
    {
      "epoch": 0.04346,
      "grad_norm": 0.7231453989893986,
      "learning_rate": 0.003,
      "loss": 4.166,
      "step": 4346
    },
    {
      "epoch": 0.04347,
      "grad_norm": 0.7413522380132354,
      "learning_rate": 0.003,
      "loss": 4.1663,
      "step": 4347
    },
    {
      "epoch": 0.04348,
      "grad_norm": 0.861089019789159,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4348
    },
    {
      "epoch": 0.04349,
      "grad_norm": 0.8929649634389639,
      "learning_rate": 0.003,
      "loss": 4.1675,
      "step": 4349
    },
    {
      "epoch": 0.0435,
      "grad_norm": 0.8422594079232042,
      "learning_rate": 0.003,
      "loss": 4.1491,
      "step": 4350
    },
    {
      "epoch": 0.04351,
      "grad_norm": 0.6766722178647709,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4351
    },
    {
      "epoch": 0.04352,
      "grad_norm": 0.6065148592107631,
      "learning_rate": 0.003,
      "loss": 4.1699,
      "step": 4352
    },
    {
      "epoch": 0.04353,
      "grad_norm": 0.650665279490602,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4353
    },
    {
      "epoch": 0.04354,
      "grad_norm": 0.6886947880787938,
      "learning_rate": 0.003,
      "loss": 4.1434,
      "step": 4354
    },
    {
      "epoch": 0.04355,
      "grad_norm": 0.671956570402557,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 4355
    },
    {
      "epoch": 0.04356,
      "grad_norm": 0.6051867008044065,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4356
    },
    {
      "epoch": 0.04357,
      "grad_norm": 0.565831993323353,
      "learning_rate": 0.003,
      "loss": 4.1356,
      "step": 4357
    },
    {
      "epoch": 0.04358,
      "grad_norm": 0.6223890122466902,
      "learning_rate": 0.003,
      "loss": 4.1266,
      "step": 4358
    },
    {
      "epoch": 0.04359,
      "grad_norm": 0.5817182346909322,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4359
    },
    {
      "epoch": 0.0436,
      "grad_norm": 0.5416327077206258,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4360
    },
    {
      "epoch": 0.04361,
      "grad_norm": 0.5828369400063332,
      "learning_rate": 0.003,
      "loss": 4.1095,
      "step": 4361
    },
    {
      "epoch": 0.04362,
      "grad_norm": 0.6659513919032584,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4362
    },
    {
      "epoch": 0.04363,
      "grad_norm": 0.6690881007982171,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4363
    },
    {
      "epoch": 0.04364,
      "grad_norm": 0.6338233307172799,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4364
    },
    {
      "epoch": 0.04365,
      "grad_norm": 0.6477441011384779,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4365
    },
    {
      "epoch": 0.04366,
      "grad_norm": 0.7001290336621013,
      "learning_rate": 0.003,
      "loss": 4.0975,
      "step": 4366
    },
    {
      "epoch": 0.04367,
      "grad_norm": 0.7750910475232337,
      "learning_rate": 0.003,
      "loss": 4.1522,
      "step": 4367
    },
    {
      "epoch": 0.04368,
      "grad_norm": 0.653538990880571,
      "learning_rate": 0.003,
      "loss": 4.135,
      "step": 4368
    },
    {
      "epoch": 0.04369,
      "grad_norm": 0.5454509649351917,
      "learning_rate": 0.003,
      "loss": 4.124,
      "step": 4369
    },
    {
      "epoch": 0.0437,
      "grad_norm": 0.587005709751278,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4370
    },
    {
      "epoch": 0.04371,
      "grad_norm": 0.6982746587986959,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4371
    },
    {
      "epoch": 0.04372,
      "grad_norm": 0.7399237590808896,
      "learning_rate": 0.003,
      "loss": 4.1211,
      "step": 4372
    },
    {
      "epoch": 0.04373,
      "grad_norm": 0.7016341160191462,
      "learning_rate": 0.003,
      "loss": 4.1446,
      "step": 4373
    },
    {
      "epoch": 0.04374,
      "grad_norm": 0.7430013647684137,
      "learning_rate": 0.003,
      "loss": 4.1899,
      "step": 4374
    },
    {
      "epoch": 0.04375,
      "grad_norm": 0.7446061707027156,
      "learning_rate": 0.003,
      "loss": 4.1043,
      "step": 4375
    },
    {
      "epoch": 0.04376,
      "grad_norm": 0.7746810507611076,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4376
    },
    {
      "epoch": 0.04377,
      "grad_norm": 0.9323940396007363,
      "learning_rate": 0.003,
      "loss": 4.1766,
      "step": 4377
    },
    {
      "epoch": 0.04378,
      "grad_norm": 1.139045470645915,
      "learning_rate": 0.003,
      "loss": 4.1725,
      "step": 4378
    },
    {
      "epoch": 0.04379,
      "grad_norm": 0.9525647154578588,
      "learning_rate": 0.003,
      "loss": 4.1848,
      "step": 4379
    },
    {
      "epoch": 0.0438,
      "grad_norm": 0.8243260327364462,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4380
    },
    {
      "epoch": 0.04381,
      "grad_norm": 0.6831625467049471,
      "learning_rate": 0.003,
      "loss": 4.1377,
      "step": 4381
    },
    {
      "epoch": 0.04382,
      "grad_norm": 0.612690019661821,
      "learning_rate": 0.003,
      "loss": 4.148,
      "step": 4382
    },
    {
      "epoch": 0.04383,
      "grad_norm": 0.540088011425151,
      "learning_rate": 0.003,
      "loss": 4.1237,
      "step": 4383
    },
    {
      "epoch": 0.04384,
      "grad_norm": 0.48067565504109466,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4384
    },
    {
      "epoch": 0.04385,
      "grad_norm": 0.4940838283566669,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4385
    },
    {
      "epoch": 0.04386,
      "grad_norm": 0.47625894049094,
      "learning_rate": 0.003,
      "loss": 4.134,
      "step": 4386
    },
    {
      "epoch": 0.04387,
      "grad_norm": 0.48055396677108986,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 4387
    },
    {
      "epoch": 0.04388,
      "grad_norm": 0.4652346965489437,
      "learning_rate": 0.003,
      "loss": 4.1166,
      "step": 4388
    },
    {
      "epoch": 0.04389,
      "grad_norm": 0.6054008250326102,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4389
    },
    {
      "epoch": 0.0439,
      "grad_norm": 0.8224358615264853,
      "learning_rate": 0.003,
      "loss": 4.1535,
      "step": 4390
    },
    {
      "epoch": 0.04391,
      "grad_norm": 0.9124868764503055,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4391
    },
    {
      "epoch": 0.04392,
      "grad_norm": 0.9806682632795687,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4392
    },
    {
      "epoch": 0.04393,
      "grad_norm": 0.9973385060246203,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4393
    },
    {
      "epoch": 0.04394,
      "grad_norm": 0.9949589217331475,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4394
    },
    {
      "epoch": 0.04395,
      "grad_norm": 0.9709607301881599,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4395
    },
    {
      "epoch": 0.04396,
      "grad_norm": 0.9645527804872946,
      "learning_rate": 0.003,
      "loss": 4.123,
      "step": 4396
    },
    {
      "epoch": 0.04397,
      "grad_norm": 0.9816350655638709,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4397
    },
    {
      "epoch": 0.04398,
      "grad_norm": 1.0455773944250841,
      "learning_rate": 0.003,
      "loss": 4.1626,
      "step": 4398
    },
    {
      "epoch": 0.04399,
      "grad_norm": 0.9119841908639416,
      "learning_rate": 0.003,
      "loss": 4.1667,
      "step": 4399
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.9349641099143394,
      "learning_rate": 0.003,
      "loss": 4.1639,
      "step": 4400
    },
    {
      "epoch": 0.04401,
      "grad_norm": 1.012943202195574,
      "learning_rate": 0.003,
      "loss": 4.1876,
      "step": 4401
    },
    {
      "epoch": 0.04402,
      "grad_norm": 0.9301815653214077,
      "learning_rate": 0.003,
      "loss": 4.1887,
      "step": 4402
    },
    {
      "epoch": 0.04403,
      "grad_norm": 0.8751459284087095,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4403
    },
    {
      "epoch": 0.04404,
      "grad_norm": 0.8722167037599222,
      "learning_rate": 0.003,
      "loss": 4.1843,
      "step": 4404
    },
    {
      "epoch": 0.04405,
      "grad_norm": 0.8424088244833199,
      "learning_rate": 0.003,
      "loss": 4.176,
      "step": 4405
    },
    {
      "epoch": 0.04406,
      "grad_norm": 0.8029456116125556,
      "learning_rate": 0.003,
      "loss": 4.1678,
      "step": 4406
    },
    {
      "epoch": 0.04407,
      "grad_norm": 0.7600665803647753,
      "learning_rate": 0.003,
      "loss": 4.1397,
      "step": 4407
    },
    {
      "epoch": 0.04408,
      "grad_norm": 0.7143442872514836,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4408
    },
    {
      "epoch": 0.04409,
      "grad_norm": 0.7965192349032346,
      "learning_rate": 0.003,
      "loss": 4.178,
      "step": 4409
    },
    {
      "epoch": 0.0441,
      "grad_norm": 0.8792702370392786,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4410
    },
    {
      "epoch": 0.04411,
      "grad_norm": 1.0243973709917655,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4411
    },
    {
      "epoch": 0.04412,
      "grad_norm": 0.8145959259792708,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4412
    },
    {
      "epoch": 0.04413,
      "grad_norm": 0.57578344484875,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4413
    },
    {
      "epoch": 0.04414,
      "grad_norm": 0.5763746372907456,
      "learning_rate": 0.003,
      "loss": 4.1858,
      "step": 4414
    },
    {
      "epoch": 0.04415,
      "grad_norm": 0.4849406384890711,
      "learning_rate": 0.003,
      "loss": 4.1431,
      "step": 4415
    },
    {
      "epoch": 0.04416,
      "grad_norm": 0.5645793903941705,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4416
    },
    {
      "epoch": 0.04417,
      "grad_norm": 0.561088827608562,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 4417
    },
    {
      "epoch": 0.04418,
      "grad_norm": 0.5718569607650594,
      "learning_rate": 0.003,
      "loss": 4.1495,
      "step": 4418
    },
    {
      "epoch": 0.04419,
      "grad_norm": 0.502350939750905,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4419
    },
    {
      "epoch": 0.0442,
      "grad_norm": 0.45571038809991676,
      "learning_rate": 0.003,
      "loss": 4.1321,
      "step": 4420
    },
    {
      "epoch": 0.04421,
      "grad_norm": 0.42375826081335205,
      "learning_rate": 0.003,
      "loss": 4.1625,
      "step": 4421
    },
    {
      "epoch": 0.04422,
      "grad_norm": 0.42008823401459217,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4422
    },
    {
      "epoch": 0.04423,
      "grad_norm": 0.46008487233587403,
      "learning_rate": 0.003,
      "loss": 4.1419,
      "step": 4423
    },
    {
      "epoch": 0.04424,
      "grad_norm": 0.5469701885775844,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 4424
    },
    {
      "epoch": 0.04425,
      "grad_norm": 0.6192720902784608,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 4425
    },
    {
      "epoch": 0.04426,
      "grad_norm": 0.6752662810050423,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4426
    },
    {
      "epoch": 0.04427,
      "grad_norm": 0.8141717149045535,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4427
    },
    {
      "epoch": 0.04428,
      "grad_norm": 1.1138698669039535,
      "learning_rate": 0.003,
      "loss": 4.1428,
      "step": 4428
    },
    {
      "epoch": 0.04429,
      "grad_norm": 1.1125457214220589,
      "learning_rate": 0.003,
      "loss": 4.1893,
      "step": 4429
    },
    {
      "epoch": 0.0443,
      "grad_norm": 0.7372479542562949,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4430
    },
    {
      "epoch": 0.04431,
      "grad_norm": 0.7590662782504862,
      "learning_rate": 0.003,
      "loss": 4.1616,
      "step": 4431
    },
    {
      "epoch": 0.04432,
      "grad_norm": 0.7851056797042649,
      "learning_rate": 0.003,
      "loss": 4.169,
      "step": 4432
    },
    {
      "epoch": 0.04433,
      "grad_norm": 0.6565615710180471,
      "learning_rate": 0.003,
      "loss": 4.1357,
      "step": 4433
    },
    {
      "epoch": 0.04434,
      "grad_norm": 0.5918951930492853,
      "learning_rate": 0.003,
      "loss": 4.1348,
      "step": 4434
    },
    {
      "epoch": 0.04435,
      "grad_norm": 0.5712466192614544,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4435
    },
    {
      "epoch": 0.04436,
      "grad_norm": 0.6540737676285047,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4436
    },
    {
      "epoch": 0.04437,
      "grad_norm": 0.6512769646169722,
      "learning_rate": 0.003,
      "loss": 4.1438,
      "step": 4437
    },
    {
      "epoch": 0.04438,
      "grad_norm": 0.6313619524940465,
      "learning_rate": 0.003,
      "loss": 4.181,
      "step": 4438
    },
    {
      "epoch": 0.04439,
      "grad_norm": 0.752951040748533,
      "learning_rate": 0.003,
      "loss": 4.1527,
      "step": 4439
    },
    {
      "epoch": 0.0444,
      "grad_norm": 0.7444628766487162,
      "learning_rate": 0.003,
      "loss": 4.1088,
      "step": 4440
    },
    {
      "epoch": 0.04441,
      "grad_norm": 0.7171166547645912,
      "learning_rate": 0.003,
      "loss": 4.1413,
      "step": 4441
    },
    {
      "epoch": 0.04442,
      "grad_norm": 0.7004680425501666,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4442
    },
    {
      "epoch": 0.04443,
      "grad_norm": 0.699693309924618,
      "learning_rate": 0.003,
      "loss": 4.1075,
      "step": 4443
    },
    {
      "epoch": 0.04444,
      "grad_norm": 0.605352884343741,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 4444
    },
    {
      "epoch": 0.04445,
      "grad_norm": 0.6910385491365711,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4445
    },
    {
      "epoch": 0.04446,
      "grad_norm": 0.6750353497967526,
      "learning_rate": 0.003,
      "loss": 4.1098,
      "step": 4446
    },
    {
      "epoch": 0.04447,
      "grad_norm": 0.5704187456661106,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4447
    },
    {
      "epoch": 0.04448,
      "grad_norm": 0.5831627691900045,
      "learning_rate": 0.003,
      "loss": 4.1085,
      "step": 4448
    },
    {
      "epoch": 0.04449,
      "grad_norm": 0.7383838513937211,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4449
    },
    {
      "epoch": 0.0445,
      "grad_norm": 0.9275667982571733,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4450
    },
    {
      "epoch": 0.04451,
      "grad_norm": 1.1278443077545208,
      "learning_rate": 0.003,
      "loss": 4.1515,
      "step": 4451
    },
    {
      "epoch": 0.04452,
      "grad_norm": 0.7200631704178471,
      "learning_rate": 0.003,
      "loss": 4.1521,
      "step": 4452
    },
    {
      "epoch": 0.04453,
      "grad_norm": 0.7319095347479152,
      "learning_rate": 0.003,
      "loss": 4.1624,
      "step": 4453
    },
    {
      "epoch": 0.04454,
      "grad_norm": 0.828360672866726,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4454
    },
    {
      "epoch": 0.04455,
      "grad_norm": 0.8149109397250537,
      "learning_rate": 0.003,
      "loss": 4.1531,
      "step": 4455
    },
    {
      "epoch": 0.04456,
      "grad_norm": 0.7949979530915787,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4456
    },
    {
      "epoch": 0.04457,
      "grad_norm": 0.7815335465443298,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4457
    },
    {
      "epoch": 0.04458,
      "grad_norm": 0.6667090111498273,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4458
    },
    {
      "epoch": 0.04459,
      "grad_norm": 0.6478410064314194,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4459
    },
    {
      "epoch": 0.0446,
      "grad_norm": 0.692693711021832,
      "learning_rate": 0.003,
      "loss": 4.1619,
      "step": 4460
    },
    {
      "epoch": 0.04461,
      "grad_norm": 0.7661349867430685,
      "learning_rate": 0.003,
      "loss": 4.1671,
      "step": 4461
    },
    {
      "epoch": 0.04462,
      "grad_norm": 0.9456365590075492,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4462
    },
    {
      "epoch": 0.04463,
      "grad_norm": 0.9219959129201606,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4463
    },
    {
      "epoch": 0.04464,
      "grad_norm": 0.8148900268740181,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4464
    },
    {
      "epoch": 0.04465,
      "grad_norm": 0.8174238600511685,
      "learning_rate": 0.003,
      "loss": 4.1486,
      "step": 4465
    },
    {
      "epoch": 0.04466,
      "grad_norm": 0.6910175481107252,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4466
    },
    {
      "epoch": 0.04467,
      "grad_norm": 0.8503966818321036,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4467
    },
    {
      "epoch": 0.04468,
      "grad_norm": 0.8010356993406599,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4468
    },
    {
      "epoch": 0.04469,
      "grad_norm": 0.7388705798425241,
      "learning_rate": 0.003,
      "loss": 4.1577,
      "step": 4469
    },
    {
      "epoch": 0.0447,
      "grad_norm": 0.776572618481807,
      "learning_rate": 0.003,
      "loss": 4.1627,
      "step": 4470
    },
    {
      "epoch": 0.04471,
      "grad_norm": 0.8915085141537737,
      "learning_rate": 0.003,
      "loss": 4.185,
      "step": 4471
    },
    {
      "epoch": 0.04472,
      "grad_norm": 0.9200136447240024,
      "learning_rate": 0.003,
      "loss": 4.1398,
      "step": 4472
    },
    {
      "epoch": 0.04473,
      "grad_norm": 0.8156139485901094,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4473
    },
    {
      "epoch": 0.04474,
      "grad_norm": 0.7239763646047743,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4474
    },
    {
      "epoch": 0.04475,
      "grad_norm": 0.6835215989709738,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 4475
    },
    {
      "epoch": 0.04476,
      "grad_norm": 0.6497813876295598,
      "learning_rate": 0.003,
      "loss": 4.1303,
      "step": 4476
    },
    {
      "epoch": 0.04477,
      "grad_norm": 0.647414838457496,
      "learning_rate": 0.003,
      "loss": 4.1467,
      "step": 4477
    },
    {
      "epoch": 0.04478,
      "grad_norm": 0.559137538294316,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4478
    },
    {
      "epoch": 0.04479,
      "grad_norm": 0.5859013586768095,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4479
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.5435317461744096,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4480
    },
    {
      "epoch": 0.04481,
      "grad_norm": 0.6011946337549432,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 4481
    },
    {
      "epoch": 0.04482,
      "grad_norm": 0.7806327420943877,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4482
    },
    {
      "epoch": 0.04483,
      "grad_norm": 1.0587276903826628,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4483
    },
    {
      "epoch": 0.04484,
      "grad_norm": 0.9760427349795051,
      "learning_rate": 0.003,
      "loss": 4.1662,
      "step": 4484
    },
    {
      "epoch": 0.04485,
      "grad_norm": 0.7488828747485865,
      "learning_rate": 0.003,
      "loss": 4.1693,
      "step": 4485
    },
    {
      "epoch": 0.04486,
      "grad_norm": 0.8084521903070014,
      "learning_rate": 0.003,
      "loss": 4.1294,
      "step": 4486
    },
    {
      "epoch": 0.04487,
      "grad_norm": 0.9117696333780938,
      "learning_rate": 0.003,
      "loss": 4.1426,
      "step": 4487
    },
    {
      "epoch": 0.04488,
      "grad_norm": 0.8996969646129346,
      "learning_rate": 0.003,
      "loss": 4.163,
      "step": 4488
    },
    {
      "epoch": 0.04489,
      "grad_norm": 0.8641616901342983,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4489
    },
    {
      "epoch": 0.0449,
      "grad_norm": 0.7280388583966139,
      "learning_rate": 0.003,
      "loss": 4.1572,
      "step": 4490
    },
    {
      "epoch": 0.04491,
      "grad_norm": 0.6733842898509934,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4491
    },
    {
      "epoch": 0.04492,
      "grad_norm": 0.6313608339838067,
      "learning_rate": 0.003,
      "loss": 4.1544,
      "step": 4492
    },
    {
      "epoch": 0.04493,
      "grad_norm": 0.6192244819984872,
      "learning_rate": 0.003,
      "loss": 4.1609,
      "step": 4493
    },
    {
      "epoch": 0.04494,
      "grad_norm": 0.5488813855675564,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4494
    },
    {
      "epoch": 0.04495,
      "grad_norm": 0.4779659159856146,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4495
    },
    {
      "epoch": 0.04496,
      "grad_norm": 0.5060747754385665,
      "learning_rate": 0.003,
      "loss": 4.1713,
      "step": 4496
    },
    {
      "epoch": 0.04497,
      "grad_norm": 0.5808549525919214,
      "learning_rate": 0.003,
      "loss": 4.1502,
      "step": 4497
    },
    {
      "epoch": 0.04498,
      "grad_norm": 0.6694153212993217,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4498
    },
    {
      "epoch": 0.04499,
      "grad_norm": 0.8090682417638635,
      "learning_rate": 0.003,
      "loss": 4.1013,
      "step": 4499
    },
    {
      "epoch": 0.045,
      "grad_norm": 0.9223841526186204,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4500
    },
    {
      "epoch": 0.04501,
      "grad_norm": 0.8349758444195646,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4501
    },
    {
      "epoch": 0.04502,
      "grad_norm": 0.7254525437975357,
      "learning_rate": 0.003,
      "loss": 4.1694,
      "step": 4502
    },
    {
      "epoch": 0.04503,
      "grad_norm": 0.649056521260754,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4503
    },
    {
      "epoch": 0.04504,
      "grad_norm": 0.713115140118379,
      "learning_rate": 0.003,
      "loss": 4.1672,
      "step": 4504
    },
    {
      "epoch": 0.04505,
      "grad_norm": 0.6117909603348726,
      "learning_rate": 0.003,
      "loss": 4.1562,
      "step": 4505
    },
    {
      "epoch": 0.04506,
      "grad_norm": 0.6145202957029064,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4506
    },
    {
      "epoch": 0.04507,
      "grad_norm": 0.5247412068715119,
      "learning_rate": 0.003,
      "loss": 4.1062,
      "step": 4507
    },
    {
      "epoch": 0.04508,
      "grad_norm": 0.5806017179220667,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4508
    },
    {
      "epoch": 0.04509,
      "grad_norm": 0.5825330876210861,
      "learning_rate": 0.003,
      "loss": 4.1497,
      "step": 4509
    },
    {
      "epoch": 0.0451,
      "grad_norm": 0.7871833251091375,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4510
    },
    {
      "epoch": 0.04511,
      "grad_norm": 0.8750417888525687,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4511
    },
    {
      "epoch": 0.04512,
      "grad_norm": 0.8564995340864695,
      "learning_rate": 0.003,
      "loss": 4.0997,
      "step": 4512
    },
    {
      "epoch": 0.04513,
      "grad_norm": 0.7843001776115731,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4513
    },
    {
      "epoch": 0.04514,
      "grad_norm": 0.6796657929720622,
      "learning_rate": 0.003,
      "loss": 4.1384,
      "step": 4514
    },
    {
      "epoch": 0.04515,
      "grad_norm": 0.6410320420606614,
      "learning_rate": 0.003,
      "loss": 4.1311,
      "step": 4515
    },
    {
      "epoch": 0.04516,
      "grad_norm": 0.6710348380128827,
      "learning_rate": 0.003,
      "loss": 4.1424,
      "step": 4516
    },
    {
      "epoch": 0.04517,
      "grad_norm": 0.8318718390934561,
      "learning_rate": 0.003,
      "loss": 4.1637,
      "step": 4517
    },
    {
      "epoch": 0.04518,
      "grad_norm": 0.895874624608404,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4518
    },
    {
      "epoch": 0.04519,
      "grad_norm": 0.9529303615004293,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4519
    },
    {
      "epoch": 0.0452,
      "grad_norm": 1.00692610224058,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4520
    },
    {
      "epoch": 0.04521,
      "grad_norm": 1.0366622726930343,
      "learning_rate": 0.003,
      "loss": 4.1524,
      "step": 4521
    },
    {
      "epoch": 0.04522,
      "grad_norm": 1.0505298007829553,
      "learning_rate": 0.003,
      "loss": 4.1571,
      "step": 4522
    },
    {
      "epoch": 0.04523,
      "grad_norm": 1.0189524405337644,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4523
    },
    {
      "epoch": 0.04524,
      "grad_norm": 1.030832446298102,
      "learning_rate": 0.003,
      "loss": 4.138,
      "step": 4524
    },
    {
      "epoch": 0.04525,
      "grad_norm": 1.0054730825905025,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4525
    },
    {
      "epoch": 0.04526,
      "grad_norm": 0.9357043959907038,
      "learning_rate": 0.003,
      "loss": 4.1777,
      "step": 4526
    },
    {
      "epoch": 0.04527,
      "grad_norm": 0.880967833594335,
      "learning_rate": 0.003,
      "loss": 4.1602,
      "step": 4527
    },
    {
      "epoch": 0.04528,
      "grad_norm": 0.95913766957737,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4528
    },
    {
      "epoch": 0.04529,
      "grad_norm": 0.7989651014070877,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4529
    },
    {
      "epoch": 0.0453,
      "grad_norm": 0.7380961288194754,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4530
    },
    {
      "epoch": 0.04531,
      "grad_norm": 0.7085687602336856,
      "learning_rate": 0.003,
      "loss": 4.1338,
      "step": 4531
    },
    {
      "epoch": 0.04532,
      "grad_norm": 0.6574885202110717,
      "learning_rate": 0.003,
      "loss": 4.16,
      "step": 4532
    },
    {
      "epoch": 0.04533,
      "grad_norm": 0.6519010842998008,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4533
    },
    {
      "epoch": 0.04534,
      "grad_norm": 0.6093622128244879,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4534
    },
    {
      "epoch": 0.04535,
      "grad_norm": 0.5837584644000219,
      "learning_rate": 0.003,
      "loss": 4.1784,
      "step": 4535
    },
    {
      "epoch": 0.04536,
      "grad_norm": 0.6897731390503988,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 4536
    },
    {
      "epoch": 0.04537,
      "grad_norm": 0.6949801331902556,
      "learning_rate": 0.003,
      "loss": 4.1517,
      "step": 4537
    },
    {
      "epoch": 0.04538,
      "grad_norm": 0.6776975735023881,
      "learning_rate": 0.003,
      "loss": 4.15,
      "step": 4538
    },
    {
      "epoch": 0.04539,
      "grad_norm": 0.678568300988415,
      "learning_rate": 0.003,
      "loss": 4.1144,
      "step": 4539
    },
    {
      "epoch": 0.0454,
      "grad_norm": 0.6603018556737656,
      "learning_rate": 0.003,
      "loss": 4.1223,
      "step": 4540
    },
    {
      "epoch": 0.04541,
      "grad_norm": 0.6133835370391497,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4541
    },
    {
      "epoch": 0.04542,
      "grad_norm": 0.635535341053592,
      "learning_rate": 0.003,
      "loss": 4.1128,
      "step": 4542
    },
    {
      "epoch": 0.04543,
      "grad_norm": 0.668577889488378,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4543
    },
    {
      "epoch": 0.04544,
      "grad_norm": 0.7427455997801305,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4544
    },
    {
      "epoch": 0.04545,
      "grad_norm": 0.8639923142163211,
      "learning_rate": 0.003,
      "loss": 4.1405,
      "step": 4545
    },
    {
      "epoch": 0.04546,
      "grad_norm": 0.6895806692659665,
      "learning_rate": 0.003,
      "loss": 4.121,
      "step": 4546
    },
    {
      "epoch": 0.04547,
      "grad_norm": 0.49153824435946125,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4547
    },
    {
      "epoch": 0.04548,
      "grad_norm": 0.5516451722786672,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4548
    },
    {
      "epoch": 0.04549,
      "grad_norm": 0.6029789439713806,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4549
    },
    {
      "epoch": 0.0455,
      "grad_norm": 0.66622853988702,
      "learning_rate": 0.003,
      "loss": 4.1547,
      "step": 4550
    },
    {
      "epoch": 0.04551,
      "grad_norm": 0.7238008534924777,
      "learning_rate": 0.003,
      "loss": 4.1669,
      "step": 4551
    },
    {
      "epoch": 0.04552,
      "grad_norm": 0.7824263722798367,
      "learning_rate": 0.003,
      "loss": 4.1509,
      "step": 4552
    },
    {
      "epoch": 0.04553,
      "grad_norm": 0.7388703999593268,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4553
    },
    {
      "epoch": 0.04554,
      "grad_norm": 0.6913086022347059,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4554
    },
    {
      "epoch": 0.04555,
      "grad_norm": 0.6442867508319275,
      "learning_rate": 0.003,
      "loss": 4.1325,
      "step": 4555
    },
    {
      "epoch": 0.04556,
      "grad_norm": 0.7238849740450599,
      "learning_rate": 0.003,
      "loss": 4.1389,
      "step": 4556
    },
    {
      "epoch": 0.04557,
      "grad_norm": 0.7976327742193281,
      "learning_rate": 0.003,
      "loss": 4.1236,
      "step": 4557
    },
    {
      "epoch": 0.04558,
      "grad_norm": 0.8032287746582655,
      "learning_rate": 0.003,
      "loss": 4.1552,
      "step": 4558
    },
    {
      "epoch": 0.04559,
      "grad_norm": 0.784841139322348,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4559
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.8095586358578389,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4560
    },
    {
      "epoch": 0.04561,
      "grad_norm": 0.6748297662204855,
      "learning_rate": 0.003,
      "loss": 4.1342,
      "step": 4561
    },
    {
      "epoch": 0.04562,
      "grad_norm": 0.6987048066021282,
      "learning_rate": 0.003,
      "loss": 4.1565,
      "step": 4562
    },
    {
      "epoch": 0.04563,
      "grad_norm": 0.772254646470346,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4563
    },
    {
      "epoch": 0.04564,
      "grad_norm": 0.8035846762926936,
      "learning_rate": 0.003,
      "loss": 4.1418,
      "step": 4564
    },
    {
      "epoch": 0.04565,
      "grad_norm": 0.7913541854291436,
      "learning_rate": 0.003,
      "loss": 4.1146,
      "step": 4565
    },
    {
      "epoch": 0.04566,
      "grad_norm": 0.7821636491006997,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4566
    },
    {
      "epoch": 0.04567,
      "grad_norm": 0.6863252350497664,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4567
    },
    {
      "epoch": 0.04568,
      "grad_norm": 0.6211213056274318,
      "learning_rate": 0.003,
      "loss": 4.112,
      "step": 4568
    },
    {
      "epoch": 0.04569,
      "grad_norm": 0.7600485305292163,
      "learning_rate": 0.003,
      "loss": 4.1442,
      "step": 4569
    },
    {
      "epoch": 0.0457,
      "grad_norm": 0.7935297140012192,
      "learning_rate": 0.003,
      "loss": 4.1368,
      "step": 4570
    },
    {
      "epoch": 0.04571,
      "grad_norm": 0.8707627695603387,
      "learning_rate": 0.003,
      "loss": 4.1688,
      "step": 4571
    },
    {
      "epoch": 0.04572,
      "grad_norm": 0.9504850878398086,
      "learning_rate": 0.003,
      "loss": 4.146,
      "step": 4572
    },
    {
      "epoch": 0.04573,
      "grad_norm": 0.9772978151032115,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4573
    },
    {
      "epoch": 0.04574,
      "grad_norm": 0.8349860867562747,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4574
    },
    {
      "epoch": 0.04575,
      "grad_norm": 0.7275515356936376,
      "learning_rate": 0.003,
      "loss": 4.1579,
      "step": 4575
    },
    {
      "epoch": 0.04576,
      "grad_norm": 0.7453152360629928,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4576
    },
    {
      "epoch": 0.04577,
      "grad_norm": 0.731228336021725,
      "learning_rate": 0.003,
      "loss": 4.1686,
      "step": 4577
    },
    {
      "epoch": 0.04578,
      "grad_norm": 0.7565651611671647,
      "learning_rate": 0.003,
      "loss": 4.1355,
      "step": 4578
    },
    {
      "epoch": 0.04579,
      "grad_norm": 0.9198271859494617,
      "learning_rate": 0.003,
      "loss": 4.1202,
      "step": 4579
    },
    {
      "epoch": 0.0458,
      "grad_norm": 1.0574643718931676,
      "learning_rate": 0.003,
      "loss": 4.168,
      "step": 4580
    },
    {
      "epoch": 0.04581,
      "grad_norm": 0.9388725861321914,
      "learning_rate": 0.003,
      "loss": 4.1802,
      "step": 4581
    },
    {
      "epoch": 0.04582,
      "grad_norm": 0.7599199575014808,
      "learning_rate": 0.003,
      "loss": 4.1742,
      "step": 4582
    },
    {
      "epoch": 0.04583,
      "grad_norm": 0.6683292234471547,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4583
    },
    {
      "epoch": 0.04584,
      "grad_norm": 0.7512487692406998,
      "learning_rate": 0.003,
      "loss": 4.1417,
      "step": 4584
    },
    {
      "epoch": 0.04585,
      "grad_norm": 0.7081640921869536,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4585
    },
    {
      "epoch": 0.04586,
      "grad_norm": 0.7168079051261622,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4586
    },
    {
      "epoch": 0.04587,
      "grad_norm": 0.814216089662895,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4587
    },
    {
      "epoch": 0.04588,
      "grad_norm": 0.8322807065043963,
      "learning_rate": 0.003,
      "loss": 4.1661,
      "step": 4588
    },
    {
      "epoch": 0.04589,
      "grad_norm": 0.8173670039992074,
      "learning_rate": 0.003,
      "loss": 4.1285,
      "step": 4589
    },
    {
      "epoch": 0.0459,
      "grad_norm": 0.7002687692328883,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4590
    },
    {
      "epoch": 0.04591,
      "grad_norm": 0.667313250694922,
      "learning_rate": 0.003,
      "loss": 4.1141,
      "step": 4591
    },
    {
      "epoch": 0.04592,
      "grad_norm": 0.7721193873262023,
      "learning_rate": 0.003,
      "loss": 4.1435,
      "step": 4592
    },
    {
      "epoch": 0.04593,
      "grad_norm": 0.719951932975662,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4593
    },
    {
      "epoch": 0.04594,
      "grad_norm": 0.624806723533564,
      "learning_rate": 0.003,
      "loss": 4.1559,
      "step": 4594
    },
    {
      "epoch": 0.04595,
      "grad_norm": 0.5756945318000921,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4595
    },
    {
      "epoch": 0.04596,
      "grad_norm": 0.5657179032543712,
      "learning_rate": 0.003,
      "loss": 4.1387,
      "step": 4596
    },
    {
      "epoch": 0.04597,
      "grad_norm": 0.6256040629924149,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4597
    },
    {
      "epoch": 0.04598,
      "grad_norm": 0.6013110035097162,
      "learning_rate": 0.003,
      "loss": 4.1762,
      "step": 4598
    },
    {
      "epoch": 0.04599,
      "grad_norm": 0.7296067063170582,
      "learning_rate": 0.003,
      "loss": 4.1254,
      "step": 4599
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.9209598285204096,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4600
    },
    {
      "epoch": 0.04601,
      "grad_norm": 1.0160633328517232,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4601
    },
    {
      "epoch": 0.04602,
      "grad_norm": 0.8169968744441843,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4602
    },
    {
      "epoch": 0.04603,
      "grad_norm": 0.659664339286615,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4603
    },
    {
      "epoch": 0.04604,
      "grad_norm": 0.6592904260797935,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4604
    },
    {
      "epoch": 0.04605,
      "grad_norm": 0.6689967757886678,
      "learning_rate": 0.003,
      "loss": 4.159,
      "step": 4605
    },
    {
      "epoch": 0.04606,
      "grad_norm": 0.6839466982971314,
      "learning_rate": 0.003,
      "loss": 4.1205,
      "step": 4606
    },
    {
      "epoch": 0.04607,
      "grad_norm": 0.6080788219356916,
      "learning_rate": 0.003,
      "loss": 4.1281,
      "step": 4607
    },
    {
      "epoch": 0.04608,
      "grad_norm": 0.6229335033564855,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4608
    },
    {
      "epoch": 0.04609,
      "grad_norm": 0.6110178676005829,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4609
    },
    {
      "epoch": 0.0461,
      "grad_norm": 0.6009057478520481,
      "learning_rate": 0.003,
      "loss": 4.1111,
      "step": 4610
    },
    {
      "epoch": 0.04611,
      "grad_norm": 0.667830986428792,
      "learning_rate": 0.003,
      "loss": 4.1615,
      "step": 4611
    },
    {
      "epoch": 0.04612,
      "grad_norm": 0.8308783400085262,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4612
    },
    {
      "epoch": 0.04613,
      "grad_norm": 0.9122587432173699,
      "learning_rate": 0.003,
      "loss": 4.1372,
      "step": 4613
    },
    {
      "epoch": 0.04614,
      "grad_norm": 1.0201054294863492,
      "learning_rate": 0.003,
      "loss": 4.1091,
      "step": 4614
    },
    {
      "epoch": 0.04615,
      "grad_norm": 0.7968952064740754,
      "learning_rate": 0.003,
      "loss": 4.145,
      "step": 4615
    },
    {
      "epoch": 0.04616,
      "grad_norm": 0.7040440287863108,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4616
    },
    {
      "epoch": 0.04617,
      "grad_norm": 0.650743607835788,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4617
    },
    {
      "epoch": 0.04618,
      "grad_norm": 0.6876364900587607,
      "learning_rate": 0.003,
      "loss": 4.1292,
      "step": 4618
    },
    {
      "epoch": 0.04619,
      "grad_norm": 0.674403145056888,
      "learning_rate": 0.003,
      "loss": 4.1306,
      "step": 4619
    },
    {
      "epoch": 0.0462,
      "grad_norm": 0.6278438951559039,
      "learning_rate": 0.003,
      "loss": 4.158,
      "step": 4620
    },
    {
      "epoch": 0.04621,
      "grad_norm": 0.5882186858354869,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4621
    },
    {
      "epoch": 0.04622,
      "grad_norm": 0.6093666817375198,
      "learning_rate": 0.003,
      "loss": 4.155,
      "step": 4622
    },
    {
      "epoch": 0.04623,
      "grad_norm": 0.5680931823986678,
      "learning_rate": 0.003,
      "loss": 4.1361,
      "step": 4623
    },
    {
      "epoch": 0.04624,
      "grad_norm": 0.6168812041909036,
      "learning_rate": 0.003,
      "loss": 4.1153,
      "step": 4624
    },
    {
      "epoch": 0.04625,
      "grad_norm": 0.6071577011256134,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4625
    },
    {
      "epoch": 0.04626,
      "grad_norm": 0.5293202352744564,
      "learning_rate": 0.003,
      "loss": 4.1344,
      "step": 4626
    },
    {
      "epoch": 0.04627,
      "grad_norm": 0.5193967115713368,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4627
    },
    {
      "epoch": 0.04628,
      "grad_norm": 0.6344196546552844,
      "learning_rate": 0.003,
      "loss": 4.1513,
      "step": 4628
    },
    {
      "epoch": 0.04629,
      "grad_norm": 0.6660491057377373,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4629
    },
    {
      "epoch": 0.0463,
      "grad_norm": 0.6585553354499009,
      "learning_rate": 0.003,
      "loss": 4.1415,
      "step": 4630
    },
    {
      "epoch": 0.04631,
      "grad_norm": 0.6065008696674108,
      "learning_rate": 0.003,
      "loss": 4.1204,
      "step": 4631
    },
    {
      "epoch": 0.04632,
      "grad_norm": 0.5310090972041924,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 4632
    },
    {
      "epoch": 0.04633,
      "grad_norm": 0.6357968891584513,
      "learning_rate": 0.003,
      "loss": 4.1147,
      "step": 4633
    },
    {
      "epoch": 0.04634,
      "grad_norm": 0.941578980575187,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4634
    },
    {
      "epoch": 0.04635,
      "grad_norm": 1.3572160948425378,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4635
    },
    {
      "epoch": 0.04636,
      "grad_norm": 0.6013442724378576,
      "learning_rate": 0.003,
      "loss": 4.1656,
      "step": 4636
    },
    {
      "epoch": 0.04637,
      "grad_norm": 0.6555645652247263,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4637
    },
    {
      "epoch": 0.04638,
      "grad_norm": 0.7328409108901823,
      "learning_rate": 0.003,
      "loss": 4.1489,
      "step": 4638
    },
    {
      "epoch": 0.04639,
      "grad_norm": 0.6867315466218923,
      "learning_rate": 0.003,
      "loss": 4.1174,
      "step": 4639
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.78928501351717,
      "learning_rate": 0.003,
      "loss": 4.1452,
      "step": 4640
    },
    {
      "epoch": 0.04641,
      "grad_norm": 0.8455497454514354,
      "learning_rate": 0.003,
      "loss": 4.1604,
      "step": 4641
    },
    {
      "epoch": 0.04642,
      "grad_norm": 0.8465515736905618,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4642
    },
    {
      "epoch": 0.04643,
      "grad_norm": 0.7201267511516922,
      "learning_rate": 0.003,
      "loss": 4.1302,
      "step": 4643
    },
    {
      "epoch": 0.04644,
      "grad_norm": 0.7720786991059095,
      "learning_rate": 0.003,
      "loss": 4.1373,
      "step": 4644
    },
    {
      "epoch": 0.04645,
      "grad_norm": 0.7658630542557853,
      "learning_rate": 0.003,
      "loss": 4.1299,
      "step": 4645
    },
    {
      "epoch": 0.04646,
      "grad_norm": 0.8677913304110078,
      "learning_rate": 0.003,
      "loss": 4.1365,
      "step": 4646
    },
    {
      "epoch": 0.04647,
      "grad_norm": 0.8914361615284006,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4647
    },
    {
      "epoch": 0.04648,
      "grad_norm": 0.8603752804558079,
      "learning_rate": 0.003,
      "loss": 4.1564,
      "step": 4648
    },
    {
      "epoch": 0.04649,
      "grad_norm": 0.8183443337243556,
      "learning_rate": 0.003,
      "loss": 4.1396,
      "step": 4649
    },
    {
      "epoch": 0.0465,
      "grad_norm": 0.6592577702353686,
      "learning_rate": 0.003,
      "loss": 4.1038,
      "step": 4650
    },
    {
      "epoch": 0.04651,
      "grad_norm": 0.6033831169642737,
      "learning_rate": 0.003,
      "loss": 4.1363,
      "step": 4651
    },
    {
      "epoch": 0.04652,
      "grad_norm": 0.6694785473055216,
      "learning_rate": 0.003,
      "loss": 4.1447,
      "step": 4652
    },
    {
      "epoch": 0.04653,
      "grad_norm": 0.6797455964877727,
      "learning_rate": 0.003,
      "loss": 4.143,
      "step": 4653
    },
    {
      "epoch": 0.04654,
      "grad_norm": 0.7400041353958778,
      "learning_rate": 0.003,
      "loss": 4.139,
      "step": 4654
    },
    {
      "epoch": 0.04655,
      "grad_norm": 0.884024246484862,
      "learning_rate": 0.003,
      "loss": 4.1483,
      "step": 4655
    },
    {
      "epoch": 0.04656,
      "grad_norm": 0.9724046681595542,
      "learning_rate": 0.003,
      "loss": 4.127,
      "step": 4656
    },
    {
      "epoch": 0.04657,
      "grad_norm": 0.9321801110747265,
      "learning_rate": 0.003,
      "loss": 4.1641,
      "step": 4657
    },
    {
      "epoch": 0.04658,
      "grad_norm": 0.7977247035986766,
      "learning_rate": 0.003,
      "loss": 4.1359,
      "step": 4658
    },
    {
      "epoch": 0.04659,
      "grad_norm": 0.7837947068767307,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4659
    },
    {
      "epoch": 0.0466,
      "grad_norm": 0.8489578033894326,
      "learning_rate": 0.003,
      "loss": 4.1561,
      "step": 4660
    },
    {
      "epoch": 0.04661,
      "grad_norm": 0.9070979393786719,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4661
    },
    {
      "epoch": 0.04662,
      "grad_norm": 0.9271278255074006,
      "learning_rate": 0.003,
      "loss": 4.1364,
      "step": 4662
    },
    {
      "epoch": 0.04663,
      "grad_norm": 0.9626508931970053,
      "learning_rate": 0.003,
      "loss": 4.1618,
      "step": 4663
    },
    {
      "epoch": 0.04664,
      "grad_norm": 1.112988241634731,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4664
    },
    {
      "epoch": 0.04665,
      "grad_norm": 0.8921340954370226,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 4665
    },
    {
      "epoch": 0.04666,
      "grad_norm": 0.7837443287711118,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4666
    },
    {
      "epoch": 0.04667,
      "grad_norm": 0.7219249423509705,
      "learning_rate": 0.003,
      "loss": 4.1315,
      "step": 4667
    },
    {
      "epoch": 0.04668,
      "grad_norm": 0.7307342741354349,
      "learning_rate": 0.003,
      "loss": 4.1076,
      "step": 4668
    },
    {
      "epoch": 0.04669,
      "grad_norm": 0.7401471208519357,
      "learning_rate": 0.003,
      "loss": 4.1599,
      "step": 4669
    },
    {
      "epoch": 0.0467,
      "grad_norm": 0.7077495687150961,
      "learning_rate": 0.003,
      "loss": 4.1097,
      "step": 4670
    },
    {
      "epoch": 0.04671,
      "grad_norm": 0.7086785987292082,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4671
    },
    {
      "epoch": 0.04672,
      "grad_norm": 0.7271653898870789,
      "learning_rate": 0.003,
      "loss": 4.1488,
      "step": 4672
    },
    {
      "epoch": 0.04673,
      "grad_norm": 0.7152687743613511,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4673
    },
    {
      "epoch": 0.04674,
      "grad_norm": 0.8387550898694738,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4674
    },
    {
      "epoch": 0.04675,
      "grad_norm": 0.9264998756521228,
      "learning_rate": 0.003,
      "loss": 4.156,
      "step": 4675
    },
    {
      "epoch": 0.04676,
      "grad_norm": 0.9760003719176258,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4676
    },
    {
      "epoch": 0.04677,
      "grad_norm": 0.9605322775687504,
      "learning_rate": 0.003,
      "loss": 4.1698,
      "step": 4677
    },
    {
      "epoch": 0.04678,
      "grad_norm": 0.760672310236161,
      "learning_rate": 0.003,
      "loss": 4.1192,
      "step": 4678
    },
    {
      "epoch": 0.04679,
      "grad_norm": 0.6288734540073927,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4679
    },
    {
      "epoch": 0.0468,
      "grad_norm": 0.5340420649165649,
      "learning_rate": 0.003,
      "loss": 4.1119,
      "step": 4680
    },
    {
      "epoch": 0.04681,
      "grad_norm": 0.5383218654638344,
      "learning_rate": 0.003,
      "loss": 4.1632,
      "step": 4681
    },
    {
      "epoch": 0.04682,
      "grad_norm": 0.5921170754815444,
      "learning_rate": 0.003,
      "loss": 4.1305,
      "step": 4682
    },
    {
      "epoch": 0.04683,
      "grad_norm": 0.5986112675201656,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4683
    },
    {
      "epoch": 0.04684,
      "grad_norm": 0.6515018805380663,
      "learning_rate": 0.003,
      "loss": 4.1463,
      "step": 4684
    },
    {
      "epoch": 0.04685,
      "grad_norm": 0.6225025584760357,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4685
    },
    {
      "epoch": 0.04686,
      "grad_norm": 0.5748567786413258,
      "learning_rate": 0.003,
      "loss": 4.1317,
      "step": 4686
    },
    {
      "epoch": 0.04687,
      "grad_norm": 0.597224880471858,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4687
    },
    {
      "epoch": 0.04688,
      "grad_norm": 0.7033215388348866,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4688
    },
    {
      "epoch": 0.04689,
      "grad_norm": 0.8369783564473339,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 4689
    },
    {
      "epoch": 0.0469,
      "grad_norm": 0.8637378056248713,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4690
    },
    {
      "epoch": 0.04691,
      "grad_norm": 0.9863994238127417,
      "learning_rate": 0.003,
      "loss": 4.1376,
      "step": 4691
    },
    {
      "epoch": 0.04692,
      "grad_norm": 1.1588055062272775,
      "learning_rate": 0.003,
      "loss": 4.1525,
      "step": 4692
    },
    {
      "epoch": 0.04693,
      "grad_norm": 1.0154258289459517,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4693
    },
    {
      "epoch": 0.04694,
      "grad_norm": 0.8542851528225356,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4694
    },
    {
      "epoch": 0.04695,
      "grad_norm": 0.7936283873091949,
      "learning_rate": 0.003,
      "loss": 4.1222,
      "step": 4695
    },
    {
      "epoch": 0.04696,
      "grad_norm": 0.8706455027386264,
      "learning_rate": 0.003,
      "loss": 4.1712,
      "step": 4696
    },
    {
      "epoch": 0.04697,
      "grad_norm": 0.888025520306328,
      "learning_rate": 0.003,
      "loss": 4.1555,
      "step": 4697
    },
    {
      "epoch": 0.04698,
      "grad_norm": 0.8024507884497375,
      "learning_rate": 0.003,
      "loss": 4.1554,
      "step": 4698
    },
    {
      "epoch": 0.04699,
      "grad_norm": 0.6716482593675086,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4699
    },
    {
      "epoch": 0.047,
      "grad_norm": 0.6875782963509716,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4700
    },
    {
      "epoch": 0.04701,
      "grad_norm": 0.6233215143134673,
      "learning_rate": 0.003,
      "loss": 4.1581,
      "step": 4701
    },
    {
      "epoch": 0.04702,
      "grad_norm": 0.5392566756700929,
      "learning_rate": 0.003,
      "loss": 4.1455,
      "step": 4702
    },
    {
      "epoch": 0.04703,
      "grad_norm": 0.5917569266603563,
      "learning_rate": 0.003,
      "loss": 4.1351,
      "step": 4703
    },
    {
      "epoch": 0.04704,
      "grad_norm": 0.6011918134456015,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4704
    },
    {
      "epoch": 0.04705,
      "grad_norm": 0.6280725769151366,
      "learning_rate": 0.003,
      "loss": 4.1567,
      "step": 4705
    },
    {
      "epoch": 0.04706,
      "grad_norm": 0.6696710280228617,
      "learning_rate": 0.003,
      "loss": 4.101,
      "step": 4706
    },
    {
      "epoch": 0.04707,
      "grad_norm": 0.5908607329438127,
      "learning_rate": 0.003,
      "loss": 4.1403,
      "step": 4707
    },
    {
      "epoch": 0.04708,
      "grad_norm": 0.5153382206986533,
      "learning_rate": 0.003,
      "loss": 4.0903,
      "step": 4708
    },
    {
      "epoch": 0.04709,
      "grad_norm": 0.513601992478386,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4709
    },
    {
      "epoch": 0.0471,
      "grad_norm": 0.6490521053147635,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4710
    },
    {
      "epoch": 0.04711,
      "grad_norm": 0.7794691577627103,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4711
    },
    {
      "epoch": 0.04712,
      "grad_norm": 0.7688497592240825,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4712
    },
    {
      "epoch": 0.04713,
      "grad_norm": 0.8439539645333411,
      "learning_rate": 0.003,
      "loss": 4.1188,
      "step": 4713
    },
    {
      "epoch": 0.04714,
      "grad_norm": 0.9146059884037103,
      "learning_rate": 0.003,
      "loss": 4.1343,
      "step": 4714
    },
    {
      "epoch": 0.04715,
      "grad_norm": 0.8646027137635436,
      "learning_rate": 0.003,
      "loss": 4.1607,
      "step": 4715
    },
    {
      "epoch": 0.04716,
      "grad_norm": 0.7274583804499275,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4716
    },
    {
      "epoch": 0.04717,
      "grad_norm": 0.683321783957918,
      "learning_rate": 0.003,
      "loss": 4.144,
      "step": 4717
    },
    {
      "epoch": 0.04718,
      "grad_norm": 0.7716759413978014,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4718
    },
    {
      "epoch": 0.04719,
      "grad_norm": 0.79629647999161,
      "learning_rate": 0.003,
      "loss": 4.1219,
      "step": 4719
    },
    {
      "epoch": 0.0472,
      "grad_norm": 0.7050760628166665,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4720
    },
    {
      "epoch": 0.04721,
      "grad_norm": 0.6252610323631939,
      "learning_rate": 0.003,
      "loss": 4.1074,
      "step": 4721
    },
    {
      "epoch": 0.04722,
      "grad_norm": 0.6123095727231639,
      "learning_rate": 0.003,
      "loss": 4.1245,
      "step": 4722
    },
    {
      "epoch": 0.04723,
      "grad_norm": 0.6308650251833371,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4723
    },
    {
      "epoch": 0.04724,
      "grad_norm": 0.6995282704649718,
      "learning_rate": 0.003,
      "loss": 4.1122,
      "step": 4724
    },
    {
      "epoch": 0.04725,
      "grad_norm": 0.822004537328782,
      "learning_rate": 0.003,
      "loss": 4.141,
      "step": 4725
    },
    {
      "epoch": 0.04726,
      "grad_norm": 0.9108754250302765,
      "learning_rate": 0.003,
      "loss": 4.1337,
      "step": 4726
    },
    {
      "epoch": 0.04727,
      "grad_norm": 0.9359931741571359,
      "learning_rate": 0.003,
      "loss": 4.1259,
      "step": 4727
    },
    {
      "epoch": 0.04728,
      "grad_norm": 0.8319271568838779,
      "learning_rate": 0.003,
      "loss": 4.1298,
      "step": 4728
    },
    {
      "epoch": 0.04729,
      "grad_norm": 0.7272073389712985,
      "learning_rate": 0.003,
      "loss": 4.1456,
      "step": 4729
    },
    {
      "epoch": 0.0473,
      "grad_norm": 0.6527677653973615,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4730
    },
    {
      "epoch": 0.04731,
      "grad_norm": 0.5930657999039424,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4731
    },
    {
      "epoch": 0.04732,
      "grad_norm": 0.6549784460880107,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4732
    },
    {
      "epoch": 0.04733,
      "grad_norm": 0.8327578035736742,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4733
    },
    {
      "epoch": 0.04734,
      "grad_norm": 1.104638767629771,
      "learning_rate": 0.003,
      "loss": 4.1512,
      "step": 4734
    },
    {
      "epoch": 0.04735,
      "grad_norm": 0.9388734582276314,
      "learning_rate": 0.003,
      "loss": 4.1679,
      "step": 4735
    },
    {
      "epoch": 0.04736,
      "grad_norm": 0.7892905763980204,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4736
    },
    {
      "epoch": 0.04737,
      "grad_norm": 0.8500450443730917,
      "learning_rate": 0.003,
      "loss": 4.1125,
      "step": 4737
    },
    {
      "epoch": 0.04738,
      "grad_norm": 0.9289248020340674,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4738
    },
    {
      "epoch": 0.04739,
      "grad_norm": 0.955981339694237,
      "learning_rate": 0.003,
      "loss": 4.0949,
      "step": 4739
    },
    {
      "epoch": 0.0474,
      "grad_norm": 0.9041807476389423,
      "learning_rate": 0.003,
      "loss": 4.1687,
      "step": 4740
    },
    {
      "epoch": 0.04741,
      "grad_norm": 0.8492001903774913,
      "learning_rate": 0.003,
      "loss": 4.1568,
      "step": 4741
    },
    {
      "epoch": 0.04742,
      "grad_norm": 0.7853348835474101,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4742
    },
    {
      "epoch": 0.04743,
      "grad_norm": 0.7795682667464989,
      "learning_rate": 0.003,
      "loss": 4.1611,
      "step": 4743
    },
    {
      "epoch": 0.04744,
      "grad_norm": 0.7267419747954841,
      "learning_rate": 0.003,
      "loss": 4.119,
      "step": 4744
    },
    {
      "epoch": 0.04745,
      "grad_norm": 0.7262044399015473,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4745
    },
    {
      "epoch": 0.04746,
      "grad_norm": 0.7299893869109523,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4746
    },
    {
      "epoch": 0.04747,
      "grad_norm": 0.806239055464935,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4747
    },
    {
      "epoch": 0.04748,
      "grad_norm": 0.6789998274727187,
      "learning_rate": 0.003,
      "loss": 4.1645,
      "step": 4748
    },
    {
      "epoch": 0.04749,
      "grad_norm": 0.54858705364636,
      "learning_rate": 0.003,
      "loss": 4.1651,
      "step": 4749
    },
    {
      "epoch": 0.0475,
      "grad_norm": 0.5060470689311919,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4750
    },
    {
      "epoch": 0.04751,
      "grad_norm": 0.5080665649296767,
      "learning_rate": 0.003,
      "loss": 4.0905,
      "step": 4751
    },
    {
      "epoch": 0.04752,
      "grad_norm": 0.5597535217525249,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4752
    },
    {
      "epoch": 0.04753,
      "grad_norm": 0.5815761039735459,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4753
    },
    {
      "epoch": 0.04754,
      "grad_norm": 0.6895819778653248,
      "learning_rate": 0.003,
      "loss": 4.1475,
      "step": 4754
    },
    {
      "epoch": 0.04755,
      "grad_norm": 0.9251374027849657,
      "learning_rate": 0.003,
      "loss": 4.1347,
      "step": 4755
    },
    {
      "epoch": 0.04756,
      "grad_norm": 1.1438681733923501,
      "learning_rate": 0.003,
      "loss": 4.1493,
      "step": 4756
    },
    {
      "epoch": 0.04757,
      "grad_norm": 0.7788109201469021,
      "learning_rate": 0.003,
      "loss": 4.1362,
      "step": 4757
    },
    {
      "epoch": 0.04758,
      "grad_norm": 0.6624956821204955,
      "learning_rate": 0.003,
      "loss": 4.1251,
      "step": 4758
    },
    {
      "epoch": 0.04759,
      "grad_norm": 0.7037195304943054,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4759
    },
    {
      "epoch": 0.0476,
      "grad_norm": 0.7411798494302422,
      "learning_rate": 0.003,
      "loss": 4.1193,
      "step": 4760
    },
    {
      "epoch": 0.04761,
      "grad_norm": 0.7054700847471357,
      "learning_rate": 0.003,
      "loss": 4.1273,
      "step": 4761
    },
    {
      "epoch": 0.04762,
      "grad_norm": 0.6226720435093628,
      "learning_rate": 0.003,
      "loss": 4.1182,
      "step": 4762
    },
    {
      "epoch": 0.04763,
      "grad_norm": 0.6886529154194588,
      "learning_rate": 0.003,
      "loss": 4.122,
      "step": 4763
    },
    {
      "epoch": 0.04764,
      "grad_norm": 0.6992577861087502,
      "learning_rate": 0.003,
      "loss": 4.1093,
      "step": 4764
    },
    {
      "epoch": 0.04765,
      "grad_norm": 0.6613811777977939,
      "learning_rate": 0.003,
      "loss": 4.1586,
      "step": 4765
    },
    {
      "epoch": 0.04766,
      "grad_norm": 0.5851808207831782,
      "learning_rate": 0.003,
      "loss": 4.1283,
      "step": 4766
    },
    {
      "epoch": 0.04767,
      "grad_norm": 0.6679898130461849,
      "learning_rate": 0.003,
      "loss": 4.1293,
      "step": 4767
    },
    {
      "epoch": 0.04768,
      "grad_norm": 0.7821493077256183,
      "learning_rate": 0.003,
      "loss": 4.1257,
      "step": 4768
    },
    {
      "epoch": 0.04769,
      "grad_norm": 0.8355601404081668,
      "learning_rate": 0.003,
      "loss": 4.0971,
      "step": 4769
    },
    {
      "epoch": 0.0477,
      "grad_norm": 0.8479665704653694,
      "learning_rate": 0.003,
      "loss": 4.1104,
      "step": 4770
    },
    {
      "epoch": 0.04771,
      "grad_norm": 0.8866254860935738,
      "learning_rate": 0.003,
      "loss": 4.1382,
      "step": 4771
    },
    {
      "epoch": 0.04772,
      "grad_norm": 0.915742333919775,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4772
    },
    {
      "epoch": 0.04773,
      "grad_norm": 0.754051301920501,
      "learning_rate": 0.003,
      "loss": 4.157,
      "step": 4773
    },
    {
      "epoch": 0.04774,
      "grad_norm": 0.7232939078296818,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4774
    },
    {
      "epoch": 0.04775,
      "grad_norm": 0.8169270534707214,
      "learning_rate": 0.003,
      "loss": 4.1399,
      "step": 4775
    },
    {
      "epoch": 0.04776,
      "grad_norm": 0.8573605280685423,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4776
    },
    {
      "epoch": 0.04777,
      "grad_norm": 0.9120618578003061,
      "learning_rate": 0.003,
      "loss": 4.1279,
      "step": 4777
    },
    {
      "epoch": 0.04778,
      "grad_norm": 0.9392445159690339,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4778
    },
    {
      "epoch": 0.04779,
      "grad_norm": 0.8671768686651342,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4779
    },
    {
      "epoch": 0.0478,
      "grad_norm": 0.8025049205614178,
      "learning_rate": 0.003,
      "loss": 4.1504,
      "step": 4780
    },
    {
      "epoch": 0.04781,
      "grad_norm": 0.7254367126713761,
      "learning_rate": 0.003,
      "loss": 4.125,
      "step": 4781
    },
    {
      "epoch": 0.04782,
      "grad_norm": 0.6859662827550447,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4782
    },
    {
      "epoch": 0.04783,
      "grad_norm": 0.6967231284903036,
      "learning_rate": 0.003,
      "loss": 4.1233,
      "step": 4783
    },
    {
      "epoch": 0.04784,
      "grad_norm": 0.7980703901010124,
      "learning_rate": 0.003,
      "loss": 4.1461,
      "step": 4784
    },
    {
      "epoch": 0.04785,
      "grad_norm": 0.8254970324649596,
      "learning_rate": 0.003,
      "loss": 4.1506,
      "step": 4785
    },
    {
      "epoch": 0.04786,
      "grad_norm": 0.8777922852113546,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4786
    },
    {
      "epoch": 0.04787,
      "grad_norm": 0.7491389893707523,
      "learning_rate": 0.003,
      "loss": 4.1458,
      "step": 4787
    },
    {
      "epoch": 0.04788,
      "grad_norm": 0.6503832601570363,
      "learning_rate": 0.003,
      "loss": 4.1261,
      "step": 4788
    },
    {
      "epoch": 0.04789,
      "grad_norm": 0.7031951667657491,
      "learning_rate": 0.003,
      "loss": 4.1151,
      "step": 4789
    },
    {
      "epoch": 0.0479,
      "grad_norm": 0.678065080300084,
      "learning_rate": 0.003,
      "loss": 4.131,
      "step": 4790
    },
    {
      "epoch": 0.04791,
      "grad_norm": 0.523098323506981,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4791
    },
    {
      "epoch": 0.04792,
      "grad_norm": 0.5260803142583377,
      "learning_rate": 0.003,
      "loss": 4.1508,
      "step": 4792
    },
    {
      "epoch": 0.04793,
      "grad_norm": 0.5215649768374346,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 4793
    },
    {
      "epoch": 0.04794,
      "grad_norm": 0.46968713617443264,
      "learning_rate": 0.003,
      "loss": 4.1301,
      "step": 4794
    },
    {
      "epoch": 0.04795,
      "grad_norm": 0.5247439754384676,
      "learning_rate": 0.003,
      "loss": 4.129,
      "step": 4795
    },
    {
      "epoch": 0.04796,
      "grad_norm": 0.6933674582411209,
      "learning_rate": 0.003,
      "loss": 4.0983,
      "step": 4796
    },
    {
      "epoch": 0.04797,
      "grad_norm": 0.9807791236957965,
      "learning_rate": 0.003,
      "loss": 4.1462,
      "step": 4797
    },
    {
      "epoch": 0.04798,
      "grad_norm": 1.1741198222289444,
      "learning_rate": 0.003,
      "loss": 4.1184,
      "step": 4798
    },
    {
      "epoch": 0.04799,
      "grad_norm": 0.6525704641811733,
      "learning_rate": 0.003,
      "loss": 4.116,
      "step": 4799
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.8089064094414999,
      "learning_rate": 0.003,
      "loss": 4.1267,
      "step": 4800
    },
    {
      "epoch": 0.04801,
      "grad_norm": 0.870503514300495,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4801
    },
    {
      "epoch": 0.04802,
      "grad_norm": 0.6635652157598942,
      "learning_rate": 0.003,
      "loss": 4.1031,
      "step": 4802
    },
    {
      "epoch": 0.04803,
      "grad_norm": 0.6200303139785031,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 4803
    },
    {
      "epoch": 0.04804,
      "grad_norm": 0.762754226201963,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4804
    },
    {
      "epoch": 0.04805,
      "grad_norm": 0.7861473887229704,
      "learning_rate": 0.003,
      "loss": 4.1476,
      "step": 4805
    },
    {
      "epoch": 0.04806,
      "grad_norm": 0.7696022521328222,
      "learning_rate": 0.003,
      "loss": 4.0929,
      "step": 4806
    },
    {
      "epoch": 0.04807,
      "grad_norm": 0.7570918480886352,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4807
    },
    {
      "epoch": 0.04808,
      "grad_norm": 0.8705313373191277,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 4808
    },
    {
      "epoch": 0.04809,
      "grad_norm": 0.8255463817975629,
      "learning_rate": 0.003,
      "loss": 4.1212,
      "step": 4809
    },
    {
      "epoch": 0.0481,
      "grad_norm": 0.8537469900625964,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4810
    },
    {
      "epoch": 0.04811,
      "grad_norm": 0.8483723617783365,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4811
    },
    {
      "epoch": 0.04812,
      "grad_norm": 0.7645031119405934,
      "learning_rate": 0.003,
      "loss": 4.1453,
      "step": 4812
    },
    {
      "epoch": 0.04813,
      "grad_norm": 0.8283116512854986,
      "learning_rate": 0.003,
      "loss": 4.1331,
      "step": 4813
    },
    {
      "epoch": 0.04814,
      "grad_norm": 0.8641578230891004,
      "learning_rate": 0.003,
      "loss": 4.0922,
      "step": 4814
    },
    {
      "epoch": 0.04815,
      "grad_norm": 0.8919893073764681,
      "learning_rate": 0.003,
      "loss": 4.1334,
      "step": 4815
    },
    {
      "epoch": 0.04816,
      "grad_norm": 0.8648426828228101,
      "learning_rate": 0.003,
      "loss": 4.1459,
      "step": 4816
    },
    {
      "epoch": 0.04817,
      "grad_norm": 0.9477238526245142,
      "learning_rate": 0.003,
      "loss": 4.1271,
      "step": 4817
    },
    {
      "epoch": 0.04818,
      "grad_norm": 0.9252788072968092,
      "learning_rate": 0.003,
      "loss": 4.1409,
      "step": 4818
    },
    {
      "epoch": 0.04819,
      "grad_norm": 0.8797819035450328,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4819
    },
    {
      "epoch": 0.0482,
      "grad_norm": 0.9035562726868283,
      "learning_rate": 0.003,
      "loss": 4.1328,
      "step": 4820
    },
    {
      "epoch": 0.04821,
      "grad_norm": 0.858921817091334,
      "learning_rate": 0.003,
      "loss": 4.1649,
      "step": 4821
    },
    {
      "epoch": 0.04822,
      "grad_norm": 0.6583705277011013,
      "learning_rate": 0.003,
      "loss": 4.1352,
      "step": 4822
    },
    {
      "epoch": 0.04823,
      "grad_norm": 0.5779379743608074,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4823
    },
    {
      "epoch": 0.04824,
      "grad_norm": 0.5505773694357159,
      "learning_rate": 0.003,
      "loss": 4.1145,
      "step": 4824
    },
    {
      "epoch": 0.04825,
      "grad_norm": 0.5013668074041548,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4825
    },
    {
      "epoch": 0.04826,
      "grad_norm": 0.5240704697199119,
      "learning_rate": 0.003,
      "loss": 4.1288,
      "step": 4826
    },
    {
      "epoch": 0.04827,
      "grad_norm": 0.533629158460227,
      "learning_rate": 0.003,
      "loss": 4.1313,
      "step": 4827
    },
    {
      "epoch": 0.04828,
      "grad_norm": 0.5992841702695002,
      "learning_rate": 0.003,
      "loss": 4.1416,
      "step": 4828
    },
    {
      "epoch": 0.04829,
      "grad_norm": 0.6683602415515053,
      "learning_rate": 0.003,
      "loss": 4.1199,
      "step": 4829
    },
    {
      "epoch": 0.0483,
      "grad_norm": 0.7131645514243471,
      "learning_rate": 0.003,
      "loss": 4.1329,
      "step": 4830
    },
    {
      "epoch": 0.04831,
      "grad_norm": 0.7918054904389603,
      "learning_rate": 0.003,
      "loss": 4.0993,
      "step": 4831
    },
    {
      "epoch": 0.04832,
      "grad_norm": 0.8874223110243422,
      "learning_rate": 0.003,
      "loss": 4.128,
      "step": 4832
    },
    {
      "epoch": 0.04833,
      "grad_norm": 0.9718339603744532,
      "learning_rate": 0.003,
      "loss": 4.1422,
      "step": 4833
    },
    {
      "epoch": 0.04834,
      "grad_norm": 0.9758671382016597,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 4834
    },
    {
      "epoch": 0.04835,
      "grad_norm": 0.9485480338404471,
      "learning_rate": 0.003,
      "loss": 4.1569,
      "step": 4835
    },
    {
      "epoch": 0.04836,
      "grad_norm": 1.0708614312540052,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4836
    },
    {
      "epoch": 0.04837,
      "grad_norm": 1.005625524837446,
      "learning_rate": 0.003,
      "loss": 4.1621,
      "step": 4837
    },
    {
      "epoch": 0.04838,
      "grad_norm": 1.142957704221482,
      "learning_rate": 0.003,
      "loss": 4.1842,
      "step": 4838
    },
    {
      "epoch": 0.04839,
      "grad_norm": 0.8876209726930161,
      "learning_rate": 0.003,
      "loss": 4.1597,
      "step": 4839
    },
    {
      "epoch": 0.0484,
      "grad_norm": 1.0000157052113117,
      "learning_rate": 0.003,
      "loss": 4.1726,
      "step": 4840
    },
    {
      "epoch": 0.04841,
      "grad_norm": 0.8947816300534793,
      "learning_rate": 0.003,
      "loss": 4.1664,
      "step": 4841
    },
    {
      "epoch": 0.04842,
      "grad_norm": 1.0306766066880264,
      "learning_rate": 0.003,
      "loss": 4.1906,
      "step": 4842
    },
    {
      "epoch": 0.04843,
      "grad_norm": 1.1670348326264735,
      "learning_rate": 0.003,
      "loss": 4.1576,
      "step": 4843
    },
    {
      "epoch": 0.04844,
      "grad_norm": 0.8379009760125189,
      "learning_rate": 0.003,
      "loss": 4.1746,
      "step": 4844
    },
    {
      "epoch": 0.04845,
      "grad_norm": 0.9239064010183352,
      "learning_rate": 0.003,
      "loss": 4.1606,
      "step": 4845
    },
    {
      "epoch": 0.04846,
      "grad_norm": 0.9084764134640881,
      "learning_rate": 0.003,
      "loss": 4.1905,
      "step": 4846
    },
    {
      "epoch": 0.04847,
      "grad_norm": 0.7704730780996573,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4847
    },
    {
      "epoch": 0.04848,
      "grad_norm": 0.8040182191975439,
      "learning_rate": 0.003,
      "loss": 4.186,
      "step": 4848
    },
    {
      "epoch": 0.04849,
      "grad_norm": 0.9074519148966281,
      "learning_rate": 0.003,
      "loss": 4.1786,
      "step": 4849
    },
    {
      "epoch": 0.0485,
      "grad_norm": 1.1771056378102565,
      "learning_rate": 0.003,
      "loss": 4.1773,
      "step": 4850
    },
    {
      "epoch": 0.04851,
      "grad_norm": 0.9654993158924701,
      "learning_rate": 0.003,
      "loss": 4.1898,
      "step": 4851
    },
    {
      "epoch": 0.04852,
      "grad_norm": 0.7654220286623955,
      "learning_rate": 0.003,
      "loss": 4.1162,
      "step": 4852
    },
    {
      "epoch": 0.04853,
      "grad_norm": 0.7291058591496458,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4853
    },
    {
      "epoch": 0.04854,
      "grad_norm": 0.6997930550604129,
      "learning_rate": 0.003,
      "loss": 4.1623,
      "step": 4854
    },
    {
      "epoch": 0.04855,
      "grad_norm": 0.8234663642795009,
      "learning_rate": 0.003,
      "loss": 4.1732,
      "step": 4855
    },
    {
      "epoch": 0.04856,
      "grad_norm": 0.7798398187689213,
      "learning_rate": 0.003,
      "loss": 4.1673,
      "step": 4856
    },
    {
      "epoch": 0.04857,
      "grad_norm": 0.6411382707024734,
      "learning_rate": 0.003,
      "loss": 4.1755,
      "step": 4857
    },
    {
      "epoch": 0.04858,
      "grad_norm": 0.6914930361794667,
      "learning_rate": 0.003,
      "loss": 4.1425,
      "step": 4858
    },
    {
      "epoch": 0.04859,
      "grad_norm": 0.742724632667153,
      "learning_rate": 0.003,
      "loss": 4.1759,
      "step": 4859
    },
    {
      "epoch": 0.0486,
      "grad_norm": 0.6805430980474091,
      "learning_rate": 0.003,
      "loss": 4.1546,
      "step": 4860
    },
    {
      "epoch": 0.04861,
      "grad_norm": 0.538868342282314,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4861
    },
    {
      "epoch": 0.04862,
      "grad_norm": 0.4415725147391537,
      "learning_rate": 0.003,
      "loss": 4.1394,
      "step": 4862
    },
    {
      "epoch": 0.04863,
      "grad_norm": 0.46517280056665583,
      "learning_rate": 0.003,
      "loss": 4.1574,
      "step": 4863
    },
    {
      "epoch": 0.04864,
      "grad_norm": 0.44851841399681913,
      "learning_rate": 0.003,
      "loss": 4.13,
      "step": 4864
    },
    {
      "epoch": 0.04865,
      "grad_norm": 0.4755135869431016,
      "learning_rate": 0.003,
      "loss": 4.1478,
      "step": 4865
    },
    {
      "epoch": 0.04866,
      "grad_norm": 0.449801980415272,
      "learning_rate": 0.003,
      "loss": 4.1507,
      "step": 4866
    },
    {
      "epoch": 0.04867,
      "grad_norm": 0.41976113144986854,
      "learning_rate": 0.003,
      "loss": 4.1172,
      "step": 4867
    },
    {
      "epoch": 0.04868,
      "grad_norm": 0.41274576265380564,
      "learning_rate": 0.003,
      "loss": 4.1258,
      "step": 4868
    },
    {
      "epoch": 0.04869,
      "grad_norm": 0.4120235509377836,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4869
    },
    {
      "epoch": 0.0487,
      "grad_norm": 0.45999567656300233,
      "learning_rate": 0.003,
      "loss": 4.1234,
      "step": 4870
    },
    {
      "epoch": 0.04871,
      "grad_norm": 0.5762534303017752,
      "learning_rate": 0.003,
      "loss": 4.1427,
      "step": 4871
    },
    {
      "epoch": 0.04872,
      "grad_norm": 0.6778749879598042,
      "learning_rate": 0.003,
      "loss": 4.1099,
      "step": 4872
    },
    {
      "epoch": 0.04873,
      "grad_norm": 0.7636826849696239,
      "learning_rate": 0.003,
      "loss": 4.1379,
      "step": 4873
    },
    {
      "epoch": 0.04874,
      "grad_norm": 0.7169955577952278,
      "learning_rate": 0.003,
      "loss": 4.1269,
      "step": 4874
    },
    {
      "epoch": 0.04875,
      "grad_norm": 0.5747097191102749,
      "learning_rate": 0.003,
      "loss": 4.118,
      "step": 4875
    },
    {
      "epoch": 0.04876,
      "grad_norm": 0.5097483392386485,
      "learning_rate": 0.003,
      "loss": 4.1148,
      "step": 4876
    },
    {
      "epoch": 0.04877,
      "grad_norm": 0.5291097317099629,
      "learning_rate": 0.003,
      "loss": 4.1183,
      "step": 4877
    },
    {
      "epoch": 0.04878,
      "grad_norm": 0.5314236989459756,
      "learning_rate": 0.003,
      "loss": 4.1007,
      "step": 4878
    },
    {
      "epoch": 0.04879,
      "grad_norm": 0.5936650310790298,
      "learning_rate": 0.003,
      "loss": 4.0979,
      "step": 4879
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.7538105145753773,
      "learning_rate": 0.003,
      "loss": 4.1118,
      "step": 4880
    },
    {
      "epoch": 0.04881,
      "grad_norm": 0.9419961284154782,
      "learning_rate": 0.003,
      "loss": 4.1176,
      "step": 4881
    },
    {
      "epoch": 0.04882,
      "grad_norm": 1.125784565290687,
      "learning_rate": 0.003,
      "loss": 4.1332,
      "step": 4882
    },
    {
      "epoch": 0.04883,
      "grad_norm": 0.7937742064689157,
      "learning_rate": 0.003,
      "loss": 4.1195,
      "step": 4883
    },
    {
      "epoch": 0.04884,
      "grad_norm": 0.7276138590729601,
      "learning_rate": 0.003,
      "loss": 4.1218,
      "step": 4884
    },
    {
      "epoch": 0.04885,
      "grad_norm": 0.7635121030741637,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4885
    },
    {
      "epoch": 0.04886,
      "grad_norm": 0.8525933214692812,
      "learning_rate": 0.003,
      "loss": 4.1333,
      "step": 4886
    },
    {
      "epoch": 0.04887,
      "grad_norm": 0.8979433474373131,
      "learning_rate": 0.003,
      "loss": 4.1469,
      "step": 4887
    },
    {
      "epoch": 0.04888,
      "grad_norm": 0.9257981316717354,
      "learning_rate": 0.003,
      "loss": 4.1543,
      "step": 4888
    },
    {
      "epoch": 0.04889,
      "grad_norm": 0.8410005218082275,
      "learning_rate": 0.003,
      "loss": 4.1282,
      "step": 4889
    },
    {
      "epoch": 0.0489,
      "grad_norm": 0.8463344038585867,
      "learning_rate": 0.003,
      "loss": 4.14,
      "step": 4890
    },
    {
      "epoch": 0.04891,
      "grad_norm": 0.7647404294541085,
      "learning_rate": 0.003,
      "loss": 4.1465,
      "step": 4891
    },
    {
      "epoch": 0.04892,
      "grad_norm": 0.7679081300937539,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4892
    },
    {
      "epoch": 0.04893,
      "grad_norm": 0.8575058455749437,
      "learning_rate": 0.003,
      "loss": 4.1449,
      "step": 4893
    },
    {
      "epoch": 0.04894,
      "grad_norm": 0.8645404495029662,
      "learning_rate": 0.003,
      "loss": 4.1584,
      "step": 4894
    },
    {
      "epoch": 0.04895,
      "grad_norm": 0.8334125833446672,
      "learning_rate": 0.003,
      "loss": 4.1318,
      "step": 4895
    },
    {
      "epoch": 0.04896,
      "grad_norm": 0.7851358889893844,
      "learning_rate": 0.003,
      "loss": 4.1436,
      "step": 4896
    },
    {
      "epoch": 0.04897,
      "grad_norm": 0.7536500770250976,
      "learning_rate": 0.003,
      "loss": 4.1523,
      "step": 4897
    },
    {
      "epoch": 0.04898,
      "grad_norm": 0.7125975228161553,
      "learning_rate": 0.003,
      "loss": 4.1385,
      "step": 4898
    },
    {
      "epoch": 0.04899,
      "grad_norm": 0.6723493590903667,
      "learning_rate": 0.003,
      "loss": 4.1252,
      "step": 4899
    },
    {
      "epoch": 0.049,
      "grad_norm": 0.5994822751068893,
      "learning_rate": 0.003,
      "loss": 4.1514,
      "step": 4900
    },
    {
      "epoch": 0.04901,
      "grad_norm": 0.562371614276483,
      "learning_rate": 0.003,
      "loss": 4.1081,
      "step": 4901
    },
    {
      "epoch": 0.04902,
      "grad_norm": 0.6664145016393446,
      "learning_rate": 0.003,
      "loss": 4.1437,
      "step": 4902
    },
    {
      "epoch": 0.04903,
      "grad_norm": 0.6848367937927413,
      "learning_rate": 0.003,
      "loss": 4.147,
      "step": 4903
    },
    {
      "epoch": 0.04904,
      "grad_norm": 0.6584184589527915,
      "learning_rate": 0.003,
      "loss": 4.1132,
      "step": 4904
    },
    {
      "epoch": 0.04905,
      "grad_norm": 0.6238214939939397,
      "learning_rate": 0.003,
      "loss": 4.11,
      "step": 4905
    },
    {
      "epoch": 0.04906,
      "grad_norm": 0.6975587494661175,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4906
    },
    {
      "epoch": 0.04907,
      "grad_norm": 0.8941666785365164,
      "learning_rate": 0.003,
      "loss": 4.1103,
      "step": 4907
    },
    {
      "epoch": 0.04908,
      "grad_norm": 1.081340571949139,
      "learning_rate": 0.003,
      "loss": 4.1381,
      "step": 4908
    },
    {
      "epoch": 0.04909,
      "grad_norm": 0.8257080561116473,
      "learning_rate": 0.003,
      "loss": 4.1335,
      "step": 4909
    },
    {
      "epoch": 0.0491,
      "grad_norm": 0.6982931611612875,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4910
    },
    {
      "epoch": 0.04911,
      "grad_norm": 0.8003274963133196,
      "learning_rate": 0.003,
      "loss": 4.1241,
      "step": 4911
    },
    {
      "epoch": 0.04912,
      "grad_norm": 0.8407576542296316,
      "learning_rate": 0.003,
      "loss": 4.1001,
      "step": 4912
    },
    {
      "epoch": 0.04913,
      "grad_norm": 0.9633535310499528,
      "learning_rate": 0.003,
      "loss": 4.1309,
      "step": 4913
    },
    {
      "epoch": 0.04914,
      "grad_norm": 0.9685795549695555,
      "learning_rate": 0.003,
      "loss": 4.1408,
      "step": 4914
    },
    {
      "epoch": 0.04915,
      "grad_norm": 0.8915068602810806,
      "learning_rate": 0.003,
      "loss": 4.1468,
      "step": 4915
    },
    {
      "epoch": 0.04916,
      "grad_norm": 0.7915988128949956,
      "learning_rate": 0.003,
      "loss": 4.1411,
      "step": 4916
    },
    {
      "epoch": 0.04917,
      "grad_norm": 0.7634607572088334,
      "learning_rate": 0.003,
      "loss": 4.0962,
      "step": 4917
    },
    {
      "epoch": 0.04918,
      "grad_norm": 0.7786335105527836,
      "learning_rate": 0.003,
      "loss": 4.1369,
      "step": 4918
    },
    {
      "epoch": 0.04919,
      "grad_norm": 0.8773870158083756,
      "learning_rate": 0.003,
      "loss": 4.1433,
      "step": 4919
    },
    {
      "epoch": 0.0492,
      "grad_norm": 0.7845152066567127,
      "learning_rate": 0.003,
      "loss": 4.1391,
      "step": 4920
    },
    {
      "epoch": 0.04921,
      "grad_norm": 0.8250729750583956,
      "learning_rate": 0.003,
      "loss": 4.1601,
      "step": 4921
    },
    {
      "epoch": 0.04922,
      "grad_norm": 0.8194188230604832,
      "learning_rate": 0.003,
      "loss": 4.1345,
      "step": 4922
    },
    {
      "epoch": 0.04923,
      "grad_norm": 0.8678442267593626,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4923
    },
    {
      "epoch": 0.04924,
      "grad_norm": 0.9322111328758901,
      "learning_rate": 0.003,
      "loss": 4.1758,
      "step": 4924
    },
    {
      "epoch": 0.04925,
      "grad_norm": 0.883656356593312,
      "learning_rate": 0.003,
      "loss": 4.1232,
      "step": 4925
    },
    {
      "epoch": 0.04926,
      "grad_norm": 0.7834220179620971,
      "learning_rate": 0.003,
      "loss": 4.1227,
      "step": 4926
    },
    {
      "epoch": 0.04927,
      "grad_norm": 0.648822263811632,
      "learning_rate": 0.003,
      "loss": 4.1563,
      "step": 4927
    },
    {
      "epoch": 0.04928,
      "grad_norm": 0.6606153717082371,
      "learning_rate": 0.003,
      "loss": 4.1175,
      "step": 4928
    },
    {
      "epoch": 0.04929,
      "grad_norm": 0.6448549657492914,
      "learning_rate": 0.003,
      "loss": 4.132,
      "step": 4929
    },
    {
      "epoch": 0.0493,
      "grad_norm": 0.6537446324241543,
      "learning_rate": 0.003,
      "loss": 4.1238,
      "step": 4930
    },
    {
      "epoch": 0.04931,
      "grad_norm": 0.6782203397244866,
      "learning_rate": 0.003,
      "loss": 4.1068,
      "step": 4931
    },
    {
      "epoch": 0.04932,
      "grad_norm": 0.7378598466407571,
      "learning_rate": 0.003,
      "loss": 4.1588,
      "step": 4932
    },
    {
      "epoch": 0.04933,
      "grad_norm": 0.7395809589970013,
      "learning_rate": 0.003,
      "loss": 4.1341,
      "step": 4933
    },
    {
      "epoch": 0.04934,
      "grad_norm": 0.7658041842655043,
      "learning_rate": 0.003,
      "loss": 4.1643,
      "step": 4934
    },
    {
      "epoch": 0.04935,
      "grad_norm": 0.9019517516492989,
      "learning_rate": 0.003,
      "loss": 4.1648,
      "step": 4935
    },
    {
      "epoch": 0.04936,
      "grad_norm": 1.1441200899996462,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4936
    },
    {
      "epoch": 0.04937,
      "grad_norm": 0.8788107449320854,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4937
    },
    {
      "epoch": 0.04938,
      "grad_norm": 0.8007894624899063,
      "learning_rate": 0.003,
      "loss": 4.1439,
      "step": 4938
    },
    {
      "epoch": 0.04939,
      "grad_norm": 0.789503524635859,
      "learning_rate": 0.003,
      "loss": 4.1548,
      "step": 4939
    },
    {
      "epoch": 0.0494,
      "grad_norm": 0.787654716377685,
      "learning_rate": 0.003,
      "loss": 4.1587,
      "step": 4940
    },
    {
      "epoch": 0.04941,
      "grad_norm": 0.6688134872012756,
      "learning_rate": 0.003,
      "loss": 4.1191,
      "step": 4941
    },
    {
      "epoch": 0.04942,
      "grad_norm": 0.5978385156801125,
      "learning_rate": 0.003,
      "loss": 4.1326,
      "step": 4942
    },
    {
      "epoch": 0.04943,
      "grad_norm": 0.6124698730462602,
      "learning_rate": 0.003,
      "loss": 4.1533,
      "step": 4943
    },
    {
      "epoch": 0.04944,
      "grad_norm": 0.5591922471734772,
      "learning_rate": 0.003,
      "loss": 4.154,
      "step": 4944
    },
    {
      "epoch": 0.04945,
      "grad_norm": 0.4939591571209979,
      "learning_rate": 0.003,
      "loss": 4.1406,
      "step": 4945
    },
    {
      "epoch": 0.04946,
      "grad_norm": 0.49333816406820896,
      "learning_rate": 0.003,
      "loss": 4.1657,
      "step": 4946
    },
    {
      "epoch": 0.04947,
      "grad_norm": 0.5127970104844136,
      "learning_rate": 0.003,
      "loss": 4.1055,
      "step": 4947
    },
    {
      "epoch": 0.04948,
      "grad_norm": 0.5804249788539954,
      "learning_rate": 0.003,
      "loss": 4.1243,
      "step": 4948
    },
    {
      "epoch": 0.04949,
      "grad_norm": 0.6106490006275131,
      "learning_rate": 0.003,
      "loss": 4.1284,
      "step": 4949
    },
    {
      "epoch": 0.0495,
      "grad_norm": 0.7326284913442733,
      "learning_rate": 0.003,
      "loss": 4.094,
      "step": 4950
    },
    {
      "epoch": 0.04951,
      "grad_norm": 0.8806738854393903,
      "learning_rate": 0.003,
      "loss": 4.1082,
      "step": 4951
    },
    {
      "epoch": 0.04952,
      "grad_norm": 0.8907989929621909,
      "learning_rate": 0.003,
      "loss": 4.1473,
      "step": 4952
    },
    {
      "epoch": 0.04953,
      "grad_norm": 0.6504370122838272,
      "learning_rate": 0.003,
      "loss": 4.1249,
      "step": 4953
    },
    {
      "epoch": 0.04954,
      "grad_norm": 0.5399865076183303,
      "learning_rate": 0.003,
      "loss": 4.1445,
      "step": 4954
    },
    {
      "epoch": 0.04955,
      "grad_norm": 0.6426116557249149,
      "learning_rate": 0.003,
      "loss": 4.1197,
      "step": 4955
    },
    {
      "epoch": 0.04956,
      "grad_norm": 0.6565119107623256,
      "learning_rate": 0.003,
      "loss": 4.1354,
      "step": 4956
    },
    {
      "epoch": 0.04957,
      "grad_norm": 0.6722789026797903,
      "learning_rate": 0.003,
      "loss": 4.106,
      "step": 4957
    },
    {
      "epoch": 0.04958,
      "grad_norm": 0.5890068452063758,
      "learning_rate": 0.003,
      "loss": 4.1201,
      "step": 4958
    },
    {
      "epoch": 0.04959,
      "grad_norm": 0.5878861479683556,
      "learning_rate": 0.003,
      "loss": 4.1239,
      "step": 4959
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.6509946110169393,
      "learning_rate": 0.003,
      "loss": 4.0952,
      "step": 4960
    },
    {
      "epoch": 0.04961,
      "grad_norm": 0.6156850509647105,
      "learning_rate": 0.003,
      "loss": 4.0992,
      "step": 4961
    },
    {
      "epoch": 0.04962,
      "grad_norm": 0.6497590683615507,
      "learning_rate": 0.003,
      "loss": 4.1161,
      "step": 4962
    },
    {
      "epoch": 0.04963,
      "grad_norm": 0.6412215380917496,
      "learning_rate": 0.003,
      "loss": 4.1519,
      "step": 4963
    },
    {
      "epoch": 0.04964,
      "grad_norm": 0.6061448543738891,
      "learning_rate": 0.003,
      "loss": 4.1224,
      "step": 4964
    },
    {
      "epoch": 0.04965,
      "grad_norm": 0.6123953168115248,
      "learning_rate": 0.003,
      "loss": 4.1231,
      "step": 4965
    },
    {
      "epoch": 0.04966,
      "grad_norm": 0.8037505368816966,
      "learning_rate": 0.003,
      "loss": 4.1008,
      "step": 4966
    },
    {
      "epoch": 0.04967,
      "grad_norm": 0.9549949754561539,
      "learning_rate": 0.003,
      "loss": 4.1401,
      "step": 4967
    },
    {
      "epoch": 0.04968,
      "grad_norm": 1.1281595822055885,
      "learning_rate": 0.003,
      "loss": 4.1707,
      "step": 4968
    },
    {
      "epoch": 0.04969,
      "grad_norm": 0.8652927562240527,
      "learning_rate": 0.003,
      "loss": 4.1378,
      "step": 4969
    },
    {
      "epoch": 0.0497,
      "grad_norm": 0.7881031556981761,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4970
    },
    {
      "epoch": 0.04971,
      "grad_norm": 0.8417863445996929,
      "learning_rate": 0.003,
      "loss": 4.153,
      "step": 4971
    },
    {
      "epoch": 0.04972,
      "grad_norm": 0.8493800636053089,
      "learning_rate": 0.003,
      "loss": 4.1289,
      "step": 4972
    },
    {
      "epoch": 0.04973,
      "grad_norm": 0.9639644128846697,
      "learning_rate": 0.003,
      "loss": 4.1826,
      "step": 4973
    },
    {
      "epoch": 0.04974,
      "grad_norm": 1.0979045911236074,
      "learning_rate": 0.003,
      "loss": 4.1492,
      "step": 4974
    },
    {
      "epoch": 0.04975,
      "grad_norm": 1.1499829958321686,
      "learning_rate": 0.003,
      "loss": 4.1556,
      "step": 4975
    },
    {
      "epoch": 0.04976,
      "grad_norm": 0.896257641249768,
      "learning_rate": 0.003,
      "loss": 4.1482,
      "step": 4976
    },
    {
      "epoch": 0.04977,
      "grad_norm": 0.8176607909447694,
      "learning_rate": 0.003,
      "loss": 4.1551,
      "step": 4977
    },
    {
      "epoch": 0.04978,
      "grad_norm": 0.8167430782012225,
      "learning_rate": 0.003,
      "loss": 4.1541,
      "step": 4978
    },
    {
      "epoch": 0.04979,
      "grad_norm": 0.8298370431983914,
      "learning_rate": 0.003,
      "loss": 4.1591,
      "step": 4979
    },
    {
      "epoch": 0.0498,
      "grad_norm": 0.8918442192162982,
      "learning_rate": 0.003,
      "loss": 4.1503,
      "step": 4980
    },
    {
      "epoch": 0.04981,
      "grad_norm": 0.8954883580668984,
      "learning_rate": 0.003,
      "loss": 4.1264,
      "step": 4981
    },
    {
      "epoch": 0.04982,
      "grad_norm": 0.8872182858380784,
      "learning_rate": 0.003,
      "loss": 4.1516,
      "step": 4982
    },
    {
      "epoch": 0.04983,
      "grad_norm": 0.7800382368496617,
      "learning_rate": 0.003,
      "loss": 4.1064,
      "step": 4983
    },
    {
      "epoch": 0.04984,
      "grad_norm": 0.7656679202701089,
      "learning_rate": 0.003,
      "loss": 4.1096,
      "step": 4984
    },
    {
      "epoch": 0.04985,
      "grad_norm": 0.8692356760179402,
      "learning_rate": 0.003,
      "loss": 4.1443,
      "step": 4985
    },
    {
      "epoch": 0.04986,
      "grad_norm": 0.8658640091365618,
      "learning_rate": 0.003,
      "loss": 4.1629,
      "step": 4986
    },
    {
      "epoch": 0.04987,
      "grad_norm": 0.9418717422739423,
      "learning_rate": 0.003,
      "loss": 4.1307,
      "step": 4987
    },
    {
      "epoch": 0.04988,
      "grad_norm": 0.9193482062490026,
      "learning_rate": 0.003,
      "loss": 4.1628,
      "step": 4988
    },
    {
      "epoch": 0.04989,
      "grad_norm": 0.8335538614629532,
      "learning_rate": 0.003,
      "loss": 4.1633,
      "step": 4989
    },
    {
      "epoch": 0.0499,
      "grad_norm": 0.7458856658289911,
      "learning_rate": 0.003,
      "loss": 4.136,
      "step": 4990
    },
    {
      "epoch": 0.04991,
      "grad_norm": 0.6549431595291397,
      "learning_rate": 0.003,
      "loss": 4.1229,
      "step": 4991
    },
    {
      "epoch": 0.04992,
      "grad_norm": 0.6708966834866974,
      "learning_rate": 0.003,
      "loss": 4.1073,
      "step": 4992
    },
    {
      "epoch": 0.04993,
      "grad_norm": 0.6367637070694386,
      "learning_rate": 0.003,
      "loss": 4.1312,
      "step": 4993
    },
    {
      "epoch": 0.04994,
      "grad_norm": 0.6415299490019786,
      "learning_rate": 0.003,
      "loss": 4.1178,
      "step": 4994
    },
    {
      "epoch": 0.04995,
      "grad_norm": 0.7355718950683062,
      "learning_rate": 0.003,
      "loss": 4.1374,
      "step": 4995
    },
    {
      "epoch": 0.04996,
      "grad_norm": 0.8795955713743572,
      "learning_rate": 0.003,
      "loss": 4.1242,
      "step": 4996
    },
    {
      "epoch": 0.04997,
      "grad_norm": 1.1069671993764165,
      "learning_rate": 0.003,
      "loss": 4.1471,
      "step": 4997
    },
    {
      "epoch": 0.04998,
      "grad_norm": 0.9565748099055139,
      "learning_rate": 0.003,
      "loss": 4.1695,
      "step": 4998
    },
    {
      "epoch": 0.04999,
      "grad_norm": 0.8361592261029797,
      "learning_rate": 0.003,
      "loss": 4.1263,
      "step": 4999
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9200109034528712,
      "learning_rate": 0.003,
      "loss": 4.1735,
      "step": 5000
    }
  ],
  "logging_steps": 1,
  "max_steps": 100000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.9821821427712e+17,
  "train_batch_size": 1024,
  "trial_name": null,
  "trial_params": null
}