{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 8017,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00012473493825620557,
      "grad_norm": 0.0,
      "learning_rate": 8.298755186721993e-08,
      "loss": 6.5433,
      "step": 1
    },
    {
      "epoch": 0.00024946987651241115,
      "grad_norm": 0.0,
      "learning_rate": 1.6597510373443985e-07,
      "loss": 6.3101,
      "step": 2
    },
    {
      "epoch": 0.00037420481476861667,
      "grad_norm": 0.0,
      "learning_rate": 2.4896265560165975e-07,
      "loss": 6.4525,
      "step": 3
    },
    {
      "epoch": 0.0004989397530248223,
      "grad_norm": 0.0,
      "learning_rate": 3.319502074688797e-07,
      "loss": 6.3739,
      "step": 4
    },
    {
      "epoch": 0.0006236746912810278,
      "grad_norm": 0.0,
      "learning_rate": 4.149377593360996e-07,
      "loss": 6.3881,
      "step": 5
    },
    {
      "epoch": 0.0007484096295372333,
      "grad_norm": 0.0,
      "learning_rate": 4.979253112033195e-07,
      "loss": 6.5473,
      "step": 6
    },
    {
      "epoch": 0.000873144567793439,
      "grad_norm": 0.0,
      "learning_rate": 5.809128630705395e-07,
      "loss": 6.6018,
      "step": 7
    },
    {
      "epoch": 0.0009978795060496446,
      "grad_norm": 0.0,
      "learning_rate": 6.639004149377594e-07,
      "loss": 6.4326,
      "step": 8
    },
    {
      "epoch": 0.00112261444430585,
      "grad_norm": 0.0,
      "learning_rate": 7.468879668049793e-07,
      "loss": 6.1861,
      "step": 9
    },
    {
      "epoch": 0.0012473493825620556,
      "grad_norm": 0.0,
      "learning_rate": 8.298755186721992e-07,
      "loss": 6.1714,
      "step": 10
    },
    {
      "epoch": 0.0013720843208182613,
      "grad_norm": 0.0,
      "learning_rate": 9.128630705394191e-07,
      "loss": 6.1402,
      "step": 11
    },
    {
      "epoch": 0.0014968192590744667,
      "grad_norm": 0.0,
      "learning_rate": 9.95850622406639e-07,
      "loss": 5.568,
      "step": 12
    },
    {
      "epoch": 0.0016215541973306723,
      "grad_norm": 0.0,
      "learning_rate": 1.078838174273859e-06,
      "loss": 5.355,
      "step": 13
    },
    {
      "epoch": 0.001746289135586878,
      "grad_norm": 0.0,
      "learning_rate": 1.161825726141079e-06,
      "loss": 5.2926,
      "step": 14
    },
    {
      "epoch": 0.0018710240738430836,
      "grad_norm": 0.0,
      "learning_rate": 1.2448132780082988e-06,
      "loss": 5.1146,
      "step": 15
    },
    {
      "epoch": 0.001995759012099289,
      "grad_norm": 0.0,
      "learning_rate": 1.3278008298755188e-06,
      "loss": 3.7858,
      "step": 16
    },
    {
      "epoch": 0.0021204939503554946,
      "grad_norm": 0.0,
      "learning_rate": 1.4107883817427386e-06,
      "loss": 3.9101,
      "step": 17
    },
    {
      "epoch": 0.0022452288886117,
      "grad_norm": 0.0,
      "learning_rate": 1.4937759336099586e-06,
      "loss": 3.8263,
      "step": 18
    },
    {
      "epoch": 0.002369963826867906,
      "grad_norm": 0.0,
      "learning_rate": 1.5767634854771784e-06,
      "loss": 3.708,
      "step": 19
    },
    {
      "epoch": 0.0024946987651241113,
      "grad_norm": 0.0,
      "learning_rate": 1.6597510373443984e-06,
      "loss": 3.7611,
      "step": 20
    },
    {
      "epoch": 0.0026194337033803167,
      "grad_norm": 0.0,
      "learning_rate": 1.7427385892116182e-06,
      "loss": 3.6356,
      "step": 21
    },
    {
      "epoch": 0.0027441686416365225,
      "grad_norm": 0.0,
      "learning_rate": 1.8257261410788382e-06,
      "loss": 3.4969,
      "step": 22
    },
    {
      "epoch": 0.002868903579892728,
      "grad_norm": 0.0,
      "learning_rate": 1.908713692946058e-06,
      "loss": 3.4411,
      "step": 23
    },
    {
      "epoch": 0.0029936385181489333,
      "grad_norm": 0.0,
      "learning_rate": 1.991701244813278e-06,
      "loss": 3.2423,
      "step": 24
    },
    {
      "epoch": 0.003118373456405139,
      "grad_norm": 0.0,
      "learning_rate": 2.074688796680498e-06,
      "loss": 3.2376,
      "step": 25
    },
    {
      "epoch": 0.0032431083946613446,
      "grad_norm": 0.0,
      "learning_rate": 2.157676348547718e-06,
      "loss": 3.2192,
      "step": 26
    },
    {
      "epoch": 0.00336784333291755,
      "grad_norm": 0.0,
      "learning_rate": 2.240663900414938e-06,
      "loss": 3.1639,
      "step": 27
    },
    {
      "epoch": 0.003492578271173756,
      "grad_norm": 0.0,
      "learning_rate": 2.323651452282158e-06,
      "loss": 3.0787,
      "step": 28
    },
    {
      "epoch": 0.0036173132094299613,
      "grad_norm": 0.0,
      "learning_rate": 2.4066390041493776e-06,
      "loss": 3.0248,
      "step": 29
    },
    {
      "epoch": 0.003742048147686167,
      "grad_norm": 0.0,
      "learning_rate": 2.4896265560165977e-06,
      "loss": 2.8591,
      "step": 30
    },
    {
      "epoch": 0.0038667830859423725,
      "grad_norm": 0.0,
      "learning_rate": 2.5726141078838172e-06,
      "loss": 2.7848,
      "step": 31
    },
    {
      "epoch": 0.003991518024198578,
      "grad_norm": 0.0,
      "learning_rate": 2.6556016597510377e-06,
      "loss": 2.8507,
      "step": 32
    },
    {
      "epoch": 0.004116252962454783,
      "grad_norm": 0.0,
      "learning_rate": 2.7385892116182577e-06,
      "loss": 2.6824,
      "step": 33
    },
    {
      "epoch": 0.004240987900710989,
      "grad_norm": 0.0,
      "learning_rate": 2.8215767634854773e-06,
      "loss": 2.5775,
      "step": 34
    },
    {
      "epoch": 0.004365722838967195,
      "grad_norm": 0.0,
      "learning_rate": 2.9045643153526977e-06,
      "loss": 2.5103,
      "step": 35
    },
    {
      "epoch": 0.0044904577772234,
      "grad_norm": 0.0,
      "learning_rate": 2.9875518672199173e-06,
      "loss": 2.4596,
      "step": 36
    },
    {
      "epoch": 0.004615192715479606,
      "grad_norm": 0.0,
      "learning_rate": 3.0705394190871373e-06,
      "loss": 2.415,
      "step": 37
    },
    {
      "epoch": 0.004739927653735812,
      "grad_norm": 0.0,
      "learning_rate": 3.153526970954357e-06,
      "loss": 2.4246,
      "step": 38
    },
    {
      "epoch": 0.004864662591992017,
      "grad_norm": 0.0,
      "learning_rate": 3.2365145228215773e-06,
      "loss": 2.4014,
      "step": 39
    },
    {
      "epoch": 0.0049893975302482225,
      "grad_norm": 0.0,
      "learning_rate": 3.319502074688797e-06,
      "loss": 2.3319,
      "step": 40
    },
    {
      "epoch": 0.005114132468504428,
      "grad_norm": 0.0,
      "learning_rate": 3.402489626556017e-06,
      "loss": 2.2356,
      "step": 41
    },
    {
      "epoch": 0.005238867406760633,
      "grad_norm": 0.0,
      "learning_rate": 3.4854771784232365e-06,
      "loss": 2.2585,
      "step": 42
    },
    {
      "epoch": 0.005363602345016839,
      "grad_norm": 0.0,
      "learning_rate": 3.568464730290457e-06,
      "loss": 2.1735,
      "step": 43
    },
    {
      "epoch": 0.005488337283273045,
      "grad_norm": 0.0,
      "learning_rate": 3.6514522821576765e-06,
      "loss": 2.1454,
      "step": 44
    },
    {
      "epoch": 0.00561307222152925,
      "grad_norm": 0.0,
      "learning_rate": 3.7344398340248965e-06,
      "loss": 2.0499,
      "step": 45
    },
    {
      "epoch": 0.005737807159785456,
      "grad_norm": 0.0,
      "learning_rate": 3.817427385892116e-06,
      "loss": 2.0182,
      "step": 46
    },
    {
      "epoch": 0.005862542098041662,
      "grad_norm": 0.0,
      "learning_rate": 3.9004149377593365e-06,
      "loss": 2.0878,
      "step": 47
    },
    {
      "epoch": 0.005987277036297867,
      "grad_norm": 0.0,
      "learning_rate": 3.983402489626556e-06,
      "loss": 2.0311,
      "step": 48
    },
    {
      "epoch": 0.0061120119745540725,
      "grad_norm": 0.0,
      "learning_rate": 4.0663900414937765e-06,
      "loss": 1.9852,
      "step": 49
    },
    {
      "epoch": 0.006236746912810278,
      "grad_norm": 0.0,
      "learning_rate": 4.149377593360996e-06,
      "loss": 2.0084,
      "step": 50
    },
    {
      "epoch": 0.006361481851066483,
      "grad_norm": 0.0,
      "learning_rate": 4.232365145228216e-06,
      "loss": 1.9556,
      "step": 51
    },
    {
      "epoch": 0.006486216789322689,
      "grad_norm": 0.0,
      "learning_rate": 4.315352697095436e-06,
      "loss": 1.9822,
      "step": 52
    },
    {
      "epoch": 0.006610951727578895,
      "grad_norm": 0.0,
      "learning_rate": 4.398340248962656e-06,
      "loss": 2.0077,
      "step": 53
    },
    {
      "epoch": 0.0067356866658351,
      "grad_norm": 0.0,
      "learning_rate": 4.481327800829876e-06,
      "loss": 1.9894,
      "step": 54
    },
    {
      "epoch": 0.006860421604091306,
      "grad_norm": 0.0,
      "learning_rate": 4.564315352697096e-06,
      "loss": 1.9154,
      "step": 55
    },
    {
      "epoch": 0.006985156542347512,
      "grad_norm": 0.0,
      "learning_rate": 4.647302904564316e-06,
      "loss": 1.9527,
      "step": 56
    },
    {
      "epoch": 0.007109891480603717,
      "grad_norm": 0.0,
      "learning_rate": 4.730290456431536e-06,
      "loss": 1.8799,
      "step": 57
    },
    {
      "epoch": 0.0072346264188599225,
      "grad_norm": 0.0,
      "learning_rate": 4.813278008298755e-06,
      "loss": 1.8721,
      "step": 58
    },
    {
      "epoch": 0.007359361357116128,
      "grad_norm": 0.0,
      "learning_rate": 4.896265560165976e-06,
      "loss": 1.8714,
      "step": 59
    },
    {
      "epoch": 0.007484096295372334,
      "grad_norm": 0.0,
      "learning_rate": 4.979253112033195e-06,
      "loss": 1.8995,
      "step": 60
    },
    {
      "epoch": 0.007608831233628539,
      "grad_norm": 0.0,
      "learning_rate": 5.062240663900415e-06,
      "loss": 1.8452,
      "step": 61
    },
    {
      "epoch": 0.007733566171884745,
      "grad_norm": 0.0,
      "learning_rate": 5.1452282157676345e-06,
      "loss": 1.8734,
      "step": 62
    },
    {
      "epoch": 0.00785830111014095,
      "grad_norm": 0.0,
      "learning_rate": 5.228215767634856e-06,
      "loss": 1.8644,
      "step": 63
    },
    {
      "epoch": 0.007983036048397157,
      "grad_norm": 0.0,
      "learning_rate": 5.311203319502075e-06,
      "loss": 1.7735,
      "step": 64
    },
    {
      "epoch": 0.008107770986653362,
      "grad_norm": 0.0,
      "learning_rate": 5.394190871369295e-06,
      "loss": 1.8558,
      "step": 65
    },
    {
      "epoch": 0.008232505924909567,
      "grad_norm": 0.0,
      "learning_rate": 5.477178423236515e-06,
      "loss": 1.869,
      "step": 66
    },
    {
      "epoch": 0.008357240863165773,
      "grad_norm": 0.0,
      "learning_rate": 5.560165975103735e-06,
      "loss": 1.8813,
      "step": 67
    },
    {
      "epoch": 0.008481975801421978,
      "grad_norm": 0.0,
      "learning_rate": 5.6431535269709545e-06,
      "loss": 1.8026,
      "step": 68
    },
    {
      "epoch": 0.008606710739678183,
      "grad_norm": 0.0,
      "learning_rate": 5.726141078838174e-06,
      "loss": 1.802,
      "step": 69
    },
    {
      "epoch": 0.00873144567793439,
      "grad_norm": 0.0,
      "learning_rate": 5.809128630705395e-06,
      "loss": 1.8185,
      "step": 70
    },
    {
      "epoch": 0.008856180616190595,
      "grad_norm": 0.0,
      "learning_rate": 5.892116182572615e-06,
      "loss": 1.7889,
      "step": 71
    },
    {
      "epoch": 0.0089809155544468,
      "grad_norm": 0.0,
      "learning_rate": 5.9751037344398345e-06,
      "loss": 1.8216,
      "step": 72
    },
    {
      "epoch": 0.009105650492703007,
      "grad_norm": 0.0,
      "learning_rate": 6.058091286307054e-06,
      "loss": 1.7133,
      "step": 73
    },
    {
      "epoch": 0.009230385430959212,
      "grad_norm": 0.0,
      "learning_rate": 6.1410788381742745e-06,
      "loss": 1.7458,
      "step": 74
    },
    {
      "epoch": 0.009355120369215417,
      "grad_norm": 0.0,
      "learning_rate": 6.224066390041494e-06,
      "loss": 1.7643,
      "step": 75
    },
    {
      "epoch": 0.009479855307471623,
      "grad_norm": 0.0,
      "learning_rate": 6.307053941908714e-06,
      "loss": 1.7807,
      "step": 76
    },
    {
      "epoch": 0.009604590245727828,
      "grad_norm": 0.0,
      "learning_rate": 6.390041493775933e-06,
      "loss": 1.7185,
      "step": 77
    },
    {
      "epoch": 0.009729325183984033,
      "grad_norm": 0.0,
      "learning_rate": 6.4730290456431546e-06,
      "loss": 1.7574,
      "step": 78
    },
    {
      "epoch": 0.00985406012224024,
      "grad_norm": 0.0,
      "learning_rate": 6.556016597510374e-06,
      "loss": 1.7463,
      "step": 79
    },
    {
      "epoch": 0.009978795060496445,
      "grad_norm": 0.0,
      "learning_rate": 6.639004149377594e-06,
      "loss": 1.8126,
      "step": 80
    },
    {
      "epoch": 0.01010352999875265,
      "grad_norm": 0.0,
      "learning_rate": 6.721991701244814e-06,
      "loss": 1.7082,
      "step": 81
    },
    {
      "epoch": 0.010228264937008857,
      "grad_norm": 0.0,
      "learning_rate": 6.804979253112034e-06,
      "loss": 1.7296,
      "step": 82
    },
    {
      "epoch": 0.010352999875265062,
      "grad_norm": 0.0,
      "learning_rate": 6.887966804979253e-06,
      "loss": 1.7787,
      "step": 83
    },
    {
      "epoch": 0.010477734813521267,
      "grad_norm": 0.0,
      "learning_rate": 6.970954356846473e-06,
      "loss": 1.6765,
      "step": 84
    },
    {
      "epoch": 0.010602469751777473,
      "grad_norm": 0.0,
      "learning_rate": 7.053941908713694e-06,
      "loss": 1.7586,
      "step": 85
    },
    {
      "epoch": 0.010727204690033678,
      "grad_norm": 0.0,
      "learning_rate": 7.136929460580914e-06,
      "loss": 1.6971,
      "step": 86
    },
    {
      "epoch": 0.010851939628289883,
      "grad_norm": 0.0,
      "learning_rate": 7.219917012448133e-06,
      "loss": 1.6469,
      "step": 87
    },
    {
      "epoch": 0.01097667456654609,
      "grad_norm": 0.0,
      "learning_rate": 7.302904564315353e-06,
      "loss": 1.7174,
      "step": 88
    },
    {
      "epoch": 0.011101409504802295,
      "grad_norm": 0.0,
      "learning_rate": 7.385892116182573e-06,
      "loss": 1.6723,
      "step": 89
    },
    {
      "epoch": 0.0112261444430585,
      "grad_norm": 0.0,
      "learning_rate": 7.468879668049793e-06,
      "loss": 1.6942,
      "step": 90
    },
    {
      "epoch": 0.011350879381314707,
      "grad_norm": 0.0,
      "learning_rate": 7.5518672199170125e-06,
      "loss": 1.6744,
      "step": 91
    },
    {
      "epoch": 0.011475614319570912,
      "grad_norm": 0.0,
      "learning_rate": 7.634854771784232e-06,
      "loss": 1.6999,
      "step": 92
    },
    {
      "epoch": 0.011600349257827117,
      "grad_norm": 0.0,
      "learning_rate": 7.717842323651453e-06,
      "loss": 1.6785,
      "step": 93
    },
    {
      "epoch": 0.011725084196083323,
      "grad_norm": 0.0,
      "learning_rate": 7.800829875518673e-06,
      "loss": 1.6731,
      "step": 94
    },
    {
      "epoch": 0.011849819134339528,
      "grad_norm": 0.0,
      "learning_rate": 7.883817427385892e-06,
      "loss": 1.6394,
      "step": 95
    },
    {
      "epoch": 0.011974554072595733,
      "grad_norm": 0.0,
      "learning_rate": 7.966804979253112e-06,
      "loss": 1.682,
      "step": 96
    },
    {
      "epoch": 0.01209928901085194,
      "grad_norm": 0.0,
      "learning_rate": 8.049792531120333e-06,
      "loss": 1.6529,
      "step": 97
    },
    {
      "epoch": 0.012224023949108145,
      "grad_norm": 0.0,
      "learning_rate": 8.132780082987553e-06,
      "loss": 1.705,
      "step": 98
    },
    {
      "epoch": 0.01234875888736435,
      "grad_norm": 0.0,
      "learning_rate": 8.215767634854772e-06,
      "loss": 1.6763,
      "step": 99
    },
    {
      "epoch": 0.012473493825620557,
      "grad_norm": 0.0,
      "learning_rate": 8.298755186721992e-06,
      "loss": 1.6498,
      "step": 100
    },
    {
      "epoch": 0.012598228763876762,
      "grad_norm": 0.0,
      "learning_rate": 8.381742738589213e-06,
      "loss": 1.6118,
      "step": 101
    },
    {
      "epoch": 0.012722963702132967,
      "grad_norm": 0.0,
      "learning_rate": 8.464730290456431e-06,
      "loss": 1.6597,
      "step": 102
    },
    {
      "epoch": 0.012847698640389173,
      "grad_norm": 0.0,
      "learning_rate": 8.547717842323652e-06,
      "loss": 1.6173,
      "step": 103
    },
    {
      "epoch": 0.012972433578645378,
      "grad_norm": 0.0,
      "learning_rate": 8.630705394190872e-06,
      "loss": 1.624,
      "step": 104
    },
    {
      "epoch": 0.013097168516901583,
      "grad_norm": 0.0,
      "learning_rate": 8.713692946058093e-06,
      "loss": 1.6705,
      "step": 105
    },
    {
      "epoch": 0.01322190345515779,
      "grad_norm": 0.0,
      "learning_rate": 8.796680497925311e-06,
      "loss": 1.6311,
      "step": 106
    },
    {
      "epoch": 0.013346638393413995,
      "grad_norm": 0.0,
      "learning_rate": 8.879668049792532e-06,
      "loss": 1.6563,
      "step": 107
    },
    {
      "epoch": 0.0134713733316702,
      "grad_norm": 0.0,
      "learning_rate": 8.962655601659752e-06,
      "loss": 1.5695,
      "step": 108
    },
    {
      "epoch": 0.013596108269926407,
      "grad_norm": 0.0,
      "learning_rate": 9.045643153526971e-06,
      "loss": 1.6551,
      "step": 109
    },
    {
      "epoch": 0.013720843208182612,
      "grad_norm": 0.0,
      "learning_rate": 9.128630705394191e-06,
      "loss": 1.5642,
      "step": 110
    },
    {
      "epoch": 0.013845578146438817,
      "grad_norm": 0.0,
      "learning_rate": 9.211618257261412e-06,
      "loss": 1.642,
      "step": 111
    },
    {
      "epoch": 0.013970313084695023,
      "grad_norm": 0.0,
      "learning_rate": 9.294605809128632e-06,
      "loss": 1.6324,
      "step": 112
    },
    {
      "epoch": 0.014095048022951228,
      "grad_norm": 0.0,
      "learning_rate": 9.377593360995851e-06,
      "loss": 1.6351,
      "step": 113
    },
    {
      "epoch": 0.014219782961207433,
      "grad_norm": 0.0,
      "learning_rate": 9.460580912863071e-06,
      "loss": 1.6178,
      "step": 114
    },
    {
      "epoch": 0.01434451789946364,
      "grad_norm": 0.0,
      "learning_rate": 9.543568464730292e-06,
      "loss": 1.5934,
      "step": 115
    },
    {
      "epoch": 0.014469252837719845,
      "grad_norm": 0.0,
      "learning_rate": 9.62655601659751e-06,
      "loss": 1.5934,
      "step": 116
    },
    {
      "epoch": 0.014593987775976052,
      "grad_norm": 0.0,
      "learning_rate": 9.709543568464731e-06,
      "loss": 1.5518,
      "step": 117
    },
    {
      "epoch": 0.014718722714232257,
      "grad_norm": 0.0,
      "learning_rate": 9.792531120331951e-06,
      "loss": 1.5626,
      "step": 118
    },
    {
      "epoch": 0.014843457652488462,
      "grad_norm": 0.0,
      "learning_rate": 9.875518672199172e-06,
      "loss": 1.6084,
      "step": 119
    },
    {
      "epoch": 0.014968192590744668,
      "grad_norm": 0.0,
      "learning_rate": 9.95850622406639e-06,
      "loss": 1.556,
      "step": 120
    },
    {
      "epoch": 0.015092927529000873,
      "grad_norm": 0.0,
      "learning_rate": 1.0041493775933611e-05,
      "loss": 1.5625,
      "step": 121
    },
    {
      "epoch": 0.015217662467257078,
      "grad_norm": 0.0,
      "learning_rate": 1.012448132780083e-05,
      "loss": 1.6091,
      "step": 122
    },
    {
      "epoch": 0.015342397405513285,
      "grad_norm": 0.0,
      "learning_rate": 1.020746887966805e-05,
      "loss": 1.6155,
      "step": 123
    },
    {
      "epoch": 0.01546713234376949,
      "grad_norm": 0.0,
      "learning_rate": 1.0290456431535269e-05,
      "loss": 1.5679,
      "step": 124
    },
    {
      "epoch": 0.015591867282025695,
      "grad_norm": 0.0,
      "learning_rate": 1.0373443983402491e-05,
      "loss": 1.6249,
      "step": 125
    },
    {
      "epoch": 0.0157166022202819,
      "grad_norm": 0.0,
      "learning_rate": 1.0456431535269711e-05,
      "loss": 1.6231,
      "step": 126
    },
    {
      "epoch": 0.015841337158538107,
      "grad_norm": 0.0,
      "learning_rate": 1.053941908713693e-05,
      "loss": 1.6146,
      "step": 127
    },
    {
      "epoch": 0.015966072096794313,
      "grad_norm": 0.0,
      "learning_rate": 1.062240663900415e-05,
      "loss": 1.6322,
      "step": 128
    },
    {
      "epoch": 0.016090807035050517,
      "grad_norm": 0.0,
      "learning_rate": 1.070539419087137e-05,
      "loss": 1.5817,
      "step": 129
    },
    {
      "epoch": 0.016215541973306723,
      "grad_norm": 0.0,
      "learning_rate": 1.078838174273859e-05,
      "loss": 1.5249,
      "step": 130
    },
    {
      "epoch": 0.01634027691156293,
      "grad_norm": 0.0,
      "learning_rate": 1.0871369294605809e-05,
      "loss": 1.558,
      "step": 131
    },
    {
      "epoch": 0.016465011849819133,
      "grad_norm": 0.0,
      "learning_rate": 1.095435684647303e-05,
      "loss": 1.6392,
      "step": 132
    },
    {
      "epoch": 0.01658974678807534,
      "grad_norm": 0.0,
      "learning_rate": 1.1037344398340251e-05,
      "loss": 1.5706,
      "step": 133
    },
    {
      "epoch": 0.016714481726331547,
      "grad_norm": 0.0,
      "learning_rate": 1.112033195020747e-05,
      "loss": 1.5438,
      "step": 134
    },
    {
      "epoch": 0.01683921666458775,
      "grad_norm": 0.0,
      "learning_rate": 1.120331950207469e-05,
      "loss": 1.5917,
      "step": 135
    },
    {
      "epoch": 0.016963951602843957,
      "grad_norm": 0.0,
      "learning_rate": 1.1286307053941909e-05,
      "loss": 1.5581,
      "step": 136
    },
    {
      "epoch": 0.017088686541100163,
      "grad_norm": 0.0,
      "learning_rate": 1.136929460580913e-05,
      "loss": 1.5807,
      "step": 137
    },
    {
      "epoch": 0.017213421479356367,
      "grad_norm": 0.0,
      "learning_rate": 1.1452282157676348e-05,
      "loss": 1.5166,
      "step": 138
    },
    {
      "epoch": 0.017338156417612573,
      "grad_norm": 0.0,
      "learning_rate": 1.1535269709543569e-05,
      "loss": 1.5847,
      "step": 139
    },
    {
      "epoch": 0.01746289135586878,
      "grad_norm": 0.0,
      "learning_rate": 1.161825726141079e-05,
      "loss": 1.5228,
      "step": 140
    },
    {
      "epoch": 0.017587626294124983,
      "grad_norm": 0.0,
      "learning_rate": 1.170124481327801e-05,
      "loss": 1.521,
      "step": 141
    },
    {
      "epoch": 0.01771236123238119,
      "grad_norm": 0.0,
      "learning_rate": 1.178423236514523e-05,
      "loss": 1.5261,
      "step": 142
    },
    {
      "epoch": 0.017837096170637397,
      "grad_norm": 0.0,
      "learning_rate": 1.1867219917012449e-05,
      "loss": 1.5991,
      "step": 143
    },
    {
      "epoch": 0.0179618311088936,
      "grad_norm": 0.0,
      "learning_rate": 1.1950207468879669e-05,
      "loss": 1.5273,
      "step": 144
    },
    {
      "epoch": 0.018086566047149807,
      "grad_norm": 0.0,
      "learning_rate": 1.2033195020746888e-05,
      "loss": 1.4863,
      "step": 145
    },
    {
      "epoch": 0.018211300985406013,
      "grad_norm": 0.0,
      "learning_rate": 1.2116182572614108e-05,
      "loss": 1.5747,
      "step": 146
    },
    {
      "epoch": 0.018336035923662217,
      "grad_norm": 0.0,
      "learning_rate": 1.219917012448133e-05,
      "loss": 1.599,
      "step": 147
    },
    {
      "epoch": 0.018460770861918423,
      "grad_norm": 0.0,
      "learning_rate": 1.2282157676348549e-05,
      "loss": 1.4974,
      "step": 148
    },
    {
      "epoch": 0.01858550580017463,
      "grad_norm": 0.0,
      "learning_rate": 1.236514522821577e-05,
      "loss": 1.4944,
      "step": 149
    },
    {
      "epoch": 0.018710240738430833,
      "grad_norm": 0.0,
      "learning_rate": 1.2448132780082988e-05,
      "loss": 1.5278,
      "step": 150
    },
    {
      "epoch": 0.01883497567668704,
      "grad_norm": 0.0,
      "learning_rate": 1.2531120331950209e-05,
      "loss": 1.5735,
      "step": 151
    },
    {
      "epoch": 0.018959710614943247,
      "grad_norm": 0.0,
      "learning_rate": 1.2614107883817427e-05,
      "loss": 1.5169,
      "step": 152
    },
    {
      "epoch": 0.01908444555319945,
      "grad_norm": 0.0,
      "learning_rate": 1.2697095435684648e-05,
      "loss": 1.4454,
      "step": 153
    },
    {
      "epoch": 0.019209180491455657,
      "grad_norm": 0.0,
      "learning_rate": 1.2780082987551867e-05,
      "loss": 1.5051,
      "step": 154
    },
    {
      "epoch": 0.019333915429711863,
      "grad_norm": 0.0,
      "learning_rate": 1.2863070539419089e-05,
      "loss": 1.5026,
      "step": 155
    },
    {
      "epoch": 0.019458650367968067,
      "grad_norm": 0.0,
      "learning_rate": 1.2946058091286309e-05,
      "loss": 1.5415,
      "step": 156
    },
    {
      "epoch": 0.019583385306224273,
      "grad_norm": 0.0,
      "learning_rate": 1.3029045643153528e-05,
      "loss": 1.5599,
      "step": 157
    },
    {
      "epoch": 0.01970812024448048,
      "grad_norm": 0.0,
      "learning_rate": 1.3112033195020748e-05,
      "loss": 1.5213,
      "step": 158
    },
    {
      "epoch": 0.019832855182736683,
      "grad_norm": 0.0,
      "learning_rate": 1.3195020746887967e-05,
      "loss": 1.4641,
      "step": 159
    },
    {
      "epoch": 0.01995759012099289,
      "grad_norm": 0.0,
      "learning_rate": 1.3278008298755187e-05,
      "loss": 1.5197,
      "step": 160
    },
    {
      "epoch": 0.020082325059249097,
      "grad_norm": 0.0,
      "learning_rate": 1.3360995850622406e-05,
      "loss": 1.536,
      "step": 161
    },
    {
      "epoch": 0.0202070599975053,
      "grad_norm": 0.0,
      "learning_rate": 1.3443983402489628e-05,
      "loss": 1.44,
      "step": 162
    },
    {
      "epoch": 0.020331794935761507,
      "grad_norm": 0.0,
      "learning_rate": 1.3526970954356849e-05,
      "loss": 1.5401,
      "step": 163
    },
    {
      "epoch": 0.020456529874017713,
      "grad_norm": 0.0,
      "learning_rate": 1.3609958506224067e-05,
      "loss": 1.5658,
      "step": 164
    },
    {
      "epoch": 0.020581264812273917,
      "grad_norm": 0.0,
      "learning_rate": 1.3692946058091288e-05,
      "loss": 1.474,
      "step": 165
    },
    {
      "epoch": 0.020705999750530123,
      "grad_norm": 0.0,
      "learning_rate": 1.3775933609958507e-05,
      "loss": 1.5051,
      "step": 166
    },
    {
      "epoch": 0.02083073468878633,
      "grad_norm": 0.0,
      "learning_rate": 1.3858921161825727e-05,
      "loss": 1.3882,
      "step": 167
    },
    {
      "epoch": 0.020955469627042533,
      "grad_norm": 0.0,
      "learning_rate": 1.3941908713692946e-05,
      "loss": 1.5312,
      "step": 168
    },
    {
      "epoch": 0.02108020456529874,
      "grad_norm": 0.0,
      "learning_rate": 1.4024896265560166e-05,
      "loss": 1.5081,
      "step": 169
    },
    {
      "epoch": 0.021204939503554947,
      "grad_norm": 0.0,
      "learning_rate": 1.4107883817427388e-05,
      "loss": 1.4842,
      "step": 170
    },
    {
      "epoch": 0.02132967444181115,
      "grad_norm": 0.0,
      "learning_rate": 1.4190871369294607e-05,
      "loss": 1.4899,
      "step": 171
    },
    {
      "epoch": 0.021454409380067357,
      "grad_norm": 0.0,
      "learning_rate": 1.4273858921161828e-05,
      "loss": 1.5141,
      "step": 172
    },
    {
      "epoch": 0.021579144318323563,
      "grad_norm": 0.0,
      "learning_rate": 1.4356846473029046e-05,
      "loss": 1.4546,
      "step": 173
    },
    {
      "epoch": 0.021703879256579767,
      "grad_norm": 0.0,
      "learning_rate": 1.4439834024896267e-05,
      "loss": 1.5677,
      "step": 174
    },
    {
      "epoch": 0.021828614194835973,
      "grad_norm": 0.0,
      "learning_rate": 1.4522821576763485e-05,
      "loss": 1.4691,
      "step": 175
    },
    {
      "epoch": 0.02195334913309218,
      "grad_norm": 0.0,
      "learning_rate": 1.4605809128630706e-05,
      "loss": 1.4057,
      "step": 176
    },
    {
      "epoch": 0.022078084071348383,
      "grad_norm": 0.0,
      "learning_rate": 1.4688796680497928e-05,
      "loss": 1.5069,
      "step": 177
    },
    {
      "epoch": 0.02220281900960459,
      "grad_norm": 0.0,
      "learning_rate": 1.4771784232365147e-05,
      "loss": 1.4609,
      "step": 178
    },
    {
      "epoch": 0.022327553947860797,
      "grad_norm": 0.0,
      "learning_rate": 1.4854771784232367e-05,
      "loss": 1.4708,
      "step": 179
    },
    {
      "epoch": 0.022452288886117,
      "grad_norm": 0.0,
      "learning_rate": 1.4937759336099586e-05,
      "loss": 1.5024,
      "step": 180
    },
    {
      "epoch": 0.022577023824373207,
      "grad_norm": 0.0,
      "learning_rate": 1.5020746887966806e-05,
      "loss": 1.4709,
      "step": 181
    },
    {
      "epoch": 0.022701758762629413,
      "grad_norm": 0.0,
      "learning_rate": 1.5103734439834025e-05,
      "loss": 1.4758,
      "step": 182
    },
    {
      "epoch": 0.022826493700885617,
      "grad_norm": 0.0,
      "learning_rate": 1.5186721991701246e-05,
      "loss": 1.4903,
      "step": 183
    },
    {
      "epoch": 0.022951228639141823,
      "grad_norm": 0.0,
      "learning_rate": 1.5269709543568464e-05,
      "loss": 1.4932,
      "step": 184
    },
    {
      "epoch": 0.02307596357739803,
      "grad_norm": 0.0,
      "learning_rate": 1.5352697095435688e-05,
      "loss": 1.4515,
      "step": 185
    },
    {
      "epoch": 0.023200698515654233,
      "grad_norm": 0.0,
      "learning_rate": 1.5435684647302905e-05,
      "loss": 1.453,
      "step": 186
    },
    {
      "epoch": 0.02332543345391044,
      "grad_norm": 0.0,
      "learning_rate": 1.5518672199170126e-05,
      "loss": 1.4594,
      "step": 187
    },
    {
      "epoch": 0.023450168392166647,
      "grad_norm": 0.0,
      "learning_rate": 1.5601659751037346e-05,
      "loss": 1.3783,
      "step": 188
    },
    {
      "epoch": 0.02357490333042285,
      "grad_norm": 0.0,
      "learning_rate": 1.5684647302904566e-05,
      "loss": 1.4978,
      "step": 189
    },
    {
      "epoch": 0.023699638268679057,
      "grad_norm": 0.0,
      "learning_rate": 1.5767634854771783e-05,
      "loss": 1.4607,
      "step": 190
    },
    {
      "epoch": 0.023824373206935263,
      "grad_norm": 0.0,
      "learning_rate": 1.5850622406639004e-05,
      "loss": 1.4478,
      "step": 191
    },
    {
      "epoch": 0.023949108145191467,
      "grad_norm": 0.0,
      "learning_rate": 1.5933609958506224e-05,
      "loss": 1.5088,
      "step": 192
    },
    {
      "epoch": 0.024073843083447673,
      "grad_norm": 0.0,
      "learning_rate": 1.6016597510373445e-05,
      "loss": 1.4428,
      "step": 193
    },
    {
      "epoch": 0.02419857802170388,
      "grad_norm": 0.0,
      "learning_rate": 1.6099585062240665e-05,
      "loss": 1.4286,
      "step": 194
    },
    {
      "epoch": 0.024323312959960083,
      "grad_norm": 0.0,
      "learning_rate": 1.6182572614107886e-05,
      "loss": 1.4267,
      "step": 195
    },
    {
      "epoch": 0.02444804789821629,
      "grad_norm": 0.0,
      "learning_rate": 1.6265560165975106e-05,
      "loss": 1.4085,
      "step": 196
    },
    {
      "epoch": 0.024572782836472497,
      "grad_norm": 0.0,
      "learning_rate": 1.6348547717842323e-05,
      "loss": 1.4354,
      "step": 197
    },
    {
      "epoch": 0.0246975177747287,
      "grad_norm": 0.0,
      "learning_rate": 1.6431535269709543e-05,
      "loss": 1.4773,
      "step": 198
    },
    {
      "epoch": 0.024822252712984907,
      "grad_norm": 0.0,
      "learning_rate": 1.6514522821576764e-05,
      "loss": 1.458,
      "step": 199
    },
    {
      "epoch": 0.024946987651241113,
      "grad_norm": 0.0,
      "learning_rate": 1.6597510373443984e-05,
      "loss": 1.4134,
      "step": 200
    },
    {
      "epoch": 0.025071722589497317,
      "grad_norm": 0.0,
      "learning_rate": 1.6680497925311205e-05,
      "loss": 1.4628,
      "step": 201
    },
    {
      "epoch": 0.025196457527753523,
      "grad_norm": 0.0,
      "learning_rate": 1.6763485477178425e-05,
      "loss": 1.4108,
      "step": 202
    },
    {
      "epoch": 0.02532119246600973,
      "grad_norm": 0.0,
      "learning_rate": 1.6846473029045646e-05,
      "loss": 1.3974,
      "step": 203
    },
    {
      "epoch": 0.025445927404265933,
      "grad_norm": 0.0,
      "learning_rate": 1.6929460580912863e-05,
      "loss": 1.4143,
      "step": 204
    },
    {
      "epoch": 0.02557066234252214,
      "grad_norm": 0.0,
      "learning_rate": 1.7012448132780083e-05,
      "loss": 1.432,
      "step": 205
    },
    {
      "epoch": 0.025695397280778347,
      "grad_norm": 0.0,
      "learning_rate": 1.7095435684647304e-05,
      "loss": 1.3939,
      "step": 206
    },
    {
      "epoch": 0.02582013221903455,
      "grad_norm": 0.0,
      "learning_rate": 1.7178423236514524e-05,
      "loss": 1.4116,
      "step": 207
    },
    {
      "epoch": 0.025944867157290757,
      "grad_norm": 0.0,
      "learning_rate": 1.7261410788381744e-05,
      "loss": 1.4785,
      "step": 208
    },
    {
      "epoch": 0.026069602095546963,
      "grad_norm": 0.0,
      "learning_rate": 1.7344398340248965e-05,
      "loss": 1.4581,
      "step": 209
    },
    {
      "epoch": 0.026194337033803167,
      "grad_norm": 0.0,
      "learning_rate": 1.7427385892116185e-05,
      "loss": 1.4322,
      "step": 210
    },
    {
      "epoch": 0.026319071972059373,
      "grad_norm": 0.0,
      "learning_rate": 1.7510373443983402e-05,
      "loss": 1.3992,
      "step": 211
    },
    {
      "epoch": 0.02644380691031558,
      "grad_norm": 0.0,
      "learning_rate": 1.7593360995850623e-05,
      "loss": 1.4504,
      "step": 212
    },
    {
      "epoch": 0.026568541848571783,
      "grad_norm": 0.0,
      "learning_rate": 1.7676348547717843e-05,
      "loss": 1.4158,
      "step": 213
    },
    {
      "epoch": 0.02669327678682799,
      "grad_norm": 0.0,
      "learning_rate": 1.7759336099585064e-05,
      "loss": 1.4429,
      "step": 214
    },
    {
      "epoch": 0.026818011725084197,
      "grad_norm": 0.0,
      "learning_rate": 1.7842323651452284e-05,
      "loss": 1.3657,
      "step": 215
    },
    {
      "epoch": 0.0269427466633404,
      "grad_norm": 0.0,
      "learning_rate": 1.7925311203319504e-05,
      "loss": 1.4196,
      "step": 216
    },
    {
      "epoch": 0.027067481601596607,
      "grad_norm": 0.0,
      "learning_rate": 1.8008298755186725e-05,
      "loss": 1.4069,
      "step": 217
    },
    {
      "epoch": 0.027192216539852813,
      "grad_norm": 0.0,
      "learning_rate": 1.8091286307053942e-05,
      "loss": 1.4005,
      "step": 218
    },
    {
      "epoch": 0.027316951478109017,
      "grad_norm": 0.0,
      "learning_rate": 1.8174273858921162e-05,
      "loss": 1.4409,
      "step": 219
    },
    {
      "epoch": 0.027441686416365223,
      "grad_norm": 0.0,
      "learning_rate": 1.8257261410788383e-05,
      "loss": 1.4544,
      "step": 220
    },
    {
      "epoch": 0.02756642135462143,
      "grad_norm": 0.0,
      "learning_rate": 1.8340248962655603e-05,
      "loss": 1.3764,
      "step": 221
    },
    {
      "epoch": 0.027691156292877633,
      "grad_norm": 0.0,
      "learning_rate": 1.8423236514522824e-05,
      "loss": 1.3881,
      "step": 222
    },
    {
      "epoch": 0.02781589123113384,
      "grad_norm": 0.0,
      "learning_rate": 1.8506224066390044e-05,
      "loss": 1.4147,
      "step": 223
    },
    {
      "epoch": 0.027940626169390047,
      "grad_norm": 0.0,
      "learning_rate": 1.8589211618257264e-05,
      "loss": 1.3798,
      "step": 224
    },
    {
      "epoch": 0.02806536110764625,
      "grad_norm": 0.0,
      "learning_rate": 1.867219917012448e-05,
      "loss": 1.3735,
      "step": 225
    },
    {
      "epoch": 0.028190096045902457,
      "grad_norm": 0.0,
      "learning_rate": 1.8755186721991702e-05,
      "loss": 1.4403,
      "step": 226
    },
    {
      "epoch": 0.028314830984158663,
      "grad_norm": 0.0,
      "learning_rate": 1.8838174273858922e-05,
      "loss": 1.4008,
      "step": 227
    },
    {
      "epoch": 0.028439565922414867,
      "grad_norm": 0.0,
      "learning_rate": 1.8921161825726143e-05,
      "loss": 1.4804,
      "step": 228
    },
    {
      "epoch": 0.028564300860671073,
      "grad_norm": 0.0,
      "learning_rate": 1.9004149377593363e-05,
      "loss": 1.4216,
      "step": 229
    },
    {
      "epoch": 0.02868903579892728,
      "grad_norm": 0.0,
      "learning_rate": 1.9087136929460584e-05,
      "loss": 1.4521,
      "step": 230
    },
    {
      "epoch": 0.028813770737183483,
      "grad_norm": 0.0,
      "learning_rate": 1.9170124481327804e-05,
      "loss": 1.482,
      "step": 231
    },
    {
      "epoch": 0.02893850567543969,
      "grad_norm": 0.0,
      "learning_rate": 1.925311203319502e-05,
      "loss": 1.423,
      "step": 232
    },
    {
      "epoch": 0.029063240613695897,
      "grad_norm": 0.0,
      "learning_rate": 1.933609958506224e-05,
      "loss": 1.3702,
      "step": 233
    },
    {
      "epoch": 0.029187975551952103,
      "grad_norm": 0.0,
      "learning_rate": 1.9419087136929462e-05,
      "loss": 1.4495,
      "step": 234
    },
    {
      "epoch": 0.029312710490208307,
      "grad_norm": 0.0,
      "learning_rate": 1.9502074688796682e-05,
      "loss": 1.4165,
      "step": 235
    },
    {
      "epoch": 0.029437445428464513,
      "grad_norm": 0.0,
      "learning_rate": 1.9585062240663903e-05,
      "loss": 1.402,
      "step": 236
    },
    {
      "epoch": 0.02956218036672072,
      "grad_norm": 0.0,
      "learning_rate": 1.966804979253112e-05,
      "loss": 1.4022,
      "step": 237
    },
    {
      "epoch": 0.029686915304976923,
      "grad_norm": 0.0,
      "learning_rate": 1.9751037344398344e-05,
      "loss": 1.415,
      "step": 238
    },
    {
      "epoch": 0.02981165024323313,
      "grad_norm": 0.0,
      "learning_rate": 1.983402489626556e-05,
      "loss": 1.393,
      "step": 239
    },
    {
      "epoch": 0.029936385181489337,
      "grad_norm": 0.0,
      "learning_rate": 1.991701244813278e-05,
      "loss": 1.4454,
      "step": 240
    },
    {
      "epoch": 0.03006112011974554,
      "grad_norm": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.444,
      "step": 241
    },
    {
      "epoch": 0.030185855058001747,
      "grad_norm": 0.0,
      "learning_rate": 1.9999999183873952e-05,
      "loss": 1.3646,
      "step": 242
    },
    {
      "epoch": 0.030310589996257953,
      "grad_norm": 0.0,
      "learning_rate": 1.9999996735495937e-05,
      "loss": 1.4145,
      "step": 243
    },
    {
      "epoch": 0.030435324934514157,
      "grad_norm": 0.0,
      "learning_rate": 1.999999265486636e-05,
      "loss": 1.4282,
      "step": 244
    },
    {
      "epoch": 0.030560059872770363,
      "grad_norm": 0.0,
      "learning_rate": 1.999998694198588e-05,
      "loss": 1.3492,
      "step": 245
    },
    {
      "epoch": 0.03068479481102657,
      "grad_norm": 0.0,
      "learning_rate": 1.999997959685543e-05,
      "loss": 1.3837,
      "step": 246
    },
    {
      "epoch": 0.030809529749282773,
      "grad_norm": 0.0,
      "learning_rate": 1.9999970619476214e-05,
      "loss": 1.3545,
      "step": 247
    },
    {
      "epoch": 0.03093426468753898,
      "grad_norm": 0.0,
      "learning_rate": 1.9999960009849694e-05,
      "loss": 1.345,
      "step": 248
    },
    {
      "epoch": 0.031058999625795187,
      "grad_norm": 0.0,
      "learning_rate": 1.9999947767977607e-05,
      "loss": 1.3299,
      "step": 249
    },
    {
      "epoch": 0.03118373456405139,
      "grad_norm": 0.0,
      "learning_rate": 1.9999933893861945e-05,
      "loss": 1.3181,
      "step": 250
    },
    {
      "epoch": 0.0313084695023076,
      "grad_norm": 0.0,
      "learning_rate": 1.9999918387504974e-05,
      "loss": 1.3889,
      "step": 251
    },
    {
      "epoch": 0.0314332044405638,
      "grad_norm": 0.0,
      "learning_rate": 1.9999901248909225e-05,
      "loss": 1.4311,
      "step": 252
    },
    {
      "epoch": 0.03155793937882001,
      "grad_norm": 0.0,
      "learning_rate": 1.9999882478077497e-05,
      "loss": 1.349,
      "step": 253
    },
    {
      "epoch": 0.03168267431707621,
      "grad_norm": 0.0,
      "learning_rate": 1.9999862075012852e-05,
      "loss": 1.4062,
      "step": 254
    },
    {
      "epoch": 0.03180740925533242,
      "grad_norm": 0.0,
      "learning_rate": 1.9999840039718622e-05,
      "loss": 1.3496,
      "step": 255
    },
    {
      "epoch": 0.03193214419358863,
      "grad_norm": 0.0,
      "learning_rate": 1.9999816372198405e-05,
      "loss": 1.364,
      "step": 256
    },
    {
      "epoch": 0.03205687913184483,
      "grad_norm": 0.0,
      "learning_rate": 1.9999791072456062e-05,
      "loss": 1.3767,
      "step": 257
    },
    {
      "epoch": 0.03218161407010103,
      "grad_norm": 0.0,
      "learning_rate": 1.9999764140495718e-05,
      "loss": 1.3467,
      "step": 258
    },
    {
      "epoch": 0.032306349008357244,
      "grad_norm": 0.0,
      "learning_rate": 1.9999735576321776e-05,
      "loss": 1.3541,
      "step": 259
    },
    {
      "epoch": 0.03243108394661345,
      "grad_norm": 0.0,
      "learning_rate": 1.99997053799389e-05,
      "loss": 1.3436,
      "step": 260
    },
    {
      "epoch": 0.03255581888486965,
      "grad_norm": 0.0,
      "learning_rate": 1.9999673551352008e-05,
      "loss": 1.4165,
      "step": 261
    },
    {
      "epoch": 0.03268055382312586,
      "grad_norm": 0.0,
      "learning_rate": 1.9999640090566304e-05,
      "loss": 1.3678,
      "step": 262
    },
    {
      "epoch": 0.03280528876138206,
      "grad_norm": 0.0,
      "learning_rate": 1.9999604997587248e-05,
      "loss": 1.3755,
      "step": 263
    },
    {
      "epoch": 0.03293002369963827,
      "grad_norm": 0.0,
      "learning_rate": 1.9999568272420575e-05,
      "loss": 1.3862,
      "step": 264
    },
    {
      "epoch": 0.03305475863789448,
      "grad_norm": 0.0,
      "learning_rate": 1.999952991507226e-05,
      "loss": 1.3936,
      "step": 265
    },
    {
      "epoch": 0.03317949357615068,
      "grad_norm": 0.0,
      "learning_rate": 1.9999489925548585e-05,
      "loss": 1.3232,
      "step": 266
    },
    {
      "epoch": 0.03330422851440688,
      "grad_norm": 0.0,
      "learning_rate": 1.9999448303856066e-05,
      "loss": 1.3867,
      "step": 267
    },
    {
      "epoch": 0.033428963452663094,
      "grad_norm": 0.0,
      "learning_rate": 1.99994050500015e-05,
      "loss": 1.3374,
      "step": 268
    },
    {
      "epoch": 0.0335536983909193,
      "grad_norm": 0.0,
      "learning_rate": 1.9999360163991946e-05,
      "loss": 1.4554,
      "step": 269
    },
    {
      "epoch": 0.0336784333291755,
      "grad_norm": 0.0,
      "learning_rate": 1.9999313645834727e-05,
      "loss": 1.3641,
      "step": 270
    },
    {
      "epoch": 0.03380316826743171,
      "grad_norm": 0.0,
      "learning_rate": 1.9999265495537443e-05,
      "loss": 1.3695,
      "step": 271
    },
    {
      "epoch": 0.03392790320568791,
      "grad_norm": 0.0,
      "learning_rate": 1.9999215713107952e-05,
      "loss": 1.3617,
      "step": 272
    },
    {
      "epoch": 0.03405263814394412,
      "grad_norm": 0.0,
      "learning_rate": 1.9999164298554375e-05,
      "loss": 1.3378,
      "step": 273
    },
    {
      "epoch": 0.03417737308220033,
      "grad_norm": 0.0,
      "learning_rate": 1.9999111251885108e-05,
      "loss": 1.377,
      "step": 274
    },
    {
      "epoch": 0.03430210802045653,
      "grad_norm": 0.0,
      "learning_rate": 1.9999056573108807e-05,
      "loss": 1.3122,
      "step": 275
    },
    {
      "epoch": 0.03442684295871273,
      "grad_norm": 0.0,
      "learning_rate": 1.99990002622344e-05,
      "loss": 1.3748,
      "step": 276
    },
    {
      "epoch": 0.034551577896968944,
      "grad_norm": 0.0,
      "learning_rate": 1.9998942319271076e-05,
      "loss": 1.3574,
      "step": 277
    },
    {
      "epoch": 0.03467631283522515,
      "grad_norm": 0.0,
      "learning_rate": 1.9998882744228294e-05,
      "loss": 1.3534,
      "step": 278
    },
    {
      "epoch": 0.03480104777348135,
      "grad_norm": 0.0,
      "learning_rate": 1.999882153711578e-05,
      "loss": 1.3472,
      "step": 279
    },
    {
      "epoch": 0.03492578271173756,
      "grad_norm": 0.0,
      "learning_rate": 1.9998758697943518e-05,
      "loss": 1.3322,
      "step": 280
    },
    {
      "epoch": 0.03505051764999376,
      "grad_norm": 0.0,
      "learning_rate": 1.9998694226721774e-05,
      "loss": 1.3991,
      "step": 281
    },
    {
      "epoch": 0.03517525258824997,
      "grad_norm": 0.0,
      "learning_rate": 1.9998628123461064e-05,
      "loss": 1.3986,
      "step": 282
    },
    {
      "epoch": 0.03529998752650618,
      "grad_norm": 0.0,
      "learning_rate": 1.999856038817218e-05,
      "loss": 1.3606,
      "step": 283
    },
    {
      "epoch": 0.03542472246476238,
      "grad_norm": 0.0,
      "learning_rate": 1.9998491020866176e-05,
      "loss": 1.3033,
      "step": 284
    },
    {
      "epoch": 0.03554945740301858,
      "grad_norm": 0.0,
      "learning_rate": 1.999842002155438e-05,
      "loss": 1.3275,
      "step": 285
    },
    {
      "epoch": 0.035674192341274794,
      "grad_norm": 0.0,
      "learning_rate": 1.9998347390248377e-05,
      "loss": 1.3367,
      "step": 286
    },
    {
      "epoch": 0.035798927279531,
      "grad_norm": 0.0,
      "learning_rate": 1.9998273126960024e-05,
      "loss": 1.3499,
      "step": 287
    },
    {
      "epoch": 0.0359236622177872,
      "grad_norm": 0.0,
      "learning_rate": 1.999819723170144e-05,
      "loss": 1.3171,
      "step": 288
    },
    {
      "epoch": 0.03604839715604341,
      "grad_norm": 0.0,
      "learning_rate": 1.9998119704485016e-05,
      "loss": 1.3367,
      "step": 289
    },
    {
      "epoch": 0.03617313209429961,
      "grad_norm": 0.0,
      "learning_rate": 1.9998040545323403e-05,
      "loss": 1.3589,
      "step": 290
    },
    {
      "epoch": 0.03629786703255582,
      "grad_norm": 0.0,
      "learning_rate": 1.9997959754229526e-05,
      "loss": 1.3679,
      "step": 291
    },
    {
      "epoch": 0.03642260197081203,
      "grad_norm": 0.0,
      "learning_rate": 1.9997877331216568e-05,
      "loss": 1.3084,
      "step": 292
    },
    {
      "epoch": 0.03654733690906823,
      "grad_norm": 0.0,
      "learning_rate": 1.9997793276297988e-05,
      "loss": 1.3152,
      "step": 293
    },
    {
      "epoch": 0.03667207184732443,
      "grad_norm": 0.0,
      "learning_rate": 1.99977075894875e-05,
      "loss": 1.4142,
      "step": 294
    },
    {
      "epoch": 0.036796806785580644,
      "grad_norm": 0.0,
      "learning_rate": 1.999762027079909e-05,
      "loss": 1.3081,
      "step": 295
    },
    {
      "epoch": 0.03692154172383685,
      "grad_norm": 0.0,
      "learning_rate": 1.9997531320247017e-05,
      "loss": 1.3402,
      "step": 296
    },
    {
      "epoch": 0.03704627666209305,
      "grad_norm": 0.0,
      "learning_rate": 1.9997440737845798e-05,
      "loss": 1.3714,
      "step": 297
    },
    {
      "epoch": 0.03717101160034926,
      "grad_norm": 0.0,
      "learning_rate": 1.9997348523610212e-05,
      "loss": 1.3283,
      "step": 298
    },
    {
      "epoch": 0.03729574653860546,
      "grad_norm": 0.0,
      "learning_rate": 1.999725467755532e-05,
      "loss": 1.3152,
      "step": 299
    },
    {
      "epoch": 0.03742048147686167,
      "grad_norm": 0.0,
      "learning_rate": 1.999715919969643e-05,
      "loss": 1.2391,
      "step": 300
    },
    {
      "epoch": 0.03754521641511788,
      "grad_norm": 0.0,
      "learning_rate": 1.9997062090049135e-05,
      "loss": 1.3388,
      "step": 301
    },
    {
      "epoch": 0.03766995135337408,
      "grad_norm": 0.0,
      "learning_rate": 1.999696334862928e-05,
      "loss": 1.3442,
      "step": 302
    },
    {
      "epoch": 0.03779468629163028,
      "grad_norm": 0.0,
      "learning_rate": 1.9996862975452986e-05,
      "loss": 1.3282,
      "step": 303
    },
    {
      "epoch": 0.037919421229886494,
      "grad_norm": 0.0,
      "learning_rate": 1.9996760970536635e-05,
      "loss": 1.3693,
      "step": 304
    },
    {
      "epoch": 0.0380441561681427,
      "grad_norm": 0.0,
      "learning_rate": 1.9996657333896875e-05,
      "loss": 1.315,
      "step": 305
    },
    {
      "epoch": 0.0381688911063989,
      "grad_norm": 0.0,
      "learning_rate": 1.9996552065550626e-05,
      "loss": 1.2988,
      "step": 306
    },
    {
      "epoch": 0.03829362604465511,
      "grad_norm": 0.0,
      "learning_rate": 1.9996445165515066e-05,
      "loss": 1.3309,
      "step": 307
    },
    {
      "epoch": 0.03841836098291131,
      "grad_norm": 0.0,
      "learning_rate": 1.9996336633807648e-05,
      "loss": 1.3785,
      "step": 308
    },
    {
      "epoch": 0.03854309592116752,
      "grad_norm": 0.0,
      "learning_rate": 1.9996226470446088e-05,
      "loss": 1.2817,
      "step": 309
    },
    {
      "epoch": 0.03866783085942373,
      "grad_norm": 0.0,
      "learning_rate": 1.999611467544836e-05,
      "loss": 1.3019,
      "step": 310
    },
    {
      "epoch": 0.03879256579767993,
      "grad_norm": 0.0,
      "learning_rate": 1.9996001248832715e-05,
      "loss": 1.3315,
      "step": 311
    },
    {
      "epoch": 0.03891730073593613,
      "grad_norm": 0.0,
      "learning_rate": 1.999588619061767e-05,
      "loss": 1.3281,
      "step": 312
    },
    {
      "epoch": 0.039042035674192344,
      "grad_norm": 0.0,
      "learning_rate": 1.9995769500822007e-05,
      "loss": 1.2834,
      "step": 313
    },
    {
      "epoch": 0.03916677061244855,
      "grad_norm": 0.0,
      "learning_rate": 1.999565117946477e-05,
      "loss": 1.3226,
      "step": 314
    },
    {
      "epoch": 0.03929150555070475,
      "grad_norm": 0.0,
      "learning_rate": 1.9995531226565267e-05,
      "loss": 1.3378,
      "step": 315
    },
    {
      "epoch": 0.03941624048896096,
      "grad_norm": 0.0,
      "learning_rate": 1.9995409642143083e-05,
      "loss": 1.2994,
      "step": 316
    },
    {
      "epoch": 0.03954097542721716,
      "grad_norm": 0.0,
      "learning_rate": 1.9995286426218065e-05,
      "loss": 1.2816,
      "step": 317
    },
    {
      "epoch": 0.03966571036547337,
      "grad_norm": 0.0,
      "learning_rate": 1.999516157881032e-05,
      "loss": 1.2961,
      "step": 318
    },
    {
      "epoch": 0.03979044530372958,
      "grad_norm": 0.0,
      "learning_rate": 1.999503509994023e-05,
      "loss": 1.304,
      "step": 319
    },
    {
      "epoch": 0.03991518024198578,
      "grad_norm": 0.0,
      "learning_rate": 1.9994906989628437e-05,
      "loss": 1.3051,
      "step": 320
    },
    {
      "epoch": 0.04003991518024198,
      "grad_norm": 0.0,
      "learning_rate": 1.9994777247895855e-05,
      "loss": 1.3011,
      "step": 321
    },
    {
      "epoch": 0.040164650118498194,
      "grad_norm": 0.0,
      "learning_rate": 1.9994645874763657e-05,
      "loss": 1.3433,
      "step": 322
    },
    {
      "epoch": 0.0402893850567544,
      "grad_norm": 0.0,
      "learning_rate": 1.999451287025329e-05,
      "loss": 1.2542,
      "step": 323
    },
    {
      "epoch": 0.0404141199950106,
      "grad_norm": 0.0,
      "learning_rate": 1.9994378234386464e-05,
      "loss": 1.2776,
      "step": 324
    },
    {
      "epoch": 0.04053885493326681,
      "grad_norm": 0.0,
      "learning_rate": 1.999424196718515e-05,
      "loss": 1.3314,
      "step": 325
    },
    {
      "epoch": 0.04066358987152301,
      "grad_norm": 0.0,
      "learning_rate": 1.9994104068671597e-05,
      "loss": 1.2867,
      "step": 326
    },
    {
      "epoch": 0.04078832480977922,
      "grad_norm": 0.0,
      "learning_rate": 1.9993964538868307e-05,
      "loss": 1.3129,
      "step": 327
    },
    {
      "epoch": 0.04091305974803543,
      "grad_norm": 0.0,
      "learning_rate": 1.9993823377798058e-05,
      "loss": 1.3098,
      "step": 328
    },
    {
      "epoch": 0.04103779468629163,
      "grad_norm": 0.0,
      "learning_rate": 1.9993680585483895e-05,
      "loss": 1.3344,
      "step": 329
    },
    {
      "epoch": 0.04116252962454783,
      "grad_norm": 0.0,
      "learning_rate": 1.9993536161949118e-05,
      "loss": 1.2674,
      "step": 330
    },
    {
      "epoch": 0.041287264562804044,
      "grad_norm": 0.0,
      "learning_rate": 1.9993390107217302e-05,
      "loss": 1.3734,
      "step": 331
    },
    {
      "epoch": 0.04141199950106025,
      "grad_norm": 0.0,
      "learning_rate": 1.9993242421312292e-05,
      "loss": 1.2497,
      "step": 332
    },
    {
      "epoch": 0.04153673443931645,
      "grad_norm": 0.0,
      "learning_rate": 1.9993093104258186e-05,
      "loss": 1.2866,
      "step": 333
    },
    {
      "epoch": 0.04166146937757266,
      "grad_norm": 0.0,
      "learning_rate": 1.9992942156079364e-05,
      "loss": 1.3242,
      "step": 334
    },
    {
      "epoch": 0.04178620431582886,
      "grad_norm": 0.0,
      "learning_rate": 1.9992789576800464e-05,
      "loss": 1.3275,
      "step": 335
    },
    {
      "epoch": 0.04191093925408507,
      "grad_norm": 0.0,
      "learning_rate": 1.9992635366446387e-05,
      "loss": 1.3239,
      "step": 336
    },
    {
      "epoch": 0.04203567419234128,
      "grad_norm": 0.0,
      "learning_rate": 1.9992479525042305e-05,
      "loss": 1.2924,
      "step": 337
    },
    {
      "epoch": 0.04216040913059748,
      "grad_norm": 0.0,
      "learning_rate": 1.999232205261365e-05,
      "loss": 1.2626,
      "step": 338
    },
    {
      "epoch": 0.04228514406885368,
      "grad_norm": 0.0,
      "learning_rate": 1.9992162949186138e-05,
      "loss": 1.3005,
      "step": 339
    },
    {
      "epoch": 0.042409879007109894,
      "grad_norm": 0.0,
      "learning_rate": 1.999200221478573e-05,
      "loss": 1.324,
      "step": 340
    },
    {
      "epoch": 0.0425346139453661,
      "grad_norm": 0.0,
      "learning_rate": 1.999183984943866e-05,
      "loss": 1.357,
      "step": 341
    },
    {
      "epoch": 0.0426593488836223,
      "grad_norm": 0.0,
      "learning_rate": 1.9991675853171438e-05,
      "loss": 1.2928,
      "step": 342
    },
    {
      "epoch": 0.04278408382187851,
      "grad_norm": 0.0,
      "learning_rate": 1.9991510226010827e-05,
      "loss": 1.2863,
      "step": 343
    },
    {
      "epoch": 0.04290881876013471,
      "grad_norm": 0.0,
      "learning_rate": 1.9991342967983863e-05,
      "loss": 1.3611,
      "step": 344
    },
    {
      "epoch": 0.04303355369839092,
      "grad_norm": 0.0,
      "learning_rate": 1.9991174079117846e-05,
      "loss": 1.2972,
      "step": 345
    },
    {
      "epoch": 0.04315828863664713,
      "grad_norm": 0.0,
      "learning_rate": 1.9991003559440342e-05,
      "loss": 1.3023,
      "step": 346
    },
    {
      "epoch": 0.04328302357490333,
      "grad_norm": 0.0,
      "learning_rate": 1.9990831408979186e-05,
      "loss": 1.3423,
      "step": 347
    },
    {
      "epoch": 0.04340775851315953,
      "grad_norm": 0.0,
      "learning_rate": 1.9990657627762474e-05,
      "loss": 1.2875,
      "step": 348
    },
    {
      "epoch": 0.043532493451415744,
      "grad_norm": 0.0,
      "learning_rate": 1.999048221581858e-05,
      "loss": 1.2739,
      "step": 349
    },
    {
      "epoch": 0.04365722838967195,
      "grad_norm": 0.0,
      "learning_rate": 1.9990305173176124e-05,
      "loss": 1.2695,
      "step": 350
    },
    {
      "epoch": 0.04378196332792815,
      "grad_norm": 0.0,
      "learning_rate": 1.9990126499864013e-05,
      "loss": 1.2989,
      "step": 351
    },
    {
      "epoch": 0.04390669826618436,
      "grad_norm": 0.0,
      "learning_rate": 1.9989946195911408e-05,
      "loss": 1.2307,
      "step": 352
    },
    {
      "epoch": 0.04403143320444056,
      "grad_norm": 0.0,
      "learning_rate": 1.9989764261347738e-05,
      "loss": 1.3085,
      "step": 353
    },
    {
      "epoch": 0.04415616814269677,
      "grad_norm": 0.0,
      "learning_rate": 1.9989580696202696e-05,
      "loss": 1.3397,
      "step": 354
    },
    {
      "epoch": 0.04428090308095298,
      "grad_norm": 0.0,
      "learning_rate": 1.9989395500506254e-05,
      "loss": 1.2844,
      "step": 355
    },
    {
      "epoch": 0.04440563801920918,
      "grad_norm": 0.0,
      "learning_rate": 1.9989208674288635e-05,
      "loss": 1.2952,
      "step": 356
    },
    {
      "epoch": 0.04453037295746538,
      "grad_norm": 0.0,
      "learning_rate": 1.998902021758033e-05,
      "loss": 1.259,
      "step": 357
    },
    {
      "epoch": 0.044655107895721594,
      "grad_norm": 0.0,
      "learning_rate": 1.9988830130412106e-05,
      "loss": 1.3077,
      "step": 358
    },
    {
      "epoch": 0.0447798428339778,
      "grad_norm": 0.0,
      "learning_rate": 1.9988638412814986e-05,
      "loss": 1.3417,
      "step": 359
    },
    {
      "epoch": 0.044904577772234,
      "grad_norm": 0.0,
      "learning_rate": 1.9988445064820267e-05,
      "loss": 1.2841,
      "step": 360
    },
    {
      "epoch": 0.04502931271049021,
      "grad_norm": 0.0,
      "learning_rate": 1.9988250086459505e-05,
      "loss": 1.3176,
      "step": 361
    },
    {
      "epoch": 0.04515404764874641,
      "grad_norm": 0.0,
      "learning_rate": 1.9988053477764527e-05,
      "loss": 1.3395,
      "step": 362
    },
    {
      "epoch": 0.04527878258700262,
      "grad_norm": 0.0,
      "learning_rate": 1.9987855238767423e-05,
      "loss": 1.3195,
      "step": 363
    },
    {
      "epoch": 0.04540351752525883,
      "grad_norm": 0.0,
      "learning_rate": 1.9987655369500554e-05,
      "loss": 1.2623,
      "step": 364
    },
    {
      "epoch": 0.04552825246351503,
      "grad_norm": 0.0,
      "learning_rate": 1.9987453869996542e-05,
      "loss": 1.3117,
      "step": 365
    },
    {
      "epoch": 0.04565298740177123,
      "grad_norm": 0.0,
      "learning_rate": 1.998725074028827e-05,
      "loss": 1.3337,
      "step": 366
    },
    {
      "epoch": 0.045777722340027444,
      "grad_norm": 0.0,
      "learning_rate": 1.9987045980408907e-05,
      "loss": 1.272,
      "step": 367
    },
    {
      "epoch": 0.04590245727828365,
      "grad_norm": 0.0,
      "learning_rate": 1.9986839590391864e-05,
      "loss": 1.2695,
      "step": 368
    },
    {
      "epoch": 0.04602719221653985,
      "grad_norm": 0.0,
      "learning_rate": 1.9986631570270835e-05,
      "loss": 1.3513,
      "step": 369
    },
    {
      "epoch": 0.04615192715479606,
      "grad_norm": 0.0,
      "learning_rate": 1.998642192007977e-05,
      "loss": 1.2675,
      "step": 370
    },
    {
      "epoch": 0.04627666209305226,
      "grad_norm": 0.0,
      "learning_rate": 1.9986210639852892e-05,
      "loss": 1.285,
      "step": 371
    },
    {
      "epoch": 0.04640139703130847,
      "grad_norm": 0.0,
      "learning_rate": 1.9985997729624687e-05,
      "loss": 1.2039,
      "step": 372
    },
    {
      "epoch": 0.04652613196956468,
      "grad_norm": 0.0,
      "learning_rate": 1.9985783189429907e-05,
      "loss": 1.259,
      "step": 373
    },
    {
      "epoch": 0.04665086690782088,
      "grad_norm": 0.0,
      "learning_rate": 1.9985567019303573e-05,
      "loss": 1.3074,
      "step": 374
    },
    {
      "epoch": 0.04677560184607708,
      "grad_norm": 0.0,
      "learning_rate": 1.9985349219280964e-05,
      "loss": 1.2448,
      "step": 375
    },
    {
      "epoch": 0.046900336784333294,
      "grad_norm": 0.0,
      "learning_rate": 1.9985129789397633e-05,
      "loss": 1.2575,
      "step": 376
    },
    {
      "epoch": 0.0470250717225895,
      "grad_norm": 0.0,
      "learning_rate": 1.9984908729689398e-05,
      "loss": 1.2684,
      "step": 377
    },
    {
      "epoch": 0.0471498066608457,
      "grad_norm": 0.0,
      "learning_rate": 1.9984686040192337e-05,
      "loss": 1.3248,
      "step": 378
    },
    {
      "epoch": 0.04727454159910191,
      "grad_norm": 0.0,
      "learning_rate": 1.9984461720942804e-05,
      "loss": 1.2474,
      "step": 379
    },
    {
      "epoch": 0.047399276537358113,
      "grad_norm": 0.0,
      "learning_rate": 1.9984235771977412e-05,
      "loss": 1.3517,
      "step": 380
    },
    {
      "epoch": 0.04752401147561432,
      "grad_norm": 0.0,
      "learning_rate": 1.9984008193333043e-05,
      "loss": 1.2417,
      "step": 381
    },
    {
      "epoch": 0.04764874641387053,
      "grad_norm": 0.0,
      "learning_rate": 1.998377898504684e-05,
      "loss": 1.2614,
      "step": 382
    },
    {
      "epoch": 0.04777348135212673,
      "grad_norm": 0.0,
      "learning_rate": 1.9983548147156214e-05,
      "loss": 1.3036,
      "step": 383
    },
    {
      "epoch": 0.04789821629038293,
      "grad_norm": 0.0,
      "learning_rate": 1.9983315679698844e-05,
      "loss": 1.2699,
      "step": 384
    },
    {
      "epoch": 0.048022951228639144,
      "grad_norm": 0.0,
      "learning_rate": 1.9983081582712684e-05,
      "loss": 1.2444,
      "step": 385
    },
    {
      "epoch": 0.04814768616689535,
      "grad_norm": 0.0,
      "learning_rate": 1.9982845856235934e-05,
      "loss": 1.2605,
      "step": 386
    },
    {
      "epoch": 0.04827242110515155,
      "grad_norm": 0.0,
      "learning_rate": 1.9982608500307074e-05,
      "loss": 1.1995,
      "step": 387
    },
    {
      "epoch": 0.04839715604340776,
      "grad_norm": 0.0,
      "learning_rate": 1.998236951496485e-05,
      "loss": 1.2644,
      "step": 388
    },
    {
      "epoch": 0.048521890981663963,
      "grad_norm": 0.0,
      "learning_rate": 1.9982128900248266e-05,
      "loss": 1.277,
      "step": 389
    },
    {
      "epoch": 0.04864662591992017,
      "grad_norm": 0.0,
      "learning_rate": 1.9981886656196595e-05,
      "loss": 1.2546,
      "step": 390
    },
    {
      "epoch": 0.04877136085817638,
      "grad_norm": 0.0,
      "learning_rate": 1.9981642782849378e-05,
      "loss": 1.2752,
      "step": 391
    },
    {
      "epoch": 0.04889609579643258,
      "grad_norm": 0.0,
      "learning_rate": 1.9981397280246427e-05,
      "loss": 1.2784,
      "step": 392
    },
    {
      "epoch": 0.04902083073468878,
      "grad_norm": 0.0,
      "learning_rate": 1.9981150148427814e-05,
      "loss": 1.2501,
      "step": 393
    },
    {
      "epoch": 0.049145565672944994,
      "grad_norm": 0.0,
      "learning_rate": 1.9980901387433866e-05,
      "loss": 1.2794,
      "step": 394
    },
    {
      "epoch": 0.0492703006112012,
      "grad_norm": 0.0,
      "learning_rate": 1.9980650997305194e-05,
      "loss": 1.2524,
      "step": 395
    },
    {
      "epoch": 0.0493950355494574,
      "grad_norm": 0.0,
      "learning_rate": 1.9980398978082673e-05,
      "loss": 1.2914,
      "step": 396
    },
    {
      "epoch": 0.04951977048771361,
      "grad_norm": 0.0,
      "learning_rate": 1.9980145329807435e-05,
      "loss": 1.2453,
      "step": 397
    },
    {
      "epoch": 0.049644505425969813,
      "grad_norm": 0.0,
      "learning_rate": 1.997989005252088e-05,
      "loss": 1.2656,
      "step": 398
    },
    {
      "epoch": 0.04976924036422602,
      "grad_norm": 0.0,
      "learning_rate": 1.9979633146264675e-05,
      "loss": 1.2875,
      "step": 399
    },
    {
      "epoch": 0.04989397530248223,
      "grad_norm": 0.0,
      "learning_rate": 1.997937461108076e-05,
      "loss": 1.2415,
      "step": 400
    },
    {
      "epoch": 0.05001871024073843,
      "grad_norm": 0.0,
      "learning_rate": 1.9979114447011323e-05,
      "loss": 1.269,
      "step": 401
    },
    {
      "epoch": 0.05014344517899463,
      "grad_norm": 0.0,
      "learning_rate": 1.997885265409884e-05,
      "loss": 1.229,
      "step": 402
    },
    {
      "epoch": 0.050268180117250844,
      "grad_norm": 0.0,
      "learning_rate": 1.9978589232386036e-05,
      "loss": 1.2219,
      "step": 403
    },
    {
      "epoch": 0.05039291505550705,
      "grad_norm": 0.0,
      "learning_rate": 1.9978324181915915e-05,
      "loss": 1.2593,
      "step": 404
    },
    {
      "epoch": 0.05051764999376325,
      "grad_norm": 0.0,
      "learning_rate": 1.997805750273173e-05,
      "loss": 1.2693,
      "step": 405
    },
    {
      "epoch": 0.05064238493201946,
      "grad_norm": 0.0,
      "learning_rate": 1.997778919487702e-05,
      "loss": 1.3044,
      "step": 406
    },
    {
      "epoch": 0.050767119870275663,
      "grad_norm": 0.0,
      "learning_rate": 1.997751925839557e-05,
      "loss": 1.292,
      "step": 407
    },
    {
      "epoch": 0.05089185480853187,
      "grad_norm": 0.0,
      "learning_rate": 1.9977247693331447e-05,
      "loss": 1.2645,
      "step": 408
    },
    {
      "epoch": 0.05101658974678808,
      "grad_norm": 0.0,
      "learning_rate": 1.997697449972898e-05,
      "loss": 1.2001,
      "step": 409
    },
    {
      "epoch": 0.05114132468504428,
      "grad_norm": 0.0,
      "learning_rate": 1.9976699677632753e-05,
      "loss": 1.3092,
      "step": 410
    },
    {
      "epoch": 0.05126605962330048,
      "grad_norm": 0.0,
      "learning_rate": 1.9976423227087626e-05,
      "loss": 1.2694,
      "step": 411
    },
    {
      "epoch": 0.051390794561556694,
      "grad_norm": 0.0,
      "learning_rate": 1.9976145148138727e-05,
      "loss": 1.2837,
      "step": 412
    },
    {
      "epoch": 0.0515155294998129,
      "grad_norm": 0.0,
      "learning_rate": 1.997586544083144e-05,
      "loss": 1.2151,
      "step": 413
    },
    {
      "epoch": 0.0516402644380691,
      "grad_norm": 0.0,
      "learning_rate": 1.9975584105211425e-05,
      "loss": 1.2963,
      "step": 414
    },
    {
      "epoch": 0.05176499937632531,
      "grad_norm": 0.0,
      "learning_rate": 1.99753011413246e-05,
      "loss": 1.3239,
      "step": 415
    },
    {
      "epoch": 0.051889734314581513,
      "grad_norm": 0.0,
      "learning_rate": 1.997501654921715e-05,
      "loss": 1.2647,
      "step": 416
    },
    {
      "epoch": 0.05201446925283772,
      "grad_norm": 0.0,
      "learning_rate": 1.9974730328935534e-05,
      "loss": 1.2288,
      "step": 417
    },
    {
      "epoch": 0.05213920419109393,
      "grad_norm": 0.0,
      "learning_rate": 1.9974442480526465e-05,
      "loss": 1.2907,
      "step": 418
    },
    {
      "epoch": 0.05226393912935013,
      "grad_norm": 0.0,
      "learning_rate": 1.997415300403693e-05,
      "loss": 1.2991,
      "step": 419
    },
    {
      "epoch": 0.05238867406760633,
      "grad_norm": 0.0,
      "learning_rate": 1.997386189951418e-05,
      "loss": 1.2482,
      "step": 420
    },
    {
      "epoch": 0.052513409005862544,
      "grad_norm": 0.0,
      "learning_rate": 1.9973569167005725e-05,
      "loss": 1.2801,
      "step": 421
    },
    {
      "epoch": 0.05263814394411875,
      "grad_norm": 0.0,
      "learning_rate": 1.997327480655935e-05,
      "loss": 1.2961,
      "step": 422
    },
    {
      "epoch": 0.05276287888237495,
      "grad_norm": 0.0,
      "learning_rate": 1.99729788182231e-05,
      "loss": 1.2311,
      "step": 423
    },
    {
      "epoch": 0.05288761382063116,
      "grad_norm": 0.0,
      "learning_rate": 1.9972681202045294e-05,
      "loss": 1.2179,
      "step": 424
    },
    {
      "epoch": 0.053012348758887363,
      "grad_norm": 0.0,
      "learning_rate": 1.9972381958074505e-05,
      "loss": 1.2805,
      "step": 425
    },
    {
      "epoch": 0.05313708369714357,
      "grad_norm": 0.0,
      "learning_rate": 1.9972081086359576e-05,
      "loss": 1.2744,
      "step": 426
    },
    {
      "epoch": 0.05326181863539978,
      "grad_norm": 0.0,
      "learning_rate": 1.997177858694962e-05,
      "loss": 1.2847,
      "step": 427
    },
    {
      "epoch": 0.05338655357365598,
      "grad_norm": 0.0,
      "learning_rate": 1.997147445989401e-05,
      "loss": 1.2783,
      "step": 428
    },
    {
      "epoch": 0.05351128851191218,
      "grad_norm": 0.0,
      "learning_rate": 1.9971168705242392e-05,
      "loss": 1.2476,
      "step": 429
    },
    {
      "epoch": 0.053636023450168394,
      "grad_norm": 0.0,
      "learning_rate": 1.9970861323044667e-05,
      "loss": 1.2692,
      "step": 430
    },
    {
      "epoch": 0.0537607583884246,
      "grad_norm": 0.0,
      "learning_rate": 1.997055231335101e-05,
      "loss": 1.2622,
      "step": 431
    },
    {
      "epoch": 0.0538854933266808,
      "grad_norm": 0.0,
      "learning_rate": 1.997024167621186e-05,
      "loss": 1.3137,
      "step": 432
    },
    {
      "epoch": 0.05401022826493701,
      "grad_norm": 0.0,
      "learning_rate": 1.996992941167792e-05,
      "loss": 1.2643,
      "step": 433
    },
    {
      "epoch": 0.054134963203193213,
      "grad_norm": 0.0,
      "learning_rate": 1.9969615519800163e-05,
      "loss": 1.2616,
      "step": 434
    },
    {
      "epoch": 0.05425969814144942,
      "grad_norm": 0.0,
      "learning_rate": 1.9969300000629815e-05,
      "loss": 1.2589,
      "step": 435
    },
    {
      "epoch": 0.05438443307970563,
      "grad_norm": 0.0,
      "learning_rate": 1.9968982854218388e-05,
      "loss": 1.3031,
      "step": 436
    },
    {
      "epoch": 0.05450916801796183,
      "grad_norm": 0.0,
      "learning_rate": 1.9968664080617637e-05,
      "loss": 1.2956,
      "step": 437
    },
    {
      "epoch": 0.05463390295621803,
      "grad_norm": 0.0,
      "learning_rate": 1.9968343679879602e-05,
      "loss": 1.2476,
      "step": 438
    },
    {
      "epoch": 0.054758637894474244,
      "grad_norm": 0.0,
      "learning_rate": 1.9968021652056578e-05,
      "loss": 1.2601,
      "step": 439
    },
    {
      "epoch": 0.05488337283273045,
      "grad_norm": 0.0,
      "learning_rate": 1.9967697997201125e-05,
      "loss": 1.2869,
      "step": 440
    },
    {
      "epoch": 0.05500810777098665,
      "grad_norm": 0.0,
      "learning_rate": 1.996737271536608e-05,
      "loss": 1.2989,
      "step": 441
    },
    {
      "epoch": 0.05513284270924286,
      "grad_norm": 0.0,
      "learning_rate": 1.996704580660453e-05,
      "loss": 1.2529,
      "step": 442
    },
    {
      "epoch": 0.055257577647499063,
      "grad_norm": 0.0,
      "learning_rate": 1.9966717270969834e-05,
      "loss": 1.2828,
      "step": 443
    },
    {
      "epoch": 0.05538231258575527,
      "grad_norm": 0.0,
      "learning_rate": 1.996638710851562e-05,
      "loss": 1.2416,
      "step": 444
    },
    {
      "epoch": 0.05550704752401148,
      "grad_norm": 0.0,
      "learning_rate": 1.996605531929578e-05,
      "loss": 1.2961,
      "step": 445
    },
    {
      "epoch": 0.05563178246226768,
      "grad_norm": 0.0,
      "learning_rate": 1.996572190336447e-05,
      "loss": 1.255,
      "step": 446
    },
    {
      "epoch": 0.05575651740052388,
      "grad_norm": 0.0,
      "learning_rate": 1.996538686077611e-05,
      "loss": 1.2572,
      "step": 447
    },
    {
      "epoch": 0.055881252338780094,
      "grad_norm": 0.0,
      "learning_rate": 1.9965050191585388e-05,
      "loss": 1.2395,
      "step": 448
    },
    {
      "epoch": 0.0560059872770363,
      "grad_norm": 0.0,
      "learning_rate": 1.996471189584726e-05,
      "loss": 1.2694,
      "step": 449
    },
    {
      "epoch": 0.0561307222152925,
      "grad_norm": 0.0,
      "learning_rate": 1.9964371973616938e-05,
      "loss": 1.2789,
      "step": 450
    },
    {
      "epoch": 0.05625545715354871,
      "grad_norm": 0.0,
      "learning_rate": 1.996403042494991e-05,
      "loss": 1.2894,
      "step": 451
    },
    {
      "epoch": 0.056380192091804913,
      "grad_norm": 0.0,
      "learning_rate": 1.9963687249901927e-05,
      "loss": 1.2087,
      "step": 452
    },
    {
      "epoch": 0.05650492703006112,
      "grad_norm": 0.0,
      "learning_rate": 1.9963342448528998e-05,
      "loss": 1.2417,
      "step": 453
    },
    {
      "epoch": 0.05662966196831733,
      "grad_norm": 0.0,
      "learning_rate": 1.996299602088741e-05,
      "loss": 1.2704,
      "step": 454
    },
    {
      "epoch": 0.05675439690657353,
      "grad_norm": 0.0,
      "learning_rate": 1.9962647967033707e-05,
      "loss": 1.2597,
      "step": 455
    },
    {
      "epoch": 0.05687913184482973,
      "grad_norm": 0.0,
      "learning_rate": 1.9962298287024695e-05,
      "loss": 1.2427,
      "step": 456
    },
    {
      "epoch": 0.057003866783085944,
      "grad_norm": 0.0,
      "learning_rate": 1.9961946980917457e-05,
      "loss": 1.276,
      "step": 457
    },
    {
      "epoch": 0.05712860172134215,
      "grad_norm": 0.0,
      "learning_rate": 1.996159404876933e-05,
      "loss": 1.237,
      "step": 458
    },
    {
      "epoch": 0.05725333665959835,
      "grad_norm": 0.0,
      "learning_rate": 1.9961239490637927e-05,
      "loss": 1.2184,
      "step": 459
    },
    {
      "epoch": 0.05737807159785456,
      "grad_norm": 0.0,
      "learning_rate": 1.9960883306581116e-05,
      "loss": 1.3028,
      "step": 460
    },
    {
      "epoch": 0.057502806536110763,
      "grad_norm": 0.0,
      "learning_rate": 1.9960525496657038e-05,
      "loss": 1.2594,
      "step": 461
    },
    {
      "epoch": 0.05762754147436697,
      "grad_norm": 0.0,
      "learning_rate": 1.9960166060924094e-05,
      "loss": 1.2484,
      "step": 462
    },
    {
      "epoch": 0.05775227641262318,
      "grad_norm": 0.0,
      "learning_rate": 1.9959804999440953e-05,
      "loss": 1.293,
      "step": 463
    },
    {
      "epoch": 0.05787701135087938,
      "grad_norm": 0.0,
      "learning_rate": 1.9959442312266555e-05,
      "loss": 1.261,
      "step": 464
    },
    {
      "epoch": 0.05800174628913559,
      "grad_norm": 0.0,
      "learning_rate": 1.9959077999460094e-05,
      "loss": 1.2582,
      "step": 465
    },
    {
      "epoch": 0.058126481227391794,
      "grad_norm": 0.0,
      "learning_rate": 1.995871206108104e-05,
      "loss": 1.269,
      "step": 466
    },
    {
      "epoch": 0.058251216165648,
      "grad_norm": 0.0,
      "learning_rate": 1.9958344497189114e-05,
      "loss": 1.2559,
      "step": 467
    },
    {
      "epoch": 0.05837595110390421,
      "grad_norm": 0.0,
      "learning_rate": 1.995797530784432e-05,
      "loss": 1.2642,
      "step": 468
    },
    {
      "epoch": 0.05850068604216041,
      "grad_norm": 0.0,
      "learning_rate": 1.9957604493106916e-05,
      "loss": 1.3278,
      "step": 469
    },
    {
      "epoch": 0.058625420980416613,
      "grad_norm": 0.0,
      "learning_rate": 1.995723205303743e-05,
      "loss": 1.2647,
      "step": 470
    },
    {
      "epoch": 0.058750155918672824,
      "grad_norm": 0.0,
      "learning_rate": 1.995685798769665e-05,
      "loss": 1.2348,
      "step": 471
    },
    {
      "epoch": 0.05887489085692903,
      "grad_norm": 0.0,
      "learning_rate": 1.9956482297145638e-05,
      "loss": 1.2587,
      "step": 472
    },
    {
      "epoch": 0.05899962579518523,
      "grad_norm": 0.0,
      "learning_rate": 1.9956104981445713e-05,
      "loss": 1.2258,
      "step": 473
    },
    {
      "epoch": 0.05912436073344144,
      "grad_norm": 0.0,
      "learning_rate": 1.9955726040658463e-05,
      "loss": 1.2298,
      "step": 474
    },
    {
      "epoch": 0.059249095671697644,
      "grad_norm": 0.0,
      "learning_rate": 1.995534547484574e-05,
      "loss": 1.2215,
      "step": 475
    },
    {
      "epoch": 0.05937383060995385,
      "grad_norm": 0.0,
      "learning_rate": 1.9954963284069663e-05,
      "loss": 1.2773,
      "step": 476
    },
    {
      "epoch": 0.05949856554821006,
      "grad_norm": 0.0,
      "learning_rate": 1.9954579468392618e-05,
      "loss": 1.3185,
      "step": 477
    },
    {
      "epoch": 0.05962330048646626,
      "grad_norm": 0.0,
      "learning_rate": 1.9954194027877247e-05,
      "loss": 1.2587,
      "step": 478
    },
    {
      "epoch": 0.059748035424722463,
      "grad_norm": 0.0,
      "learning_rate": 1.9953806962586468e-05,
      "loss": 1.2447,
      "step": 479
    },
    {
      "epoch": 0.059872770362978674,
      "grad_norm": 0.0,
      "learning_rate": 1.9953418272583456e-05,
      "loss": 1.2195,
      "step": 480
    },
    {
      "epoch": 0.05999750530123488,
      "grad_norm": 0.0,
      "learning_rate": 1.9953027957931658e-05,
      "loss": 1.2139,
      "step": 481
    },
    {
      "epoch": 0.06012224023949108,
      "grad_norm": 0.0,
      "learning_rate": 1.9952636018694787e-05,
      "loss": 1.247,
      "step": 482
    },
    {
      "epoch": 0.06024697517774729,
      "grad_norm": 0.0,
      "learning_rate": 1.995224245493681e-05,
      "loss": 1.2773,
      "step": 483
    },
    {
      "epoch": 0.060371710116003494,
      "grad_norm": 0.0,
      "learning_rate": 1.995184726672197e-05,
      "loss": 1.2763,
      "step": 484
    },
    {
      "epoch": 0.0604964450542597,
      "grad_norm": 0.0,
      "learning_rate": 1.9951450454114773e-05,
      "loss": 1.2572,
      "step": 485
    },
    {
      "epoch": 0.06062117999251591,
      "grad_norm": 0.0,
      "learning_rate": 1.9951052017179985e-05,
      "loss": 1.2889,
      "step": 486
    },
    {
      "epoch": 0.06074591493077211,
      "grad_norm": 0.0,
      "learning_rate": 1.9950651955982646e-05,
      "loss": 1.2281,
      "step": 487
    },
    {
      "epoch": 0.060870649869028313,
      "grad_norm": 0.0,
      "learning_rate": 1.995025027058805e-05,
      "loss": 1.2619,
      "step": 488
    },
    {
      "epoch": 0.060995384807284524,
      "grad_norm": 0.0,
      "learning_rate": 1.994984696106177e-05,
      "loss": 1.2244,
      "step": 489
    },
    {
      "epoch": 0.06112011974554073,
      "grad_norm": 0.0,
      "learning_rate": 1.9949442027469628e-05,
      "loss": 1.2619,
      "step": 490
    },
    {
      "epoch": 0.06124485468379693,
      "grad_norm": 0.0,
      "learning_rate": 1.994903546987772e-05,
      "loss": 1.2383,
      "step": 491
    },
    {
      "epoch": 0.06136958962205314,
      "grad_norm": 0.0,
      "learning_rate": 1.9948627288352415e-05,
      "loss": 1.2615,
      "step": 492
    },
    {
      "epoch": 0.061494324560309344,
      "grad_norm": 0.0,
      "learning_rate": 1.9948217482960334e-05,
      "loss": 1.1972,
      "step": 493
    },
    {
      "epoch": 0.06161905949856555,
      "grad_norm": 0.0,
      "learning_rate": 1.9947806053768363e-05,
      "loss": 1.2298,
      "step": 494
    },
    {
      "epoch": 0.06174379443682176,
      "grad_norm": 0.0,
      "learning_rate": 1.994739300084366e-05,
      "loss": 1.2018,
      "step": 495
    },
    {
      "epoch": 0.06186852937507796,
      "grad_norm": 0.0,
      "learning_rate": 1.9946978324253652e-05,
      "loss": 1.2576,
      "step": 496
    },
    {
      "epoch": 0.061993264313334163,
      "grad_norm": 0.0,
      "learning_rate": 1.9946562024066018e-05,
      "loss": 1.2521,
      "step": 497
    },
    {
      "epoch": 0.062117999251590374,
      "grad_norm": 0.0,
      "learning_rate": 1.9946144100348707e-05,
      "loss": 1.2777,
      "step": 498
    },
    {
      "epoch": 0.06224273418984658,
      "grad_norm": 0.0,
      "learning_rate": 1.994572455316994e-05,
      "loss": 1.2727,
      "step": 499
    },
    {
      "epoch": 0.06236746912810278,
      "grad_norm": 0.0,
      "learning_rate": 1.9945303382598194e-05,
      "loss": 1.1839,
      "step": 500
    },
    {
      "epoch": 0.06249220406635899,
      "grad_norm": 0.0,
      "learning_rate": 1.994488058870222e-05,
      "loss": 1.2328,
      "step": 501
    },
    {
      "epoch": 0.0626169390046152,
      "grad_norm": 0.0,
      "learning_rate": 1.9944456171551024e-05,
      "loss": 1.2331,
      "step": 502
    },
    {
      "epoch": 0.0627416739428714,
      "grad_norm": 0.0,
      "learning_rate": 1.994403013121388e-05,
      "loss": 1.2455,
      "step": 503
    },
    {
      "epoch": 0.0628664088811276,
      "grad_norm": 0.0,
      "learning_rate": 1.9943602467760334e-05,
      "loss": 1.2375,
      "step": 504
    },
    {
      "epoch": 0.06299114381938381,
      "grad_norm": 0.0,
      "learning_rate": 1.9943173181260187e-05,
      "loss": 1.2629,
      "step": 505
    },
    {
      "epoch": 0.06311587875764002,
      "grad_norm": 0.0,
      "learning_rate": 1.9942742271783507e-05,
      "loss": 1.2928,
      "step": 506
    },
    {
      "epoch": 0.06324061369589622,
      "grad_norm": 0.0,
      "learning_rate": 1.994230973940064e-05,
      "loss": 1.2674,
      "step": 507
    },
    {
      "epoch": 0.06336534863415243,
      "grad_norm": 0.0,
      "learning_rate": 1.994187558418218e-05,
      "loss": 1.2476,
      "step": 508
    },
    {
      "epoch": 0.06349008357240864,
      "grad_norm": 0.0,
      "learning_rate": 1.9941439806198985e-05,
      "loss": 1.2244,
      "step": 509
    },
    {
      "epoch": 0.06361481851066483,
      "grad_norm": 0.0,
      "learning_rate": 1.994100240552219e-05,
      "loss": 1.2773,
      "step": 510
    },
    {
      "epoch": 0.06373955344892104,
      "grad_norm": 0.0,
      "learning_rate": 1.9940563382223196e-05,
      "loss": 1.2494,
      "step": 511
    },
    {
      "epoch": 0.06386428838717725,
      "grad_norm": 0.0,
      "learning_rate": 1.994012273637366e-05,
      "loss": 1.2431,
      "step": 512
    },
    {
      "epoch": 0.06398902332543345,
      "grad_norm": 0.0,
      "learning_rate": 1.99396804680455e-05,
      "loss": 1.2261,
      "step": 513
    },
    {
      "epoch": 0.06411375826368966,
      "grad_norm": 0.0,
      "learning_rate": 1.9939236577310913e-05,
      "loss": 1.2366,
      "step": 514
    },
    {
      "epoch": 0.06423849320194587,
      "grad_norm": 0.0,
      "learning_rate": 1.9938791064242348e-05,
      "loss": 1.2431,
      "step": 515
    },
    {
      "epoch": 0.06436322814020207,
      "grad_norm": 0.0,
      "learning_rate": 1.993834392891252e-05,
      "loss": 1.2,
      "step": 516
    },
    {
      "epoch": 0.06448796307845828,
      "grad_norm": 0.0,
      "learning_rate": 1.9937895171394427e-05,
      "loss": 1.2552,
      "step": 517
    },
    {
      "epoch": 0.06461269801671449,
      "grad_norm": 0.0,
      "learning_rate": 1.993744479176131e-05,
      "loss": 1.2407,
      "step": 518
    },
    {
      "epoch": 0.06473743295497068,
      "grad_norm": 0.0,
      "learning_rate": 1.9936992790086675e-05,
      "loss": 1.1943,
      "step": 519
    },
    {
      "epoch": 0.0648621678932269,
      "grad_norm": 0.0,
      "learning_rate": 1.993653916644431e-05,
      "loss": 1.2048,
      "step": 520
    },
    {
      "epoch": 0.0649869028314831,
      "grad_norm": 0.0,
      "learning_rate": 1.9936083920908254e-05,
      "loss": 1.2027,
      "step": 521
    },
    {
      "epoch": 0.0651116377697393,
      "grad_norm": 0.0,
      "learning_rate": 1.9935627053552814e-05,
      "loss": 1.239,
      "step": 522
    },
    {
      "epoch": 0.06523637270799551,
      "grad_norm": 0.0,
      "learning_rate": 1.9935168564452562e-05,
      "loss": 1.1902,
      "step": 523
    },
    {
      "epoch": 0.06536110764625172,
      "grad_norm": 0.0,
      "learning_rate": 1.993470845368234e-05,
      "loss": 1.2988,
      "step": 524
    },
    {
      "epoch": 0.06548584258450792,
      "grad_norm": 0.0,
      "learning_rate": 1.9934246721317244e-05,
      "loss": 1.274,
      "step": 525
    },
    {
      "epoch": 0.06561057752276413,
      "grad_norm": 0.0,
      "learning_rate": 1.993378336743264e-05,
      "loss": 1.2507,
      "step": 526
    },
    {
      "epoch": 0.06573531246102034,
      "grad_norm": 0.0,
      "learning_rate": 1.9933318392104166e-05,
      "loss": 1.1744,
      "step": 527
    },
    {
      "epoch": 0.06586004739927653,
      "grad_norm": 0.0,
      "learning_rate": 1.9932851795407705e-05,
      "loss": 1.2147,
      "step": 528
    },
    {
      "epoch": 0.06598478233753274,
      "grad_norm": 0.0,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 1.1765,
      "step": 529
    },
    {
      "epoch": 0.06610951727578895,
      "grad_norm": 0.0,
      "learning_rate": 1.9931913738215763e-05,
      "loss": 1.2337,
      "step": 530
    },
    {
      "epoch": 0.06623425221404515,
      "grad_norm": 0.0,
      "learning_rate": 1.993144227787339e-05,
      "loss": 1.2313,
      "step": 531
    },
    {
      "epoch": 0.06635898715230136,
      "grad_norm": 0.0,
      "learning_rate": 1.9930969196469266e-05,
      "loss": 1.2144,
      "step": 532
    },
    {
      "epoch": 0.06648372209055757,
      "grad_norm": 0.0,
      "learning_rate": 1.9930494494080612e-05,
      "loss": 1.2574,
      "step": 533
    },
    {
      "epoch": 0.06660845702881377,
      "grad_norm": 0.0,
      "learning_rate": 1.9930018170784912e-05,
      "loss": 1.2599,
      "step": 534
    },
    {
      "epoch": 0.06673319196706998,
      "grad_norm": 0.0,
      "learning_rate": 1.9929540226659907e-05,
      "loss": 1.26,
      "step": 535
    },
    {
      "epoch": 0.06685792690532619,
      "grad_norm": 0.0,
      "learning_rate": 1.992906066178362e-05,
      "loss": 1.2598,
      "step": 536
    },
    {
      "epoch": 0.06698266184358238,
      "grad_norm": 0.0,
      "learning_rate": 1.992857947623432e-05,
      "loss": 1.2049,
      "step": 537
    },
    {
      "epoch": 0.0671073967818386,
      "grad_norm": 0.0,
      "learning_rate": 1.9928096670090552e-05,
      "loss": 1.2517,
      "step": 538
    },
    {
      "epoch": 0.0672321317200948,
      "grad_norm": 0.0,
      "learning_rate": 1.992761224343112e-05,
      "loss": 1.2235,
      "step": 539
    },
    {
      "epoch": 0.067356866658351,
      "grad_norm": 0.0,
      "learning_rate": 1.99271261963351e-05,
      "loss": 1.263,
      "step": 540
    },
    {
      "epoch": 0.06748160159660721,
      "grad_norm": 0.0,
      "learning_rate": 1.992663852888182e-05,
      "loss": 1.1687,
      "step": 541
    },
    {
      "epoch": 0.06760633653486342,
      "grad_norm": 0.0,
      "learning_rate": 1.9926149241150886e-05,
      "loss": 1.1635,
      "step": 542
    },
    {
      "epoch": 0.06773107147311962,
      "grad_norm": 0.0,
      "learning_rate": 1.9925658333222155e-05,
      "loss": 1.2097,
      "step": 543
    },
    {
      "epoch": 0.06785580641137583,
      "grad_norm": 0.0,
      "learning_rate": 1.992516580517576e-05,
      "loss": 1.1664,
      "step": 544
    },
    {
      "epoch": 0.06798054134963204,
      "grad_norm": 0.0,
      "learning_rate": 1.9924671657092097e-05,
      "loss": 1.222,
      "step": 545
    },
    {
      "epoch": 0.06810527628788823,
      "grad_norm": 0.0,
      "learning_rate": 1.992417588905182e-05,
      "loss": 1.1945,
      "step": 546
    },
    {
      "epoch": 0.06823001122614444,
      "grad_norm": 0.0,
      "learning_rate": 1.992367850113585e-05,
      "loss": 1.2369,
      "step": 547
    },
    {
      "epoch": 0.06835474616440065,
      "grad_norm": 0.0,
      "learning_rate": 1.992317949342537e-05,
      "loss": 1.2869,
      "step": 548
    },
    {
      "epoch": 0.06847948110265685,
      "grad_norm": 0.0,
      "learning_rate": 1.992267886600184e-05,
      "loss": 1.2296,
      "step": 549
    },
    {
      "epoch": 0.06860421604091306,
      "grad_norm": 0.0,
      "learning_rate": 1.9922176618946966e-05,
      "loss": 1.2031,
      "step": 550
    },
    {
      "epoch": 0.06872895097916927,
      "grad_norm": 0.0,
      "learning_rate": 1.9921672752342737e-05,
      "loss": 1.1726,
      "step": 551
    },
    {
      "epoch": 0.06885368591742547,
      "grad_norm": 0.0,
      "learning_rate": 1.9921167266271385e-05,
      "loss": 1.2414,
      "step": 552
    },
    {
      "epoch": 0.06897842085568168,
      "grad_norm": 0.0,
      "learning_rate": 1.9920660160815423e-05,
      "loss": 1.2369,
      "step": 553
    },
    {
      "epoch": 0.06910315579393789,
      "grad_norm": 0.0,
      "learning_rate": 1.992015143605763e-05,
      "loss": 1.2369,
      "step": 554
    },
    {
      "epoch": 0.06922789073219408,
      "grad_norm": 0.0,
      "learning_rate": 1.9919641092081033e-05,
      "loss": 1.1799,
      "step": 555
    },
    {
      "epoch": 0.0693526256704503,
      "grad_norm": 0.0,
      "learning_rate": 1.9919129128968938e-05,
      "loss": 1.203,
      "step": 556
    },
    {
      "epoch": 0.0694773606087065,
      "grad_norm": 0.0,
      "learning_rate": 1.9918615546804908e-05,
      "loss": 1.1923,
      "step": 557
    },
    {
      "epoch": 0.0696020955469627,
      "grad_norm": 0.0,
      "learning_rate": 1.991810034567278e-05,
      "loss": 1.2296,
      "step": 558
    },
    {
      "epoch": 0.06972683048521891,
      "grad_norm": 0.0,
      "learning_rate": 1.9917583525656633e-05,
      "loss": 1.2398,
      "step": 559
    },
    {
      "epoch": 0.06985156542347512,
      "grad_norm": 0.0,
      "learning_rate": 1.991706508684084e-05,
      "loss": 1.2308,
      "step": 560
    },
    {
      "epoch": 0.06997630036173132,
      "grad_norm": 0.0,
      "learning_rate": 1.991654502931001e-05,
      "loss": 1.2459,
      "step": 561
    },
    {
      "epoch": 0.07010103529998753,
      "grad_norm": 0.0,
      "learning_rate": 1.9916023353149042e-05,
      "loss": 1.2299,
      "step": 562
    },
    {
      "epoch": 0.07022577023824374,
      "grad_norm": 0.0,
      "learning_rate": 1.9915500058443084e-05,
      "loss": 1.2391,
      "step": 563
    },
    {
      "epoch": 0.07035050517649993,
      "grad_norm": 0.0,
      "learning_rate": 1.9914975145277545e-05,
      "loss": 1.2303,
      "step": 564
    },
    {
      "epoch": 0.07047524011475614,
      "grad_norm": 0.0,
      "learning_rate": 1.9914448613738107e-05,
      "loss": 1.1888,
      "step": 565
    },
    {
      "epoch": 0.07059997505301235,
      "grad_norm": 0.0,
      "learning_rate": 1.9913920463910713e-05,
      "loss": 1.2513,
      "step": 566
    },
    {
      "epoch": 0.07072470999126855,
      "grad_norm": 0.0,
      "learning_rate": 1.9913390695881572e-05,
      "loss": 1.2361,
      "step": 567
    },
    {
      "epoch": 0.07084944492952476,
      "grad_norm": 0.0,
      "learning_rate": 1.991285930973715e-05,
      "loss": 1.231,
      "step": 568
    },
    {
      "epoch": 0.07097417986778097,
      "grad_norm": 0.0,
      "learning_rate": 1.991232630556419e-05,
      "loss": 1.2686,
      "step": 569
    },
    {
      "epoch": 0.07109891480603717,
      "grad_norm": 0.0,
      "learning_rate": 1.991179168344969e-05,
      "loss": 1.176,
      "step": 570
    },
    {
      "epoch": 0.07122364974429338,
      "grad_norm": 0.0,
      "learning_rate": 1.991125544348091e-05,
      "loss": 1.2385,
      "step": 571
    },
    {
      "epoch": 0.07134838468254959,
      "grad_norm": 0.0,
      "learning_rate": 1.9910717585745383e-05,
      "loss": 1.2343,
      "step": 572
    },
    {
      "epoch": 0.07147311962080578,
      "grad_norm": 0.0,
      "learning_rate": 1.9910178110330894e-05,
      "loss": 1.2128,
      "step": 573
    },
    {
      "epoch": 0.071597854559062,
      "grad_norm": 0.0,
      "learning_rate": 1.9909637017325508e-05,
      "loss": 1.2366,
      "step": 574
    },
    {
      "epoch": 0.0717225894973182,
      "grad_norm": 0.0,
      "learning_rate": 1.990909430681754e-05,
      "loss": 1.23,
      "step": 575
    },
    {
      "epoch": 0.0718473244355744,
      "grad_norm": 0.0,
      "learning_rate": 1.9908549978895572e-05,
      "loss": 1.1457,
      "step": 576
    },
    {
      "epoch": 0.07197205937383061,
      "grad_norm": 0.0,
      "learning_rate": 1.9908004033648452e-05,
      "loss": 1.1935,
      "step": 577
    },
    {
      "epoch": 0.07209679431208682,
      "grad_norm": 0.0,
      "learning_rate": 1.9907456471165297e-05,
      "loss": 1.1779,
      "step": 578
    },
    {
      "epoch": 0.07222152925034302,
      "grad_norm": 0.0,
      "learning_rate": 1.9906907291535483e-05,
      "loss": 1.2126,
      "step": 579
    },
    {
      "epoch": 0.07234626418859923,
      "grad_norm": 0.0,
      "learning_rate": 1.9906356494848644e-05,
      "loss": 1.1645,
      "step": 580
    },
    {
      "epoch": 0.07247099912685544,
      "grad_norm": 0.0,
      "learning_rate": 1.990580408119469e-05,
      "loss": 1.1953,
      "step": 581
    },
    {
      "epoch": 0.07259573406511163,
      "grad_norm": 0.0,
      "learning_rate": 1.9905250050663786e-05,
      "loss": 1.1985,
      "step": 582
    },
    {
      "epoch": 0.07272046900336784,
      "grad_norm": 0.0,
      "learning_rate": 1.9904694403346362e-05,
      "loss": 1.1872,
      "step": 583
    },
    {
      "epoch": 0.07284520394162405,
      "grad_norm": 0.0,
      "learning_rate": 1.9904137139333114e-05,
      "loss": 1.207,
      "step": 584
    },
    {
      "epoch": 0.07296993887988025,
      "grad_norm": 0.0,
      "learning_rate": 1.9903578258715003e-05,
      "loss": 1.1853,
      "step": 585
    },
    {
      "epoch": 0.07309467381813646,
      "grad_norm": 0.0,
      "learning_rate": 1.9903017761583253e-05,
      "loss": 1.2363,
      "step": 586
    },
    {
      "epoch": 0.07321940875639267,
      "grad_norm": 0.0,
      "learning_rate": 1.990245564802935e-05,
      "loss": 1.1988,
      "step": 587
    },
    {
      "epoch": 0.07334414369464887,
      "grad_norm": 0.0,
      "learning_rate": 1.9901891918145048e-05,
      "loss": 1.3107,
      "step": 588
    },
    {
      "epoch": 0.07346887863290508,
      "grad_norm": 0.0,
      "learning_rate": 1.990132657202236e-05,
      "loss": 1.2249,
      "step": 589
    },
    {
      "epoch": 0.07359361357116129,
      "grad_norm": 0.0,
      "learning_rate": 1.9900759609753562e-05,
      "loss": 1.2184,
      "step": 590
    },
    {
      "epoch": 0.07371834850941748,
      "grad_norm": 0.0,
      "learning_rate": 1.9900191031431202e-05,
      "loss": 1.2606,
      "step": 591
    },
    {
      "epoch": 0.0738430834476737,
      "grad_norm": 0.0,
      "learning_rate": 1.989962083714808e-05,
      "loss": 1.2549,
      "step": 592
    },
    {
      "epoch": 0.0739678183859299,
      "grad_norm": 0.0,
      "learning_rate": 1.9899049026997272e-05,
      "loss": 1.249,
      "step": 593
    },
    {
      "epoch": 0.0740925533241861,
      "grad_norm": 0.0,
      "learning_rate": 1.9898475601072107e-05,
      "loss": 1.2637,
      "step": 594
    },
    {
      "epoch": 0.07421728826244231,
      "grad_norm": 0.0,
      "learning_rate": 1.9897900559466188e-05,
      "loss": 1.2268,
      "step": 595
    },
    {
      "epoch": 0.07434202320069852,
      "grad_norm": 0.0,
      "learning_rate": 1.989732390227337e-05,
      "loss": 1.2541,
      "step": 596
    },
    {
      "epoch": 0.07446675813895472,
      "grad_norm": 0.0,
      "learning_rate": 1.989674562958778e-05,
      "loss": 1.1697,
      "step": 597
    },
    {
      "epoch": 0.07459149307721093,
      "grad_norm": 0.0,
      "learning_rate": 1.989616574150381e-05,
      "loss": 1.2124,
      "step": 598
    },
    {
      "epoch": 0.07471622801546714,
      "grad_norm": 0.0,
      "learning_rate": 1.989558423811611e-05,
      "loss": 1.2001,
      "step": 599
    },
    {
      "epoch": 0.07484096295372333,
      "grad_norm": 0.0,
      "learning_rate": 1.9895001119519596e-05,
      "loss": 1.2385,
      "step": 600
    },
    {
      "epoch": 0.07496569789197954,
      "grad_norm": 0.0,
      "learning_rate": 1.9894416385809444e-05,
      "loss": 1.1965,
      "step": 601
    },
    {
      "epoch": 0.07509043283023575,
      "grad_norm": 0.0,
      "learning_rate": 1.9893830037081108e-05,
      "loss": 1.2083,
      "step": 602
    },
    {
      "epoch": 0.07521516776849195,
      "grad_norm": 0.0,
      "learning_rate": 1.9893242073430282e-05,
      "loss": 1.214,
      "step": 603
    },
    {
      "epoch": 0.07533990270674816,
      "grad_norm": 0.0,
      "learning_rate": 1.9892652494952948e-05,
      "loss": 1.2242,
      "step": 604
    },
    {
      "epoch": 0.07546463764500437,
      "grad_norm": 0.0,
      "learning_rate": 1.989206130174533e-05,
      "loss": 1.1852,
      "step": 605
    },
    {
      "epoch": 0.07558937258326057,
      "grad_norm": 0.0,
      "learning_rate": 1.9891468493903932e-05,
      "loss": 1.2327,
      "step": 606
    },
    {
      "epoch": 0.07571410752151678,
      "grad_norm": 0.0,
      "learning_rate": 1.989087407152551e-05,
      "loss": 1.2151,
      "step": 607
    },
    {
      "epoch": 0.07583884245977299,
      "grad_norm": 0.0,
      "learning_rate": 1.9890278034707095e-05,
      "loss": 1.2453,
      "step": 608
    },
    {
      "epoch": 0.07596357739802918,
      "grad_norm": 0.0,
      "learning_rate": 1.9889680383545974e-05,
      "loss": 1.2363,
      "step": 609
    },
    {
      "epoch": 0.0760883123362854,
      "grad_norm": 0.0,
      "learning_rate": 1.9889081118139694e-05,
      "loss": 1.2146,
      "step": 610
    },
    {
      "epoch": 0.0762130472745416,
      "grad_norm": 0.0,
      "learning_rate": 1.988848023858607e-05,
      "loss": 1.1613,
      "step": 611
    },
    {
      "epoch": 0.0763377822127978,
      "grad_norm": 0.0,
      "learning_rate": 1.9887877744983194e-05,
      "loss": 1.2213,
      "step": 612
    },
    {
      "epoch": 0.07646251715105401,
      "grad_norm": 0.0,
      "learning_rate": 1.988727363742939e-05,
      "loss": 1.1848,
      "step": 613
    },
    {
      "epoch": 0.07658725208931022,
      "grad_norm": 0.0,
      "learning_rate": 1.9886667916023273e-05,
      "loss": 1.19,
      "step": 614
    },
    {
      "epoch": 0.07671198702756642,
      "grad_norm": 0.0,
      "learning_rate": 1.9886060580863713e-05,
      "loss": 1.2294,
      "step": 615
    },
    {
      "epoch": 0.07683672196582263,
      "grad_norm": 0.0,
      "learning_rate": 1.9885451632049836e-05,
      "loss": 1.207,
      "step": 616
    },
    {
      "epoch": 0.07696145690407884,
      "grad_norm": 0.0,
      "learning_rate": 1.9884841069681045e-05,
      "loss": 1.2822,
      "step": 617
    },
    {
      "epoch": 0.07708619184233503,
      "grad_norm": 0.0,
      "learning_rate": 1.9884228893856995e-05,
      "loss": 1.1977,
      "step": 618
    },
    {
      "epoch": 0.07721092678059124,
      "grad_norm": 0.0,
      "learning_rate": 1.988361510467761e-05,
      "loss": 1.2163,
      "step": 619
    },
    {
      "epoch": 0.07733566171884745,
      "grad_norm": 0.0,
      "learning_rate": 1.9882999702243073e-05,
      "loss": 1.2084,
      "step": 620
    },
    {
      "epoch": 0.07746039665710365,
      "grad_norm": 0.0,
      "learning_rate": 1.9882382686653837e-05,
      "loss": 1.2591,
      "step": 621
    },
    {
      "epoch": 0.07758513159535986,
      "grad_norm": 0.0,
      "learning_rate": 1.9881764058010615e-05,
      "loss": 1.1577,
      "step": 622
    },
    {
      "epoch": 0.07770986653361607,
      "grad_norm": 0.0,
      "learning_rate": 1.988114381641438e-05,
      "loss": 1.1819,
      "step": 623
    },
    {
      "epoch": 0.07783460147187227,
      "grad_norm": 0.0,
      "learning_rate": 1.988052196196637e-05,
      "loss": 1.2627,
      "step": 624
    },
    {
      "epoch": 0.07795933641012848,
      "grad_norm": 0.0,
      "learning_rate": 1.9879898494768093e-05,
      "loss": 1.2303,
      "step": 625
    },
    {
      "epoch": 0.07808407134838469,
      "grad_norm": 0.0,
      "learning_rate": 1.9879273414921306e-05,
      "loss": 1.1792,
      "step": 626
    },
    {
      "epoch": 0.07820880628664088,
      "grad_norm": 0.0,
      "learning_rate": 1.9878646722528044e-05,
      "loss": 1.2153,
      "step": 627
    },
    {
      "epoch": 0.0783335412248971,
      "grad_norm": 0.0,
      "learning_rate": 1.9878018417690602e-05,
      "loss": 1.1952,
      "step": 628
    },
    {
      "epoch": 0.0784582761631533,
      "grad_norm": 0.0,
      "learning_rate": 1.9877388500511526e-05,
      "loss": 1.2,
      "step": 629
    },
    {
      "epoch": 0.0785830111014095,
      "grad_norm": 0.0,
      "learning_rate": 1.987675697109364e-05,
      "loss": 1.1473,
      "step": 630
    },
    {
      "epoch": 0.07870774603966571,
      "grad_norm": 0.0,
      "learning_rate": 1.9876123829540027e-05,
      "loss": 1.2223,
      "step": 631
    },
    {
      "epoch": 0.07883248097792192,
      "grad_norm": 0.0,
      "learning_rate": 1.987548907595403e-05,
      "loss": 1.1847,
      "step": 632
    },
    {
      "epoch": 0.07895721591617812,
      "grad_norm": 0.0,
      "learning_rate": 1.9874852710439252e-05,
      "loss": 1.1965,
      "step": 633
    },
    {
      "epoch": 0.07908195085443433,
      "grad_norm": 0.0,
      "learning_rate": 1.987421473309957e-05,
      "loss": 1.1344,
      "step": 634
    },
    {
      "epoch": 0.07920668579269054,
      "grad_norm": 0.0,
      "learning_rate": 1.9873575144039115e-05,
      "loss": 1.2201,
      "step": 635
    },
    {
      "epoch": 0.07933142073094673,
      "grad_norm": 0.0,
      "learning_rate": 1.9872933943362285e-05,
      "loss": 1.1869,
      "step": 636
    },
    {
      "epoch": 0.07945615566920294,
      "grad_norm": 0.0,
      "learning_rate": 1.9872291131173743e-05,
      "loss": 1.1909,
      "step": 637
    },
    {
      "epoch": 0.07958089060745915,
      "grad_norm": 0.0,
      "learning_rate": 1.9871646707578406e-05,
      "loss": 1.2021,
      "step": 638
    },
    {
      "epoch": 0.07970562554571535,
      "grad_norm": 0.0,
      "learning_rate": 1.9871000672681468e-05,
      "loss": 1.1743,
      "step": 639
    },
    {
      "epoch": 0.07983036048397156,
      "grad_norm": 0.0,
      "learning_rate": 1.9870353026588368e-05,
      "loss": 1.2263,
      "step": 640
    },
    {
      "epoch": 0.07995509542222777,
      "grad_norm": 0.0,
      "learning_rate": 1.986970376940483e-05,
      "loss": 1.2064,
      "step": 641
    },
    {
      "epoch": 0.08007983036048397,
      "grad_norm": 0.0,
      "learning_rate": 1.986905290123682e-05,
      "loss": 1.1981,
      "step": 642
    },
    {
      "epoch": 0.08020456529874018,
      "grad_norm": 0.0,
      "learning_rate": 1.9868400422190577e-05,
      "loss": 1.1763,
      "step": 643
    },
    {
      "epoch": 0.08032930023699639,
      "grad_norm": 0.0,
      "learning_rate": 1.9867746332372607e-05,
      "loss": 1.1863,
      "step": 644
    },
    {
      "epoch": 0.08045403517525258,
      "grad_norm": 0.0,
      "learning_rate": 1.9867090631889666e-05,
      "loss": 1.1505,
      "step": 645
    },
    {
      "epoch": 0.0805787701135088,
      "grad_norm": 0.0,
      "learning_rate": 1.9866433320848793e-05,
      "loss": 1.1888,
      "step": 646
    },
    {
      "epoch": 0.080703505051765,
      "grad_norm": 0.0,
      "learning_rate": 1.986577439935727e-05,
      "loss": 1.2263,
      "step": 647
    },
    {
      "epoch": 0.0808282399900212,
      "grad_norm": 0.0,
      "learning_rate": 1.9865113867522643e-05,
      "loss": 1.1926,
      "step": 648
    },
    {
      "epoch": 0.08095297492827741,
      "grad_norm": 0.0,
      "learning_rate": 1.986445172545274e-05,
      "loss": 1.2014,
      "step": 649
    },
    {
      "epoch": 0.08107770986653362,
      "grad_norm": 0.0,
      "learning_rate": 1.9863787973255634e-05,
      "loss": 1.1607,
      "step": 650
    },
    {
      "epoch": 0.08120244480478982,
      "grad_norm": 0.0,
      "learning_rate": 1.9863122611039664e-05,
      "loss": 1.1933,
      "step": 651
    },
    {
      "epoch": 0.08132717974304603,
      "grad_norm": 0.0,
      "learning_rate": 1.9862455638913437e-05,
      "loss": 1.2343,
      "step": 652
    },
    {
      "epoch": 0.08145191468130224,
      "grad_norm": 0.0,
      "learning_rate": 1.9861787056985823e-05,
      "loss": 1.2011,
      "step": 653
    },
    {
      "epoch": 0.08157664961955843,
      "grad_norm": 0.0,
      "learning_rate": 1.986111686536594e-05,
      "loss": 1.1979,
      "step": 654
    },
    {
      "epoch": 0.08170138455781464,
      "grad_norm": 0.0,
      "learning_rate": 1.9860445064163193e-05,
      "loss": 1.2124,
      "step": 655
    },
    {
      "epoch": 0.08182611949607085,
      "grad_norm": 0.0,
      "learning_rate": 1.985977165348723e-05,
      "loss": 1.1718,
      "step": 656
    },
    {
      "epoch": 0.08195085443432705,
      "grad_norm": 0.0,
      "learning_rate": 1.9859096633447965e-05,
      "loss": 1.1924,
      "step": 657
    },
    {
      "epoch": 0.08207558937258326,
      "grad_norm": 0.0,
      "learning_rate": 1.985842000415559e-05,
      "loss": 1.1591,
      "step": 658
    },
    {
      "epoch": 0.08220032431083947,
      "grad_norm": 0.0,
      "learning_rate": 1.9857741765720536e-05,
      "loss": 1.1977,
      "step": 659
    },
    {
      "epoch": 0.08232505924909567,
      "grad_norm": 0.0,
      "learning_rate": 1.9857061918253514e-05,
      "loss": 1.2072,
      "step": 660
    },
    {
      "epoch": 0.08244979418735188,
      "grad_norm": 0.0,
      "learning_rate": 1.985638046186549e-05,
      "loss": 1.2318,
      "step": 661
    },
    {
      "epoch": 0.08257452912560809,
      "grad_norm": 0.0,
      "learning_rate": 1.98556973966677e-05,
      "loss": 1.1452,
      "step": 662
    },
    {
      "epoch": 0.08269926406386428,
      "grad_norm": 0.0,
      "learning_rate": 1.9855012722771637e-05,
      "loss": 1.1579,
      "step": 663
    },
    {
      "epoch": 0.0828239990021205,
      "grad_norm": 0.0,
      "learning_rate": 1.985432644028905e-05,
      "loss": 1.2145,
      "step": 664
    },
    {
      "epoch": 0.0829487339403767,
      "grad_norm": 0.0,
      "learning_rate": 1.9853638549331962e-05,
      "loss": 1.1775,
      "step": 665
    },
    {
      "epoch": 0.0830734688786329,
      "grad_norm": 0.0,
      "learning_rate": 1.9852949050012655e-05,
      "loss": 1.1882,
      "step": 666
    },
    {
      "epoch": 0.08319820381688911,
      "grad_norm": 0.0,
      "learning_rate": 1.9852257942443672e-05,
      "loss": 1.1849,
      "step": 667
    },
    {
      "epoch": 0.08332293875514532,
      "grad_norm": 0.0,
      "learning_rate": 1.9851565226737817e-05,
      "loss": 1.2077,
      "step": 668
    },
    {
      "epoch": 0.08344767369340152,
      "grad_norm": 0.0,
      "learning_rate": 1.9850870903008162e-05,
      "loss": 1.192,
      "step": 669
    },
    {
      "epoch": 0.08357240863165773,
      "grad_norm": 0.0,
      "learning_rate": 1.9850174971368033e-05,
      "loss": 1.2045,
      "step": 670
    },
    {
      "epoch": 0.08369714356991394,
      "grad_norm": 0.0,
      "learning_rate": 1.9849477431931032e-05,
      "loss": 1.2484,
      "step": 671
    },
    {
      "epoch": 0.08382187850817013,
      "grad_norm": 0.0,
      "learning_rate": 1.984877828481101e-05,
      "loss": 1.2075,
      "step": 672
    },
    {
      "epoch": 0.08394661344642634,
      "grad_norm": 0.0,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 1.1991,
      "step": 673
    },
    {
      "epoch": 0.08407134838468255,
      "grad_norm": 0.0,
      "learning_rate": 1.984737516797863e-05,
      "loss": 1.1903,
      "step": 674
    },
    {
      "epoch": 0.08419608332293875,
      "grad_norm": 0.0,
      "learning_rate": 1.984667119849531e-05,
      "loss": 1.1947,
      "step": 675
    },
    {
      "epoch": 0.08432081826119496,
      "grad_norm": 0.0,
      "learning_rate": 1.984596562178701e-05,
      "loss": 1.1864,
      "step": 676
    },
    {
      "epoch": 0.08444555319945117,
      "grad_norm": 0.0,
      "learning_rate": 1.9845258437968905e-05,
      "loss": 1.1992,
      "step": 677
    },
    {
      "epoch": 0.08457028813770737,
      "grad_norm": 0.0,
      "learning_rate": 1.9844549647156428e-05,
      "loss": 1.131,
      "step": 678
    },
    {
      "epoch": 0.08469502307596358,
      "grad_norm": 0.0,
      "learning_rate": 1.984383924946527e-05,
      "loss": 1.198,
      "step": 679
    },
    {
      "epoch": 0.08481975801421979,
      "grad_norm": 0.0,
      "learning_rate": 1.984312724501138e-05,
      "loss": 1.2231,
      "step": 680
    },
    {
      "epoch": 0.08494449295247598,
      "grad_norm": 0.0,
      "learning_rate": 1.984241363391098e-05,
      "loss": 1.1506,
      "step": 681
    },
    {
      "epoch": 0.0850692278907322,
      "grad_norm": 0.0,
      "learning_rate": 1.9841698416280554e-05,
      "loss": 1.1846,
      "step": 682
    },
    {
      "epoch": 0.0851939628289884,
      "grad_norm": 0.0,
      "learning_rate": 1.9840981592236838e-05,
      "loss": 1.2164,
      "step": 683
    },
    {
      "epoch": 0.0853186977672446,
      "grad_norm": 0.0,
      "learning_rate": 1.9840263161896832e-05,
      "loss": 1.2046,
      "step": 684
    },
    {
      "epoch": 0.08544343270550081,
      "grad_norm": 0.0,
      "learning_rate": 1.983954312537781e-05,
      "loss": 1.1409,
      "step": 685
    },
    {
      "epoch": 0.08556816764375702,
      "grad_norm": 0.0,
      "learning_rate": 1.9838821482797295e-05,
      "loss": 1.2094,
      "step": 686
    },
    {
      "epoch": 0.08569290258201322,
      "grad_norm": 0.0,
      "learning_rate": 1.9838098234273078e-05,
      "loss": 1.1367,
      "step": 687
    },
    {
      "epoch": 0.08581763752026943,
      "grad_norm": 0.0,
      "learning_rate": 1.9837373379923212e-05,
      "loss": 1.1882,
      "step": 688
    },
    {
      "epoch": 0.08594237245852564,
      "grad_norm": 0.0,
      "learning_rate": 1.9836646919866012e-05,
      "loss": 1.1634,
      "step": 689
    },
    {
      "epoch": 0.08606710739678183,
      "grad_norm": 0.0,
      "learning_rate": 1.9835918854220056e-05,
      "loss": 1.169,
      "step": 690
    },
    {
      "epoch": 0.08619184233503804,
      "grad_norm": 0.0,
      "learning_rate": 1.983518918310418e-05,
      "loss": 1.1628,
      "step": 691
    },
    {
      "epoch": 0.08631657727329425,
      "grad_norm": 0.0,
      "learning_rate": 1.9834457906637484e-05,
      "loss": 1.1784,
      "step": 692
    },
    {
      "epoch": 0.08644131221155045,
      "grad_norm": 0.0,
      "learning_rate": 1.9833725024939335e-05,
      "loss": 1.1805,
      "step": 693
    },
    {
      "epoch": 0.08656604714980666,
      "grad_norm": 0.0,
      "learning_rate": 1.9832990538129353e-05,
      "loss": 1.2038,
      "step": 694
    },
    {
      "epoch": 0.08669078208806287,
      "grad_norm": 0.0,
      "learning_rate": 1.9832254446327427e-05,
      "loss": 1.1686,
      "step": 695
    },
    {
      "epoch": 0.08681551702631907,
      "grad_norm": 0.0,
      "learning_rate": 1.9831516749653708e-05,
      "loss": 1.1612,
      "step": 696
    },
    {
      "epoch": 0.08694025196457528,
      "grad_norm": 0.0,
      "learning_rate": 1.9830777448228603e-05,
      "loss": 1.1744,
      "step": 697
    },
    {
      "epoch": 0.08706498690283149,
      "grad_norm": 0.0,
      "learning_rate": 1.9830036542172783e-05,
      "loss": 1.1868,
      "step": 698
    },
    {
      "epoch": 0.08718972184108768,
      "grad_norm": 0.0,
      "learning_rate": 1.982929403160719e-05,
      "loss": 1.207,
      "step": 699
    },
    {
      "epoch": 0.0873144567793439,
      "grad_norm": 0.0,
      "learning_rate": 1.9828549916653013e-05,
      "loss": 1.189,
      "step": 700
    },
    {
      "epoch": 0.0874391917176001,
      "grad_norm": 0.0,
      "learning_rate": 1.9827804197431716e-05,
      "loss": 1.2008,
      "step": 701
    },
    {
      "epoch": 0.0875639266558563,
      "grad_norm": 0.0,
      "learning_rate": 1.9827056874065013e-05,
      "loss": 1.2109,
      "step": 702
    },
    {
      "epoch": 0.08768866159411251,
      "grad_norm": 0.0,
      "learning_rate": 1.982630794667489e-05,
      "loss": 1.1947,
      "step": 703
    },
    {
      "epoch": 0.08781339653236872,
      "grad_norm": 0.0,
      "learning_rate": 1.9825557415383593e-05,
      "loss": 1.1593,
      "step": 704
    },
    {
      "epoch": 0.08793813147062492,
      "grad_norm": 0.0,
      "learning_rate": 1.9824805280313625e-05,
      "loss": 1.2243,
      "step": 705
    },
    {
      "epoch": 0.08806286640888113,
      "grad_norm": 0.0,
      "learning_rate": 1.982405154158775e-05,
      "loss": 1.177,
      "step": 706
    },
    {
      "epoch": 0.08818760134713734,
      "grad_norm": 0.0,
      "learning_rate": 1.9823296199329004e-05,
      "loss": 1.1767,
      "step": 707
    },
    {
      "epoch": 0.08831233628539353,
      "grad_norm": 0.0,
      "learning_rate": 1.982253925366067e-05,
      "loss": 1.2191,
      "step": 708
    },
    {
      "epoch": 0.08843707122364974,
      "grad_norm": 0.0,
      "learning_rate": 1.982178070470631e-05,
      "loss": 1.178,
      "step": 709
    },
    {
      "epoch": 0.08856180616190595,
      "grad_norm": 0.0,
      "learning_rate": 1.982102055258973e-05,
      "loss": 1.1719,
      "step": 710
    },
    {
      "epoch": 0.08868654110016215,
      "grad_norm": 0.0,
      "learning_rate": 1.982025879743501e-05,
      "loss": 1.2379,
      "step": 711
    },
    {
      "epoch": 0.08881127603841836,
      "grad_norm": 0.0,
      "learning_rate": 1.9819495439366488e-05,
      "loss": 1.165,
      "step": 712
    },
    {
      "epoch": 0.08893601097667457,
      "grad_norm": 0.0,
      "learning_rate": 1.9818730478508762e-05,
      "loss": 1.2231,
      "step": 713
    },
    {
      "epoch": 0.08906074591493077,
      "grad_norm": 0.0,
      "learning_rate": 1.9817963914986692e-05,
      "loss": 1.2211,
      "step": 714
    },
    {
      "epoch": 0.08918548085318698,
      "grad_norm": 0.0,
      "learning_rate": 1.98171957489254e-05,
      "loss": 1.2188,
      "step": 715
    },
    {
      "epoch": 0.08931021579144319,
      "grad_norm": 0.0,
      "learning_rate": 1.9816425980450273e-05,
      "loss": 1.1769,
      "step": 716
    },
    {
      "epoch": 0.08943495072969938,
      "grad_norm": 0.0,
      "learning_rate": 1.9815654609686955e-05,
      "loss": 1.2183,
      "step": 717
    },
    {
      "epoch": 0.0895596856679556,
      "grad_norm": 0.0,
      "learning_rate": 1.9814881636761355e-05,
      "loss": 1.1766,
      "step": 718
    },
    {
      "epoch": 0.0896844206062118,
      "grad_norm": 0.0,
      "learning_rate": 1.9814107061799642e-05,
      "loss": 1.2082,
      "step": 719
    },
    {
      "epoch": 0.089809155544468,
      "grad_norm": 0.0,
      "learning_rate": 1.981333088492824e-05,
      "loss": 1.1694,
      "step": 720
    },
    {
      "epoch": 0.08993389048272421,
      "grad_norm": 0.0,
      "learning_rate": 1.9812553106273848e-05,
      "loss": 1.1217,
      "step": 721
    },
    {
      "epoch": 0.09005862542098042,
      "grad_norm": 0.0,
      "learning_rate": 1.9811773725963416e-05,
      "loss": 1.2319,
      "step": 722
    },
    {
      "epoch": 0.09018336035923662,
      "grad_norm": 0.0,
      "learning_rate": 1.9810992744124155e-05,
      "loss": 1.2042,
      "step": 723
    },
    {
      "epoch": 0.09030809529749283,
      "grad_norm": 0.0,
      "learning_rate": 1.981021016088355e-05,
      "loss": 1.1952,
      "step": 724
    },
    {
      "epoch": 0.09043283023574904,
      "grad_norm": 0.0,
      "learning_rate": 1.9809425976369327e-05,
      "loss": 1.1862,
      "step": 725
    },
    {
      "epoch": 0.09055756517400523,
      "grad_norm": 0.0,
      "learning_rate": 1.9808640190709493e-05,
      "loss": 1.2151,
      "step": 726
    },
    {
      "epoch": 0.09068230011226144,
      "grad_norm": 0.0,
      "learning_rate": 1.9807852804032306e-05,
      "loss": 1.1803,
      "step": 727
    },
    {
      "epoch": 0.09080703505051765,
      "grad_norm": 0.0,
      "learning_rate": 1.9807063816466287e-05,
      "loss": 1.1787,
      "step": 728
    },
    {
      "epoch": 0.09093176998877385,
      "grad_norm": 0.0,
      "learning_rate": 1.9806273228140216e-05,
      "loss": 1.1207,
      "step": 729
    },
    {
      "epoch": 0.09105650492703006,
      "grad_norm": 0.0,
      "learning_rate": 1.9805481039183144e-05,
      "loss": 1.1974,
      "step": 730
    },
    {
      "epoch": 0.09118123986528627,
      "grad_norm": 0.0,
      "learning_rate": 1.9804687249724366e-05,
      "loss": 1.1583,
      "step": 731
    },
    {
      "epoch": 0.09130597480354247,
      "grad_norm": 0.0,
      "learning_rate": 1.980389185989346e-05,
      "loss": 1.1604,
      "step": 732
    },
    {
      "epoch": 0.09143070974179868,
      "grad_norm": 0.0,
      "learning_rate": 1.980309486982024e-05,
      "loss": 1.1466,
      "step": 733
    },
    {
      "epoch": 0.09155544468005489,
      "grad_norm": 0.0,
      "learning_rate": 1.980229627963481e-05,
      "loss": 1.1974,
      "step": 734
    },
    {
      "epoch": 0.09168017961831108,
      "grad_norm": 0.0,
      "learning_rate": 1.980149608946751e-05,
      "loss": 1.1596,
      "step": 735
    },
    {
      "epoch": 0.0918049145565673,
      "grad_norm": 0.0,
      "learning_rate": 1.980069429944895e-05,
      "loss": 1.1762,
      "step": 736
    },
    {
      "epoch": 0.0919296494948235,
      "grad_norm": 0.0,
      "learning_rate": 1.9799890909710013e-05,
      "loss": 1.1832,
      "step": 737
    },
    {
      "epoch": 0.0920543844330797,
      "grad_norm": 0.0,
      "learning_rate": 1.9799085920381823e-05,
      "loss": 1.1881,
      "step": 738
    },
    {
      "epoch": 0.09217911937133591,
      "grad_norm": 0.0,
      "learning_rate": 1.9798279331595776e-05,
      "loss": 1.2147,
      "step": 739
    },
    {
      "epoch": 0.09230385430959212,
      "grad_norm": 0.0,
      "learning_rate": 1.979747114348353e-05,
      "loss": 1.1695,
      "step": 740
    },
    {
      "epoch": 0.09242858924784832,
      "grad_norm": 0.0,
      "learning_rate": 1.9796661356177e-05,
      "loss": 1.1387,
      "step": 741
    },
    {
      "epoch": 0.09255332418610453,
      "grad_norm": 0.0,
      "learning_rate": 1.9795849969808366e-05,
      "loss": 1.1421,
      "step": 742
    },
    {
      "epoch": 0.09267805912436074,
      "grad_norm": 0.0,
      "learning_rate": 1.9795036984510062e-05,
      "loss": 1.2062,
      "step": 743
    },
    {
      "epoch": 0.09280279406261693,
      "grad_norm": 0.0,
      "learning_rate": 1.9794222400414793e-05,
      "loss": 1.1871,
      "step": 744
    },
    {
      "epoch": 0.09292752900087314,
      "grad_norm": 0.0,
      "learning_rate": 1.9793406217655516e-05,
      "loss": 1.2336,
      "step": 745
    },
    {
      "epoch": 0.09305226393912935,
      "grad_norm": 0.0,
      "learning_rate": 1.9792588436365453e-05,
      "loss": 1.1676,
      "step": 746
    },
    {
      "epoch": 0.09317699887738555,
      "grad_norm": 0.0,
      "learning_rate": 1.9791769056678093e-05,
      "loss": 1.1769,
      "step": 747
    },
    {
      "epoch": 0.09330173381564176,
      "grad_norm": 0.0,
      "learning_rate": 1.9790948078727167e-05,
      "loss": 1.2313,
      "step": 748
    },
    {
      "epoch": 0.09342646875389797,
      "grad_norm": 0.0,
      "learning_rate": 1.9790125502646692e-05,
      "loss": 1.1524,
      "step": 749
    },
    {
      "epoch": 0.09355120369215417,
      "grad_norm": 0.0,
      "learning_rate": 1.978930132857092e-05,
      "loss": 1.1907,
      "step": 750
    },
    {
      "epoch": 0.09367593863041038,
      "grad_norm": 0.0,
      "learning_rate": 1.978847555663439e-05,
      "loss": 1.15,
      "step": 751
    },
    {
      "epoch": 0.09380067356866659,
      "grad_norm": 0.0,
      "learning_rate": 1.9787648186971882e-05,
      "loss": 1.2393,
      "step": 752
    },
    {
      "epoch": 0.09392540850692278,
      "grad_norm": 0.0,
      "learning_rate": 1.9786819219718443e-05,
      "loss": 1.1828,
      "step": 753
    },
    {
      "epoch": 0.094050143445179,
      "grad_norm": 0.0,
      "learning_rate": 1.9785988655009386e-05,
      "loss": 1.1817,
      "step": 754
    },
    {
      "epoch": 0.0941748783834352,
      "grad_norm": 0.0,
      "learning_rate": 1.9785156492980275e-05,
      "loss": 1.1761,
      "step": 755
    },
    {
      "epoch": 0.0942996133216914,
      "grad_norm": 0.0,
      "learning_rate": 1.9784322733766942e-05,
      "loss": 1.1775,
      "step": 756
    },
    {
      "epoch": 0.09442434825994761,
      "grad_norm": 0.0,
      "learning_rate": 1.9783487377505477e-05,
      "loss": 1.1375,
      "step": 757
    },
    {
      "epoch": 0.09454908319820382,
      "grad_norm": 0.0,
      "learning_rate": 1.9782650424332234e-05,
      "loss": 1.1846,
      "step": 758
    },
    {
      "epoch": 0.09467381813646002,
      "grad_norm": 0.0,
      "learning_rate": 1.9781811874383822e-05,
      "loss": 1.1158,
      "step": 759
    },
    {
      "epoch": 0.09479855307471623,
      "grad_norm": 0.0,
      "learning_rate": 1.9780971727797112e-05,
      "loss": 1.1422,
      "step": 760
    },
    {
      "epoch": 0.09492328801297244,
      "grad_norm": 0.0,
      "learning_rate": 1.978012998470924e-05,
      "loss": 1.1235,
      "step": 761
    },
    {
      "epoch": 0.09504802295122863,
      "grad_norm": 0.0,
      "learning_rate": 1.9779286645257597e-05,
      "loss": 1.1921,
      "step": 762
    },
    {
      "epoch": 0.09517275788948484,
      "grad_norm": 0.0,
      "learning_rate": 1.977844170957984e-05,
      "loss": 1.156,
      "step": 763
    },
    {
      "epoch": 0.09529749282774105,
      "grad_norm": 0.0,
      "learning_rate": 1.9777595177813884e-05,
      "loss": 1.2195,
      "step": 764
    },
    {
      "epoch": 0.09542222776599725,
      "grad_norm": 0.0,
      "learning_rate": 1.9776747050097908e-05,
      "loss": 1.1604,
      "step": 765
    },
    {
      "epoch": 0.09554696270425346,
      "grad_norm": 0.0,
      "learning_rate": 1.9775897326570334e-05,
      "loss": 1.1834,
      "step": 766
    },
    {
      "epoch": 0.09567169764250967,
      "grad_norm": 0.0,
      "learning_rate": 1.977504600736987e-05,
      "loss": 1.1632,
      "step": 767
    },
    {
      "epoch": 0.09579643258076587,
      "grad_norm": 0.0,
      "learning_rate": 1.9774193092635476e-05,
      "loss": 1.1789,
      "step": 768
    },
    {
      "epoch": 0.09592116751902208,
      "grad_norm": 0.0,
      "learning_rate": 1.9773338582506357e-05,
      "loss": 1.1508,
      "step": 769
    },
    {
      "epoch": 0.09604590245727829,
      "grad_norm": 0.0,
      "learning_rate": 1.9772482477121995e-05,
      "loss": 1.1376,
      "step": 770
    },
    {
      "epoch": 0.09617063739553448,
      "grad_norm": 0.0,
      "learning_rate": 1.977162477662214e-05,
      "loss": 1.1842,
      "step": 771
    },
    {
      "epoch": 0.0962953723337907,
      "grad_norm": 0.0,
      "learning_rate": 1.977076548114677e-05,
      "loss": 1.1817,
      "step": 772
    },
    {
      "epoch": 0.0964201072720469,
      "grad_norm": 0.0,
      "learning_rate": 1.976990459083616e-05,
      "loss": 1.1416,
      "step": 773
    },
    {
      "epoch": 0.0965448422103031,
      "grad_norm": 0.0,
      "learning_rate": 1.976904210583082e-05,
      "loss": 1.1573,
      "step": 774
    },
    {
      "epoch": 0.09666957714855931,
      "grad_norm": 0.0,
      "learning_rate": 1.976817802627153e-05,
      "loss": 1.1388,
      "step": 775
    },
    {
      "epoch": 0.09679431208681552,
      "grad_norm": 0.0,
      "learning_rate": 1.9767312352299335e-05,
      "loss": 1.0713,
      "step": 776
    },
    {
      "epoch": 0.09691904702507172,
      "grad_norm": 0.0,
      "learning_rate": 1.9766445084055532e-05,
      "loss": 1.1738,
      "step": 777
    },
    {
      "epoch": 0.09704378196332793,
      "grad_norm": 0.0,
      "learning_rate": 1.976557622168168e-05,
      "loss": 1.1918,
      "step": 778
    },
    {
      "epoch": 0.09716851690158414,
      "grad_norm": 0.0,
      "learning_rate": 1.9764705765319602e-05,
      "loss": 1.1734,
      "step": 779
    },
    {
      "epoch": 0.09729325183984033,
      "grad_norm": 0.0,
      "learning_rate": 1.9763833715111373e-05,
      "loss": 1.2237,
      "step": 780
    },
    {
      "epoch": 0.09741798677809654,
      "grad_norm": 0.0,
      "learning_rate": 1.9762960071199334e-05,
      "loss": 1.1747,
      "step": 781
    },
    {
      "epoch": 0.09754272171635275,
      "grad_norm": 0.0,
      "learning_rate": 1.9762084833726095e-05,
      "loss": 1.1552,
      "step": 782
    },
    {
      "epoch": 0.09766745665460895,
      "grad_norm": 0.0,
      "learning_rate": 1.9761208002834506e-05,
      "loss": 1.1713,
      "step": 783
    },
    {
      "epoch": 0.09779219159286516,
      "grad_norm": 0.0,
      "learning_rate": 1.9760329578667696e-05,
      "loss": 1.223,
      "step": 784
    },
    {
      "epoch": 0.09791692653112137,
      "grad_norm": 0.0,
      "learning_rate": 1.9759449561369036e-05,
      "loss": 1.2293,
      "step": 785
    },
    {
      "epoch": 0.09804166146937757,
      "grad_norm": 0.0,
      "learning_rate": 1.975856795108218e-05,
      "loss": 1.1682,
      "step": 786
    },
    {
      "epoch": 0.09816639640763378,
      "grad_norm": 0.0,
      "learning_rate": 1.975768474795102e-05,
      "loss": 1.1513,
      "step": 787
    },
    {
      "epoch": 0.09829113134588999,
      "grad_norm": 0.0,
      "learning_rate": 1.9756799952119715e-05,
      "loss": 1.1646,
      "step": 788
    },
    {
      "epoch": 0.09841586628414618,
      "grad_norm": 0.0,
      "learning_rate": 1.9755913563732696e-05,
      "loss": 1.1568,
      "step": 789
    },
    {
      "epoch": 0.0985406012224024,
      "grad_norm": 0.0,
      "learning_rate": 1.975502558293464e-05,
      "loss": 1.1302,
      "step": 790
    },
    {
      "epoch": 0.0986653361606586,
      "grad_norm": 0.0,
      "learning_rate": 1.9754136009870478e-05,
      "loss": 1.1457,
      "step": 791
    },
    {
      "epoch": 0.0987900710989148,
      "grad_norm": 0.0,
      "learning_rate": 1.9753244844685425e-05,
      "loss": 1.1809,
      "step": 792
    },
    {
      "epoch": 0.09891480603717101,
      "grad_norm": 0.0,
      "learning_rate": 1.9752352087524934e-05,
      "loss": 1.1301,
      "step": 793
    },
    {
      "epoch": 0.09903954097542722,
      "grad_norm": 0.0,
      "learning_rate": 1.9751457738534726e-05,
      "loss": 1.2039,
      "step": 794
    },
    {
      "epoch": 0.09916427591368342,
      "grad_norm": 0.0,
      "learning_rate": 1.975056179786078e-05,
      "loss": 1.1862,
      "step": 795
    },
    {
      "epoch": 0.09928901085193963,
      "grad_norm": 0.0,
      "learning_rate": 1.9749664265649346e-05,
      "loss": 1.1899,
      "step": 796
    },
    {
      "epoch": 0.09941374579019584,
      "grad_norm": 0.0,
      "learning_rate": 1.9748765142046908e-05,
      "loss": 1.232,
      "step": 797
    },
    {
      "epoch": 0.09953848072845203,
      "grad_norm": 0.0,
      "learning_rate": 1.974786442720024e-05,
      "loss": 1.169,
      "step": 798
    },
    {
      "epoch": 0.09966321566670824,
      "grad_norm": 0.0,
      "learning_rate": 1.974696212125635e-05,
      "loss": 1.1332,
      "step": 799
    },
    {
      "epoch": 0.09978795060496445,
      "grad_norm": 0.0,
      "learning_rate": 1.9746058224362527e-05,
      "loss": 1.2009,
      "step": 800
    },
    {
      "epoch": 0.09991268554322065,
      "grad_norm": 0.0,
      "learning_rate": 1.9745152736666303e-05,
      "loss": 1.1968,
      "step": 801
    },
    {
      "epoch": 0.10003742048147686,
      "grad_norm": 0.0,
      "learning_rate": 1.9744245658315477e-05,
      "loss": 1.1263,
      "step": 802
    },
    {
      "epoch": 0.10016215541973307,
      "grad_norm": 0.0,
      "learning_rate": 1.9743336989458108e-05,
      "loss": 1.1061,
      "step": 803
    },
    {
      "epoch": 0.10028689035798927,
      "grad_norm": 0.0,
      "learning_rate": 1.9742426730242517e-05,
      "loss": 1.1357,
      "step": 804
    },
    {
      "epoch": 0.10041162529624548,
      "grad_norm": 0.0,
      "learning_rate": 1.9741514880817275e-05,
      "loss": 1.2657,
      "step": 805
    },
    {
      "epoch": 0.10053636023450169,
      "grad_norm": 0.0,
      "learning_rate": 1.9740601441331228e-05,
      "loss": 1.1085,
      "step": 806
    },
    {
      "epoch": 0.10066109517275788,
      "grad_norm": 0.0,
      "learning_rate": 1.9739686411933463e-05,
      "loss": 1.1648,
      "step": 807
    },
    {
      "epoch": 0.1007858301110141,
      "grad_norm": 0.0,
      "learning_rate": 1.9738769792773338e-05,
      "loss": 1.124,
      "step": 808
    },
    {
      "epoch": 0.1009105650492703,
      "grad_norm": 0.0,
      "learning_rate": 1.973785158400047e-05,
      "loss": 1.1553,
      "step": 809
    },
    {
      "epoch": 0.1010352999875265,
      "grad_norm": 0.0,
      "learning_rate": 1.9736931785764736e-05,
      "loss": 1.2189,
      "step": 810
    },
    {
      "epoch": 0.10116003492578271,
      "grad_norm": 0.0,
      "learning_rate": 1.9736010398216266e-05,
      "loss": 1.1865,
      "step": 811
    },
    {
      "epoch": 0.10128476986403892,
      "grad_norm": 0.0,
      "learning_rate": 1.973508742150546e-05,
      "loss": 1.2295,
      "step": 812
    },
    {
      "epoch": 0.10140950480229512,
      "grad_norm": 0.0,
      "learning_rate": 1.973416285578296e-05,
      "loss": 1.1547,
      "step": 813
    },
    {
      "epoch": 0.10153423974055133,
      "grad_norm": 0.0,
      "learning_rate": 1.9733236701199693e-05,
      "loss": 1.2123,
      "step": 814
    },
    {
      "epoch": 0.10165897467880754,
      "grad_norm": 0.0,
      "learning_rate": 1.9732308957906815e-05,
      "loss": 1.1651,
      "step": 815
    },
    {
      "epoch": 0.10178370961706373,
      "grad_norm": 0.0,
      "learning_rate": 1.9731379626055768e-05,
      "loss": 1.1578,
      "step": 816
    },
    {
      "epoch": 0.10190844455531994,
      "grad_norm": 0.0,
      "learning_rate": 1.973044870579824e-05,
      "loss": 1.2133,
      "step": 817
    },
    {
      "epoch": 0.10203317949357615,
      "grad_norm": 0.0,
      "learning_rate": 1.972951619728618e-05,
      "loss": 1.19,
      "step": 818
    },
    {
      "epoch": 0.10215791443183235,
      "grad_norm": 0.0,
      "learning_rate": 1.9728582100671797e-05,
      "loss": 1.2229,
      "step": 819
    },
    {
      "epoch": 0.10228264937008856,
      "grad_norm": 0.0,
      "learning_rate": 1.9727646416107558e-05,
      "loss": 1.1694,
      "step": 820
    },
    {
      "epoch": 0.10240738430834477,
      "grad_norm": 0.0,
      "learning_rate": 1.9726709143746194e-05,
      "loss": 1.1113,
      "step": 821
    },
    {
      "epoch": 0.10253211924660097,
      "grad_norm": 0.0,
      "learning_rate": 1.9725770283740684e-05,
      "loss": 1.1811,
      "step": 822
    },
    {
      "epoch": 0.10265685418485718,
      "grad_norm": 0.0,
      "learning_rate": 1.9724829836244282e-05,
      "loss": 1.1442,
      "step": 823
    },
    {
      "epoch": 0.10278158912311339,
      "grad_norm": 0.0,
      "learning_rate": 1.9723887801410486e-05,
      "loss": 1.1212,
      "step": 824
    },
    {
      "epoch": 0.10290632406136958,
      "grad_norm": 0.0,
      "learning_rate": 1.9722944179393068e-05,
      "loss": 1.141,
      "step": 825
    },
    {
      "epoch": 0.1030310589996258,
      "grad_norm": 0.0,
      "learning_rate": 1.972199897034604e-05,
      "loss": 1.1598,
      "step": 826
    },
    {
      "epoch": 0.103155793937882,
      "grad_norm": 0.0,
      "learning_rate": 1.9721052174423693e-05,
      "loss": 1.1316,
      "step": 827
    },
    {
      "epoch": 0.1032805288761382,
      "grad_norm": 0.0,
      "learning_rate": 1.9720103791780564e-05,
      "loss": 1.1224,
      "step": 828
    },
    {
      "epoch": 0.10340526381439441,
      "grad_norm": 0.0,
      "learning_rate": 1.9719153822571456e-05,
      "loss": 1.1039,
      "step": 829
    },
    {
      "epoch": 0.10352999875265062,
      "grad_norm": 0.0,
      "learning_rate": 1.9718202266951425e-05,
      "loss": 1.1315,
      "step": 830
    },
    {
      "epoch": 0.10365473369090682,
      "grad_norm": 0.0,
      "learning_rate": 1.9717249125075784e-05,
      "loss": 1.1592,
      "step": 831
    },
    {
      "epoch": 0.10377946862916303,
      "grad_norm": 0.0,
      "learning_rate": 1.971629439710012e-05,
      "loss": 1.1478,
      "step": 832
    },
    {
      "epoch": 0.10390420356741924,
      "grad_norm": 0.0,
      "learning_rate": 1.971533808318027e-05,
      "loss": 1.2011,
      "step": 833
    },
    {
      "epoch": 0.10402893850567543,
      "grad_norm": 0.0,
      "learning_rate": 1.9714380183472314e-05,
      "loss": 1.1466,
      "step": 834
    },
    {
      "epoch": 0.10415367344393164,
      "grad_norm": 0.0,
      "learning_rate": 1.9713420698132614e-05,
      "loss": 1.1855,
      "step": 835
    },
    {
      "epoch": 0.10427840838218785,
      "grad_norm": 0.0,
      "learning_rate": 1.9712459627317787e-05,
      "loss": 1.0987,
      "step": 836
    },
    {
      "epoch": 0.10440314332044405,
      "grad_norm": 0.0,
      "learning_rate": 1.9711496971184696e-05,
      "loss": 1.1396,
      "step": 837
    },
    {
      "epoch": 0.10452787825870026,
      "grad_norm": 0.0,
      "learning_rate": 1.9710532729890474e-05,
      "loss": 1.1682,
      "step": 838
    },
    {
      "epoch": 0.10465261319695647,
      "grad_norm": 0.0,
      "learning_rate": 1.970956690359251e-05,
      "loss": 1.1401,
      "step": 839
    },
    {
      "epoch": 0.10477734813521267,
      "grad_norm": 0.0,
      "learning_rate": 1.970859949244845e-05,
      "loss": 1.1691,
      "step": 840
    },
    {
      "epoch": 0.10490208307346888,
      "grad_norm": 0.0,
      "learning_rate": 1.9707630496616203e-05,
      "loss": 1.1299,
      "step": 841
    },
    {
      "epoch": 0.10502681801172509,
      "grad_norm": 0.0,
      "learning_rate": 1.9706659916253927e-05,
      "loss": 1.1951,
      "step": 842
    },
    {
      "epoch": 0.10515155294998128,
      "grad_norm": 0.0,
      "learning_rate": 1.970568775152005e-05,
      "loss": 1.1178,
      "step": 843
    },
    {
      "epoch": 0.1052762878882375,
      "grad_norm": 0.0,
      "learning_rate": 1.9704714002573255e-05,
      "loss": 1.18,
      "step": 844
    },
    {
      "epoch": 0.1054010228264937,
      "grad_norm": 0.0,
      "learning_rate": 1.970373866957248e-05,
      "loss": 1.1635,
      "step": 845
    },
    {
      "epoch": 0.1055257577647499,
      "grad_norm": 0.0,
      "learning_rate": 1.9702761752676923e-05,
      "loss": 1.1217,
      "step": 846
    },
    {
      "epoch": 0.10565049270300611,
      "grad_norm": 0.0,
      "learning_rate": 1.9701783252046043e-05,
      "loss": 1.1174,
      "step": 847
    },
    {
      "epoch": 0.10577522764126232,
      "grad_norm": 0.0,
      "learning_rate": 1.9700803167839556e-05,
      "loss": 1.1561,
      "step": 848
    },
    {
      "epoch": 0.10589996257951852,
      "grad_norm": 0.0,
      "learning_rate": 1.9699821500217436e-05,
      "loss": 1.1074,
      "step": 849
    },
    {
      "epoch": 0.10602469751777473,
      "grad_norm": 0.0,
      "learning_rate": 1.969883824933992e-05,
      "loss": 1.1841,
      "step": 850
    },
    {
      "epoch": 0.10614943245603094,
      "grad_norm": 0.0,
      "learning_rate": 1.9697853415367486e-05,
      "loss": 1.2502,
      "step": 851
    },
    {
      "epoch": 0.10627416739428713,
      "grad_norm": 0.0,
      "learning_rate": 1.9696866998460898e-05,
      "loss": 1.1191,
      "step": 852
    },
    {
      "epoch": 0.10639890233254334,
      "grad_norm": 0.0,
      "learning_rate": 1.9695878998781162e-05,
      "loss": 1.1336,
      "step": 853
    },
    {
      "epoch": 0.10652363727079955,
      "grad_norm": 0.0,
      "learning_rate": 1.969488941648954e-05,
      "loss": 1.1342,
      "step": 854
    },
    {
      "epoch": 0.10664837220905575,
      "grad_norm": 0.0,
      "learning_rate": 1.969389825174756e-05,
      "loss": 1.156,
      "step": 855
    },
    {
      "epoch": 0.10677310714731196,
      "grad_norm": 0.0,
      "learning_rate": 1.9692905504717e-05,
      "loss": 1.2362,
      "step": 856
    },
    {
      "epoch": 0.10689784208556817,
      "grad_norm": 0.0,
      "learning_rate": 1.9691911175559905e-05,
      "loss": 1.1404,
      "step": 857
    },
    {
      "epoch": 0.10702257702382437,
      "grad_norm": 0.0,
      "learning_rate": 1.9690915264438576e-05,
      "loss": 1.1473,
      "step": 858
    },
    {
      "epoch": 0.10714731196208058,
      "grad_norm": 0.0,
      "learning_rate": 1.968991777151557e-05,
      "loss": 1.1728,
      "step": 859
    },
    {
      "epoch": 0.10727204690033679,
      "grad_norm": 0.0,
      "learning_rate": 1.96889186969537e-05,
      "loss": 1.2058,
      "step": 860
    },
    {
      "epoch": 0.10739678183859298,
      "grad_norm": 0.0,
      "learning_rate": 1.9687918040916046e-05,
      "loss": 1.1978,
      "step": 861
    },
    {
      "epoch": 0.1075215167768492,
      "grad_norm": 0.0,
      "learning_rate": 1.9686915803565934e-05,
      "loss": 1.1532,
      "step": 862
    },
    {
      "epoch": 0.1076462517151054,
      "grad_norm": 0.0,
      "learning_rate": 1.9685911985066955e-05,
      "loss": 1.1667,
      "step": 863
    },
    {
      "epoch": 0.1077709866533616,
      "grad_norm": 0.0,
      "learning_rate": 1.9684906585582965e-05,
      "loss": 1.1577,
      "step": 864
    },
    {
      "epoch": 0.10789572159161781,
      "grad_norm": 0.0,
      "learning_rate": 1.9683899605278062e-05,
      "loss": 1.1359,
      "step": 865
    },
    {
      "epoch": 0.10802045652987402,
      "grad_norm": 0.0,
      "learning_rate": 1.968289104431661e-05,
      "loss": 1.1571,
      "step": 866
    },
    {
      "epoch": 0.10814519146813022,
      "grad_norm": 0.0,
      "learning_rate": 1.968188090286324e-05,
      "loss": 1.1749,
      "step": 867
    },
    {
      "epoch": 0.10826992640638643,
      "grad_norm": 0.0,
      "learning_rate": 1.968086918108283e-05,
      "loss": 1.1034,
      "step": 868
    },
    {
      "epoch": 0.10839466134464264,
      "grad_norm": 0.0,
      "learning_rate": 1.9679855879140512e-05,
      "loss": 1.1777,
      "step": 869
    },
    {
      "epoch": 0.10851939628289883,
      "grad_norm": 0.0,
      "learning_rate": 1.967884099720169e-05,
      "loss": 1.1433,
      "step": 870
    },
    {
      "epoch": 0.10864413122115504,
      "grad_norm": 0.0,
      "learning_rate": 1.9677824535432012e-05,
      "loss": 1.1396,
      "step": 871
    },
    {
      "epoch": 0.10876886615941125,
      "grad_norm": 0.0,
      "learning_rate": 1.9676806493997395e-05,
      "loss": 1.1601,
      "step": 872
    },
    {
      "epoch": 0.10889360109766745,
      "grad_norm": 0.0,
      "learning_rate": 1.9675786873064005e-05,
      "loss": 1.1752,
      "step": 873
    },
    {
      "epoch": 0.10901833603592366,
      "grad_norm": 0.0,
      "learning_rate": 1.9674765672798274e-05,
      "loss": 1.194,
      "step": 874
    },
    {
      "epoch": 0.10914307097417987,
      "grad_norm": 0.0,
      "learning_rate": 1.9673742893366885e-05,
      "loss": 1.1336,
      "step": 875
    },
    {
      "epoch": 0.10926780591243607,
      "grad_norm": 0.0,
      "learning_rate": 1.9672718534936785e-05,
      "loss": 1.0703,
      "step": 876
    },
    {
      "epoch": 0.10939254085069228,
      "grad_norm": 0.0,
      "learning_rate": 1.9671692597675167e-05,
      "loss": 1.1487,
      "step": 877
    },
    {
      "epoch": 0.10951727578894849,
      "grad_norm": 0.0,
      "learning_rate": 1.9670665081749497e-05,
      "loss": 1.1387,
      "step": 878
    },
    {
      "epoch": 0.10964201072720468,
      "grad_norm": 0.0,
      "learning_rate": 1.966963598732749e-05,
      "loss": 1.1354,
      "step": 879
    },
    {
      "epoch": 0.1097667456654609,
      "grad_norm": 0.0,
      "learning_rate": 1.9668605314577124e-05,
      "loss": 1.1581,
      "step": 880
    },
    {
      "epoch": 0.1098914806037171,
      "grad_norm": 0.0,
      "learning_rate": 1.9667573063666622e-05,
      "loss": 1.1854,
      "step": 881
    },
    {
      "epoch": 0.1100162155419733,
      "grad_norm": 0.0,
      "learning_rate": 1.966653923476448e-05,
      "loss": 1.1091,
      "step": 882
    },
    {
      "epoch": 0.11014095048022951,
      "grad_norm": 0.0,
      "learning_rate": 1.966550382803944e-05,
      "loss": 1.1012,
      "step": 883
    },
    {
      "epoch": 0.11026568541848572,
      "grad_norm": 0.0,
      "learning_rate": 1.966446684366051e-05,
      "loss": 1.1457,
      "step": 884
    },
    {
      "epoch": 0.11039042035674192,
      "grad_norm": 0.0,
      "learning_rate": 1.9663428281796954e-05,
      "loss": 1.0981,
      "step": 885
    },
    {
      "epoch": 0.11051515529499813,
      "grad_norm": 0.0,
      "learning_rate": 1.9662388142618285e-05,
      "loss": 1.1329,
      "step": 886
    },
    {
      "epoch": 0.11063989023325434,
      "grad_norm": 0.0,
      "learning_rate": 1.9661346426294284e-05,
      "loss": 1.1448,
      "step": 887
    },
    {
      "epoch": 0.11076462517151053,
      "grad_norm": 0.0,
      "learning_rate": 1.9660303132994988e-05,
      "loss": 1.166,
      "step": 888
    },
    {
      "epoch": 0.11088936010976674,
      "grad_norm": 0.0,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 1.1692,
      "step": 889
    },
    {
      "epoch": 0.11101409504802295,
      "grad_norm": 0.0,
      "learning_rate": 1.9658211816151923e-05,
      "loss": 1.1055,
      "step": 890
    },
    {
      "epoch": 0.11113882998627915,
      "grad_norm": 0.0,
      "learning_rate": 1.965716379294951e-05,
      "loss": 1.1489,
      "step": 891
    },
    {
      "epoch": 0.11126356492453536,
      "grad_norm": 0.0,
      "learning_rate": 1.9656114193454514e-05,
      "loss": 1.1968,
      "step": 892
    },
    {
      "epoch": 0.11138829986279157,
      "grad_norm": 0.0,
      "learning_rate": 1.9655063017838253e-05,
      "loss": 1.1433,
      "step": 893
    },
    {
      "epoch": 0.11151303480104777,
      "grad_norm": 0.0,
      "learning_rate": 1.96540102662723e-05,
      "loss": 1.1445,
      "step": 894
    },
    {
      "epoch": 0.11163776973930398,
      "grad_norm": 0.0,
      "learning_rate": 1.96529559389285e-05,
      "loss": 1.1156,
      "step": 895
    },
    {
      "epoch": 0.11176250467756019,
      "grad_norm": 0.0,
      "learning_rate": 1.9651900035978938e-05,
      "loss": 1.1377,
      "step": 896
    },
    {
      "epoch": 0.11188723961581638,
      "grad_norm": 0.0,
      "learning_rate": 1.9650842557595968e-05,
      "loss": 1.1505,
      "step": 897
    },
    {
      "epoch": 0.1120119745540726,
      "grad_norm": 0.0,
      "learning_rate": 1.9649783503952196e-05,
      "loss": 1.1513,
      "step": 898
    },
    {
      "epoch": 0.1121367094923288,
      "grad_norm": 0.0,
      "learning_rate": 1.9648722875220487e-05,
      "loss": 1.1648,
      "step": 899
    },
    {
      "epoch": 0.112261444430585,
      "grad_norm": 0.0,
      "learning_rate": 1.9647660671573957e-05,
      "loss": 1.2138,
      "step": 900
    },
    {
      "epoch": 0.11238617936884121,
      "grad_norm": 0.0,
      "learning_rate": 1.9646596893185995e-05,
      "loss": 1.124,
      "step": 901
    },
    {
      "epoch": 0.11251091430709742,
      "grad_norm": 0.0,
      "learning_rate": 1.964553154023023e-05,
      "loss": 1.1405,
      "step": 902
    },
    {
      "epoch": 0.11263564924535362,
      "grad_norm": 0.0,
      "learning_rate": 1.9644464612880555e-05,
      "loss": 1.132,
      "step": 903
    },
    {
      "epoch": 0.11276038418360983,
      "grad_norm": 0.0,
      "learning_rate": 1.964339611131112e-05,
      "loss": 1.2382,
      "step": 904
    },
    {
      "epoch": 0.11288511912186604,
      "grad_norm": 0.0,
      "learning_rate": 1.964232603569633e-05,
      "loss": 1.146,
      "step": 905
    },
    {
      "epoch": 0.11300985406012223,
      "grad_norm": 0.0,
      "learning_rate": 1.9641254386210847e-05,
      "loss": 1.1538,
      "step": 906
    },
    {
      "epoch": 0.11313458899837844,
      "grad_norm": 0.0,
      "learning_rate": 1.9640181163029597e-05,
      "loss": 1.1144,
      "step": 907
    },
    {
      "epoch": 0.11325932393663465,
      "grad_norm": 0.0,
      "learning_rate": 1.9639106366327753e-05,
      "loss": 1.1154,
      "step": 908
    },
    {
      "epoch": 0.11338405887489085,
      "grad_norm": 0.0,
      "learning_rate": 1.963802999628075e-05,
      "loss": 1.1006,
      "step": 909
    },
    {
      "epoch": 0.11350879381314706,
      "grad_norm": 0.0,
      "learning_rate": 1.9636952053064277e-05,
      "loss": 1.1512,
      "step": 910
    },
    {
      "epoch": 0.11363352875140327,
      "grad_norm": 0.0,
      "learning_rate": 1.9635872536854283e-05,
      "loss": 1.129,
      "step": 911
    },
    {
      "epoch": 0.11375826368965947,
      "grad_norm": 0.0,
      "learning_rate": 1.9634791447826972e-05,
      "loss": 1.1462,
      "step": 912
    },
    {
      "epoch": 0.11388299862791568,
      "grad_norm": 0.0,
      "learning_rate": 1.9633708786158803e-05,
      "loss": 1.1263,
      "step": 913
    },
    {
      "epoch": 0.11400773356617189,
      "grad_norm": 0.0,
      "learning_rate": 1.96326245520265e-05,
      "loss": 1.181,
      "step": 914
    },
    {
      "epoch": 0.11413246850442808,
      "grad_norm": 0.0,
      "learning_rate": 1.963153874560703e-05,
      "loss": 1.1162,
      "step": 915
    },
    {
      "epoch": 0.1142572034426843,
      "grad_norm": 0.0,
      "learning_rate": 1.963045136707763e-05,
      "loss": 1.1803,
      "step": 916
    },
    {
      "epoch": 0.1143819383809405,
      "grad_norm": 0.0,
      "learning_rate": 1.962936241661578e-05,
      "loss": 1.1123,
      "step": 917
    },
    {
      "epoch": 0.1145066733191967,
      "grad_norm": 0.0,
      "learning_rate": 1.9628271894399233e-05,
      "loss": 1.1047,
      "step": 918
    },
    {
      "epoch": 0.11463140825745291,
      "grad_norm": 0.0,
      "learning_rate": 1.9627179800605984e-05,
      "loss": 1.1256,
      "step": 919
    },
    {
      "epoch": 0.11475614319570912,
      "grad_norm": 0.0,
      "learning_rate": 1.962608613541429e-05,
      "loss": 1.1491,
      "step": 920
    },
    {
      "epoch": 0.11488087813396532,
      "grad_norm": 0.0,
      "learning_rate": 1.9624990899002666e-05,
      "loss": 1.1099,
      "step": 921
    },
    {
      "epoch": 0.11500561307222153,
      "grad_norm": 0.0,
      "learning_rate": 1.9623894091549886e-05,
      "loss": 1.1712,
      "step": 922
    },
    {
      "epoch": 0.11513034801047774,
      "grad_norm": 0.0,
      "learning_rate": 1.9622795713234973e-05,
      "loss": 1.0951,
      "step": 923
    },
    {
      "epoch": 0.11525508294873393,
      "grad_norm": 0.0,
      "learning_rate": 1.9621695764237206e-05,
      "loss": 1.1622,
      "step": 924
    },
    {
      "epoch": 0.11537981788699014,
      "grad_norm": 0.0,
      "learning_rate": 1.9620594244736133e-05,
      "loss": 1.1812,
      "step": 925
    },
    {
      "epoch": 0.11550455282524635,
      "grad_norm": 0.0,
      "learning_rate": 1.9619491154911545e-05,
      "loss": 1.153,
      "step": 926
    },
    {
      "epoch": 0.11562928776350256,
      "grad_norm": 0.0,
      "learning_rate": 1.9618386494943493e-05,
      "loss": 1.1707,
      "step": 927
    },
    {
      "epoch": 0.11575402270175876,
      "grad_norm": 0.0,
      "learning_rate": 1.9617280265012287e-05,
      "loss": 1.1652,
      "step": 928
    },
    {
      "epoch": 0.11587875764001497,
      "grad_norm": 0.0,
      "learning_rate": 1.9616172465298492e-05,
      "loss": 1.1451,
      "step": 929
    },
    {
      "epoch": 0.11600349257827118,
      "grad_norm": 0.0,
      "learning_rate": 1.9615063095982928e-05,
      "loss": 1.1559,
      "step": 930
    },
    {
      "epoch": 0.11612822751652738,
      "grad_norm": 0.0,
      "learning_rate": 1.9613952157246674e-05,
      "loss": 1.1521,
      "step": 931
    },
    {
      "epoch": 0.11625296245478359,
      "grad_norm": 0.0,
      "learning_rate": 1.961283964927106e-05,
      "loss": 1.1152,
      "step": 932
    },
    {
      "epoch": 0.1163776973930398,
      "grad_norm": 0.0,
      "learning_rate": 1.9611725572237677e-05,
      "loss": 1.1444,
      "step": 933
    },
    {
      "epoch": 0.116502432331296,
      "grad_norm": 0.0,
      "learning_rate": 1.9610609926328373e-05,
      "loss": 1.1536,
      "step": 934
    },
    {
      "epoch": 0.1166271672695522,
      "grad_norm": 0.0,
      "learning_rate": 1.9609492711725245e-05,
      "loss": 1.165,
      "step": 935
    },
    {
      "epoch": 0.11675190220780841,
      "grad_norm": 0.0,
      "learning_rate": 1.960837392861065e-05,
      "loss": 1.0875,
      "step": 936
    },
    {
      "epoch": 0.11687663714606461,
      "grad_norm": 0.0,
      "learning_rate": 1.9607253577167206e-05,
      "loss": 1.1947,
      "step": 937
    },
    {
      "epoch": 0.11700137208432082,
      "grad_norm": 0.0,
      "learning_rate": 1.9606131657577783e-05,
      "loss": 1.1163,
      "step": 938
    },
    {
      "epoch": 0.11712610702257703,
      "grad_norm": 0.0,
      "learning_rate": 1.9605008170025498e-05,
      "loss": 1.1686,
      "step": 939
    },
    {
      "epoch": 0.11725084196083323,
      "grad_norm": 0.0,
      "learning_rate": 1.9603883114693744e-05,
      "loss": 1.1438,
      "step": 940
    },
    {
      "epoch": 0.11737557689908944,
      "grad_norm": 0.0,
      "learning_rate": 1.9602756491766154e-05,
      "loss": 1.1252,
      "step": 941
    },
    {
      "epoch": 0.11750031183734565,
      "grad_norm": 0.0,
      "learning_rate": 1.9601628301426617e-05,
      "loss": 1.1245,
      "step": 942
    },
    {
      "epoch": 0.11762504677560184,
      "grad_norm": 0.0,
      "learning_rate": 1.960049854385929e-05,
      "loss": 1.1531,
      "step": 943
    },
    {
      "epoch": 0.11774978171385805,
      "grad_norm": 0.0,
      "learning_rate": 1.959936721924857e-05,
      "loss": 1.1717,
      "step": 944
    },
    {
      "epoch": 0.11787451665211426,
      "grad_norm": 0.0,
      "learning_rate": 1.959823432777912e-05,
      "loss": 1.1388,
      "step": 945
    },
    {
      "epoch": 0.11799925159037046,
      "grad_norm": 0.0,
      "learning_rate": 1.959709986963586e-05,
      "loss": 1.1534,
      "step": 946
    },
    {
      "epoch": 0.11812398652862667,
      "grad_norm": 0.0,
      "learning_rate": 1.959596384500396e-05,
      "loss": 1.1587,
      "step": 947
    },
    {
      "epoch": 0.11824872146688288,
      "grad_norm": 0.0,
      "learning_rate": 1.9594826254068846e-05,
      "loss": 1.152,
      "step": 948
    },
    {
      "epoch": 0.11837345640513908,
      "grad_norm": 0.0,
      "learning_rate": 1.9593687097016202e-05,
      "loss": 1.1289,
      "step": 949
    },
    {
      "epoch": 0.11849819134339529,
      "grad_norm": 0.0,
      "learning_rate": 1.959254637403197e-05,
      "loss": 1.169,
      "step": 950
    },
    {
      "epoch": 0.1186229262816515,
      "grad_norm": 0.0,
      "learning_rate": 1.9591404085302345e-05,
      "loss": 1.1627,
      "step": 951
    },
    {
      "epoch": 0.1187476612199077,
      "grad_norm": 0.0,
      "learning_rate": 1.9590260231013774e-05,
      "loss": 1.1162,
      "step": 952
    },
    {
      "epoch": 0.1188723961581639,
      "grad_norm": 0.0,
      "learning_rate": 1.9589114811352962e-05,
      "loss": 1.1175,
      "step": 953
    },
    {
      "epoch": 0.11899713109642011,
      "grad_norm": 0.0,
      "learning_rate": 1.9587967826506875e-05,
      "loss": 1.1419,
      "step": 954
    },
    {
      "epoch": 0.11912186603467631,
      "grad_norm": 0.0,
      "learning_rate": 1.9586819276662728e-05,
      "loss": 1.1202,
      "step": 955
    },
    {
      "epoch": 0.11924660097293252,
      "grad_norm": 0.0,
      "learning_rate": 1.958566916200799e-05,
      "loss": 1.1266,
      "step": 956
    },
    {
      "epoch": 0.11937133591118873,
      "grad_norm": 0.0,
      "learning_rate": 1.9584517482730394e-05,
      "loss": 1.1601,
      "step": 957
    },
    {
      "epoch": 0.11949607084944493,
      "grad_norm": 0.0,
      "learning_rate": 1.9583364239017916e-05,
      "loss": 1.1688,
      "step": 958
    },
    {
      "epoch": 0.11962080578770114,
      "grad_norm": 0.0,
      "learning_rate": 1.9582209431058806e-05,
      "loss": 1.1367,
      "step": 959
    },
    {
      "epoch": 0.11974554072595735,
      "grad_norm": 0.0,
      "learning_rate": 1.9581053059041547e-05,
      "loss": 1.1303,
      "step": 960
    },
    {
      "epoch": 0.11987027566421354,
      "grad_norm": 0.0,
      "learning_rate": 1.957989512315489e-05,
      "loss": 1.0973,
      "step": 961
    },
    {
      "epoch": 0.11999501060246975,
      "grad_norm": 0.0,
      "learning_rate": 1.9578735623587843e-05,
      "loss": 1.1673,
      "step": 962
    },
    {
      "epoch": 0.12011974554072596,
      "grad_norm": 0.0,
      "learning_rate": 1.9577574560529663e-05,
      "loss": 1.2002,
      "step": 963
    },
    {
      "epoch": 0.12024448047898216,
      "grad_norm": 0.0,
      "learning_rate": 1.9576411934169867e-05,
      "loss": 1.1511,
      "step": 964
    },
    {
      "epoch": 0.12036921541723837,
      "grad_norm": 0.0,
      "learning_rate": 1.9575247744698222e-05,
      "loss": 1.1388,
      "step": 965
    },
    {
      "epoch": 0.12049395035549458,
      "grad_norm": 0.0,
      "learning_rate": 1.9574081992304753e-05,
      "loss": 1.09,
      "step": 966
    },
    {
      "epoch": 0.12061868529375078,
      "grad_norm": 0.0,
      "learning_rate": 1.9572914677179743e-05,
      "loss": 1.1336,
      "step": 967
    },
    {
      "epoch": 0.12074342023200699,
      "grad_norm": 0.0,
      "learning_rate": 1.957174579951373e-05,
      "loss": 1.093,
      "step": 968
    },
    {
      "epoch": 0.1208681551702632,
      "grad_norm": 0.0,
      "learning_rate": 1.9570575359497493e-05,
      "loss": 1.1816,
      "step": 969
    },
    {
      "epoch": 0.1209928901085194,
      "grad_norm": 0.0,
      "learning_rate": 1.956940335732209e-05,
      "loss": 1.1122,
      "step": 970
    },
    {
      "epoch": 0.1211176250467756,
      "grad_norm": 0.0,
      "learning_rate": 1.9568229793178816e-05,
      "loss": 1.1463,
      "step": 971
    },
    {
      "epoch": 0.12124235998503181,
      "grad_norm": 0.0,
      "learning_rate": 1.9567054667259222e-05,
      "loss": 1.142,
      "step": 972
    },
    {
      "epoch": 0.12136709492328801,
      "grad_norm": 0.0,
      "learning_rate": 1.9565877979755125e-05,
      "loss": 1.1704,
      "step": 973
    },
    {
      "epoch": 0.12149182986154422,
      "grad_norm": 0.0,
      "learning_rate": 1.9564699730858584e-05,
      "loss": 1.1838,
      "step": 974
    },
    {
      "epoch": 0.12161656479980043,
      "grad_norm": 0.0,
      "learning_rate": 1.9563519920761926e-05,
      "loss": 1.1336,
      "step": 975
    },
    {
      "epoch": 0.12174129973805663,
      "grad_norm": 0.0,
      "learning_rate": 1.956233854965772e-05,
      "loss": 1.1126,
      "step": 976
    },
    {
      "epoch": 0.12186603467631284,
      "grad_norm": 0.0,
      "learning_rate": 1.95611556177388e-05,
      "loss": 1.1157,
      "step": 977
    },
    {
      "epoch": 0.12199076961456905,
      "grad_norm": 0.0,
      "learning_rate": 1.9559971125198246e-05,
      "loss": 1.1095,
      "step": 978
    },
    {
      "epoch": 0.12211550455282524,
      "grad_norm": 0.0,
      "learning_rate": 1.9558785072229395e-05,
      "loss": 1.1709,
      "step": 979
    },
    {
      "epoch": 0.12224023949108145,
      "grad_norm": 0.0,
      "learning_rate": 1.955759745902585e-05,
      "loss": 1.1411,
      "step": 980
    },
    {
      "epoch": 0.12236497442933766,
      "grad_norm": 0.0,
      "learning_rate": 1.9556408285781452e-05,
      "loss": 1.1383,
      "step": 981
    },
    {
      "epoch": 0.12248970936759386,
      "grad_norm": 0.0,
      "learning_rate": 1.9555217552690308e-05,
      "loss": 1.1391,
      "step": 982
    },
    {
      "epoch": 0.12261444430585007,
      "grad_norm": 0.0,
      "learning_rate": 1.9554025259946772e-05,
      "loss": 1.1447,
      "step": 983
    },
    {
      "epoch": 0.12273917924410628,
      "grad_norm": 0.0,
      "learning_rate": 1.9552831407745457e-05,
      "loss": 1.1413,
      "step": 984
    },
    {
      "epoch": 0.12286391418236248,
      "grad_norm": 0.0,
      "learning_rate": 1.955163599628123e-05,
      "loss": 1.1329,
      "step": 985
    },
    {
      "epoch": 0.12298864912061869,
      "grad_norm": 0.0,
      "learning_rate": 1.9550439025749217e-05,
      "loss": 1.1285,
      "step": 986
    },
    {
      "epoch": 0.1231133840588749,
      "grad_norm": 0.0,
      "learning_rate": 1.954924049634479e-05,
      "loss": 1.1336,
      "step": 987
    },
    {
      "epoch": 0.1232381189971311,
      "grad_norm": 0.0,
      "learning_rate": 1.9548040408263575e-05,
      "loss": 1.1578,
      "step": 988
    },
    {
      "epoch": 0.1233628539353873,
      "grad_norm": 0.0,
      "learning_rate": 1.954683876170146e-05,
      "loss": 1.1119,
      "step": 989
    },
    {
      "epoch": 0.12348758887364351,
      "grad_norm": 0.0,
      "learning_rate": 1.954563555685459e-05,
      "loss": 1.0908,
      "step": 990
    },
    {
      "epoch": 0.12361232381189971,
      "grad_norm": 0.0,
      "learning_rate": 1.9544430793919347e-05,
      "loss": 1.1221,
      "step": 991
    },
    {
      "epoch": 0.12373705875015592,
      "grad_norm": 0.0,
      "learning_rate": 1.954322447309239e-05,
      "loss": 1.153,
      "step": 992
    },
    {
      "epoch": 0.12386179368841213,
      "grad_norm": 0.0,
      "learning_rate": 1.9542016594570614e-05,
      "loss": 1.0948,
      "step": 993
    },
    {
      "epoch": 0.12398652862666833,
      "grad_norm": 0.0,
      "learning_rate": 1.9540807158551175e-05,
      "loss": 1.1051,
      "step": 994
    },
    {
      "epoch": 0.12411126356492454,
      "grad_norm": 0.0,
      "learning_rate": 1.9539596165231487e-05,
      "loss": 1.1083,
      "step": 995
    },
    {
      "epoch": 0.12423599850318075,
      "grad_norm": 0.0,
      "learning_rate": 1.953838361480921e-05,
      "loss": 1.194,
      "step": 996
    },
    {
      "epoch": 0.12436073344143694,
      "grad_norm": 0.0,
      "learning_rate": 1.953716950748227e-05,
      "loss": 1.0966,
      "step": 997
    },
    {
      "epoch": 0.12448546837969315,
      "grad_norm": 0.0,
      "learning_rate": 1.9535953843448834e-05,
      "loss": 1.0894,
      "step": 998
    },
    {
      "epoch": 0.12461020331794936,
      "grad_norm": 0.0,
      "learning_rate": 1.9534736622907334e-05,
      "loss": 1.1391,
      "step": 999
    },
    {
      "epoch": 0.12473493825620556,
      "grad_norm": 0.0,
      "learning_rate": 1.9533517846056442e-05,
      "loss": 1.1064,
      "step": 1000
    },
    {
      "epoch": 0.12485967319446177,
      "grad_norm": 0.0,
      "learning_rate": 1.95322975130951e-05,
      "loss": 1.1433,
      "step": 1001
    },
    {
      "epoch": 0.12498440813271798,
      "grad_norm": 0.0,
      "learning_rate": 1.95310756242225e-05,
      "loss": 1.1501,
      "step": 1002
    },
    {
      "epoch": 0.12510914307097418,
      "grad_norm": 0.0,
      "learning_rate": 1.952985217963808e-05,
      "loss": 1.1732,
      "step": 1003
    },
    {
      "epoch": 0.1252338780092304,
      "grad_norm": 0.0,
      "learning_rate": 1.9528627179541534e-05,
      "loss": 1.1597,
      "step": 1004
    },
    {
      "epoch": 0.1253586129474866,
      "grad_norm": 0.0,
      "learning_rate": 1.952740062413282e-05,
      "loss": 1.1587,
      "step": 1005
    },
    {
      "epoch": 0.1254833478857428,
      "grad_norm": 0.0,
      "learning_rate": 1.952617251361214e-05,
      "loss": 1.1582,
      "step": 1006
    },
    {
      "epoch": 0.125608082823999,
      "grad_norm": 0.0,
      "learning_rate": 1.952494284817995e-05,
      "loss": 1.1231,
      "step": 1007
    },
    {
      "epoch": 0.1257328177622552,
      "grad_norm": 0.0,
      "learning_rate": 1.9523711628036967e-05,
      "loss": 1.1287,
      "step": 1008
    },
    {
      "epoch": 0.1258575527005114,
      "grad_norm": 0.0,
      "learning_rate": 1.9522478853384154e-05,
      "loss": 1.1177,
      "step": 1009
    },
    {
      "epoch": 0.12598228763876762,
      "grad_norm": 0.0,
      "learning_rate": 1.9521244524422732e-05,
      "loss": 1.0999,
      "step": 1010
    },
    {
      "epoch": 0.12610702257702383,
      "grad_norm": 0.0,
      "learning_rate": 1.9520008641354176e-05,
      "loss": 1.1496,
      "step": 1011
    },
    {
      "epoch": 0.12623175751528004,
      "grad_norm": 0.0,
      "learning_rate": 1.951877120438021e-05,
      "loss": 1.1302,
      "step": 1012
    },
    {
      "epoch": 0.12635649245353622,
      "grad_norm": 0.0,
      "learning_rate": 1.9517532213702816e-05,
      "loss": 1.1575,
      "step": 1013
    },
    {
      "epoch": 0.12648122739179243,
      "grad_norm": 0.0,
      "learning_rate": 1.951629166952423e-05,
      "loss": 1.1348,
      "step": 1014
    },
    {
      "epoch": 0.12660596233004864,
      "grad_norm": 0.0,
      "learning_rate": 1.951504957204694e-05,
      "loss": 1.1499,
      "step": 1015
    },
    {
      "epoch": 0.12673069726830485,
      "grad_norm": 0.0,
      "learning_rate": 1.9513805921473685e-05,
      "loss": 1.1314,
      "step": 1016
    },
    {
      "epoch": 0.12685543220656106,
      "grad_norm": 0.0,
      "learning_rate": 1.9512560718007462e-05,
      "loss": 1.1384,
      "step": 1017
    },
    {
      "epoch": 0.12698016714481727,
      "grad_norm": 0.0,
      "learning_rate": 1.9511313961851522e-05,
      "loss": 1.1023,
      "step": 1018
    },
    {
      "epoch": 0.12710490208307346,
      "grad_norm": 0.0,
      "learning_rate": 1.951006565320936e-05,
      "loss": 1.0817,
      "step": 1019
    },
    {
      "epoch": 0.12722963702132967,
      "grad_norm": 0.0,
      "learning_rate": 1.9508815792284742e-05,
      "loss": 1.1211,
      "step": 1020
    },
    {
      "epoch": 0.12735437195958588,
      "grad_norm": 0.0,
      "learning_rate": 1.9507564379281666e-05,
      "loss": 1.1528,
      "step": 1021
    },
    {
      "epoch": 0.1274791068978421,
      "grad_norm": 0.0,
      "learning_rate": 1.95063114144044e-05,
      "loss": 1.1334,
      "step": 1022
    },
    {
      "epoch": 0.1276038418360983,
      "grad_norm": 0.0,
      "learning_rate": 1.9505056897857462e-05,
      "loss": 1.1118,
      "step": 1023
    },
    {
      "epoch": 0.1277285767743545,
      "grad_norm": 0.0,
      "learning_rate": 1.9503800829845613e-05,
      "loss": 1.1127,
      "step": 1024
    },
    {
      "epoch": 0.1278533117126107,
      "grad_norm": 0.0,
      "learning_rate": 1.950254321057388e-05,
      "loss": 1.1389,
      "step": 1025
    },
    {
      "epoch": 0.1279780466508669,
      "grad_norm": 0.0,
      "learning_rate": 1.9501284040247538e-05,
      "loss": 1.1227,
      "step": 1026
    },
    {
      "epoch": 0.1281027815891231,
      "grad_norm": 0.0,
      "learning_rate": 1.9500023319072113e-05,
      "loss": 1.1492,
      "step": 1027
    },
    {
      "epoch": 0.12822751652737932,
      "grad_norm": 0.0,
      "learning_rate": 1.949876104725339e-05,
      "loss": 1.0503,
      "step": 1028
    },
    {
      "epoch": 0.12835225146563553,
      "grad_norm": 0.0,
      "learning_rate": 1.9497497224997402e-05,
      "loss": 1.1046,
      "step": 1029
    },
    {
      "epoch": 0.12847698640389174,
      "grad_norm": 0.0,
      "learning_rate": 1.9496231852510434e-05,
      "loss": 1.1636,
      "step": 1030
    },
    {
      "epoch": 0.12860172134214792,
      "grad_norm": 0.0,
      "learning_rate": 1.9494964929999032e-05,
      "loss": 1.1369,
      "step": 1031
    },
    {
      "epoch": 0.12872645628040413,
      "grad_norm": 0.0,
      "learning_rate": 1.9493696457669982e-05,
      "loss": 1.096,
      "step": 1032
    },
    {
      "epoch": 0.12885119121866034,
      "grad_norm": 0.0,
      "learning_rate": 1.949242643573034e-05,
      "loss": 1.1121,
      "step": 1033
    },
    {
      "epoch": 0.12897592615691655,
      "grad_norm": 0.0,
      "learning_rate": 1.94911548643874e-05,
      "loss": 1.125,
      "step": 1034
    },
    {
      "epoch": 0.12910066109517276,
      "grad_norm": 0.0,
      "learning_rate": 1.9489881743848714e-05,
      "loss": 1.1212,
      "step": 1035
    },
    {
      "epoch": 0.12922539603342897,
      "grad_norm": 0.0,
      "learning_rate": 1.9488607074322094e-05,
      "loss": 1.1142,
      "step": 1036
    },
    {
      "epoch": 0.12935013097168516,
      "grad_norm": 0.0,
      "learning_rate": 1.9487330856015587e-05,
      "loss": 1.1235,
      "step": 1037
    },
    {
      "epoch": 0.12947486590994137,
      "grad_norm": 0.0,
      "learning_rate": 1.948605308913751e-05,
      "loss": 1.1715,
      "step": 1038
    },
    {
      "epoch": 0.12959960084819758,
      "grad_norm": 0.0,
      "learning_rate": 1.948477377389643e-05,
      "loss": 1.1627,
      "step": 1039
    },
    {
      "epoch": 0.1297243357864538,
      "grad_norm": 0.0,
      "learning_rate": 1.948349291050116e-05,
      "loss": 1.1719,
      "step": 1040
    },
    {
      "epoch": 0.12984907072471,
      "grad_norm": 0.0,
      "learning_rate": 1.9482210499160767e-05,
      "loss": 1.1182,
      "step": 1041
    },
    {
      "epoch": 0.1299738056629662,
      "grad_norm": 0.0,
      "learning_rate": 1.9480926540084578e-05,
      "loss": 1.0823,
      "step": 1042
    },
    {
      "epoch": 0.1300985406012224,
      "grad_norm": 0.0,
      "learning_rate": 1.9479641033482163e-05,
      "loss": 1.1308,
      "step": 1043
    },
    {
      "epoch": 0.1302232755394786,
      "grad_norm": 0.0,
      "learning_rate": 1.9478353979563352e-05,
      "loss": 1.1437,
      "step": 1044
    },
    {
      "epoch": 0.1303480104777348,
      "grad_norm": 0.0,
      "learning_rate": 1.9477065378538222e-05,
      "loss": 1.0979,
      "step": 1045
    },
    {
      "epoch": 0.13047274541599102,
      "grad_norm": 0.0,
      "learning_rate": 1.947577523061711e-05,
      "loss": 1.141,
      "step": 1046
    },
    {
      "epoch": 0.13059748035424723,
      "grad_norm": 0.0,
      "learning_rate": 1.9474483536010592e-05,
      "loss": 1.17,
      "step": 1047
    },
    {
      "epoch": 0.13072221529250344,
      "grad_norm": 0.0,
      "learning_rate": 1.9473190294929514e-05,
      "loss": 1.0991,
      "step": 1048
    },
    {
      "epoch": 0.13084695023075962,
      "grad_norm": 0.0,
      "learning_rate": 1.9471895507584963e-05,
      "loss": 1.1973,
      "step": 1049
    },
    {
      "epoch": 0.13097168516901583,
      "grad_norm": 0.0,
      "learning_rate": 1.947059917418828e-05,
      "loss": 1.1357,
      "step": 1050
    },
    {
      "epoch": 0.13109642010727204,
      "grad_norm": 0.0,
      "learning_rate": 1.946930129495106e-05,
      "loss": 1.1282,
      "step": 1051
    },
    {
      "epoch": 0.13122115504552825,
      "grad_norm": 0.0,
      "learning_rate": 1.9468001870085145e-05,
      "loss": 1.1503,
      "step": 1052
    },
    {
      "epoch": 0.13134588998378446,
      "grad_norm": 0.0,
      "learning_rate": 1.946670089980264e-05,
      "loss": 1.1082,
      "step": 1053
    },
    {
      "epoch": 0.13147062492204067,
      "grad_norm": 0.0,
      "learning_rate": 1.9465398384315893e-05,
      "loss": 1.161,
      "step": 1054
    },
    {
      "epoch": 0.13159535986029686,
      "grad_norm": 0.0,
      "learning_rate": 1.9464094323837514e-05,
      "loss": 1.1192,
      "step": 1055
    },
    {
      "epoch": 0.13172009479855307,
      "grad_norm": 0.0,
      "learning_rate": 1.9462788718580347e-05,
      "loss": 1.1282,
      "step": 1056
    },
    {
      "epoch": 0.13184482973680928,
      "grad_norm": 0.0,
      "learning_rate": 1.946148156875751e-05,
      "loss": 1.11,
      "step": 1057
    },
    {
      "epoch": 0.1319695646750655,
      "grad_norm": 0.0,
      "learning_rate": 1.946017287458235e-05,
      "loss": 1.132,
      "step": 1058
    },
    {
      "epoch": 0.1320942996133217,
      "grad_norm": 0.0,
      "learning_rate": 1.9458862636268494e-05,
      "loss": 1.1124,
      "step": 1059
    },
    {
      "epoch": 0.1322190345515779,
      "grad_norm": 0.0,
      "learning_rate": 1.9457550854029797e-05,
      "loss": 1.1679,
      "step": 1060
    },
    {
      "epoch": 0.1323437694898341,
      "grad_norm": 0.0,
      "learning_rate": 1.9456237528080377e-05,
      "loss": 1.1478,
      "step": 1061
    },
    {
      "epoch": 0.1324685044280903,
      "grad_norm": 0.0,
      "learning_rate": 1.9454922658634604e-05,
      "loss": 1.0773,
      "step": 1062
    },
    {
      "epoch": 0.1325932393663465,
      "grad_norm": 0.0,
      "learning_rate": 1.9453606245907094e-05,
      "loss": 1.1056,
      "step": 1063
    },
    {
      "epoch": 0.13271797430460272,
      "grad_norm": 0.0,
      "learning_rate": 1.9452288290112725e-05,
      "loss": 1.1262,
      "step": 1064
    },
    {
      "epoch": 0.13284270924285893,
      "grad_norm": 0.0,
      "learning_rate": 1.945096879146661e-05,
      "loss": 1.1281,
      "step": 1065
    },
    {
      "epoch": 0.13296744418111514,
      "grad_norm": 0.0,
      "learning_rate": 1.944964775018413e-05,
      "loss": 1.1214,
      "step": 1066
    },
    {
      "epoch": 0.13309217911937132,
      "grad_norm": 0.0,
      "learning_rate": 1.944832516648092e-05,
      "loss": 1.1413,
      "step": 1067
    },
    {
      "epoch": 0.13321691405762753,
      "grad_norm": 0.0,
      "learning_rate": 1.944700104057285e-05,
      "loss": 1.1123,
      "step": 1068
    },
    {
      "epoch": 0.13334164899588374,
      "grad_norm": 0.0,
      "learning_rate": 1.944567537267605e-05,
      "loss": 1.0934,
      "step": 1069
    },
    {
      "epoch": 0.13346638393413995,
      "grad_norm": 0.0,
      "learning_rate": 1.9444348163006904e-05,
      "loss": 1.1306,
      "step": 1070
    },
    {
      "epoch": 0.13359111887239616,
      "grad_norm": 0.0,
      "learning_rate": 1.944301941178205e-05,
      "loss": 1.1304,
      "step": 1071
    },
    {
      "epoch": 0.13371585381065237,
      "grad_norm": 0.0,
      "learning_rate": 1.944168911921837e-05,
      "loss": 1.1074,
      "step": 1072
    },
    {
      "epoch": 0.13384058874890856,
      "grad_norm": 0.0,
      "learning_rate": 1.9440357285533e-05,
      "loss": 1.0921,
      "step": 1073
    },
    {
      "epoch": 0.13396532368716477,
      "grad_norm": 0.0,
      "learning_rate": 1.9439023910943336e-05,
      "loss": 1.1569,
      "step": 1074
    },
    {
      "epoch": 0.13409005862542098,
      "grad_norm": 0.0,
      "learning_rate": 1.9437688995667008e-05,
      "loss": 1.0784,
      "step": 1075
    },
    {
      "epoch": 0.1342147935636772,
      "grad_norm": 0.0,
      "learning_rate": 1.9436352539921916e-05,
      "loss": 1.1296,
      "step": 1076
    },
    {
      "epoch": 0.1343395285019334,
      "grad_norm": 0.0,
      "learning_rate": 1.94350145439262e-05,
      "loss": 1.0693,
      "step": 1077
    },
    {
      "epoch": 0.1344642634401896,
      "grad_norm": 0.0,
      "learning_rate": 1.9433675007898255e-05,
      "loss": 1.1578,
      "step": 1078
    },
    {
      "epoch": 0.1345889983784458,
      "grad_norm": 0.0,
      "learning_rate": 1.9432333932056726e-05,
      "loss": 1.1414,
      "step": 1079
    },
    {
      "epoch": 0.134713733316702,
      "grad_norm": 0.0,
      "learning_rate": 1.9430991316620512e-05,
      "loss": 1.2287,
      "step": 1080
    },
    {
      "epoch": 0.1348384682549582,
      "grad_norm": 0.0,
      "learning_rate": 1.942964716180876e-05,
      "loss": 1.0994,
      "step": 1081
    },
    {
      "epoch": 0.13496320319321442,
      "grad_norm": 0.0,
      "learning_rate": 1.9428301467840875e-05,
      "loss": 1.1297,
      "step": 1082
    },
    {
      "epoch": 0.13508793813147063,
      "grad_norm": 0.0,
      "learning_rate": 1.94269542349365e-05,
      "loss": 1.1508,
      "step": 1083
    },
    {
      "epoch": 0.13521267306972684,
      "grad_norm": 0.0,
      "learning_rate": 1.9425605463315546e-05,
      "loss": 1.0755,
      "step": 1084
    },
    {
      "epoch": 0.13533740800798302,
      "grad_norm": 0.0,
      "learning_rate": 1.942425515319816e-05,
      "loss": 1.1158,
      "step": 1085
    },
    {
      "epoch": 0.13546214294623923,
      "grad_norm": 0.0,
      "learning_rate": 1.9422903304804747e-05,
      "loss": 1.1375,
      "step": 1086
    },
    {
      "epoch": 0.13558687788449544,
      "grad_norm": 0.0,
      "learning_rate": 1.9421549918355965e-05,
      "loss": 1.0972,
      "step": 1087
    },
    {
      "epoch": 0.13571161282275165,
      "grad_norm": 0.0,
      "learning_rate": 1.942019499407272e-05,
      "loss": 1.1236,
      "step": 1088
    },
    {
      "epoch": 0.13583634776100786,
      "grad_norm": 0.0,
      "learning_rate": 1.9418838532176173e-05,
      "loss": 1.1385,
      "step": 1089
    },
    {
      "epoch": 0.13596108269926407,
      "grad_norm": 0.0,
      "learning_rate": 1.9417480532887733e-05,
      "loss": 1.0836,
      "step": 1090
    },
    {
      "epoch": 0.13608581763752026,
      "grad_norm": 0.0,
      "learning_rate": 1.941612099642905e-05,
      "loss": 1.0772,
      "step": 1091
    },
    {
      "epoch": 0.13621055257577647,
      "grad_norm": 0.0,
      "learning_rate": 1.9414759923022045e-05,
      "loss": 1.1735,
      "step": 1092
    },
    {
      "epoch": 0.13633528751403268,
      "grad_norm": 0.0,
      "learning_rate": 1.9413397312888876e-05,
      "loss": 1.0814,
      "step": 1093
    },
    {
      "epoch": 0.1364600224522889,
      "grad_norm": 0.0,
      "learning_rate": 1.9412033166251954e-05,
      "loss": 1.1389,
      "step": 1094
    },
    {
      "epoch": 0.1365847573905451,
      "grad_norm": 0.0,
      "learning_rate": 1.9410667483333944e-05,
      "loss": 1.0884,
      "step": 1095
    },
    {
      "epoch": 0.1367094923288013,
      "grad_norm": 0.0,
      "learning_rate": 1.9409300264357757e-05,
      "loss": 1.0577,
      "step": 1096
    },
    {
      "epoch": 0.1368342272670575,
      "grad_norm": 0.0,
      "learning_rate": 1.940793150954656e-05,
      "loss": 1.1443,
      "step": 1097
    },
    {
      "epoch": 0.1369589622053137,
      "grad_norm": 0.0,
      "learning_rate": 1.940656121912377e-05,
      "loss": 1.0989,
      "step": 1098
    },
    {
      "epoch": 0.1370836971435699,
      "grad_norm": 0.0,
      "learning_rate": 1.940518939331305e-05,
      "loss": 1.063,
      "step": 1099
    },
    {
      "epoch": 0.13720843208182612,
      "grad_norm": 0.0,
      "learning_rate": 1.9403816032338318e-05,
      "loss": 1.1621,
      "step": 1100
    },
    {
      "epoch": 0.13733316702008233,
      "grad_norm": 0.0,
      "learning_rate": 1.9402441136423738e-05,
      "loss": 1.1752,
      "step": 1101
    },
    {
      "epoch": 0.13745790195833854,
      "grad_norm": 0.0,
      "learning_rate": 1.9401064705793735e-05,
      "loss": 1.1501,
      "step": 1102
    },
    {
      "epoch": 0.13758263689659472,
      "grad_norm": 0.0,
      "learning_rate": 1.9399686740672967e-05,
      "loss": 1.1507,
      "step": 1103
    },
    {
      "epoch": 0.13770737183485093,
      "grad_norm": 0.0,
      "learning_rate": 1.9398307241286364e-05,
      "loss": 1.1157,
      "step": 1104
    },
    {
      "epoch": 0.13783210677310714,
      "grad_norm": 0.0,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 1.0677,
      "step": 1105
    },
    {
      "epoch": 0.13795684171136335,
      "grad_norm": 0.0,
      "learning_rate": 1.9395543640616553e-05,
      "loss": 1.1637,
      "step": 1106
    },
    {
      "epoch": 0.13808157664961956,
      "grad_norm": 0.0,
      "learning_rate": 1.9394159539784444e-05,
      "loss": 1.0835,
      "step": 1107
    },
    {
      "epoch": 0.13820631158787577,
      "grad_norm": 0.0,
      "learning_rate": 1.9392773905588668e-05,
      "loss": 1.081,
      "step": 1108
    },
    {
      "epoch": 0.13833104652613196,
      "grad_norm": 0.0,
      "learning_rate": 1.93913867382554e-05,
      "loss": 1.1475,
      "step": 1109
    },
    {
      "epoch": 0.13845578146438817,
      "grad_norm": 0.0,
      "learning_rate": 1.9389998038011068e-05,
      "loss": 1.1231,
      "step": 1110
    },
    {
      "epoch": 0.13858051640264438,
      "grad_norm": 0.0,
      "learning_rate": 1.9388607805082328e-05,
      "loss": 1.0812,
      "step": 1111
    },
    {
      "epoch": 0.1387052513409006,
      "grad_norm": 0.0,
      "learning_rate": 1.938721603969611e-05,
      "loss": 1.146,
      "step": 1112
    },
    {
      "epoch": 0.1388299862791568,
      "grad_norm": 0.0,
      "learning_rate": 1.9385822742079585e-05,
      "loss": 1.0993,
      "step": 1113
    },
    {
      "epoch": 0.138954721217413,
      "grad_norm": 0.0,
      "learning_rate": 1.9384427912460172e-05,
      "loss": 1.1314,
      "step": 1114
    },
    {
      "epoch": 0.1390794561556692,
      "grad_norm": 0.0,
      "learning_rate": 1.9383031551065546e-05,
      "loss": 1.1138,
      "step": 1115
    },
    {
      "epoch": 0.1392041910939254,
      "grad_norm": 0.0,
      "learning_rate": 1.9381633658123625e-05,
      "loss": 1.1517,
      "step": 1116
    },
    {
      "epoch": 0.1393289260321816,
      "grad_norm": 0.0,
      "learning_rate": 1.938023423386258e-05,
      "loss": 1.1195,
      "step": 1117
    },
    {
      "epoch": 0.13945366097043782,
      "grad_norm": 0.0,
      "learning_rate": 1.9378833278510835e-05,
      "loss": 1.1632,
      "step": 1118
    },
    {
      "epoch": 0.13957839590869403,
      "grad_norm": 0.0,
      "learning_rate": 1.937743079229706e-05,
      "loss": 1.1023,
      "step": 1119
    },
    {
      "epoch": 0.13970313084695024,
      "grad_norm": 0.0,
      "learning_rate": 1.9376026775450172e-05,
      "loss": 1.188,
      "step": 1120
    },
    {
      "epoch": 0.13982786578520642,
      "grad_norm": 0.0,
      "learning_rate": 1.9374621228199353e-05,
      "loss": 1.1246,
      "step": 1121
    },
    {
      "epoch": 0.13995260072346263,
      "grad_norm": 0.0,
      "learning_rate": 1.9373214150774014e-05,
      "loss": 1.0856,
      "step": 1122
    },
    {
      "epoch": 0.14007733566171884,
      "grad_norm": 0.0,
      "learning_rate": 1.9371805543403826e-05,
      "loss": 1.1137,
      "step": 1123
    },
    {
      "epoch": 0.14020207059997505,
      "grad_norm": 0.0,
      "learning_rate": 1.9370395406318713e-05,
      "loss": 1.0517,
      "step": 1124
    },
    {
      "epoch": 0.14032680553823126,
      "grad_norm": 0.0,
      "learning_rate": 1.9368983739748842e-05,
      "loss": 1.1367,
      "step": 1125
    },
    {
      "epoch": 0.14045154047648747,
      "grad_norm": 0.0,
      "learning_rate": 1.9367570543924635e-05,
      "loss": 1.1535,
      "step": 1126
    },
    {
      "epoch": 0.14057627541474366,
      "grad_norm": 0.0,
      "learning_rate": 1.936615581907676e-05,
      "loss": 1.1021,
      "step": 1127
    },
    {
      "epoch": 0.14070101035299987,
      "grad_norm": 0.0,
      "learning_rate": 1.9364739565436137e-05,
      "loss": 1.0551,
      "step": 1128
    },
    {
      "epoch": 0.14082574529125608,
      "grad_norm": 0.0,
      "learning_rate": 1.9363321783233933e-05,
      "loss": 1.0974,
      "step": 1129
    },
    {
      "epoch": 0.1409504802295123,
      "grad_norm": 0.0,
      "learning_rate": 1.9361902472701563e-05,
      "loss": 1.1214,
      "step": 1130
    },
    {
      "epoch": 0.1410752151677685,
      "grad_norm": 0.0,
      "learning_rate": 1.9360481634070704e-05,
      "loss": 1.1726,
      "step": 1131
    },
    {
      "epoch": 0.1411999501060247,
      "grad_norm": 0.0,
      "learning_rate": 1.935905926757326e-05,
      "loss": 1.1374,
      "step": 1132
    },
    {
      "epoch": 0.1413246850442809,
      "grad_norm": 0.0,
      "learning_rate": 1.9357635373441402e-05,
      "loss": 1.1175,
      "step": 1133
    },
    {
      "epoch": 0.1414494199825371,
      "grad_norm": 0.0,
      "learning_rate": 1.935620995190755e-05,
      "loss": 1.1196,
      "step": 1134
    },
    {
      "epoch": 0.1415741549207933,
      "grad_norm": 0.0,
      "learning_rate": 1.9354783003204366e-05,
      "loss": 1.1663,
      "step": 1135
    },
    {
      "epoch": 0.14169888985904952,
      "grad_norm": 0.0,
      "learning_rate": 1.935335452756476e-05,
      "loss": 1.1566,
      "step": 1136
    },
    {
      "epoch": 0.14182362479730573,
      "grad_norm": 0.0,
      "learning_rate": 1.93519245252219e-05,
      "loss": 1.0581,
      "step": 1137
    },
    {
      "epoch": 0.14194835973556194,
      "grad_norm": 0.0,
      "learning_rate": 1.9350492996409195e-05,
      "loss": 1.1406,
      "step": 1138
    },
    {
      "epoch": 0.14207309467381812,
      "grad_norm": 0.0,
      "learning_rate": 1.934905994136031e-05,
      "loss": 1.148,
      "step": 1139
    },
    {
      "epoch": 0.14219782961207433,
      "grad_norm": 0.0,
      "learning_rate": 1.9347625360309153e-05,
      "loss": 1.0833,
      "step": 1140
    },
    {
      "epoch": 0.14232256455033054,
      "grad_norm": 0.0,
      "learning_rate": 1.9346189253489888e-05,
      "loss": 1.1359,
      "step": 1141
    },
    {
      "epoch": 0.14244729948858675,
      "grad_norm": 0.0,
      "learning_rate": 1.9344751621136915e-05,
      "loss": 1.1354,
      "step": 1142
    },
    {
      "epoch": 0.14257203442684296,
      "grad_norm": 0.0,
      "learning_rate": 1.9343312463484903e-05,
      "loss": 1.1611,
      "step": 1143
    },
    {
      "epoch": 0.14269676936509917,
      "grad_norm": 0.0,
      "learning_rate": 1.934187178076875e-05,
      "loss": 1.1539,
      "step": 1144
    },
    {
      "epoch": 0.14282150430335536,
      "grad_norm": 0.0,
      "learning_rate": 1.934042957322362e-05,
      "loss": 1.127,
      "step": 1145
    },
    {
      "epoch": 0.14294623924161157,
      "grad_norm": 0.0,
      "learning_rate": 1.9338985841084907e-05,
      "loss": 1.1159,
      "step": 1146
    },
    {
      "epoch": 0.14307097417986778,
      "grad_norm": 0.0,
      "learning_rate": 1.933754058458827e-05,
      "loss": 1.0938,
      "step": 1147
    },
    {
      "epoch": 0.143195709118124,
      "grad_norm": 0.0,
      "learning_rate": 1.9336093803969616e-05,
      "loss": 1.12,
      "step": 1148
    },
    {
      "epoch": 0.1433204440563802,
      "grad_norm": 0.0,
      "learning_rate": 1.933464549946509e-05,
      "loss": 1.1476,
      "step": 1149
    },
    {
      "epoch": 0.1434451789946364,
      "grad_norm": 0.0,
      "learning_rate": 1.9333195671311093e-05,
      "loss": 1.0917,
      "step": 1150
    },
    {
      "epoch": 0.1435699139328926,
      "grad_norm": 0.0,
      "learning_rate": 1.9331744319744272e-05,
      "loss": 1.1102,
      "step": 1151
    },
    {
      "epoch": 0.1436946488711488,
      "grad_norm": 0.0,
      "learning_rate": 1.933029144500153e-05,
      "loss": 1.1028,
      "step": 1152
    },
    {
      "epoch": 0.143819383809405,
      "grad_norm": 0.0,
      "learning_rate": 1.932883704732001e-05,
      "loss": 1.1347,
      "step": 1153
    },
    {
      "epoch": 0.14394411874766122,
      "grad_norm": 0.0,
      "learning_rate": 1.9327381126937098e-05,
      "loss": 1.0719,
      "step": 1154
    },
    {
      "epoch": 0.14406885368591743,
      "grad_norm": 0.0,
      "learning_rate": 1.932592368409045e-05,
      "loss": 1.0831,
      "step": 1155
    },
    {
      "epoch": 0.14419358862417364,
      "grad_norm": 0.0,
      "learning_rate": 1.9324464719017947e-05,
      "loss": 1.0972,
      "step": 1156
    },
    {
      "epoch": 0.14431832356242982,
      "grad_norm": 0.0,
      "learning_rate": 1.932300423195774e-05,
      "loss": 1.0684,
      "step": 1157
    },
    {
      "epoch": 0.14444305850068603,
      "grad_norm": 0.0,
      "learning_rate": 1.9321542223148207e-05,
      "loss": 1.0786,
      "step": 1158
    },
    {
      "epoch": 0.14456779343894224,
      "grad_norm": 0.0,
      "learning_rate": 1.932007869282799e-05,
      "loss": 1.1027,
      "step": 1159
    },
    {
      "epoch": 0.14469252837719845,
      "grad_norm": 0.0,
      "learning_rate": 1.931861364123597e-05,
      "loss": 1.1092,
      "step": 1160
    },
    {
      "epoch": 0.14481726331545466,
      "grad_norm": 0.0,
      "learning_rate": 1.9317147068611282e-05,
      "loss": 1.0734,
      "step": 1161
    },
    {
      "epoch": 0.14494199825371087,
      "grad_norm": 0.0,
      "learning_rate": 1.931567897519331e-05,
      "loss": 1.1152,
      "step": 1162
    },
    {
      "epoch": 0.14506673319196706,
      "grad_norm": 0.0,
      "learning_rate": 1.931420936122168e-05,
      "loss": 1.0743,
      "step": 1163
    },
    {
      "epoch": 0.14519146813022327,
      "grad_norm": 0.0,
      "learning_rate": 1.9312738226936277e-05,
      "loss": 1.1153,
      "step": 1164
    },
    {
      "epoch": 0.14531620306847948,
      "grad_norm": 0.0,
      "learning_rate": 1.931126557257722e-05,
      "loss": 1.131,
      "step": 1165
    },
    {
      "epoch": 0.1454409380067357,
      "grad_norm": 0.0,
      "learning_rate": 1.930979139838489e-05,
      "loss": 1.1659,
      "step": 1166
    },
    {
      "epoch": 0.1455656729449919,
      "grad_norm": 0.0,
      "learning_rate": 1.93083157045999e-05,
      "loss": 1.1132,
      "step": 1167
    },
    {
      "epoch": 0.1456904078832481,
      "grad_norm": 0.0,
      "learning_rate": 1.9306838491463126e-05,
      "loss": 1.1274,
      "step": 1168
    },
    {
      "epoch": 0.1458151428215043,
      "grad_norm": 0.0,
      "learning_rate": 1.9305359759215686e-05,
      "loss": 1.1355,
      "step": 1169
    },
    {
      "epoch": 0.1459398777597605,
      "grad_norm": 0.0,
      "learning_rate": 1.930387950809895e-05,
      "loss": 1.1355,
      "step": 1170
    },
    {
      "epoch": 0.1460646126980167,
      "grad_norm": 0.0,
      "learning_rate": 1.9302397738354526e-05,
      "loss": 1.1082,
      "step": 1171
    },
    {
      "epoch": 0.14618934763627292,
      "grad_norm": 0.0,
      "learning_rate": 1.9300914450224278e-05,
      "loss": 1.1711,
      "step": 1172
    },
    {
      "epoch": 0.14631408257452913,
      "grad_norm": 0.0,
      "learning_rate": 1.929942964395032e-05,
      "loss": 1.1451,
      "step": 1173
    },
    {
      "epoch": 0.14643881751278534,
      "grad_norm": 0.0,
      "learning_rate": 1.9297943319775007e-05,
      "loss": 1.0723,
      "step": 1174
    },
    {
      "epoch": 0.14656355245104152,
      "grad_norm": 0.0,
      "learning_rate": 1.929645547794094e-05,
      "loss": 1.0819,
      "step": 1175
    },
    {
      "epoch": 0.14668828738929773,
      "grad_norm": 0.0,
      "learning_rate": 1.929496611869098e-05,
      "loss": 1.0957,
      "step": 1176
    },
    {
      "epoch": 0.14681302232755394,
      "grad_norm": 0.0,
      "learning_rate": 1.9293475242268224e-05,
      "loss": 1.1206,
      "step": 1177
    },
    {
      "epoch": 0.14693775726581015,
      "grad_norm": 0.0,
      "learning_rate": 1.9291982848916025e-05,
      "loss": 1.1448,
      "step": 1178
    },
    {
      "epoch": 0.14706249220406636,
      "grad_norm": 0.0,
      "learning_rate": 1.929048893887797e-05,
      "loss": 1.0689,
      "step": 1179
    },
    {
      "epoch": 0.14718722714232257,
      "grad_norm": 0.0,
      "learning_rate": 1.928899351239791e-05,
      "loss": 1.0981,
      "step": 1180
    },
    {
      "epoch": 0.14731196208057876,
      "grad_norm": 0.0,
      "learning_rate": 1.9287496569719937e-05,
      "loss": 1.1598,
      "step": 1181
    },
    {
      "epoch": 0.14743669701883497,
      "grad_norm": 0.0,
      "learning_rate": 1.9285998111088384e-05,
      "loss": 1.0743,
      "step": 1182
    },
    {
      "epoch": 0.14756143195709118,
      "grad_norm": 0.0,
      "learning_rate": 1.928449813674784e-05,
      "loss": 1.1013,
      "step": 1183
    },
    {
      "epoch": 0.1476861668953474,
      "grad_norm": 0.0,
      "learning_rate": 1.9282996646943146e-05,
      "loss": 1.1148,
      "step": 1184
    },
    {
      "epoch": 0.1478109018336036,
      "grad_norm": 0.0,
      "learning_rate": 1.9281493641919368e-05,
      "loss": 1.069,
      "step": 1185
    },
    {
      "epoch": 0.1479356367718598,
      "grad_norm": 0.0,
      "learning_rate": 1.9279989121921846e-05,
      "loss": 1.1031,
      "step": 1186
    },
    {
      "epoch": 0.148060371710116,
      "grad_norm": 0.0,
      "learning_rate": 1.9278483087196155e-05,
      "loss": 1.1352,
      "step": 1187
    },
    {
      "epoch": 0.1481851066483722,
      "grad_norm": 0.0,
      "learning_rate": 1.927697553798811e-05,
      "loss": 1.0962,
      "step": 1188
    },
    {
      "epoch": 0.1483098415866284,
      "grad_norm": 0.0,
      "learning_rate": 1.9275466474543787e-05,
      "loss": 1.1633,
      "step": 1189
    },
    {
      "epoch": 0.14843457652488462,
      "grad_norm": 0.0,
      "learning_rate": 1.9273955897109504e-05,
      "loss": 1.1169,
      "step": 1190
    },
    {
      "epoch": 0.14855931146314083,
      "grad_norm": 0.0,
      "learning_rate": 1.927244380593182e-05,
      "loss": 1.1146,
      "step": 1191
    },
    {
      "epoch": 0.14868404640139704,
      "grad_norm": 0.0,
      "learning_rate": 1.9270930201257557e-05,
      "loss": 1.1356,
      "step": 1192
    },
    {
      "epoch": 0.14880878133965322,
      "grad_norm": 0.0,
      "learning_rate": 1.9269415083333758e-05,
      "loss": 1.113,
      "step": 1193
    },
    {
      "epoch": 0.14893351627790943,
      "grad_norm": 0.0,
      "learning_rate": 1.9267898452407744e-05,
      "loss": 1.1076,
      "step": 1194
    },
    {
      "epoch": 0.14905825121616564,
      "grad_norm": 0.0,
      "learning_rate": 1.9266380308727054e-05,
      "loss": 1.0924,
      "step": 1195
    },
    {
      "epoch": 0.14918298615442185,
      "grad_norm": 0.0,
      "learning_rate": 1.9264860652539498e-05,
      "loss": 1.1425,
      "step": 1196
    },
    {
      "epoch": 0.14930772109267806,
      "grad_norm": 0.0,
      "learning_rate": 1.9263339484093115e-05,
      "loss": 1.0823,
      "step": 1197
    },
    {
      "epoch": 0.14943245603093427,
      "grad_norm": 0.0,
      "learning_rate": 1.92618168036362e-05,
      "loss": 1.1124,
      "step": 1198
    },
    {
      "epoch": 0.14955719096919046,
      "grad_norm": 0.0,
      "learning_rate": 1.9260292611417296e-05,
      "loss": 1.1527,
      "step": 1199
    },
    {
      "epoch": 0.14968192590744667,
      "grad_norm": 0.0,
      "learning_rate": 1.925876690768519e-05,
      "loss": 1.0595,
      "step": 1200
    },
    {
      "epoch": 0.14980666084570288,
      "grad_norm": 0.0,
      "learning_rate": 1.9257239692688907e-05,
      "loss": 1.1128,
      "step": 1201
    },
    {
      "epoch": 0.1499313957839591,
      "grad_norm": 0.0,
      "learning_rate": 1.9255710966677733e-05,
      "loss": 1.1177,
      "step": 1202
    },
    {
      "epoch": 0.1500561307222153,
      "grad_norm": 0.0,
      "learning_rate": 1.9254180729901194e-05,
      "loss": 1.0519,
      "step": 1203
    },
    {
      "epoch": 0.1501808656604715,
      "grad_norm": 0.0,
      "learning_rate": 1.9252648982609065e-05,
      "loss": 1.0798,
      "step": 1204
    },
    {
      "epoch": 0.1503056005987277,
      "grad_norm": 0.0,
      "learning_rate": 1.9251115725051367e-05,
      "loss": 1.1395,
      "step": 1205
    },
    {
      "epoch": 0.1504303355369839,
      "grad_norm": 0.0,
      "learning_rate": 1.9249580957478357e-05,
      "loss": 1.1436,
      "step": 1206
    },
    {
      "epoch": 0.1505550704752401,
      "grad_norm": 0.0,
      "learning_rate": 1.9248044680140558e-05,
      "loss": 1.1339,
      "step": 1207
    },
    {
      "epoch": 0.15067980541349632,
      "grad_norm": 0.0,
      "learning_rate": 1.9246506893288725e-05,
      "loss": 1.1102,
      "step": 1208
    },
    {
      "epoch": 0.15080454035175253,
      "grad_norm": 0.0,
      "learning_rate": 1.9244967597173863e-05,
      "loss": 1.1331,
      "step": 1209
    },
    {
      "epoch": 0.15092927529000874,
      "grad_norm": 0.0,
      "learning_rate": 1.9243426792047225e-05,
      "loss": 1.0789,
      "step": 1210
    },
    {
      "epoch": 0.15105401022826492,
      "grad_norm": 0.0,
      "learning_rate": 1.924188447816031e-05,
      "loss": 1.0248,
      "step": 1211
    },
    {
      "epoch": 0.15117874516652113,
      "grad_norm": 0.0,
      "learning_rate": 1.924034065576486e-05,
      "loss": 1.1033,
      "step": 1212
    },
    {
      "epoch": 0.15130348010477734,
      "grad_norm": 0.0,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 1.071,
      "step": 1213
    },
    {
      "epoch": 0.15142821504303355,
      "grad_norm": 0.0,
      "learning_rate": 1.923724848645657e-05,
      "loss": 1.0978,
      "step": 1214
    },
    {
      "epoch": 0.15155294998128976,
      "grad_norm": 0.0,
      "learning_rate": 1.923570014004845e-05,
      "loss": 1.0932,
      "step": 1215
    },
    {
      "epoch": 0.15167768491954597,
      "grad_norm": 0.0,
      "learning_rate": 1.9234150286141238e-05,
      "loss": 1.1018,
      "step": 1216
    },
    {
      "epoch": 0.15180241985780216,
      "grad_norm": 0.0,
      "learning_rate": 1.9232598924987904e-05,
      "loss": 1.1327,
      "step": 1217
    },
    {
      "epoch": 0.15192715479605837,
      "grad_norm": 0.0,
      "learning_rate": 1.923104605684168e-05,
      "loss": 1.1135,
      "step": 1218
    },
    {
      "epoch": 0.15205188973431458,
      "grad_norm": 0.0,
      "learning_rate": 1.9229491681956016e-05,
      "loss": 1.0512,
      "step": 1219
    },
    {
      "epoch": 0.1521766246725708,
      "grad_norm": 0.0,
      "learning_rate": 1.922793580058464e-05,
      "loss": 1.0743,
      "step": 1220
    },
    {
      "epoch": 0.152301359610827,
      "grad_norm": 0.0,
      "learning_rate": 1.9226378412981507e-05,
      "loss": 1.1266,
      "step": 1221
    },
    {
      "epoch": 0.1524260945490832,
      "grad_norm": 0.0,
      "learning_rate": 1.9224819519400825e-05,
      "loss": 1.0974,
      "step": 1222
    },
    {
      "epoch": 0.1525508294873394,
      "grad_norm": 0.0,
      "learning_rate": 1.9223259120097035e-05,
      "loss": 1.1084,
      "step": 1223
    },
    {
      "epoch": 0.1526755644255956,
      "grad_norm": 0.0,
      "learning_rate": 1.9221697215324843e-05,
      "loss": 1.1021,
      "step": 1224
    },
    {
      "epoch": 0.1528002993638518,
      "grad_norm": 0.0,
      "learning_rate": 1.9220133805339186e-05,
      "loss": 1.0809,
      "step": 1225
    },
    {
      "epoch": 0.15292503430210802,
      "grad_norm": 0.0,
      "learning_rate": 1.9218568890395252e-05,
      "loss": 1.1212,
      "step": 1226
    },
    {
      "epoch": 0.15304976924036423,
      "grad_norm": 0.0,
      "learning_rate": 1.921700247074848e-05,
      "loss": 1.1368,
      "step": 1227
    },
    {
      "epoch": 0.15317450417862044,
      "grad_norm": 0.0,
      "learning_rate": 1.9215434546654544e-05,
      "loss": 1.1316,
      "step": 1228
    },
    {
      "epoch": 0.15329923911687662,
      "grad_norm": 0.0,
      "learning_rate": 1.921386511836937e-05,
      "loss": 1.1229,
      "step": 1229
    },
    {
      "epoch": 0.15342397405513283,
      "grad_norm": 0.0,
      "learning_rate": 1.921229418614913e-05,
      "loss": 1.1356,
      "step": 1230
    },
    {
      "epoch": 0.15354870899338904,
      "grad_norm": 0.0,
      "learning_rate": 1.9210721750250237e-05,
      "loss": 1.0967,
      "step": 1231
    },
    {
      "epoch": 0.15367344393164525,
      "grad_norm": 0.0,
      "learning_rate": 1.920914781092935e-05,
      "loss": 1.1164,
      "step": 1232
    },
    {
      "epoch": 0.15379817886990146,
      "grad_norm": 0.0,
      "learning_rate": 1.9207572368443386e-05,
      "loss": 1.1003,
      "step": 1233
    },
    {
      "epoch": 0.15392291380815767,
      "grad_norm": 0.0,
      "learning_rate": 1.9205995423049482e-05,
      "loss": 1.1532,
      "step": 1234
    },
    {
      "epoch": 0.15404764874641388,
      "grad_norm": 0.0,
      "learning_rate": 1.920441697500505e-05,
      "loss": 1.0597,
      "step": 1235
    },
    {
      "epoch": 0.15417238368467007,
      "grad_norm": 0.0,
      "learning_rate": 1.9202837024567723e-05,
      "loss": 1.114,
      "step": 1236
    },
    {
      "epoch": 0.15429711862292628,
      "grad_norm": 0.0,
      "learning_rate": 1.9201255571995392e-05,
      "loss": 1.0953,
      "step": 1237
    },
    {
      "epoch": 0.1544218535611825,
      "grad_norm": 0.0,
      "learning_rate": 1.9199672617546187e-05,
      "loss": 1.0832,
      "step": 1238
    },
    {
      "epoch": 0.1545465884994387,
      "grad_norm": 0.0,
      "learning_rate": 1.9198088161478492e-05,
      "loss": 1.0857,
      "step": 1239
    },
    {
      "epoch": 0.1546713234376949,
      "grad_norm": 0.0,
      "learning_rate": 1.9196502204050925e-05,
      "loss": 1.1247,
      "step": 1240
    },
    {
      "epoch": 0.15479605837595112,
      "grad_norm": 0.0,
      "learning_rate": 1.9194914745522354e-05,
      "loss": 1.1256,
      "step": 1241
    },
    {
      "epoch": 0.1549207933142073,
      "grad_norm": 0.0,
      "learning_rate": 1.9193325786151895e-05,
      "loss": 1.1067,
      "step": 1242
    },
    {
      "epoch": 0.1550455282524635,
      "grad_norm": 0.0,
      "learning_rate": 1.9191735326198907e-05,
      "loss": 1.0957,
      "step": 1243
    },
    {
      "epoch": 0.15517026319071972,
      "grad_norm": 0.0,
      "learning_rate": 1.919014336592299e-05,
      "loss": 1.1548,
      "step": 1244
    },
    {
      "epoch": 0.15529499812897593,
      "grad_norm": 0.0,
      "learning_rate": 1.9188549905583993e-05,
      "loss": 1.0478,
      "step": 1245
    },
    {
      "epoch": 0.15541973306723214,
      "grad_norm": 0.0,
      "learning_rate": 1.918695494544201e-05,
      "loss": 1.085,
      "step": 1246
    },
    {
      "epoch": 0.15554446800548835,
      "grad_norm": 0.0,
      "learning_rate": 1.9185358485757376e-05,
      "loss": 1.0943,
      "step": 1247
    },
    {
      "epoch": 0.15566920294374453,
      "grad_norm": 0.0,
      "learning_rate": 1.918376052679068e-05,
      "loss": 1.1112,
      "step": 1248
    },
    {
      "epoch": 0.15579393788200074,
      "grad_norm": 0.0,
      "learning_rate": 1.9182161068802742e-05,
      "loss": 1.0272,
      "step": 1249
    },
    {
      "epoch": 0.15591867282025695,
      "grad_norm": 0.0,
      "learning_rate": 1.9180560112054637e-05,
      "loss": 1.0465,
      "step": 1250
    },
    {
      "epoch": 0.15604340775851316,
      "grad_norm": 0.0,
      "learning_rate": 1.917895765680768e-05,
      "loss": 1.0991,
      "step": 1251
    },
    {
      "epoch": 0.15616814269676937,
      "grad_norm": 0.0,
      "learning_rate": 1.9177353703323435e-05,
      "loss": 1.0752,
      "step": 1252
    },
    {
      "epoch": 0.15629287763502558,
      "grad_norm": 0.0,
      "learning_rate": 1.9175748251863707e-05,
      "loss": 1.0523,
      "step": 1253
    },
    {
      "epoch": 0.15641761257328177,
      "grad_norm": 0.0,
      "learning_rate": 1.917414130269054e-05,
      "loss": 1.0725,
      "step": 1254
    },
    {
      "epoch": 0.15654234751153798,
      "grad_norm": 0.0,
      "learning_rate": 1.917253285606624e-05,
      "loss": 1.1162,
      "step": 1255
    },
    {
      "epoch": 0.1566670824497942,
      "grad_norm": 0.0,
      "learning_rate": 1.9170922912253335e-05,
      "loss": 1.0798,
      "step": 1256
    },
    {
      "epoch": 0.1567918173880504,
      "grad_norm": 0.0,
      "learning_rate": 1.916931147151461e-05,
      "loss": 1.0971,
      "step": 1257
    },
    {
      "epoch": 0.1569165523263066,
      "grad_norm": 0.0,
      "learning_rate": 1.9167698534113105e-05,
      "loss": 1.1087,
      "step": 1258
    },
    {
      "epoch": 0.15704128726456282,
      "grad_norm": 0.0,
      "learning_rate": 1.9166084100312078e-05,
      "loss": 1.1137,
      "step": 1259
    },
    {
      "epoch": 0.157166022202819,
      "grad_norm": 0.0,
      "learning_rate": 1.916446817037505e-05,
      "loss": 1.0835,
      "step": 1260
    },
    {
      "epoch": 0.1572907571410752,
      "grad_norm": 0.0,
      "learning_rate": 1.916285074456578e-05,
      "loss": 1.1114,
      "step": 1261
    },
    {
      "epoch": 0.15741549207933142,
      "grad_norm": 0.0,
      "learning_rate": 1.9161231823148275e-05,
      "loss": 1.1233,
      "step": 1262
    },
    {
      "epoch": 0.15754022701758763,
      "grad_norm": 0.0,
      "learning_rate": 1.9159611406386786e-05,
      "loss": 1.0485,
      "step": 1263
    },
    {
      "epoch": 0.15766496195584384,
      "grad_norm": 0.0,
      "learning_rate": 1.9157989494545802e-05,
      "loss": 1.1349,
      "step": 1264
    },
    {
      "epoch": 0.15778969689410005,
      "grad_norm": 0.0,
      "learning_rate": 1.9156366087890062e-05,
      "loss": 1.1579,
      "step": 1265
    },
    {
      "epoch": 0.15791443183235623,
      "grad_norm": 0.0,
      "learning_rate": 1.9154741186684544e-05,
      "loss": 1.0762,
      "step": 1266
    },
    {
      "epoch": 0.15803916677061244,
      "grad_norm": 0.0,
      "learning_rate": 1.9153114791194475e-05,
      "loss": 1.1057,
      "step": 1267
    },
    {
      "epoch": 0.15816390170886865,
      "grad_norm": 0.0,
      "learning_rate": 1.9151486901685322e-05,
      "loss": 1.0982,
      "step": 1268
    },
    {
      "epoch": 0.15828863664712486,
      "grad_norm": 0.0,
      "learning_rate": 1.91498575184228e-05,
      "loss": 1.1293,
      "step": 1269
    },
    {
      "epoch": 0.15841337158538107,
      "grad_norm": 0.0,
      "learning_rate": 1.9148226641672864e-05,
      "loss": 1.1032,
      "step": 1270
    },
    {
      "epoch": 0.15853810652363728,
      "grad_norm": 0.0,
      "learning_rate": 1.9146594271701716e-05,
      "loss": 1.1407,
      "step": 1271
    },
    {
      "epoch": 0.15866284146189347,
      "grad_norm": 0.0,
      "learning_rate": 1.9144960408775794e-05,
      "loss": 1.133,
      "step": 1272
    },
    {
      "epoch": 0.15878757640014968,
      "grad_norm": 0.0,
      "learning_rate": 1.9143325053161795e-05,
      "loss": 1.1591,
      "step": 1273
    },
    {
      "epoch": 0.1589123113384059,
      "grad_norm": 0.0,
      "learning_rate": 1.9141688205126644e-05,
      "loss": 1.1544,
      "step": 1274
    },
    {
      "epoch": 0.1590370462766621,
      "grad_norm": 0.0,
      "learning_rate": 1.9140049864937517e-05,
      "loss": 1.0991,
      "step": 1275
    },
    {
      "epoch": 0.1591617812149183,
      "grad_norm": 0.0,
      "learning_rate": 1.9138410032861833e-05,
      "loss": 1.1302,
      "step": 1276
    },
    {
      "epoch": 0.15928651615317452,
      "grad_norm": 0.0,
      "learning_rate": 1.9136768709167254e-05,
      "loss": 1.1472,
      "step": 1277
    },
    {
      "epoch": 0.1594112510914307,
      "grad_norm": 0.0,
      "learning_rate": 1.9135125894121685e-05,
      "loss": 1.123,
      "step": 1278
    },
    {
      "epoch": 0.1595359860296869,
      "grad_norm": 0.0,
      "learning_rate": 1.9133481587993274e-05,
      "loss": 1.1298,
      "step": 1279
    },
    {
      "epoch": 0.15966072096794312,
      "grad_norm": 0.0,
      "learning_rate": 1.9131835791050417e-05,
      "loss": 1.0552,
      "step": 1280
    },
    {
      "epoch": 0.15978545590619933,
      "grad_norm": 0.0,
      "learning_rate": 1.9130188503561744e-05,
      "loss": 1.1166,
      "step": 1281
    },
    {
      "epoch": 0.15991019084445554,
      "grad_norm": 0.0,
      "learning_rate": 1.9128539725796137e-05,
      "loss": 1.0523,
      "step": 1282
    },
    {
      "epoch": 0.16003492578271175,
      "grad_norm": 0.0,
      "learning_rate": 1.9126889458022713e-05,
      "loss": 1.097,
      "step": 1283
    },
    {
      "epoch": 0.16015966072096793,
      "grad_norm": 0.0,
      "learning_rate": 1.912523770051085e-05,
      "loss": 1.1073,
      "step": 1284
    },
    {
      "epoch": 0.16028439565922414,
      "grad_norm": 0.0,
      "learning_rate": 1.9123584453530145e-05,
      "loss": 1.0787,
      "step": 1285
    },
    {
      "epoch": 0.16040913059748035,
      "grad_norm": 0.0,
      "learning_rate": 1.912192971735045e-05,
      "loss": 1.0543,
      "step": 1286
    },
    {
      "epoch": 0.16053386553573656,
      "grad_norm": 0.0,
      "learning_rate": 1.9120273492241864e-05,
      "loss": 1.0903,
      "step": 1287
    },
    {
      "epoch": 0.16065860047399277,
      "grad_norm": 0.0,
      "learning_rate": 1.911861577847473e-05,
      "loss": 1.1365,
      "step": 1288
    },
    {
      "epoch": 0.16078333541224898,
      "grad_norm": 0.0,
      "learning_rate": 1.911695657631961e-05,
      "loss": 1.0463,
      "step": 1289
    },
    {
      "epoch": 0.16090807035050517,
      "grad_norm": 0.0,
      "learning_rate": 1.911529588604735e-05,
      "loss": 1.1128,
      "step": 1290
    },
    {
      "epoch": 0.16103280528876138,
      "grad_norm": 0.0,
      "learning_rate": 1.9113633707929002e-05,
      "loss": 1.0598,
      "step": 1291
    },
    {
      "epoch": 0.1611575402270176,
      "grad_norm": 0.0,
      "learning_rate": 1.911197004223588e-05,
      "loss": 1.0971,
      "step": 1292
    },
    {
      "epoch": 0.1612822751652738,
      "grad_norm": 0.0,
      "learning_rate": 1.911030488923953e-05,
      "loss": 1.0961,
      "step": 1293
    },
    {
      "epoch": 0.16140701010353,
      "grad_norm": 0.0,
      "learning_rate": 1.910863824921176e-05,
      "loss": 1.1146,
      "step": 1294
    },
    {
      "epoch": 0.16153174504178622,
      "grad_norm": 0.0,
      "learning_rate": 1.91069701224246e-05,
      "loss": 1.09,
      "step": 1295
    },
    {
      "epoch": 0.1616564799800424,
      "grad_norm": 0.0,
      "learning_rate": 1.9105300509150327e-05,
      "loss": 1.1108,
      "step": 1296
    },
    {
      "epoch": 0.1617812149182986,
      "grad_norm": 0.0,
      "learning_rate": 1.9103629409661468e-05,
      "loss": 1.0477,
      "step": 1297
    },
    {
      "epoch": 0.16190594985655482,
      "grad_norm": 0.0,
      "learning_rate": 1.910195682423079e-05,
      "loss": 1.0647,
      "step": 1298
    },
    {
      "epoch": 0.16203068479481103,
      "grad_norm": 0.0,
      "learning_rate": 1.91002827531313e-05,
      "loss": 1.1565,
      "step": 1299
    },
    {
      "epoch": 0.16215541973306724,
      "grad_norm": 0.0,
      "learning_rate": 1.9098607196636245e-05,
      "loss": 1.0322,
      "step": 1300
    },
    {
      "epoch": 0.16228015467132345,
      "grad_norm": 0.0,
      "learning_rate": 1.9096930155019124e-05,
      "loss": 1.0819,
      "step": 1301
    },
    {
      "epoch": 0.16240488960957963,
      "grad_norm": 0.0,
      "learning_rate": 1.9095251628553666e-05,
      "loss": 1.1268,
      "step": 1302
    },
    {
      "epoch": 0.16252962454783584,
      "grad_norm": 0.0,
      "learning_rate": 1.9093571617513853e-05,
      "loss": 1.0998,
      "step": 1303
    },
    {
      "epoch": 0.16265435948609205,
      "grad_norm": 0.0,
      "learning_rate": 1.9091890122173903e-05,
      "loss": 1.1774,
      "step": 1304
    },
    {
      "epoch": 0.16277909442434826,
      "grad_norm": 0.0,
      "learning_rate": 1.9090207142808283e-05,
      "loss": 1.0237,
      "step": 1305
    },
    {
      "epoch": 0.16290382936260447,
      "grad_norm": 0.0,
      "learning_rate": 1.9088522679691694e-05,
      "loss": 1.0875,
      "step": 1306
    },
    {
      "epoch": 0.16302856430086068,
      "grad_norm": 0.0,
      "learning_rate": 1.9086836733099084e-05,
      "loss": 1.1166,
      "step": 1307
    },
    {
      "epoch": 0.16315329923911687,
      "grad_norm": 0.0,
      "learning_rate": 1.908514930330564e-05,
      "loss": 1.1014,
      "step": 1308
    },
    {
      "epoch": 0.16327803417737308,
      "grad_norm": 0.0,
      "learning_rate": 1.9083460390586796e-05,
      "loss": 1.1115,
      "step": 1309
    },
    {
      "epoch": 0.1634027691156293,
      "grad_norm": 0.0,
      "learning_rate": 1.908176999521822e-05,
      "loss": 1.0933,
      "step": 1310
    },
    {
      "epoch": 0.1635275040538855,
      "grad_norm": 0.0,
      "learning_rate": 1.9080078117475833e-05,
      "loss": 1.1218,
      "step": 1311
    },
    {
      "epoch": 0.1636522389921417,
      "grad_norm": 0.0,
      "learning_rate": 1.907838475763579e-05,
      "loss": 1.0502,
      "step": 1312
    },
    {
      "epoch": 0.16377697393039792,
      "grad_norm": 0.0,
      "learning_rate": 1.907668991597449e-05,
      "loss": 1.1126,
      "step": 1313
    },
    {
      "epoch": 0.1639017088686541,
      "grad_norm": 0.0,
      "learning_rate": 1.9074993592768575e-05,
      "loss": 1.1447,
      "step": 1314
    },
    {
      "epoch": 0.1640264438069103,
      "grad_norm": 0.0,
      "learning_rate": 1.907329578829492e-05,
      "loss": 1.0973,
      "step": 1315
    },
    {
      "epoch": 0.16415117874516652,
      "grad_norm": 0.0,
      "learning_rate": 1.907159650283066e-05,
      "loss": 1.0943,
      "step": 1316
    },
    {
      "epoch": 0.16427591368342273,
      "grad_norm": 0.0,
      "learning_rate": 1.9069895736653158e-05,
      "loss": 1.1549,
      "step": 1317
    },
    {
      "epoch": 0.16440064862167894,
      "grad_norm": 0.0,
      "learning_rate": 1.906819349004002e-05,
      "loss": 1.1193,
      "step": 1318
    },
    {
      "epoch": 0.16452538355993515,
      "grad_norm": 0.0,
      "learning_rate": 1.9066489763269098e-05,
      "loss": 1.1334,
      "step": 1319
    },
    {
      "epoch": 0.16465011849819133,
      "grad_norm": 0.0,
      "learning_rate": 1.906478455661848e-05,
      "loss": 1.0882,
      "step": 1320
    },
    {
      "epoch": 0.16477485343644754,
      "grad_norm": 0.0,
      "learning_rate": 1.9063077870366504e-05,
      "loss": 1.1157,
      "step": 1321
    },
    {
      "epoch": 0.16489958837470375,
      "grad_norm": 0.0,
      "learning_rate": 1.9061369704791736e-05,
      "loss": 1.1333,
      "step": 1322
    },
    {
      "epoch": 0.16502432331295996,
      "grad_norm": 0.0,
      "learning_rate": 1.9059660060172994e-05,
      "loss": 1.0904,
      "step": 1323
    },
    {
      "epoch": 0.16514905825121617,
      "grad_norm": 0.0,
      "learning_rate": 1.9057948936789343e-05,
      "loss": 1.0921,
      "step": 1324
    },
    {
      "epoch": 0.16527379318947238,
      "grad_norm": 0.0,
      "learning_rate": 1.9056236334920074e-05,
      "loss": 1.1128,
      "step": 1325
    },
    {
      "epoch": 0.16539852812772857,
      "grad_norm": 0.0,
      "learning_rate": 1.9054522254844728e-05,
      "loss": 1.1062,
      "step": 1326
    },
    {
      "epoch": 0.16552326306598478,
      "grad_norm": 0.0,
      "learning_rate": 1.905280669684309e-05,
      "loss": 1.1171,
      "step": 1327
    },
    {
      "epoch": 0.165647998004241,
      "grad_norm": 0.0,
      "learning_rate": 1.9051089661195177e-05,
      "loss": 1.1261,
      "step": 1328
    },
    {
      "epoch": 0.1657727329424972,
      "grad_norm": 0.0,
      "learning_rate": 1.9049371148181253e-05,
      "loss": 1.1324,
      "step": 1329
    },
    {
      "epoch": 0.1658974678807534,
      "grad_norm": 0.0,
      "learning_rate": 1.9047651158081827e-05,
      "loss": 1.0923,
      "step": 1330
    },
    {
      "epoch": 0.16602220281900962,
      "grad_norm": 0.0,
      "learning_rate": 1.904592969117764e-05,
      "loss": 1.0813,
      "step": 1331
    },
    {
      "epoch": 0.1661469377572658,
      "grad_norm": 0.0,
      "learning_rate": 1.9044206747749684e-05,
      "loss": 1.1255,
      "step": 1332
    },
    {
      "epoch": 0.166271672695522,
      "grad_norm": 0.0,
      "learning_rate": 1.9042482328079178e-05,
      "loss": 1.0924,
      "step": 1333
    },
    {
      "epoch": 0.16639640763377822,
      "grad_norm": 0.0,
      "learning_rate": 1.90407564324476e-05,
      "loss": 1.1175,
      "step": 1334
    },
    {
      "epoch": 0.16652114257203443,
      "grad_norm": 0.0,
      "learning_rate": 1.9039029061136657e-05,
      "loss": 1.1462,
      "step": 1335
    },
    {
      "epoch": 0.16664587751029064,
      "grad_norm": 0.0,
      "learning_rate": 1.90373002144283e-05,
      "loss": 1.0465,
      "step": 1336
    },
    {
      "epoch": 0.16677061244854685,
      "grad_norm": 0.0,
      "learning_rate": 1.9035569892604715e-05,
      "loss": 1.1395,
      "step": 1337
    },
    {
      "epoch": 0.16689534738680303,
      "grad_norm": 0.0,
      "learning_rate": 1.903383809594834e-05,
      "loss": 1.1051,
      "step": 1338
    },
    {
      "epoch": 0.16702008232505924,
      "grad_norm": 0.0,
      "learning_rate": 1.9032104824741843e-05,
      "loss": 1.082,
      "step": 1339
    },
    {
      "epoch": 0.16714481726331545,
      "grad_norm": 0.0,
      "learning_rate": 1.9030370079268143e-05,
      "loss": 1.1196,
      "step": 1340
    },
    {
      "epoch": 0.16726955220157166,
      "grad_norm": 0.0,
      "learning_rate": 1.902863385981039e-05,
      "loss": 1.1096,
      "step": 1341
    },
    {
      "epoch": 0.16739428713982787,
      "grad_norm": 0.0,
      "learning_rate": 1.9026896166651986e-05,
      "loss": 1.1545,
      "step": 1342
    },
    {
      "epoch": 0.16751902207808408,
      "grad_norm": 0.0,
      "learning_rate": 1.9025157000076556e-05,
      "loss": 1.0746,
      "step": 1343
    },
    {
      "epoch": 0.16764375701634027,
      "grad_norm": 0.0,
      "learning_rate": 1.9023416360367984e-05,
      "loss": 1.1173,
      "step": 1344
    },
    {
      "epoch": 0.16776849195459648,
      "grad_norm": 0.0,
      "learning_rate": 1.902167424781038e-05,
      "loss": 1.036,
      "step": 1345
    },
    {
      "epoch": 0.1678932268928527,
      "grad_norm": 0.0,
      "learning_rate": 1.90199306626881e-05,
      "loss": 1.0863,
      "step": 1346
    },
    {
      "epoch": 0.1680179618311089,
      "grad_norm": 0.0,
      "learning_rate": 1.9018185605285752e-05,
      "loss": 1.0074,
      "step": 1347
    },
    {
      "epoch": 0.1681426967693651,
      "grad_norm": 0.0,
      "learning_rate": 1.901643907588816e-05,
      "loss": 1.107,
      "step": 1348
    },
    {
      "epoch": 0.16826743170762132,
      "grad_norm": 0.0,
      "learning_rate": 1.9014691074780414e-05,
      "loss": 1.1003,
      "step": 1349
    },
    {
      "epoch": 0.1683921666458775,
      "grad_norm": 0.0,
      "learning_rate": 1.901294160224782e-05,
      "loss": 1.0199,
      "step": 1350
    },
    {
      "epoch": 0.1685169015841337,
      "grad_norm": 0.0,
      "learning_rate": 1.9011190658575948e-05,
      "loss": 1.1606,
      "step": 1351
    },
    {
      "epoch": 0.16864163652238992,
      "grad_norm": 0.0,
      "learning_rate": 1.900943824405058e-05,
      "loss": 1.0434,
      "step": 1352
    },
    {
      "epoch": 0.16876637146064613,
      "grad_norm": 0.0,
      "learning_rate": 1.900768435895777e-05,
      "loss": 1.1013,
      "step": 1353
    },
    {
      "epoch": 0.16889110639890234,
      "grad_norm": 0.0,
      "learning_rate": 1.9005929003583793e-05,
      "loss": 1.1176,
      "step": 1354
    },
    {
      "epoch": 0.16901584133715855,
      "grad_norm": 0.0,
      "learning_rate": 1.900417217821516e-05,
      "loss": 1.0814,
      "step": 1355
    },
    {
      "epoch": 0.16914057627541473,
      "grad_norm": 0.0,
      "learning_rate": 1.9002413883138633e-05,
      "loss": 1.0319,
      "step": 1356
    },
    {
      "epoch": 0.16926531121367094,
      "grad_norm": 0.0,
      "learning_rate": 1.900065411864121e-05,
      "loss": 1.1037,
      "step": 1357
    },
    {
      "epoch": 0.16939004615192715,
      "grad_norm": 0.0,
      "learning_rate": 1.8998892885010134e-05,
      "loss": 1.0065,
      "step": 1358
    },
    {
      "epoch": 0.16951478109018336,
      "grad_norm": 0.0,
      "learning_rate": 1.8997130182532876e-05,
      "loss": 1.0375,
      "step": 1359
    },
    {
      "epoch": 0.16963951602843957,
      "grad_norm": 0.0,
      "learning_rate": 1.899536601149716e-05,
      "loss": 1.047,
      "step": 1360
    },
    {
      "epoch": 0.16976425096669578,
      "grad_norm": 0.0,
      "learning_rate": 1.8993600372190933e-05,
      "loss": 1.0829,
      "step": 1361
    },
    {
      "epoch": 0.16988898590495197,
      "grad_norm": 0.0,
      "learning_rate": 1.89918332649024e-05,
      "loss": 1.1296,
      "step": 1362
    },
    {
      "epoch": 0.17001372084320818,
      "grad_norm": 0.0,
      "learning_rate": 1.8990064689919995e-05,
      "loss": 1.1029,
      "step": 1363
    },
    {
      "epoch": 0.1701384557814644,
      "grad_norm": 0.0,
      "learning_rate": 1.8988294647532395e-05,
      "loss": 1.1035,
      "step": 1364
    },
    {
      "epoch": 0.1702631907197206,
      "grad_norm": 0.0,
      "learning_rate": 1.8986523138028513e-05,
      "loss": 1.0854,
      "step": 1365
    },
    {
      "epoch": 0.1703879256579768,
      "grad_norm": 0.0,
      "learning_rate": 1.898475016169751e-05,
      "loss": 1.0822,
      "step": 1366
    },
    {
      "epoch": 0.17051266059623302,
      "grad_norm": 0.0,
      "learning_rate": 1.8982975718828774e-05,
      "loss": 1.0766,
      "step": 1367
    },
    {
      "epoch": 0.1706373955344892,
      "grad_norm": 0.0,
      "learning_rate": 1.898119980971194e-05,
      "loss": 1.0591,
      "step": 1368
    },
    {
      "epoch": 0.1707621304727454,
      "grad_norm": 0.0,
      "learning_rate": 1.897942243463688e-05,
      "loss": 1.0745,
      "step": 1369
    },
    {
      "epoch": 0.17088686541100162,
      "grad_norm": 0.0,
      "learning_rate": 1.8977643593893716e-05,
      "loss": 1.0724,
      "step": 1370
    },
    {
      "epoch": 0.17101160034925783,
      "grad_norm": 0.0,
      "learning_rate": 1.8975863287772787e-05,
      "loss": 1.0937,
      "step": 1371
    },
    {
      "epoch": 0.17113633528751404,
      "grad_norm": 0.0,
      "learning_rate": 1.897408151656469e-05,
      "loss": 1.0926,
      "step": 1372
    },
    {
      "epoch": 0.17126107022577025,
      "grad_norm": 0.0,
      "learning_rate": 1.8972298280560256e-05,
      "loss": 1.0537,
      "step": 1373
    },
    {
      "epoch": 0.17138580516402643,
      "grad_norm": 0.0,
      "learning_rate": 1.897051358005055e-05,
      "loss": 1.047,
      "step": 1374
    },
    {
      "epoch": 0.17151054010228264,
      "grad_norm": 0.0,
      "learning_rate": 1.8968727415326885e-05,
      "loss": 1.0871,
      "step": 1375
    },
    {
      "epoch": 0.17163527504053885,
      "grad_norm": 0.0,
      "learning_rate": 1.8966939786680806e-05,
      "loss": 1.0905,
      "step": 1376
    },
    {
      "epoch": 0.17176000997879506,
      "grad_norm": 0.0,
      "learning_rate": 1.8965150694404094e-05,
      "loss": 1.0739,
      "step": 1377
    },
    {
      "epoch": 0.17188474491705127,
      "grad_norm": 0.0,
      "learning_rate": 1.896336013878878e-05,
      "loss": 1.0827,
      "step": 1378
    },
    {
      "epoch": 0.17200947985530748,
      "grad_norm": 0.0,
      "learning_rate": 1.8961568120127128e-05,
      "loss": 1.076,
      "step": 1379
    },
    {
      "epoch": 0.17213421479356367,
      "grad_norm": 0.0,
      "learning_rate": 1.8959774638711638e-05,
      "loss": 1.0509,
      "step": 1380
    },
    {
      "epoch": 0.17225894973181988,
      "grad_norm": 0.0,
      "learning_rate": 1.895797969483505e-05,
      "loss": 1.0731,
      "step": 1381
    },
    {
      "epoch": 0.1723836846700761,
      "grad_norm": 0.0,
      "learning_rate": 1.895618328879035e-05,
      "loss": 1.143,
      "step": 1382
    },
    {
      "epoch": 0.1725084196083323,
      "grad_norm": 0.0,
      "learning_rate": 1.8954385420870754e-05,
      "loss": 1.0736,
      "step": 1383
    },
    {
      "epoch": 0.1726331545465885,
      "grad_norm": 0.0,
      "learning_rate": 1.895258609136972e-05,
      "loss": 1.0663,
      "step": 1384
    },
    {
      "epoch": 0.17275788948484472,
      "grad_norm": 0.0,
      "learning_rate": 1.8950785300580935e-05,
      "loss": 1.0704,
      "step": 1385
    },
    {
      "epoch": 0.1728826244231009,
      "grad_norm": 0.0,
      "learning_rate": 1.8948983048798345e-05,
      "loss": 1.031,
      "step": 1386
    },
    {
      "epoch": 0.1730073593613571,
      "grad_norm": 0.0,
      "learning_rate": 1.8947179336316122e-05,
      "loss": 1.1143,
      "step": 1387
    },
    {
      "epoch": 0.17313209429961332,
      "grad_norm": 0.0,
      "learning_rate": 1.894537416342867e-05,
      "loss": 1.0447,
      "step": 1388
    },
    {
      "epoch": 0.17325682923786953,
      "grad_norm": 0.0,
      "learning_rate": 1.894356753043064e-05,
      "loss": 1.1046,
      "step": 1389
    },
    {
      "epoch": 0.17338156417612574,
      "grad_norm": 0.0,
      "learning_rate": 1.8941759437616933e-05,
      "loss": 1.0935,
      "step": 1390
    },
    {
      "epoch": 0.17350629911438195,
      "grad_norm": 0.0,
      "learning_rate": 1.8939949885282657e-05,
      "loss": 1.0546,
      "step": 1391
    },
    {
      "epoch": 0.17363103405263813,
      "grad_norm": 0.0,
      "learning_rate": 1.8938138873723187e-05,
      "loss": 1.1051,
      "step": 1392
    },
    {
      "epoch": 0.17375576899089434,
      "grad_norm": 0.0,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 1.1052,
      "step": 1393
    },
    {
      "epoch": 0.17388050392915055,
      "grad_norm": 0.0,
      "learning_rate": 1.893451247411131e-05,
      "loss": 1.0769,
      "step": 1394
    },
    {
      "epoch": 0.17400523886740676,
      "grad_norm": 0.0,
      "learning_rate": 1.893269708665082e-05,
      "loss": 1.0574,
      "step": 1395
    },
    {
      "epoch": 0.17412997380566297,
      "grad_norm": 0.0,
      "learning_rate": 1.8930880241148973e-05,
      "loss": 1.1541,
      "step": 1396
    },
    {
      "epoch": 0.17425470874391918,
      "grad_norm": 0.0,
      "learning_rate": 1.8929061937902328e-05,
      "loss": 1.0252,
      "step": 1397
    },
    {
      "epoch": 0.17437944368217537,
      "grad_norm": 0.0,
      "learning_rate": 1.892724217720767e-05,
      "loss": 1.0589,
      "step": 1398
    },
    {
      "epoch": 0.17450417862043158,
      "grad_norm": 0.0,
      "learning_rate": 1.8925420959362037e-05,
      "loss": 1.0907,
      "step": 1399
    },
    {
      "epoch": 0.1746289135586878,
      "grad_norm": 0.0,
      "learning_rate": 1.8923598284662695e-05,
      "loss": 1.1174,
      "step": 1400
    },
    {
      "epoch": 0.174753648496944,
      "grad_norm": 0.0,
      "learning_rate": 1.8921774153407148e-05,
      "loss": 1.04,
      "step": 1401
    },
    {
      "epoch": 0.1748783834352002,
      "grad_norm": 0.0,
      "learning_rate": 1.8919948565893144e-05,
      "loss": 1.0381,
      "step": 1402
    },
    {
      "epoch": 0.17500311837345642,
      "grad_norm": 0.0,
      "learning_rate": 1.891812152241866e-05,
      "loss": 1.0702,
      "step": 1403
    },
    {
      "epoch": 0.1751278533117126,
      "grad_norm": 0.0,
      "learning_rate": 1.8916293023281925e-05,
      "loss": 1.0489,
      "step": 1404
    },
    {
      "epoch": 0.1752525882499688,
      "grad_norm": 0.0,
      "learning_rate": 1.8914463068781388e-05,
      "loss": 1.0726,
      "step": 1405
    },
    {
      "epoch": 0.17537732318822502,
      "grad_norm": 0.0,
      "learning_rate": 1.8912631659215745e-05,
      "loss": 1.0724,
      "step": 1406
    },
    {
      "epoch": 0.17550205812648123,
      "grad_norm": 0.0,
      "learning_rate": 1.891079879488393e-05,
      "loss": 1.0695,
      "step": 1407
    },
    {
      "epoch": 0.17562679306473744,
      "grad_norm": 0.0,
      "learning_rate": 1.890896447608511e-05,
      "loss": 1.0921,
      "step": 1408
    },
    {
      "epoch": 0.17575152800299365,
      "grad_norm": 0.0,
      "learning_rate": 1.8907128703118696e-05,
      "loss": 1.1122,
      "step": 1409
    },
    {
      "epoch": 0.17587626294124983,
      "grad_norm": 0.0,
      "learning_rate": 1.8905291476284328e-05,
      "loss": 1.0702,
      "step": 1410
    },
    {
      "epoch": 0.17600099787950604,
      "grad_norm": 0.0,
      "learning_rate": 1.8903452795881893e-05,
      "loss": 1.1124,
      "step": 1411
    },
    {
      "epoch": 0.17612573281776225,
      "grad_norm": 0.0,
      "learning_rate": 1.89016126622115e-05,
      "loss": 1.0876,
      "step": 1412
    },
    {
      "epoch": 0.17625046775601846,
      "grad_norm": 0.0,
      "learning_rate": 1.8899771075573522e-05,
      "loss": 1.1089,
      "step": 1413
    },
    {
      "epoch": 0.17637520269427467,
      "grad_norm": 0.0,
      "learning_rate": 1.8897928036268535e-05,
      "loss": 1.0761,
      "step": 1414
    },
    {
      "epoch": 0.17649993763253088,
      "grad_norm": 0.0,
      "learning_rate": 1.8896083544597377e-05,
      "loss": 1.1004,
      "step": 1415
    },
    {
      "epoch": 0.17662467257078707,
      "grad_norm": 0.0,
      "learning_rate": 1.889423760086112e-05,
      "loss": 1.1368,
      "step": 1416
    },
    {
      "epoch": 0.17674940750904328,
      "grad_norm": 0.0,
      "learning_rate": 1.8892390205361063e-05,
      "loss": 1.0333,
      "step": 1417
    },
    {
      "epoch": 0.1768741424472995,
      "grad_norm": 0.0,
      "learning_rate": 1.889054135839875e-05,
      "loss": 1.0995,
      "step": 1418
    },
    {
      "epoch": 0.1769988773855557,
      "grad_norm": 0.0,
      "learning_rate": 1.8888691060275955e-05,
      "loss": 1.0582,
      "step": 1419
    },
    {
      "epoch": 0.1771236123238119,
      "grad_norm": 0.0,
      "learning_rate": 1.8886839311294695e-05,
      "loss": 1.059,
      "step": 1420
    },
    {
      "epoch": 0.17724834726206812,
      "grad_norm": 0.0,
      "learning_rate": 1.8884986111757223e-05,
      "loss": 1.0849,
      "step": 1421
    },
    {
      "epoch": 0.1773730822003243,
      "grad_norm": 0.0,
      "learning_rate": 1.8883131461966034e-05,
      "loss": 1.0658,
      "step": 1422
    },
    {
      "epoch": 0.1774978171385805,
      "grad_norm": 0.0,
      "learning_rate": 1.8881275362223845e-05,
      "loss": 1.047,
      "step": 1423
    },
    {
      "epoch": 0.17762255207683672,
      "grad_norm": 0.0,
      "learning_rate": 1.887941781283362e-05,
      "loss": 1.1441,
      "step": 1424
    },
    {
      "epoch": 0.17774728701509293,
      "grad_norm": 0.0,
      "learning_rate": 1.8877558814098564e-05,
      "loss": 1.0356,
      "step": 1425
    },
    {
      "epoch": 0.17787202195334914,
      "grad_norm": 0.0,
      "learning_rate": 1.88756983663221e-05,
      "loss": 1.0798,
      "step": 1426
    },
    {
      "epoch": 0.17799675689160535,
      "grad_norm": 0.0,
      "learning_rate": 1.8873836469807912e-05,
      "loss": 1.1231,
      "step": 1427
    },
    {
      "epoch": 0.17812149182986153,
      "grad_norm": 0.0,
      "learning_rate": 1.88719731248599e-05,
      "loss": 1.0354,
      "step": 1428
    },
    {
      "epoch": 0.17824622676811774,
      "grad_norm": 0.0,
      "learning_rate": 1.887010833178222e-05,
      "loss": 1.0972,
      "step": 1429
    },
    {
      "epoch": 0.17837096170637395,
      "grad_norm": 0.0,
      "learning_rate": 1.886824209087924e-05,
      "loss": 1.0746,
      "step": 1430
    },
    {
      "epoch": 0.17849569664463016,
      "grad_norm": 0.0,
      "learning_rate": 1.8866374402455587e-05,
      "loss": 1.1135,
      "step": 1431
    },
    {
      "epoch": 0.17862043158288637,
      "grad_norm": 0.0,
      "learning_rate": 1.886450526681611e-05,
      "loss": 1.0591,
      "step": 1432
    },
    {
      "epoch": 0.17874516652114258,
      "grad_norm": 0.0,
      "learning_rate": 1.88626346842659e-05,
      "loss": 1.1603,
      "step": 1433
    },
    {
      "epoch": 0.17886990145939877,
      "grad_norm": 0.0,
      "learning_rate": 1.8860762655110284e-05,
      "loss": 1.1093,
      "step": 1434
    },
    {
      "epoch": 0.17899463639765498,
      "grad_norm": 0.0,
      "learning_rate": 1.8858889179654824e-05,
      "loss": 1.0933,
      "step": 1435
    },
    {
      "epoch": 0.1791193713359112,
      "grad_norm": 0.0,
      "learning_rate": 1.8857014258205323e-05,
      "loss": 1.0944,
      "step": 1436
    },
    {
      "epoch": 0.1792441062741674,
      "grad_norm": 0.0,
      "learning_rate": 1.8855137891067805e-05,
      "loss": 1.1029,
      "step": 1437
    },
    {
      "epoch": 0.1793688412124236,
      "grad_norm": 0.0,
      "learning_rate": 1.885326007854855e-05,
      "loss": 1.0897,
      "step": 1438
    },
    {
      "epoch": 0.17949357615067982,
      "grad_norm": 0.0,
      "learning_rate": 1.885138082095406e-05,
      "loss": 1.0703,
      "step": 1439
    },
    {
      "epoch": 0.179618311088936,
      "grad_norm": 0.0,
      "learning_rate": 1.8849500118591077e-05,
      "loss": 1.1392,
      "step": 1440
    },
    {
      "epoch": 0.1797430460271922,
      "grad_norm": 0.0,
      "learning_rate": 1.8847617971766577e-05,
      "loss": 1.0533,
      "step": 1441
    },
    {
      "epoch": 0.17986778096544842,
      "grad_norm": 0.0,
      "learning_rate": 1.8845734380787783e-05,
      "loss": 1.0478,
      "step": 1442
    },
    {
      "epoch": 0.17999251590370463,
      "grad_norm": 0.0,
      "learning_rate": 1.8843849345962135e-05,
      "loss": 1.0838,
      "step": 1443
    },
    {
      "epoch": 0.18011725084196084,
      "grad_norm": 0.0,
      "learning_rate": 1.8841962867597325e-05,
      "loss": 1.1238,
      "step": 1444
    },
    {
      "epoch": 0.18024198578021705,
      "grad_norm": 0.0,
      "learning_rate": 1.8840074946001264e-05,
      "loss": 1.0972,
      "step": 1445
    },
    {
      "epoch": 0.18036672071847323,
      "grad_norm": 0.0,
      "learning_rate": 1.883818558148212e-05,
      "loss": 1.0812,
      "step": 1446
    },
    {
      "epoch": 0.18049145565672944,
      "grad_norm": 0.0,
      "learning_rate": 1.883629477434828e-05,
      "loss": 1.0488,
      "step": 1447
    },
    {
      "epoch": 0.18061619059498565,
      "grad_norm": 0.0,
      "learning_rate": 1.8834402524908366e-05,
      "loss": 1.0634,
      "step": 1448
    },
    {
      "epoch": 0.18074092553324186,
      "grad_norm": 0.0,
      "learning_rate": 1.883250883347125e-05,
      "loss": 1.0744,
      "step": 1449
    },
    {
      "epoch": 0.18086566047149807,
      "grad_norm": 0.0,
      "learning_rate": 1.8830613700346026e-05,
      "loss": 1.088,
      "step": 1450
    },
    {
      "epoch": 0.18099039540975428,
      "grad_norm": 0.0,
      "learning_rate": 1.8828717125842026e-05,
      "loss": 1.0777,
      "step": 1451
    },
    {
      "epoch": 0.18111513034801047,
      "grad_norm": 0.0,
      "learning_rate": 1.8826819110268822e-05,
      "loss": 1.0946,
      "step": 1452
    },
    {
      "epoch": 0.18123986528626668,
      "grad_norm": 0.0,
      "learning_rate": 1.8824919653936217e-05,
      "loss": 1.1098,
      "step": 1453
    },
    {
      "epoch": 0.1813646002245229,
      "grad_norm": 0.0,
      "learning_rate": 1.8823018757154246e-05,
      "loss": 1.1247,
      "step": 1454
    },
    {
      "epoch": 0.1814893351627791,
      "grad_norm": 0.0,
      "learning_rate": 1.8821116420233186e-05,
      "loss": 1.1171,
      "step": 1455
    },
    {
      "epoch": 0.1816140701010353,
      "grad_norm": 0.0,
      "learning_rate": 1.881921264348355e-05,
      "loss": 1.1489,
      "step": 1456
    },
    {
      "epoch": 0.18173880503929152,
      "grad_norm": 0.0,
      "learning_rate": 1.881730742721608e-05,
      "loss": 1.0879,
      "step": 1457
    },
    {
      "epoch": 0.1818635399775477,
      "grad_norm": 0.0,
      "learning_rate": 1.8815400771741753e-05,
      "loss": 1.0917,
      "step": 1458
    },
    {
      "epoch": 0.1819882749158039,
      "grad_norm": 0.0,
      "learning_rate": 1.881349267737179e-05,
      "loss": 1.1207,
      "step": 1459
    },
    {
      "epoch": 0.18211300985406012,
      "grad_norm": 0.0,
      "learning_rate": 1.881158314441763e-05,
      "loss": 1.0839,
      "step": 1460
    },
    {
      "epoch": 0.18223774479231633,
      "grad_norm": 0.0,
      "learning_rate": 1.8809672173190964e-05,
      "loss": 1.0378,
      "step": 1461
    },
    {
      "epoch": 0.18236247973057254,
      "grad_norm": 0.0,
      "learning_rate": 1.880775976400371e-05,
      "loss": 1.0733,
      "step": 1462
    },
    {
      "epoch": 0.18248721466882875,
      "grad_norm": 0.0,
      "learning_rate": 1.8805845917168016e-05,
      "loss": 1.0969,
      "step": 1463
    },
    {
      "epoch": 0.18261194960708493,
      "grad_norm": 0.0,
      "learning_rate": 1.8803930632996277e-05,
      "loss": 1.1077,
      "step": 1464
    },
    {
      "epoch": 0.18273668454534114,
      "grad_norm": 0.0,
      "learning_rate": 1.880201391180111e-05,
      "loss": 1.1169,
      "step": 1465
    },
    {
      "epoch": 0.18286141948359735,
      "grad_norm": 0.0,
      "learning_rate": 1.8800095753895382e-05,
      "loss": 1.04,
      "step": 1466
    },
    {
      "epoch": 0.18298615442185356,
      "grad_norm": 0.0,
      "learning_rate": 1.879817615959217e-05,
      "loss": 1.0999,
      "step": 1467
    },
    {
      "epoch": 0.18311088936010977,
      "grad_norm": 0.0,
      "learning_rate": 1.8796255129204812e-05,
      "loss": 1.0215,
      "step": 1468
    },
    {
      "epoch": 0.18323562429836598,
      "grad_norm": 0.0,
      "learning_rate": 1.8794332663046866e-05,
      "loss": 1.0378,
      "step": 1469
    },
    {
      "epoch": 0.18336035923662217,
      "grad_norm": 0.0,
      "learning_rate": 1.879240876143212e-05,
      "loss": 1.0362,
      "step": 1470
    },
    {
      "epoch": 0.18348509417487838,
      "grad_norm": 0.0,
      "learning_rate": 1.8790483424674614e-05,
      "loss": 1.0607,
      "step": 1471
    },
    {
      "epoch": 0.1836098291131346,
      "grad_norm": 0.0,
      "learning_rate": 1.8788556653088604e-05,
      "loss": 1.0526,
      "step": 1472
    },
    {
      "epoch": 0.1837345640513908,
      "grad_norm": 0.0,
      "learning_rate": 1.8786628446988594e-05,
      "loss": 1.0973,
      "step": 1473
    },
    {
      "epoch": 0.183859298989647,
      "grad_norm": 0.0,
      "learning_rate": 1.8784698806689308e-05,
      "loss": 1.0463,
      "step": 1474
    },
    {
      "epoch": 0.18398403392790322,
      "grad_norm": 0.0,
      "learning_rate": 1.8782767732505713e-05,
      "loss": 1.047,
      "step": 1475
    },
    {
      "epoch": 0.1841087688661594,
      "grad_norm": 0.0,
      "learning_rate": 1.8780835224753017e-05,
      "loss": 1.0589,
      "step": 1476
    },
    {
      "epoch": 0.1842335038044156,
      "grad_norm": 0.0,
      "learning_rate": 1.8778901283746646e-05,
      "loss": 1.0659,
      "step": 1477
    },
    {
      "epoch": 0.18435823874267182,
      "grad_norm": 0.0,
      "learning_rate": 1.877696590980227e-05,
      "loss": 1.0853,
      "step": 1478
    },
    {
      "epoch": 0.18448297368092803,
      "grad_norm": 0.0,
      "learning_rate": 1.8775029103235795e-05,
      "loss": 1.0867,
      "step": 1479
    },
    {
      "epoch": 0.18460770861918424,
      "grad_norm": 0.0,
      "learning_rate": 1.877309086436335e-05,
      "loss": 1.1204,
      "step": 1480
    },
    {
      "epoch": 0.18473244355744045,
      "grad_norm": 0.0,
      "learning_rate": 1.8771151193501306e-05,
      "loss": 1.0491,
      "step": 1481
    },
    {
      "epoch": 0.18485717849569663,
      "grad_norm": 0.0,
      "learning_rate": 1.8769210090966274e-05,
      "loss": 1.0539,
      "step": 1482
    },
    {
      "epoch": 0.18498191343395284,
      "grad_norm": 0.0,
      "learning_rate": 1.876726755707508e-05,
      "loss": 1.0667,
      "step": 1483
    },
    {
      "epoch": 0.18510664837220905,
      "grad_norm": 0.0,
      "learning_rate": 1.8765323592144797e-05,
      "loss": 1.0991,
      "step": 1484
    },
    {
      "epoch": 0.18523138331046526,
      "grad_norm": 0.0,
      "learning_rate": 1.8763378196492735e-05,
      "loss": 1.0963,
      "step": 1485
    },
    {
      "epoch": 0.18535611824872147,
      "grad_norm": 0.0,
      "learning_rate": 1.8761431370436427e-05,
      "loss": 1.0121,
      "step": 1486
    },
    {
      "epoch": 0.18548085318697768,
      "grad_norm": 0.0,
      "learning_rate": 1.8759483114293644e-05,
      "loss": 1.0708,
      "step": 1487
    },
    {
      "epoch": 0.18560558812523387,
      "grad_norm": 0.0,
      "learning_rate": 1.875753342838239e-05,
      "loss": 1.118,
      "step": 1488
    },
    {
      "epoch": 0.18573032306349008,
      "grad_norm": 0.0,
      "learning_rate": 1.8755582313020912e-05,
      "loss": 1.1093,
      "step": 1489
    },
    {
      "epoch": 0.1858550580017463,
      "grad_norm": 0.0,
      "learning_rate": 1.875362976852767e-05,
      "loss": 1.0793,
      "step": 1490
    },
    {
      "epoch": 0.1859797929400025,
      "grad_norm": 0.0,
      "learning_rate": 1.8751675795221365e-05,
      "loss": 1.1116,
      "step": 1491
    },
    {
      "epoch": 0.1861045278782587,
      "grad_norm": 0.0,
      "learning_rate": 1.8749720393420948e-05,
      "loss": 1.1434,
      "step": 1492
    },
    {
      "epoch": 0.18622926281651492,
      "grad_norm": 0.0,
      "learning_rate": 1.8747763563445584e-05,
      "loss": 1.0105,
      "step": 1493
    },
    {
      "epoch": 0.1863539977547711,
      "grad_norm": 0.0,
      "learning_rate": 1.8745805305614674e-05,
      "loss": 1.028,
      "step": 1494
    },
    {
      "epoch": 0.1864787326930273,
      "grad_norm": 0.0,
      "learning_rate": 1.874384562024786e-05,
      "loss": 1.0831,
      "step": 1495
    },
    {
      "epoch": 0.18660346763128352,
      "grad_norm": 0.0,
      "learning_rate": 1.8741884507665006e-05,
      "loss": 1.1368,
      "step": 1496
    },
    {
      "epoch": 0.18672820256953973,
      "grad_norm": 0.0,
      "learning_rate": 1.8739921968186222e-05,
      "loss": 1.0619,
      "step": 1497
    },
    {
      "epoch": 0.18685293750779594,
      "grad_norm": 0.0,
      "learning_rate": 1.8737958002131837e-05,
      "loss": 1.0663,
      "step": 1498
    },
    {
      "epoch": 0.18697767244605215,
      "grad_norm": 0.0,
      "learning_rate": 1.8735992609822427e-05,
      "loss": 1.1168,
      "step": 1499
    },
    {
      "epoch": 0.18710240738430833,
      "grad_norm": 0.0,
      "learning_rate": 1.873402579157879e-05,
      "loss": 1.1177,
      "step": 1500
    },
    {
      "epoch": 0.18722714232256454,
      "grad_norm": 0.0,
      "learning_rate": 1.8732057547721962e-05,
      "loss": 1.1316,
      "step": 1501
    },
    {
      "epoch": 0.18735187726082075,
      "grad_norm": 0.0,
      "learning_rate": 1.8730087878573205e-05,
      "loss": 1.1219,
      "step": 1502
    },
    {
      "epoch": 0.18747661219907696,
      "grad_norm": 0.0,
      "learning_rate": 1.872811678445402e-05,
      "loss": 1.0596,
      "step": 1503
    },
    {
      "epoch": 0.18760134713733317,
      "grad_norm": 0.0,
      "learning_rate": 1.8726144265686145e-05,
      "loss": 1.0369,
      "step": 1504
    },
    {
      "epoch": 0.18772608207558938,
      "grad_norm": 0.0,
      "learning_rate": 1.8724170322591536e-05,
      "loss": 1.0898,
      "step": 1505
    },
    {
      "epoch": 0.18785081701384557,
      "grad_norm": 0.0,
      "learning_rate": 1.8722194955492404e-05,
      "loss": 1.1114,
      "step": 1506
    },
    {
      "epoch": 0.18797555195210178,
      "grad_norm": 0.0,
      "learning_rate": 1.8720218164711164e-05,
      "loss": 1.0838,
      "step": 1507
    },
    {
      "epoch": 0.188100286890358,
      "grad_norm": 0.0,
      "learning_rate": 1.8718239950570484e-05,
      "loss": 1.1091,
      "step": 1508
    },
    {
      "epoch": 0.1882250218286142,
      "grad_norm": 0.0,
      "learning_rate": 1.871626031339326e-05,
      "loss": 1.0798,
      "step": 1509
    },
    {
      "epoch": 0.1883497567668704,
      "grad_norm": 0.0,
      "learning_rate": 1.8714279253502616e-05,
      "loss": 1.1107,
      "step": 1510
    },
    {
      "epoch": 0.18847449170512662,
      "grad_norm": 0.0,
      "learning_rate": 1.8712296771221912e-05,
      "loss": 1.1,
      "step": 1511
    },
    {
      "epoch": 0.1885992266433828,
      "grad_norm": 0.0,
      "learning_rate": 1.871031286687474e-05,
      "loss": 1.0681,
      "step": 1512
    },
    {
      "epoch": 0.188723961581639,
      "grad_norm": 0.0,
      "learning_rate": 1.870832754078492e-05,
      "loss": 1.0743,
      "step": 1513
    },
    {
      "epoch": 0.18884869651989522,
      "grad_norm": 0.0,
      "learning_rate": 1.8706340793276516e-05,
      "loss": 1.091,
      "step": 1514
    },
    {
      "epoch": 0.18897343145815143,
      "grad_norm": 0.0,
      "learning_rate": 1.8704352624673804e-05,
      "loss": 1.0858,
      "step": 1515
    },
    {
      "epoch": 0.18909816639640764,
      "grad_norm": 0.0,
      "learning_rate": 1.870236303530131e-05,
      "loss": 1.0533,
      "step": 1516
    },
    {
      "epoch": 0.18922290133466385,
      "grad_norm": 0.0,
      "learning_rate": 1.870037202548378e-05,
      "loss": 1.0962,
      "step": 1517
    },
    {
      "epoch": 0.18934763627292003,
      "grad_norm": 0.0,
      "learning_rate": 1.8698379595546202e-05,
      "loss": 1.0539,
      "step": 1518
    },
    {
      "epoch": 0.18947237121117624,
      "grad_norm": 0.0,
      "learning_rate": 1.8696385745813793e-05,
      "loss": 1.0449,
      "step": 1519
    },
    {
      "epoch": 0.18959710614943245,
      "grad_norm": 0.0,
      "learning_rate": 1.869439047661199e-05,
      "loss": 1.0658,
      "step": 1520
    },
    {
      "epoch": 0.18972184108768866,
      "grad_norm": 0.0,
      "learning_rate": 1.8692393788266477e-05,
      "loss": 1.0408,
      "step": 1521
    },
    {
      "epoch": 0.18984657602594487,
      "grad_norm": 0.0,
      "learning_rate": 1.869039568110317e-05,
      "loss": 1.0401,
      "step": 1522
    },
    {
      "epoch": 0.18997131096420108,
      "grad_norm": 0.0,
      "learning_rate": 1.86883961554482e-05,
      "loss": 1.1052,
      "step": 1523
    },
    {
      "epoch": 0.19009604590245727,
      "grad_norm": 0.0,
      "learning_rate": 1.8686395211627944e-05,
      "loss": 1.0633,
      "step": 1524
    },
    {
      "epoch": 0.19022078084071348,
      "grad_norm": 0.0,
      "learning_rate": 1.8684392849969006e-05,
      "loss": 1.1292,
      "step": 1525
    },
    {
      "epoch": 0.1903455157789697,
      "grad_norm": 0.0,
      "learning_rate": 1.8682389070798225e-05,
      "loss": 1.1032,
      "step": 1526
    },
    {
      "epoch": 0.1904702507172259,
      "grad_norm": 0.0,
      "learning_rate": 1.8680383874442664e-05,
      "loss": 1.0854,
      "step": 1527
    },
    {
      "epoch": 0.1905949856554821,
      "grad_norm": 0.0,
      "learning_rate": 1.8678377261229624e-05,
      "loss": 1.0839,
      "step": 1528
    },
    {
      "epoch": 0.19071972059373832,
      "grad_norm": 0.0,
      "learning_rate": 1.8676369231486635e-05,
      "loss": 1.1194,
      "step": 1529
    },
    {
      "epoch": 0.1908444555319945,
      "grad_norm": 0.0,
      "learning_rate": 1.8674359785541455e-05,
      "loss": 1.0814,
      "step": 1530
    },
    {
      "epoch": 0.1909691904702507,
      "grad_norm": 0.0,
      "learning_rate": 1.8672348923722083e-05,
      "loss": 1.0432,
      "step": 1531
    },
    {
      "epoch": 0.19109392540850692,
      "grad_norm": 0.0,
      "learning_rate": 1.8670336646356735e-05,
      "loss": 1.1014,
      "step": 1532
    },
    {
      "epoch": 0.19121866034676313,
      "grad_norm": 0.0,
      "learning_rate": 1.8668322953773866e-05,
      "loss": 1.0537,
      "step": 1533
    },
    {
      "epoch": 0.19134339528501934,
      "grad_norm": 0.0,
      "learning_rate": 1.8666307846302168e-05,
      "loss": 1.0986,
      "step": 1534
    },
    {
      "epoch": 0.19146813022327555,
      "grad_norm": 0.0,
      "learning_rate": 1.866429132427055e-05,
      "loss": 1.0339,
      "step": 1535
    },
    {
      "epoch": 0.19159286516153173,
      "grad_norm": 0.0,
      "learning_rate": 1.8662273388008167e-05,
      "loss": 1.0679,
      "step": 1536
    },
    {
      "epoch": 0.19171760009978794,
      "grad_norm": 0.0,
      "learning_rate": 1.866025403784439e-05,
      "loss": 1.1282,
      "step": 1537
    },
    {
      "epoch": 0.19184233503804415,
      "grad_norm": 0.0,
      "learning_rate": 1.8658233274108827e-05,
      "loss": 1.1358,
      "step": 1538
    },
    {
      "epoch": 0.19196706997630036,
      "grad_norm": 0.0,
      "learning_rate": 1.8656211097131324e-05,
      "loss": 1.0809,
      "step": 1539
    },
    {
      "epoch": 0.19209180491455657,
      "grad_norm": 0.0,
      "learning_rate": 1.8654187507241946e-05,
      "loss": 1.0386,
      "step": 1540
    },
    {
      "epoch": 0.19221653985281278,
      "grad_norm": 0.0,
      "learning_rate": 1.8652162504771e-05,
      "loss": 1.076,
      "step": 1541
    },
    {
      "epoch": 0.19234127479106897,
      "grad_norm": 0.0,
      "learning_rate": 1.8650136090049013e-05,
      "loss": 1.0389,
      "step": 1542
    },
    {
      "epoch": 0.19246600972932518,
      "grad_norm": 0.0,
      "learning_rate": 1.8648108263406742e-05,
      "loss": 1.0712,
      "step": 1543
    },
    {
      "epoch": 0.1925907446675814,
      "grad_norm": 0.0,
      "learning_rate": 1.864607902517519e-05,
      "loss": 1.0837,
      "step": 1544
    },
    {
      "epoch": 0.1927154796058376,
      "grad_norm": 0.0,
      "learning_rate": 1.8644048375685574e-05,
      "loss": 1.0863,
      "step": 1545
    },
    {
      "epoch": 0.1928402145440938,
      "grad_norm": 0.0,
      "learning_rate": 1.8642016315269345e-05,
      "loss": 1.0874,
      "step": 1546
    },
    {
      "epoch": 0.19296494948235002,
      "grad_norm": 0.0,
      "learning_rate": 1.863998284425819e-05,
      "loss": 1.0649,
      "step": 1547
    },
    {
      "epoch": 0.1930896844206062,
      "grad_norm": 0.0,
      "learning_rate": 1.8637947962984024e-05,
      "loss": 1.1398,
      "step": 1548
    },
    {
      "epoch": 0.1932144193588624,
      "grad_norm": 0.0,
      "learning_rate": 1.8635911671778987e-05,
      "loss": 1.112,
      "step": 1549
    },
    {
      "epoch": 0.19333915429711862,
      "grad_norm": 0.0,
      "learning_rate": 1.8633873970975457e-05,
      "loss": 1.0519,
      "step": 1550
    },
    {
      "epoch": 0.19346388923537483,
      "grad_norm": 0.0,
      "learning_rate": 1.863183486090604e-05,
      "loss": 1.0951,
      "step": 1551
    },
    {
      "epoch": 0.19358862417363104,
      "grad_norm": 0.0,
      "learning_rate": 1.8629794341903558e-05,
      "loss": 1.1238,
      "step": 1552
    },
    {
      "epoch": 0.19371335911188725,
      "grad_norm": 0.0,
      "learning_rate": 1.8627752414301087e-05,
      "loss": 1.0462,
      "step": 1553
    },
    {
      "epoch": 0.19383809405014343,
      "grad_norm": 0.0,
      "learning_rate": 1.862570907843192e-05,
      "loss": 1.0547,
      "step": 1554
    },
    {
      "epoch": 0.19396282898839964,
      "grad_norm": 0.0,
      "learning_rate": 1.8623664334629576e-05,
      "loss": 1.1024,
      "step": 1555
    },
    {
      "epoch": 0.19408756392665585,
      "grad_norm": 0.0,
      "learning_rate": 1.8621618183227813e-05,
      "loss": 1.0678,
      "step": 1556
    },
    {
      "epoch": 0.19421229886491206,
      "grad_norm": 0.0,
      "learning_rate": 1.861957062456061e-05,
      "loss": 1.0571,
      "step": 1557
    },
    {
      "epoch": 0.19433703380316827,
      "grad_norm": 0.0,
      "learning_rate": 1.8617521658962183e-05,
      "loss": 1.0641,
      "step": 1558
    },
    {
      "epoch": 0.19446176874142448,
      "grad_norm": 0.0,
      "learning_rate": 1.8615471286766977e-05,
      "loss": 1.0466,
      "step": 1559
    },
    {
      "epoch": 0.19458650367968067,
      "grad_norm": 0.0,
      "learning_rate": 1.861341950830966e-05,
      "loss": 1.1001,
      "step": 1560
    },
    {
      "epoch": 0.19471123861793688,
      "grad_norm": 0.0,
      "learning_rate": 1.8611366323925138e-05,
      "loss": 1.0632,
      "step": 1561
    },
    {
      "epoch": 0.1948359735561931,
      "grad_norm": 0.0,
      "learning_rate": 1.860931173394854e-05,
      "loss": 1.0538,
      "step": 1562
    },
    {
      "epoch": 0.1949607084944493,
      "grad_norm": 0.0,
      "learning_rate": 1.860725573871523e-05,
      "loss": 1.1649,
      "step": 1563
    },
    {
      "epoch": 0.1950854434327055,
      "grad_norm": 0.0,
      "learning_rate": 1.860519833856079e-05,
      "loss": 1.0408,
      "step": 1564
    },
    {
      "epoch": 0.19521017837096172,
      "grad_norm": 0.0,
      "learning_rate": 1.860313953382105e-05,
      "loss": 1.0565,
      "step": 1565
    },
    {
      "epoch": 0.1953349133092179,
      "grad_norm": 0.0,
      "learning_rate": 1.8601079324832053e-05,
      "loss": 1.0662,
      "step": 1566
    },
    {
      "epoch": 0.1954596482474741,
      "grad_norm": 0.0,
      "learning_rate": 1.859901771193008e-05,
      "loss": 1.0413,
      "step": 1567
    },
    {
      "epoch": 0.19558438318573032,
      "grad_norm": 0.0,
      "learning_rate": 1.8596954695451633e-05,
      "loss": 1.0889,
      "step": 1568
    },
    {
      "epoch": 0.19570911812398653,
      "grad_norm": 0.0,
      "learning_rate": 1.8594890275733453e-05,
      "loss": 1.1284,
      "step": 1569
    },
    {
      "epoch": 0.19583385306224274,
      "grad_norm": 0.0,
      "learning_rate": 1.8592824453112504e-05,
      "loss": 1.0399,
      "step": 1570
    },
    {
      "epoch": 0.19595858800049895,
      "grad_norm": 0.0,
      "learning_rate": 1.8590757227925982e-05,
      "loss": 1.0642,
      "step": 1571
    },
    {
      "epoch": 0.19608332293875513,
      "grad_norm": 0.0,
      "learning_rate": 1.8588688600511307e-05,
      "loss": 1.049,
      "step": 1572
    },
    {
      "epoch": 0.19620805787701134,
      "grad_norm": 0.0,
      "learning_rate": 1.8586618571206133e-05,
      "loss": 1.0874,
      "step": 1573
    },
    {
      "epoch": 0.19633279281526755,
      "grad_norm": 0.0,
      "learning_rate": 1.858454714034834e-05,
      "loss": 1.1126,
      "step": 1574
    },
    {
      "epoch": 0.19645752775352376,
      "grad_norm": 0.0,
      "learning_rate": 1.8582474308276038e-05,
      "loss": 1.0148,
      "step": 1575
    },
    {
      "epoch": 0.19658226269177997,
      "grad_norm": 0.0,
      "learning_rate": 1.858040007532757e-05,
      "loss": 1.0712,
      "step": 1576
    },
    {
      "epoch": 0.19670699763003618,
      "grad_norm": 0.0,
      "learning_rate": 1.8578324441841496e-05,
      "loss": 1.1087,
      "step": 1577
    },
    {
      "epoch": 0.19683173256829237,
      "grad_norm": 0.0,
      "learning_rate": 1.8576247408156614e-05,
      "loss": 1.1197,
      "step": 1578
    },
    {
      "epoch": 0.19695646750654858,
      "grad_norm": 0.0,
      "learning_rate": 1.857416897461195e-05,
      "loss": 1.0804,
      "step": 1579
    },
    {
      "epoch": 0.1970812024448048,
      "grad_norm": 0.0,
      "learning_rate": 1.8572089141546755e-05,
      "loss": 1.0625,
      "step": 1580
    },
    {
      "epoch": 0.197205937383061,
      "grad_norm": 0.0,
      "learning_rate": 1.8570007909300514e-05,
      "loss": 1.0641,
      "step": 1581
    },
    {
      "epoch": 0.1973306723213172,
      "grad_norm": 0.0,
      "learning_rate": 1.856792527821293e-05,
      "loss": 1.1301,
      "step": 1582
    },
    {
      "epoch": 0.19745540725957342,
      "grad_norm": 0.0,
      "learning_rate": 1.8565841248623946e-05,
      "loss": 1.0518,
      "step": 1583
    },
    {
      "epoch": 0.1975801421978296,
      "grad_norm": 0.0,
      "learning_rate": 1.8563755820873724e-05,
      "loss": 1.0531,
      "step": 1584
    },
    {
      "epoch": 0.1977048771360858,
      "grad_norm": 0.0,
      "learning_rate": 1.8561668995302668e-05,
      "loss": 1.0209,
      "step": 1585
    },
    {
      "epoch": 0.19782961207434202,
      "grad_norm": 0.0,
      "learning_rate": 1.8559580772251386e-05,
      "loss": 1.0415,
      "step": 1586
    },
    {
      "epoch": 0.19795434701259823,
      "grad_norm": 0.0,
      "learning_rate": 1.855749115206074e-05,
      "loss": 1.0919,
      "step": 1587
    },
    {
      "epoch": 0.19807908195085444,
      "grad_norm": 0.0,
      "learning_rate": 1.8555400135071803e-05,
      "loss": 1.0613,
      "step": 1588
    },
    {
      "epoch": 0.19820381688911065,
      "grad_norm": 0.0,
      "learning_rate": 1.8553307721625887e-05,
      "loss": 1.1086,
      "step": 1589
    },
    {
      "epoch": 0.19832855182736683,
      "grad_norm": 0.0,
      "learning_rate": 1.8551213912064517e-05,
      "loss": 1.086,
      "step": 1590
    },
    {
      "epoch": 0.19845328676562304,
      "grad_norm": 0.0,
      "learning_rate": 1.854911870672947e-05,
      "loss": 1.0793,
      "step": 1591
    },
    {
      "epoch": 0.19857802170387925,
      "grad_norm": 0.0,
      "learning_rate": 1.854702210596272e-05,
      "loss": 1.0174,
      "step": 1592
    },
    {
      "epoch": 0.19870275664213546,
      "grad_norm": 0.0,
      "learning_rate": 1.8544924110106495e-05,
      "loss": 1.0386,
      "step": 1593
    },
    {
      "epoch": 0.19882749158039167,
      "grad_norm": 0.0,
      "learning_rate": 1.8542824719503244e-05,
      "loss": 1.0414,
      "step": 1594
    },
    {
      "epoch": 0.19895222651864788,
      "grad_norm": 0.0,
      "learning_rate": 1.854072393449563e-05,
      "loss": 1.035,
      "step": 1595
    },
    {
      "epoch": 0.19907696145690407,
      "grad_norm": 0.0,
      "learning_rate": 1.853862175542656e-05,
      "loss": 1.0575,
      "step": 1596
    },
    {
      "epoch": 0.19920169639516028,
      "grad_norm": 0.0,
      "learning_rate": 1.8536518182639166e-05,
      "loss": 1.1028,
      "step": 1597
    },
    {
      "epoch": 0.1993264313334165,
      "grad_norm": 0.0,
      "learning_rate": 1.8534413216476795e-05,
      "loss": 1.0816,
      "step": 1598
    },
    {
      "epoch": 0.1994511662716727,
      "grad_norm": 0.0,
      "learning_rate": 1.853230685728304e-05,
      "loss": 1.0747,
      "step": 1599
    },
    {
      "epoch": 0.1995759012099289,
      "grad_norm": 0.0,
      "learning_rate": 1.8530199105401705e-05,
      "loss": 1.06,
      "step": 1600
    },
    {
      "epoch": 0.19970063614818512,
      "grad_norm": 0.0,
      "learning_rate": 1.852808996117683e-05,
      "loss": 1.0965,
      "step": 1601
    },
    {
      "epoch": 0.1998253710864413,
      "grad_norm": 0.0,
      "learning_rate": 1.8525979424952683e-05,
      "loss": 1.1099,
      "step": 1602
    },
    {
      "epoch": 0.1999501060246975,
      "grad_norm": 0.0,
      "learning_rate": 1.8523867497073757e-05,
      "loss": 1.0651,
      "step": 1603
    },
    {
      "epoch": 0.20007484096295372,
      "grad_norm": 0.0,
      "learning_rate": 1.8521754177884766e-05,
      "loss": 1.0689,
      "step": 1604
    },
    {
      "epoch": 0.20019957590120993,
      "grad_norm": 0.0,
      "learning_rate": 1.8519639467730667e-05,
      "loss": 1.0924,
      "step": 1605
    },
    {
      "epoch": 0.20032431083946614,
      "grad_norm": 0.0,
      "learning_rate": 1.8517523366956624e-05,
      "loss": 1.036,
      "step": 1606
    },
    {
      "epoch": 0.20044904577772235,
      "grad_norm": 0.0,
      "learning_rate": 1.8515405875908042e-05,
      "loss": 1.0772,
      "step": 1607
    },
    {
      "epoch": 0.20057378071597853,
      "grad_norm": 0.0,
      "learning_rate": 1.8513286994930547e-05,
      "loss": 1.0556,
      "step": 1608
    },
    {
      "epoch": 0.20069851565423474,
      "grad_norm": 0.0,
      "learning_rate": 1.8511166724369997e-05,
      "loss": 1.0713,
      "step": 1609
    },
    {
      "epoch": 0.20082325059249095,
      "grad_norm": 0.0,
      "learning_rate": 1.8509045064572474e-05,
      "loss": 1.0811,
      "step": 1610
    },
    {
      "epoch": 0.20094798553074716,
      "grad_norm": 0.0,
      "learning_rate": 1.8506922015884284e-05,
      "loss": 1.0316,
      "step": 1611
    },
    {
      "epoch": 0.20107272046900337,
      "grad_norm": 0.0,
      "learning_rate": 1.8504797578651962e-05,
      "loss": 1.0536,
      "step": 1612
    },
    {
      "epoch": 0.20119745540725958,
      "grad_norm": 0.0,
      "learning_rate": 1.8502671753222272e-05,
      "loss": 1.0129,
      "step": 1613
    },
    {
      "epoch": 0.20132219034551577,
      "grad_norm": 0.0,
      "learning_rate": 1.85005445399422e-05,
      "loss": 1.0415,
      "step": 1614
    },
    {
      "epoch": 0.20144692528377198,
      "grad_norm": 0.0,
      "learning_rate": 1.849841593915896e-05,
      "loss": 1.1267,
      "step": 1615
    },
    {
      "epoch": 0.2015716602220282,
      "grad_norm": 0.0,
      "learning_rate": 1.8496285951219995e-05,
      "loss": 1.0178,
      "step": 1616
    },
    {
      "epoch": 0.2016963951602844,
      "grad_norm": 0.0,
      "learning_rate": 1.8494154576472976e-05,
      "loss": 1.0657,
      "step": 1617
    },
    {
      "epoch": 0.2018211300985406,
      "grad_norm": 0.0,
      "learning_rate": 1.849202181526579e-05,
      "loss": 1.078,
      "step": 1618
    },
    {
      "epoch": 0.20194586503679682,
      "grad_norm": 0.0,
      "learning_rate": 1.8489887667946562e-05,
      "loss": 1.0604,
      "step": 1619
    },
    {
      "epoch": 0.202070599975053,
      "grad_norm": 0.0,
      "learning_rate": 1.848775213486364e-05,
      "loss": 1.0413,
      "step": 1620
    },
    {
      "epoch": 0.2021953349133092,
      "grad_norm": 0.0,
      "learning_rate": 1.848561521636559e-05,
      "loss": 1.0245,
      "step": 1621
    },
    {
      "epoch": 0.20232006985156542,
      "grad_norm": 0.0,
      "learning_rate": 1.848347691280122e-05,
      "loss": 1.0816,
      "step": 1622
    },
    {
      "epoch": 0.20244480478982163,
      "grad_norm": 0.0,
      "learning_rate": 1.848133722451955e-05,
      "loss": 1.0846,
      "step": 1623
    },
    {
      "epoch": 0.20256953972807784,
      "grad_norm": 0.0,
      "learning_rate": 1.847919615186983e-05,
      "loss": 1.0825,
      "step": 1624
    },
    {
      "epoch": 0.20269427466633405,
      "grad_norm": 0.0,
      "learning_rate": 1.847705369520154e-05,
      "loss": 1.0566,
      "step": 1625
    },
    {
      "epoch": 0.20281900960459023,
      "grad_norm": 0.0,
      "learning_rate": 1.847490985486438e-05,
      "loss": 1.0039,
      "step": 1626
    },
    {
      "epoch": 0.20294374454284644,
      "grad_norm": 0.0,
      "learning_rate": 1.847276463120828e-05,
      "loss": 1.0716,
      "step": 1627
    },
    {
      "epoch": 0.20306847948110265,
      "grad_norm": 0.0,
      "learning_rate": 1.8470618024583398e-05,
      "loss": 1.0642,
      "step": 1628
    },
    {
      "epoch": 0.20319321441935886,
      "grad_norm": 0.0,
      "learning_rate": 1.8468470035340108e-05,
      "loss": 1.0486,
      "step": 1629
    },
    {
      "epoch": 0.20331794935761507,
      "grad_norm": 0.0,
      "learning_rate": 1.846632066382902e-05,
      "loss": 1.0711,
      "step": 1630
    },
    {
      "epoch": 0.20344268429587128,
      "grad_norm": 0.0,
      "learning_rate": 1.846416991040097e-05,
      "loss": 1.0886,
      "step": 1631
    },
    {
      "epoch": 0.20356741923412747,
      "grad_norm": 0.0,
      "learning_rate": 1.8462017775407005e-05,
      "loss": 1.0622,
      "step": 1632
    },
    {
      "epoch": 0.20369215417238368,
      "grad_norm": 0.0,
      "learning_rate": 1.845986425919841e-05,
      "loss": 1.0455,
      "step": 1633
    },
    {
      "epoch": 0.2038168891106399,
      "grad_norm": 0.0,
      "learning_rate": 1.8457709362126703e-05,
      "loss": 1.0657,
      "step": 1634
    },
    {
      "epoch": 0.2039416240488961,
      "grad_norm": 0.0,
      "learning_rate": 1.8455553084543603e-05,
      "loss": 1.0691,
      "step": 1635
    },
    {
      "epoch": 0.2040663589871523,
      "grad_norm": 0.0,
      "learning_rate": 1.8453395426801083e-05,
      "loss": 1.0806,
      "step": 1636
    },
    {
      "epoch": 0.20419109392540852,
      "grad_norm": 0.0,
      "learning_rate": 1.845123638925132e-05,
      "loss": 0.9941,
      "step": 1637
    },
    {
      "epoch": 0.2043158288636647,
      "grad_norm": 0.0,
      "learning_rate": 1.844907597224672e-05,
      "loss": 1.0901,
      "step": 1638
    },
    {
      "epoch": 0.2044405638019209,
      "grad_norm": 0.0,
      "learning_rate": 1.8446914176139925e-05,
      "loss": 1.0677,
      "step": 1639
    },
    {
      "epoch": 0.20456529874017712,
      "grad_norm": 0.0,
      "learning_rate": 1.8444751001283786e-05,
      "loss": 1.0966,
      "step": 1640
    },
    {
      "epoch": 0.20469003367843333,
      "grad_norm": 0.0,
      "learning_rate": 1.8442586448031396e-05,
      "loss": 1.0779,
      "step": 1641
    },
    {
      "epoch": 0.20481476861668954,
      "grad_norm": 0.0,
      "learning_rate": 1.8440420516736063e-05,
      "loss": 1.109,
      "step": 1642
    },
    {
      "epoch": 0.20493950355494575,
      "grad_norm": 0.0,
      "learning_rate": 1.8438253207751318e-05,
      "loss": 1.0352,
      "step": 1643
    },
    {
      "epoch": 0.20506423849320193,
      "grad_norm": 0.0,
      "learning_rate": 1.8436084521430922e-05,
      "loss": 1.0361,
      "step": 1644
    },
    {
      "epoch": 0.20518897343145814,
      "grad_norm": 0.0,
      "learning_rate": 1.843391445812886e-05,
      "loss": 1.0868,
      "step": 1645
    },
    {
      "epoch": 0.20531370836971435,
      "grad_norm": 0.0,
      "learning_rate": 1.843174301819934e-05,
      "loss": 1.0713,
      "step": 1646
    },
    {
      "epoch": 0.20543844330797056,
      "grad_norm": 0.0,
      "learning_rate": 1.8429570201996797e-05,
      "loss": 1.084,
      "step": 1647
    },
    {
      "epoch": 0.20556317824622677,
      "grad_norm": 0.0,
      "learning_rate": 1.8427396009875887e-05,
      "loss": 1.004,
      "step": 1648
    },
    {
      "epoch": 0.20568791318448298,
      "grad_norm": 0.0,
      "learning_rate": 1.8425220442191496e-05,
      "loss": 1.0465,
      "step": 1649
    },
    {
      "epoch": 0.20581264812273917,
      "grad_norm": 0.0,
      "learning_rate": 1.8423043499298733e-05,
      "loss": 1.0842,
      "step": 1650
    },
    {
      "epoch": 0.20593738306099538,
      "grad_norm": 0.0,
      "learning_rate": 1.8420865181552923e-05,
      "loss": 1.1353,
      "step": 1651
    },
    {
      "epoch": 0.2060621179992516,
      "grad_norm": 0.0,
      "learning_rate": 1.841868548930963e-05,
      "loss": 1.1208,
      "step": 1652
    },
    {
      "epoch": 0.2061868529375078,
      "grad_norm": 0.0,
      "learning_rate": 1.8416504422924628e-05,
      "loss": 1.0475,
      "step": 1653
    },
    {
      "epoch": 0.206311587875764,
      "grad_norm": 0.0,
      "learning_rate": 1.841432198275393e-05,
      "loss": 1.1025,
      "step": 1654
    },
    {
      "epoch": 0.20643632281402022,
      "grad_norm": 0.0,
      "learning_rate": 1.8412138169153754e-05,
      "loss": 1.0612,
      "step": 1655
    },
    {
      "epoch": 0.2065610577522764,
      "grad_norm": 0.0,
      "learning_rate": 1.8409952982480567e-05,
      "loss": 1.0598,
      "step": 1656
    },
    {
      "epoch": 0.2066857926905326,
      "grad_norm": 0.0,
      "learning_rate": 1.8407766423091033e-05,
      "loss": 1.0782,
      "step": 1657
    },
    {
      "epoch": 0.20681052762878882,
      "grad_norm": 0.0,
      "learning_rate": 1.8405578491342063e-05,
      "loss": 1.042,
      "step": 1658
    },
    {
      "epoch": 0.20693526256704503,
      "grad_norm": 0.0,
      "learning_rate": 1.8403389187590778e-05,
      "loss": 1.1315,
      "step": 1659
    },
    {
      "epoch": 0.20705999750530124,
      "grad_norm": 0.0,
      "learning_rate": 1.8401198512194532e-05,
      "loss": 1.0449,
      "step": 1660
    },
    {
      "epoch": 0.20718473244355745,
      "grad_norm": 0.0,
      "learning_rate": 1.8399006465510898e-05,
      "loss": 1.0621,
      "step": 1661
    },
    {
      "epoch": 0.20730946738181363,
      "grad_norm": 0.0,
      "learning_rate": 1.8396813047897665e-05,
      "loss": 1.0567,
      "step": 1662
    },
    {
      "epoch": 0.20743420232006984,
      "grad_norm": 0.0,
      "learning_rate": 1.8394618259712866e-05,
      "loss": 1.0788,
      "step": 1663
    },
    {
      "epoch": 0.20755893725832605,
      "grad_norm": 0.0,
      "learning_rate": 1.8392422101314736e-05,
      "loss": 1.0184,
      "step": 1664
    },
    {
      "epoch": 0.20768367219658226,
      "grad_norm": 0.0,
      "learning_rate": 1.8390224573061748e-05,
      "loss": 1.0219,
      "step": 1665
    },
    {
      "epoch": 0.20780840713483847,
      "grad_norm": 0.0,
      "learning_rate": 1.8388025675312595e-05,
      "loss": 1.0719,
      "step": 1666
    },
    {
      "epoch": 0.20793314207309468,
      "grad_norm": 0.0,
      "learning_rate": 1.838582540842619e-05,
      "loss": 1.0353,
      "step": 1667
    },
    {
      "epoch": 0.20805787701135087,
      "grad_norm": 0.0,
      "learning_rate": 1.838362377276167e-05,
      "loss": 1.064,
      "step": 1668
    },
    {
      "epoch": 0.20818261194960708,
      "grad_norm": 0.0,
      "learning_rate": 1.8381420768678406e-05,
      "loss": 1.0695,
      "step": 1669
    },
    {
      "epoch": 0.2083073468878633,
      "grad_norm": 0.0,
      "learning_rate": 1.8379216396535976e-05,
      "loss": 1.0538,
      "step": 1670
    },
    {
      "epoch": 0.2084320818261195,
      "grad_norm": 0.0,
      "learning_rate": 1.837701065669419e-05,
      "loss": 1.0476,
      "step": 1671
    },
    {
      "epoch": 0.2085568167643757,
      "grad_norm": 0.0,
      "learning_rate": 1.837480354951308e-05,
      "loss": 1.0655,
      "step": 1672
    },
    {
      "epoch": 0.20868155170263192,
      "grad_norm": 0.0,
      "learning_rate": 1.8372595075352902e-05,
      "loss": 1.0234,
      "step": 1673
    },
    {
      "epoch": 0.2088062866408881,
      "grad_norm": 0.0,
      "learning_rate": 1.8370385234574142e-05,
      "loss": 1.03,
      "step": 1674
    },
    {
      "epoch": 0.2089310215791443,
      "grad_norm": 0.0,
      "learning_rate": 1.8368174027537492e-05,
      "loss": 1.065,
      "step": 1675
    },
    {
      "epoch": 0.20905575651740052,
      "grad_norm": 0.0,
      "learning_rate": 1.836596145460388e-05,
      "loss": 1.0373,
      "step": 1676
    },
    {
      "epoch": 0.20918049145565673,
      "grad_norm": 0.0,
      "learning_rate": 1.8363747516134455e-05,
      "loss": 1.0232,
      "step": 1677
    },
    {
      "epoch": 0.20930522639391294,
      "grad_norm": 0.0,
      "learning_rate": 1.8361532212490586e-05,
      "loss": 1.0644,
      "step": 1678
    },
    {
      "epoch": 0.20942996133216915,
      "grad_norm": 0.0,
      "learning_rate": 1.835931554403387e-05,
      "loss": 1.0871,
      "step": 1679
    },
    {
      "epoch": 0.20955469627042533,
      "grad_norm": 0.0,
      "learning_rate": 1.8357097511126114e-05,
      "loss": 1.134,
      "step": 1680
    },
    {
      "epoch": 0.20967943120868154,
      "grad_norm": 0.0,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 1.0775,
      "step": 1681
    },
    {
      "epoch": 0.20980416614693775,
      "grad_norm": 0.0,
      "learning_rate": 1.8352657353405883e-05,
      "loss": 1.1148,
      "step": 1682
    },
    {
      "epoch": 0.20992890108519396,
      "grad_norm": 0.0,
      "learning_rate": 1.8350435229318154e-05,
      "loss": 1.0582,
      "step": 1683
    },
    {
      "epoch": 0.21005363602345017,
      "grad_norm": 0.0,
      "learning_rate": 1.834821174222888e-05,
      "loss": 1.0549,
      "step": 1684
    },
    {
      "epoch": 0.21017837096170638,
      "grad_norm": 0.0,
      "learning_rate": 1.834598689250099e-05,
      "loss": 1.0771,
      "step": 1685
    },
    {
      "epoch": 0.21030310589996257,
      "grad_norm": 0.0,
      "learning_rate": 1.8343760680497643e-05,
      "loss": 1.0136,
      "step": 1686
    },
    {
      "epoch": 0.21042784083821878,
      "grad_norm": 0.0,
      "learning_rate": 1.8341533106582204e-05,
      "loss": 1.0212,
      "step": 1687
    },
    {
      "epoch": 0.210552575776475,
      "grad_norm": 0.0,
      "learning_rate": 1.8339304171118274e-05,
      "loss": 1.0731,
      "step": 1688
    },
    {
      "epoch": 0.2106773107147312,
      "grad_norm": 0.0,
      "learning_rate": 1.8337073874469673e-05,
      "loss": 1.0267,
      "step": 1689
    },
    {
      "epoch": 0.2108020456529874,
      "grad_norm": 0.0,
      "learning_rate": 1.833484221700044e-05,
      "loss": 1.0687,
      "step": 1690
    },
    {
      "epoch": 0.21092678059124362,
      "grad_norm": 0.0,
      "learning_rate": 1.8332609199074832e-05,
      "loss": 1.0291,
      "step": 1691
    },
    {
      "epoch": 0.2110515155294998,
      "grad_norm": 0.0,
      "learning_rate": 1.8330374821057342e-05,
      "loss": 1.0953,
      "step": 1692
    },
    {
      "epoch": 0.211176250467756,
      "grad_norm": 0.0,
      "learning_rate": 1.8328139083312675e-05,
      "loss": 1.0832,
      "step": 1693
    },
    {
      "epoch": 0.21130098540601222,
      "grad_norm": 0.0,
      "learning_rate": 1.8325901986205756e-05,
      "loss": 1.0494,
      "step": 1694
    },
    {
      "epoch": 0.21142572034426843,
      "grad_norm": 0.0,
      "learning_rate": 1.8323663530101737e-05,
      "loss": 1.0809,
      "step": 1695
    },
    {
      "epoch": 0.21155045528252464,
      "grad_norm": 0.0,
      "learning_rate": 1.8321423715365992e-05,
      "loss": 1.0677,
      "step": 1696
    },
    {
      "epoch": 0.21167519022078085,
      "grad_norm": 0.0,
      "learning_rate": 1.8319182542364117e-05,
      "loss": 1.0394,
      "step": 1697
    },
    {
      "epoch": 0.21179992515903703,
      "grad_norm": 0.0,
      "learning_rate": 1.8316940011461925e-05,
      "loss": 1.0475,
      "step": 1698
    },
    {
      "epoch": 0.21192466009729324,
      "grad_norm": 0.0,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 1.0775,
      "step": 1699
    },
    {
      "epoch": 0.21204939503554945,
      "grad_norm": 0.0,
      "learning_rate": 1.8312450877420964e-05,
      "loss": 1.0693,
      "step": 1700
    },
    {
      "epoch": 0.21217412997380566,
      "grad_norm": 0.0,
      "learning_rate": 1.8310204275014932e-05,
      "loss": 1.0285,
      "step": 1701
    },
    {
      "epoch": 0.21229886491206187,
      "grad_norm": 0.0,
      "learning_rate": 1.8307956316174066e-05,
      "loss": 1.0729,
      "step": 1702
    },
    {
      "epoch": 0.21242359985031808,
      "grad_norm": 0.0,
      "learning_rate": 1.8305707001265287e-05,
      "loss": 1.0228,
      "step": 1703
    },
    {
      "epoch": 0.21254833478857427,
      "grad_norm": 0.0,
      "learning_rate": 1.830345633065574e-05,
      "loss": 1.0825,
      "step": 1704
    },
    {
      "epoch": 0.21267306972683048,
      "grad_norm": 0.0,
      "learning_rate": 1.830120430471279e-05,
      "loss": 1.0604,
      "step": 1705
    },
    {
      "epoch": 0.2127978046650867,
      "grad_norm": 0.0,
      "learning_rate": 1.8298950923804024e-05,
      "loss": 1.077,
      "step": 1706
    },
    {
      "epoch": 0.2129225396033429,
      "grad_norm": 0.0,
      "learning_rate": 1.8296696188297253e-05,
      "loss": 1.0313,
      "step": 1707
    },
    {
      "epoch": 0.2130472745415991,
      "grad_norm": 0.0,
      "learning_rate": 1.8294440098560508e-05,
      "loss": 1.0716,
      "step": 1708
    },
    {
      "epoch": 0.21317200947985532,
      "grad_norm": 0.0,
      "learning_rate": 1.8292182654962035e-05,
      "loss": 1.0835,
      "step": 1709
    },
    {
      "epoch": 0.2132967444181115,
      "grad_norm": 0.0,
      "learning_rate": 1.828992385787031e-05,
      "loss": 1.075,
      "step": 1710
    },
    {
      "epoch": 0.2134214793563677,
      "grad_norm": 0.0,
      "learning_rate": 1.828766370765402e-05,
      "loss": 1.0415,
      "step": 1711
    },
    {
      "epoch": 0.21354621429462392,
      "grad_norm": 0.0,
      "learning_rate": 1.828540220468208e-05,
      "loss": 1.0789,
      "step": 1712
    },
    {
      "epoch": 0.21367094923288013,
      "grad_norm": 0.0,
      "learning_rate": 1.8283139349323632e-05,
      "loss": 1.0117,
      "step": 1713
    },
    {
      "epoch": 0.21379568417113634,
      "grad_norm": 0.0,
      "learning_rate": 1.8280875141948026e-05,
      "loss": 1.0335,
      "step": 1714
    },
    {
      "epoch": 0.21392041910939255,
      "grad_norm": 0.0,
      "learning_rate": 1.827860958292483e-05,
      "loss": 1.127,
      "step": 1715
    },
    {
      "epoch": 0.21404515404764873,
      "grad_norm": 0.0,
      "learning_rate": 1.8276342672623854e-05,
      "loss": 1.0576,
      "step": 1716
    },
    {
      "epoch": 0.21416988898590494,
      "grad_norm": 0.0,
      "learning_rate": 1.8274074411415104e-05,
      "loss": 1.0973,
      "step": 1717
    },
    {
      "epoch": 0.21429462392416115,
      "grad_norm": 0.0,
      "learning_rate": 1.8271804799668828e-05,
      "loss": 1.0323,
      "step": 1718
    },
    {
      "epoch": 0.21441935886241736,
      "grad_norm": 0.0,
      "learning_rate": 1.8269533837755472e-05,
      "loss": 1.0776,
      "step": 1719
    },
    {
      "epoch": 0.21454409380067357,
      "grad_norm": 0.0,
      "learning_rate": 1.8267261526045723e-05,
      "loss": 1.0785,
      "step": 1720
    },
    {
      "epoch": 0.21466882873892978,
      "grad_norm": 0.0,
      "learning_rate": 1.8264987864910475e-05,
      "loss": 1.1081,
      "step": 1721
    },
    {
      "epoch": 0.21479356367718597,
      "grad_norm": 0.0,
      "learning_rate": 1.826271285472085e-05,
      "loss": 1.0186,
      "step": 1722
    },
    {
      "epoch": 0.21491829861544218,
      "grad_norm": 0.0,
      "learning_rate": 1.8260436495848185e-05,
      "loss": 1.0309,
      "step": 1723
    },
    {
      "epoch": 0.2150430335536984,
      "grad_norm": 0.0,
      "learning_rate": 1.8258158788664038e-05,
      "loss": 1.0893,
      "step": 1724
    },
    {
      "epoch": 0.2151677684919546,
      "grad_norm": 0.0,
      "learning_rate": 1.825587973354019e-05,
      "loss": 1.0594,
      "step": 1725
    },
    {
      "epoch": 0.2152925034302108,
      "grad_norm": 0.0,
      "learning_rate": 1.8253599330848638e-05,
      "loss": 1.0323,
      "step": 1726
    },
    {
      "epoch": 0.21541723836846702,
      "grad_norm": 0.0,
      "learning_rate": 1.825131758096161e-05,
      "loss": 1.0757,
      "step": 1727
    },
    {
      "epoch": 0.2155419733067232,
      "grad_norm": 0.0,
      "learning_rate": 1.8249034484251532e-05,
      "loss": 1.0111,
      "step": 1728
    },
    {
      "epoch": 0.2156667082449794,
      "grad_norm": 0.0,
      "learning_rate": 1.824675004109107e-05,
      "loss": 1.0685,
      "step": 1729
    },
    {
      "epoch": 0.21579144318323562,
      "grad_norm": 0.0,
      "learning_rate": 1.8244464251853103e-05,
      "loss": 1.0949,
      "step": 1730
    },
    {
      "epoch": 0.21591617812149183,
      "grad_norm": 0.0,
      "learning_rate": 1.8242177116910726e-05,
      "loss": 1.0654,
      "step": 1731
    },
    {
      "epoch": 0.21604091305974804,
      "grad_norm": 0.0,
      "learning_rate": 1.8239888636637263e-05,
      "loss": 1.0486,
      "step": 1732
    },
    {
      "epoch": 0.21616564799800425,
      "grad_norm": 0.0,
      "learning_rate": 1.8237598811406244e-05,
      "loss": 1.0951,
      "step": 1733
    },
    {
      "epoch": 0.21629038293626043,
      "grad_norm": 0.0,
      "learning_rate": 1.8235307641591435e-05,
      "loss": 1.0359,
      "step": 1734
    },
    {
      "epoch": 0.21641511787451664,
      "grad_norm": 0.0,
      "learning_rate": 1.8233015127566805e-05,
      "loss": 0.9988,
      "step": 1735
    },
    {
      "epoch": 0.21653985281277285,
      "grad_norm": 0.0,
      "learning_rate": 1.823072126970655e-05,
      "loss": 1.0466,
      "step": 1736
    },
    {
      "epoch": 0.21666458775102906,
      "grad_norm": 0.0,
      "learning_rate": 1.8228426068385094e-05,
      "loss": 1.0734,
      "step": 1737
    },
    {
      "epoch": 0.21678932268928527,
      "grad_norm": 0.0,
      "learning_rate": 1.8226129523977066e-05,
      "loss": 1.0767,
      "step": 1738
    },
    {
      "epoch": 0.21691405762754148,
      "grad_norm": 0.0,
      "learning_rate": 1.822383163685732e-05,
      "loss": 1.0911,
      "step": 1739
    },
    {
      "epoch": 0.21703879256579767,
      "grad_norm": 0.0,
      "learning_rate": 1.8221532407400926e-05,
      "loss": 1.0609,
      "step": 1740
    },
    {
      "epoch": 0.21716352750405388,
      "grad_norm": 0.0,
      "learning_rate": 1.8219231835983184e-05,
      "loss": 1.0646,
      "step": 1741
    },
    {
      "epoch": 0.2172882624423101,
      "grad_norm": 0.0,
      "learning_rate": 1.8216929922979596e-05,
      "loss": 1.0392,
      "step": 1742
    },
    {
      "epoch": 0.2174129973805663,
      "grad_norm": 0.0,
      "learning_rate": 1.82146266687659e-05,
      "loss": 1.0475,
      "step": 1743
    },
    {
      "epoch": 0.2175377323188225,
      "grad_norm": 0.0,
      "learning_rate": 1.8212322073718042e-05,
      "loss": 1.0326,
      "step": 1744
    },
    {
      "epoch": 0.21766246725707872,
      "grad_norm": 0.0,
      "learning_rate": 1.8210016138212186e-05,
      "loss": 1.0787,
      "step": 1745
    },
    {
      "epoch": 0.2177872021953349,
      "grad_norm": 0.0,
      "learning_rate": 1.8207708862624727e-05,
      "loss": 1.0443,
      "step": 1746
    },
    {
      "epoch": 0.2179119371335911,
      "grad_norm": 0.0,
      "learning_rate": 1.8205400247332266e-05,
      "loss": 1.0383,
      "step": 1747
    },
    {
      "epoch": 0.21803667207184732,
      "grad_norm": 0.0,
      "learning_rate": 1.8203090292711626e-05,
      "loss": 0.9978,
      "step": 1748
    },
    {
      "epoch": 0.21816140701010353,
      "grad_norm": 0.0,
      "learning_rate": 1.820077899913985e-05,
      "loss": 1.0588,
      "step": 1749
    },
    {
      "epoch": 0.21828614194835974,
      "grad_norm": 0.0,
      "learning_rate": 1.8198466366994208e-05,
      "loss": 1.0815,
      "step": 1750
    },
    {
      "epoch": 0.21841087688661595,
      "grad_norm": 0.0,
      "learning_rate": 1.8196152396652165e-05,
      "loss": 1.1064,
      "step": 1751
    },
    {
      "epoch": 0.21853561182487213,
      "grad_norm": 0.0,
      "learning_rate": 1.8193837088491432e-05,
      "loss": 1.0391,
      "step": 1752
    },
    {
      "epoch": 0.21866034676312834,
      "grad_norm": 0.0,
      "learning_rate": 1.819152044288992e-05,
      "loss": 1.102,
      "step": 1753
    },
    {
      "epoch": 0.21878508170138455,
      "grad_norm": 0.0,
      "learning_rate": 1.818920246022576e-05,
      "loss": 1.0462,
      "step": 1754
    },
    {
      "epoch": 0.21890981663964076,
      "grad_norm": 0.0,
      "learning_rate": 1.8186883140877317e-05,
      "loss": 1.0641,
      "step": 1755
    },
    {
      "epoch": 0.21903455157789697,
      "grad_norm": 0.0,
      "learning_rate": 1.818456248522315e-05,
      "loss": 1.0448,
      "step": 1756
    },
    {
      "epoch": 0.21915928651615318,
      "grad_norm": 0.0,
      "learning_rate": 1.8182240493642056e-05,
      "loss": 1.0791,
      "step": 1757
    },
    {
      "epoch": 0.21928402145440937,
      "grad_norm": 0.0,
      "learning_rate": 1.8179917166513042e-05,
      "loss": 1.0948,
      "step": 1758
    },
    {
      "epoch": 0.21940875639266558,
      "grad_norm": 0.0,
      "learning_rate": 1.817759250421533e-05,
      "loss": 1.0094,
      "step": 1759
    },
    {
      "epoch": 0.2195334913309218,
      "grad_norm": 0.0,
      "learning_rate": 1.8175266507128365e-05,
      "loss": 1.0816,
      "step": 1760
    },
    {
      "epoch": 0.219658226269178,
      "grad_norm": 0.0,
      "learning_rate": 1.8172939175631807e-05,
      "loss": 1.0272,
      "step": 1761
    },
    {
      "epoch": 0.2197829612074342,
      "grad_norm": 0.0,
      "learning_rate": 1.817061051010554e-05,
      "loss": 1.0652,
      "step": 1762
    },
    {
      "epoch": 0.21990769614569042,
      "grad_norm": 0.0,
      "learning_rate": 1.8168280510929656e-05,
      "loss": 1.0679,
      "step": 1763
    },
    {
      "epoch": 0.2200324310839466,
      "grad_norm": 0.0,
      "learning_rate": 1.816594917848447e-05,
      "loss": 1.0148,
      "step": 1764
    },
    {
      "epoch": 0.2201571660222028,
      "grad_norm": 0.0,
      "learning_rate": 1.816361651315052e-05,
      "loss": 1.0828,
      "step": 1765
    },
    {
      "epoch": 0.22028190096045902,
      "grad_norm": 0.0,
      "learning_rate": 1.816128251530855e-05,
      "loss": 1.0452,
      "step": 1766
    },
    {
      "epoch": 0.22040663589871523,
      "grad_norm": 0.0,
      "learning_rate": 1.8158947185339525e-05,
      "loss": 1.089,
      "step": 1767
    },
    {
      "epoch": 0.22053137083697144,
      "grad_norm": 0.0,
      "learning_rate": 1.8156610523624637e-05,
      "loss": 1.0167,
      "step": 1768
    },
    {
      "epoch": 0.22065610577522765,
      "grad_norm": 0.0,
      "learning_rate": 1.8154272530545283e-05,
      "loss": 1.0865,
      "step": 1769
    },
    {
      "epoch": 0.22078084071348383,
      "grad_norm": 0.0,
      "learning_rate": 1.8151933206483087e-05,
      "loss": 1.0165,
      "step": 1770
    },
    {
      "epoch": 0.22090557565174004,
      "grad_norm": 0.0,
      "learning_rate": 1.8149592551819883e-05,
      "loss": 1.0798,
      "step": 1771
    },
    {
      "epoch": 0.22103031058999625,
      "grad_norm": 0.0,
      "learning_rate": 1.814725056693772e-05,
      "loss": 1.06,
      "step": 1772
    },
    {
      "epoch": 0.22115504552825246,
      "grad_norm": 0.0,
      "learning_rate": 1.8144907252218875e-05,
      "loss": 1.079,
      "step": 1773
    },
    {
      "epoch": 0.22127978046650867,
      "grad_norm": 0.0,
      "learning_rate": 1.8142562608045834e-05,
      "loss": 1.0981,
      "step": 1774
    },
    {
      "epoch": 0.22140451540476488,
      "grad_norm": 0.0,
      "learning_rate": 1.8140216634801304e-05,
      "loss": 1.0974,
      "step": 1775
    },
    {
      "epoch": 0.22152925034302107,
      "grad_norm": 0.0,
      "learning_rate": 1.8137869332868207e-05,
      "loss": 1.1015,
      "step": 1776
    },
    {
      "epoch": 0.22165398528127728,
      "grad_norm": 0.0,
      "learning_rate": 1.8135520702629677e-05,
      "loss": 1.0685,
      "step": 1777
    },
    {
      "epoch": 0.2217787202195335,
      "grad_norm": 0.0,
      "learning_rate": 1.8133170744469074e-05,
      "loss": 1.0343,
      "step": 1778
    },
    {
      "epoch": 0.2219034551577897,
      "grad_norm": 0.0,
      "learning_rate": 1.813081945876997e-05,
      "loss": 1.0254,
      "step": 1779
    },
    {
      "epoch": 0.2220281900960459,
      "grad_norm": 0.0,
      "learning_rate": 1.8128466845916156e-05,
      "loss": 1.0472,
      "step": 1780
    },
    {
      "epoch": 0.22215292503430212,
      "grad_norm": 0.0,
      "learning_rate": 1.8126112906291632e-05,
      "loss": 1.117,
      "step": 1781
    },
    {
      "epoch": 0.2222776599725583,
      "grad_norm": 0.0,
      "learning_rate": 1.812375764028062e-05,
      "loss": 1.0663,
      "step": 1782
    },
    {
      "epoch": 0.2224023949108145,
      "grad_norm": 0.0,
      "learning_rate": 1.812140104826757e-05,
      "loss": 0.9966,
      "step": 1783
    },
    {
      "epoch": 0.22252712984907072,
      "grad_norm": 0.0,
      "learning_rate": 1.8119043130637127e-05,
      "loss": 1.0028,
      "step": 1784
    },
    {
      "epoch": 0.22265186478732693,
      "grad_norm": 0.0,
      "learning_rate": 1.8116683887774164e-05,
      "loss": 1.1522,
      "step": 1785
    },
    {
      "epoch": 0.22277659972558314,
      "grad_norm": 0.0,
      "learning_rate": 1.811432332006377e-05,
      "loss": 1.0181,
      "step": 1786
    },
    {
      "epoch": 0.22290133466383935,
      "grad_norm": 0.0,
      "learning_rate": 1.811196142789125e-05,
      "loss": 1.0021,
      "step": 1787
    },
    {
      "epoch": 0.22302606960209553,
      "grad_norm": 0.0,
      "learning_rate": 1.810959821164212e-05,
      "loss": 1.0306,
      "step": 1788
    },
    {
      "epoch": 0.22315080454035174,
      "grad_norm": 0.0,
      "learning_rate": 1.8107233671702123e-05,
      "loss": 1.0308,
      "step": 1789
    },
    {
      "epoch": 0.22327553947860795,
      "grad_norm": 0.0,
      "learning_rate": 1.810486780845721e-05,
      "loss": 1.0576,
      "step": 1790
    },
    {
      "epoch": 0.22340027441686416,
      "grad_norm": 0.0,
      "learning_rate": 1.8102500622293546e-05,
      "loss": 1.0626,
      "step": 1791
    },
    {
      "epoch": 0.22352500935512037,
      "grad_norm": 0.0,
      "learning_rate": 1.810013211359752e-05,
      "loss": 1.0759,
      "step": 1792
    },
    {
      "epoch": 0.22364974429337658,
      "grad_norm": 0.0,
      "learning_rate": 1.8097762282755726e-05,
      "loss": 1.0497,
      "step": 1793
    },
    {
      "epoch": 0.22377447923163277,
      "grad_norm": 0.0,
      "learning_rate": 1.809539113015499e-05,
      "loss": 1.0741,
      "step": 1794
    },
    {
      "epoch": 0.22389921416988898,
      "grad_norm": 0.0,
      "learning_rate": 1.8093018656182333e-05,
      "loss": 1.1189,
      "step": 1795
    },
    {
      "epoch": 0.2240239491081452,
      "grad_norm": 0.0,
      "learning_rate": 1.8090644861225008e-05,
      "loss": 1.0378,
      "step": 1796
    },
    {
      "epoch": 0.2241486840464014,
      "grad_norm": 0.0,
      "learning_rate": 1.808826974567048e-05,
      "loss": 1.0766,
      "step": 1797
    },
    {
      "epoch": 0.2242734189846576,
      "grad_norm": 0.0,
      "learning_rate": 1.808589330990642e-05,
      "loss": 1.0382,
      "step": 1798
    },
    {
      "epoch": 0.22439815392291382,
      "grad_norm": 0.0,
      "learning_rate": 1.8083515554320734e-05,
      "loss": 1.0745,
      "step": 1799
    },
    {
      "epoch": 0.22452288886117,
      "grad_norm": 0.0,
      "learning_rate": 1.8081136479301522e-05,
      "loss": 1.0518,
      "step": 1800
    },
    {
      "epoch": 0.2246476237994262,
      "grad_norm": 0.0,
      "learning_rate": 1.8078756085237112e-05,
      "loss": 1.1021,
      "step": 1801
    },
    {
      "epoch": 0.22477235873768242,
      "grad_norm": 0.0,
      "learning_rate": 1.8076374372516045e-05,
      "loss": 1.0914,
      "step": 1802
    },
    {
      "epoch": 0.22489709367593863,
      "grad_norm": 0.0,
      "learning_rate": 1.8073991341527078e-05,
      "loss": 1.0106,
      "step": 1803
    },
    {
      "epoch": 0.22502182861419484,
      "grad_norm": 0.0,
      "learning_rate": 1.807160699265918e-05,
      "loss": 1.0367,
      "step": 1804
    },
    {
      "epoch": 0.22514656355245105,
      "grad_norm": 0.0,
      "learning_rate": 1.8069221326301534e-05,
      "loss": 1.0063,
      "step": 1805
    },
    {
      "epoch": 0.22527129849070723,
      "grad_norm": 0.0,
      "learning_rate": 1.8066834342843546e-05,
      "loss": 1.0144,
      "step": 1806
    },
    {
      "epoch": 0.22539603342896344,
      "grad_norm": 0.0,
      "learning_rate": 1.806444604267483e-05,
      "loss": 1.0192,
      "step": 1807
    },
    {
      "epoch": 0.22552076836721965,
      "grad_norm": 0.0,
      "learning_rate": 1.8062056426185214e-05,
      "loss": 1.0716,
      "step": 1808
    },
    {
      "epoch": 0.22564550330547586,
      "grad_norm": 0.0,
      "learning_rate": 1.8059665493764745e-05,
      "loss": 1.0638,
      "step": 1809
    },
    {
      "epoch": 0.22577023824373207,
      "grad_norm": 0.0,
      "learning_rate": 1.8057273245803686e-05,
      "loss": 1.0258,
      "step": 1810
    },
    {
      "epoch": 0.22589497318198828,
      "grad_norm": 0.0,
      "learning_rate": 1.805487968269251e-05,
      "loss": 1.0302,
      "step": 1811
    },
    {
      "epoch": 0.22601970812024447,
      "grad_norm": 0.0,
      "learning_rate": 1.8052484804821907e-05,
      "loss": 1.0575,
      "step": 1812
    },
    {
      "epoch": 0.22614444305850068,
      "grad_norm": 0.0,
      "learning_rate": 1.8050088612582785e-05,
      "loss": 1.0584,
      "step": 1813
    },
    {
      "epoch": 0.2262691779967569,
      "grad_norm": 0.0,
      "learning_rate": 1.8047691106366256e-05,
      "loss": 1.0707,
      "step": 1814
    },
    {
      "epoch": 0.2263939129350131,
      "grad_norm": 0.0,
      "learning_rate": 1.804529228656366e-05,
      "loss": 1.0364,
      "step": 1815
    },
    {
      "epoch": 0.2265186478732693,
      "grad_norm": 0.0,
      "learning_rate": 1.8042892153566543e-05,
      "loss": 1.0766,
      "step": 1816
    },
    {
      "epoch": 0.22664338281152552,
      "grad_norm": 0.0,
      "learning_rate": 1.8040490707766664e-05,
      "loss": 1.0434,
      "step": 1817
    },
    {
      "epoch": 0.2267681177497817,
      "grad_norm": 0.0,
      "learning_rate": 1.8038087949556003e-05,
      "loss": 1.0752,
      "step": 1818
    },
    {
      "epoch": 0.2268928526880379,
      "grad_norm": 0.0,
      "learning_rate": 1.803568387932675e-05,
      "loss": 1.0541,
      "step": 1819
    },
    {
      "epoch": 0.22701758762629412,
      "grad_norm": 0.0,
      "learning_rate": 1.803327849747131e-05,
      "loss": 1.0537,
      "step": 1820
    },
    {
      "epoch": 0.22714232256455033,
      "grad_norm": 0.0,
      "learning_rate": 1.80308718043823e-05,
      "loss": 1.0695,
      "step": 1821
    },
    {
      "epoch": 0.22726705750280654,
      "grad_norm": 0.0,
      "learning_rate": 1.8028463800452553e-05,
      "loss": 1.0857,
      "step": 1822
    },
    {
      "epoch": 0.22739179244106275,
      "grad_norm": 0.0,
      "learning_rate": 1.8026054486075116e-05,
      "loss": 1.0704,
      "step": 1823
    },
    {
      "epoch": 0.22751652737931893,
      "grad_norm": 0.0,
      "learning_rate": 1.8023643861643255e-05,
      "loss": 1.0532,
      "step": 1824
    },
    {
      "epoch": 0.22764126231757514,
      "grad_norm": 0.0,
      "learning_rate": 1.802123192755044e-05,
      "loss": 1.1274,
      "step": 1825
    },
    {
      "epoch": 0.22776599725583135,
      "grad_norm": 0.0,
      "learning_rate": 1.801881868419036e-05,
      "loss": 1.0894,
      "step": 1826
    },
    {
      "epoch": 0.22789073219408756,
      "grad_norm": 0.0,
      "learning_rate": 1.801640413195692e-05,
      "loss": 1.0532,
      "step": 1827
    },
    {
      "epoch": 0.22801546713234377,
      "grad_norm": 0.0,
      "learning_rate": 1.8013988271244227e-05,
      "loss": 1.0669,
      "step": 1828
    },
    {
      "epoch": 0.22814020207059998,
      "grad_norm": 0.0,
      "learning_rate": 1.8011571102446622e-05,
      "loss": 1.0504,
      "step": 1829
    },
    {
      "epoch": 0.22826493700885617,
      "grad_norm": 0.0,
      "learning_rate": 1.800915262595864e-05,
      "loss": 1.0608,
      "step": 1830
    },
    {
      "epoch": 0.22838967194711238,
      "grad_norm": 0.0,
      "learning_rate": 1.800673284217504e-05,
      "loss": 1.0586,
      "step": 1831
    },
    {
      "epoch": 0.2285144068853686,
      "grad_norm": 0.0,
      "learning_rate": 1.8004311751490792e-05,
      "loss": 1.0126,
      "step": 1832
    },
    {
      "epoch": 0.2286391418236248,
      "grad_norm": 0.0,
      "learning_rate": 1.800188935430108e-05,
      "loss": 1.0164,
      "step": 1833
    },
    {
      "epoch": 0.228763876761881,
      "grad_norm": 0.0,
      "learning_rate": 1.7999465651001297e-05,
      "loss": 1.0937,
      "step": 1834
    },
    {
      "epoch": 0.22888861170013722,
      "grad_norm": 0.0,
      "learning_rate": 1.7997040641987057e-05,
      "loss": 1.0739,
      "step": 1835
    },
    {
      "epoch": 0.2290133466383934,
      "grad_norm": 0.0,
      "learning_rate": 1.7994614327654178e-05,
      "loss": 1.0069,
      "step": 1836
    },
    {
      "epoch": 0.2291380815766496,
      "grad_norm": 0.0,
      "learning_rate": 1.7992186708398697e-05,
      "loss": 1.0046,
      "step": 1837
    },
    {
      "epoch": 0.22926281651490582,
      "grad_norm": 0.0,
      "learning_rate": 1.7989757784616862e-05,
      "loss": 1.0585,
      "step": 1838
    },
    {
      "epoch": 0.22938755145316203,
      "grad_norm": 0.0,
      "learning_rate": 1.798732755670514e-05,
      "loss": 1.0752,
      "step": 1839
    },
    {
      "epoch": 0.22951228639141824,
      "grad_norm": 0.0,
      "learning_rate": 1.79848960250602e-05,
      "loss": 1.0277,
      "step": 1840
    },
    {
      "epoch": 0.22963702132967445,
      "grad_norm": 0.0,
      "learning_rate": 1.7982463190078928e-05,
      "loss": 1.0604,
      "step": 1841
    },
    {
      "epoch": 0.22976175626793063,
      "grad_norm": 0.0,
      "learning_rate": 1.7980029052158432e-05,
      "loss": 1.0631,
      "step": 1842
    },
    {
      "epoch": 0.22988649120618684,
      "grad_norm": 0.0,
      "learning_rate": 1.7977593611696017e-05,
      "loss": 1.0158,
      "step": 1843
    },
    {
      "epoch": 0.23001122614444305,
      "grad_norm": 0.0,
      "learning_rate": 1.7975156869089207e-05,
      "loss": 1.0368,
      "step": 1844
    },
    {
      "epoch": 0.23013596108269926,
      "grad_norm": 0.0,
      "learning_rate": 1.7972718824735748e-05,
      "loss": 1.0525,
      "step": 1845
    },
    {
      "epoch": 0.23026069602095547,
      "grad_norm": 0.0,
      "learning_rate": 1.7970279479033583e-05,
      "loss": 1.0644,
      "step": 1846
    },
    {
      "epoch": 0.23038543095921168,
      "grad_norm": 0.0,
      "learning_rate": 1.7967838832380878e-05,
      "loss": 1.0251,
      "step": 1847
    },
    {
      "epoch": 0.23051016589746787,
      "grad_norm": 0.0,
      "learning_rate": 1.7965396885176008e-05,
      "loss": 1.1198,
      "step": 1848
    },
    {
      "epoch": 0.23063490083572408,
      "grad_norm": 0.0,
      "learning_rate": 1.7962953637817556e-05,
      "loss": 1.0072,
      "step": 1849
    },
    {
      "epoch": 0.2307596357739803,
      "grad_norm": 0.0,
      "learning_rate": 1.796050909070433e-05,
      "loss": 1.0572,
      "step": 1850
    },
    {
      "epoch": 0.2308843707122365,
      "grad_norm": 0.0,
      "learning_rate": 1.795806324423534e-05,
      "loss": 1.0542,
      "step": 1851
    },
    {
      "epoch": 0.2310091056504927,
      "grad_norm": 0.0,
      "learning_rate": 1.79556160988098e-05,
      "loss": 1.0456,
      "step": 1852
    },
    {
      "epoch": 0.23113384058874892,
      "grad_norm": 0.0,
      "learning_rate": 1.7953167654827157e-05,
      "loss": 1.0407,
      "step": 1853
    },
    {
      "epoch": 0.23125857552700513,
      "grad_norm": 0.0,
      "learning_rate": 1.795071791268705e-05,
      "loss": 1.0536,
      "step": 1854
    },
    {
      "epoch": 0.2313833104652613,
      "grad_norm": 0.0,
      "learning_rate": 1.7948266872789348e-05,
      "loss": 1.0666,
      "step": 1855
    },
    {
      "epoch": 0.23150804540351752,
      "grad_norm": 0.0,
      "learning_rate": 1.7945814535534116e-05,
      "loss": 1.0382,
      "step": 1856
    },
    {
      "epoch": 0.23163278034177373,
      "grad_norm": 0.0,
      "learning_rate": 1.794336090132164e-05,
      "loss": 1.036,
      "step": 1857
    },
    {
      "epoch": 0.23175751528002994,
      "grad_norm": 0.0,
      "learning_rate": 1.794090597055241e-05,
      "loss": 1.0613,
      "step": 1858
    },
    {
      "epoch": 0.23188225021828615,
      "grad_norm": 0.0,
      "learning_rate": 1.7938449743627143e-05,
      "loss": 1.0689,
      "step": 1859
    },
    {
      "epoch": 0.23200698515654236,
      "grad_norm": 0.0,
      "learning_rate": 1.7935992220946746e-05,
      "loss": 1.0841,
      "step": 1860
    },
    {
      "epoch": 0.23213172009479854,
      "grad_norm": 0.0,
      "learning_rate": 1.7933533402912354e-05,
      "loss": 1.0845,
      "step": 1861
    },
    {
      "epoch": 0.23225645503305475,
      "grad_norm": 0.0,
      "learning_rate": 1.7931073289925308e-05,
      "loss": 1.0859,
      "step": 1862
    },
    {
      "epoch": 0.23238118997131096,
      "grad_norm": 0.0,
      "learning_rate": 1.7928611882387157e-05,
      "loss": 1.0575,
      "step": 1863
    },
    {
      "epoch": 0.23250592490956717,
      "grad_norm": 0.0,
      "learning_rate": 1.792614918069967e-05,
      "loss": 1.0638,
      "step": 1864
    },
    {
      "epoch": 0.23263065984782338,
      "grad_norm": 0.0,
      "learning_rate": 1.792368518526482e-05,
      "loss": 1.0422,
      "step": 1865
    },
    {
      "epoch": 0.2327553947860796,
      "grad_norm": 0.0,
      "learning_rate": 1.7921219896484793e-05,
      "loss": 1.0571,
      "step": 1866
    },
    {
      "epoch": 0.23288012972433578,
      "grad_norm": 0.0,
      "learning_rate": 1.7918753314761987e-05,
      "loss": 1.0613,
      "step": 1867
    },
    {
      "epoch": 0.233004864662592,
      "grad_norm": 0.0,
      "learning_rate": 1.7916285440499005e-05,
      "loss": 1.0913,
      "step": 1868
    },
    {
      "epoch": 0.2331295996008482,
      "grad_norm": 0.0,
      "learning_rate": 1.7913816274098678e-05,
      "loss": 1.0973,
      "step": 1869
    },
    {
      "epoch": 0.2332543345391044,
      "grad_norm": 0.0,
      "learning_rate": 1.791134581596402e-05,
      "loss": 1.0691,
      "step": 1870
    },
    {
      "epoch": 0.23337906947736062,
      "grad_norm": 0.0,
      "learning_rate": 1.7908874066498286e-05,
      "loss": 0.9701,
      "step": 1871
    },
    {
      "epoch": 0.23350380441561683,
      "grad_norm": 0.0,
      "learning_rate": 1.7906401026104923e-05,
      "loss": 1.0584,
      "step": 1872
    },
    {
      "epoch": 0.233628539353873,
      "grad_norm": 0.0,
      "learning_rate": 1.7903926695187595e-05,
      "loss": 1.084,
      "step": 1873
    },
    {
      "epoch": 0.23375327429212922,
      "grad_norm": 0.0,
      "learning_rate": 1.7901451074150173e-05,
      "loss": 1.0734,
      "step": 1874
    },
    {
      "epoch": 0.23387800923038543,
      "grad_norm": 0.0,
      "learning_rate": 1.789897416339674e-05,
      "loss": 1.0825,
      "step": 1875
    },
    {
      "epoch": 0.23400274416864164,
      "grad_norm": 0.0,
      "learning_rate": 1.7896495963331588e-05,
      "loss": 1.033,
      "step": 1876
    },
    {
      "epoch": 0.23412747910689785,
      "grad_norm": 0.0,
      "learning_rate": 1.789401647435923e-05,
      "loss": 1.003,
      "step": 1877
    },
    {
      "epoch": 0.23425221404515406,
      "grad_norm": 0.0,
      "learning_rate": 1.7891535696884372e-05,
      "loss": 1.1092,
      "step": 1878
    },
    {
      "epoch": 0.23437694898341024,
      "grad_norm": 0.0,
      "learning_rate": 1.7889053631311947e-05,
      "loss": 1.1016,
      "step": 1879
    },
    {
      "epoch": 0.23450168392166645,
      "grad_norm": 0.0,
      "learning_rate": 1.7886570278047087e-05,
      "loss": 1.0772,
      "step": 1880
    },
    {
      "epoch": 0.23462641885992266,
      "grad_norm": 0.0,
      "learning_rate": 1.788408563749514e-05,
      "loss": 1.0347,
      "step": 1881
    },
    {
      "epoch": 0.23475115379817887,
      "grad_norm": 0.0,
      "learning_rate": 1.7881599710061657e-05,
      "loss": 1.0842,
      "step": 1882
    },
    {
      "epoch": 0.23487588873643508,
      "grad_norm": 0.0,
      "learning_rate": 1.7879112496152408e-05,
      "loss": 1.0596,
      "step": 1883
    },
    {
      "epoch": 0.2350006236746913,
      "grad_norm": 0.0,
      "learning_rate": 1.787662399617337e-05,
      "loss": 1.0836,
      "step": 1884
    },
    {
      "epoch": 0.23512535861294748,
      "grad_norm": 0.0,
      "learning_rate": 1.7874134210530725e-05,
      "loss": 1.0551,
      "step": 1885
    },
    {
      "epoch": 0.2352500935512037,
      "grad_norm": 0.0,
      "learning_rate": 1.7871643139630872e-05,
      "loss": 1.0454,
      "step": 1886
    },
    {
      "epoch": 0.2353748284894599,
      "grad_norm": 0.0,
      "learning_rate": 1.7869150783880415e-05,
      "loss": 1.0806,
      "step": 1887
    },
    {
      "epoch": 0.2354995634277161,
      "grad_norm": 0.0,
      "learning_rate": 1.786665714368617e-05,
      "loss": 1.0546,
      "step": 1888
    },
    {
      "epoch": 0.23562429836597232,
      "grad_norm": 0.0,
      "learning_rate": 1.7864162219455164e-05,
      "loss": 1.0266,
      "step": 1889
    },
    {
      "epoch": 0.23574903330422853,
      "grad_norm": 0.0,
      "learning_rate": 1.786166601159463e-05,
      "loss": 1.019,
      "step": 1890
    },
    {
      "epoch": 0.2358737682424847,
      "grad_norm": 0.0,
      "learning_rate": 1.7859168520512008e-05,
      "loss": 1.0535,
      "step": 1891
    },
    {
      "epoch": 0.23599850318074092,
      "grad_norm": 0.0,
      "learning_rate": 1.7856669746614955e-05,
      "loss": 1.0523,
      "step": 1892
    },
    {
      "epoch": 0.23612323811899713,
      "grad_norm": 0.0,
      "learning_rate": 1.7854169690311338e-05,
      "loss": 1.0672,
      "step": 1893
    },
    {
      "epoch": 0.23624797305725334,
      "grad_norm": 0.0,
      "learning_rate": 1.785166835200922e-05,
      "loss": 1.0343,
      "step": 1894
    },
    {
      "epoch": 0.23637270799550955,
      "grad_norm": 0.0,
      "learning_rate": 1.784916573211689e-05,
      "loss": 1.0745,
      "step": 1895
    },
    {
      "epoch": 0.23649744293376576,
      "grad_norm": 0.0,
      "learning_rate": 1.784666183104284e-05,
      "loss": 1.0105,
      "step": 1896
    },
    {
      "epoch": 0.23662217787202194,
      "grad_norm": 0.0,
      "learning_rate": 1.784415664919576e-05,
      "loss": 1.0205,
      "step": 1897
    },
    {
      "epoch": 0.23674691281027815,
      "grad_norm": 0.0,
      "learning_rate": 1.7841650186984565e-05,
      "loss": 1.0307,
      "step": 1898
    },
    {
      "epoch": 0.23687164774853436,
      "grad_norm": 0.0,
      "learning_rate": 1.7839142444818375e-05,
      "loss": 1.1057,
      "step": 1899
    },
    {
      "epoch": 0.23699638268679057,
      "grad_norm": 0.0,
      "learning_rate": 1.7836633423106515e-05,
      "loss": 1.0036,
      "step": 1900
    },
    {
      "epoch": 0.23712111762504678,
      "grad_norm": 0.0,
      "learning_rate": 1.783412312225852e-05,
      "loss": 1.0777,
      "step": 1901
    },
    {
      "epoch": 0.237245852563303,
      "grad_norm": 0.0,
      "learning_rate": 1.783161154268413e-05,
      "loss": 1.113,
      "step": 1902
    },
    {
      "epoch": 0.23737058750155918,
      "grad_norm": 0.0,
      "learning_rate": 1.7829098684793303e-05,
      "loss": 1.0341,
      "step": 1903
    },
    {
      "epoch": 0.2374953224398154,
      "grad_norm": 0.0,
      "learning_rate": 1.7826584548996205e-05,
      "loss": 1.1264,
      "step": 1904
    },
    {
      "epoch": 0.2376200573780716,
      "grad_norm": 0.0,
      "learning_rate": 1.78240691357032e-05,
      "loss": 1.0705,
      "step": 1905
    },
    {
      "epoch": 0.2377447923163278,
      "grad_norm": 0.0,
      "learning_rate": 1.782155244532487e-05,
      "loss": 1.0731,
      "step": 1906
    },
    {
      "epoch": 0.23786952725458402,
      "grad_norm": 0.0,
      "learning_rate": 1.7819034478271993e-05,
      "loss": 1.0722,
      "step": 1907
    },
    {
      "epoch": 0.23799426219284023,
      "grad_norm": 0.0,
      "learning_rate": 1.7816515234955578e-05,
      "loss": 1.0335,
      "step": 1908
    },
    {
      "epoch": 0.2381189971310964,
      "grad_norm": 0.0,
      "learning_rate": 1.7813994715786825e-05,
      "loss": 1.0423,
      "step": 1909
    },
    {
      "epoch": 0.23824373206935262,
      "grad_norm": 0.0,
      "learning_rate": 1.7811472921177143e-05,
      "loss": 1.0686,
      "step": 1910
    },
    {
      "epoch": 0.23836846700760883,
      "grad_norm": 0.0,
      "learning_rate": 1.7808949851538152e-05,
      "loss": 1.0398,
      "step": 1911
    },
    {
      "epoch": 0.23849320194586504,
      "grad_norm": 0.0,
      "learning_rate": 1.7806425507281684e-05,
      "loss": 1.046,
      "step": 1912
    },
    {
      "epoch": 0.23861793688412125,
      "grad_norm": 0.0,
      "learning_rate": 1.7803899888819777e-05,
      "loss": 1.0593,
      "step": 1913
    },
    {
      "epoch": 0.23874267182237746,
      "grad_norm": 0.0,
      "learning_rate": 1.780137299656467e-05,
      "loss": 1.0556,
      "step": 1914
    },
    {
      "epoch": 0.23886740676063364,
      "grad_norm": 0.0,
      "learning_rate": 1.7798844830928818e-05,
      "loss": 1.0759,
      "step": 1915
    },
    {
      "epoch": 0.23899214169888985,
      "grad_norm": 0.0,
      "learning_rate": 1.7796315392324884e-05,
      "loss": 1.0449,
      "step": 1916
    },
    {
      "epoch": 0.23911687663714606,
      "grad_norm": 0.0,
      "learning_rate": 1.7793784681165734e-05,
      "loss": 1.0227,
      "step": 1917
    },
    {
      "epoch": 0.23924161157540227,
      "grad_norm": 0.0,
      "learning_rate": 1.7791252697864444e-05,
      "loss": 1.0389,
      "step": 1918
    },
    {
      "epoch": 0.23936634651365848,
      "grad_norm": 0.0,
      "learning_rate": 1.7788719442834293e-05,
      "loss": 0.991,
      "step": 1919
    },
    {
      "epoch": 0.2394910814519147,
      "grad_norm": 0.0,
      "learning_rate": 1.7786184916488784e-05,
      "loss": 1.0585,
      "step": 1920
    },
    {
      "epoch": 0.23961581639017088,
      "grad_norm": 0.0,
      "learning_rate": 1.7783649119241603e-05,
      "loss": 1.0001,
      "step": 1921
    },
    {
      "epoch": 0.2397405513284271,
      "grad_norm": 0.0,
      "learning_rate": 1.778111205150666e-05,
      "loss": 1.0545,
      "step": 1922
    },
    {
      "epoch": 0.2398652862666833,
      "grad_norm": 0.0,
      "learning_rate": 1.777857371369807e-05,
      "loss": 1.0389,
      "step": 1923
    },
    {
      "epoch": 0.2399900212049395,
      "grad_norm": 0.0,
      "learning_rate": 1.7776034106230156e-05,
      "loss": 1.0438,
      "step": 1924
    },
    {
      "epoch": 0.24011475614319572,
      "grad_norm": 0.0,
      "learning_rate": 1.777349322951744e-05,
      "loss": 1.0849,
      "step": 1925
    },
    {
      "epoch": 0.24023949108145193,
      "grad_norm": 0.0,
      "learning_rate": 1.7770951083974662e-05,
      "loss": 1.0171,
      "step": 1926
    },
    {
      "epoch": 0.2403642260197081,
      "grad_norm": 0.0,
      "learning_rate": 1.7768407670016766e-05,
      "loss": 1.0455,
      "step": 1927
    },
    {
      "epoch": 0.24048896095796432,
      "grad_norm": 0.0,
      "learning_rate": 1.7765862988058893e-05,
      "loss": 1.0357,
      "step": 1928
    },
    {
      "epoch": 0.24061369589622053,
      "grad_norm": 0.0,
      "learning_rate": 1.7763317038516407e-05,
      "loss": 1.0214,
      "step": 1929
    },
    {
      "epoch": 0.24073843083447674,
      "grad_norm": 0.0,
      "learning_rate": 1.7760769821804867e-05,
      "loss": 0.9917,
      "step": 1930
    },
    {
      "epoch": 0.24086316577273295,
      "grad_norm": 0.0,
      "learning_rate": 1.7758221338340043e-05,
      "loss": 1.009,
      "step": 1931
    },
    {
      "epoch": 0.24098790071098916,
      "grad_norm": 0.0,
      "learning_rate": 1.7755671588537913e-05,
      "loss": 1.0064,
      "step": 1932
    },
    {
      "epoch": 0.24111263564924534,
      "grad_norm": 0.0,
      "learning_rate": 1.775312057281466e-05,
      "loss": 1.0238,
      "step": 1933
    },
    {
      "epoch": 0.24123737058750155,
      "grad_norm": 0.0,
      "learning_rate": 1.7750568291586676e-05,
      "loss": 1.0815,
      "step": 1934
    },
    {
      "epoch": 0.24136210552575776,
      "grad_norm": 0.0,
      "learning_rate": 1.7748014745270556e-05,
      "loss": 1.0509,
      "step": 1935
    },
    {
      "epoch": 0.24148684046401397,
      "grad_norm": 0.0,
      "learning_rate": 1.7745459934283103e-05,
      "loss": 1.0606,
      "step": 1936
    },
    {
      "epoch": 0.24161157540227018,
      "grad_norm": 0.0,
      "learning_rate": 1.7742903859041324e-05,
      "loss": 1.0353,
      "step": 1937
    },
    {
      "epoch": 0.2417363103405264,
      "grad_norm": 0.0,
      "learning_rate": 1.7740346519962443e-05,
      "loss": 1.0705,
      "step": 1938
    },
    {
      "epoch": 0.24186104527878258,
      "grad_norm": 0.0,
      "learning_rate": 1.7737787917463875e-05,
      "loss": 1.04,
      "step": 1939
    },
    {
      "epoch": 0.2419857802170388,
      "grad_norm": 0.0,
      "learning_rate": 1.7735228051963248e-05,
      "loss": 1.0318,
      "step": 1940
    },
    {
      "epoch": 0.242110515155295,
      "grad_norm": 0.0,
      "learning_rate": 1.77326669238784e-05,
      "loss": 1.0293,
      "step": 1941
    },
    {
      "epoch": 0.2422352500935512,
      "grad_norm": 0.0,
      "learning_rate": 1.773010453362737e-05,
      "loss": 1.0164,
      "step": 1942
    },
    {
      "epoch": 0.24235998503180742,
      "grad_norm": 0.0,
      "learning_rate": 1.7727540881628407e-05,
      "loss": 1.0253,
      "step": 1943
    },
    {
      "epoch": 0.24248471997006363,
      "grad_norm": 0.0,
      "learning_rate": 1.7724975968299963e-05,
      "loss": 1.0181,
      "step": 1944
    },
    {
      "epoch": 0.2426094549083198,
      "grad_norm": 0.0,
      "learning_rate": 1.7722409794060692e-05,
      "loss": 1.0122,
      "step": 1945
    },
    {
      "epoch": 0.24273418984657602,
      "grad_norm": 0.0,
      "learning_rate": 1.7719842359329463e-05,
      "loss": 1.0059,
      "step": 1946
    },
    {
      "epoch": 0.24285892478483223,
      "grad_norm": 0.0,
      "learning_rate": 1.7717273664525347e-05,
      "loss": 1.0058,
      "step": 1947
    },
    {
      "epoch": 0.24298365972308844,
      "grad_norm": 0.0,
      "learning_rate": 1.7714703710067615e-05,
      "loss": 1.0389,
      "step": 1948
    },
    {
      "epoch": 0.24310839466134465,
      "grad_norm": 0.0,
      "learning_rate": 1.771213249637575e-05,
      "loss": 1.1012,
      "step": 1949
    },
    {
      "epoch": 0.24323312959960086,
      "grad_norm": 0.0,
      "learning_rate": 1.7709560023869444e-05,
      "loss": 1.0652,
      "step": 1950
    },
    {
      "epoch": 0.24335786453785704,
      "grad_norm": 0.0,
      "learning_rate": 1.770698629296858e-05,
      "loss": 1.0384,
      "step": 1951
    },
    {
      "epoch": 0.24348259947611325,
      "grad_norm": 0.0,
      "learning_rate": 1.7704411304093267e-05,
      "loss": 1.0513,
      "step": 1952
    },
    {
      "epoch": 0.24360733441436946,
      "grad_norm": 0.0,
      "learning_rate": 1.77018350576638e-05,
      "loss": 1.0316,
      "step": 1953
    },
    {
      "epoch": 0.24373206935262567,
      "grad_norm": 0.0,
      "learning_rate": 1.7699257554100688e-05,
      "loss": 1.0608,
      "step": 1954
    },
    {
      "epoch": 0.24385680429088188,
      "grad_norm": 0.0,
      "learning_rate": 1.7696678793824644e-05,
      "loss": 1.0107,
      "step": 1955
    },
    {
      "epoch": 0.2439815392291381,
      "grad_norm": 0.0,
      "learning_rate": 1.769409877725659e-05,
      "loss": 1.0363,
      "step": 1956
    },
    {
      "epoch": 0.24410627416739428,
      "grad_norm": 0.0,
      "learning_rate": 1.7691517504817652e-05,
      "loss": 1.079,
      "step": 1957
    },
    {
      "epoch": 0.2442310091056505,
      "grad_norm": 0.0,
      "learning_rate": 1.7688934976929152e-05,
      "loss": 1.0403,
      "step": 1958
    },
    {
      "epoch": 0.2443557440439067,
      "grad_norm": 0.0,
      "learning_rate": 1.768635119401263e-05,
      "loss": 1.0248,
      "step": 1959
    },
    {
      "epoch": 0.2444804789821629,
      "grad_norm": 0.0,
      "learning_rate": 1.7683766156489817e-05,
      "loss": 1.0669,
      "step": 1960
    },
    {
      "epoch": 0.24460521392041912,
      "grad_norm": 0.0,
      "learning_rate": 1.768117986478266e-05,
      "loss": 1.0017,
      "step": 1961
    },
    {
      "epoch": 0.24472994885867533,
      "grad_norm": 0.0,
      "learning_rate": 1.7678592319313313e-05,
      "loss": 1.0614,
      "step": 1962
    },
    {
      "epoch": 0.2448546837969315,
      "grad_norm": 0.0,
      "learning_rate": 1.7676003520504117e-05,
      "loss": 1.0339,
      "step": 1963
    },
    {
      "epoch": 0.24497941873518772,
      "grad_norm": 0.0,
      "learning_rate": 1.767341346877764e-05,
      "loss": 1.0492,
      "step": 1964
    },
    {
      "epoch": 0.24510415367344393,
      "grad_norm": 0.0,
      "learning_rate": 1.7670822164556642e-05,
      "loss": 1.0699,
      "step": 1965
    },
    {
      "epoch": 0.24522888861170014,
      "grad_norm": 0.0,
      "learning_rate": 1.7668229608264077e-05,
      "loss": 1.0203,
      "step": 1966
    },
    {
      "epoch": 0.24535362354995635,
      "grad_norm": 0.0,
      "learning_rate": 1.7665635800323134e-05,
      "loss": 1.0448,
      "step": 1967
    },
    {
      "epoch": 0.24547835848821256,
      "grad_norm": 0.0,
      "learning_rate": 1.7663040741157172e-05,
      "loss": 1.0418,
      "step": 1968
    },
    {
      "epoch": 0.24560309342646874,
      "grad_norm": 0.0,
      "learning_rate": 1.766044443118978e-05,
      "loss": 1.0342,
      "step": 1969
    },
    {
      "epoch": 0.24572782836472495,
      "grad_norm": 0.0,
      "learning_rate": 1.765784687084474e-05,
      "loss": 1.0018,
      "step": 1970
    },
    {
      "epoch": 0.24585256330298116,
      "grad_norm": 0.0,
      "learning_rate": 1.765524806054604e-05,
      "loss": 1.0264,
      "step": 1971
    },
    {
      "epoch": 0.24597729824123737,
      "grad_norm": 0.0,
      "learning_rate": 1.765264800071786e-05,
      "loss": 1.0605,
      "step": 1972
    },
    {
      "epoch": 0.24610203317949358,
      "grad_norm": 0.0,
      "learning_rate": 1.765004669178461e-05,
      "loss": 1.0491,
      "step": 1973
    },
    {
      "epoch": 0.2462267681177498,
      "grad_norm": 0.0,
      "learning_rate": 1.7647444134170883e-05,
      "loss": 1.0637,
      "step": 1974
    },
    {
      "epoch": 0.24635150305600598,
      "grad_norm": 0.0,
      "learning_rate": 1.764484032830148e-05,
      "loss": 1.0332,
      "step": 1975
    },
    {
      "epoch": 0.2464762379942622,
      "grad_norm": 0.0,
      "learning_rate": 1.7642235274601408e-05,
      "loss": 1.0574,
      "step": 1976
    },
    {
      "epoch": 0.2466009729325184,
      "grad_norm": 0.0,
      "learning_rate": 1.7639628973495884e-05,
      "loss": 0.9905,
      "step": 1977
    },
    {
      "epoch": 0.2467257078707746,
      "grad_norm": 0.0,
      "learning_rate": 1.763702142541032e-05,
      "loss": 1.0384,
      "step": 1978
    },
    {
      "epoch": 0.24685044280903082,
      "grad_norm": 0.0,
      "learning_rate": 1.7634412630770326e-05,
      "loss": 1.0585,
      "step": 1979
    },
    {
      "epoch": 0.24697517774728703,
      "grad_norm": 0.0,
      "learning_rate": 1.7631802590001726e-05,
      "loss": 1.0838,
      "step": 1980
    },
    {
      "epoch": 0.2470999126855432,
      "grad_norm": 0.0,
      "learning_rate": 1.7629191303530553e-05,
      "loss": 1.0504,
      "step": 1981
    },
    {
      "epoch": 0.24722464762379942,
      "grad_norm": 0.0,
      "learning_rate": 1.7626578771783026e-05,
      "loss": 1.0369,
      "step": 1982
    },
    {
      "epoch": 0.24734938256205563,
      "grad_norm": 0.0,
      "learning_rate": 1.762396499518558e-05,
      "loss": 1.0474,
      "step": 1983
    },
    {
      "epoch": 0.24747411750031184,
      "grad_norm": 0.0,
      "learning_rate": 1.762134997416484e-05,
      "loss": 1.0356,
      "step": 1984
    },
    {
      "epoch": 0.24759885243856805,
      "grad_norm": 0.0,
      "learning_rate": 1.761873370914766e-05,
      "loss": 1.105,
      "step": 1985
    },
    {
      "epoch": 0.24772358737682426,
      "grad_norm": 0.0,
      "learning_rate": 1.761611620056107e-05,
      "loss": 1.0509,
      "step": 1986
    },
    {
      "epoch": 0.24784832231508044,
      "grad_norm": 0.0,
      "learning_rate": 1.761349744883231e-05,
      "loss": 1.0278,
      "step": 1987
    },
    {
      "epoch": 0.24797305725333665,
      "grad_norm": 0.0,
      "learning_rate": 1.761087745438884e-05,
      "loss": 1.077,
      "step": 1988
    },
    {
      "epoch": 0.24809779219159286,
      "grad_norm": 0.0,
      "learning_rate": 1.7608256217658294e-05,
      "loss": 1.0538,
      "step": 1989
    },
    {
      "epoch": 0.24822252712984907,
      "grad_norm": 0.0,
      "learning_rate": 1.7605633739068532e-05,
      "loss": 1.0074,
      "step": 1990
    },
    {
      "epoch": 0.24834726206810528,
      "grad_norm": 0.0,
      "learning_rate": 1.7603010019047607e-05,
      "loss": 1.0436,
      "step": 1991
    },
    {
      "epoch": 0.2484719970063615,
      "grad_norm": 0.0,
      "learning_rate": 1.7600385058023774e-05,
      "loss": 1.0406,
      "step": 1992
    },
    {
      "epoch": 0.24859673194461768,
      "grad_norm": 0.0,
      "learning_rate": 1.7597758856425497e-05,
      "loss": 1.0258,
      "step": 1993
    },
    {
      "epoch": 0.2487214668828739,
      "grad_norm": 0.0,
      "learning_rate": 1.7595131414681434e-05,
      "loss": 1.0504,
      "step": 1994
    },
    {
      "epoch": 0.2488462018211301,
      "grad_norm": 0.0,
      "learning_rate": 1.7592502733220454e-05,
      "loss": 1.0988,
      "step": 1995
    },
    {
      "epoch": 0.2489709367593863,
      "grad_norm": 0.0,
      "learning_rate": 1.758987281247162e-05,
      "loss": 0.9772,
      "step": 1996
    },
    {
      "epoch": 0.24909567169764252,
      "grad_norm": 0.0,
      "learning_rate": 1.7587241652864206e-05,
      "loss": 1.0398,
      "step": 1997
    },
    {
      "epoch": 0.24922040663589873,
      "grad_norm": 0.0,
      "learning_rate": 1.758460925482768e-05,
      "loss": 1.0791,
      "step": 1998
    },
    {
      "epoch": 0.2493451415741549,
      "grad_norm": 0.0,
      "learning_rate": 1.7581975618791714e-05,
      "loss": 1.0831,
      "step": 1999
    },
    {
      "epoch": 0.24946987651241112,
      "grad_norm": 0.0,
      "learning_rate": 1.757934074518619e-05,
      "loss": 1.0541,
      "step": 2000
    },
    {
      "epoch": 0.24959461145066733,
      "grad_norm": 0.0,
      "learning_rate": 1.757670463444118e-05,
      "loss": 1.0402,
      "step": 2001
    },
    {
      "epoch": 0.24971934638892354,
      "grad_norm": 0.0,
      "learning_rate": 1.7574067286986968e-05,
      "loss": 1.0178,
      "step": 2002
    },
    {
      "epoch": 0.24984408132717975,
      "grad_norm": 0.0,
      "learning_rate": 1.7571428703254033e-05,
      "loss": 1.0368,
      "step": 2003
    },
    {
      "epoch": 0.24996881626543596,
      "grad_norm": 0.0,
      "learning_rate": 1.756878888367306e-05,
      "loss": 0.9941,
      "step": 2004
    },
    {
      "epoch": 0.25009355120369214,
      "grad_norm": 0.0,
      "learning_rate": 1.756614782867493e-05,
      "loss": 1.0632,
      "step": 2005
    },
    {
      "epoch": 0.25021828614194835,
      "grad_norm": 0.0,
      "learning_rate": 1.7563505538690734e-05,
      "loss": 1.05,
      "step": 2006
    },
    {
      "epoch": 0.25034302108020456,
      "grad_norm": 0.0,
      "learning_rate": 1.756086201415176e-05,
      "loss": 1.0007,
      "step": 2007
    },
    {
      "epoch": 0.2504677560184608,
      "grad_norm": 0.0,
      "learning_rate": 1.7558217255489498e-05,
      "loss": 1.1402,
      "step": 2008
    },
    {
      "epoch": 0.250592490956717,
      "grad_norm": 0.0,
      "learning_rate": 1.755557126313564e-05,
      "loss": 1.0883,
      "step": 2009
    },
    {
      "epoch": 0.2507172258949732,
      "grad_norm": 0.0,
      "learning_rate": 1.7552924037522074e-05,
      "loss": 1.0716,
      "step": 2010
    },
    {
      "epoch": 0.2508419608332294,
      "grad_norm": 0.0,
      "learning_rate": 1.7550275579080893e-05,
      "loss": 1.0205,
      "step": 2011
    },
    {
      "epoch": 0.2509666957714856,
      "grad_norm": 0.0,
      "learning_rate": 1.7547625888244403e-05,
      "loss": 1.0099,
      "step": 2012
    },
    {
      "epoch": 0.2510914307097418,
      "grad_norm": 0.0,
      "learning_rate": 1.7544974965445088e-05,
      "loss": 1.0875,
      "step": 2013
    },
    {
      "epoch": 0.251216165647998,
      "grad_norm": 0.0,
      "learning_rate": 1.7542322811115657e-05,
      "loss": 1.0384,
      "step": 2014
    },
    {
      "epoch": 0.2513409005862542,
      "grad_norm": 0.0,
      "learning_rate": 1.7539669425688997e-05,
      "loss": 1.0665,
      "step": 2015
    },
    {
      "epoch": 0.2514656355245104,
      "grad_norm": 0.0,
      "learning_rate": 1.7537014809598214e-05,
      "loss": 0.9946,
      "step": 2016
    },
    {
      "epoch": 0.2515903704627666,
      "grad_norm": 0.0,
      "learning_rate": 1.7534358963276606e-05,
      "loss": 1.0807,
      "step": 2017
    },
    {
      "epoch": 0.2517151054010228,
      "grad_norm": 0.0,
      "learning_rate": 1.753170188715768e-05,
      "loss": 1.0503,
      "step": 2018
    },
    {
      "epoch": 0.25183984033927903,
      "grad_norm": 0.0,
      "learning_rate": 1.752904358167513e-05,
      "loss": 1.0395,
      "step": 2019
    },
    {
      "epoch": 0.25196457527753524,
      "grad_norm": 0.0,
      "learning_rate": 1.752638404726286e-05,
      "loss": 1.0218,
      "step": 2020
    },
    {
      "epoch": 0.25208931021579145,
      "grad_norm": 0.0,
      "learning_rate": 1.7523723284354977e-05,
      "loss": 1.0505,
      "step": 2021
    },
    {
      "epoch": 0.25221404515404766,
      "grad_norm": 0.0,
      "learning_rate": 1.752106129338578e-05,
      "loss": 1.1192,
      "step": 2022
    },
    {
      "epoch": 0.25233878009230387,
      "grad_norm": 0.0,
      "learning_rate": 1.7518398074789776e-05,
      "loss": 1.0198,
      "step": 2023
    },
    {
      "epoch": 0.2524635150305601,
      "grad_norm": 0.0,
      "learning_rate": 1.751573362900167e-05,
      "loss": 1.0149,
      "step": 2024
    },
    {
      "epoch": 0.2525882499688163,
      "grad_norm": 0.0,
      "learning_rate": 1.751306795645636e-05,
      "loss": 1.0477,
      "step": 2025
    },
    {
      "epoch": 0.25271298490707245,
      "grad_norm": 0.0,
      "learning_rate": 1.751040105758896e-05,
      "loss": 1.0708,
      "step": 2026
    },
    {
      "epoch": 0.25283771984532866,
      "grad_norm": 0.0,
      "learning_rate": 1.750773293283477e-05,
      "loss": 1.0599,
      "step": 2027
    },
    {
      "epoch": 0.25296245478358487,
      "grad_norm": 0.0,
      "learning_rate": 1.7505063582629297e-05,
      "loss": 1.0773,
      "step": 2028
    },
    {
      "epoch": 0.2530871897218411,
      "grad_norm": 0.0,
      "learning_rate": 1.7502393007408246e-05,
      "loss": 1.0446,
      "step": 2029
    },
    {
      "epoch": 0.2532119246600973,
      "grad_norm": 0.0,
      "learning_rate": 1.7499721207607523e-05,
      "loss": 1.0508,
      "step": 2030
    },
    {
      "epoch": 0.2533366595983535,
      "grad_norm": 0.0,
      "learning_rate": 1.749704818366323e-05,
      "loss": 1.054,
      "step": 2031
    },
    {
      "epoch": 0.2534613945366097,
      "grad_norm": 0.0,
      "learning_rate": 1.7494373936011674e-05,
      "loss": 1.0502,
      "step": 2032
    },
    {
      "epoch": 0.2535861294748659,
      "grad_norm": 0.0,
      "learning_rate": 1.749169846508936e-05,
      "loss": 1.0714,
      "step": 2033
    },
    {
      "epoch": 0.25371086441312213,
      "grad_norm": 0.0,
      "learning_rate": 1.7489021771332995e-05,
      "loss": 1.0472,
      "step": 2034
    },
    {
      "epoch": 0.25383559935137834,
      "grad_norm": 0.0,
      "learning_rate": 1.7486343855179476e-05,
      "loss": 1.0397,
      "step": 2035
    },
    {
      "epoch": 0.25396033428963455,
      "grad_norm": 0.0,
      "learning_rate": 1.748366471706591e-05,
      "loss": 1.0257,
      "step": 2036
    },
    {
      "epoch": 0.25408506922789076,
      "grad_norm": 0.0,
      "learning_rate": 1.74809843574296e-05,
      "loss": 1.0556,
      "step": 2037
    },
    {
      "epoch": 0.2542098041661469,
      "grad_norm": 0.0,
      "learning_rate": 1.747830277670805e-05,
      "loss": 1.0969,
      "step": 2038
    },
    {
      "epoch": 0.2543345391044031,
      "grad_norm": 0.0,
      "learning_rate": 1.747561997533896e-05,
      "loss": 1.0175,
      "step": 2039
    },
    {
      "epoch": 0.25445927404265933,
      "grad_norm": 0.0,
      "learning_rate": 1.747293595376023e-05,
      "loss": 1.0243,
      "step": 2040
    },
    {
      "epoch": 0.25458400898091554,
      "grad_norm": 0.0,
      "learning_rate": 1.7470250712409963e-05,
      "loss": 1.0682,
      "step": 2041
    },
    {
      "epoch": 0.25470874391917175,
      "grad_norm": 0.0,
      "learning_rate": 1.7467564251726452e-05,
      "loss": 1.0456,
      "step": 2042
    },
    {
      "epoch": 0.25483347885742796,
      "grad_norm": 0.0,
      "learning_rate": 1.74648765721482e-05,
      "loss": 1.0018,
      "step": 2043
    },
    {
      "epoch": 0.2549582137956842,
      "grad_norm": 0.0,
      "learning_rate": 1.7462187674113903e-05,
      "loss": 1.021,
      "step": 2044
    },
    {
      "epoch": 0.2550829487339404,
      "grad_norm": 0.0,
      "learning_rate": 1.7459497558062456e-05,
      "loss": 1.0639,
      "step": 2045
    },
    {
      "epoch": 0.2552076836721966,
      "grad_norm": 0.0,
      "learning_rate": 1.745680622443296e-05,
      "loss": 1.095,
      "step": 2046
    },
    {
      "epoch": 0.2553324186104528,
      "grad_norm": 0.0,
      "learning_rate": 1.7454113673664695e-05,
      "loss": 1.0722,
      "step": 2047
    },
    {
      "epoch": 0.255457153548709,
      "grad_norm": 0.0,
      "learning_rate": 1.7451419906197168e-05,
      "loss": 1.018,
      "step": 2048
    },
    {
      "epoch": 0.2555818884869652,
      "grad_norm": 0.0,
      "learning_rate": 1.744872492247006e-05,
      "loss": 1.0902,
      "step": 2049
    },
    {
      "epoch": 0.2557066234252214,
      "grad_norm": 0.0,
      "learning_rate": 1.7446028722923266e-05,
      "loss": 1.0073,
      "step": 2050
    },
    {
      "epoch": 0.2558313583634776,
      "grad_norm": 0.0,
      "learning_rate": 1.7443331307996868e-05,
      "loss": 1.0358,
      "step": 2051
    },
    {
      "epoch": 0.2559560933017338,
      "grad_norm": 0.0,
      "learning_rate": 1.7440632678131157e-05,
      "loss": 1.0642,
      "step": 2052
    },
    {
      "epoch": 0.25608082823999,
      "grad_norm": 0.0,
      "learning_rate": 1.7437932833766614e-05,
      "loss": 0.9788,
      "step": 2053
    },
    {
      "epoch": 0.2562055631782462,
      "grad_norm": 0.0,
      "learning_rate": 1.7435231775343924e-05,
      "loss": 1.0323,
      "step": 2054
    },
    {
      "epoch": 0.25633029811650243,
      "grad_norm": 0.0,
      "learning_rate": 1.7432529503303967e-05,
      "loss": 1.0799,
      "step": 2055
    },
    {
      "epoch": 0.25645503305475864,
      "grad_norm": 0.0,
      "learning_rate": 1.742982601808782e-05,
      "loss": 1.0148,
      "step": 2056
    },
    {
      "epoch": 0.25657976799301485,
      "grad_norm": 0.0,
      "learning_rate": 1.7427121320136765e-05,
      "loss": 1.08,
      "step": 2057
    },
    {
      "epoch": 0.25670450293127106,
      "grad_norm": 0.0,
      "learning_rate": 1.7424415409892274e-05,
      "loss": 1.0112,
      "step": 2058
    },
    {
      "epoch": 0.25682923786952727,
      "grad_norm": 0.0,
      "learning_rate": 1.7421708287796017e-05,
      "loss": 1.045,
      "step": 2059
    },
    {
      "epoch": 0.2569539728077835,
      "grad_norm": 0.0,
      "learning_rate": 1.741899995428987e-05,
      "loss": 1.0506,
      "step": 2060
    },
    {
      "epoch": 0.2570787077460397,
      "grad_norm": 0.0,
      "learning_rate": 1.74162904098159e-05,
      "loss": 0.986,
      "step": 2061
    },
    {
      "epoch": 0.25720344268429585,
      "grad_norm": 0.0,
      "learning_rate": 1.741357965481637e-05,
      "loss": 1.0139,
      "step": 2062
    },
    {
      "epoch": 0.25732817762255206,
      "grad_norm": 0.0,
      "learning_rate": 1.741086768973374e-05,
      "loss": 1.0351,
      "step": 2063
    },
    {
      "epoch": 0.25745291256080827,
      "grad_norm": 0.0,
      "learning_rate": 1.7408154515010684e-05,
      "loss": 0.9952,
      "step": 2064
    },
    {
      "epoch": 0.2575776474990645,
      "grad_norm": 0.0,
      "learning_rate": 1.740544013109005e-05,
      "loss": 0.9996,
      "step": 2065
    },
    {
      "epoch": 0.2577023824373207,
      "grad_norm": 0.0,
      "learning_rate": 1.7402724538414895e-05,
      "loss": 1.0306,
      "step": 2066
    },
    {
      "epoch": 0.2578271173755769,
      "grad_norm": 0.0,
      "learning_rate": 1.740000773742847e-05,
      "loss": 1.0632,
      "step": 2067
    },
    {
      "epoch": 0.2579518523138331,
      "grad_norm": 0.0,
      "learning_rate": 1.7397289728574234e-05,
      "loss": 1.0592,
      "step": 2068
    },
    {
      "epoch": 0.2580765872520893,
      "grad_norm": 0.0,
      "learning_rate": 1.7394570512295832e-05,
      "loss": 1.0702,
      "step": 2069
    },
    {
      "epoch": 0.25820132219034553,
      "grad_norm": 0.0,
      "learning_rate": 1.73918500890371e-05,
      "loss": 0.9931,
      "step": 2070
    },
    {
      "epoch": 0.25832605712860174,
      "grad_norm": 0.0,
      "learning_rate": 1.738912845924209e-05,
      "loss": 1.0194,
      "step": 2071
    },
    {
      "epoch": 0.25845079206685795,
      "grad_norm": 0.0,
      "learning_rate": 1.738640562335504e-05,
      "loss": 1.0778,
      "step": 2072
    },
    {
      "epoch": 0.25857552700511416,
      "grad_norm": 0.0,
      "learning_rate": 1.7383681581820374e-05,
      "loss": 1.0613,
      "step": 2073
    },
    {
      "epoch": 0.2587002619433703,
      "grad_norm": 0.0,
      "learning_rate": 1.7380956335082735e-05,
      "loss": 1.0577,
      "step": 2074
    },
    {
      "epoch": 0.2588249968816265,
      "grad_norm": 0.0,
      "learning_rate": 1.737822988358695e-05,
      "loss": 0.9935,
      "step": 2075
    },
    {
      "epoch": 0.25894973181988273,
      "grad_norm": 0.0,
      "learning_rate": 1.7375502227778044e-05,
      "loss": 0.9846,
      "step": 2076
    },
    {
      "epoch": 0.25907446675813894,
      "grad_norm": 0.0,
      "learning_rate": 1.737277336810124e-05,
      "loss": 1.0452,
      "step": 2077
    },
    {
      "epoch": 0.25919920169639515,
      "grad_norm": 0.0,
      "learning_rate": 1.7370043305001957e-05,
      "loss": 1.073,
      "step": 2078
    },
    {
      "epoch": 0.25932393663465136,
      "grad_norm": 0.0,
      "learning_rate": 1.7367312038925804e-05,
      "loss": 1.0515,
      "step": 2079
    },
    {
      "epoch": 0.2594486715729076,
      "grad_norm": 0.0,
      "learning_rate": 1.7364579570318602e-05,
      "loss": 1.0362,
      "step": 2080
    },
    {
      "epoch": 0.2595734065111638,
      "grad_norm": 0.0,
      "learning_rate": 1.7361845899626355e-05,
      "loss": 0.9715,
      "step": 2081
    },
    {
      "epoch": 0.25969814144942,
      "grad_norm": 0.0,
      "learning_rate": 1.7359111027295262e-05,
      "loss": 1.0001,
      "step": 2082
    },
    {
      "epoch": 0.2598228763876762,
      "grad_norm": 0.0,
      "learning_rate": 1.7356374953771727e-05,
      "loss": 0.9842,
      "step": 2083
    },
    {
      "epoch": 0.2599476113259324,
      "grad_norm": 0.0,
      "learning_rate": 1.735363767950235e-05,
      "loss": 1.0044,
      "step": 2084
    },
    {
      "epoch": 0.2600723462641886,
      "grad_norm": 0.0,
      "learning_rate": 1.7350899204933917e-05,
      "loss": 1.041,
      "step": 2085
    },
    {
      "epoch": 0.2601970812024448,
      "grad_norm": 0.0,
      "learning_rate": 1.7348159530513424e-05,
      "loss": 1.0394,
      "step": 2086
    },
    {
      "epoch": 0.260321816140701,
      "grad_norm": 0.0,
      "learning_rate": 1.7345418656688043e-05,
      "loss": 0.9928,
      "step": 2087
    },
    {
      "epoch": 0.2604465510789572,
      "grad_norm": 0.0,
      "learning_rate": 1.734267658390516e-05,
      "loss": 1.0208,
      "step": 2088
    },
    {
      "epoch": 0.2605712860172134,
      "grad_norm": 0.0,
      "learning_rate": 1.7339933312612353e-05,
      "loss": 1.0699,
      "step": 2089
    },
    {
      "epoch": 0.2606960209554696,
      "grad_norm": 0.0,
      "learning_rate": 1.733718884325739e-05,
      "loss": 1.0412,
      "step": 2090
    },
    {
      "epoch": 0.26082075589372583,
      "grad_norm": 0.0,
      "learning_rate": 1.7334443176288243e-05,
      "loss": 1.0697,
      "step": 2091
    },
    {
      "epoch": 0.26094549083198204,
      "grad_norm": 0.0,
      "learning_rate": 1.7331696312153062e-05,
      "loss": 1.078,
      "step": 2092
    },
    {
      "epoch": 0.26107022577023825,
      "grad_norm": 0.0,
      "learning_rate": 1.7328948251300216e-05,
      "loss": 1.0429,
      "step": 2093
    },
    {
      "epoch": 0.26119496070849446,
      "grad_norm": 0.0,
      "learning_rate": 1.7326198994178255e-05,
      "loss": 1.0895,
      "step": 2094
    },
    {
      "epoch": 0.26131969564675067,
      "grad_norm": 0.0,
      "learning_rate": 1.7323448541235922e-05,
      "loss": 1.0625,
      "step": 2095
    },
    {
      "epoch": 0.2614444305850069,
      "grad_norm": 0.0,
      "learning_rate": 1.7320696892922164e-05,
      "loss": 1.1006,
      "step": 2096
    },
    {
      "epoch": 0.2615691655232631,
      "grad_norm": 0.0,
      "learning_rate": 1.7317944049686125e-05,
      "loss": 1.0483,
      "step": 2097
    },
    {
      "epoch": 0.26169390046151925,
      "grad_norm": 0.0,
      "learning_rate": 1.7315190011977126e-05,
      "loss": 1.0591,
      "step": 2098
    },
    {
      "epoch": 0.26181863539977546,
      "grad_norm": 0.0,
      "learning_rate": 1.7312434780244704e-05,
      "loss": 1.0325,
      "step": 2099
    },
    {
      "epoch": 0.26194337033803167,
      "grad_norm": 0.0,
      "learning_rate": 1.7309678354938583e-05,
      "loss": 1.021,
      "step": 2100
    },
    {
      "epoch": 0.2620681052762879,
      "grad_norm": 0.0,
      "learning_rate": 1.7306920736508676e-05,
      "loss": 0.9987,
      "step": 2101
    },
    {
      "epoch": 0.2621928402145441,
      "grad_norm": 0.0,
      "learning_rate": 1.7304161925405097e-05,
      "loss": 1.0558,
      "step": 2102
    },
    {
      "epoch": 0.2623175751528003,
      "grad_norm": 0.0,
      "learning_rate": 1.7301401922078162e-05,
      "loss": 1.081,
      "step": 2103
    },
    {
      "epoch": 0.2624423100910565,
      "grad_norm": 0.0,
      "learning_rate": 1.7298640726978357e-05,
      "loss": 1.0436,
      "step": 2104
    },
    {
      "epoch": 0.2625670450293127,
      "grad_norm": 0.0,
      "learning_rate": 1.7295878340556394e-05,
      "loss": 1.0958,
      "step": 2105
    },
    {
      "epoch": 0.26269177996756893,
      "grad_norm": 0.0,
      "learning_rate": 1.7293114763263154e-05,
      "loss": 1.0205,
      "step": 2106
    },
    {
      "epoch": 0.26281651490582514,
      "grad_norm": 0.0,
      "learning_rate": 1.7290349995549726e-05,
      "loss": 0.9927,
      "step": 2107
    },
    {
      "epoch": 0.26294124984408135,
      "grad_norm": 0.0,
      "learning_rate": 1.7287584037867394e-05,
      "loss": 1.0033,
      "step": 2108
    },
    {
      "epoch": 0.26306598478233756,
      "grad_norm": 0.0,
      "learning_rate": 1.7284816890667627e-05,
      "loss": 1.0375,
      "step": 2109
    },
    {
      "epoch": 0.2631907197205937,
      "grad_norm": 0.0,
      "learning_rate": 1.7282048554402095e-05,
      "loss": 1.0207,
      "step": 2110
    },
    {
      "epoch": 0.2633154546588499,
      "grad_norm": 0.0,
      "learning_rate": 1.7279279029522654e-05,
      "loss": 1.0704,
      "step": 2111
    },
    {
      "epoch": 0.26344018959710613,
      "grad_norm": 0.0,
      "learning_rate": 1.7276508316481367e-05,
      "loss": 0.9905,
      "step": 2112
    },
    {
      "epoch": 0.26356492453536234,
      "grad_norm": 0.0,
      "learning_rate": 1.7273736415730488e-05,
      "loss": 1.029,
      "step": 2113
    },
    {
      "epoch": 0.26368965947361855,
      "grad_norm": 0.0,
      "learning_rate": 1.7270963327722455e-05,
      "loss": 1.0445,
      "step": 2114
    },
    {
      "epoch": 0.26381439441187476,
      "grad_norm": 0.0,
      "learning_rate": 1.7268189052909905e-05,
      "loss": 1.0506,
      "step": 2115
    },
    {
      "epoch": 0.263939129350131,
      "grad_norm": 0.0,
      "learning_rate": 1.726541359174567e-05,
      "loss": 1.0249,
      "step": 2116
    },
    {
      "epoch": 0.2640638642883872,
      "grad_norm": 0.0,
      "learning_rate": 1.726263694468278e-05,
      "loss": 1.043,
      "step": 2117
    },
    {
      "epoch": 0.2641885992266434,
      "grad_norm": 0.0,
      "learning_rate": 1.7259859112174452e-05,
      "loss": 1.0287,
      "step": 2118
    },
    {
      "epoch": 0.2643133341648996,
      "grad_norm": 0.0,
      "learning_rate": 1.7257080094674092e-05,
      "loss": 1.0448,
      "step": 2119
    },
    {
      "epoch": 0.2644380691031558,
      "grad_norm": 0.0,
      "learning_rate": 1.7254299892635313e-05,
      "loss": 1.0267,
      "step": 2120
    },
    {
      "epoch": 0.264562804041412,
      "grad_norm": 0.0,
      "learning_rate": 1.7251518506511914e-05,
      "loss": 1.0485,
      "step": 2121
    },
    {
      "epoch": 0.2646875389796682,
      "grad_norm": 0.0,
      "learning_rate": 1.7248735936757882e-05,
      "loss": 1.0694,
      "step": 2122
    },
    {
      "epoch": 0.2648122739179244,
      "grad_norm": 0.0,
      "learning_rate": 1.7245952183827403e-05,
      "loss": 1.0693,
      "step": 2123
    },
    {
      "epoch": 0.2649370088561806,
      "grad_norm": 0.0,
      "learning_rate": 1.724316724817486e-05,
      "loss": 1.0311,
      "step": 2124
    },
    {
      "epoch": 0.2650617437944368,
      "grad_norm": 0.0,
      "learning_rate": 1.7240381130254826e-05,
      "loss": 1.0325,
      "step": 2125
    },
    {
      "epoch": 0.265186478732693,
      "grad_norm": 0.0,
      "learning_rate": 1.7237593830522063e-05,
      "loss": 0.9892,
      "step": 2126
    },
    {
      "epoch": 0.26531121367094923,
      "grad_norm": 0.0,
      "learning_rate": 1.7234805349431526e-05,
      "loss": 1.0152,
      "step": 2127
    },
    {
      "epoch": 0.26543594860920544,
      "grad_norm": 0.0,
      "learning_rate": 1.7232015687438365e-05,
      "loss": 1.0378,
      "step": 2128
    },
    {
      "epoch": 0.26556068354746165,
      "grad_norm": 0.0,
      "learning_rate": 1.722922484499793e-05,
      "loss": 1.0679,
      "step": 2129
    },
    {
      "epoch": 0.26568541848571786,
      "grad_norm": 0.0,
      "learning_rate": 1.722643282256575e-05,
      "loss": 1.0302,
      "step": 2130
    },
    {
      "epoch": 0.26581015342397407,
      "grad_norm": 0.0,
      "learning_rate": 1.7223639620597556e-05,
      "loss": 1.0699,
      "step": 2131
    },
    {
      "epoch": 0.2659348883622303,
      "grad_norm": 0.0,
      "learning_rate": 1.7220845239549272e-05,
      "loss": 1.0538,
      "step": 2132
    },
    {
      "epoch": 0.2660596233004865,
      "grad_norm": 0.0,
      "learning_rate": 1.7218049679877003e-05,
      "loss": 1.0489,
      "step": 2133
    },
    {
      "epoch": 0.26618435823874265,
      "grad_norm": 0.0,
      "learning_rate": 1.7215252942037067e-05,
      "loss": 1.0483,
      "step": 2134
    },
    {
      "epoch": 0.26630909317699886,
      "grad_norm": 0.0,
      "learning_rate": 1.721245502648595e-05,
      "loss": 0.9804,
      "step": 2135
    },
    {
      "epoch": 0.26643382811525507,
      "grad_norm": 0.0,
      "learning_rate": 1.7209655933680348e-05,
      "loss": 1.0565,
      "step": 2136
    },
    {
      "epoch": 0.2665585630535113,
      "grad_norm": 0.0,
      "learning_rate": 1.7206855664077147e-05,
      "loss": 1.0517,
      "step": 2137
    },
    {
      "epoch": 0.2666832979917675,
      "grad_norm": 0.0,
      "learning_rate": 1.7204054218133418e-05,
      "loss": 1.1176,
      "step": 2138
    },
    {
      "epoch": 0.2668080329300237,
      "grad_norm": 0.0,
      "learning_rate": 1.7201251596306424e-05,
      "loss": 1.0041,
      "step": 2139
    },
    {
      "epoch": 0.2669327678682799,
      "grad_norm": 0.0,
      "learning_rate": 1.719844779905363e-05,
      "loss": 1.0355,
      "step": 2140
    },
    {
      "epoch": 0.2670575028065361,
      "grad_norm": 0.0,
      "learning_rate": 1.719564282683268e-05,
      "loss": 1.0079,
      "step": 2141
    },
    {
      "epoch": 0.26718223774479233,
      "grad_norm": 0.0,
      "learning_rate": 1.719283668010142e-05,
      "loss": 1.0486,
      "step": 2142
    },
    {
      "epoch": 0.26730697268304854,
      "grad_norm": 0.0,
      "learning_rate": 1.7190029359317884e-05,
      "loss": 1.0339,
      "step": 2143
    },
    {
      "epoch": 0.26743170762130475,
      "grad_norm": 0.0,
      "learning_rate": 1.7187220864940296e-05,
      "loss": 1.0123,
      "step": 2144
    },
    {
      "epoch": 0.26755644255956096,
      "grad_norm": 0.0,
      "learning_rate": 1.7184411197427077e-05,
      "loss": 1.0841,
      "step": 2145
    },
    {
      "epoch": 0.2676811774978171,
      "grad_norm": 0.0,
      "learning_rate": 1.7181600357236828e-05,
      "loss": 1.0549,
      "step": 2146
    },
    {
      "epoch": 0.2678059124360733,
      "grad_norm": 0.0,
      "learning_rate": 1.717878834482836e-05,
      "loss": 1.081,
      "step": 2147
    },
    {
      "epoch": 0.26793064737432953,
      "grad_norm": 0.0,
      "learning_rate": 1.717597516066065e-05,
      "loss": 1.0499,
      "step": 2148
    },
    {
      "epoch": 0.26805538231258574,
      "grad_norm": 0.0,
      "learning_rate": 1.7173160805192895e-05,
      "loss": 1.032,
      "step": 2149
    },
    {
      "epoch": 0.26818011725084195,
      "grad_norm": 0.0,
      "learning_rate": 1.7170345278884463e-05,
      "loss": 1.0324,
      "step": 2150
    },
    {
      "epoch": 0.26830485218909816,
      "grad_norm": 0.0,
      "learning_rate": 1.716752858219491e-05,
      "loss": 1.0209,
      "step": 2151
    },
    {
      "epoch": 0.2684295871273544,
      "grad_norm": 0.0,
      "learning_rate": 1.7164710715584007e-05,
      "loss": 0.9964,
      "step": 2152
    },
    {
      "epoch": 0.2685543220656106,
      "grad_norm": 0.0,
      "learning_rate": 1.7161891679511695e-05,
      "loss": 1.0085,
      "step": 2153
    },
    {
      "epoch": 0.2686790570038668,
      "grad_norm": 0.0,
      "learning_rate": 1.7159071474438106e-05,
      "loss": 0.9963,
      "step": 2154
    },
    {
      "epoch": 0.268803791942123,
      "grad_norm": 0.0,
      "learning_rate": 1.7156250100823575e-05,
      "loss": 1.0962,
      "step": 2155
    },
    {
      "epoch": 0.2689285268803792,
      "grad_norm": 0.0,
      "learning_rate": 1.7153427559128618e-05,
      "loss": 1.0332,
      "step": 2156
    },
    {
      "epoch": 0.2690532618186354,
      "grad_norm": 0.0,
      "learning_rate": 1.715060384981395e-05,
      "loss": 1.049,
      "step": 2157
    },
    {
      "epoch": 0.2691779967568916,
      "grad_norm": 0.0,
      "learning_rate": 1.7147778973340466e-05,
      "loss": 1.0364,
      "step": 2158
    },
    {
      "epoch": 0.2693027316951478,
      "grad_norm": 0.0,
      "learning_rate": 1.714495293016926e-05,
      "loss": 0.9749,
      "step": 2159
    },
    {
      "epoch": 0.269427466633404,
      "grad_norm": 0.0,
      "learning_rate": 1.714212572076161e-05,
      "loss": 1.0424,
      "step": 2160
    },
    {
      "epoch": 0.2695522015716602,
      "grad_norm": 0.0,
      "learning_rate": 1.7139297345578992e-05,
      "loss": 1.0395,
      "step": 2161
    },
    {
      "epoch": 0.2696769365099164,
      "grad_norm": 0.0,
      "learning_rate": 1.713646780508307e-05,
      "loss": 1.0376,
      "step": 2162
    },
    {
      "epoch": 0.26980167144817263,
      "grad_norm": 0.0,
      "learning_rate": 1.7133637099735686e-05,
      "loss": 1.0269,
      "step": 2163
    },
    {
      "epoch": 0.26992640638642884,
      "grad_norm": 0.0,
      "learning_rate": 1.7130805229998894e-05,
      "loss": 1.0745,
      "step": 2164
    },
    {
      "epoch": 0.27005114132468505,
      "grad_norm": 0.0,
      "learning_rate": 1.712797219633492e-05,
      "loss": 1.0544,
      "step": 2165
    },
    {
      "epoch": 0.27017587626294126,
      "grad_norm": 0.0,
      "learning_rate": 1.7125137999206188e-05,
      "loss": 1.0075,
      "step": 2166
    },
    {
      "epoch": 0.27030061120119747,
      "grad_norm": 0.0,
      "learning_rate": 1.7122302639075313e-05,
      "loss": 1.0248,
      "step": 2167
    },
    {
      "epoch": 0.2704253461394537,
      "grad_norm": 0.0,
      "learning_rate": 1.7119466116405093e-05,
      "loss": 1.0545,
      "step": 2168
    },
    {
      "epoch": 0.2705500810777099,
      "grad_norm": 0.0,
      "learning_rate": 1.7116628431658525e-05,
      "loss": 1.0202,
      "step": 2169
    },
    {
      "epoch": 0.27067481601596605,
      "grad_norm": 0.0,
      "learning_rate": 1.7113789585298786e-05,
      "loss": 1.0505,
      "step": 2170
    },
    {
      "epoch": 0.27079955095422226,
      "grad_norm": 0.0,
      "learning_rate": 1.7110949577789247e-05,
      "loss": 1.0464,
      "step": 2171
    },
    {
      "epoch": 0.27092428589247847,
      "grad_norm": 0.0,
      "learning_rate": 1.7108108409593473e-05,
      "loss": 1.0601,
      "step": 2172
    },
    {
      "epoch": 0.2710490208307347,
      "grad_norm": 0.0,
      "learning_rate": 1.710526608117521e-05,
      "loss": 1.0619,
      "step": 2173
    },
    {
      "epoch": 0.2711737557689909,
      "grad_norm": 0.0,
      "learning_rate": 1.7102422592998403e-05,
      "loss": 1.0149,
      "step": 2174
    },
    {
      "epoch": 0.2712984907072471,
      "grad_norm": 0.0,
      "learning_rate": 1.7099577945527177e-05,
      "loss": 1.0522,
      "step": 2175
    },
    {
      "epoch": 0.2714232256455033,
      "grad_norm": 0.0,
      "learning_rate": 1.7096732139225853e-05,
      "loss": 1.1013,
      "step": 2176
    },
    {
      "epoch": 0.2715479605837595,
      "grad_norm": 0.0,
      "learning_rate": 1.709388517455893e-05,
      "loss": 1.0446,
      "step": 2177
    },
    {
      "epoch": 0.27167269552201573,
      "grad_norm": 0.0,
      "learning_rate": 1.7091037051991113e-05,
      "loss": 1.0505,
      "step": 2178
    },
    {
      "epoch": 0.27179743046027194,
      "grad_norm": 0.0,
      "learning_rate": 1.7088187771987286e-05,
      "loss": 1.0534,
      "step": 2179
    },
    {
      "epoch": 0.27192216539852815,
      "grad_norm": 0.0,
      "learning_rate": 1.7085337335012524e-05,
      "loss": 1.0076,
      "step": 2180
    },
    {
      "epoch": 0.27204690033678436,
      "grad_norm": 0.0,
      "learning_rate": 1.708248574153209e-05,
      "loss": 1.0013,
      "step": 2181
    },
    {
      "epoch": 0.2721716352750405,
      "grad_norm": 0.0,
      "learning_rate": 1.707963299201143e-05,
      "loss": 1.0926,
      "step": 2182
    },
    {
      "epoch": 0.2722963702132967,
      "grad_norm": 0.0,
      "learning_rate": 1.707677908691619e-05,
      "loss": 1.0205,
      "step": 2183
    },
    {
      "epoch": 0.27242110515155293,
      "grad_norm": 0.0,
      "learning_rate": 1.7073924026712202e-05,
      "loss": 1.073,
      "step": 2184
    },
    {
      "epoch": 0.27254584008980914,
      "grad_norm": 0.0,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 1.0273,
      "step": 2185
    },
    {
      "epoch": 0.27267057502806535,
      "grad_norm": 0.0,
      "learning_rate": 1.7068210442842226e-05,
      "loss": 1.059,
      "step": 2186
    },
    {
      "epoch": 0.27279530996632156,
      "grad_norm": 0.0,
      "learning_rate": 1.7065351920108838e-05,
      "loss": 0.9956,
      "step": 2187
    },
    {
      "epoch": 0.2729200449045778,
      "grad_norm": 0.0,
      "learning_rate": 1.7062492244131907e-05,
      "loss": 1.0386,
      "step": 2188
    },
    {
      "epoch": 0.273044779842834,
      "grad_norm": 0.0,
      "learning_rate": 1.705963141537819e-05,
      "loss": 1.0091,
      "step": 2189
    },
    {
      "epoch": 0.2731695147810902,
      "grad_norm": 0.0,
      "learning_rate": 1.705676943431466e-05,
      "loss": 0.9931,
      "step": 2190
    },
    {
      "epoch": 0.2732942497193464,
      "grad_norm": 0.0,
      "learning_rate": 1.7053906301408457e-05,
      "loss": 1.0548,
      "step": 2191
    },
    {
      "epoch": 0.2734189846576026,
      "grad_norm": 0.0,
      "learning_rate": 1.7051042017126916e-05,
      "loss": 0.9886,
      "step": 2192
    },
    {
      "epoch": 0.2735437195958588,
      "grad_norm": 0.0,
      "learning_rate": 1.7048176581937562e-05,
      "loss": 0.9875,
      "step": 2193
    },
    {
      "epoch": 0.273668454534115,
      "grad_norm": 0.0,
      "learning_rate": 1.7045309996308107e-05,
      "loss": 1.0535,
      "step": 2194
    },
    {
      "epoch": 0.2737931894723712,
      "grad_norm": 0.0,
      "learning_rate": 1.704244226070645e-05,
      "loss": 0.9821,
      "step": 2195
    },
    {
      "epoch": 0.2739179244106274,
      "grad_norm": 0.0,
      "learning_rate": 1.7039573375600675e-05,
      "loss": 0.9991,
      "step": 2196
    },
    {
      "epoch": 0.2740426593488836,
      "grad_norm": 0.0,
      "learning_rate": 1.703670334145906e-05,
      "loss": 0.9826,
      "step": 2197
    },
    {
      "epoch": 0.2741673942871398,
      "grad_norm": 0.0,
      "learning_rate": 1.7033832158750065e-05,
      "loss": 1.0357,
      "step": 2198
    },
    {
      "epoch": 0.27429212922539603,
      "grad_norm": 0.0,
      "learning_rate": 1.703095982794234e-05,
      "loss": 0.9983,
      "step": 2199
    },
    {
      "epoch": 0.27441686416365224,
      "grad_norm": 0.0,
      "learning_rate": 1.7028086349504722e-05,
      "loss": 1.0385,
      "step": 2200
    },
    {
      "epoch": 0.27454159910190845,
      "grad_norm": 0.0,
      "learning_rate": 1.7025211723906233e-05,
      "loss": 0.9922,
      "step": 2201
    },
    {
      "epoch": 0.27466633404016466,
      "grad_norm": 0.0,
      "learning_rate": 1.7022335951616085e-05,
      "loss": 1.0187,
      "step": 2202
    },
    {
      "epoch": 0.27479106897842087,
      "grad_norm": 0.0,
      "learning_rate": 1.7019459033103684e-05,
      "loss": 0.9928,
      "step": 2203
    },
    {
      "epoch": 0.2749158039166771,
      "grad_norm": 0.0,
      "learning_rate": 1.7016580968838604e-05,
      "loss": 1.0882,
      "step": 2204
    },
    {
      "epoch": 0.2750405388549333,
      "grad_norm": 0.0,
      "learning_rate": 1.7013701759290622e-05,
      "loss": 1.0305,
      "step": 2205
    },
    {
      "epoch": 0.27516527379318945,
      "grad_norm": 0.0,
      "learning_rate": 1.70108214049297e-05,
      "loss": 1.0257,
      "step": 2206
    },
    {
      "epoch": 0.27529000873144566,
      "grad_norm": 0.0,
      "learning_rate": 1.7007939906225985e-05,
      "loss": 1.0525,
      "step": 2207
    },
    {
      "epoch": 0.27541474366970187,
      "grad_norm": 0.0,
      "learning_rate": 1.7005057263649806e-05,
      "loss": 1.0075,
      "step": 2208
    },
    {
      "epoch": 0.2755394786079581,
      "grad_norm": 0.0,
      "learning_rate": 1.7002173477671685e-05,
      "loss": 1.0337,
      "step": 2209
    },
    {
      "epoch": 0.2756642135462143,
      "grad_norm": 0.0,
      "learning_rate": 1.699928854876233e-05,
      "loss": 1.0646,
      "step": 2210
    },
    {
      "epoch": 0.2757889484844705,
      "grad_norm": 0.0,
      "learning_rate": 1.6996402477392635e-05,
      "loss": 1.0233,
      "step": 2211
    },
    {
      "epoch": 0.2759136834227267,
      "grad_norm": 0.0,
      "learning_rate": 1.699351526403367e-05,
      "loss": 0.9703,
      "step": 2212
    },
    {
      "epoch": 0.2760384183609829,
      "grad_norm": 0.0,
      "learning_rate": 1.6990626909156716e-05,
      "loss": 1.0333,
      "step": 2213
    },
    {
      "epoch": 0.27616315329923913,
      "grad_norm": 0.0,
      "learning_rate": 1.6987737413233214e-05,
      "loss": 1.0067,
      "step": 2214
    },
    {
      "epoch": 0.27628788823749534,
      "grad_norm": 0.0,
      "learning_rate": 1.6984846776734806e-05,
      "loss": 0.9981,
      "step": 2215
    },
    {
      "epoch": 0.27641262317575155,
      "grad_norm": 0.0,
      "learning_rate": 1.6981955000133318e-05,
      "loss": 1.0132,
      "step": 2216
    },
    {
      "epoch": 0.27653735811400776,
      "grad_norm": 0.0,
      "learning_rate": 1.6979062083900762e-05,
      "loss": 1.0439,
      "step": 2217
    },
    {
      "epoch": 0.2766620930522639,
      "grad_norm": 0.0,
      "learning_rate": 1.697616802850933e-05,
      "loss": 1.045,
      "step": 2218
    },
    {
      "epoch": 0.2767868279905201,
      "grad_norm": 0.0,
      "learning_rate": 1.6973272834431407e-05,
      "loss": 1.0732,
      "step": 2219
    },
    {
      "epoch": 0.27691156292877633,
      "grad_norm": 0.0,
      "learning_rate": 1.6970376502139563e-05,
      "loss": 1.0227,
      "step": 2220
    },
    {
      "epoch": 0.27703629786703254,
      "grad_norm": 0.0,
      "learning_rate": 1.6967479032106552e-05,
      "loss": 1.0334,
      "step": 2221
    },
    {
      "epoch": 0.27716103280528875,
      "grad_norm": 0.0,
      "learning_rate": 1.6964580424805314e-05,
      "loss": 1.049,
      "step": 2222
    },
    {
      "epoch": 0.27728576774354496,
      "grad_norm": 0.0,
      "learning_rate": 1.6961680680708974e-05,
      "loss": 0.9975,
      "step": 2223
    },
    {
      "epoch": 0.2774105026818012,
      "grad_norm": 0.0,
      "learning_rate": 1.6958779800290847e-05,
      "loss": 1.0186,
      "step": 2224
    },
    {
      "epoch": 0.2775352376200574,
      "grad_norm": 0.0,
      "learning_rate": 1.6955877784024418e-05,
      "loss": 1.0505,
      "step": 2225
    },
    {
      "epoch": 0.2776599725583136,
      "grad_norm": 0.0,
      "learning_rate": 1.6952974632383385e-05,
      "loss": 1.0351,
      "step": 2226
    },
    {
      "epoch": 0.2777847074965698,
      "grad_norm": 0.0,
      "learning_rate": 1.6950070345841608e-05,
      "loss": 1.0464,
      "step": 2227
    },
    {
      "epoch": 0.277909442434826,
      "grad_norm": 0.0,
      "learning_rate": 1.694716492487314e-05,
      "loss": 1.0185,
      "step": 2228
    },
    {
      "epoch": 0.2780341773730822,
      "grad_norm": 0.0,
      "learning_rate": 1.694425836995222e-05,
      "loss": 1.0251,
      "step": 2229
    },
    {
      "epoch": 0.2781589123113384,
      "grad_norm": 0.0,
      "learning_rate": 1.694135068155327e-05,
      "loss": 1.0055,
      "step": 2230
    },
    {
      "epoch": 0.2782836472495946,
      "grad_norm": 0.0,
      "learning_rate": 1.6938441860150895e-05,
      "loss": 0.9866,
      "step": 2231
    },
    {
      "epoch": 0.2784083821878508,
      "grad_norm": 0.0,
      "learning_rate": 1.6935531906219895e-05,
      "loss": 1.1086,
      "step": 2232
    },
    {
      "epoch": 0.278533117126107,
      "grad_norm": 0.0,
      "learning_rate": 1.6932620820235243e-05,
      "loss": 1.056,
      "step": 2233
    },
    {
      "epoch": 0.2786578520643632,
      "grad_norm": 0.0,
      "learning_rate": 1.6929708602672104e-05,
      "loss": 1.0002,
      "step": 2234
    },
    {
      "epoch": 0.27878258700261943,
      "grad_norm": 0.0,
      "learning_rate": 1.6926795254005823e-05,
      "loss": 1.0297,
      "step": 2235
    },
    {
      "epoch": 0.27890732194087564,
      "grad_norm": 0.0,
      "learning_rate": 1.6923880774711934e-05,
      "loss": 1.0433,
      "step": 2236
    },
    {
      "epoch": 0.27903205687913185,
      "grad_norm": 0.0,
      "learning_rate": 1.692096516526615e-05,
      "loss": 1.0558,
      "step": 2237
    },
    {
      "epoch": 0.27915679181738806,
      "grad_norm": 0.0,
      "learning_rate": 1.6918048426144374e-05,
      "loss": 1.0444,
      "step": 2238
    },
    {
      "epoch": 0.27928152675564427,
      "grad_norm": 0.0,
      "learning_rate": 1.6915130557822698e-05,
      "loss": 0.9788,
      "step": 2239
    },
    {
      "epoch": 0.2794062616939005,
      "grad_norm": 0.0,
      "learning_rate": 1.6912211560777377e-05,
      "loss": 1.0077,
      "step": 2240
    },
    {
      "epoch": 0.2795309966321567,
      "grad_norm": 0.0,
      "learning_rate": 1.690929143548488e-05,
      "loss": 1.0146,
      "step": 2241
    },
    {
      "epoch": 0.27965573157041285,
      "grad_norm": 0.0,
      "learning_rate": 1.6906370182421836e-05,
      "loss": 0.9622,
      "step": 2242
    },
    {
      "epoch": 0.27978046650866906,
      "grad_norm": 0.0,
      "learning_rate": 1.6903447802065066e-05,
      "loss": 1.0155,
      "step": 2243
    },
    {
      "epoch": 0.27990520144692527,
      "grad_norm": 0.0,
      "learning_rate": 1.6900524294891587e-05,
      "loss": 1.0151,
      "step": 2244
    },
    {
      "epoch": 0.2800299363851815,
      "grad_norm": 0.0,
      "learning_rate": 1.6897599661378578e-05,
      "loss": 1.0334,
      "step": 2245
    },
    {
      "epoch": 0.2801546713234377,
      "grad_norm": 0.0,
      "learning_rate": 1.689467390200342e-05,
      "loss": 1.0062,
      "step": 2246
    },
    {
      "epoch": 0.2802794062616939,
      "grad_norm": 0.0,
      "learning_rate": 1.689174701724366e-05,
      "loss": 1.0281,
      "step": 2247
    },
    {
      "epoch": 0.2804041411999501,
      "grad_norm": 0.0,
      "learning_rate": 1.6888819007577054e-05,
      "loss": 0.9931,
      "step": 2248
    },
    {
      "epoch": 0.2805288761382063,
      "grad_norm": 0.0,
      "learning_rate": 1.6885889873481514e-05,
      "loss": 1.0306,
      "step": 2249
    },
    {
      "epoch": 0.28065361107646253,
      "grad_norm": 0.0,
      "learning_rate": 1.688295961543516e-05,
      "loss": 1.0182,
      "step": 2250
    },
    {
      "epoch": 0.28077834601471874,
      "grad_norm": 0.0,
      "learning_rate": 1.6880028233916277e-05,
      "loss": 1.0193,
      "step": 2251
    },
    {
      "epoch": 0.28090308095297495,
      "grad_norm": 0.0,
      "learning_rate": 1.6877095729403338e-05,
      "loss": 0.9956,
      "step": 2252
    },
    {
      "epoch": 0.28102781589123116,
      "grad_norm": 0.0,
      "learning_rate": 1.687416210237501e-05,
      "loss": 1.0124,
      "step": 2253
    },
    {
      "epoch": 0.2811525508294873,
      "grad_norm": 0.0,
      "learning_rate": 1.6871227353310127e-05,
      "loss": 1.0179,
      "step": 2254
    },
    {
      "epoch": 0.2812772857677435,
      "grad_norm": 0.0,
      "learning_rate": 1.686829148268772e-05,
      "loss": 1.0405,
      "step": 2255
    },
    {
      "epoch": 0.28140202070599973,
      "grad_norm": 0.0,
      "learning_rate": 1.6865354490986995e-05,
      "loss": 0.9925,
      "step": 2256
    },
    {
      "epoch": 0.28152675564425594,
      "grad_norm": 0.0,
      "learning_rate": 1.686241637868734e-05,
      "loss": 1.0484,
      "step": 2257
    },
    {
      "epoch": 0.28165149058251215,
      "grad_norm": 0.0,
      "learning_rate": 1.685947714626833e-05,
      "loss": 1.0004,
      "step": 2258
    },
    {
      "epoch": 0.28177622552076836,
      "grad_norm": 0.0,
      "learning_rate": 1.6856536794209725e-05,
      "loss": 0.9401,
      "step": 2259
    },
    {
      "epoch": 0.2819009604590246,
      "grad_norm": 0.0,
      "learning_rate": 1.6853595322991464e-05,
      "loss": 1.0136,
      "step": 2260
    },
    {
      "epoch": 0.2820256953972808,
      "grad_norm": 0.0,
      "learning_rate": 1.6850652733093666e-05,
      "loss": 1.0258,
      "step": 2261
    },
    {
      "epoch": 0.282150430335537,
      "grad_norm": 0.0,
      "learning_rate": 1.684770902499664e-05,
      "loss": 1.0055,
      "step": 2262
    },
    {
      "epoch": 0.2822751652737932,
      "grad_norm": 0.0,
      "learning_rate": 1.6844764199180868e-05,
      "loss": 0.9902,
      "step": 2263
    },
    {
      "epoch": 0.2823999002120494,
      "grad_norm": 0.0,
      "learning_rate": 1.6841818256127027e-05,
      "loss": 1.0207,
      "step": 2264
    },
    {
      "epoch": 0.2825246351503056,
      "grad_norm": 0.0,
      "learning_rate": 1.6838871196315962e-05,
      "loss": 1.0362,
      "step": 2265
    },
    {
      "epoch": 0.2826493700885618,
      "grad_norm": 0.0,
      "learning_rate": 1.6835923020228714e-05,
      "loss": 1.0816,
      "step": 2266
    },
    {
      "epoch": 0.282774105026818,
      "grad_norm": 0.0,
      "learning_rate": 1.6832973728346497e-05,
      "loss": 1.0409,
      "step": 2267
    },
    {
      "epoch": 0.2828988399650742,
      "grad_norm": 0.0,
      "learning_rate": 1.683002332115071e-05,
      "loss": 1.0271,
      "step": 2268
    },
    {
      "epoch": 0.2830235749033304,
      "grad_norm": 0.0,
      "learning_rate": 1.6827071799122928e-05,
      "loss": 1.0429,
      "step": 2269
    },
    {
      "epoch": 0.2831483098415866,
      "grad_norm": 0.0,
      "learning_rate": 1.682411916274492e-05,
      "loss": 1.0512,
      "step": 2270
    },
    {
      "epoch": 0.28327304477984283,
      "grad_norm": 0.0,
      "learning_rate": 1.682116541249863e-05,
      "loss": 0.9693,
      "step": 2271
    },
    {
      "epoch": 0.28339777971809904,
      "grad_norm": 0.0,
      "learning_rate": 1.6818210548866188e-05,
      "loss": 0.9948,
      "step": 2272
    },
    {
      "epoch": 0.28352251465635525,
      "grad_norm": 0.0,
      "learning_rate": 1.6815254572329895e-05,
      "loss": 1.0251,
      "step": 2273
    },
    {
      "epoch": 0.28364724959461146,
      "grad_norm": 0.0,
      "learning_rate": 1.6812297483372243e-05,
      "loss": 1.0298,
      "step": 2274
    },
    {
      "epoch": 0.28377198453286767,
      "grad_norm": 0.0,
      "learning_rate": 1.6809339282475905e-05,
      "loss": 1.0606,
      "step": 2275
    },
    {
      "epoch": 0.2838967194711239,
      "grad_norm": 0.0,
      "learning_rate": 1.6806379970123736e-05,
      "loss": 0.9482,
      "step": 2276
    },
    {
      "epoch": 0.2840214544093801,
      "grad_norm": 0.0,
      "learning_rate": 1.6803419546798766e-05,
      "loss": 1.044,
      "step": 2277
    },
    {
      "epoch": 0.28414618934763625,
      "grad_norm": 0.0,
      "learning_rate": 1.6800458012984213e-05,
      "loss": 1.0825,
      "step": 2278
    },
    {
      "epoch": 0.28427092428589246,
      "grad_norm": 0.0,
      "learning_rate": 1.6797495369163477e-05,
      "loss": 1.0572,
      "step": 2279
    },
    {
      "epoch": 0.28439565922414867,
      "grad_norm": 0.0,
      "learning_rate": 1.679453161582013e-05,
      "loss": 1.0083,
      "step": 2280
    },
    {
      "epoch": 0.2845203941624049,
      "grad_norm": 0.0,
      "learning_rate": 1.6791566753437935e-05,
      "loss": 0.9875,
      "step": 2281
    },
    {
      "epoch": 0.2846451291006611,
      "grad_norm": 0.0,
      "learning_rate": 1.6788600782500828e-05,
      "loss": 1.0413,
      "step": 2282
    },
    {
      "epoch": 0.2847698640389173,
      "grad_norm": 0.0,
      "learning_rate": 1.6785633703492936e-05,
      "loss": 1.0202,
      "step": 2283
    },
    {
      "epoch": 0.2848945989771735,
      "grad_norm": 0.0,
      "learning_rate": 1.678266551689856e-05,
      "loss": 1.0311,
      "step": 2284
    },
    {
      "epoch": 0.2850193339154297,
      "grad_norm": 0.0,
      "learning_rate": 1.677969622320218e-05,
      "loss": 1.0199,
      "step": 2285
    },
    {
      "epoch": 0.28514406885368593,
      "grad_norm": 0.0,
      "learning_rate": 1.6776725822888466e-05,
      "loss": 1.0151,
      "step": 2286
    },
    {
      "epoch": 0.28526880379194214,
      "grad_norm": 0.0,
      "learning_rate": 1.677375431644225e-05,
      "loss": 1.0655,
      "step": 2287
    },
    {
      "epoch": 0.28539353873019835,
      "grad_norm": 0.0,
      "learning_rate": 1.677078170434857e-05,
      "loss": 1.077,
      "step": 2288
    },
    {
      "epoch": 0.28551827366845456,
      "grad_norm": 0.0,
      "learning_rate": 1.676780798709262e-05,
      "loss": 1.0368,
      "step": 2289
    },
    {
      "epoch": 0.2856430086067107,
      "grad_norm": 0.0,
      "learning_rate": 1.6764833165159796e-05,
      "loss": 1.0522,
      "step": 2290
    },
    {
      "epoch": 0.2857677435449669,
      "grad_norm": 0.0,
      "learning_rate": 1.6761857239035658e-05,
      "loss": 1.0127,
      "step": 2291
    },
    {
      "epoch": 0.28589247848322313,
      "grad_norm": 0.0,
      "learning_rate": 1.6758880209205952e-05,
      "loss": 1.0405,
      "step": 2292
    },
    {
      "epoch": 0.28601721342147934,
      "grad_norm": 0.0,
      "learning_rate": 1.6755902076156606e-05,
      "loss": 1.0336,
      "step": 2293
    },
    {
      "epoch": 0.28614194835973555,
      "grad_norm": 0.0,
      "learning_rate": 1.6752922840373724e-05,
      "loss": 1.0287,
      "step": 2294
    },
    {
      "epoch": 0.28626668329799176,
      "grad_norm": 0.0,
      "learning_rate": 1.6749942502343594e-05,
      "loss": 1.0679,
      "step": 2295
    },
    {
      "epoch": 0.286391418236248,
      "grad_norm": 0.0,
      "learning_rate": 1.6746961062552686e-05,
      "loss": 1.0688,
      "step": 2296
    },
    {
      "epoch": 0.2865161531745042,
      "grad_norm": 0.0,
      "learning_rate": 1.674397852148764e-05,
      "loss": 0.9617,
      "step": 2297
    },
    {
      "epoch": 0.2866408881127604,
      "grad_norm": 0.0,
      "learning_rate": 1.6740994879635284e-05,
      "loss": 0.9916,
      "step": 2298
    },
    {
      "epoch": 0.2867656230510166,
      "grad_norm": 0.0,
      "learning_rate": 1.6738010137482628e-05,
      "loss": 1.0038,
      "step": 2299
    },
    {
      "epoch": 0.2868903579892728,
      "grad_norm": 0.0,
      "learning_rate": 1.673502429551685e-05,
      "loss": 1.0137,
      "step": 2300
    },
    {
      "epoch": 0.287015092927529,
      "grad_norm": 0.0,
      "learning_rate": 1.673203735422532e-05,
      "loss": 1.0545,
      "step": 2301
    },
    {
      "epoch": 0.2871398278657852,
      "grad_norm": 0.0,
      "learning_rate": 1.6729049314095578e-05,
      "loss": 1.0169,
      "step": 2302
    },
    {
      "epoch": 0.2872645628040414,
      "grad_norm": 0.0,
      "learning_rate": 1.672606017561535e-05,
      "loss": 1.0199,
      "step": 2303
    },
    {
      "epoch": 0.2873892977422976,
      "grad_norm": 0.0,
      "learning_rate": 1.672306993927254e-05,
      "loss": 1.0529,
      "step": 2304
    },
    {
      "epoch": 0.2875140326805538,
      "grad_norm": 0.0,
      "learning_rate": 1.6720078605555227e-05,
      "loss": 1.0676,
      "step": 2305
    },
    {
      "epoch": 0.28763876761881,
      "grad_norm": 0.0,
      "learning_rate": 1.671708617495167e-05,
      "loss": 1.0542,
      "step": 2306
    },
    {
      "epoch": 0.28776350255706623,
      "grad_norm": 0.0,
      "learning_rate": 1.671409264795032e-05,
      "loss": 0.994,
      "step": 2307
    },
    {
      "epoch": 0.28788823749532244,
      "grad_norm": 0.0,
      "learning_rate": 1.6711098025039782e-05,
      "loss": 1.0224,
      "step": 2308
    },
    {
      "epoch": 0.28801297243357865,
      "grad_norm": 0.0,
      "learning_rate": 1.6708102306708864e-05,
      "loss": 1.0598,
      "step": 2309
    },
    {
      "epoch": 0.28813770737183486,
      "grad_norm": 0.0,
      "learning_rate": 1.6705105493446535e-05,
      "loss": 1.0208,
      "step": 2310
    },
    {
      "epoch": 0.28826244231009107,
      "grad_norm": 0.0,
      "learning_rate": 1.670210758574196e-05,
      "loss": 1.0292,
      "step": 2311
    },
    {
      "epoch": 0.2883871772483473,
      "grad_norm": 0.0,
      "learning_rate": 1.6699108584084463e-05,
      "loss": 1.0208,
      "step": 2312
    },
    {
      "epoch": 0.2885119121866035,
      "grad_norm": 0.0,
      "learning_rate": 1.6696108488963568e-05,
      "loss": 1.0272,
      "step": 2313
    },
    {
      "epoch": 0.28863664712485965,
      "grad_norm": 0.0,
      "learning_rate": 1.6693107300868953e-05,
      "loss": 1.0766,
      "step": 2314
    },
    {
      "epoch": 0.28876138206311586,
      "grad_norm": 0.0,
      "learning_rate": 1.66901050202905e-05,
      "loss": 1.0099,
      "step": 2315
    },
    {
      "epoch": 0.28888611700137207,
      "grad_norm": 0.0,
      "learning_rate": 1.6687101647718245e-05,
      "loss": 0.994,
      "step": 2316
    },
    {
      "epoch": 0.2890108519396283,
      "grad_norm": 0.0,
      "learning_rate": 1.6684097183642425e-05,
      "loss": 1.0413,
      "step": 2317
    },
    {
      "epoch": 0.2891355868778845,
      "grad_norm": 0.0,
      "learning_rate": 1.6681091628553437e-05,
      "loss": 1.0023,
      "step": 2318
    },
    {
      "epoch": 0.2892603218161407,
      "grad_norm": 0.0,
      "learning_rate": 1.667808498294187e-05,
      "loss": 1.0259,
      "step": 2319
    },
    {
      "epoch": 0.2893850567543969,
      "grad_norm": 0.0,
      "learning_rate": 1.6675077247298475e-05,
      "loss": 1.0317,
      "step": 2320
    },
    {
      "epoch": 0.2895097916926531,
      "grad_norm": 0.0,
      "learning_rate": 1.6672068422114195e-05,
      "loss": 0.9926,
      "step": 2321
    },
    {
      "epoch": 0.28963452663090933,
      "grad_norm": 0.0,
      "learning_rate": 1.666905850788015e-05,
      "loss": 0.9816,
      "step": 2322
    },
    {
      "epoch": 0.28975926156916554,
      "grad_norm": 0.0,
      "learning_rate": 1.666604750508763e-05,
      "loss": 1.0344,
      "step": 2323
    },
    {
      "epoch": 0.28988399650742175,
      "grad_norm": 0.0,
      "learning_rate": 1.666303541422811e-05,
      "loss": 0.9785,
      "step": 2324
    },
    {
      "epoch": 0.29000873144567796,
      "grad_norm": 0.0,
      "learning_rate": 1.666002223579323e-05,
      "loss": 0.9879,
      "step": 2325
    },
    {
      "epoch": 0.2901334663839341,
      "grad_norm": 0.0,
      "learning_rate": 1.6657007970274824e-05,
      "loss": 1.0147,
      "step": 2326
    },
    {
      "epoch": 0.2902582013221903,
      "grad_norm": 0.0,
      "learning_rate": 1.66539926181649e-05,
      "loss": 1.0073,
      "step": 2327
    },
    {
      "epoch": 0.29038293626044653,
      "grad_norm": 0.0,
      "learning_rate": 1.665097617995563e-05,
      "loss": 1.0125,
      "step": 2328
    },
    {
      "epoch": 0.29050767119870274,
      "grad_norm": 0.0,
      "learning_rate": 1.6647958656139377e-05,
      "loss": 1.0361,
      "step": 2329
    },
    {
      "epoch": 0.29063240613695895,
      "grad_norm": 0.0,
      "learning_rate": 1.6644940047208683e-05,
      "loss": 1.0575,
      "step": 2330
    },
    {
      "epoch": 0.29075714107521516,
      "grad_norm": 0.0,
      "learning_rate": 1.664192035365625e-05,
      "loss": 1.0472,
      "step": 2331
    },
    {
      "epoch": 0.2908818760134714,
      "grad_norm": 0.0,
      "learning_rate": 1.6638899575974975e-05,
      "loss": 1.0345,
      "step": 2332
    },
    {
      "epoch": 0.2910066109517276,
      "grad_norm": 0.0,
      "learning_rate": 1.6635877714657925e-05,
      "loss": 1.0259,
      "step": 2333
    },
    {
      "epoch": 0.2911313458899838,
      "grad_norm": 0.0,
      "learning_rate": 1.6632854770198337e-05,
      "loss": 1.0354,
      "step": 2334
    },
    {
      "epoch": 0.29125608082824,
      "grad_norm": 0.0,
      "learning_rate": 1.6629830743089645e-05,
      "loss": 1.0009,
      "step": 2335
    },
    {
      "epoch": 0.2913808157664962,
      "grad_norm": 0.0,
      "learning_rate": 1.6626805633825432e-05,
      "loss": 1.0139,
      "step": 2336
    },
    {
      "epoch": 0.2915055507047524,
      "grad_norm": 0.0,
      "learning_rate": 1.662377944289948e-05,
      "loss": 1.0245,
      "step": 2337
    },
    {
      "epoch": 0.2916302856430086,
      "grad_norm": 0.0,
      "learning_rate": 1.662075217080574e-05,
      "loss": 0.9966,
      "step": 2338
    },
    {
      "epoch": 0.2917550205812648,
      "grad_norm": 0.0,
      "learning_rate": 1.6617723818038337e-05,
      "loss": 0.9838,
      "step": 2339
    },
    {
      "epoch": 0.291879755519521,
      "grad_norm": 0.0,
      "learning_rate": 1.6614694385091575e-05,
      "loss": 1.02,
      "step": 2340
    },
    {
      "epoch": 0.2920044904577772,
      "grad_norm": 0.0,
      "learning_rate": 1.6611663872459932e-05,
      "loss": 1.0288,
      "step": 2341
    },
    {
      "epoch": 0.2921292253960334,
      "grad_norm": 0.0,
      "learning_rate": 1.660863228063807e-05,
      "loss": 1.0041,
      "step": 2342
    },
    {
      "epoch": 0.29225396033428963,
      "grad_norm": 0.0,
      "learning_rate": 1.6605599610120816e-05,
      "loss": 1.0245,
      "step": 2343
    },
    {
      "epoch": 0.29237869527254584,
      "grad_norm": 0.0,
      "learning_rate": 1.6602565861403176e-05,
      "loss": 1.0032,
      "step": 2344
    },
    {
      "epoch": 0.29250343021080205,
      "grad_norm": 0.0,
      "learning_rate": 1.659953103498034e-05,
      "loss": 0.9969,
      "step": 2345
    },
    {
      "epoch": 0.29262816514905826,
      "grad_norm": 0.0,
      "learning_rate": 1.6596495131347667e-05,
      "loss": 1.0124,
      "step": 2346
    },
    {
      "epoch": 0.29275290008731447,
      "grad_norm": 0.0,
      "learning_rate": 1.659345815100069e-05,
      "loss": 1.0403,
      "step": 2347
    },
    {
      "epoch": 0.2928776350255707,
      "grad_norm": 0.0,
      "learning_rate": 1.6590420094435125e-05,
      "loss": 0.9636,
      "step": 2348
    },
    {
      "epoch": 0.2930023699638269,
      "grad_norm": 0.0,
      "learning_rate": 1.6587380962146854e-05,
      "loss": 1.0059,
      "step": 2349
    },
    {
      "epoch": 0.29312710490208305,
      "grad_norm": 0.0,
      "learning_rate": 1.6584340754631945e-05,
      "loss": 1.0682,
      "step": 2350
    },
    {
      "epoch": 0.29325183984033926,
      "grad_norm": 0.0,
      "learning_rate": 1.6581299472386633e-05,
      "loss": 1.0183,
      "step": 2351
    },
    {
      "epoch": 0.29337657477859547,
      "grad_norm": 0.0,
      "learning_rate": 1.6578257115907334e-05,
      "loss": 1.0179,
      "step": 2352
    },
    {
      "epoch": 0.2935013097168517,
      "grad_norm": 0.0,
      "learning_rate": 1.657521368569064e-05,
      "loss": 1.065,
      "step": 2353
    },
    {
      "epoch": 0.2936260446551079,
      "grad_norm": 0.0,
      "learning_rate": 1.657216918223331e-05,
      "loss": 0.9933,
      "step": 2354
    },
    {
      "epoch": 0.2937507795933641,
      "grad_norm": 0.0,
      "learning_rate": 1.6569123606032284e-05,
      "loss": 1.0508,
      "step": 2355
    },
    {
      "epoch": 0.2938755145316203,
      "grad_norm": 0.0,
      "learning_rate": 1.656607695758468e-05,
      "loss": 0.9886,
      "step": 2356
    },
    {
      "epoch": 0.2940002494698765,
      "grad_norm": 0.0,
      "learning_rate": 1.6563029237387787e-05,
      "loss": 0.9817,
      "step": 2357
    },
    {
      "epoch": 0.29412498440813273,
      "grad_norm": 0.0,
      "learning_rate": 1.6559980445939068e-05,
      "loss": 1.025,
      "step": 2358
    },
    {
      "epoch": 0.29424971934638894,
      "grad_norm": 0.0,
      "learning_rate": 1.6556930583736167e-05,
      "loss": 1.0131,
      "step": 2359
    },
    {
      "epoch": 0.29437445428464515,
      "grad_norm": 0.0,
      "learning_rate": 1.6553879651276894e-05,
      "loss": 1.0271,
      "step": 2360
    },
    {
      "epoch": 0.29449918922290136,
      "grad_norm": 0.0,
      "learning_rate": 1.6550827649059238e-05,
      "loss": 1.0384,
      "step": 2361
    },
    {
      "epoch": 0.2946239241611575,
      "grad_norm": 0.0,
      "learning_rate": 1.6547774577581365e-05,
      "loss": 1.0047,
      "step": 2362
    },
    {
      "epoch": 0.2947486590994137,
      "grad_norm": 0.0,
      "learning_rate": 1.6544720437341616e-05,
      "loss": 1.0226,
      "step": 2363
    },
    {
      "epoch": 0.29487339403766993,
      "grad_norm": 0.0,
      "learning_rate": 1.6541665228838496e-05,
      "loss": 0.9743,
      "step": 2364
    },
    {
      "epoch": 0.29499812897592614,
      "grad_norm": 0.0,
      "learning_rate": 1.6538608952570698e-05,
      "loss": 0.9861,
      "step": 2365
    },
    {
      "epoch": 0.29512286391418235,
      "grad_norm": 0.0,
      "learning_rate": 1.6535551609037083e-05,
      "loss": 1.0048,
      "step": 2366
    },
    {
      "epoch": 0.29524759885243856,
      "grad_norm": 0.0,
      "learning_rate": 1.653249319873668e-05,
      "loss": 1.0002,
      "step": 2367
    },
    {
      "epoch": 0.2953723337906948,
      "grad_norm": 0.0,
      "learning_rate": 1.652943372216871e-05,
      "loss": 1.0351,
      "step": 2368
    },
    {
      "epoch": 0.295497068728951,
      "grad_norm": 0.0,
      "learning_rate": 1.6526373179832546e-05,
      "loss": 1.0409,
      "step": 2369
    },
    {
      "epoch": 0.2956218036672072,
      "grad_norm": 0.0,
      "learning_rate": 1.6523311572227753e-05,
      "loss": 0.9831,
      "step": 2370
    },
    {
      "epoch": 0.2957465386054634,
      "grad_norm": 0.0,
      "learning_rate": 1.6520248899854057e-05,
      "loss": 1.0408,
      "step": 2371
    },
    {
      "epoch": 0.2958712735437196,
      "grad_norm": 0.0,
      "learning_rate": 1.6517185163211367e-05,
      "loss": 1.0264,
      "step": 2372
    },
    {
      "epoch": 0.2959960084819758,
      "grad_norm": 0.0,
      "learning_rate": 1.6514120362799763e-05,
      "loss": 0.9919,
      "step": 2373
    },
    {
      "epoch": 0.296120743420232,
      "grad_norm": 0.0,
      "learning_rate": 1.6511054499119493e-05,
      "loss": 0.9897,
      "step": 2374
    },
    {
      "epoch": 0.2962454783584882,
      "grad_norm": 0.0,
      "learning_rate": 1.6507987572670986e-05,
      "loss": 1.0611,
      "step": 2375
    },
    {
      "epoch": 0.2963702132967444,
      "grad_norm": 0.0,
      "learning_rate": 1.6504919583954845e-05,
      "loss": 1.0452,
      "step": 2376
    },
    {
      "epoch": 0.2964949482350006,
      "grad_norm": 0.0,
      "learning_rate": 1.650185053347184e-05,
      "loss": 1.0485,
      "step": 2377
    },
    {
      "epoch": 0.2966196831732568,
      "grad_norm": 0.0,
      "learning_rate": 1.6498780421722912e-05,
      "loss": 1.0598,
      "step": 2378
    },
    {
      "epoch": 0.29674441811151303,
      "grad_norm": 0.0,
      "learning_rate": 1.6495709249209185e-05,
      "loss": 1.0163,
      "step": 2379
    },
    {
      "epoch": 0.29686915304976924,
      "grad_norm": 0.0,
      "learning_rate": 1.649263701643196e-05,
      "loss": 1.0265,
      "step": 2380
    },
    {
      "epoch": 0.29699388798802545,
      "grad_norm": 0.0,
      "learning_rate": 1.648956372389269e-05,
      "loss": 1.0057,
      "step": 2381
    },
    {
      "epoch": 0.29711862292628166,
      "grad_norm": 0.0,
      "learning_rate": 1.648648937209302e-05,
      "loss": 1.0496,
      "step": 2382
    },
    {
      "epoch": 0.29724335786453787,
      "grad_norm": 0.0,
      "learning_rate": 1.6483413961534764e-05,
      "loss": 0.9819,
      "step": 2383
    },
    {
      "epoch": 0.2973680928027941,
      "grad_norm": 0.0,
      "learning_rate": 1.64803374927199e-05,
      "loss": 1.0014,
      "step": 2384
    },
    {
      "epoch": 0.2974928277410503,
      "grad_norm": 0.0,
      "learning_rate": 1.647725996615059e-05,
      "loss": 1.0174,
      "step": 2385
    },
    {
      "epoch": 0.29761756267930645,
      "grad_norm": 0.0,
      "learning_rate": 1.6474181382329165e-05,
      "loss": 1.0011,
      "step": 2386
    },
    {
      "epoch": 0.29774229761756266,
      "grad_norm": 0.0,
      "learning_rate": 1.647110174175812e-05,
      "loss": 1.015,
      "step": 2387
    },
    {
      "epoch": 0.29786703255581887,
      "grad_norm": 0.0,
      "learning_rate": 1.6468021044940142e-05,
      "loss": 1.0096,
      "step": 2388
    },
    {
      "epoch": 0.2979917674940751,
      "grad_norm": 0.0,
      "learning_rate": 1.6464939292378066e-05,
      "loss": 0.9837,
      "step": 2389
    },
    {
      "epoch": 0.2981165024323313,
      "grad_norm": 0.0,
      "learning_rate": 1.6461856484574917e-05,
      "loss": 0.9836,
      "step": 2390
    },
    {
      "epoch": 0.2982412373705875,
      "grad_norm": 0.0,
      "learning_rate": 1.6458772622033894e-05,
      "loss": 1.0068,
      "step": 2391
    },
    {
      "epoch": 0.2983659723088437,
      "grad_norm": 0.0,
      "learning_rate": 1.6455687705258348e-05,
      "loss": 1.0248,
      "step": 2392
    },
    {
      "epoch": 0.2984907072470999,
      "grad_norm": 0.0,
      "learning_rate": 1.6452601734751827e-05,
      "loss": 1.0562,
      "step": 2393
    },
    {
      "epoch": 0.29861544218535613,
      "grad_norm": 0.0,
      "learning_rate": 1.644951471101803e-05,
      "loss": 1.0007,
      "step": 2394
    },
    {
      "epoch": 0.29874017712361234,
      "grad_norm": 0.0,
      "learning_rate": 1.6446426634560838e-05,
      "loss": 1.0401,
      "step": 2395
    },
    {
      "epoch": 0.29886491206186855,
      "grad_norm": 0.0,
      "learning_rate": 1.644333750588431e-05,
      "loss": 0.9795,
      "step": 2396
    },
    {
      "epoch": 0.29898964700012476,
      "grad_norm": 0.0,
      "learning_rate": 1.6440247325492665e-05,
      "loss": 1.0668,
      "step": 2397
    },
    {
      "epoch": 0.2991143819383809,
      "grad_norm": 0.0,
      "learning_rate": 1.6437156093890297e-05,
      "loss": 0.9592,
      "step": 2398
    },
    {
      "epoch": 0.2992391168766371,
      "grad_norm": 0.0,
      "learning_rate": 1.6434063811581774e-05,
      "loss": 0.9987,
      "step": 2399
    },
    {
      "epoch": 0.29936385181489333,
      "grad_norm": 0.0,
      "learning_rate": 1.6430970479071837e-05,
      "loss": 1.0369,
      "step": 2400
    },
    {
      "epoch": 0.29948858675314954,
      "grad_norm": 0.0,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 0.9575,
      "step": 2401
    },
    {
      "epoch": 0.29961332169140575,
      "grad_norm": 0.0,
      "learning_rate": 1.6424780665467526e-05,
      "loss": 1.0227,
      "step": 2402
    },
    {
      "epoch": 0.29973805662966196,
      "grad_norm": 0.0,
      "learning_rate": 1.642168418538349e-05,
      "loss": 1.0428,
      "step": 2403
    },
    {
      "epoch": 0.2998627915679182,
      "grad_norm": 0.0,
      "learning_rate": 1.6418586657118698e-05,
      "loss": 0.9911,
      "step": 2404
    },
    {
      "epoch": 0.2999875265061744,
      "grad_norm": 0.0,
      "learning_rate": 1.6415488081178753e-05,
      "loss": 1.008,
      "step": 2405
    },
    {
      "epoch": 0.3001122614444306,
      "grad_norm": 0.0,
      "learning_rate": 1.641238845806942e-05,
      "loss": 1.0141,
      "step": 2406
    },
    {
      "epoch": 0.3002369963826868,
      "grad_norm": 0.0,
      "learning_rate": 1.6409287788296637e-05,
      "loss": 0.9955,
      "step": 2407
    },
    {
      "epoch": 0.300361731320943,
      "grad_norm": 0.0,
      "learning_rate": 1.640618607236651e-05,
      "loss": 1.0393,
      "step": 2408
    },
    {
      "epoch": 0.3004864662591992,
      "grad_norm": 0.0,
      "learning_rate": 1.640308331078532e-05,
      "loss": 0.9804,
      "step": 2409
    },
    {
      "epoch": 0.3006112011974554,
      "grad_norm": 0.0,
      "learning_rate": 1.6399979504059506e-05,
      "loss": 1.0451,
      "step": 2410
    },
    {
      "epoch": 0.3007359361357116,
      "grad_norm": 0.0,
      "learning_rate": 1.63968746526957e-05,
      "loss": 0.9881,
      "step": 2411
    },
    {
      "epoch": 0.3008606710739678,
      "grad_norm": 0.0,
      "learning_rate": 1.639376875720068e-05,
      "loss": 1.0557,
      "step": 2412
    },
    {
      "epoch": 0.300985406012224,
      "grad_norm": 0.0,
      "learning_rate": 1.6390661818081417e-05,
      "loss": 1.0807,
      "step": 2413
    },
    {
      "epoch": 0.3011101409504802,
      "grad_norm": 0.0,
      "learning_rate": 1.6387553835845038e-05,
      "loss": 1.0597,
      "step": 2414
    },
    {
      "epoch": 0.30123487588873643,
      "grad_norm": 0.0,
      "learning_rate": 1.6384444810998843e-05,
      "loss": 1.0127,
      "step": 2415
    },
    {
      "epoch": 0.30135961082699264,
      "grad_norm": 0.0,
      "learning_rate": 1.6381334744050304e-05,
      "loss": 1.018,
      "step": 2416
    },
    {
      "epoch": 0.30148434576524885,
      "grad_norm": 0.0,
      "learning_rate": 1.637822363550706e-05,
      "loss": 1.0214,
      "step": 2417
    },
    {
      "epoch": 0.30160908070350506,
      "grad_norm": 0.0,
      "learning_rate": 1.637511148587693e-05,
      "loss": 1.0481,
      "step": 2418
    },
    {
      "epoch": 0.30173381564176127,
      "grad_norm": 0.0,
      "learning_rate": 1.6371998295667885e-05,
      "loss": 0.9625,
      "step": 2419
    },
    {
      "epoch": 0.3018585505800175,
      "grad_norm": 0.0,
      "learning_rate": 1.6368884065388085e-05,
      "loss": 0.9967,
      "step": 2420
    },
    {
      "epoch": 0.3019832855182737,
      "grad_norm": 0.0,
      "learning_rate": 1.6365768795545844e-05,
      "loss": 1.0345,
      "step": 2421
    },
    {
      "epoch": 0.30210802045652985,
      "grad_norm": 0.0,
      "learning_rate": 1.636265248664965e-05,
      "loss": 0.9973,
      "step": 2422
    },
    {
      "epoch": 0.30223275539478606,
      "grad_norm": 0.0,
      "learning_rate": 1.6359535139208178e-05,
      "loss": 0.9979,
      "step": 2423
    },
    {
      "epoch": 0.30235749033304227,
      "grad_norm": 0.0,
      "learning_rate": 1.6356416753730244e-05,
      "loss": 1.048,
      "step": 2424
    },
    {
      "epoch": 0.3024822252712985,
      "grad_norm": 0.0,
      "learning_rate": 1.635329733072485e-05,
      "loss": 1.0165,
      "step": 2425
    },
    {
      "epoch": 0.3026069602095547,
      "grad_norm": 0.0,
      "learning_rate": 1.635017687070117e-05,
      "loss": 1.0066,
      "step": 2426
    },
    {
      "epoch": 0.3027316951478109,
      "grad_norm": 0.0,
      "learning_rate": 1.6347055374168536e-05,
      "loss": 1.011,
      "step": 2427
    },
    {
      "epoch": 0.3028564300860671,
      "grad_norm": 0.0,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 1.0102,
      "step": 2428
    },
    {
      "epoch": 0.3029811650243233,
      "grad_norm": 0.0,
      "learning_rate": 1.6340809273614606e-05,
      "loss": 1.047,
      "step": 2429
    },
    {
      "epoch": 0.30310589996257953,
      "grad_norm": 0.0,
      "learning_rate": 1.6337684670612834e-05,
      "loss": 1.0546,
      "step": 2430
    },
    {
      "epoch": 0.30323063490083574,
      "grad_norm": 0.0,
      "learning_rate": 1.6334559033141152e-05,
      "loss": 1.0154,
      "step": 2431
    },
    {
      "epoch": 0.30335536983909195,
      "grad_norm": 0.0,
      "learning_rate": 1.633143236170974e-05,
      "loss": 1.0009,
      "step": 2432
    },
    {
      "epoch": 0.30348010477734816,
      "grad_norm": 0.0,
      "learning_rate": 1.6328304656828953e-05,
      "loss": 1.0109,
      "step": 2433
    },
    {
      "epoch": 0.3036048397156043,
      "grad_norm": 0.0,
      "learning_rate": 1.6325175919009307e-05,
      "loss": 1.0341,
      "step": 2434
    },
    {
      "epoch": 0.3037295746538605,
      "grad_norm": 0.0,
      "learning_rate": 1.63220461487615e-05,
      "loss": 1.016,
      "step": 2435
    },
    {
      "epoch": 0.30385430959211673,
      "grad_norm": 0.0,
      "learning_rate": 1.631891534659638e-05,
      "loss": 1.0428,
      "step": 2436
    },
    {
      "epoch": 0.30397904453037294,
      "grad_norm": 0.0,
      "learning_rate": 1.6315783513024977e-05,
      "loss": 1.0559,
      "step": 2437
    },
    {
      "epoch": 0.30410377946862915,
      "grad_norm": 0.0,
      "learning_rate": 1.6312650648558485e-05,
      "loss": 0.9891,
      "step": 2438
    },
    {
      "epoch": 0.30422851440688536,
      "grad_norm": 0.0,
      "learning_rate": 1.6309516753708264e-05,
      "loss": 0.9947,
      "step": 2439
    },
    {
      "epoch": 0.3043532493451416,
      "grad_norm": 0.0,
      "learning_rate": 1.630638182898585e-05,
      "loss": 1.0301,
      "step": 2440
    },
    {
      "epoch": 0.3044779842833978,
      "grad_norm": 0.0,
      "learning_rate": 1.6303245874902936e-05,
      "loss": 1.0273,
      "step": 2441
    },
    {
      "epoch": 0.304602719221654,
      "grad_norm": 0.0,
      "learning_rate": 1.6300108891971395e-05,
      "loss": 0.9869,
      "step": 2442
    },
    {
      "epoch": 0.3047274541599102,
      "grad_norm": 0.0,
      "learning_rate": 1.6296970880703257e-05,
      "loss": 1.0428,
      "step": 2443
    },
    {
      "epoch": 0.3048521890981664,
      "grad_norm": 0.0,
      "learning_rate": 1.6293831841610725e-05,
      "loss": 1.0374,
      "step": 2444
    },
    {
      "epoch": 0.3049769240364226,
      "grad_norm": 0.0,
      "learning_rate": 1.6290691775206168e-05,
      "loss": 1.0274,
      "step": 2445
    },
    {
      "epoch": 0.3051016589746788,
      "grad_norm": 0.0,
      "learning_rate": 1.6287550682002126e-05,
      "loss": 1.0411,
      "step": 2446
    },
    {
      "epoch": 0.305226393912935,
      "grad_norm": 0.0,
      "learning_rate": 1.6284408562511305e-05,
      "loss": 1.0266,
      "step": 2447
    },
    {
      "epoch": 0.3053511288511912,
      "grad_norm": 0.0,
      "learning_rate": 1.628126541724658e-05,
      "loss": 1.0041,
      "step": 2448
    },
    {
      "epoch": 0.3054758637894474,
      "grad_norm": 0.0,
      "learning_rate": 1.627812124672099e-05,
      "loss": 1.0221,
      "step": 2449
    },
    {
      "epoch": 0.3056005987277036,
      "grad_norm": 0.0,
      "learning_rate": 1.6274976051447734e-05,
      "loss": 1.0,
      "step": 2450
    },
    {
      "epoch": 0.30572533366595983,
      "grad_norm": 0.0,
      "learning_rate": 1.62718298319402e-05,
      "loss": 1.052,
      "step": 2451
    },
    {
      "epoch": 0.30585006860421604,
      "grad_norm": 0.0,
      "learning_rate": 1.6268682588711927e-05,
      "loss": 1.0807,
      "step": 2452
    },
    {
      "epoch": 0.30597480354247225,
      "grad_norm": 0.0,
      "learning_rate": 1.626553432227662e-05,
      "loss": 0.984,
      "step": 2453
    },
    {
      "epoch": 0.30609953848072846,
      "grad_norm": 0.0,
      "learning_rate": 1.6262385033148164e-05,
      "loss": 0.9954,
      "step": 2454
    },
    {
      "epoch": 0.30622427341898467,
      "grad_norm": 0.0,
      "learning_rate": 1.6259234721840595e-05,
      "loss": 1.0011,
      "step": 2455
    },
    {
      "epoch": 0.3063490083572409,
      "grad_norm": 0.0,
      "learning_rate": 1.625608338886812e-05,
      "loss": 1.0323,
      "step": 2456
    },
    {
      "epoch": 0.3064737432954971,
      "grad_norm": 0.0,
      "learning_rate": 1.6252931034745122e-05,
      "loss": 1.0077,
      "step": 2457
    },
    {
      "epoch": 0.30659847823375325,
      "grad_norm": 0.0,
      "learning_rate": 1.6249777659986146e-05,
      "loss": 0.9702,
      "step": 2458
    },
    {
      "epoch": 0.30672321317200946,
      "grad_norm": 0.0,
      "learning_rate": 1.62466232651059e-05,
      "loss": 0.9739,
      "step": 2459
    },
    {
      "epoch": 0.30684794811026567,
      "grad_norm": 0.0,
      "learning_rate": 1.6243467850619262e-05,
      "loss": 1.0168,
      "step": 2460
    },
    {
      "epoch": 0.3069726830485219,
      "grad_norm": 0.0,
      "learning_rate": 1.6240311417041274e-05,
      "loss": 1.0536,
      "step": 2461
    },
    {
      "epoch": 0.3070974179867781,
      "grad_norm": 0.0,
      "learning_rate": 1.623715396488714e-05,
      "loss": 0.9934,
      "step": 2462
    },
    {
      "epoch": 0.3072221529250343,
      "grad_norm": 0.0,
      "learning_rate": 1.6233995494672242e-05,
      "loss": 0.9783,
      "step": 2463
    },
    {
      "epoch": 0.3073468878632905,
      "grad_norm": 0.0,
      "learning_rate": 1.6230836006912127e-05,
      "loss": 1.0294,
      "step": 2464
    },
    {
      "epoch": 0.3074716228015467,
      "grad_norm": 0.0,
      "learning_rate": 1.6227675502122492e-05,
      "loss": 1.0033,
      "step": 2465
    },
    {
      "epoch": 0.30759635773980293,
      "grad_norm": 0.0,
      "learning_rate": 1.622451398081922e-05,
      "loss": 1.0207,
      "step": 2466
    },
    {
      "epoch": 0.30772109267805914,
      "grad_norm": 0.0,
      "learning_rate": 1.6221351443518343e-05,
      "loss": 1.0155,
      "step": 2467
    },
    {
      "epoch": 0.30784582761631535,
      "grad_norm": 0.0,
      "learning_rate": 1.621818789073607e-05,
      "loss": 1.0469,
      "step": 2468
    },
    {
      "epoch": 0.30797056255457156,
      "grad_norm": 0.0,
      "learning_rate": 1.621502332298878e-05,
      "loss": 0.9771,
      "step": 2469
    },
    {
      "epoch": 0.30809529749282777,
      "grad_norm": 0.0,
      "learning_rate": 1.6211857740793e-05,
      "loss": 0.9978,
      "step": 2470
    },
    {
      "epoch": 0.3082200324310839,
      "grad_norm": 0.0,
      "learning_rate": 1.620869114466544e-05,
      "loss": 0.9917,
      "step": 2471
    },
    {
      "epoch": 0.30834476736934013,
      "grad_norm": 0.0,
      "learning_rate": 1.6205523535122963e-05,
      "loss": 1.0526,
      "step": 2472
    },
    {
      "epoch": 0.30846950230759634,
      "grad_norm": 0.0,
      "learning_rate": 1.6202354912682602e-05,
      "loss": 1.06,
      "step": 2473
    },
    {
      "epoch": 0.30859423724585255,
      "grad_norm": 0.0,
      "learning_rate": 1.619918527786156e-05,
      "loss": 1.0306,
      "step": 2474
    },
    {
      "epoch": 0.30871897218410876,
      "grad_norm": 0.0,
      "learning_rate": 1.6196014631177203e-05,
      "loss": 0.975,
      "step": 2475
    },
    {
      "epoch": 0.308843707122365,
      "grad_norm": 0.0,
      "learning_rate": 1.6192842973147053e-05,
      "loss": 1.0581,
      "step": 2476
    },
    {
      "epoch": 0.3089684420606212,
      "grad_norm": 0.0,
      "learning_rate": 1.6189670304288814e-05,
      "loss": 1.0027,
      "step": 2477
    },
    {
      "epoch": 0.3090931769988774,
      "grad_norm": 0.0,
      "learning_rate": 1.6186496625120338e-05,
      "loss": 1.0669,
      "step": 2478
    },
    {
      "epoch": 0.3092179119371336,
      "grad_norm": 0.0,
      "learning_rate": 1.6183321936159654e-05,
      "loss": 0.9937,
      "step": 2479
    },
    {
      "epoch": 0.3093426468753898,
      "grad_norm": 0.0,
      "learning_rate": 1.6180146237924945e-05,
      "loss": 1.0345,
      "step": 2480
    },
    {
      "epoch": 0.309467381813646,
      "grad_norm": 0.0,
      "learning_rate": 1.6176969530934573e-05,
      "loss": 0.9873,
      "step": 2481
    },
    {
      "epoch": 0.30959211675190224,
      "grad_norm": 0.0,
      "learning_rate": 1.6173791815707053e-05,
      "loss": 1.0618,
      "step": 2482
    },
    {
      "epoch": 0.3097168516901584,
      "grad_norm": 0.0,
      "learning_rate": 1.6170613092761064e-05,
      "loss": 0.9951,
      "step": 2483
    },
    {
      "epoch": 0.3098415866284146,
      "grad_norm": 0.0,
      "learning_rate": 1.6167433362615463e-05,
      "loss": 0.9702,
      "step": 2484
    },
    {
      "epoch": 0.3099663215666708,
      "grad_norm": 0.0,
      "learning_rate": 1.6164252625789255e-05,
      "loss": 0.9784,
      "step": 2485
    },
    {
      "epoch": 0.310091056504927,
      "grad_norm": 0.0,
      "learning_rate": 1.616107088280162e-05,
      "loss": 0.994,
      "step": 2486
    },
    {
      "epoch": 0.31021579144318323,
      "grad_norm": 0.0,
      "learning_rate": 1.6157888134171896e-05,
      "loss": 0.9925,
      "step": 2487
    },
    {
      "epoch": 0.31034052638143944,
      "grad_norm": 0.0,
      "learning_rate": 1.6154704380419587e-05,
      "loss": 1.0161,
      "step": 2488
    },
    {
      "epoch": 0.31046526131969565,
      "grad_norm": 0.0,
      "learning_rate": 1.615151962206437e-05,
      "loss": 1.0179,
      "step": 2489
    },
    {
      "epoch": 0.31058999625795186,
      "grad_norm": 0.0,
      "learning_rate": 1.6148333859626063e-05,
      "loss": 1.0141,
      "step": 2490
    },
    {
      "epoch": 0.31071473119620807,
      "grad_norm": 0.0,
      "learning_rate": 1.6145147093624677e-05,
      "loss": 1.0485,
      "step": 2491
    },
    {
      "epoch": 0.3108394661344643,
      "grad_norm": 0.0,
      "learning_rate": 1.6141959324580366e-05,
      "loss": 0.9776,
      "step": 2492
    },
    {
      "epoch": 0.3109642010727205,
      "grad_norm": 0.0,
      "learning_rate": 1.613877055301346e-05,
      "loss": 1.0243,
      "step": 2493
    },
    {
      "epoch": 0.3110889360109767,
      "grad_norm": 0.0,
      "learning_rate": 1.6135580779444438e-05,
      "loss": 1.049,
      "step": 2494
    },
    {
      "epoch": 0.31121367094923286,
      "grad_norm": 0.0,
      "learning_rate": 1.6132390004393954e-05,
      "loss": 1.0525,
      "step": 2495
    },
    {
      "epoch": 0.31133840588748907,
      "grad_norm": 0.0,
      "learning_rate": 1.6129198228382826e-05,
      "loss": 1.0552,
      "step": 2496
    },
    {
      "epoch": 0.3114631408257453,
      "grad_norm": 0.0,
      "learning_rate": 1.6126005451932028e-05,
      "loss": 0.9828,
      "step": 2497
    },
    {
      "epoch": 0.3115878757640015,
      "grad_norm": 0.0,
      "learning_rate": 1.6122811675562712e-05,
      "loss": 1.0339,
      "step": 2498
    },
    {
      "epoch": 0.3117126107022577,
      "grad_norm": 0.0,
      "learning_rate": 1.611961689979617e-05,
      "loss": 1.0209,
      "step": 2499
    },
    {
      "epoch": 0.3118373456405139,
      "grad_norm": 0.0,
      "learning_rate": 1.6116421125153876e-05,
      "loss": 1.0252,
      "step": 2500
    },
    {
      "epoch": 0.3119620805787701,
      "grad_norm": 0.0,
      "learning_rate": 1.611322435215746e-05,
      "loss": 1.036,
      "step": 2501
    },
    {
      "epoch": 0.31208681551702633,
      "grad_norm": 0.0,
      "learning_rate": 1.6110026581328715e-05,
      "loss": 0.9988,
      "step": 2502
    },
    {
      "epoch": 0.31221155045528254,
      "grad_norm": 0.0,
      "learning_rate": 1.61068278131896e-05,
      "loss": 0.9658,
      "step": 2503
    },
    {
      "epoch": 0.31233628539353875,
      "grad_norm": 0.0,
      "learning_rate": 1.6103628048262236e-05,
      "loss": 1.0228,
      "step": 2504
    },
    {
      "epoch": 0.31246102033179496,
      "grad_norm": 0.0,
      "learning_rate": 1.61004272870689e-05,
      "loss": 0.9389,
      "step": 2505
    },
    {
      "epoch": 0.31258575527005117,
      "grad_norm": 0.0,
      "learning_rate": 1.6097225530132044e-05,
      "loss": 1.0081,
      "step": 2506
    },
    {
      "epoch": 0.3127104902083073,
      "grad_norm": 0.0,
      "learning_rate": 1.609402277797427e-05,
      "loss": 1.0306,
      "step": 2507
    },
    {
      "epoch": 0.31283522514656353,
      "grad_norm": 0.0,
      "learning_rate": 1.6090819031118344e-05,
      "loss": 0.9998,
      "step": 2508
    },
    {
      "epoch": 0.31295996008481974,
      "grad_norm": 0.0,
      "learning_rate": 1.608761429008721e-05,
      "loss": 1.0615,
      "step": 2509
    },
    {
      "epoch": 0.31308469502307595,
      "grad_norm": 0.0,
      "learning_rate": 1.6084408555403952e-05,
      "loss": 1.0166,
      "step": 2510
    },
    {
      "epoch": 0.31320942996133216,
      "grad_norm": 0.0,
      "learning_rate": 1.608120182759183e-05,
      "loss": 0.9958,
      "step": 2511
    },
    {
      "epoch": 0.3133341648995884,
      "grad_norm": 0.0,
      "learning_rate": 1.6077994107174267e-05,
      "loss": 1.0546,
      "step": 2512
    },
    {
      "epoch": 0.3134588998378446,
      "grad_norm": 0.0,
      "learning_rate": 1.6074785394674835e-05,
      "loss": 0.9834,
      "step": 2513
    },
    {
      "epoch": 0.3135836347761008,
      "grad_norm": 0.0,
      "learning_rate": 1.607157569061729e-05,
      "loss": 0.9852,
      "step": 2514
    },
    {
      "epoch": 0.313708369714357,
      "grad_norm": 0.0,
      "learning_rate": 1.6068364995525522e-05,
      "loss": 1.0413,
      "step": 2515
    },
    {
      "epoch": 0.3138331046526132,
      "grad_norm": 0.0,
      "learning_rate": 1.6065153309923607e-05,
      "loss": 1.0214,
      "step": 2516
    },
    {
      "epoch": 0.3139578395908694,
      "grad_norm": 0.0,
      "learning_rate": 1.606194063433577e-05,
      "loss": 0.9851,
      "step": 2517
    },
    {
      "epoch": 0.31408257452912564,
      "grad_norm": 0.0,
      "learning_rate": 1.60587269692864e-05,
      "loss": 0.9813,
      "step": 2518
    },
    {
      "epoch": 0.3142073094673818,
      "grad_norm": 0.0,
      "learning_rate": 1.605551231530005e-05,
      "loss": 0.9957,
      "step": 2519
    },
    {
      "epoch": 0.314332044405638,
      "grad_norm": 0.0,
      "learning_rate": 1.6052296672901433e-05,
      "loss": 1.0111,
      "step": 2520
    },
    {
      "epoch": 0.3144567793438942,
      "grad_norm": 0.0,
      "learning_rate": 1.6049080042615417e-05,
      "loss": 0.9929,
      "step": 2521
    },
    {
      "epoch": 0.3145815142821504,
      "grad_norm": 0.0,
      "learning_rate": 1.6045862424967046e-05,
      "loss": 0.9869,
      "step": 2522
    },
    {
      "epoch": 0.31470624922040663,
      "grad_norm": 0.0,
      "learning_rate": 1.604264382048151e-05,
      "loss": 0.9884,
      "step": 2523
    },
    {
      "epoch": 0.31483098415866284,
      "grad_norm": 0.0,
      "learning_rate": 1.6039424229684168e-05,
      "loss": 1.0231,
      "step": 2524
    },
    {
      "epoch": 0.31495571909691905,
      "grad_norm": 0.0,
      "learning_rate": 1.6036203653100538e-05,
      "loss": 1.0537,
      "step": 2525
    },
    {
      "epoch": 0.31508045403517526,
      "grad_norm": 0.0,
      "learning_rate": 1.6032982091256302e-05,
      "loss": 1.0585,
      "step": 2526
    },
    {
      "epoch": 0.31520518897343147,
      "grad_norm": 0.0,
      "learning_rate": 1.6029759544677298e-05,
      "loss": 1.0329,
      "step": 2527
    },
    {
      "epoch": 0.3153299239116877,
      "grad_norm": 0.0,
      "learning_rate": 1.6026536013889526e-05,
      "loss": 1.0321,
      "step": 2528
    },
    {
      "epoch": 0.3154546588499439,
      "grad_norm": 0.0,
      "learning_rate": 1.602331149941915e-05,
      "loss": 1.0478,
      "step": 2529
    },
    {
      "epoch": 0.3155793937882001,
      "grad_norm": 0.0,
      "learning_rate": 1.602008600179249e-05,
      "loss": 1.0217,
      "step": 2530
    },
    {
      "epoch": 0.31570412872645626,
      "grad_norm": 0.0,
      "learning_rate": 1.6016859521536024e-05,
      "loss": 1.0548,
      "step": 2531
    },
    {
      "epoch": 0.31582886366471247,
      "grad_norm": 0.0,
      "learning_rate": 1.6013632059176405e-05,
      "loss": 1.0491,
      "step": 2532
    },
    {
      "epoch": 0.3159535986029687,
      "grad_norm": 0.0,
      "learning_rate": 1.601040361524043e-05,
      "loss": 0.9926,
      "step": 2533
    },
    {
      "epoch": 0.3160783335412249,
      "grad_norm": 0.0,
      "learning_rate": 1.6007174190255062e-05,
      "loss": 1.0202,
      "step": 2534
    },
    {
      "epoch": 0.3162030684794811,
      "grad_norm": 0.0,
      "learning_rate": 1.600394378474743e-05,
      "loss": 0.981,
      "step": 2535
    },
    {
      "epoch": 0.3163278034177373,
      "grad_norm": 0.0,
      "learning_rate": 1.6000712399244813e-05,
      "loss": 1.0019,
      "step": 2536
    },
    {
      "epoch": 0.3164525383559935,
      "grad_norm": 0.0,
      "learning_rate": 1.5997480034274652e-05,
      "loss": 0.9475,
      "step": 2537
    },
    {
      "epoch": 0.31657727329424973,
      "grad_norm": 0.0,
      "learning_rate": 1.5994246690364557e-05,
      "loss": 1.0027,
      "step": 2538
    },
    {
      "epoch": 0.31670200823250594,
      "grad_norm": 0.0,
      "learning_rate": 1.5991012368042286e-05,
      "loss": 1.0111,
      "step": 2539
    },
    {
      "epoch": 0.31682674317076215,
      "grad_norm": 0.0,
      "learning_rate": 1.5987777067835762e-05,
      "loss": 1.0557,
      "step": 2540
    },
    {
      "epoch": 0.31695147810901836,
      "grad_norm": 0.0,
      "learning_rate": 1.598454079027307e-05,
      "loss": 0.9889,
      "step": 2541
    },
    {
      "epoch": 0.31707621304727457,
      "grad_norm": 0.0,
      "learning_rate": 1.598130353588246e-05,
      "loss": 1.0487,
      "step": 2542
    },
    {
      "epoch": 0.3172009479855307,
      "grad_norm": 0.0,
      "learning_rate": 1.5978065305192316e-05,
      "loss": 1.0004,
      "step": 2543
    },
    {
      "epoch": 0.31732568292378693,
      "grad_norm": 0.0,
      "learning_rate": 1.5974826098731213e-05,
      "loss": 0.9892,
      "step": 2544
    },
    {
      "epoch": 0.31745041786204314,
      "grad_norm": 0.0,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.9392,
      "step": 2545
    },
    {
      "epoch": 0.31757515280029935,
      "grad_norm": 0.0,
      "learning_rate": 1.596834476061115e-05,
      "loss": 1.0153,
      "step": 2546
    },
    {
      "epoch": 0.31769988773855556,
      "grad_norm": 0.0,
      "learning_rate": 1.5965102630010112e-05,
      "loss": 0.9817,
      "step": 2547
    },
    {
      "epoch": 0.3178246226768118,
      "grad_norm": 0.0,
      "learning_rate": 1.596185952575394e-05,
      "loss": 0.9983,
      "step": 2548
    },
    {
      "epoch": 0.317949357615068,
      "grad_norm": 0.0,
      "learning_rate": 1.5958615448372003e-05,
      "loss": 1.0619,
      "step": 2549
    },
    {
      "epoch": 0.3180740925533242,
      "grad_norm": 0.0,
      "learning_rate": 1.5955370398393806e-05,
      "loss": 1.0442,
      "step": 2550
    },
    {
      "epoch": 0.3181988274915804,
      "grad_norm": 0.0,
      "learning_rate": 1.5952124376349025e-05,
      "loss": 1.0214,
      "step": 2551
    },
    {
      "epoch": 0.3183235624298366,
      "grad_norm": 0.0,
      "learning_rate": 1.5948877382767497e-05,
      "loss": 0.9828,
      "step": 2552
    },
    {
      "epoch": 0.3184482973680928,
      "grad_norm": 0.0,
      "learning_rate": 1.5945629418179205e-05,
      "loss": 1.0358,
      "step": 2553
    },
    {
      "epoch": 0.31857303230634904,
      "grad_norm": 0.0,
      "learning_rate": 1.5942380483114305e-05,
      "loss": 1.0445,
      "step": 2554
    },
    {
      "epoch": 0.3186977672446052,
      "grad_norm": 0.0,
      "learning_rate": 1.5939130578103104e-05,
      "loss": 1.069,
      "step": 2555
    },
    {
      "epoch": 0.3188225021828614,
      "grad_norm": 0.0,
      "learning_rate": 1.5935879703676067e-05,
      "loss": 1.0102,
      "step": 2556
    },
    {
      "epoch": 0.3189472371211176,
      "grad_norm": 0.0,
      "learning_rate": 1.5932627860363824e-05,
      "loss": 0.9692,
      "step": 2557
    },
    {
      "epoch": 0.3190719720593738,
      "grad_norm": 0.0,
      "learning_rate": 1.592937504869715e-05,
      "loss": 1.0272,
      "step": 2558
    },
    {
      "epoch": 0.31919670699763003,
      "grad_norm": 0.0,
      "learning_rate": 1.592612126920699e-05,
      "loss": 1.0056,
      "step": 2559
    },
    {
      "epoch": 0.31932144193588624,
      "grad_norm": 0.0,
      "learning_rate": 1.592286652242444e-05,
      "loss": 1.007,
      "step": 2560
    },
    {
      "epoch": 0.31944617687414245,
      "grad_norm": 0.0,
      "learning_rate": 1.591961080888076e-05,
      "loss": 1.0677,
      "step": 2561
    },
    {
      "epoch": 0.31957091181239866,
      "grad_norm": 0.0,
      "learning_rate": 1.5916354129107364e-05,
      "loss": 1.0062,
      "step": 2562
    },
    {
      "epoch": 0.31969564675065487,
      "grad_norm": 0.0,
      "learning_rate": 1.5913096483635827e-05,
      "loss": 0.9742,
      "step": 2563
    },
    {
      "epoch": 0.3198203816889111,
      "grad_norm": 0.0,
      "learning_rate": 1.590983787299787e-05,
      "loss": 1.0364,
      "step": 2564
    },
    {
      "epoch": 0.3199451166271673,
      "grad_norm": 0.0,
      "learning_rate": 1.590657829772539e-05,
      "loss": 1.0192,
      "step": 2565
    },
    {
      "epoch": 0.3200698515654235,
      "grad_norm": 0.0,
      "learning_rate": 1.5903317758350424e-05,
      "loss": 1.0484,
      "step": 2566
    },
    {
      "epoch": 0.32019458650367966,
      "grad_norm": 0.0,
      "learning_rate": 1.5900056255405184e-05,
      "loss": 1.0025,
      "step": 2567
    },
    {
      "epoch": 0.32031932144193587,
      "grad_norm": 0.0,
      "learning_rate": 1.589679378942202e-05,
      "loss": 0.9388,
      "step": 2568
    },
    {
      "epoch": 0.3204440563801921,
      "grad_norm": 0.0,
      "learning_rate": 1.589353036093345e-05,
      "loss": 1.0127,
      "step": 2569
    },
    {
      "epoch": 0.3205687913184483,
      "grad_norm": 0.0,
      "learning_rate": 1.5890265970472153e-05,
      "loss": 1.0195,
      "step": 2570
    },
    {
      "epoch": 0.3206935262567045,
      "grad_norm": 0.0,
      "learning_rate": 1.5887000618570955e-05,
      "loss": 1.0047,
      "step": 2571
    },
    {
      "epoch": 0.3208182611949607,
      "grad_norm": 0.0,
      "learning_rate": 1.5883734305762846e-05,
      "loss": 1.0053,
      "step": 2572
    },
    {
      "epoch": 0.3209429961332169,
      "grad_norm": 0.0,
      "learning_rate": 1.5880467032580974e-05,
      "loss": 0.9613,
      "step": 2573
    },
    {
      "epoch": 0.32106773107147313,
      "grad_norm": 0.0,
      "learning_rate": 1.587719879955863e-05,
      "loss": 1.0245,
      "step": 2574
    },
    {
      "epoch": 0.32119246600972934,
      "grad_norm": 0.0,
      "learning_rate": 1.5873929607229283e-05,
      "loss": 1.0025,
      "step": 2575
    },
    {
      "epoch": 0.32131720094798555,
      "grad_norm": 0.0,
      "learning_rate": 1.5870659456126542e-05,
      "loss": 1.0279,
      "step": 2576
    },
    {
      "epoch": 0.32144193588624176,
      "grad_norm": 0.0,
      "learning_rate": 1.586738834678418e-05,
      "loss": 1.0534,
      "step": 2577
    },
    {
      "epoch": 0.32156667082449797,
      "grad_norm": 0.0,
      "learning_rate": 1.5864116279736124e-05,
      "loss": 1.031,
      "step": 2578
    },
    {
      "epoch": 0.3216914057627541,
      "grad_norm": 0.0,
      "learning_rate": 1.5860843255516457e-05,
      "loss": 1.0364,
      "step": 2579
    },
    {
      "epoch": 0.32181614070101033,
      "grad_norm": 0.0,
      "learning_rate": 1.585756927465942e-05,
      "loss": 1.0211,
      "step": 2580
    },
    {
      "epoch": 0.32194087563926654,
      "grad_norm": 0.0,
      "learning_rate": 1.5854294337699407e-05,
      "loss": 1.0328,
      "step": 2581
    },
    {
      "epoch": 0.32206561057752275,
      "grad_norm": 0.0,
      "learning_rate": 1.5851018445170973e-05,
      "loss": 1.0587,
      "step": 2582
    },
    {
      "epoch": 0.32219034551577896,
      "grad_norm": 0.0,
      "learning_rate": 1.5847741597608825e-05,
      "loss": 1.0187,
      "step": 2583
    },
    {
      "epoch": 0.3223150804540352,
      "grad_norm": 0.0,
      "learning_rate": 1.584446379554783e-05,
      "loss": 0.9978,
      "step": 2584
    },
    {
      "epoch": 0.3224398153922914,
      "grad_norm": 0.0,
      "learning_rate": 1.5841185039523004e-05,
      "loss": 1.0153,
      "step": 2585
    },
    {
      "epoch": 0.3225645503305476,
      "grad_norm": 0.0,
      "learning_rate": 1.5837905330069525e-05,
      "loss": 1.0243,
      "step": 2586
    },
    {
      "epoch": 0.3226892852688038,
      "grad_norm": 0.0,
      "learning_rate": 1.583462466772272e-05,
      "loss": 1.0225,
      "step": 2587
    },
    {
      "epoch": 0.32281402020706,
      "grad_norm": 0.0,
      "learning_rate": 1.5831343053018082e-05,
      "loss": 1.0413,
      "step": 2588
    },
    {
      "epoch": 0.3229387551453162,
      "grad_norm": 0.0,
      "learning_rate": 1.582806048649125e-05,
      "loss": 1.0011,
      "step": 2589
    },
    {
      "epoch": 0.32306349008357244,
      "grad_norm": 0.0,
      "learning_rate": 1.5824776968678024e-05,
      "loss": 0.9667,
      "step": 2590
    },
    {
      "epoch": 0.3231882250218286,
      "grad_norm": 0.0,
      "learning_rate": 1.5821492500114353e-05,
      "loss": 0.975,
      "step": 2591
    },
    {
      "epoch": 0.3233129599600848,
      "grad_norm": 0.0,
      "learning_rate": 1.5818207081336345e-05,
      "loss": 1.0046,
      "step": 2592
    },
    {
      "epoch": 0.323437694898341,
      "grad_norm": 0.0,
      "learning_rate": 1.5814920712880267e-05,
      "loss": 1.0814,
      "step": 2593
    },
    {
      "epoch": 0.3235624298365972,
      "grad_norm": 0.0,
      "learning_rate": 1.5811633395282538e-05,
      "loss": 0.9965,
      "step": 2594
    },
    {
      "epoch": 0.32368716477485343,
      "grad_norm": 0.0,
      "learning_rate": 1.5808345129079726e-05,
      "loss": 0.9569,
      "step": 2595
    },
    {
      "epoch": 0.32381189971310964,
      "grad_norm": 0.0,
      "learning_rate": 1.580505591480856e-05,
      "loss": 1.0159,
      "step": 2596
    },
    {
      "epoch": 0.32393663465136585,
      "grad_norm": 0.0,
      "learning_rate": 1.5801765753005923e-05,
      "loss": 1.0349,
      "step": 2597
    },
    {
      "epoch": 0.32406136958962206,
      "grad_norm": 0.0,
      "learning_rate": 1.579847464420886e-05,
      "loss": 1.0623,
      "step": 2598
    },
    {
      "epoch": 0.32418610452787827,
      "grad_norm": 0.0,
      "learning_rate": 1.5795182588954553e-05,
      "loss": 1.0163,
      "step": 2599
    },
    {
      "epoch": 0.3243108394661345,
      "grad_norm": 0.0,
      "learning_rate": 1.5791889587780347e-05,
      "loss": 1.0243,
      "step": 2600
    },
    {
      "epoch": 0.3244355744043907,
      "grad_norm": 0.0,
      "learning_rate": 1.578859564122375e-05,
      "loss": 1.0512,
      "step": 2601
    },
    {
      "epoch": 0.3245603093426469,
      "grad_norm": 0.0,
      "learning_rate": 1.578530074982242e-05,
      "loss": 0.9724,
      "step": 2602
    },
    {
      "epoch": 0.32468504428090306,
      "grad_norm": 0.0,
      "learning_rate": 1.5782004914114155e-05,
      "loss": 1.0165,
      "step": 2603
    },
    {
      "epoch": 0.32480977921915927,
      "grad_norm": 0.0,
      "learning_rate": 1.5778708134636928e-05,
      "loss": 0.9998,
      "step": 2604
    },
    {
      "epoch": 0.3249345141574155,
      "grad_norm": 0.0,
      "learning_rate": 1.577541041192885e-05,
      "loss": 1.0343,
      "step": 2605
    },
    {
      "epoch": 0.3250592490956717,
      "grad_norm": 0.0,
      "learning_rate": 1.57721117465282e-05,
      "loss": 1.0035,
      "step": 2606
    },
    {
      "epoch": 0.3251839840339279,
      "grad_norm": 0.0,
      "learning_rate": 1.5768812138973394e-05,
      "loss": 1.0304,
      "step": 2607
    },
    {
      "epoch": 0.3253087189721841,
      "grad_norm": 0.0,
      "learning_rate": 1.576551158980302e-05,
      "loss": 1.022,
      "step": 2608
    },
    {
      "epoch": 0.3254334539104403,
      "grad_norm": 0.0,
      "learning_rate": 1.5762210099555804e-05,
      "loss": 1.0522,
      "step": 2609
    },
    {
      "epoch": 0.32555818884869653,
      "grad_norm": 0.0,
      "learning_rate": 1.5758907668770637e-05,
      "loss": 1.0108,
      "step": 2610
    },
    {
      "epoch": 0.32568292378695274,
      "grad_norm": 0.0,
      "learning_rate": 1.5755604297986556e-05,
      "loss": 1.0151,
      "step": 2611
    },
    {
      "epoch": 0.32580765872520895,
      "grad_norm": 0.0,
      "learning_rate": 1.5752299987742757e-05,
      "loss": 1.0331,
      "step": 2612
    },
    {
      "epoch": 0.32593239366346516,
      "grad_norm": 0.0,
      "learning_rate": 1.5748994738578582e-05,
      "loss": 0.9443,
      "step": 2613
    },
    {
      "epoch": 0.32605712860172137,
      "grad_norm": 0.0,
      "learning_rate": 1.5745688551033538e-05,
      "loss": 0.9856,
      "step": 2614
    },
    {
      "epoch": 0.3261818635399775,
      "grad_norm": 0.0,
      "learning_rate": 1.574238142564727e-05,
      "loss": 1.0215,
      "step": 2615
    },
    {
      "epoch": 0.32630659847823373,
      "grad_norm": 0.0,
      "learning_rate": 1.5739073362959594e-05,
      "loss": 0.9971,
      "step": 2616
    },
    {
      "epoch": 0.32643133341648994,
      "grad_norm": 0.0,
      "learning_rate": 1.573576436351046e-05,
      "loss": 1.0112,
      "step": 2617
    },
    {
      "epoch": 0.32655606835474615,
      "grad_norm": 0.0,
      "learning_rate": 1.573245442783999e-05,
      "loss": 0.9812,
      "step": 2618
    },
    {
      "epoch": 0.32668080329300236,
      "grad_norm": 0.0,
      "learning_rate": 1.5729143556488438e-05,
      "loss": 1.0275,
      "step": 2619
    },
    {
      "epoch": 0.3268055382312586,
      "grad_norm": 0.0,
      "learning_rate": 1.572583174999623e-05,
      "loss": 1.0038,
      "step": 2620
    },
    {
      "epoch": 0.3269302731695148,
      "grad_norm": 0.0,
      "learning_rate": 1.572251900890393e-05,
      "loss": 1.007,
      "step": 2621
    },
    {
      "epoch": 0.327055008107771,
      "grad_norm": 0.0,
      "learning_rate": 1.5719205333752263e-05,
      "loss": 0.9774,
      "step": 2622
    },
    {
      "epoch": 0.3271797430460272,
      "grad_norm": 0.0,
      "learning_rate": 1.5715890725082106e-05,
      "loss": 1.0112,
      "step": 2623
    },
    {
      "epoch": 0.3273044779842834,
      "grad_norm": 0.0,
      "learning_rate": 1.5712575183434486e-05,
      "loss": 0.9699,
      "step": 2624
    },
    {
      "epoch": 0.3274292129225396,
      "grad_norm": 0.0,
      "learning_rate": 1.5709258709350585e-05,
      "loss": 0.9986,
      "step": 2625
    },
    {
      "epoch": 0.32755394786079584,
      "grad_norm": 0.0,
      "learning_rate": 1.570594130337173e-05,
      "loss": 0.9975,
      "step": 2626
    },
    {
      "epoch": 0.327678682799052,
      "grad_norm": 0.0,
      "learning_rate": 1.570262296603941e-05,
      "loss": 1.0167,
      "step": 2627
    },
    {
      "epoch": 0.3278034177373082,
      "grad_norm": 0.0,
      "learning_rate": 1.569930369789526e-05,
      "loss": 1.0209,
      "step": 2628
    },
    {
      "epoch": 0.3279281526755644,
      "grad_norm": 0.0,
      "learning_rate": 1.5695983499481065e-05,
      "loss": 0.9849,
      "step": 2629
    },
    {
      "epoch": 0.3280528876138206,
      "grad_norm": 0.0,
      "learning_rate": 1.5692662371338772e-05,
      "loss": 0.9992,
      "step": 2630
    },
    {
      "epoch": 0.32817762255207683,
      "grad_norm": 0.0,
      "learning_rate": 1.5689340314010467e-05,
      "loss": 1.0177,
      "step": 2631
    },
    {
      "epoch": 0.32830235749033304,
      "grad_norm": 0.0,
      "learning_rate": 1.5686017328038393e-05,
      "loss": 0.983,
      "step": 2632
    },
    {
      "epoch": 0.32842709242858925,
      "grad_norm": 0.0,
      "learning_rate": 1.568269341396495e-05,
      "loss": 0.985,
      "step": 2633
    },
    {
      "epoch": 0.32855182736684546,
      "grad_norm": 0.0,
      "learning_rate": 1.567936857233268e-05,
      "loss": 0.9748,
      "step": 2634
    },
    {
      "epoch": 0.32867656230510167,
      "grad_norm": 0.0,
      "learning_rate": 1.567604280368429e-05,
      "loss": 0.9637,
      "step": 2635
    },
    {
      "epoch": 0.3288012972433579,
      "grad_norm": 0.0,
      "learning_rate": 1.5672716108562614e-05,
      "loss": 1.0336,
      "step": 2636
    },
    {
      "epoch": 0.3289260321816141,
      "grad_norm": 0.0,
      "learning_rate": 1.5669388487510663e-05,
      "loss": 1.0201,
      "step": 2637
    },
    {
      "epoch": 0.3290507671198703,
      "grad_norm": 0.0,
      "learning_rate": 1.566605994107159e-05,
      "loss": 0.9937,
      "step": 2638
    },
    {
      "epoch": 0.32917550205812646,
      "grad_norm": 0.0,
      "learning_rate": 1.5662730469788688e-05,
      "loss": 1.0079,
      "step": 2639
    },
    {
      "epoch": 0.32930023699638267,
      "grad_norm": 0.0,
      "learning_rate": 1.565940007420542e-05,
      "loss": 1.0063,
      "step": 2640
    },
    {
      "epoch": 0.3294249719346389,
      "grad_norm": 0.0,
      "learning_rate": 1.5656068754865388e-05,
      "loss": 1.0341,
      "step": 2641
    },
    {
      "epoch": 0.3295497068728951,
      "grad_norm": 0.0,
      "learning_rate": 1.5652736512312345e-05,
      "loss": 1.0027,
      "step": 2642
    },
    {
      "epoch": 0.3296744418111513,
      "grad_norm": 0.0,
      "learning_rate": 1.5649403347090196e-05,
      "loss": 1.0074,
      "step": 2643
    },
    {
      "epoch": 0.3297991767494075,
      "grad_norm": 0.0,
      "learning_rate": 1.5646069259743007e-05,
      "loss": 1.0157,
      "step": 2644
    },
    {
      "epoch": 0.3299239116876637,
      "grad_norm": 0.0,
      "learning_rate": 1.564273425081497e-05,
      "loss": 1.0167,
      "step": 2645
    },
    {
      "epoch": 0.33004864662591993,
      "grad_norm": 0.0,
      "learning_rate": 1.563939832085046e-05,
      "loss": 0.9613,
      "step": 2646
    },
    {
      "epoch": 0.33017338156417614,
      "grad_norm": 0.0,
      "learning_rate": 1.5636061470393968e-05,
      "loss": 1.0092,
      "step": 2647
    },
    {
      "epoch": 0.33029811650243235,
      "grad_norm": 0.0,
      "learning_rate": 1.5632723699990165e-05,
      "loss": 0.9869,
      "step": 2648
    },
    {
      "epoch": 0.33042285144068856,
      "grad_norm": 0.0,
      "learning_rate": 1.5629385010183852e-05,
      "loss": 0.9851,
      "step": 2649
    },
    {
      "epoch": 0.33054758637894477,
      "grad_norm": 0.0,
      "learning_rate": 1.5626045401519986e-05,
      "loss": 0.9859,
      "step": 2650
    },
    {
      "epoch": 0.3306723213172009,
      "grad_norm": 0.0,
      "learning_rate": 1.562270487454368e-05,
      "loss": 0.9885,
      "step": 2651
    },
    {
      "epoch": 0.33079705625545713,
      "grad_norm": 0.0,
      "learning_rate": 1.5619363429800195e-05,
      "loss": 1.0053,
      "step": 2652
    },
    {
      "epoch": 0.33092179119371334,
      "grad_norm": 0.0,
      "learning_rate": 1.561602106783493e-05,
      "loss": 1.0278,
      "step": 2653
    },
    {
      "epoch": 0.33104652613196955,
      "grad_norm": 0.0,
      "learning_rate": 1.5612677789193452e-05,
      "loss": 1.0602,
      "step": 2654
    },
    {
      "epoch": 0.33117126107022576,
      "grad_norm": 0.0,
      "learning_rate": 1.5609333594421458e-05,
      "loss": 1.0249,
      "step": 2655
    },
    {
      "epoch": 0.331295996008482,
      "grad_norm": 0.0,
      "learning_rate": 1.5605988484064818e-05,
      "loss": 1.0241,
      "step": 2656
    },
    {
      "epoch": 0.3314207309467382,
      "grad_norm": 0.0,
      "learning_rate": 1.5602642458669527e-05,
      "loss": 0.9801,
      "step": 2657
    },
    {
      "epoch": 0.3315454658849944,
      "grad_norm": 0.0,
      "learning_rate": 1.559929551878175e-05,
      "loss": 1.0203,
      "step": 2658
    },
    {
      "epoch": 0.3316702008232506,
      "grad_norm": 0.0,
      "learning_rate": 1.559594766494778e-05,
      "loss": 1.0043,
      "step": 2659
    },
    {
      "epoch": 0.3317949357615068,
      "grad_norm": 0.0,
      "learning_rate": 1.5592598897714078e-05,
      "loss": 0.9899,
      "step": 2660
    },
    {
      "epoch": 0.331919670699763,
      "grad_norm": 0.0,
      "learning_rate": 1.558924921762725e-05,
      "loss": 0.97,
      "step": 2661
    },
    {
      "epoch": 0.33204440563801924,
      "grad_norm": 0.0,
      "learning_rate": 1.5585898625234047e-05,
      "loss": 0.971,
      "step": 2662
    },
    {
      "epoch": 0.3321691405762754,
      "grad_norm": 0.0,
      "learning_rate": 1.5582547121081368e-05,
      "loss": 1.021,
      "step": 2663
    },
    {
      "epoch": 0.3322938755145316,
      "grad_norm": 0.0,
      "learning_rate": 1.5579194705716263e-05,
      "loss": 1.0446,
      "step": 2664
    },
    {
      "epoch": 0.3324186104527878,
      "grad_norm": 0.0,
      "learning_rate": 1.557584137968593e-05,
      "loss": 1.0086,
      "step": 2665
    },
    {
      "epoch": 0.332543345391044,
      "grad_norm": 0.0,
      "learning_rate": 1.5572487143537717e-05,
      "loss": 0.9814,
      "step": 2666
    },
    {
      "epoch": 0.33266808032930023,
      "grad_norm": 0.0,
      "learning_rate": 1.556913199781912e-05,
      "loss": 0.9872,
      "step": 2667
    },
    {
      "epoch": 0.33279281526755644,
      "grad_norm": 0.0,
      "learning_rate": 1.556577594307779e-05,
      "loss": 0.9912,
      "step": 2668
    },
    {
      "epoch": 0.33291755020581265,
      "grad_norm": 0.0,
      "learning_rate": 1.556241897986151e-05,
      "loss": 0.9919,
      "step": 2669
    },
    {
      "epoch": 0.33304228514406886,
      "grad_norm": 0.0,
      "learning_rate": 1.5559061108718225e-05,
      "loss": 0.9407,
      "step": 2670
    },
    {
      "epoch": 0.33316702008232507,
      "grad_norm": 0.0,
      "learning_rate": 1.5555702330196024e-05,
      "loss": 1.0375,
      "step": 2671
    },
    {
      "epoch": 0.3332917550205813,
      "grad_norm": 0.0,
      "learning_rate": 1.5552342644843142e-05,
      "loss": 1.0659,
      "step": 2672
    },
    {
      "epoch": 0.3334164899588375,
      "grad_norm": 0.0,
      "learning_rate": 1.554898205320797e-05,
      "loss": 1.0327,
      "step": 2673
    },
    {
      "epoch": 0.3335412248970937,
      "grad_norm": 0.0,
      "learning_rate": 1.5545620555839036e-05,
      "loss": 0.9462,
      "step": 2674
    },
    {
      "epoch": 0.33366595983534986,
      "grad_norm": 0.0,
      "learning_rate": 1.5542258153285025e-05,
      "loss": 0.9917,
      "step": 2675
    },
    {
      "epoch": 0.33379069477360607,
      "grad_norm": 0.0,
      "learning_rate": 1.553889484609476e-05,
      "loss": 0.9822,
      "step": 2676
    },
    {
      "epoch": 0.3339154297118623,
      "grad_norm": 0.0,
      "learning_rate": 1.5535530634817227e-05,
      "loss": 1.0097,
      "step": 2677
    },
    {
      "epoch": 0.3340401646501185,
      "grad_norm": 0.0,
      "learning_rate": 1.5532165520001537e-05,
      "loss": 1.0045,
      "step": 2678
    },
    {
      "epoch": 0.3341648995883747,
      "grad_norm": 0.0,
      "learning_rate": 1.5528799502196975e-05,
      "loss": 0.9702,
      "step": 2679
    },
    {
      "epoch": 0.3342896345266309,
      "grad_norm": 0.0,
      "learning_rate": 1.552543258195295e-05,
      "loss": 1.0045,
      "step": 2680
    },
    {
      "epoch": 0.3344143694648871,
      "grad_norm": 0.0,
      "learning_rate": 1.5522064759819037e-05,
      "loss": 0.9841,
      "step": 2681
    },
    {
      "epoch": 0.33453910440314333,
      "grad_norm": 0.0,
      "learning_rate": 1.551869603634494e-05,
      "loss": 1.0172,
      "step": 2682
    },
    {
      "epoch": 0.33466383934139954,
      "grad_norm": 0.0,
      "learning_rate": 1.5515326412080527e-05,
      "loss": 1.0014,
      "step": 2683
    },
    {
      "epoch": 0.33478857427965575,
      "grad_norm": 0.0,
      "learning_rate": 1.5511955887575803e-05,
      "loss": 1.0391,
      "step": 2684
    },
    {
      "epoch": 0.33491330921791196,
      "grad_norm": 0.0,
      "learning_rate": 1.5508584463380923e-05,
      "loss": 0.9775,
      "step": 2685
    },
    {
      "epoch": 0.33503804415616817,
      "grad_norm": 0.0,
      "learning_rate": 1.5505212140046186e-05,
      "loss": 1.0322,
      "step": 2686
    },
    {
      "epoch": 0.3351627790944243,
      "grad_norm": 0.0,
      "learning_rate": 1.5501838918122043e-05,
      "loss": 1.007,
      "step": 2687
    },
    {
      "epoch": 0.33528751403268053,
      "grad_norm": 0.0,
      "learning_rate": 1.5498464798159088e-05,
      "loss": 0.9868,
      "step": 2688
    },
    {
      "epoch": 0.33541224897093674,
      "grad_norm": 0.0,
      "learning_rate": 1.5495089780708062e-05,
      "loss": 1.0296,
      "step": 2689
    },
    {
      "epoch": 0.33553698390919295,
      "grad_norm": 0.0,
      "learning_rate": 1.5491713866319852e-05,
      "loss": 1.0114,
      "step": 2690
    },
    {
      "epoch": 0.33566171884744916,
      "grad_norm": 0.0,
      "learning_rate": 1.5488337055545497e-05,
      "loss": 1.0364,
      "step": 2691
    },
    {
      "epoch": 0.3357864537857054,
      "grad_norm": 0.0,
      "learning_rate": 1.5484959348936174e-05,
      "loss": 0.9922,
      "step": 2692
    },
    {
      "epoch": 0.3359111887239616,
      "grad_norm": 0.0,
      "learning_rate": 1.5481580747043208e-05,
      "loss": 1.045,
      "step": 2693
    },
    {
      "epoch": 0.3360359236622178,
      "grad_norm": 0.0,
      "learning_rate": 1.5478201250418074e-05,
      "loss": 0.9512,
      "step": 2694
    },
    {
      "epoch": 0.336160658600474,
      "grad_norm": 0.0,
      "learning_rate": 1.5474820859612394e-05,
      "loss": 0.9585,
      "step": 2695
    },
    {
      "epoch": 0.3362853935387302,
      "grad_norm": 0.0,
      "learning_rate": 1.547143957517793e-05,
      "loss": 1.0041,
      "step": 2696
    },
    {
      "epoch": 0.3364101284769864,
      "grad_norm": 0.0,
      "learning_rate": 1.546805739766659e-05,
      "loss": 1.0173,
      "step": 2697
    },
    {
      "epoch": 0.33653486341524264,
      "grad_norm": 0.0,
      "learning_rate": 1.5464674327630437e-05,
      "loss": 1.031,
      "step": 2698
    },
    {
      "epoch": 0.3366595983534988,
      "grad_norm": 0.0,
      "learning_rate": 1.546129036562167e-05,
      "loss": 1.0128,
      "step": 2699
    },
    {
      "epoch": 0.336784333291755,
      "grad_norm": 0.0,
      "learning_rate": 1.5457905512192634e-05,
      "loss": 0.9932,
      "step": 2700
    },
    {
      "epoch": 0.3369090682300112,
      "grad_norm": 0.0,
      "learning_rate": 1.5454519767895826e-05,
      "loss": 1.0376,
      "step": 2701
    },
    {
      "epoch": 0.3370338031682674,
      "grad_norm": 0.0,
      "learning_rate": 1.5451133133283886e-05,
      "loss": 0.9697,
      "step": 2702
    },
    {
      "epoch": 0.33715853810652363,
      "grad_norm": 0.0,
      "learning_rate": 1.5447745608909595e-05,
      "loss": 0.9579,
      "step": 2703
    },
    {
      "epoch": 0.33728327304477984,
      "grad_norm": 0.0,
      "learning_rate": 1.5444357195325885e-05,
      "loss": 0.9458,
      "step": 2704
    },
    {
      "epoch": 0.33740800798303605,
      "grad_norm": 0.0,
      "learning_rate": 1.5440967893085827e-05,
      "loss": 0.967,
      "step": 2705
    },
    {
      "epoch": 0.33753274292129226,
      "grad_norm": 0.0,
      "learning_rate": 1.5437577702742646e-05,
      "loss": 1.0007,
      "step": 2706
    },
    {
      "epoch": 0.33765747785954847,
      "grad_norm": 0.0,
      "learning_rate": 1.54341866248497e-05,
      "loss": 1.0336,
      "step": 2707
    },
    {
      "epoch": 0.3377822127978047,
      "grad_norm": 0.0,
      "learning_rate": 1.5430794659960505e-05,
      "loss": 1.0902,
      "step": 2708
    },
    {
      "epoch": 0.3379069477360609,
      "grad_norm": 0.0,
      "learning_rate": 1.5427401808628714e-05,
      "loss": 0.964,
      "step": 2709
    },
    {
      "epoch": 0.3380316826743171,
      "grad_norm": 0.0,
      "learning_rate": 1.542400807140812e-05,
      "loss": 0.9523,
      "step": 2710
    },
    {
      "epoch": 0.33815641761257326,
      "grad_norm": 0.0,
      "learning_rate": 1.5420613448852667e-05,
      "loss": 0.9623,
      "step": 2711
    },
    {
      "epoch": 0.33828115255082947,
      "grad_norm": 0.0,
      "learning_rate": 1.541721794151645e-05,
      "loss": 0.9673,
      "step": 2712
    },
    {
      "epoch": 0.3384058874890857,
      "grad_norm": 0.0,
      "learning_rate": 1.5413821549953697e-05,
      "loss": 1.0096,
      "step": 2713
    },
    {
      "epoch": 0.3385306224273419,
      "grad_norm": 0.0,
      "learning_rate": 1.541042427471879e-05,
      "loss": 0.9572,
      "step": 2714
    },
    {
      "epoch": 0.3386553573655981,
      "grad_norm": 0.0,
      "learning_rate": 1.5407026116366238e-05,
      "loss": 1.0304,
      "step": 2715
    },
    {
      "epoch": 0.3387800923038543,
      "grad_norm": 0.0,
      "learning_rate": 1.5403627075450717e-05,
      "loss": 1.0092,
      "step": 2716
    },
    {
      "epoch": 0.3389048272421105,
      "grad_norm": 0.0,
      "learning_rate": 1.540022715252703e-05,
      "loss": 1.0069,
      "step": 2717
    },
    {
      "epoch": 0.33902956218036673,
      "grad_norm": 0.0,
      "learning_rate": 1.5396826348150134e-05,
      "loss": 0.9823,
      "step": 2718
    },
    {
      "epoch": 0.33915429711862294,
      "grad_norm": 0.0,
      "learning_rate": 1.539342466287513e-05,
      "loss": 1.0404,
      "step": 2719
    },
    {
      "epoch": 0.33927903205687915,
      "grad_norm": 0.0,
      "learning_rate": 1.5390022097257248e-05,
      "loss": 1.0079,
      "step": 2720
    },
    {
      "epoch": 0.33940376699513536,
      "grad_norm": 0.0,
      "learning_rate": 1.538661865185188e-05,
      "loss": 0.9925,
      "step": 2721
    },
    {
      "epoch": 0.33952850193339157,
      "grad_norm": 0.0,
      "learning_rate": 1.538321432721455e-05,
      "loss": 1.0124,
      "step": 2722
    },
    {
      "epoch": 0.3396532368716477,
      "grad_norm": 0.0,
      "learning_rate": 1.5379809123900936e-05,
      "loss": 1.035,
      "step": 2723
    },
    {
      "epoch": 0.33977797180990393,
      "grad_norm": 0.0,
      "learning_rate": 1.5376403042466846e-05,
      "loss": 1.0668,
      "step": 2724
    },
    {
      "epoch": 0.33990270674816014,
      "grad_norm": 0.0,
      "learning_rate": 1.5372996083468242e-05,
      "loss": 1.0162,
      "step": 2725
    },
    {
      "epoch": 0.34002744168641635,
      "grad_norm": 0.0,
      "learning_rate": 1.5369588247461224e-05,
      "loss": 0.9666,
      "step": 2726
    },
    {
      "epoch": 0.34015217662467256,
      "grad_norm": 0.0,
      "learning_rate": 1.5366179535002033e-05,
      "loss": 1.0114,
      "step": 2727
    },
    {
      "epoch": 0.3402769115629288,
      "grad_norm": 0.0,
      "learning_rate": 1.5362769946647068e-05,
      "loss": 0.9696,
      "step": 2728
    },
    {
      "epoch": 0.340401646501185,
      "grad_norm": 0.0,
      "learning_rate": 1.535935948295285e-05,
      "loss": 1.0251,
      "step": 2729
    },
    {
      "epoch": 0.3405263814394412,
      "grad_norm": 0.0,
      "learning_rate": 1.5355948144476056e-05,
      "loss": 0.943,
      "step": 2730
    },
    {
      "epoch": 0.3406511163776974,
      "grad_norm": 0.0,
      "learning_rate": 1.5352535931773506e-05,
      "loss": 0.9749,
      "step": 2731
    },
    {
      "epoch": 0.3407758513159536,
      "grad_norm": 0.0,
      "learning_rate": 1.534912284540215e-05,
      "loss": 0.9967,
      "step": 2732
    },
    {
      "epoch": 0.3409005862542098,
      "grad_norm": 0.0,
      "learning_rate": 1.5345708885919095e-05,
      "loss": 1.0092,
      "step": 2733
    },
    {
      "epoch": 0.34102532119246604,
      "grad_norm": 0.0,
      "learning_rate": 1.534229405388159e-05,
      "loss": 1.0158,
      "step": 2734
    },
    {
      "epoch": 0.3411500561307222,
      "grad_norm": 0.0,
      "learning_rate": 1.533887834984701e-05,
      "loss": 0.9763,
      "step": 2735
    },
    {
      "epoch": 0.3412747910689784,
      "grad_norm": 0.0,
      "learning_rate": 1.5335461774372898e-05,
      "loss": 1.0088,
      "step": 2736
    },
    {
      "epoch": 0.3413995260072346,
      "grad_norm": 0.0,
      "learning_rate": 1.5332044328016916e-05,
      "loss": 0.9853,
      "step": 2737
    },
    {
      "epoch": 0.3415242609454908,
      "grad_norm": 0.0,
      "learning_rate": 1.5328626011336874e-05,
      "loss": 1.0425,
      "step": 2738
    },
    {
      "epoch": 0.34164899588374703,
      "grad_norm": 0.0,
      "learning_rate": 1.532520682489074e-05,
      "loss": 1.0607,
      "step": 2739
    },
    {
      "epoch": 0.34177373082200324,
      "grad_norm": 0.0,
      "learning_rate": 1.5321786769236604e-05,
      "loss": 1.0344,
      "step": 2740
    },
    {
      "epoch": 0.34189846576025945,
      "grad_norm": 0.0,
      "learning_rate": 1.5318365844932702e-05,
      "loss": 1.0292,
      "step": 2741
    },
    {
      "epoch": 0.34202320069851566,
      "grad_norm": 0.0,
      "learning_rate": 1.531494405253742e-05,
      "loss": 0.9895,
      "step": 2742
    },
    {
      "epoch": 0.34214793563677187,
      "grad_norm": 0.0,
      "learning_rate": 1.5311521392609283e-05,
      "loss": 1.0076,
      "step": 2743
    },
    {
      "epoch": 0.3422726705750281,
      "grad_norm": 0.0,
      "learning_rate": 1.530809786570695e-05,
      "loss": 1.0054,
      "step": 2744
    },
    {
      "epoch": 0.3423974055132843,
      "grad_norm": 0.0,
      "learning_rate": 1.5304673472389227e-05,
      "loss": 1.014,
      "step": 2745
    },
    {
      "epoch": 0.3425221404515405,
      "grad_norm": 0.0,
      "learning_rate": 1.5301248213215065e-05,
      "loss": 1.0007,
      "step": 2746
    },
    {
      "epoch": 0.34264687538979666,
      "grad_norm": 0.0,
      "learning_rate": 1.5297822088743552e-05,
      "loss": 0.9896,
      "step": 2747
    },
    {
      "epoch": 0.34277161032805287,
      "grad_norm": 0.0,
      "learning_rate": 1.5294395099533913e-05,
      "loss": 1.0277,
      "step": 2748
    },
    {
      "epoch": 0.3428963452663091,
      "grad_norm": 0.0,
      "learning_rate": 1.5290967246145528e-05,
      "loss": 0.9899,
      "step": 2749
    },
    {
      "epoch": 0.3430210802045653,
      "grad_norm": 0.0,
      "learning_rate": 1.52875385291379e-05,
      "loss": 1.0151,
      "step": 2750
    },
    {
      "epoch": 0.3431458151428215,
      "grad_norm": 0.0,
      "learning_rate": 1.5284108949070687e-05,
      "loss": 0.9973,
      "step": 2751
    },
    {
      "epoch": 0.3432705500810777,
      "grad_norm": 0.0,
      "learning_rate": 1.528067850650368e-05,
      "loss": 0.9389,
      "step": 2752
    },
    {
      "epoch": 0.3433952850193339,
      "grad_norm": 0.0,
      "learning_rate": 1.5277247201996818e-05,
      "loss": 0.9974,
      "step": 2753
    },
    {
      "epoch": 0.34352001995759013,
      "grad_norm": 0.0,
      "learning_rate": 1.5273815036110177e-05,
      "loss": 1.0054,
      "step": 2754
    },
    {
      "epoch": 0.34364475489584634,
      "grad_norm": 0.0,
      "learning_rate": 1.5270382009403963e-05,
      "loss": 0.9171,
      "step": 2755
    },
    {
      "epoch": 0.34376948983410255,
      "grad_norm": 0.0,
      "learning_rate": 1.5266948122438544e-05,
      "loss": 1.0608,
      "step": 2756
    },
    {
      "epoch": 0.34389422477235876,
      "grad_norm": 0.0,
      "learning_rate": 1.5263513375774413e-05,
      "loss": 1.0108,
      "step": 2757
    },
    {
      "epoch": 0.34401895971061497,
      "grad_norm": 0.0,
      "learning_rate": 1.5260077769972204e-05,
      "loss": 0.9701,
      "step": 2758
    },
    {
      "epoch": 0.3441436946488711,
      "grad_norm": 0.0,
      "learning_rate": 1.52566413055927e-05,
      "loss": 0.9974,
      "step": 2759
    },
    {
      "epoch": 0.34426842958712733,
      "grad_norm": 0.0,
      "learning_rate": 1.5253203983196812e-05,
      "loss": 1.0154,
      "step": 2760
    },
    {
      "epoch": 0.34439316452538354,
      "grad_norm": 0.0,
      "learning_rate": 1.5249765803345602e-05,
      "loss": 1.0549,
      "step": 2761
    },
    {
      "epoch": 0.34451789946363975,
      "grad_norm": 0.0,
      "learning_rate": 1.524632676660027e-05,
      "loss": 1.0566,
      "step": 2762
    },
    {
      "epoch": 0.34464263440189596,
      "grad_norm": 0.0,
      "learning_rate": 1.5242886873522149e-05,
      "loss": 0.9954,
      "step": 2763
    },
    {
      "epoch": 0.3447673693401522,
      "grad_norm": 0.0,
      "learning_rate": 1.5239446124672717e-05,
      "loss": 0.9936,
      "step": 2764
    },
    {
      "epoch": 0.3448921042784084,
      "grad_norm": 0.0,
      "learning_rate": 1.5236004520613592e-05,
      "loss": 0.975,
      "step": 2765
    },
    {
      "epoch": 0.3450168392166646,
      "grad_norm": 0.0,
      "learning_rate": 1.5232562061906533e-05,
      "loss": 0.9909,
      "step": 2766
    },
    {
      "epoch": 0.3451415741549208,
      "grad_norm": 0.0,
      "learning_rate": 1.5229118749113432e-05,
      "loss": 1.08,
      "step": 2767
    },
    {
      "epoch": 0.345266309093177,
      "grad_norm": 0.0,
      "learning_rate": 1.5225674582796321e-05,
      "loss": 0.9894,
      "step": 2768
    },
    {
      "epoch": 0.3453910440314332,
      "grad_norm": 0.0,
      "learning_rate": 1.5222229563517385e-05,
      "loss": 1.0201,
      "step": 2769
    },
    {
      "epoch": 0.34551577896968944,
      "grad_norm": 0.0,
      "learning_rate": 1.5218783691838935e-05,
      "loss": 1.0026,
      "step": 2770
    },
    {
      "epoch": 0.3456405139079456,
      "grad_norm": 0.0,
      "learning_rate": 1.5215336968323415e-05,
      "loss": 0.9778,
      "step": 2771
    },
    {
      "epoch": 0.3457652488462018,
      "grad_norm": 0.0,
      "learning_rate": 1.5211889393533427e-05,
      "loss": 0.9991,
      "step": 2772
    },
    {
      "epoch": 0.345889983784458,
      "grad_norm": 0.0,
      "learning_rate": 1.52084409680317e-05,
      "loss": 1.0153,
      "step": 2773
    },
    {
      "epoch": 0.3460147187227142,
      "grad_norm": 0.0,
      "learning_rate": 1.5204991692381099e-05,
      "loss": 0.9828,
      "step": 2774
    },
    {
      "epoch": 0.34613945366097043,
      "grad_norm": 0.0,
      "learning_rate": 1.5201541567144639e-05,
      "loss": 1.0393,
      "step": 2775
    },
    {
      "epoch": 0.34626418859922664,
      "grad_norm": 0.0,
      "learning_rate": 1.5198090592885469e-05,
      "loss": 0.9762,
      "step": 2776
    },
    {
      "epoch": 0.34638892353748285,
      "grad_norm": 0.0,
      "learning_rate": 1.5194638770166865e-05,
      "loss": 1.0246,
      "step": 2777
    },
    {
      "epoch": 0.34651365847573906,
      "grad_norm": 0.0,
      "learning_rate": 1.5191186099552261e-05,
      "loss": 1.0668,
      "step": 2778
    },
    {
      "epoch": 0.34663839341399527,
      "grad_norm": 0.0,
      "learning_rate": 1.5187732581605217e-05,
      "loss": 1.0404,
      "step": 2779
    },
    {
      "epoch": 0.3467631283522515,
      "grad_norm": 0.0,
      "learning_rate": 1.5184278216889431e-05,
      "loss": 0.9436,
      "step": 2780
    },
    {
      "epoch": 0.3468878632905077,
      "grad_norm": 0.0,
      "learning_rate": 1.5180823005968746e-05,
      "loss": 1.046,
      "step": 2781
    },
    {
      "epoch": 0.3470125982287639,
      "grad_norm": 0.0,
      "learning_rate": 1.517736694940714e-05,
      "loss": 0.9809,
      "step": 2782
    },
    {
      "epoch": 0.34713733316702006,
      "grad_norm": 0.0,
      "learning_rate": 1.5173910047768728e-05,
      "loss": 0.99,
      "step": 2783
    },
    {
      "epoch": 0.34726206810527627,
      "grad_norm": 0.0,
      "learning_rate": 1.517045230161776e-05,
      "loss": 1.009,
      "step": 2784
    },
    {
      "epoch": 0.3473868030435325,
      "grad_norm": 0.0,
      "learning_rate": 1.5166993711518631e-05,
      "loss": 0.9996,
      "step": 2785
    },
    {
      "epoch": 0.3475115379817887,
      "grad_norm": 0.0,
      "learning_rate": 1.5163534278035869e-05,
      "loss": 0.9775,
      "step": 2786
    },
    {
      "epoch": 0.3476362729200449,
      "grad_norm": 0.0,
      "learning_rate": 1.5160074001734142e-05,
      "loss": 1.0312,
      "step": 2787
    },
    {
      "epoch": 0.3477610078583011,
      "grad_norm": 0.0,
      "learning_rate": 1.515661288317825e-05,
      "loss": 1.0535,
      "step": 2788
    },
    {
      "epoch": 0.3478857427965573,
      "grad_norm": 0.0,
      "learning_rate": 1.5153150922933141e-05,
      "loss": 0.977,
      "step": 2789
    },
    {
      "epoch": 0.34801047773481353,
      "grad_norm": 0.0,
      "learning_rate": 1.514968812156389e-05,
      "loss": 0.9796,
      "step": 2790
    },
    {
      "epoch": 0.34813521267306974,
      "grad_norm": 0.0,
      "learning_rate": 1.5146224479635715e-05,
      "loss": 1.0475,
      "step": 2791
    },
    {
      "epoch": 0.34825994761132595,
      "grad_norm": 0.0,
      "learning_rate": 1.5142759997713969e-05,
      "loss": 1.0165,
      "step": 2792
    },
    {
      "epoch": 0.34838468254958216,
      "grad_norm": 0.0,
      "learning_rate": 1.5139294676364146e-05,
      "loss": 0.9553,
      "step": 2793
    },
    {
      "epoch": 0.34850941748783837,
      "grad_norm": 0.0,
      "learning_rate": 1.5135828516151868e-05,
      "loss": 0.9831,
      "step": 2794
    },
    {
      "epoch": 0.3486341524260945,
      "grad_norm": 0.0,
      "learning_rate": 1.5132361517642902e-05,
      "loss": 1.0291,
      "step": 2795
    },
    {
      "epoch": 0.34875888736435073,
      "grad_norm": 0.0,
      "learning_rate": 1.5128893681403152e-05,
      "loss": 1.0299,
      "step": 2796
    },
    {
      "epoch": 0.34888362230260694,
      "grad_norm": 0.0,
      "learning_rate": 1.5125425007998653e-05,
      "loss": 1.033,
      "step": 2797
    },
    {
      "epoch": 0.34900835724086315,
      "grad_norm": 0.0,
      "learning_rate": 1.5121955497995581e-05,
      "loss": 0.9655,
      "step": 2798
    },
    {
      "epoch": 0.34913309217911936,
      "grad_norm": 0.0,
      "learning_rate": 1.5118485151960252e-05,
      "loss": 0.9885,
      "step": 2799
    },
    {
      "epoch": 0.3492578271173756,
      "grad_norm": 0.0,
      "learning_rate": 1.5115013970459108e-05,
      "loss": 1.0125,
      "step": 2800
    },
    {
      "epoch": 0.3493825620556318,
      "grad_norm": 0.0,
      "learning_rate": 1.5111541954058733e-05,
      "loss": 1.0099,
      "step": 2801
    },
    {
      "epoch": 0.349507296993888,
      "grad_norm": 0.0,
      "learning_rate": 1.5108069103325852e-05,
      "loss": 0.9761,
      "step": 2802
    },
    {
      "epoch": 0.3496320319321442,
      "grad_norm": 0.0,
      "learning_rate": 1.5104595418827317e-05,
      "loss": 0.984,
      "step": 2803
    },
    {
      "epoch": 0.3497567668704004,
      "grad_norm": 0.0,
      "learning_rate": 1.5101120901130126e-05,
      "loss": 0.9781,
      "step": 2804
    },
    {
      "epoch": 0.3498815018086566,
      "grad_norm": 0.0,
      "learning_rate": 1.5097645550801404e-05,
      "loss": 0.9944,
      "step": 2805
    },
    {
      "epoch": 0.35000623674691284,
      "grad_norm": 0.0,
      "learning_rate": 1.509416936840842e-05,
      "loss": 0.927,
      "step": 2806
    },
    {
      "epoch": 0.350130971685169,
      "grad_norm": 0.0,
      "learning_rate": 1.509069235451857e-05,
      "loss": 1.0058,
      "step": 2807
    },
    {
      "epoch": 0.3502557066234252,
      "grad_norm": 0.0,
      "learning_rate": 1.5087214509699388e-05,
      "loss": 1.0173,
      "step": 2808
    },
    {
      "epoch": 0.3503804415616814,
      "grad_norm": 0.0,
      "learning_rate": 1.5083735834518556e-05,
      "loss": 1.0044,
      "step": 2809
    },
    {
      "epoch": 0.3505051764999376,
      "grad_norm": 0.0,
      "learning_rate": 1.5080256329543877e-05,
      "loss": 0.9838,
      "step": 2810
    },
    {
      "epoch": 0.35062991143819383,
      "grad_norm": 0.0,
      "learning_rate": 1.5076775995343286e-05,
      "loss": 0.9809,
      "step": 2811
    },
    {
      "epoch": 0.35075464637645004,
      "grad_norm": 0.0,
      "learning_rate": 1.5073294832484872e-05,
      "loss": 1.0173,
      "step": 2812
    },
    {
      "epoch": 0.35087938131470625,
      "grad_norm": 0.0,
      "learning_rate": 1.5069812841536844e-05,
      "loss": 0.9986,
      "step": 2813
    },
    {
      "epoch": 0.35100411625296246,
      "grad_norm": 0.0,
      "learning_rate": 1.5066330023067549e-05,
      "loss": 1.0107,
      "step": 2814
    },
    {
      "epoch": 0.35112885119121867,
      "grad_norm": 0.0,
      "learning_rate": 1.5062846377645476e-05,
      "loss": 0.9821,
      "step": 2815
    },
    {
      "epoch": 0.3512535861294749,
      "grad_norm": 0.0,
      "learning_rate": 1.5059361905839236e-05,
      "loss": 0.9763,
      "step": 2816
    },
    {
      "epoch": 0.3513783210677311,
      "grad_norm": 0.0,
      "learning_rate": 1.505587660821759e-05,
      "loss": 1.0172,
      "step": 2817
    },
    {
      "epoch": 0.3515030560059873,
      "grad_norm": 0.0,
      "learning_rate": 1.5052390485349424e-05,
      "loss": 0.9506,
      "step": 2818
    },
    {
      "epoch": 0.35162779094424346,
      "grad_norm": 0.0,
      "learning_rate": 1.5048903537803759e-05,
      "loss": 1.0249,
      "step": 2819
    },
    {
      "epoch": 0.35175252588249967,
      "grad_norm": 0.0,
      "learning_rate": 1.5045415766149755e-05,
      "loss": 0.976,
      "step": 2820
    },
    {
      "epoch": 0.3518772608207559,
      "grad_norm": 0.0,
      "learning_rate": 1.5041927170956707e-05,
      "loss": 1.0174,
      "step": 2821
    },
    {
      "epoch": 0.3520019957590121,
      "grad_norm": 0.0,
      "learning_rate": 1.5038437752794035e-05,
      "loss": 0.9818,
      "step": 2822
    },
    {
      "epoch": 0.3521267306972683,
      "grad_norm": 0.0,
      "learning_rate": 1.5034947512231304e-05,
      "loss": 0.9849,
      "step": 2823
    },
    {
      "epoch": 0.3522514656355245,
      "grad_norm": 0.0,
      "learning_rate": 1.5031456449838207e-05,
      "loss": 1.0105,
      "step": 2824
    },
    {
      "epoch": 0.3523762005737807,
      "grad_norm": 0.0,
      "learning_rate": 1.5027964566184578e-05,
      "loss": 0.9842,
      "step": 2825
    },
    {
      "epoch": 0.35250093551203693,
      "grad_norm": 0.0,
      "learning_rate": 1.5024471861840378e-05,
      "loss": 0.934,
      "step": 2826
    },
    {
      "epoch": 0.35262567045029314,
      "grad_norm": 0.0,
      "learning_rate": 1.5020978337375702e-05,
      "loss": 0.9912,
      "step": 2827
    },
    {
      "epoch": 0.35275040538854935,
      "grad_norm": 0.0,
      "learning_rate": 1.5017483993360782e-05,
      "loss": 0.9825,
      "step": 2828
    },
    {
      "epoch": 0.35287514032680556,
      "grad_norm": 0.0,
      "learning_rate": 1.5013988830365986e-05,
      "loss": 1.0161,
      "step": 2829
    },
    {
      "epoch": 0.35299987526506177,
      "grad_norm": 0.0,
      "learning_rate": 1.501049284896181e-05,
      "loss": 1.0253,
      "step": 2830
    },
    {
      "epoch": 0.3531246102033179,
      "grad_norm": 0.0,
      "learning_rate": 1.5006996049718887e-05,
      "loss": 1.0009,
      "step": 2831
    },
    {
      "epoch": 0.35324934514157413,
      "grad_norm": 0.0,
      "learning_rate": 1.5003498433207987e-05,
      "loss": 0.9749,
      "step": 2832
    },
    {
      "epoch": 0.35337408007983034,
      "grad_norm": 0.0,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.9799,
      "step": 2833
    },
    {
      "epoch": 0.35349881501808655,
      "grad_norm": 0.0,
      "learning_rate": 1.4996500750665968e-05,
      "loss": 1.0003,
      "step": 2834
    },
    {
      "epoch": 0.35362354995634276,
      "grad_norm": 0.0,
      "learning_rate": 1.4993000685777051e-05,
      "loss": 0.9438,
      "step": 2835
    },
    {
      "epoch": 0.353748284894599,
      "grad_norm": 0.0,
      "learning_rate": 1.4989499805904548e-05,
      "loss": 0.9778,
      "step": 2836
    },
    {
      "epoch": 0.3538730198328552,
      "grad_norm": 0.0,
      "learning_rate": 1.498599811161989e-05,
      "loss": 0.9979,
      "step": 2837
    },
    {
      "epoch": 0.3539977547711114,
      "grad_norm": 0.0,
      "learning_rate": 1.4982495603494648e-05,
      "loss": 1.0163,
      "step": 2838
    },
    {
      "epoch": 0.3541224897093676,
      "grad_norm": 0.0,
      "learning_rate": 1.4978992282100518e-05,
      "loss": 1.0,
      "step": 2839
    },
    {
      "epoch": 0.3542472246476238,
      "grad_norm": 0.0,
      "learning_rate": 1.497548814800932e-05,
      "loss": 1.0519,
      "step": 2840
    },
    {
      "epoch": 0.35437195958588,
      "grad_norm": 0.0,
      "learning_rate": 1.497198320179303e-05,
      "loss": 0.9876,
      "step": 2841
    },
    {
      "epoch": 0.35449669452413624,
      "grad_norm": 0.0,
      "learning_rate": 1.4968477444023739e-05,
      "loss": 1.0283,
      "step": 2842
    },
    {
      "epoch": 0.3546214294623924,
      "grad_norm": 0.0,
      "learning_rate": 1.496497087527367e-05,
      "loss": 0.9821,
      "step": 2843
    },
    {
      "epoch": 0.3547461644006486,
      "grad_norm": 0.0,
      "learning_rate": 1.4961463496115194e-05,
      "loss": 0.9752,
      "step": 2844
    },
    {
      "epoch": 0.3548708993389048,
      "grad_norm": 0.0,
      "learning_rate": 1.4957955307120796e-05,
      "loss": 1.0184,
      "step": 2845
    },
    {
      "epoch": 0.354995634277161,
      "grad_norm": 0.0,
      "learning_rate": 1.4954446308863098e-05,
      "loss": 0.9585,
      "step": 2846
    },
    {
      "epoch": 0.35512036921541723,
      "grad_norm": 0.0,
      "learning_rate": 1.4950936501914863e-05,
      "loss": 0.9566,
      "step": 2847
    },
    {
      "epoch": 0.35524510415367344,
      "grad_norm": 0.0,
      "learning_rate": 1.4947425886848983e-05,
      "loss": 0.961,
      "step": 2848
    },
    {
      "epoch": 0.35536983909192965,
      "grad_norm": 0.0,
      "learning_rate": 1.494391446423847e-05,
      "loss": 0.9629,
      "step": 2849
    },
    {
      "epoch": 0.35549457403018586,
      "grad_norm": 0.0,
      "learning_rate": 1.4940402234656484e-05,
      "loss": 0.9813,
      "step": 2850
    },
    {
      "epoch": 0.35561930896844207,
      "grad_norm": 0.0,
      "learning_rate": 1.4936889198676303e-05,
      "loss": 0.9803,
      "step": 2851
    },
    {
      "epoch": 0.3557440439066983,
      "grad_norm": 0.0,
      "learning_rate": 1.4933375356871349e-05,
      "loss": 1.0245,
      "step": 2852
    },
    {
      "epoch": 0.3558687788449545,
      "grad_norm": 0.0,
      "learning_rate": 1.4929860709815166e-05,
      "loss": 0.9981,
      "step": 2853
    },
    {
      "epoch": 0.3559935137832107,
      "grad_norm": 0.0,
      "learning_rate": 1.4926345258081433e-05,
      "loss": 0.9748,
      "step": 2854
    },
    {
      "epoch": 0.35611824872146686,
      "grad_norm": 0.0,
      "learning_rate": 1.4922829002243968e-05,
      "loss": 0.9947,
      "step": 2855
    },
    {
      "epoch": 0.35624298365972307,
      "grad_norm": 0.0,
      "learning_rate": 1.49193119428767e-05,
      "loss": 1.0089,
      "step": 2856
    },
    {
      "epoch": 0.3563677185979793,
      "grad_norm": 0.0,
      "learning_rate": 1.4915794080553706e-05,
      "loss": 0.9712,
      "step": 2857
    },
    {
      "epoch": 0.3564924535362355,
      "grad_norm": 0.0,
      "learning_rate": 1.4912275415849195e-05,
      "loss": 0.9885,
      "step": 2858
    },
    {
      "epoch": 0.3566171884744917,
      "grad_norm": 0.0,
      "learning_rate": 1.4908755949337499e-05,
      "loss": 1.0503,
      "step": 2859
    },
    {
      "epoch": 0.3567419234127479,
      "grad_norm": 0.0,
      "learning_rate": 1.4905235681593079e-05,
      "loss": 1.0304,
      "step": 2860
    },
    {
      "epoch": 0.3568666583510041,
      "grad_norm": 0.0,
      "learning_rate": 1.490171461319054e-05,
      "loss": 1.0446,
      "step": 2861
    },
    {
      "epoch": 0.35699139328926033,
      "grad_norm": 0.0,
      "learning_rate": 1.4898192744704605e-05,
      "loss": 0.9798,
      "step": 2862
    },
    {
      "epoch": 0.35711612822751654,
      "grad_norm": 0.0,
      "learning_rate": 1.4894670076710127e-05,
      "loss": 0.9971,
      "step": 2863
    },
    {
      "epoch": 0.35724086316577275,
      "grad_norm": 0.0,
      "learning_rate": 1.4891146609782102e-05,
      "loss": 0.9864,
      "step": 2864
    },
    {
      "epoch": 0.35736559810402896,
      "grad_norm": 0.0,
      "learning_rate": 1.4887622344495643e-05,
      "loss": 0.9692,
      "step": 2865
    },
    {
      "epoch": 0.35749033304228517,
      "grad_norm": 0.0,
      "learning_rate": 1.4884097281426004e-05,
      "loss": 1.0061,
      "step": 2866
    },
    {
      "epoch": 0.3576150679805413,
      "grad_norm": 0.0,
      "learning_rate": 1.488057142114856e-05,
      "loss": 1.0409,
      "step": 2867
    },
    {
      "epoch": 0.35773980291879753,
      "grad_norm": 0.0,
      "learning_rate": 1.4877044764238821e-05,
      "loss": 1.0067,
      "step": 2868
    },
    {
      "epoch": 0.35786453785705374,
      "grad_norm": 0.0,
      "learning_rate": 1.4873517311272425e-05,
      "loss": 1.0363,
      "step": 2869
    },
    {
      "epoch": 0.35798927279530995,
      "grad_norm": 0.0,
      "learning_rate": 1.4869989062825143e-05,
      "loss": 0.9545,
      "step": 2870
    },
    {
      "epoch": 0.35811400773356616,
      "grad_norm": 0.0,
      "learning_rate": 1.4866460019472877e-05,
      "loss": 0.9931,
      "step": 2871
    },
    {
      "epoch": 0.3582387426718224,
      "grad_norm": 0.0,
      "learning_rate": 1.4862930181791654e-05,
      "loss": 1.0165,
      "step": 2872
    },
    {
      "epoch": 0.3583634776100786,
      "grad_norm": 0.0,
      "learning_rate": 1.4859399550357629e-05,
      "loss": 0.963,
      "step": 2873
    },
    {
      "epoch": 0.3584882125483348,
      "grad_norm": 0.0,
      "learning_rate": 1.4855868125747093e-05,
      "loss": 0.9482,
      "step": 2874
    },
    {
      "epoch": 0.358612947486591,
      "grad_norm": 0.0,
      "learning_rate": 1.4852335908536464e-05,
      "loss": 0.9399,
      "step": 2875
    },
    {
      "epoch": 0.3587376824248472,
      "grad_norm": 0.0,
      "learning_rate": 1.4848802899302288e-05,
      "loss": 0.9738,
      "step": 2876
    },
    {
      "epoch": 0.3588624173631034,
      "grad_norm": 0.0,
      "learning_rate": 1.4845269098621237e-05,
      "loss": 0.9525,
      "step": 2877
    },
    {
      "epoch": 0.35898715230135964,
      "grad_norm": 0.0,
      "learning_rate": 1.484173450707013e-05,
      "loss": 1.0399,
      "step": 2878
    },
    {
      "epoch": 0.3591118872396158,
      "grad_norm": 0.0,
      "learning_rate": 1.4838199125225885e-05,
      "loss": 1.0152,
      "step": 2879
    },
    {
      "epoch": 0.359236622177872,
      "grad_norm": 0.0,
      "learning_rate": 1.4834662953665572e-05,
      "loss": 0.9662,
      "step": 2880
    },
    {
      "epoch": 0.3593613571161282,
      "grad_norm": 0.0,
      "learning_rate": 1.4831125992966386e-05,
      "loss": 1.0186,
      "step": 2881
    },
    {
      "epoch": 0.3594860920543844,
      "grad_norm": 0.0,
      "learning_rate": 1.4827588243705647e-05,
      "loss": 0.9814,
      "step": 2882
    },
    {
      "epoch": 0.35961082699264063,
      "grad_norm": 0.0,
      "learning_rate": 1.48240497064608e-05,
      "loss": 0.9667,
      "step": 2883
    },
    {
      "epoch": 0.35973556193089684,
      "grad_norm": 0.0,
      "learning_rate": 1.4820510381809432e-05,
      "loss": 0.9956,
      "step": 2884
    },
    {
      "epoch": 0.35986029686915305,
      "grad_norm": 0.0,
      "learning_rate": 1.4816970270329242e-05,
      "loss": 0.9922,
      "step": 2885
    },
    {
      "epoch": 0.35998503180740926,
      "grad_norm": 0.0,
      "learning_rate": 1.481342937259807e-05,
      "loss": 1.0126,
      "step": 2886
    },
    {
      "epoch": 0.36010976674566547,
      "grad_norm": 0.0,
      "learning_rate": 1.4809887689193878e-05,
      "loss": 0.9904,
      "step": 2887
    },
    {
      "epoch": 0.3602345016839217,
      "grad_norm": 0.0,
      "learning_rate": 1.480634522069476e-05,
      "loss": 0.9835,
      "step": 2888
    },
    {
      "epoch": 0.3603592366221779,
      "grad_norm": 0.0,
      "learning_rate": 1.4802801967678932e-05,
      "loss": 1.0033,
      "step": 2889
    },
    {
      "epoch": 0.3604839715604341,
      "grad_norm": 0.0,
      "learning_rate": 1.4799257930724747e-05,
      "loss": 0.9804,
      "step": 2890
    },
    {
      "epoch": 0.36060870649869026,
      "grad_norm": 0.0,
      "learning_rate": 1.4795713110410679e-05,
      "loss": 0.9716,
      "step": 2891
    },
    {
      "epoch": 0.36073344143694647,
      "grad_norm": 0.0,
      "learning_rate": 1.4792167507315332e-05,
      "loss": 0.9334,
      "step": 2892
    },
    {
      "epoch": 0.3608581763752027,
      "grad_norm": 0.0,
      "learning_rate": 1.4788621122017434e-05,
      "loss": 0.9822,
      "step": 2893
    },
    {
      "epoch": 0.3609829113134589,
      "grad_norm": 0.0,
      "learning_rate": 1.4785073955095855e-05,
      "loss": 0.9925,
      "step": 2894
    },
    {
      "epoch": 0.3611076462517151,
      "grad_norm": 0.0,
      "learning_rate": 1.4781526007129575e-05,
      "loss": 1.0099,
      "step": 2895
    },
    {
      "epoch": 0.3612323811899713,
      "grad_norm": 0.0,
      "learning_rate": 1.4777977278697704e-05,
      "loss": 1.0453,
      "step": 2896
    },
    {
      "epoch": 0.3613571161282275,
      "grad_norm": 0.0,
      "learning_rate": 1.4774427770379492e-05,
      "loss": 0.9504,
      "step": 2897
    },
    {
      "epoch": 0.36148185106648373,
      "grad_norm": 0.0,
      "learning_rate": 1.4770877482754304e-05,
      "loss": 1.0245,
      "step": 2898
    },
    {
      "epoch": 0.36160658600473994,
      "grad_norm": 0.0,
      "learning_rate": 1.4767326416401639e-05,
      "loss": 1.0735,
      "step": 2899
    },
    {
      "epoch": 0.36173132094299615,
      "grad_norm": 0.0,
      "learning_rate": 1.476377457190112e-05,
      "loss": 0.9711,
      "step": 2900
    },
    {
      "epoch": 0.36185605588125236,
      "grad_norm": 0.0,
      "learning_rate": 1.4760221949832497e-05,
      "loss": 0.9436,
      "step": 2901
    },
    {
      "epoch": 0.36198079081950857,
      "grad_norm": 0.0,
      "learning_rate": 1.4756668550775643e-05,
      "loss": 1.0249,
      "step": 2902
    },
    {
      "epoch": 0.3621055257577647,
      "grad_norm": 0.0,
      "learning_rate": 1.475311437531057e-05,
      "loss": 1.0056,
      "step": 2903
    },
    {
      "epoch": 0.36223026069602093,
      "grad_norm": 0.0,
      "learning_rate": 1.4749559424017405e-05,
      "loss": 1.0012,
      "step": 2904
    },
    {
      "epoch": 0.36235499563427714,
      "grad_norm": 0.0,
      "learning_rate": 1.4746003697476406e-05,
      "loss": 0.9912,
      "step": 2905
    },
    {
      "epoch": 0.36247973057253335,
      "grad_norm": 0.0,
      "learning_rate": 1.4742447196267957e-05,
      "loss": 0.9979,
      "step": 2906
    },
    {
      "epoch": 0.36260446551078956,
      "grad_norm": 0.0,
      "learning_rate": 1.473888992097257e-05,
      "loss": 1.0139,
      "step": 2907
    },
    {
      "epoch": 0.3627292004490458,
      "grad_norm": 0.0,
      "learning_rate": 1.4735331872170882e-05,
      "loss": 0.9583,
      "step": 2908
    },
    {
      "epoch": 0.362853935387302,
      "grad_norm": 0.0,
      "learning_rate": 1.4731773050443654e-05,
      "loss": 0.9923,
      "step": 2909
    },
    {
      "epoch": 0.3629786703255582,
      "grad_norm": 0.0,
      "learning_rate": 1.4728213456371777e-05,
      "loss": 0.989,
      "step": 2910
    },
    {
      "epoch": 0.3631034052638144,
      "grad_norm": 0.0,
      "learning_rate": 1.4724653090536266e-05,
      "loss": 1.0014,
      "step": 2911
    },
    {
      "epoch": 0.3632281402020706,
      "grad_norm": 0.0,
      "learning_rate": 1.4721091953518264e-05,
      "loss": 0.922,
      "step": 2912
    },
    {
      "epoch": 0.3633528751403268,
      "grad_norm": 0.0,
      "learning_rate": 1.4717530045899038e-05,
      "loss": 0.9941,
      "step": 2913
    },
    {
      "epoch": 0.36347761007858304,
      "grad_norm": 0.0,
      "learning_rate": 1.4713967368259981e-05,
      "loss": 1.0087,
      "step": 2914
    },
    {
      "epoch": 0.3636023450168392,
      "grad_norm": 0.0,
      "learning_rate": 1.4710403921182605e-05,
      "loss": 0.9785,
      "step": 2915
    },
    {
      "epoch": 0.3637270799550954,
      "grad_norm": 0.0,
      "learning_rate": 1.4706839705248564e-05,
      "loss": 1.0208,
      "step": 2916
    },
    {
      "epoch": 0.3638518148933516,
      "grad_norm": 0.0,
      "learning_rate": 1.4703274721039626e-05,
      "loss": 0.9719,
      "step": 2917
    },
    {
      "epoch": 0.3639765498316078,
      "grad_norm": 0.0,
      "learning_rate": 1.4699708969137684e-05,
      "loss": 0.9952,
      "step": 2918
    },
    {
      "epoch": 0.36410128476986403,
      "grad_norm": 0.0,
      "learning_rate": 1.4696142450124758e-05,
      "loss": 0.9925,
      "step": 2919
    },
    {
      "epoch": 0.36422601970812024,
      "grad_norm": 0.0,
      "learning_rate": 1.4692575164582997e-05,
      "loss": 1.0146,
      "step": 2920
    },
    {
      "epoch": 0.36435075464637645,
      "grad_norm": 0.0,
      "learning_rate": 1.4689007113094667e-05,
      "loss": 1.0125,
      "step": 2921
    },
    {
      "epoch": 0.36447548958463266,
      "grad_norm": 0.0,
      "learning_rate": 1.468543829624217e-05,
      "loss": 1.0407,
      "step": 2922
    },
    {
      "epoch": 0.36460022452288887,
      "grad_norm": 0.0,
      "learning_rate": 1.4681868714608021e-05,
      "loss": 1.0328,
      "step": 2923
    },
    {
      "epoch": 0.3647249594611451,
      "grad_norm": 0.0,
      "learning_rate": 1.467829836877487e-05,
      "loss": 0.9793,
      "step": 2924
    },
    {
      "epoch": 0.3648496943994013,
      "grad_norm": 0.0,
      "learning_rate": 1.4674727259325487e-05,
      "loss": 1.0099,
      "step": 2925
    },
    {
      "epoch": 0.3649744293376575,
      "grad_norm": 0.0,
      "learning_rate": 1.4671155386842764e-05,
      "loss": 0.9585,
      "step": 2926
    },
    {
      "epoch": 0.36509916427591366,
      "grad_norm": 0.0,
      "learning_rate": 1.4667582751909723e-05,
      "loss": 1.0338,
      "step": 2927
    },
    {
      "epoch": 0.36522389921416987,
      "grad_norm": 0.0,
      "learning_rate": 1.466400935510951e-05,
      "loss": 0.933,
      "step": 2928
    },
    {
      "epoch": 0.3653486341524261,
      "grad_norm": 0.0,
      "learning_rate": 1.4660435197025391e-05,
      "loss": 0.9981,
      "step": 2929
    },
    {
      "epoch": 0.3654733690906823,
      "grad_norm": 0.0,
      "learning_rate": 1.4656860278240756e-05,
      "loss": 0.9828,
      "step": 2930
    },
    {
      "epoch": 0.3655981040289385,
      "grad_norm": 0.0,
      "learning_rate": 1.4653284599339128e-05,
      "loss": 0.976,
      "step": 2931
    },
    {
      "epoch": 0.3657228389671947,
      "grad_norm": 0.0,
      "learning_rate": 1.4649708160904142e-05,
      "loss": 0.9517,
      "step": 2932
    },
    {
      "epoch": 0.3658475739054509,
      "grad_norm": 0.0,
      "learning_rate": 1.4646130963519567e-05,
      "loss": 0.9611,
      "step": 2933
    },
    {
      "epoch": 0.36597230884370713,
      "grad_norm": 0.0,
      "learning_rate": 1.4642553007769292e-05,
      "loss": 0.9719,
      "step": 2934
    },
    {
      "epoch": 0.36609704378196334,
      "grad_norm": 0.0,
      "learning_rate": 1.4638974294237325e-05,
      "loss": 1.0356,
      "step": 2935
    },
    {
      "epoch": 0.36622177872021955,
      "grad_norm": 0.0,
      "learning_rate": 1.4635394823507807e-05,
      "loss": 0.9767,
      "step": 2936
    },
    {
      "epoch": 0.36634651365847576,
      "grad_norm": 0.0,
      "learning_rate": 1.4631814596164995e-05,
      "loss": 0.9099,
      "step": 2937
    },
    {
      "epoch": 0.36647124859673197,
      "grad_norm": 0.0,
      "learning_rate": 1.4628233612793277e-05,
      "loss": 0.9964,
      "step": 2938
    },
    {
      "epoch": 0.3665959835349881,
      "grad_norm": 0.0,
      "learning_rate": 1.4624651873977152e-05,
      "loss": 1.0148,
      "step": 2939
    },
    {
      "epoch": 0.36672071847324433,
      "grad_norm": 0.0,
      "learning_rate": 1.4621069380301256e-05,
      "loss": 1.0399,
      "step": 2940
    },
    {
      "epoch": 0.36684545341150054,
      "grad_norm": 0.0,
      "learning_rate": 1.4617486132350343e-05,
      "loss": 0.9567,
      "step": 2941
    },
    {
      "epoch": 0.36697018834975675,
      "grad_norm": 0.0,
      "learning_rate": 1.4613902130709282e-05,
      "loss": 0.9391,
      "step": 2942
    },
    {
      "epoch": 0.36709492328801296,
      "grad_norm": 0.0,
      "learning_rate": 1.461031737596308e-05,
      "loss": 1.0167,
      "step": 2943
    },
    {
      "epoch": 0.3672196582262692,
      "grad_norm": 0.0,
      "learning_rate": 1.4606731868696857e-05,
      "loss": 0.9683,
      "step": 2944
    },
    {
      "epoch": 0.3673443931645254,
      "grad_norm": 0.0,
      "learning_rate": 1.460314560949586e-05,
      "loss": 0.9692,
      "step": 2945
    },
    {
      "epoch": 0.3674691281027816,
      "grad_norm": 0.0,
      "learning_rate": 1.4599558598945452e-05,
      "loss": 0.9964,
      "step": 2946
    },
    {
      "epoch": 0.3675938630410378,
      "grad_norm": 0.0,
      "learning_rate": 1.4595970837631126e-05,
      "loss": 0.9467,
      "step": 2947
    },
    {
      "epoch": 0.367718597979294,
      "grad_norm": 0.0,
      "learning_rate": 1.4592382326138498e-05,
      "loss": 0.9851,
      "step": 2948
    },
    {
      "epoch": 0.3678433329175502,
      "grad_norm": 0.0,
      "learning_rate": 1.45887930650533e-05,
      "loss": 0.9967,
      "step": 2949
    },
    {
      "epoch": 0.36796806785580644,
      "grad_norm": 0.0,
      "learning_rate": 1.458520305496139e-05,
      "loss": 0.9696,
      "step": 2950
    },
    {
      "epoch": 0.3680928027940626,
      "grad_norm": 0.0,
      "learning_rate": 1.4581612296448752e-05,
      "loss": 1.0167,
      "step": 2951
    },
    {
      "epoch": 0.3682175377323188,
      "grad_norm": 0.0,
      "learning_rate": 1.4578020790101484e-05,
      "loss": 1.0345,
      "step": 2952
    },
    {
      "epoch": 0.368342272670575,
      "grad_norm": 0.0,
      "learning_rate": 1.4574428536505809e-05,
      "loss": 1.0029,
      "step": 2953
    },
    {
      "epoch": 0.3684670076088312,
      "grad_norm": 0.0,
      "learning_rate": 1.457083553624808e-05,
      "loss": 0.959,
      "step": 2954
    },
    {
      "epoch": 0.36859174254708743,
      "grad_norm": 0.0,
      "learning_rate": 1.4567241789914758e-05,
      "loss": 0.9828,
      "step": 2955
    },
    {
      "epoch": 0.36871647748534364,
      "grad_norm": 0.0,
      "learning_rate": 1.4563647298092438e-05,
      "loss": 0.9585,
      "step": 2956
    },
    {
      "epoch": 0.36884121242359985,
      "grad_norm": 0.0,
      "learning_rate": 1.456005206136783e-05,
      "loss": 0.9818,
      "step": 2957
    },
    {
      "epoch": 0.36896594736185606,
      "grad_norm": 0.0,
      "learning_rate": 1.455645608032777e-05,
      "loss": 1.0333,
      "step": 2958
    },
    {
      "epoch": 0.36909068230011227,
      "grad_norm": 0.0,
      "learning_rate": 1.4552859355559205e-05,
      "loss": 1.03,
      "step": 2959
    },
    {
      "epoch": 0.3692154172383685,
      "grad_norm": 0.0,
      "learning_rate": 1.454926188764922e-05,
      "loss": 1.0312,
      "step": 2960
    },
    {
      "epoch": 0.3693401521766247,
      "grad_norm": 0.0,
      "learning_rate": 1.4545663677185007e-05,
      "loss": 0.9644,
      "step": 2961
    },
    {
      "epoch": 0.3694648871148809,
      "grad_norm": 0.0,
      "learning_rate": 1.4542064724753886e-05,
      "loss": 1.0274,
      "step": 2962
    },
    {
      "epoch": 0.36958962205313706,
      "grad_norm": 0.0,
      "learning_rate": 1.4538465030943298e-05,
      "loss": 1.0076,
      "step": 2963
    },
    {
      "epoch": 0.36971435699139327,
      "grad_norm": 0.0,
      "learning_rate": 1.4534864596340806e-05,
      "loss": 0.9479,
      "step": 2964
    },
    {
      "epoch": 0.3698390919296495,
      "grad_norm": 0.0,
      "learning_rate": 1.4531263421534083e-05,
      "loss": 1.0408,
      "step": 2965
    },
    {
      "epoch": 0.3699638268679057,
      "grad_norm": 0.0,
      "learning_rate": 1.452766150711094e-05,
      "loss": 1.0255,
      "step": 2966
    },
    {
      "epoch": 0.3700885618061619,
      "grad_norm": 0.0,
      "learning_rate": 1.4524058853659297e-05,
      "loss": 0.9856,
      "step": 2967
    },
    {
      "epoch": 0.3702132967444181,
      "grad_norm": 0.0,
      "learning_rate": 1.45204554617672e-05,
      "loss": 1.0415,
      "step": 2968
    },
    {
      "epoch": 0.3703380316826743,
      "grad_norm": 0.0,
      "learning_rate": 1.4516851332022808e-05,
      "loss": 0.9953,
      "step": 2969
    },
    {
      "epoch": 0.37046276662093053,
      "grad_norm": 0.0,
      "learning_rate": 1.451324646501441e-05,
      "loss": 1.0328,
      "step": 2970
    },
    {
      "epoch": 0.37058750155918674,
      "grad_norm": 0.0,
      "learning_rate": 1.4509640861330415e-05,
      "loss": 0.9855,
      "step": 2971
    },
    {
      "epoch": 0.37071223649744295,
      "grad_norm": 0.0,
      "learning_rate": 1.4506034521559337e-05,
      "loss": 0.9958,
      "step": 2972
    },
    {
      "epoch": 0.37083697143569916,
      "grad_norm": 0.0,
      "learning_rate": 1.4502427446289833e-05,
      "loss": 1.0174,
      "step": 2973
    },
    {
      "epoch": 0.37096170637395537,
      "grad_norm": 0.0,
      "learning_rate": 1.4498819636110665e-05,
      "loss": 0.9964,
      "step": 2974
    },
    {
      "epoch": 0.3710864413122115,
      "grad_norm": 0.0,
      "learning_rate": 1.4495211091610716e-05,
      "loss": 1.0322,
      "step": 2975
    },
    {
      "epoch": 0.37121117625046773,
      "grad_norm": 0.0,
      "learning_rate": 1.4491601813378995e-05,
      "loss": 0.9638,
      "step": 2976
    },
    {
      "epoch": 0.37133591118872394,
      "grad_norm": 0.0,
      "learning_rate": 1.4487991802004625e-05,
      "loss": 0.9745,
      "step": 2977
    },
    {
      "epoch": 0.37146064612698015,
      "grad_norm": 0.0,
      "learning_rate": 1.4484381058076849e-05,
      "loss": 0.9817,
      "step": 2978
    },
    {
      "epoch": 0.37158538106523636,
      "grad_norm": 0.0,
      "learning_rate": 1.4480769582185036e-05,
      "loss": 1.0031,
      "step": 2979
    },
    {
      "epoch": 0.3717101160034926,
      "grad_norm": 0.0,
      "learning_rate": 1.4477157374918665e-05,
      "loss": 1.0024,
      "step": 2980
    },
    {
      "epoch": 0.3718348509417488,
      "grad_norm": 0.0,
      "learning_rate": 1.4473544436867344e-05,
      "loss": 1.0173,
      "step": 2981
    },
    {
      "epoch": 0.371959585880005,
      "grad_norm": 0.0,
      "learning_rate": 1.4469930768620792e-05,
      "loss": 0.9673,
      "step": 2982
    },
    {
      "epoch": 0.3720843208182612,
      "grad_norm": 0.0,
      "learning_rate": 1.4466316370768853e-05,
      "loss": 0.9887,
      "step": 2983
    },
    {
      "epoch": 0.3722090557565174,
      "grad_norm": 0.0,
      "learning_rate": 1.446270124390149e-05,
      "loss": 0.9361,
      "step": 2984
    },
    {
      "epoch": 0.3723337906947736,
      "grad_norm": 0.0,
      "learning_rate": 1.4459085388608778e-05,
      "loss": 0.9535,
      "step": 2985
    },
    {
      "epoch": 0.37245852563302984,
      "grad_norm": 0.0,
      "learning_rate": 1.4455468805480918e-05,
      "loss": 0.992,
      "step": 2986
    },
    {
      "epoch": 0.372583260571286,
      "grad_norm": 0.0,
      "learning_rate": 1.4451851495108224e-05,
      "loss": 0.9618,
      "step": 2987
    },
    {
      "epoch": 0.3727079955095422,
      "grad_norm": 0.0,
      "learning_rate": 1.4448233458081137e-05,
      "loss": 0.9839,
      "step": 2988
    },
    {
      "epoch": 0.3728327304477984,
      "grad_norm": 0.0,
      "learning_rate": 1.4444614694990212e-05,
      "loss": 0.9588,
      "step": 2989
    },
    {
      "epoch": 0.3729574653860546,
      "grad_norm": 0.0,
      "learning_rate": 1.4440995206426118e-05,
      "loss": 1.0113,
      "step": 2990
    },
    {
      "epoch": 0.37308220032431083,
      "grad_norm": 0.0,
      "learning_rate": 1.4437374992979654e-05,
      "loss": 0.9995,
      "step": 2991
    },
    {
      "epoch": 0.37320693526256704,
      "grad_norm": 0.0,
      "learning_rate": 1.4433754055241722e-05,
      "loss": 1.015,
      "step": 2992
    },
    {
      "epoch": 0.37333167020082325,
      "grad_norm": 0.0,
      "learning_rate": 1.4430132393803353e-05,
      "loss": 0.9556,
      "step": 2993
    },
    {
      "epoch": 0.37345640513907946,
      "grad_norm": 0.0,
      "learning_rate": 1.4426510009255697e-05,
      "loss": 1.0231,
      "step": 2994
    },
    {
      "epoch": 0.37358114007733567,
      "grad_norm": 0.0,
      "learning_rate": 1.4422886902190014e-05,
      "loss": 0.9579,
      "step": 2995
    },
    {
      "epoch": 0.3737058750155919,
      "grad_norm": 0.0,
      "learning_rate": 1.4419263073197689e-05,
      "loss": 1.0043,
      "step": 2996
    },
    {
      "epoch": 0.3738306099538481,
      "grad_norm": 0.0,
      "learning_rate": 1.4415638522870223e-05,
      "loss": 0.9983,
      "step": 2997
    },
    {
      "epoch": 0.3739553448921043,
      "grad_norm": 0.0,
      "learning_rate": 1.441201325179923e-05,
      "loss": 1.0113,
      "step": 2998
    },
    {
      "epoch": 0.37408007983036046,
      "grad_norm": 0.0,
      "learning_rate": 1.4408387260576449e-05,
      "loss": 0.9667,
      "step": 2999
    },
    {
      "epoch": 0.37420481476861667,
      "grad_norm": 0.0,
      "learning_rate": 1.4404760549793733e-05,
      "loss": 1.0073,
      "step": 3000
    },
    {
      "epoch": 0.3743295497068729,
      "grad_norm": 0.0,
      "learning_rate": 1.440113312004305e-05,
      "loss": 0.9697,
      "step": 3001
    },
    {
      "epoch": 0.3744542846451291,
      "grad_norm": 0.0,
      "learning_rate": 1.4397504971916491e-05,
      "loss": 0.9757,
      "step": 3002
    },
    {
      "epoch": 0.3745790195833853,
      "grad_norm": 0.0,
      "learning_rate": 1.4393876106006259e-05,
      "loss": 0.9475,
      "step": 3003
    },
    {
      "epoch": 0.3747037545216415,
      "grad_norm": 0.0,
      "learning_rate": 1.4390246522904681e-05,
      "loss": 1.0173,
      "step": 3004
    },
    {
      "epoch": 0.3748284894598977,
      "grad_norm": 0.0,
      "learning_rate": 1.4386616223204189e-05,
      "loss": 0.9797,
      "step": 3005
    },
    {
      "epoch": 0.37495322439815393,
      "grad_norm": 0.0,
      "learning_rate": 1.4382985207497344e-05,
      "loss": 1.0025,
      "step": 3006
    },
    {
      "epoch": 0.37507795933641014,
      "grad_norm": 0.0,
      "learning_rate": 1.4379353476376821e-05,
      "loss": 1.0467,
      "step": 3007
    },
    {
      "epoch": 0.37520269427466635,
      "grad_norm": 0.0,
      "learning_rate": 1.4375721030435408e-05,
      "loss": 0.9644,
      "step": 3008
    },
    {
      "epoch": 0.37532742921292256,
      "grad_norm": 0.0,
      "learning_rate": 1.437208787026601e-05,
      "loss": 0.987,
      "step": 3009
    },
    {
      "epoch": 0.37545216415117877,
      "grad_norm": 0.0,
      "learning_rate": 1.4368453996461651e-05,
      "loss": 1.0218,
      "step": 3010
    },
    {
      "epoch": 0.3755768990894349,
      "grad_norm": 0.0,
      "learning_rate": 1.436481940961547e-05,
      "loss": 1.0074,
      "step": 3011
    },
    {
      "epoch": 0.37570163402769113,
      "grad_norm": 0.0,
      "learning_rate": 1.4361184110320729e-05,
      "loss": 1.0206,
      "step": 3012
    },
    {
      "epoch": 0.37582636896594734,
      "grad_norm": 0.0,
      "learning_rate": 1.4357548099170794e-05,
      "loss": 0.9973,
      "step": 3013
    },
    {
      "epoch": 0.37595110390420355,
      "grad_norm": 0.0,
      "learning_rate": 1.435391137675916e-05,
      "loss": 0.9567,
      "step": 3014
    },
    {
      "epoch": 0.37607583884245976,
      "grad_norm": 0.0,
      "learning_rate": 1.4350273943679422e-05,
      "loss": 1.027,
      "step": 3015
    },
    {
      "epoch": 0.376200573780716,
      "grad_norm": 0.0,
      "learning_rate": 1.434663580052531e-05,
      "loss": 1.0106,
      "step": 3016
    },
    {
      "epoch": 0.3763253087189722,
      "grad_norm": 0.0,
      "learning_rate": 1.4342996947890657e-05,
      "loss": 0.9757,
      "step": 3017
    },
    {
      "epoch": 0.3764500436572284,
      "grad_norm": 0.0,
      "learning_rate": 1.4339357386369414e-05,
      "loss": 0.9881,
      "step": 3018
    },
    {
      "epoch": 0.3765747785954846,
      "grad_norm": 0.0,
      "learning_rate": 1.4335717116555654e-05,
      "loss": 0.937,
      "step": 3019
    },
    {
      "epoch": 0.3766995135337408,
      "grad_norm": 0.0,
      "learning_rate": 1.4332076139043557e-05,
      "loss": 1.0019,
      "step": 3020
    },
    {
      "epoch": 0.376824248471997,
      "grad_norm": 0.0,
      "learning_rate": 1.432843445442742e-05,
      "loss": 1.0064,
      "step": 3021
    },
    {
      "epoch": 0.37694898341025324,
      "grad_norm": 0.0,
      "learning_rate": 1.4324792063301662e-05,
      "loss": 0.9876,
      "step": 3022
    },
    {
      "epoch": 0.3770737183485094,
      "grad_norm": 0.0,
      "learning_rate": 1.4321148966260814e-05,
      "loss": 0.986,
      "step": 3023
    },
    {
      "epoch": 0.3771984532867656,
      "grad_norm": 0.0,
      "learning_rate": 1.4317505163899518e-05,
      "loss": 0.9915,
      "step": 3024
    },
    {
      "epoch": 0.3773231882250218,
      "grad_norm": 0.0,
      "learning_rate": 1.4313860656812537e-05,
      "loss": 1.025,
      "step": 3025
    },
    {
      "epoch": 0.377447923163278,
      "grad_norm": 0.0,
      "learning_rate": 1.4310215445594741e-05,
      "loss": 0.9767,
      "step": 3026
    },
    {
      "epoch": 0.37757265810153423,
      "grad_norm": 0.0,
      "learning_rate": 1.430656953084113e-05,
      "loss": 0.971,
      "step": 3027
    },
    {
      "epoch": 0.37769739303979044,
      "grad_norm": 0.0,
      "learning_rate": 1.43029229131468e-05,
      "loss": 0.9737,
      "step": 3028
    },
    {
      "epoch": 0.37782212797804665,
      "grad_norm": 0.0,
      "learning_rate": 1.4299275593106973e-05,
      "loss": 1.0346,
      "step": 3029
    },
    {
      "epoch": 0.37794686291630286,
      "grad_norm": 0.0,
      "learning_rate": 1.4295627571316989e-05,
      "loss": 1.0182,
      "step": 3030
    },
    {
      "epoch": 0.37807159785455907,
      "grad_norm": 0.0,
      "learning_rate": 1.4291978848372292e-05,
      "loss": 0.9776,
      "step": 3031
    },
    {
      "epoch": 0.3781963327928153,
      "grad_norm": 0.0,
      "learning_rate": 1.4288329424868446e-05,
      "loss": 0.9522,
      "step": 3032
    },
    {
      "epoch": 0.3783210677310715,
      "grad_norm": 0.0,
      "learning_rate": 1.428467930140113e-05,
      "loss": 0.9716,
      "step": 3033
    },
    {
      "epoch": 0.3784458026693277,
      "grad_norm": 0.0,
      "learning_rate": 1.4281028478566135e-05,
      "loss": 1.0126,
      "step": 3034
    },
    {
      "epoch": 0.37857053760758386,
      "grad_norm": 0.0,
      "learning_rate": 1.4277376956959368e-05,
      "loss": 1.0134,
      "step": 3035
    },
    {
      "epoch": 0.37869527254584007,
      "grad_norm": 0.0,
      "learning_rate": 1.427372473717685e-05,
      "loss": 0.9809,
      "step": 3036
    },
    {
      "epoch": 0.3788200074840963,
      "grad_norm": 0.0,
      "learning_rate": 1.4270071819814717e-05,
      "loss": 0.9358,
      "step": 3037
    },
    {
      "epoch": 0.3789447424223525,
      "grad_norm": 0.0,
      "learning_rate": 1.4266418205469212e-05,
      "loss": 0.9935,
      "step": 3038
    },
    {
      "epoch": 0.3790694773606087,
      "grad_norm": 0.0,
      "learning_rate": 1.42627638947367e-05,
      "loss": 0.9901,
      "step": 3039
    },
    {
      "epoch": 0.3791942122988649,
      "grad_norm": 0.0,
      "learning_rate": 1.425910888821366e-05,
      "loss": 0.9794,
      "step": 3040
    },
    {
      "epoch": 0.3793189472371211,
      "grad_norm": 0.0,
      "learning_rate": 1.4255453186496674e-05,
      "loss": 1.0176,
      "step": 3041
    },
    {
      "epoch": 0.37944368217537733,
      "grad_norm": 0.0,
      "learning_rate": 1.4251796790182453e-05,
      "loss": 0.9847,
      "step": 3042
    },
    {
      "epoch": 0.37956841711363354,
      "grad_norm": 0.0,
      "learning_rate": 1.4248139699867807e-05,
      "loss": 0.9724,
      "step": 3043
    },
    {
      "epoch": 0.37969315205188975,
      "grad_norm": 0.0,
      "learning_rate": 1.4244481916149665e-05,
      "loss": 0.9954,
      "step": 3044
    },
    {
      "epoch": 0.37981788699014596,
      "grad_norm": 0.0,
      "learning_rate": 1.4240823439625074e-05,
      "loss": 0.9797,
      "step": 3045
    },
    {
      "epoch": 0.37994262192840217,
      "grad_norm": 0.0,
      "learning_rate": 1.4237164270891185e-05,
      "loss": 1.0049,
      "step": 3046
    },
    {
      "epoch": 0.3800673568666583,
      "grad_norm": 0.0,
      "learning_rate": 1.4233504410545271e-05,
      "loss": 0.9807,
      "step": 3047
    },
    {
      "epoch": 0.38019209180491453,
      "grad_norm": 0.0,
      "learning_rate": 1.4229843859184709e-05,
      "loss": 1.0228,
      "step": 3048
    },
    {
      "epoch": 0.38031682674317074,
      "grad_norm": 0.0,
      "learning_rate": 1.4226182617406996e-05,
      "loss": 1.0118,
      "step": 3049
    },
    {
      "epoch": 0.38044156168142695,
      "grad_norm": 0.0,
      "learning_rate": 1.4222520685809739e-05,
      "loss": 0.9834,
      "step": 3050
    },
    {
      "epoch": 0.38056629661968316,
      "grad_norm": 0.0,
      "learning_rate": 1.4218858064990655e-05,
      "loss": 0.9491,
      "step": 3051
    },
    {
      "epoch": 0.3806910315579394,
      "grad_norm": 0.0,
      "learning_rate": 1.421519475554758e-05,
      "loss": 1.0132,
      "step": 3052
    },
    {
      "epoch": 0.3808157664961956,
      "grad_norm": 0.0,
      "learning_rate": 1.4211530758078455e-05,
      "loss": 0.9888,
      "step": 3053
    },
    {
      "epoch": 0.3809405014344518,
      "grad_norm": 0.0,
      "learning_rate": 1.4207866073181343e-05,
      "loss": 1.0014,
      "step": 3054
    },
    {
      "epoch": 0.381065236372708,
      "grad_norm": 0.0,
      "learning_rate": 1.4204200701454403e-05,
      "loss": 1.0023,
      "step": 3055
    },
    {
      "epoch": 0.3811899713109642,
      "grad_norm": 0.0,
      "learning_rate": 1.420053464349592e-05,
      "loss": 1.0,
      "step": 3056
    },
    {
      "epoch": 0.3813147062492204,
      "grad_norm": 0.0,
      "learning_rate": 1.4196867899904292e-05,
      "loss": 0.9602,
      "step": 3057
    },
    {
      "epoch": 0.38143944118747664,
      "grad_norm": 0.0,
      "learning_rate": 1.4193200471278019e-05,
      "loss": 1.0216,
      "step": 3058
    },
    {
      "epoch": 0.3815641761257328,
      "grad_norm": 0.0,
      "learning_rate": 1.4189532358215718e-05,
      "loss": 1.0133,
      "step": 3059
    },
    {
      "epoch": 0.381688911063989,
      "grad_norm": 0.0,
      "learning_rate": 1.4185863561316121e-05,
      "loss": 0.9568,
      "step": 3060
    },
    {
      "epoch": 0.3818136460022452,
      "grad_norm": 0.0,
      "learning_rate": 1.4182194081178065e-05,
      "loss": 0.9772,
      "step": 3061
    },
    {
      "epoch": 0.3819383809405014,
      "grad_norm": 0.0,
      "learning_rate": 1.4178523918400502e-05,
      "loss": 0.9679,
      "step": 3062
    },
    {
      "epoch": 0.38206311587875763,
      "grad_norm": 0.0,
      "learning_rate": 1.4174853073582498e-05,
      "loss": 0.9641,
      "step": 3063
    },
    {
      "epoch": 0.38218785081701384,
      "grad_norm": 0.0,
      "learning_rate": 1.4171181547323221e-05,
      "loss": 0.9453,
      "step": 3064
    },
    {
      "epoch": 0.38231258575527005,
      "grad_norm": 0.0,
      "learning_rate": 1.4167509340221961e-05,
      "loss": 0.9787,
      "step": 3065
    },
    {
      "epoch": 0.38243732069352626,
      "grad_norm": 0.0,
      "learning_rate": 1.4163836452878115e-05,
      "loss": 0.9799,
      "step": 3066
    },
    {
      "epoch": 0.38256205563178247,
      "grad_norm": 0.0,
      "learning_rate": 1.4160162885891193e-05,
      "loss": 0.9952,
      "step": 3067
    },
    {
      "epoch": 0.3826867905700387,
      "grad_norm": 0.0,
      "learning_rate": 1.4156488639860806e-05,
      "loss": 0.9505,
      "step": 3068
    },
    {
      "epoch": 0.3828115255082949,
      "grad_norm": 0.0,
      "learning_rate": 1.415281371538669e-05,
      "loss": 0.9478,
      "step": 3069
    },
    {
      "epoch": 0.3829362604465511,
      "grad_norm": 0.0,
      "learning_rate": 1.4149138113068687e-05,
      "loss": 0.9617,
      "step": 3070
    },
    {
      "epoch": 0.38306099538480726,
      "grad_norm": 0.0,
      "learning_rate": 1.4145461833506745e-05,
      "loss": 1.0045,
      "step": 3071
    },
    {
      "epoch": 0.38318573032306347,
      "grad_norm": 0.0,
      "learning_rate": 1.4141784877300921e-05,
      "loss": 0.9752,
      "step": 3072
    },
    {
      "epoch": 0.3833104652613197,
      "grad_norm": 0.0,
      "learning_rate": 1.4138107245051394e-05,
      "loss": 0.9869,
      "step": 3073
    },
    {
      "epoch": 0.3834352001995759,
      "grad_norm": 0.0,
      "learning_rate": 1.4134428937358441e-05,
      "loss": 0.9524,
      "step": 3074
    },
    {
      "epoch": 0.3835599351378321,
      "grad_norm": 0.0,
      "learning_rate": 1.4130749954822458e-05,
      "loss": 0.9767,
      "step": 3075
    },
    {
      "epoch": 0.3836846700760883,
      "grad_norm": 0.0,
      "learning_rate": 1.4127070298043949e-05,
      "loss": 1.0055,
      "step": 3076
    },
    {
      "epoch": 0.3838094050143445,
      "grad_norm": 0.0,
      "learning_rate": 1.4123389967623525e-05,
      "loss": 0.9509,
      "step": 3077
    },
    {
      "epoch": 0.38393413995260073,
      "grad_norm": 0.0,
      "learning_rate": 1.4119708964161905e-05,
      "loss": 0.9935,
      "step": 3078
    },
    {
      "epoch": 0.38405887489085694,
      "grad_norm": 0.0,
      "learning_rate": 1.4116027288259924e-05,
      "loss": 1.002,
      "step": 3079
    },
    {
      "epoch": 0.38418360982911315,
      "grad_norm": 0.0,
      "learning_rate": 1.4112344940518528e-05,
      "loss": 0.9392,
      "step": 3080
    },
    {
      "epoch": 0.38430834476736936,
      "grad_norm": 0.0,
      "learning_rate": 1.4108661921538766e-05,
      "loss": 0.9782,
      "step": 3081
    },
    {
      "epoch": 0.38443307970562557,
      "grad_norm": 0.0,
      "learning_rate": 1.4104978231921798e-05,
      "loss": 1.0338,
      "step": 3082
    },
    {
      "epoch": 0.3845578146438817,
      "grad_norm": 0.0,
      "learning_rate": 1.4101293872268897e-05,
      "loss": 1.0271,
      "step": 3083
    },
    {
      "epoch": 0.38468254958213793,
      "grad_norm": 0.0,
      "learning_rate": 1.4097608843181445e-05,
      "loss": 0.9719,
      "step": 3084
    },
    {
      "epoch": 0.38480728452039414,
      "grad_norm": 0.0,
      "learning_rate": 1.4093923145260926e-05,
      "loss": 0.9556,
      "step": 3085
    },
    {
      "epoch": 0.38493201945865035,
      "grad_norm": 0.0,
      "learning_rate": 1.4090236779108946e-05,
      "loss": 0.9877,
      "step": 3086
    },
    {
      "epoch": 0.38505675439690656,
      "grad_norm": 0.0,
      "learning_rate": 1.408654974532721e-05,
      "loss": 0.9998,
      "step": 3087
    },
    {
      "epoch": 0.3851814893351628,
      "grad_norm": 0.0,
      "learning_rate": 1.4082862044517529e-05,
      "loss": 0.9605,
      "step": 3088
    },
    {
      "epoch": 0.385306224273419,
      "grad_norm": 0.0,
      "learning_rate": 1.4079173677281836e-05,
      "loss": 0.9805,
      "step": 3089
    },
    {
      "epoch": 0.3854309592116752,
      "grad_norm": 0.0,
      "learning_rate": 1.4075484644222166e-05,
      "loss": 0.9659,
      "step": 3090
    },
    {
      "epoch": 0.3855556941499314,
      "grad_norm": 0.0,
      "learning_rate": 1.4071794945940657e-05,
      "loss": 0.9757,
      "step": 3091
    },
    {
      "epoch": 0.3856804290881876,
      "grad_norm": 0.0,
      "learning_rate": 1.4068104583039563e-05,
      "loss": 0.9821,
      "step": 3092
    },
    {
      "epoch": 0.3858051640264438,
      "grad_norm": 0.0,
      "learning_rate": 1.4064413556121253e-05,
      "loss": 0.984,
      "step": 3093
    },
    {
      "epoch": 0.38592989896470004,
      "grad_norm": 0.0,
      "learning_rate": 1.4060721865788178e-05,
      "loss": 0.9969,
      "step": 3094
    },
    {
      "epoch": 0.38605463390295625,
      "grad_norm": 0.0,
      "learning_rate": 1.4057029512642926e-05,
      "loss": 0.94,
      "step": 3095
    },
    {
      "epoch": 0.3861793688412124,
      "grad_norm": 0.0,
      "learning_rate": 1.4053336497288183e-05,
      "loss": 0.9915,
      "step": 3096
    },
    {
      "epoch": 0.3863041037794686,
      "grad_norm": 0.0,
      "learning_rate": 1.4049642820326737e-05,
      "loss": 0.9828,
      "step": 3097
    },
    {
      "epoch": 0.3864288387177248,
      "grad_norm": 0.0,
      "learning_rate": 1.4045948482361494e-05,
      "loss": 0.981,
      "step": 3098
    },
    {
      "epoch": 0.38655357365598103,
      "grad_norm": 0.0,
      "learning_rate": 1.4042253483995459e-05,
      "loss": 0.9483,
      "step": 3099
    },
    {
      "epoch": 0.38667830859423724,
      "grad_norm": 0.0,
      "learning_rate": 1.4038557825831754e-05,
      "loss": 0.9687,
      "step": 3100
    },
    {
      "epoch": 0.38680304353249345,
      "grad_norm": 0.0,
      "learning_rate": 1.4034861508473595e-05,
      "loss": 1.0194,
      "step": 3101
    },
    {
      "epoch": 0.38692777847074966,
      "grad_norm": 0.0,
      "learning_rate": 1.4031164532524322e-05,
      "loss": 0.9242,
      "step": 3102
    },
    {
      "epoch": 0.38705251340900587,
      "grad_norm": 0.0,
      "learning_rate": 1.4027466898587375e-05,
      "loss": 0.9959,
      "step": 3103
    },
    {
      "epoch": 0.3871772483472621,
      "grad_norm": 0.0,
      "learning_rate": 1.4023768607266294e-05,
      "loss": 1.0282,
      "step": 3104
    },
    {
      "epoch": 0.3873019832855183,
      "grad_norm": 0.0,
      "learning_rate": 1.4020069659164741e-05,
      "loss": 0.9968,
      "step": 3105
    },
    {
      "epoch": 0.3874267182237745,
      "grad_norm": 0.0,
      "learning_rate": 1.4016370054886474e-05,
      "loss": 0.9893,
      "step": 3106
    },
    {
      "epoch": 0.3875514531620307,
      "grad_norm": 0.0,
      "learning_rate": 1.4012669795035359e-05,
      "loss": 0.9615,
      "step": 3107
    },
    {
      "epoch": 0.38767618810028687,
      "grad_norm": 0.0,
      "learning_rate": 1.4008968880215377e-05,
      "loss": 0.9916,
      "step": 3108
    },
    {
      "epoch": 0.3878009230385431,
      "grad_norm": 0.0,
      "learning_rate": 1.4005267311030608e-05,
      "loss": 0.9918,
      "step": 3109
    },
    {
      "epoch": 0.3879256579767993,
      "grad_norm": 0.0,
      "learning_rate": 1.4001565088085242e-05,
      "loss": 0.9782,
      "step": 3110
    },
    {
      "epoch": 0.3880503929150555,
      "grad_norm": 0.0,
      "learning_rate": 1.3997862211983573e-05,
      "loss": 0.9948,
      "step": 3111
    },
    {
      "epoch": 0.3881751278533117,
      "grad_norm": 0.0,
      "learning_rate": 1.3994158683330006e-05,
      "loss": 0.9972,
      "step": 3112
    },
    {
      "epoch": 0.3882998627915679,
      "grad_norm": 0.0,
      "learning_rate": 1.399045450272905e-05,
      "loss": 0.9498,
      "step": 3113
    },
    {
      "epoch": 0.38842459772982413,
      "grad_norm": 0.0,
      "learning_rate": 1.398674967078532e-05,
      "loss": 0.994,
      "step": 3114
    },
    {
      "epoch": 0.38854933266808034,
      "grad_norm": 0.0,
      "learning_rate": 1.3983044188103541e-05,
      "loss": 1.0164,
      "step": 3115
    },
    {
      "epoch": 0.38867406760633655,
      "grad_norm": 0.0,
      "learning_rate": 1.3979338055288536e-05,
      "loss": 0.987,
      "step": 3116
    },
    {
      "epoch": 0.38879880254459276,
      "grad_norm": 0.0,
      "learning_rate": 1.3975631272945243e-05,
      "loss": 0.9887,
      "step": 3117
    },
    {
      "epoch": 0.38892353748284897,
      "grad_norm": 0.0,
      "learning_rate": 1.3971923841678698e-05,
      "loss": 0.9671,
      "step": 3118
    },
    {
      "epoch": 0.3890482724211052,
      "grad_norm": 0.0,
      "learning_rate": 1.3968215762094054e-05,
      "loss": 0.9526,
      "step": 3119
    },
    {
      "epoch": 0.38917300735936133,
      "grad_norm": 0.0,
      "learning_rate": 1.3964507034796558e-05,
      "loss": 1.0171,
      "step": 3120
    },
    {
      "epoch": 0.38929774229761754,
      "grad_norm": 0.0,
      "learning_rate": 1.396079766039157e-05,
      "loss": 0.9535,
      "step": 3121
    },
    {
      "epoch": 0.38942247723587375,
      "grad_norm": 0.0,
      "learning_rate": 1.3957087639484552e-05,
      "loss": 0.9566,
      "step": 3122
    },
    {
      "epoch": 0.38954721217412996,
      "grad_norm": 0.0,
      "learning_rate": 1.3953376972681075e-05,
      "loss": 0.9827,
      "step": 3123
    },
    {
      "epoch": 0.3896719471123862,
      "grad_norm": 0.0,
      "learning_rate": 1.394966566058681e-05,
      "loss": 0.9889,
      "step": 3124
    },
    {
      "epoch": 0.3897966820506424,
      "grad_norm": 0.0,
      "learning_rate": 1.3945953703807538e-05,
      "loss": 0.9427,
      "step": 3125
    },
    {
      "epoch": 0.3899214169888986,
      "grad_norm": 0.0,
      "learning_rate": 1.394224110294915e-05,
      "loss": 0.9887,
      "step": 3126
    },
    {
      "epoch": 0.3900461519271548,
      "grad_norm": 0.0,
      "learning_rate": 1.3938527858617625e-05,
      "loss": 0.9784,
      "step": 3127
    },
    {
      "epoch": 0.390170886865411,
      "grad_norm": 0.0,
      "learning_rate": 1.3934813971419066e-05,
      "loss": 0.985,
      "step": 3128
    },
    {
      "epoch": 0.3902956218036672,
      "grad_norm": 0.0,
      "learning_rate": 1.3931099441959671e-05,
      "loss": 1.0162,
      "step": 3129
    },
    {
      "epoch": 0.39042035674192344,
      "grad_norm": 0.0,
      "learning_rate": 1.3927384270845744e-05,
      "loss": 0.9494,
      "step": 3130
    },
    {
      "epoch": 0.39054509168017965,
      "grad_norm": 0.0,
      "learning_rate": 1.3923668458683692e-05,
      "loss": 0.9986,
      "step": 3131
    },
    {
      "epoch": 0.3906698266184358,
      "grad_norm": 0.0,
      "learning_rate": 1.3919952006080037e-05,
      "loss": 0.9688,
      "step": 3132
    },
    {
      "epoch": 0.390794561556692,
      "grad_norm": 0.0,
      "learning_rate": 1.3916234913641393e-05,
      "loss": 1.0352,
      "step": 3133
    },
    {
      "epoch": 0.3909192964949482,
      "grad_norm": 0.0,
      "learning_rate": 1.391251718197448e-05,
      "loss": 0.9956,
      "step": 3134
    },
    {
      "epoch": 0.39104403143320443,
      "grad_norm": 0.0,
      "learning_rate": 1.3908798811686131e-05,
      "loss": 1.0133,
      "step": 3135
    },
    {
      "epoch": 0.39116876637146064,
      "grad_norm": 0.0,
      "learning_rate": 1.3905079803383277e-05,
      "loss": 1.0227,
      "step": 3136
    },
    {
      "epoch": 0.39129350130971685,
      "grad_norm": 0.0,
      "learning_rate": 1.390136015767295e-05,
      "loss": 0.9668,
      "step": 3137
    },
    {
      "epoch": 0.39141823624797306,
      "grad_norm": 0.0,
      "learning_rate": 1.3897639875162297e-05,
      "loss": 0.9886,
      "step": 3138
    },
    {
      "epoch": 0.39154297118622927,
      "grad_norm": 0.0,
      "learning_rate": 1.3893918956458554e-05,
      "loss": 1.0659,
      "step": 3139
    },
    {
      "epoch": 0.3916677061244855,
      "grad_norm": 0.0,
      "learning_rate": 1.3890197402169073e-05,
      "loss": 0.9825,
      "step": 3140
    },
    {
      "epoch": 0.3917924410627417,
      "grad_norm": 0.0,
      "learning_rate": 1.3886475212901301e-05,
      "loss": 0.9552,
      "step": 3141
    },
    {
      "epoch": 0.3919171760009979,
      "grad_norm": 0.0,
      "learning_rate": 1.38827523892628e-05,
      "loss": 0.9722,
      "step": 3142
    },
    {
      "epoch": 0.3920419109392541,
      "grad_norm": 0.0,
      "learning_rate": 1.3879028931861227e-05,
      "loss": 1.0113,
      "step": 3143
    },
    {
      "epoch": 0.39216664587751027,
      "grad_norm": 0.0,
      "learning_rate": 1.387530484130434e-05,
      "loss": 0.9917,
      "step": 3144
    },
    {
      "epoch": 0.3922913808157665,
      "grad_norm": 0.0,
      "learning_rate": 1.3871580118200007e-05,
      "loss": 0.9677,
      "step": 3145
    },
    {
      "epoch": 0.3924161157540227,
      "grad_norm": 0.0,
      "learning_rate": 1.3867854763156198e-05,
      "loss": 1.0043,
      "step": 3146
    },
    {
      "epoch": 0.3925408506922789,
      "grad_norm": 0.0,
      "learning_rate": 1.3864128776780983e-05,
      "loss": 1.0103,
      "step": 3147
    },
    {
      "epoch": 0.3926655856305351,
      "grad_norm": 0.0,
      "learning_rate": 1.3860402159682535e-05,
      "loss": 0.9191,
      "step": 3148
    },
    {
      "epoch": 0.3927903205687913,
      "grad_norm": 0.0,
      "learning_rate": 1.385667491246914e-05,
      "loss": 1.0082,
      "step": 3149
    },
    {
      "epoch": 0.39291505550704753,
      "grad_norm": 0.0,
      "learning_rate": 1.385294703574917e-05,
      "loss": 1.0177,
      "step": 3150
    },
    {
      "epoch": 0.39303979044530374,
      "grad_norm": 0.0,
      "learning_rate": 1.3849218530131112e-05,
      "loss": 1.0003,
      "step": 3151
    },
    {
      "epoch": 0.39316452538355995,
      "grad_norm": 0.0,
      "learning_rate": 1.3845489396223552e-05,
      "loss": 0.9592,
      "step": 3152
    },
    {
      "epoch": 0.39328926032181616,
      "grad_norm": 0.0,
      "learning_rate": 1.3841759634635177e-05,
      "loss": 0.994,
      "step": 3153
    },
    {
      "epoch": 0.39341399526007237,
      "grad_norm": 0.0,
      "learning_rate": 1.383802924597478e-05,
      "loss": 0.9481,
      "step": 3154
    },
    {
      "epoch": 0.3935387301983286,
      "grad_norm": 0.0,
      "learning_rate": 1.3834298230851252e-05,
      "loss": 1.0097,
      "step": 3155
    },
    {
      "epoch": 0.39366346513658473,
      "grad_norm": 0.0,
      "learning_rate": 1.3830566589873595e-05,
      "loss": 1.0487,
      "step": 3156
    },
    {
      "epoch": 0.39378820007484094,
      "grad_norm": 0.0,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.9492,
      "step": 3157
    },
    {
      "epoch": 0.39391293501309715,
      "grad_norm": 0.0,
      "learning_rate": 1.3823101432792368e-05,
      "loss": 0.9724,
      "step": 3158
    },
    {
      "epoch": 0.39403766995135336,
      "grad_norm": 0.0,
      "learning_rate": 1.3819367917907304e-05,
      "loss": 1.037,
      "step": 3159
    },
    {
      "epoch": 0.3941624048896096,
      "grad_norm": 0.0,
      "learning_rate": 1.381563377960511e-05,
      "loss": 1.0117,
      "step": 3160
    },
    {
      "epoch": 0.3942871398278658,
      "grad_norm": 0.0,
      "learning_rate": 1.3811899018495291e-05,
      "loss": 0.9664,
      "step": 3161
    },
    {
      "epoch": 0.394411874766122,
      "grad_norm": 0.0,
      "learning_rate": 1.3808163635187454e-05,
      "loss": 1.0267,
      "step": 3162
    },
    {
      "epoch": 0.3945366097043782,
      "grad_norm": 0.0,
      "learning_rate": 1.380442763029131e-05,
      "loss": 0.9626,
      "step": 3163
    },
    {
      "epoch": 0.3946613446426344,
      "grad_norm": 0.0,
      "learning_rate": 1.3800691004416669e-05,
      "loss": 0.952,
      "step": 3164
    },
    {
      "epoch": 0.3947860795808906,
      "grad_norm": 0.0,
      "learning_rate": 1.379695375817344e-05,
      "loss": 0.9734,
      "step": 3165
    },
    {
      "epoch": 0.39491081451914684,
      "grad_norm": 0.0,
      "learning_rate": 1.3793215892171636e-05,
      "loss": 0.9918,
      "step": 3166
    },
    {
      "epoch": 0.39503554945740305,
      "grad_norm": 0.0,
      "learning_rate": 1.3789477407021372e-05,
      "loss": 0.9655,
      "step": 3167
    },
    {
      "epoch": 0.3951602843956592,
      "grad_norm": 0.0,
      "learning_rate": 1.3785738303332862e-05,
      "loss": 0.9759,
      "step": 3168
    },
    {
      "epoch": 0.3952850193339154,
      "grad_norm": 0.0,
      "learning_rate": 1.3781998581716427e-05,
      "loss": 0.9183,
      "step": 3169
    },
    {
      "epoch": 0.3954097542721716,
      "grad_norm": 0.0,
      "learning_rate": 1.3778258242782477e-05,
      "loss": 1.0107,
      "step": 3170
    },
    {
      "epoch": 0.39553448921042783,
      "grad_norm": 0.0,
      "learning_rate": 1.3774517287141531e-05,
      "loss": 0.9764,
      "step": 3171
    },
    {
      "epoch": 0.39565922414868404,
      "grad_norm": 0.0,
      "learning_rate": 1.3770775715404212e-05,
      "loss": 0.9632,
      "step": 3172
    },
    {
      "epoch": 0.39578395908694025,
      "grad_norm": 0.0,
      "learning_rate": 1.3767033528181232e-05,
      "loss": 0.9521,
      "step": 3173
    },
    {
      "epoch": 0.39590869402519646,
      "grad_norm": 0.0,
      "learning_rate": 1.3763290726083419e-05,
      "loss": 0.9208,
      "step": 3174
    },
    {
      "epoch": 0.39603342896345267,
      "grad_norm": 0.0,
      "learning_rate": 1.3759547309721681e-05,
      "loss": 1.043,
      "step": 3175
    },
    {
      "epoch": 0.3961581639017089,
      "grad_norm": 0.0,
      "learning_rate": 1.375580327970705e-05,
      "loss": 0.9707,
      "step": 3176
    },
    {
      "epoch": 0.3962828988399651,
      "grad_norm": 0.0,
      "learning_rate": 1.3752058636650638e-05,
      "loss": 1.0261,
      "step": 3177
    },
    {
      "epoch": 0.3964076337782213,
      "grad_norm": 0.0,
      "learning_rate": 1.3748313381163667e-05,
      "loss": 0.9977,
      "step": 3178
    },
    {
      "epoch": 0.3965323687164775,
      "grad_norm": 0.0,
      "learning_rate": 1.374456751385746e-05,
      "loss": 1.0012,
      "step": 3179
    },
    {
      "epoch": 0.39665710365473367,
      "grad_norm": 0.0,
      "learning_rate": 1.3740821035343432e-05,
      "loss": 0.9868,
      "step": 3180
    },
    {
      "epoch": 0.3967818385929899,
      "grad_norm": 0.0,
      "learning_rate": 1.3737073946233106e-05,
      "loss": 1.01,
      "step": 3181
    },
    {
      "epoch": 0.3969065735312461,
      "grad_norm": 0.0,
      "learning_rate": 1.37333262471381e-05,
      "loss": 0.962,
      "step": 3182
    },
    {
      "epoch": 0.3970313084695023,
      "grad_norm": 0.0,
      "learning_rate": 1.372957793867014e-05,
      "loss": 1.0073,
      "step": 3183
    },
    {
      "epoch": 0.3971560434077585,
      "grad_norm": 0.0,
      "learning_rate": 1.372582902144103e-05,
      "loss": 0.9918,
      "step": 3184
    },
    {
      "epoch": 0.3972807783460147,
      "grad_norm": 0.0,
      "learning_rate": 1.3722079496062702e-05,
      "loss": 0.9594,
      "step": 3185
    },
    {
      "epoch": 0.39740551328427093,
      "grad_norm": 0.0,
      "learning_rate": 1.3718329363147165e-05,
      "loss": 0.9517,
      "step": 3186
    },
    {
      "epoch": 0.39753024822252714,
      "grad_norm": 0.0,
      "learning_rate": 1.3714578623306535e-05,
      "loss": 0.9567,
      "step": 3187
    },
    {
      "epoch": 0.39765498316078335,
      "grad_norm": 0.0,
      "learning_rate": 1.371082727715303e-05,
      "loss": 0.9545,
      "step": 3188
    },
    {
      "epoch": 0.39777971809903956,
      "grad_norm": 0.0,
      "learning_rate": 1.3707075325298971e-05,
      "loss": 1.0223,
      "step": 3189
    },
    {
      "epoch": 0.39790445303729577,
      "grad_norm": 0.0,
      "learning_rate": 1.3703322768356758e-05,
      "loss": 0.9758,
      "step": 3190
    },
    {
      "epoch": 0.398029187975552,
      "grad_norm": 0.0,
      "learning_rate": 1.3699569606938909e-05,
      "loss": 0.9617,
      "step": 3191
    },
    {
      "epoch": 0.39815392291380813,
      "grad_norm": 0.0,
      "learning_rate": 1.3695815841658035e-05,
      "loss": 0.9634,
      "step": 3192
    },
    {
      "epoch": 0.39827865785206434,
      "grad_norm": 0.0,
      "learning_rate": 1.3692061473126845e-05,
      "loss": 0.9851,
      "step": 3193
    },
    {
      "epoch": 0.39840339279032055,
      "grad_norm": 0.0,
      "learning_rate": 1.3688306501958149e-05,
      "loss": 1.0094,
      "step": 3194
    },
    {
      "epoch": 0.39852812772857676,
      "grad_norm": 0.0,
      "learning_rate": 1.3684550928764846e-05,
      "loss": 0.965,
      "step": 3195
    },
    {
      "epoch": 0.398652862666833,
      "grad_norm": 0.0,
      "learning_rate": 1.3680794754159947e-05,
      "loss": 0.9625,
      "step": 3196
    },
    {
      "epoch": 0.3987775976050892,
      "grad_norm": 0.0,
      "learning_rate": 1.3677037978756551e-05,
      "loss": 0.9661,
      "step": 3197
    },
    {
      "epoch": 0.3989023325433454,
      "grad_norm": 0.0,
      "learning_rate": 1.3673280603167856e-05,
      "loss": 0.9705,
      "step": 3198
    },
    {
      "epoch": 0.3990270674816016,
      "grad_norm": 0.0,
      "learning_rate": 1.3669522628007168e-05,
      "loss": 0.9842,
      "step": 3199
    },
    {
      "epoch": 0.3991518024198578,
      "grad_norm": 0.0,
      "learning_rate": 1.366576405388788e-05,
      "loss": 0.9805,
      "step": 3200
    },
    {
      "epoch": 0.399276537358114,
      "grad_norm": 0.0,
      "learning_rate": 1.3662004881423481e-05,
      "loss": 0.9413,
      "step": 3201
    },
    {
      "epoch": 0.39940127229637024,
      "grad_norm": 0.0,
      "learning_rate": 1.3658245111227571e-05,
      "loss": 0.9387,
      "step": 3202
    },
    {
      "epoch": 0.39952600723462645,
      "grad_norm": 0.0,
      "learning_rate": 1.3654484743913829e-05,
      "loss": 1.0036,
      "step": 3203
    },
    {
      "epoch": 0.3996507421728826,
      "grad_norm": 0.0,
      "learning_rate": 1.3650723780096051e-05,
      "loss": 0.9709,
      "step": 3204
    },
    {
      "epoch": 0.3997754771111388,
      "grad_norm": 0.0,
      "learning_rate": 1.364696222038812e-05,
      "loss": 0.9495,
      "step": 3205
    },
    {
      "epoch": 0.399900212049395,
      "grad_norm": 0.0,
      "learning_rate": 1.3643200065404015e-05,
      "loss": 0.9752,
      "step": 3206
    },
    {
      "epoch": 0.40002494698765123,
      "grad_norm": 0.0,
      "learning_rate": 1.3639437315757813e-05,
      "loss": 0.9768,
      "step": 3207
    },
    {
      "epoch": 0.40014968192590744,
      "grad_norm": 0.0,
      "learning_rate": 1.363567397206369e-05,
      "loss": 0.9364,
      "step": 3208
    },
    {
      "epoch": 0.40027441686416365,
      "grad_norm": 0.0,
      "learning_rate": 1.3631910034935923e-05,
      "loss": 0.966,
      "step": 3209
    },
    {
      "epoch": 0.40039915180241986,
      "grad_norm": 0.0,
      "learning_rate": 1.3628145504988875e-05,
      "loss": 1.0138,
      "step": 3210
    },
    {
      "epoch": 0.40052388674067607,
      "grad_norm": 0.0,
      "learning_rate": 1.3624380382837017e-05,
      "loss": 0.9138,
      "step": 3211
    },
    {
      "epoch": 0.4006486216789323,
      "grad_norm": 0.0,
      "learning_rate": 1.3620614669094915e-05,
      "loss": 1.0184,
      "step": 3212
    },
    {
      "epoch": 0.4007733566171885,
      "grad_norm": 0.0,
      "learning_rate": 1.3616848364377217e-05,
      "loss": 0.9764,
      "step": 3213
    },
    {
      "epoch": 0.4008980915554447,
      "grad_norm": 0.0,
      "learning_rate": 1.3613081469298687e-05,
      "loss": 1.0089,
      "step": 3214
    },
    {
      "epoch": 0.4010228264937009,
      "grad_norm": 0.0,
      "learning_rate": 1.3609313984474178e-05,
      "loss": 0.992,
      "step": 3215
    },
    {
      "epoch": 0.40114756143195707,
      "grad_norm": 0.0,
      "learning_rate": 1.3605545910518636e-05,
      "loss": 0.9763,
      "step": 3216
    },
    {
      "epoch": 0.4012722963702133,
      "grad_norm": 0.0,
      "learning_rate": 1.3601777248047105e-05,
      "loss": 0.9488,
      "step": 3217
    },
    {
      "epoch": 0.4013970313084695,
      "grad_norm": 0.0,
      "learning_rate": 1.3598007997674726e-05,
      "loss": 1.0048,
      "step": 3218
    },
    {
      "epoch": 0.4015217662467257,
      "grad_norm": 0.0,
      "learning_rate": 1.3594238160016741e-05,
      "loss": 1.0139,
      "step": 3219
    },
    {
      "epoch": 0.4016465011849819,
      "grad_norm": 0.0,
      "learning_rate": 1.3590467735688475e-05,
      "loss": 0.9416,
      "step": 3220
    },
    {
      "epoch": 0.4017712361232381,
      "grad_norm": 0.0,
      "learning_rate": 1.3586696725305357e-05,
      "loss": 0.9595,
      "step": 3221
    },
    {
      "epoch": 0.40189597106149433,
      "grad_norm": 0.0,
      "learning_rate": 1.358292512948292e-05,
      "loss": 1.0081,
      "step": 3222
    },
    {
      "epoch": 0.40202070599975054,
      "grad_norm": 0.0,
      "learning_rate": 1.357915294883677e-05,
      "loss": 0.9855,
      "step": 3223
    },
    {
      "epoch": 0.40214544093800675,
      "grad_norm": 0.0,
      "learning_rate": 1.3575380183982633e-05,
      "loss": 0.9934,
      "step": 3224
    },
    {
      "epoch": 0.40227017587626296,
      "grad_norm": 0.0,
      "learning_rate": 1.3571606835536315e-05,
      "loss": 1.0198,
      "step": 3225
    },
    {
      "epoch": 0.40239491081451917,
      "grad_norm": 0.0,
      "learning_rate": 1.3567832904113718e-05,
      "loss": 0.9924,
      "step": 3226
    },
    {
      "epoch": 0.4025196457527754,
      "grad_norm": 0.0,
      "learning_rate": 1.356405839033085e-05,
      "loss": 0.966,
      "step": 3227
    },
    {
      "epoch": 0.40264438069103153,
      "grad_norm": 0.0,
      "learning_rate": 1.3560283294803801e-05,
      "loss": 1.004,
      "step": 3228
    },
    {
      "epoch": 0.40276911562928774,
      "grad_norm": 0.0,
      "learning_rate": 1.3556507618148769e-05,
      "loss": 0.9422,
      "step": 3229
    },
    {
      "epoch": 0.40289385056754395,
      "grad_norm": 0.0,
      "learning_rate": 1.3552731360982028e-05,
      "loss": 0.9666,
      "step": 3230
    },
    {
      "epoch": 0.40301858550580016,
      "grad_norm": 0.0,
      "learning_rate": 1.3548954523919969e-05,
      "loss": 1.0024,
      "step": 3231
    },
    {
      "epoch": 0.4031433204440564,
      "grad_norm": 0.0,
      "learning_rate": 1.3545177107579062e-05,
      "loss": 0.9876,
      "step": 3232
    },
    {
      "epoch": 0.4032680553823126,
      "grad_norm": 0.0,
      "learning_rate": 1.3541399112575877e-05,
      "loss": 0.9556,
      "step": 3233
    },
    {
      "epoch": 0.4033927903205688,
      "grad_norm": 0.0,
      "learning_rate": 1.3537620539527079e-05,
      "loss": 0.9703,
      "step": 3234
    },
    {
      "epoch": 0.403517525258825,
      "grad_norm": 0.0,
      "learning_rate": 1.3533841389049429e-05,
      "loss": 0.997,
      "step": 3235
    },
    {
      "epoch": 0.4036422601970812,
      "grad_norm": 0.0,
      "learning_rate": 1.3530061661759772e-05,
      "loss": 0.9855,
      "step": 3236
    },
    {
      "epoch": 0.4037669951353374,
      "grad_norm": 0.0,
      "learning_rate": 1.3526281358275061e-05,
      "loss": 0.9595,
      "step": 3237
    },
    {
      "epoch": 0.40389173007359364,
      "grad_norm": 0.0,
      "learning_rate": 1.3522500479212337e-05,
      "loss": 0.977,
      "step": 3238
    },
    {
      "epoch": 0.40401646501184985,
      "grad_norm": 0.0,
      "learning_rate": 1.3518719025188734e-05,
      "loss": 0.9571,
      "step": 3239
    },
    {
      "epoch": 0.404141199950106,
      "grad_norm": 0.0,
      "learning_rate": 1.3514936996821475e-05,
      "loss": 0.9967,
      "step": 3240
    },
    {
      "epoch": 0.4042659348883622,
      "grad_norm": 0.0,
      "learning_rate": 1.351115439472789e-05,
      "loss": 0.9753,
      "step": 3241
    },
    {
      "epoch": 0.4043906698266184,
      "grad_norm": 0.0,
      "learning_rate": 1.3507371219525391e-05,
      "loss": 0.9819,
      "step": 3242
    },
    {
      "epoch": 0.40451540476487463,
      "grad_norm": 0.0,
      "learning_rate": 1.350358747183149e-05,
      "loss": 0.9919,
      "step": 3243
    },
    {
      "epoch": 0.40464013970313084,
      "grad_norm": 0.0,
      "learning_rate": 1.3499803152263787e-05,
      "loss": 1.041,
      "step": 3244
    },
    {
      "epoch": 0.40476487464138705,
      "grad_norm": 0.0,
      "learning_rate": 1.3496018261439985e-05,
      "loss": 0.9784,
      "step": 3245
    },
    {
      "epoch": 0.40488960957964326,
      "grad_norm": 0.0,
      "learning_rate": 1.3492232799977862e-05,
      "loss": 0.9538,
      "step": 3246
    },
    {
      "epoch": 0.40501434451789947,
      "grad_norm": 0.0,
      "learning_rate": 1.3488446768495309e-05,
      "loss": 0.9814,
      "step": 3247
    },
    {
      "epoch": 0.4051390794561557,
      "grad_norm": 0.0,
      "learning_rate": 1.3484660167610302e-05,
      "loss": 1.0288,
      "step": 3248
    },
    {
      "epoch": 0.4052638143944119,
      "grad_norm": 0.0,
      "learning_rate": 1.3480872997940906e-05,
      "loss": 1.0033,
      "step": 3249
    },
    {
      "epoch": 0.4053885493326681,
      "grad_norm": 0.0,
      "learning_rate": 1.3477085260105284e-05,
      "loss": 0.9826,
      "step": 3250
    },
    {
      "epoch": 0.4055132842709243,
      "grad_norm": 0.0,
      "learning_rate": 1.3473296954721695e-05,
      "loss": 1.0501,
      "step": 3251
    },
    {
      "epoch": 0.40563801920918047,
      "grad_norm": 0.0,
      "learning_rate": 1.3469508082408478e-05,
      "loss": 0.982,
      "step": 3252
    },
    {
      "epoch": 0.4057627541474367,
      "grad_norm": 0.0,
      "learning_rate": 1.3465718643784076e-05,
      "loss": 0.978,
      "step": 3253
    },
    {
      "epoch": 0.4058874890856929,
      "grad_norm": 0.0,
      "learning_rate": 1.3461928639467021e-05,
      "loss": 0.9963,
      "step": 3254
    },
    {
      "epoch": 0.4060122240239491,
      "grad_norm": 0.0,
      "learning_rate": 1.3458138070075942e-05,
      "loss": 1.0015,
      "step": 3255
    },
    {
      "epoch": 0.4061369589622053,
      "grad_norm": 0.0,
      "learning_rate": 1.3454346936229547e-05,
      "loss": 1.0006,
      "step": 3256
    },
    {
      "epoch": 0.4062616939004615,
      "grad_norm": 0.0,
      "learning_rate": 1.3450555238546648e-05,
      "loss": 0.9848,
      "step": 3257
    },
    {
      "epoch": 0.40638642883871773,
      "grad_norm": 0.0,
      "learning_rate": 1.3446762977646147e-05,
      "loss": 0.9208,
      "step": 3258
    },
    {
      "epoch": 0.40651116377697394,
      "grad_norm": 0.0,
      "learning_rate": 1.3442970154147035e-05,
      "loss": 0.9922,
      "step": 3259
    },
    {
      "epoch": 0.40663589871523015,
      "grad_norm": 0.0,
      "learning_rate": 1.3439176768668396e-05,
      "loss": 0.9871,
      "step": 3260
    },
    {
      "epoch": 0.40676063365348636,
      "grad_norm": 0.0,
      "learning_rate": 1.3435382821829408e-05,
      "loss": 0.9676,
      "step": 3261
    },
    {
      "epoch": 0.40688536859174257,
      "grad_norm": 0.0,
      "learning_rate": 1.3431588314249341e-05,
      "loss": 0.991,
      "step": 3262
    },
    {
      "epoch": 0.4070101035299988,
      "grad_norm": 0.0,
      "learning_rate": 1.3427793246547547e-05,
      "loss": 1.0465,
      "step": 3263
    },
    {
      "epoch": 0.40713483846825493,
      "grad_norm": 0.0,
      "learning_rate": 1.3423997619343483e-05,
      "loss": 0.9862,
      "step": 3264
    },
    {
      "epoch": 0.40725957340651114,
      "grad_norm": 0.0,
      "learning_rate": 1.342020143325669e-05,
      "loss": 0.9564,
      "step": 3265
    },
    {
      "epoch": 0.40738430834476735,
      "grad_norm": 0.0,
      "learning_rate": 1.3416404688906797e-05,
      "loss": 0.9957,
      "step": 3266
    },
    {
      "epoch": 0.40750904328302356,
      "grad_norm": 0.0,
      "learning_rate": 1.3412607386913532e-05,
      "loss": 0.9201,
      "step": 3267
    },
    {
      "epoch": 0.4076337782212798,
      "grad_norm": 0.0,
      "learning_rate": 1.3408809527896714e-05,
      "loss": 1.0086,
      "step": 3268
    },
    {
      "epoch": 0.407758513159536,
      "grad_norm": 0.0,
      "learning_rate": 1.3405011112476244e-05,
      "loss": 0.9862,
      "step": 3269
    },
    {
      "epoch": 0.4078832480977922,
      "grad_norm": 0.0,
      "learning_rate": 1.340121214127212e-05,
      "loss": 0.9405,
      "step": 3270
    },
    {
      "epoch": 0.4080079830360484,
      "grad_norm": 0.0,
      "learning_rate": 1.339741261490443e-05,
      "loss": 0.9376,
      "step": 3271
    },
    {
      "epoch": 0.4081327179743046,
      "grad_norm": 0.0,
      "learning_rate": 1.3393612533993354e-05,
      "loss": 0.9668,
      "step": 3272
    },
    {
      "epoch": 0.4082574529125608,
      "grad_norm": 0.0,
      "learning_rate": 1.338981189915916e-05,
      "loss": 0.9637,
      "step": 3273
    },
    {
      "epoch": 0.40838218785081704,
      "grad_norm": 0.0,
      "learning_rate": 1.3386010711022206e-05,
      "loss": 0.9999,
      "step": 3274
    },
    {
      "epoch": 0.40850692278907325,
      "grad_norm": 0.0,
      "learning_rate": 1.3382208970202943e-05,
      "loss": 0.9754,
      "step": 3275
    },
    {
      "epoch": 0.4086316577273294,
      "grad_norm": 0.0,
      "learning_rate": 1.3378406677321914e-05,
      "loss": 0.9569,
      "step": 3276
    },
    {
      "epoch": 0.4087563926655856,
      "grad_norm": 0.0,
      "learning_rate": 1.3374603832999741e-05,
      "loss": 0.9692,
      "step": 3277
    },
    {
      "epoch": 0.4088811276038418,
      "grad_norm": 0.0,
      "learning_rate": 1.3370800437857154e-05,
      "loss": 0.9739,
      "step": 3278
    },
    {
      "epoch": 0.40900586254209803,
      "grad_norm": 0.0,
      "learning_rate": 1.3366996492514956e-05,
      "loss": 0.9653,
      "step": 3279
    },
    {
      "epoch": 0.40913059748035424,
      "grad_norm": 0.0,
      "learning_rate": 1.336319199759405e-05,
      "loss": 1.0045,
      "step": 3280
    },
    {
      "epoch": 0.40925533241861045,
      "grad_norm": 0.0,
      "learning_rate": 1.3359386953715423e-05,
      "loss": 0.9631,
      "step": 3281
    },
    {
      "epoch": 0.40938006735686666,
      "grad_norm": 0.0,
      "learning_rate": 1.3355581361500156e-05,
      "loss": 0.9982,
      "step": 3282
    },
    {
      "epoch": 0.40950480229512287,
      "grad_norm": 0.0,
      "learning_rate": 1.3351775221569416e-05,
      "loss": 0.9522,
      "step": 3283
    },
    {
      "epoch": 0.4096295372333791,
      "grad_norm": 0.0,
      "learning_rate": 1.3347968534544467e-05,
      "loss": 1.0326,
      "step": 3284
    },
    {
      "epoch": 0.4097542721716353,
      "grad_norm": 0.0,
      "learning_rate": 1.3344161301046648e-05,
      "loss": 1.0264,
      "step": 3285
    },
    {
      "epoch": 0.4098790071098915,
      "grad_norm": 0.0,
      "learning_rate": 1.3340353521697396e-05,
      "loss": 0.945,
      "step": 3286
    },
    {
      "epoch": 0.4100037420481477,
      "grad_norm": 0.0,
      "learning_rate": 1.333654519711824e-05,
      "loss": 0.9292,
      "step": 3287
    },
    {
      "epoch": 0.41012847698640387,
      "grad_norm": 0.0,
      "learning_rate": 1.3332736327930798e-05,
      "loss": 0.9846,
      "step": 3288
    },
    {
      "epoch": 0.4102532119246601,
      "grad_norm": 0.0,
      "learning_rate": 1.3328926914756765e-05,
      "loss": 0.9719,
      "step": 3289
    },
    {
      "epoch": 0.4103779468629163,
      "grad_norm": 0.0,
      "learning_rate": 1.3325116958217944e-05,
      "loss": 1.0035,
      "step": 3290
    },
    {
      "epoch": 0.4105026818011725,
      "grad_norm": 0.0,
      "learning_rate": 1.3321306458936204e-05,
      "loss": 0.9325,
      "step": 3291
    },
    {
      "epoch": 0.4106274167394287,
      "grad_norm": 0.0,
      "learning_rate": 1.3317495417533523e-05,
      "loss": 0.966,
      "step": 3292
    },
    {
      "epoch": 0.4107521516776849,
      "grad_norm": 0.0,
      "learning_rate": 1.3313683834631954e-05,
      "loss": 0.9942,
      "step": 3293
    },
    {
      "epoch": 0.41087688661594113,
      "grad_norm": 0.0,
      "learning_rate": 1.3309871710853646e-05,
      "loss": 0.9606,
      "step": 3294
    },
    {
      "epoch": 0.41100162155419734,
      "grad_norm": 0.0,
      "learning_rate": 1.3306059046820836e-05,
      "loss": 1.0202,
      "step": 3295
    },
    {
      "epoch": 0.41112635649245355,
      "grad_norm": 0.0,
      "learning_rate": 1.3302245843155843e-05,
      "loss": 0.9717,
      "step": 3296
    },
    {
      "epoch": 0.41125109143070976,
      "grad_norm": 0.0,
      "learning_rate": 1.3298432100481078e-05,
      "loss": 0.9596,
      "step": 3297
    },
    {
      "epoch": 0.41137582636896597,
      "grad_norm": 0.0,
      "learning_rate": 1.3294617819419044e-05,
      "loss": 0.9696,
      "step": 3298
    },
    {
      "epoch": 0.4115005613072222,
      "grad_norm": 0.0,
      "learning_rate": 1.3290803000592323e-05,
      "loss": 1.0161,
      "step": 3299
    },
    {
      "epoch": 0.41162529624547833,
      "grad_norm": 0.0,
      "learning_rate": 1.328698764462359e-05,
      "loss": 0.9859,
      "step": 3300
    },
    {
      "epoch": 0.41175003118373454,
      "grad_norm": 0.0,
      "learning_rate": 1.3283171752135614e-05,
      "loss": 1.0292,
      "step": 3301
    },
    {
      "epoch": 0.41187476612199075,
      "grad_norm": 0.0,
      "learning_rate": 1.3279355323751234e-05,
      "loss": 0.9557,
      "step": 3302
    },
    {
      "epoch": 0.41199950106024696,
      "grad_norm": 0.0,
      "learning_rate": 1.3275538360093397e-05,
      "loss": 1.0402,
      "step": 3303
    },
    {
      "epoch": 0.4121242359985032,
      "grad_norm": 0.0,
      "learning_rate": 1.327172086178512e-05,
      "loss": 1.013,
      "step": 3304
    },
    {
      "epoch": 0.4122489709367594,
      "grad_norm": 0.0,
      "learning_rate": 1.3267902829449523e-05,
      "loss": 1.0211,
      "step": 3305
    },
    {
      "epoch": 0.4123737058750156,
      "grad_norm": 0.0,
      "learning_rate": 1.3264084263709798e-05,
      "loss": 0.9713,
      "step": 3306
    },
    {
      "epoch": 0.4124984408132718,
      "grad_norm": 0.0,
      "learning_rate": 1.3260265165189234e-05,
      "loss": 0.9679,
      "step": 3307
    },
    {
      "epoch": 0.412623175751528,
      "grad_norm": 0.0,
      "learning_rate": 1.3256445534511209e-05,
      "loss": 0.958,
      "step": 3308
    },
    {
      "epoch": 0.4127479106897842,
      "grad_norm": 0.0,
      "learning_rate": 1.3252625372299172e-05,
      "loss": 0.9831,
      "step": 3309
    },
    {
      "epoch": 0.41287264562804044,
      "grad_norm": 0.0,
      "learning_rate": 1.3248804679176679e-05,
      "loss": 0.9826,
      "step": 3310
    },
    {
      "epoch": 0.41299738056629665,
      "grad_norm": 0.0,
      "learning_rate": 1.324498345576736e-05,
      "loss": 1.0188,
      "step": 3311
    },
    {
      "epoch": 0.4131221155045528,
      "grad_norm": 0.0,
      "learning_rate": 1.3241161702694933e-05,
      "loss": 0.9678,
      "step": 3312
    },
    {
      "epoch": 0.413246850442809,
      "grad_norm": 0.0,
      "learning_rate": 1.3237339420583213e-05,
      "loss": 0.918,
      "step": 3313
    },
    {
      "epoch": 0.4133715853810652,
      "grad_norm": 0.0,
      "learning_rate": 1.3233516610056082e-05,
      "loss": 0.9525,
      "step": 3314
    },
    {
      "epoch": 0.41349632031932143,
      "grad_norm": 0.0,
      "learning_rate": 1.3229693271737525e-05,
      "loss": 0.9662,
      "step": 3315
    },
    {
      "epoch": 0.41362105525757764,
      "grad_norm": 0.0,
      "learning_rate": 1.3225869406251603e-05,
      "loss": 0.9819,
      "step": 3316
    },
    {
      "epoch": 0.41374579019583385,
      "grad_norm": 0.0,
      "learning_rate": 1.3222045014222474e-05,
      "loss": 1.0782,
      "step": 3317
    },
    {
      "epoch": 0.41387052513409006,
      "grad_norm": 0.0,
      "learning_rate": 1.3218220096274373e-05,
      "loss": 0.9712,
      "step": 3318
    },
    {
      "epoch": 0.41399526007234627,
      "grad_norm": 0.0,
      "learning_rate": 1.3214394653031616e-05,
      "loss": 0.9891,
      "step": 3319
    },
    {
      "epoch": 0.4141199950106025,
      "grad_norm": 0.0,
      "learning_rate": 1.3210568685118621e-05,
      "loss": 0.9657,
      "step": 3320
    },
    {
      "epoch": 0.4142447299488587,
      "grad_norm": 0.0,
      "learning_rate": 1.3206742193159878e-05,
      "loss": 1.0027,
      "step": 3321
    },
    {
      "epoch": 0.4143694648871149,
      "grad_norm": 0.0,
      "learning_rate": 1.3202915177779967e-05,
      "loss": 0.9132,
      "step": 3322
    },
    {
      "epoch": 0.4144941998253711,
      "grad_norm": 0.0,
      "learning_rate": 1.3199087639603553e-05,
      "loss": 0.9376,
      "step": 3323
    },
    {
      "epoch": 0.41461893476362727,
      "grad_norm": 0.0,
      "learning_rate": 1.319525957925539e-05,
      "loss": 1.0008,
      "step": 3324
    },
    {
      "epoch": 0.4147436697018835,
      "grad_norm": 0.0,
      "learning_rate": 1.319143099736031e-05,
      "loss": 0.949,
      "step": 3325
    },
    {
      "epoch": 0.4148684046401397,
      "grad_norm": 0.0,
      "learning_rate": 1.3187601894543235e-05,
      "loss": 0.935,
      "step": 3326
    },
    {
      "epoch": 0.4149931395783959,
      "grad_norm": 0.0,
      "learning_rate": 1.3183772271429174e-05,
      "loss": 0.9845,
      "step": 3327
    },
    {
      "epoch": 0.4151178745166521,
      "grad_norm": 0.0,
      "learning_rate": 1.3179942128643216e-05,
      "loss": 0.9711,
      "step": 3328
    },
    {
      "epoch": 0.4152426094549083,
      "grad_norm": 0.0,
      "learning_rate": 1.3176111466810532e-05,
      "loss": 1.0098,
      "step": 3329
    },
    {
      "epoch": 0.41536734439316453,
      "grad_norm": 0.0,
      "learning_rate": 1.3172280286556394e-05,
      "loss": 0.9763,
      "step": 3330
    },
    {
      "epoch": 0.41549207933142074,
      "grad_norm": 0.0,
      "learning_rate": 1.3168448588506137e-05,
      "loss": 0.9655,
      "step": 3331
    },
    {
      "epoch": 0.41561681426967695,
      "grad_norm": 0.0,
      "learning_rate": 1.3164616373285194e-05,
      "loss": 0.9915,
      "step": 3332
    },
    {
      "epoch": 0.41574154920793316,
      "grad_norm": 0.0,
      "learning_rate": 1.316078364151908e-05,
      "loss": 0.9332,
      "step": 3333
    },
    {
      "epoch": 0.41586628414618937,
      "grad_norm": 0.0,
      "learning_rate": 1.3156950393833393e-05,
      "loss": 1.0026,
      "step": 3334
    },
    {
      "epoch": 0.4159910190844456,
      "grad_norm": 0.0,
      "learning_rate": 1.3153116630853813e-05,
      "loss": 0.9388,
      "step": 3335
    },
    {
      "epoch": 0.41611575402270173,
      "grad_norm": 0.0,
      "learning_rate": 1.3149282353206112e-05,
      "loss": 0.9679,
      "step": 3336
    },
    {
      "epoch": 0.41624048896095794,
      "grad_norm": 0.0,
      "learning_rate": 1.3145447561516138e-05,
      "loss": 0.9391,
      "step": 3337
    },
    {
      "epoch": 0.41636522389921415,
      "grad_norm": 0.0,
      "learning_rate": 1.3141612256409825e-05,
      "loss": 0.9821,
      "step": 3338
    },
    {
      "epoch": 0.41648995883747036,
      "grad_norm": 0.0,
      "learning_rate": 1.3137776438513193e-05,
      "loss": 0.9738,
      "step": 3339
    },
    {
      "epoch": 0.4166146937757266,
      "grad_norm": 0.0,
      "learning_rate": 1.3133940108452341e-05,
      "loss": 0.9998,
      "step": 3340
    },
    {
      "epoch": 0.4167394287139828,
      "grad_norm": 0.0,
      "learning_rate": 1.3130103266853462e-05,
      "loss": 1.0078,
      "step": 3341
    },
    {
      "epoch": 0.416864163652239,
      "grad_norm": 0.0,
      "learning_rate": 1.3126265914342814e-05,
      "loss": 0.9664,
      "step": 3342
    },
    {
      "epoch": 0.4169888985904952,
      "grad_norm": 0.0,
      "learning_rate": 1.3122428051546759e-05,
      "loss": 0.9335,
      "step": 3343
    },
    {
      "epoch": 0.4171136335287514,
      "grad_norm": 0.0,
      "learning_rate": 1.3118589679091732e-05,
      "loss": 0.9833,
      "step": 3344
    },
    {
      "epoch": 0.4172383684670076,
      "grad_norm": 0.0,
      "learning_rate": 1.3114750797604248e-05,
      "loss": 0.9464,
      "step": 3345
    },
    {
      "epoch": 0.41736310340526384,
      "grad_norm": 0.0,
      "learning_rate": 1.3110911407710909e-05,
      "loss": 0.9253,
      "step": 3346
    },
    {
      "epoch": 0.41748783834352005,
      "grad_norm": 0.0,
      "learning_rate": 1.3107071510038407e-05,
      "loss": 0.9595,
      "step": 3347
    },
    {
      "epoch": 0.4176125732817762,
      "grad_norm": 0.0,
      "learning_rate": 1.3103231105213505e-05,
      "loss": 1.0093,
      "step": 3348
    },
    {
      "epoch": 0.4177373082200324,
      "grad_norm": 0.0,
      "learning_rate": 1.3099390193863053e-05,
      "loss": 1.0055,
      "step": 3349
    },
    {
      "epoch": 0.4178620431582886,
      "grad_norm": 0.0,
      "learning_rate": 1.3095548776613987e-05,
      "loss": 1.002,
      "step": 3350
    },
    {
      "epoch": 0.41798677809654483,
      "grad_norm": 0.0,
      "learning_rate": 1.3091706854093323e-05,
      "loss": 0.972,
      "step": 3351
    },
    {
      "epoch": 0.41811151303480104,
      "grad_norm": 0.0,
      "learning_rate": 1.3087864426928158e-05,
      "loss": 1.0143,
      "step": 3352
    },
    {
      "epoch": 0.41823624797305725,
      "grad_norm": 0.0,
      "learning_rate": 1.3084021495745673e-05,
      "loss": 0.981,
      "step": 3353
    },
    {
      "epoch": 0.41836098291131346,
      "grad_norm": 0.0,
      "learning_rate": 1.3080178061173135e-05,
      "loss": 0.9726,
      "step": 3354
    },
    {
      "epoch": 0.41848571784956967,
      "grad_norm": 0.0,
      "learning_rate": 1.3076334123837884e-05,
      "loss": 1.0101,
      "step": 3355
    },
    {
      "epoch": 0.4186104527878259,
      "grad_norm": 0.0,
      "learning_rate": 1.3072489684367348e-05,
      "loss": 0.9661,
      "step": 3356
    },
    {
      "epoch": 0.4187351877260821,
      "grad_norm": 0.0,
      "learning_rate": 1.3068644743389043e-05,
      "loss": 0.9868,
      "step": 3357
    },
    {
      "epoch": 0.4188599226643383,
      "grad_norm": 0.0,
      "learning_rate": 1.3064799301530556e-05,
      "loss": 0.9527,
      "step": 3358
    },
    {
      "epoch": 0.4189846576025945,
      "grad_norm": 0.0,
      "learning_rate": 1.3060953359419556e-05,
      "loss": 0.9969,
      "step": 3359
    },
    {
      "epoch": 0.41910939254085067,
      "grad_norm": 0.0,
      "learning_rate": 1.3057106917683803e-05,
      "loss": 0.9082,
      "step": 3360
    },
    {
      "epoch": 0.4192341274791069,
      "grad_norm": 0.0,
      "learning_rate": 1.3053259976951134e-05,
      "loss": 0.9957,
      "step": 3361
    },
    {
      "epoch": 0.4193588624173631,
      "grad_norm": 0.0,
      "learning_rate": 1.3049412537849459e-05,
      "loss": 0.9363,
      "step": 3362
    },
    {
      "epoch": 0.4194835973556193,
      "grad_norm": 0.0,
      "learning_rate": 1.3045564601006789e-05,
      "loss": 0.9809,
      "step": 3363
    },
    {
      "epoch": 0.4196083322938755,
      "grad_norm": 0.0,
      "learning_rate": 1.3041716167051197e-05,
      "loss": 0.957,
      "step": 3364
    },
    {
      "epoch": 0.4197330672321317,
      "grad_norm": 0.0,
      "learning_rate": 1.3037867236610841e-05,
      "loss": 0.9818,
      "step": 3365
    },
    {
      "epoch": 0.41985780217038793,
      "grad_norm": 0.0,
      "learning_rate": 1.303401781031397e-05,
      "loss": 0.9462,
      "step": 3366
    },
    {
      "epoch": 0.41998253710864414,
      "grad_norm": 0.0,
      "learning_rate": 1.3030167888788908e-05,
      "loss": 1.0061,
      "step": 3367
    },
    {
      "epoch": 0.42010727204690035,
      "grad_norm": 0.0,
      "learning_rate": 1.3026317472664052e-05,
      "loss": 0.9403,
      "step": 3368
    },
    {
      "epoch": 0.42023200698515656,
      "grad_norm": 0.0,
      "learning_rate": 1.3022466562567894e-05,
      "loss": 0.9711,
      "step": 3369
    },
    {
      "epoch": 0.42035674192341277,
      "grad_norm": 0.0,
      "learning_rate": 1.3018615159128995e-05,
      "loss": 1.0123,
      "step": 3370
    },
    {
      "epoch": 0.420481476861669,
      "grad_norm": 0.0,
      "learning_rate": 1.3014763262976007e-05,
      "loss": 0.9474,
      "step": 3371
    },
    {
      "epoch": 0.42060621179992513,
      "grad_norm": 0.0,
      "learning_rate": 1.3010910874737648e-05,
      "loss": 0.9982,
      "step": 3372
    },
    {
      "epoch": 0.42073094673818134,
      "grad_norm": 0.0,
      "learning_rate": 1.300705799504273e-05,
      "loss": 0.9958,
      "step": 3373
    },
    {
      "epoch": 0.42085568167643755,
      "grad_norm": 0.0,
      "learning_rate": 1.3003204624520145e-05,
      "loss": 0.9617,
      "step": 3374
    },
    {
      "epoch": 0.42098041661469376,
      "grad_norm": 0.0,
      "learning_rate": 1.2999350763798853e-05,
      "loss": 1.0738,
      "step": 3375
    },
    {
      "epoch": 0.42110515155295,
      "grad_norm": 0.0,
      "learning_rate": 1.29954964135079e-05,
      "loss": 0.9649,
      "step": 3376
    },
    {
      "epoch": 0.4212298864912062,
      "grad_norm": 0.0,
      "learning_rate": 1.2991641574276419e-05,
      "loss": 0.96,
      "step": 3377
    },
    {
      "epoch": 0.4213546214294624,
      "grad_norm": 0.0,
      "learning_rate": 1.2987786246733611e-05,
      "loss": 0.9957,
      "step": 3378
    },
    {
      "epoch": 0.4214793563677186,
      "grad_norm": 0.0,
      "learning_rate": 1.2983930431508766e-05,
      "loss": 0.9929,
      "step": 3379
    },
    {
      "epoch": 0.4216040913059748,
      "grad_norm": 0.0,
      "learning_rate": 1.2980074129231254e-05,
      "loss": 0.948,
      "step": 3380
    },
    {
      "epoch": 0.421728826244231,
      "grad_norm": 0.0,
      "learning_rate": 1.2976217340530517e-05,
      "loss": 0.9329,
      "step": 3381
    },
    {
      "epoch": 0.42185356118248724,
      "grad_norm": 0.0,
      "learning_rate": 1.2972360066036078e-05,
      "loss": 0.9542,
      "step": 3382
    },
    {
      "epoch": 0.42197829612074345,
      "grad_norm": 0.0,
      "learning_rate": 1.2968502306377541e-05,
      "loss": 1.0123,
      "step": 3383
    },
    {
      "epoch": 0.4221030310589996,
      "grad_norm": 0.0,
      "learning_rate": 1.2964644062184597e-05,
      "loss": 0.9891,
      "step": 3384
    },
    {
      "epoch": 0.4222277659972558,
      "grad_norm": 0.0,
      "learning_rate": 1.2960785334087e-05,
      "loss": 0.936,
      "step": 3385
    },
    {
      "epoch": 0.422352500935512,
      "grad_norm": 0.0,
      "learning_rate": 1.2956926122714597e-05,
      "loss": 0.9238,
      "step": 3386
    },
    {
      "epoch": 0.42247723587376823,
      "grad_norm": 0.0,
      "learning_rate": 1.295306642869731e-05,
      "loss": 0.9796,
      "step": 3387
    },
    {
      "epoch": 0.42260197081202444,
      "grad_norm": 0.0,
      "learning_rate": 1.2949206252665132e-05,
      "loss": 0.9864,
      "step": 3388
    },
    {
      "epoch": 0.42272670575028065,
      "grad_norm": 0.0,
      "learning_rate": 1.2945345595248142e-05,
      "loss": 0.9495,
      "step": 3389
    },
    {
      "epoch": 0.42285144068853686,
      "grad_norm": 0.0,
      "learning_rate": 1.2941484457076503e-05,
      "loss": 0.9546,
      "step": 3390
    },
    {
      "epoch": 0.42297617562679307,
      "grad_norm": 0.0,
      "learning_rate": 1.2937622838780444e-05,
      "loss": 0.9985,
      "step": 3391
    },
    {
      "epoch": 0.4231009105650493,
      "grad_norm": 0.0,
      "learning_rate": 1.2933760740990284e-05,
      "loss": 0.9158,
      "step": 3392
    },
    {
      "epoch": 0.4232256455033055,
      "grad_norm": 0.0,
      "learning_rate": 1.2929898164336407e-05,
      "loss": 0.9402,
      "step": 3393
    },
    {
      "epoch": 0.4233503804415617,
      "grad_norm": 0.0,
      "learning_rate": 1.2926035109449293e-05,
      "loss": 0.9277,
      "step": 3394
    },
    {
      "epoch": 0.4234751153798179,
      "grad_norm": 0.0,
      "learning_rate": 1.2922171576959479e-05,
      "loss": 0.9485,
      "step": 3395
    },
    {
      "epoch": 0.42359985031807407,
      "grad_norm": 0.0,
      "learning_rate": 1.2918307567497596e-05,
      "loss": 0.9937,
      "step": 3396
    },
    {
      "epoch": 0.4237245852563303,
      "grad_norm": 0.0,
      "learning_rate": 1.2914443081694353e-05,
      "loss": 0.9304,
      "step": 3397
    },
    {
      "epoch": 0.4238493201945865,
      "grad_norm": 0.0,
      "learning_rate": 1.2910578120180524e-05,
      "loss": 0.9483,
      "step": 3398
    },
    {
      "epoch": 0.4239740551328427,
      "grad_norm": 0.0,
      "learning_rate": 1.2906712683586968e-05,
      "loss": 0.9738,
      "step": 3399
    },
    {
      "epoch": 0.4240987900710989,
      "grad_norm": 0.0,
      "learning_rate": 1.2902846772544625e-05,
      "loss": 0.9746,
      "step": 3400
    },
    {
      "epoch": 0.4242235250093551,
      "grad_norm": 0.0,
      "learning_rate": 1.289898038768451e-05,
      "loss": 0.9915,
      "step": 3401
    },
    {
      "epoch": 0.42434825994761133,
      "grad_norm": 0.0,
      "learning_rate": 1.289511352963771e-05,
      "loss": 0.9905,
      "step": 3402
    },
    {
      "epoch": 0.42447299488586754,
      "grad_norm": 0.0,
      "learning_rate": 1.28912461990354e-05,
      "loss": 0.9626,
      "step": 3403
    },
    {
      "epoch": 0.42459772982412375,
      "grad_norm": 0.0,
      "learning_rate": 1.2887378396508822e-05,
      "loss": 0.964,
      "step": 3404
    },
    {
      "epoch": 0.42472246476237996,
      "grad_norm": 0.0,
      "learning_rate": 1.2883510122689297e-05,
      "loss": 0.9625,
      "step": 3405
    },
    {
      "epoch": 0.42484719970063617,
      "grad_norm": 0.0,
      "learning_rate": 1.2879641378208229e-05,
      "loss": 0.9928,
      "step": 3406
    },
    {
      "epoch": 0.4249719346388924,
      "grad_norm": 0.0,
      "learning_rate": 1.2875772163697094e-05,
      "loss": 0.9273,
      "step": 3407
    },
    {
      "epoch": 0.42509666957714853,
      "grad_norm": 0.0,
      "learning_rate": 1.2871902479787439e-05,
      "loss": 0.9696,
      "step": 3408
    },
    {
      "epoch": 0.42522140451540474,
      "grad_norm": 0.0,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 0.9538,
      "step": 3409
    },
    {
      "epoch": 0.42534613945366095,
      "grad_norm": 0.0,
      "learning_rate": 1.2864161706299187e-05,
      "loss": 0.9322,
      "step": 3410
    },
    {
      "epoch": 0.42547087439191716,
      "grad_norm": 0.0,
      "learning_rate": 1.286029061798408e-05,
      "loss": 0.9607,
      "step": 3411
    },
    {
      "epoch": 0.4255956093301734,
      "grad_norm": 0.0,
      "learning_rate": 1.285641906279743e-05,
      "loss": 1.0039,
      "step": 3412
    },
    {
      "epoch": 0.4257203442684296,
      "grad_norm": 0.0,
      "learning_rate": 1.2852547041371181e-05,
      "loss": 1.01,
      "step": 3413
    },
    {
      "epoch": 0.4258450792066858,
      "grad_norm": 0.0,
      "learning_rate": 1.2848674554337346e-05,
      "loss": 0.9915,
      "step": 3414
    },
    {
      "epoch": 0.425969814144942,
      "grad_norm": 0.0,
      "learning_rate": 1.2844801602328007e-05,
      "loss": 0.977,
      "step": 3415
    },
    {
      "epoch": 0.4260945490831982,
      "grad_norm": 0.0,
      "learning_rate": 1.2840928185975327e-05,
      "loss": 1.0159,
      "step": 3416
    },
    {
      "epoch": 0.4262192840214544,
      "grad_norm": 0.0,
      "learning_rate": 1.2837054305911551e-05,
      "loss": 0.9784,
      "step": 3417
    },
    {
      "epoch": 0.42634401895971064,
      "grad_norm": 0.0,
      "learning_rate": 1.2833179962768988e-05,
      "loss": 0.9816,
      "step": 3418
    },
    {
      "epoch": 0.42646875389796685,
      "grad_norm": 0.0,
      "learning_rate": 1.2829305157180033e-05,
      "loss": 0.9545,
      "step": 3419
    },
    {
      "epoch": 0.426593488836223,
      "grad_norm": 0.0,
      "learning_rate": 1.282542988977715e-05,
      "loss": 0.9429,
      "step": 3420
    },
    {
      "epoch": 0.4267182237744792,
      "grad_norm": 0.0,
      "learning_rate": 1.2821554161192879e-05,
      "loss": 0.9452,
      "step": 3421
    },
    {
      "epoch": 0.4268429587127354,
      "grad_norm": 0.0,
      "learning_rate": 1.2817677972059838e-05,
      "loss": 0.9926,
      "step": 3422
    },
    {
      "epoch": 0.42696769365099163,
      "grad_norm": 0.0,
      "learning_rate": 1.2813801323010719e-05,
      "loss": 0.9275,
      "step": 3423
    },
    {
      "epoch": 0.42709242858924784,
      "grad_norm": 0.0,
      "learning_rate": 1.280992421467829e-05,
      "loss": 1.0094,
      "step": 3424
    },
    {
      "epoch": 0.42721716352750405,
      "grad_norm": 0.0,
      "learning_rate": 1.280604664769539e-05,
      "loss": 0.9568,
      "step": 3425
    },
    {
      "epoch": 0.42734189846576026,
      "grad_norm": 0.0,
      "learning_rate": 1.2802168622694938e-05,
      "loss": 0.9901,
      "step": 3426
    },
    {
      "epoch": 0.42746663340401647,
      "grad_norm": 0.0,
      "learning_rate": 1.2798290140309924e-05,
      "loss": 0.9292,
      "step": 3427
    },
    {
      "epoch": 0.4275913683422727,
      "grad_norm": 0.0,
      "learning_rate": 1.2794411201173414e-05,
      "loss": 0.999,
      "step": 3428
    },
    {
      "epoch": 0.4277161032805289,
      "grad_norm": 0.0,
      "learning_rate": 1.2790531805918546e-05,
      "loss": 0.9397,
      "step": 3429
    },
    {
      "epoch": 0.4278408382187851,
      "grad_norm": 0.0,
      "learning_rate": 1.2786651955178545e-05,
      "loss": 1.0181,
      "step": 3430
    },
    {
      "epoch": 0.4279655731570413,
      "grad_norm": 0.0,
      "learning_rate": 1.2782771649586689e-05,
      "loss": 0.9423,
      "step": 3431
    },
    {
      "epoch": 0.42809030809529747,
      "grad_norm": 0.0,
      "learning_rate": 1.2778890889776349e-05,
      "loss": 1.0065,
      "step": 3432
    },
    {
      "epoch": 0.4282150430335537,
      "grad_norm": 0.0,
      "learning_rate": 1.2775009676380959e-05,
      "loss": 1.0111,
      "step": 3433
    },
    {
      "epoch": 0.4283397779718099,
      "grad_norm": 0.0,
      "learning_rate": 1.2771128010034032e-05,
      "loss": 0.9946,
      "step": 3434
    },
    {
      "epoch": 0.4284645129100661,
      "grad_norm": 0.0,
      "learning_rate": 1.2767245891369156e-05,
      "loss": 0.9979,
      "step": 3435
    },
    {
      "epoch": 0.4285892478483223,
      "grad_norm": 0.0,
      "learning_rate": 1.2763363321019986e-05,
      "loss": 0.9975,
      "step": 3436
    },
    {
      "epoch": 0.4287139827865785,
      "grad_norm": 0.0,
      "learning_rate": 1.2759480299620262e-05,
      "loss": 0.9689,
      "step": 3437
    },
    {
      "epoch": 0.42883871772483473,
      "grad_norm": 0.0,
      "learning_rate": 1.2755596827803784e-05,
      "loss": 0.93,
      "step": 3438
    },
    {
      "epoch": 0.42896345266309094,
      "grad_norm": 0.0,
      "learning_rate": 1.2751712906204434e-05,
      "loss": 0.9797,
      "step": 3439
    },
    {
      "epoch": 0.42908818760134715,
      "grad_norm": 0.0,
      "learning_rate": 1.274782853545617e-05,
      "loss": 0.9024,
      "step": 3440
    },
    {
      "epoch": 0.42921292253960336,
      "grad_norm": 0.0,
      "learning_rate": 1.2743943716193017e-05,
      "loss": 0.9292,
      "step": 3441
    },
    {
      "epoch": 0.42933765747785957,
      "grad_norm": 0.0,
      "learning_rate": 1.2740058449049073e-05,
      "loss": 0.9948,
      "step": 3442
    },
    {
      "epoch": 0.4294623924161158,
      "grad_norm": 0.0,
      "learning_rate": 1.2736172734658516e-05,
      "loss": 0.9626,
      "step": 3443
    },
    {
      "epoch": 0.42958712735437193,
      "grad_norm": 0.0,
      "learning_rate": 1.273228657365559e-05,
      "loss": 0.9573,
      "step": 3444
    },
    {
      "epoch": 0.42971186229262814,
      "grad_norm": 0.0,
      "learning_rate": 1.2728399966674612e-05,
      "loss": 0.9927,
      "step": 3445
    },
    {
      "epoch": 0.42983659723088435,
      "grad_norm": 0.0,
      "learning_rate": 1.2724512914349979e-05,
      "loss": 0.9902,
      "step": 3446
    },
    {
      "epoch": 0.42996133216914056,
      "grad_norm": 0.0,
      "learning_rate": 1.2720625417316153e-05,
      "loss": 0.95,
      "step": 3447
    },
    {
      "epoch": 0.4300860671073968,
      "grad_norm": 0.0,
      "learning_rate": 1.271673747620767e-05,
      "loss": 0.9866,
      "step": 3448
    },
    {
      "epoch": 0.430210802045653,
      "grad_norm": 0.0,
      "learning_rate": 1.2712849091659147e-05,
      "loss": 0.9574,
      "step": 3449
    },
    {
      "epoch": 0.4303355369839092,
      "grad_norm": 0.0,
      "learning_rate": 1.2708960264305261e-05,
      "loss": 0.9589,
      "step": 3450
    },
    {
      "epoch": 0.4304602719221654,
      "grad_norm": 0.0,
      "learning_rate": 1.2705070994780763e-05,
      "loss": 0.9908,
      "step": 3451
    },
    {
      "epoch": 0.4305850068604216,
      "grad_norm": 0.0,
      "learning_rate": 1.2701181283720487e-05,
      "loss": 0.9819,
      "step": 3452
    },
    {
      "epoch": 0.4307097417986778,
      "grad_norm": 0.0,
      "learning_rate": 1.269729113175933e-05,
      "loss": 0.9634,
      "step": 3453
    },
    {
      "epoch": 0.43083447673693404,
      "grad_norm": 0.0,
      "learning_rate": 1.2693400539532263e-05,
      "loss": 0.969,
      "step": 3454
    },
    {
      "epoch": 0.43095921167519025,
      "grad_norm": 0.0,
      "learning_rate": 1.2689509507674324e-05,
      "loss": 0.9598,
      "step": 3455
    },
    {
      "epoch": 0.4310839466134464,
      "grad_norm": 0.0,
      "learning_rate": 1.2685618036820633e-05,
      "loss": 0.9557,
      "step": 3456
    },
    {
      "epoch": 0.4312086815517026,
      "grad_norm": 0.0,
      "learning_rate": 1.2681726127606374e-05,
      "loss": 0.969,
      "step": 3457
    },
    {
      "epoch": 0.4313334164899588,
      "grad_norm": 0.0,
      "learning_rate": 1.2677833780666806e-05,
      "loss": 0.9605,
      "step": 3458
    },
    {
      "epoch": 0.43145815142821503,
      "grad_norm": 0.0,
      "learning_rate": 1.2673940996637255e-05,
      "loss": 1.0218,
      "step": 3459
    },
    {
      "epoch": 0.43158288636647124,
      "grad_norm": 0.0,
      "learning_rate": 1.267004777615313e-05,
      "loss": 0.9712,
      "step": 3460
    },
    {
      "epoch": 0.43170762130472745,
      "grad_norm": 0.0,
      "learning_rate": 1.266615411984989e-05,
      "loss": 0.9409,
      "step": 3461
    },
    {
      "epoch": 0.43183235624298366,
      "grad_norm": 0.0,
      "learning_rate": 1.2662260028363084e-05,
      "loss": 0.9351,
      "step": 3462
    },
    {
      "epoch": 0.43195709118123987,
      "grad_norm": 0.0,
      "learning_rate": 1.2658365502328329e-05,
      "loss": 0.9786,
      "step": 3463
    },
    {
      "epoch": 0.4320818261194961,
      "grad_norm": 0.0,
      "learning_rate": 1.2654470542381305e-05,
      "loss": 0.9518,
      "step": 3464
    },
    {
      "epoch": 0.4322065610577523,
      "grad_norm": 0.0,
      "learning_rate": 1.2650575149157773e-05,
      "loss": 0.9123,
      "step": 3465
    },
    {
      "epoch": 0.4323312959960085,
      "grad_norm": 0.0,
      "learning_rate": 1.2646679323293553e-05,
      "loss": 1.0002,
      "step": 3466
    },
    {
      "epoch": 0.4324560309342647,
      "grad_norm": 0.0,
      "learning_rate": 1.2642783065424546e-05,
      "loss": 1.0283,
      "step": 3467
    },
    {
      "epoch": 0.43258076587252087,
      "grad_norm": 0.0,
      "learning_rate": 1.2638886376186719e-05,
      "loss": 0.9589,
      "step": 3468
    },
    {
      "epoch": 0.4327055008107771,
      "grad_norm": 0.0,
      "learning_rate": 1.2634989256216108e-05,
      "loss": 0.967,
      "step": 3469
    },
    {
      "epoch": 0.4328302357490333,
      "grad_norm": 0.0,
      "learning_rate": 1.2631091706148824e-05,
      "loss": 0.9934,
      "step": 3470
    },
    {
      "epoch": 0.4329549706872895,
      "grad_norm": 0.0,
      "learning_rate": 1.2627193726621045e-05,
      "loss": 1.0199,
      "step": 3471
    },
    {
      "epoch": 0.4330797056255457,
      "grad_norm": 0.0,
      "learning_rate": 1.2623295318269018e-05,
      "loss": 0.9486,
      "step": 3472
    },
    {
      "epoch": 0.4332044405638019,
      "grad_norm": 0.0,
      "learning_rate": 1.261939648172906e-05,
      "loss": 0.9873,
      "step": 3473
    },
    {
      "epoch": 0.43332917550205813,
      "grad_norm": 0.0,
      "learning_rate": 1.2615497217637565e-05,
      "loss": 1.0065,
      "step": 3474
    },
    {
      "epoch": 0.43345391044031434,
      "grad_norm": 0.0,
      "learning_rate": 1.2611597526630983e-05,
      "loss": 0.9482,
      "step": 3475
    },
    {
      "epoch": 0.43357864537857055,
      "grad_norm": 0.0,
      "learning_rate": 1.2607697409345851e-05,
      "loss": 0.9828,
      "step": 3476
    },
    {
      "epoch": 0.43370338031682676,
      "grad_norm": 0.0,
      "learning_rate": 1.2603796866418762e-05,
      "loss": 0.9745,
      "step": 3477
    },
    {
      "epoch": 0.43382811525508297,
      "grad_norm": 0.0,
      "learning_rate": 1.2599895898486381e-05,
      "loss": 0.9499,
      "step": 3478
    },
    {
      "epoch": 0.4339528501933392,
      "grad_norm": 0.0,
      "learning_rate": 1.2595994506185446e-05,
      "loss": 0.966,
      "step": 3479
    },
    {
      "epoch": 0.43407758513159533,
      "grad_norm": 0.0,
      "learning_rate": 1.2592092690152765e-05,
      "loss": 0.928,
      "step": 3480
    },
    {
      "epoch": 0.43420232006985154,
      "grad_norm": 0.0,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 0.9769,
      "step": 3481
    },
    {
      "epoch": 0.43432705500810775,
      "grad_norm": 0.0,
      "learning_rate": 1.2584287789439725e-05,
      "loss": 0.9664,
      "step": 3482
    },
    {
      "epoch": 0.43445178994636396,
      "grad_norm": 0.0,
      "learning_rate": 1.2580384706033324e-05,
      "loss": 0.9561,
      "step": 3483
    },
    {
      "epoch": 0.4345765248846202,
      "grad_norm": 0.0,
      "learning_rate": 1.2576481201443087e-05,
      "loss": 0.9593,
      "step": 3484
    },
    {
      "epoch": 0.4347012598228764,
      "grad_norm": 0.0,
      "learning_rate": 1.2572577276306163e-05,
      "loss": 0.9454,
      "step": 3485
    },
    {
      "epoch": 0.4348259947611326,
      "grad_norm": 0.0,
      "learning_rate": 1.2568672931259778e-05,
      "loss": 0.9535,
      "step": 3486
    },
    {
      "epoch": 0.4349507296993888,
      "grad_norm": 0.0,
      "learning_rate": 1.256476816694121e-05,
      "loss": 0.9495,
      "step": 3487
    },
    {
      "epoch": 0.435075464637645,
      "grad_norm": 0.0,
      "learning_rate": 1.2560862983987825e-05,
      "loss": 1.038,
      "step": 3488
    },
    {
      "epoch": 0.4352001995759012,
      "grad_norm": 0.0,
      "learning_rate": 1.2556957383037038e-05,
      "loss": 0.9563,
      "step": 3489
    },
    {
      "epoch": 0.43532493451415744,
      "grad_norm": 0.0,
      "learning_rate": 1.2553051364726347e-05,
      "loss": 0.9618,
      "step": 3490
    },
    {
      "epoch": 0.43544966945241365,
      "grad_norm": 0.0,
      "learning_rate": 1.254914492969331e-05,
      "loss": 0.9761,
      "step": 3491
    },
    {
      "epoch": 0.4355744043906698,
      "grad_norm": 0.0,
      "learning_rate": 1.2545238078575557e-05,
      "loss": 1.0031,
      "step": 3492
    },
    {
      "epoch": 0.435699139328926,
      "grad_norm": 0.0,
      "learning_rate": 1.2541330812010786e-05,
      "loss": 0.9482,
      "step": 3493
    },
    {
      "epoch": 0.4358238742671822,
      "grad_norm": 0.0,
      "learning_rate": 1.253742313063676e-05,
      "loss": 0.9401,
      "step": 3494
    },
    {
      "epoch": 0.43594860920543843,
      "grad_norm": 0.0,
      "learning_rate": 1.253351503509131e-05,
      "loss": 0.9929,
      "step": 3495
    },
    {
      "epoch": 0.43607334414369464,
      "grad_norm": 0.0,
      "learning_rate": 1.2529606526012338e-05,
      "loss": 1.0057,
      "step": 3496
    },
    {
      "epoch": 0.43619807908195085,
      "grad_norm": 0.0,
      "learning_rate": 1.2525697604037807e-05,
      "loss": 1.006,
      "step": 3497
    },
    {
      "epoch": 0.43632281402020706,
      "grad_norm": 0.0,
      "learning_rate": 1.2521788269805757e-05,
      "loss": 0.9039,
      "step": 3498
    },
    {
      "epoch": 0.43644754895846327,
      "grad_norm": 0.0,
      "learning_rate": 1.2517878523954287e-05,
      "loss": 0.9701,
      "step": 3499
    },
    {
      "epoch": 0.4365722838967195,
      "grad_norm": 0.0,
      "learning_rate": 1.2513968367121567e-05,
      "loss": 1.0131,
      "step": 3500
    },
    {
      "epoch": 0.4366970188349757,
      "grad_norm": 0.0,
      "learning_rate": 1.2510057799945832e-05,
      "loss": 0.9729,
      "step": 3501
    },
    {
      "epoch": 0.4368217537732319,
      "grad_norm": 0.0,
      "learning_rate": 1.2506146823065386e-05,
      "loss": 1.0178,
      "step": 3502
    },
    {
      "epoch": 0.4369464887114881,
      "grad_norm": 0.0,
      "learning_rate": 1.2502235437118597e-05,
      "loss": 0.9274,
      "step": 3503
    },
    {
      "epoch": 0.43707122364974427,
      "grad_norm": 0.0,
      "learning_rate": 1.2498323642743906e-05,
      "loss": 0.9752,
      "step": 3504
    },
    {
      "epoch": 0.4371959585880005,
      "grad_norm": 0.0,
      "learning_rate": 1.2494411440579814e-05,
      "loss": 0.9826,
      "step": 3505
    },
    {
      "epoch": 0.4373206935262567,
      "grad_norm": 0.0,
      "learning_rate": 1.2490498831264892e-05,
      "loss": 0.9658,
      "step": 3506
    },
    {
      "epoch": 0.4374454284645129,
      "grad_norm": 0.0,
      "learning_rate": 1.2486585815437775e-05,
      "loss": 0.9483,
      "step": 3507
    },
    {
      "epoch": 0.4375701634027691,
      "grad_norm": 0.0,
      "learning_rate": 1.2482672393737164e-05,
      "loss": 0.9848,
      "step": 3508
    },
    {
      "epoch": 0.4376948983410253,
      "grad_norm": 0.0,
      "learning_rate": 1.2478758566801834e-05,
      "loss": 0.947,
      "step": 3509
    },
    {
      "epoch": 0.43781963327928153,
      "grad_norm": 0.0,
      "learning_rate": 1.2474844335270613e-05,
      "loss": 0.9839,
      "step": 3510
    },
    {
      "epoch": 0.43794436821753774,
      "grad_norm": 0.0,
      "learning_rate": 1.247092969978241e-05,
      "loss": 0.9765,
      "step": 3511
    },
    {
      "epoch": 0.43806910315579395,
      "grad_norm": 0.0,
      "learning_rate": 1.2467014660976185e-05,
      "loss": 1.0252,
      "step": 3512
    },
    {
      "epoch": 0.43819383809405016,
      "grad_norm": 0.0,
      "learning_rate": 1.2463099219490977e-05,
      "loss": 1.0336,
      "step": 3513
    },
    {
      "epoch": 0.43831857303230637,
      "grad_norm": 0.0,
      "learning_rate": 1.245918337596588e-05,
      "loss": 1.0399,
      "step": 3514
    },
    {
      "epoch": 0.4384433079705626,
      "grad_norm": 0.0,
      "learning_rate": 1.2455267131040058e-05,
      "loss": 0.998,
      "step": 3515
    },
    {
      "epoch": 0.43856804290881873,
      "grad_norm": 0.0,
      "learning_rate": 1.245135048535275e-05,
      "loss": 1.0037,
      "step": 3516
    },
    {
      "epoch": 0.43869277784707494,
      "grad_norm": 0.0,
      "learning_rate": 1.2447433439543239e-05,
      "loss": 0.9442,
      "step": 3517
    },
    {
      "epoch": 0.43881751278533115,
      "grad_norm": 0.0,
      "learning_rate": 1.2443515994250892e-05,
      "loss": 0.9792,
      "step": 3518
    },
    {
      "epoch": 0.43894224772358736,
      "grad_norm": 0.0,
      "learning_rate": 1.2439598150115138e-05,
      "loss": 0.9353,
      "step": 3519
    },
    {
      "epoch": 0.4390669826618436,
      "grad_norm": 0.0,
      "learning_rate": 1.243567990777546e-05,
      "loss": 0.9689,
      "step": 3520
    },
    {
      "epoch": 0.4391917176000998,
      "grad_norm": 0.0,
      "learning_rate": 1.2431761267871419e-05,
      "loss": 0.9947,
      "step": 3521
    },
    {
      "epoch": 0.439316452538356,
      "grad_norm": 0.0,
      "learning_rate": 1.2427842231042636e-05,
      "loss": 0.9636,
      "step": 3522
    },
    {
      "epoch": 0.4394411874766122,
      "grad_norm": 0.0,
      "learning_rate": 1.2423922797928794e-05,
      "loss": 0.9197,
      "step": 3523
    },
    {
      "epoch": 0.4395659224148684,
      "grad_norm": 0.0,
      "learning_rate": 1.2420002969169642e-05,
      "loss": 1.0126,
      "step": 3524
    },
    {
      "epoch": 0.4396906573531246,
      "grad_norm": 0.0,
      "learning_rate": 1.2416082745405001e-05,
      "loss": 0.9496,
      "step": 3525
    },
    {
      "epoch": 0.43981539229138084,
      "grad_norm": 0.0,
      "learning_rate": 1.2412162127274748e-05,
      "loss": 0.9894,
      "step": 3526
    },
    {
      "epoch": 0.43994012722963705,
      "grad_norm": 0.0,
      "learning_rate": 1.240824111541882e-05,
      "loss": 0.9462,
      "step": 3527
    },
    {
      "epoch": 0.4400648621678932,
      "grad_norm": 0.0,
      "learning_rate": 1.2404319710477238e-05,
      "loss": 0.9094,
      "step": 3528
    },
    {
      "epoch": 0.4401895971061494,
      "grad_norm": 0.0,
      "learning_rate": 1.2400397913090061e-05,
      "loss": 0.9447,
      "step": 3529
    },
    {
      "epoch": 0.4403143320444056,
      "grad_norm": 0.0,
      "learning_rate": 1.2396475723897432e-05,
      "loss": 0.9984,
      "step": 3530
    },
    {
      "epoch": 0.44043906698266183,
      "grad_norm": 0.0,
      "learning_rate": 1.239255314353955e-05,
      "loss": 0.9543,
      "step": 3531
    },
    {
      "epoch": 0.44056380192091804,
      "grad_norm": 0.0,
      "learning_rate": 1.2388630172656681e-05,
      "loss": 0.9943,
      "step": 3532
    },
    {
      "epoch": 0.44068853685917425,
      "grad_norm": 0.0,
      "learning_rate": 1.238470681188915e-05,
      "loss": 0.9707,
      "step": 3533
    },
    {
      "epoch": 0.44081327179743046,
      "grad_norm": 0.0,
      "learning_rate": 1.2380783061877348e-05,
      "loss": 0.9753,
      "step": 3534
    },
    {
      "epoch": 0.44093800673568667,
      "grad_norm": 0.0,
      "learning_rate": 1.2376858923261732e-05,
      "loss": 0.9293,
      "step": 3535
    },
    {
      "epoch": 0.4410627416739429,
      "grad_norm": 0.0,
      "learning_rate": 1.2372934396682822e-05,
      "loss": 1.022,
      "step": 3536
    },
    {
      "epoch": 0.4411874766121991,
      "grad_norm": 0.0,
      "learning_rate": 1.2369009482781191e-05,
      "loss": 0.9523,
      "step": 3537
    },
    {
      "epoch": 0.4413122115504553,
      "grad_norm": 0.0,
      "learning_rate": 1.2365084182197493e-05,
      "loss": 0.9512,
      "step": 3538
    },
    {
      "epoch": 0.4414369464887115,
      "grad_norm": 0.0,
      "learning_rate": 1.2361158495572436e-05,
      "loss": 0.9812,
      "step": 3539
    },
    {
      "epoch": 0.44156168142696767,
      "grad_norm": 0.0,
      "learning_rate": 1.2357232423546786e-05,
      "loss": 0.9633,
      "step": 3540
    },
    {
      "epoch": 0.4416864163652239,
      "grad_norm": 0.0,
      "learning_rate": 1.2353305966761377e-05,
      "loss": 1.0156,
      "step": 3541
    },
    {
      "epoch": 0.4418111513034801,
      "grad_norm": 0.0,
      "learning_rate": 1.2349379125857109e-05,
      "loss": 0.993,
      "step": 3542
    },
    {
      "epoch": 0.4419358862417363,
      "grad_norm": 0.0,
      "learning_rate": 1.2345451901474941e-05,
      "loss": 1.0018,
      "step": 3543
    },
    {
      "epoch": 0.4420606211799925,
      "grad_norm": 0.0,
      "learning_rate": 1.2341524294255893e-05,
      "loss": 0.9537,
      "step": 3544
    },
    {
      "epoch": 0.4421853561182487,
      "grad_norm": 0.0,
      "learning_rate": 1.2337596304841053e-05,
      "loss": 0.9651,
      "step": 3545
    },
    {
      "epoch": 0.44231009105650493,
      "grad_norm": 0.0,
      "learning_rate": 1.2333667933871565e-05,
      "loss": 0.9624,
      "step": 3546
    },
    {
      "epoch": 0.44243482599476114,
      "grad_norm": 0.0,
      "learning_rate": 1.2329739181988635e-05,
      "loss": 0.9792,
      "step": 3547
    },
    {
      "epoch": 0.44255956093301735,
      "grad_norm": 0.0,
      "learning_rate": 1.2325810049833539e-05,
      "loss": 0.9534,
      "step": 3548
    },
    {
      "epoch": 0.44268429587127356,
      "grad_norm": 0.0,
      "learning_rate": 1.2321880538047615e-05,
      "loss": 1.0021,
      "step": 3549
    },
    {
      "epoch": 0.44280903080952977,
      "grad_norm": 0.0,
      "learning_rate": 1.231795064727225e-05,
      "loss": 0.9226,
      "step": 3550
    },
    {
      "epoch": 0.442933765747786,
      "grad_norm": 0.0,
      "learning_rate": 1.23140203781489e-05,
      "loss": 1.0251,
      "step": 3551
    },
    {
      "epoch": 0.44305850068604213,
      "grad_norm": 0.0,
      "learning_rate": 1.2310089731319093e-05,
      "loss": 0.9552,
      "step": 3552
    },
    {
      "epoch": 0.44318323562429834,
      "grad_norm": 0.0,
      "learning_rate": 1.2306158707424402e-05,
      "loss": 0.9304,
      "step": 3553
    },
    {
      "epoch": 0.44330797056255455,
      "grad_norm": 0.0,
      "learning_rate": 1.2302227307106474e-05,
      "loss": 0.9634,
      "step": 3554
    },
    {
      "epoch": 0.44343270550081076,
      "grad_norm": 0.0,
      "learning_rate": 1.229829553100701e-05,
      "loss": 0.9898,
      "step": 3555
    },
    {
      "epoch": 0.443557440439067,
      "grad_norm": 0.0,
      "learning_rate": 1.2294363379767777e-05,
      "loss": 0.9973,
      "step": 3556
    },
    {
      "epoch": 0.4436821753773232,
      "grad_norm": 0.0,
      "learning_rate": 1.2290430854030597e-05,
      "loss": 0.9487,
      "step": 3557
    },
    {
      "epoch": 0.4438069103155794,
      "grad_norm": 0.0,
      "learning_rate": 1.2286497954437362e-05,
      "loss": 1.0067,
      "step": 3558
    },
    {
      "epoch": 0.4439316452538356,
      "grad_norm": 0.0,
      "learning_rate": 1.2282564681630018e-05,
      "loss": 0.9397,
      "step": 3559
    },
    {
      "epoch": 0.4440563801920918,
      "grad_norm": 0.0,
      "learning_rate": 1.2278631036250575e-05,
      "loss": 1.028,
      "step": 3560
    },
    {
      "epoch": 0.444181115130348,
      "grad_norm": 0.0,
      "learning_rate": 1.2274697018941101e-05,
      "loss": 0.97,
      "step": 3561
    },
    {
      "epoch": 0.44430585006860424,
      "grad_norm": 0.0,
      "learning_rate": 1.2270762630343734e-05,
      "loss": 0.9506,
      "step": 3562
    },
    {
      "epoch": 0.44443058500686045,
      "grad_norm": 0.0,
      "learning_rate": 1.2266827871100658e-05,
      "loss": 1.0113,
      "step": 3563
    },
    {
      "epoch": 0.4445553199451166,
      "grad_norm": 0.0,
      "learning_rate": 1.2262892741854124e-05,
      "loss": 0.9949,
      "step": 3564
    },
    {
      "epoch": 0.4446800548833728,
      "grad_norm": 0.0,
      "learning_rate": 1.2258957243246447e-05,
      "loss": 0.9504,
      "step": 3565
    },
    {
      "epoch": 0.444804789821629,
      "grad_norm": 0.0,
      "learning_rate": 1.2255021375920005e-05,
      "loss": 0.9439,
      "step": 3566
    },
    {
      "epoch": 0.44492952475988523,
      "grad_norm": 0.0,
      "learning_rate": 1.2251085140517222e-05,
      "loss": 0.9601,
      "step": 3567
    },
    {
      "epoch": 0.44505425969814144,
      "grad_norm": 0.0,
      "learning_rate": 1.2247148537680595e-05,
      "loss": 0.9292,
      "step": 3568
    },
    {
      "epoch": 0.44517899463639765,
      "grad_norm": 0.0,
      "learning_rate": 1.2243211568052678e-05,
      "loss": 0.9659,
      "step": 3569
    },
    {
      "epoch": 0.44530372957465386,
      "grad_norm": 0.0,
      "learning_rate": 1.2239274232276079e-05,
      "loss": 0.9722,
      "step": 3570
    },
    {
      "epoch": 0.44542846451291007,
      "grad_norm": 0.0,
      "learning_rate": 1.2235336530993475e-05,
      "loss": 1.0019,
      "step": 3571
    },
    {
      "epoch": 0.4455531994511663,
      "grad_norm": 0.0,
      "learning_rate": 1.22313984648476e-05,
      "loss": 0.9643,
      "step": 3572
    },
    {
      "epoch": 0.4456779343894225,
      "grad_norm": 0.0,
      "learning_rate": 1.2227460034481239e-05,
      "loss": 0.9801,
      "step": 3573
    },
    {
      "epoch": 0.4458026693276787,
      "grad_norm": 0.0,
      "learning_rate": 1.2223521240537247e-05,
      "loss": 0.9663,
      "step": 3574
    },
    {
      "epoch": 0.4459274042659349,
      "grad_norm": 0.0,
      "learning_rate": 1.221958208365853e-05,
      "loss": 0.9519,
      "step": 3575
    },
    {
      "epoch": 0.44605213920419107,
      "grad_norm": 0.0,
      "learning_rate": 1.2215642564488066e-05,
      "loss": 0.9248,
      "step": 3576
    },
    {
      "epoch": 0.4461768741424473,
      "grad_norm": 0.0,
      "learning_rate": 1.2211702683668878e-05,
      "loss": 0.9829,
      "step": 3577
    },
    {
      "epoch": 0.4463016090807035,
      "grad_norm": 0.0,
      "learning_rate": 1.2207762441844055e-05,
      "loss": 0.9714,
      "step": 3578
    },
    {
      "epoch": 0.4464263440189597,
      "grad_norm": 0.0,
      "learning_rate": 1.2203821839656748e-05,
      "loss": 0.9718,
      "step": 3579
    },
    {
      "epoch": 0.4465510789572159,
      "grad_norm": 0.0,
      "learning_rate": 1.2199880877750157e-05,
      "loss": 0.9699,
      "step": 3580
    },
    {
      "epoch": 0.4466758138954721,
      "grad_norm": 0.0,
      "learning_rate": 1.2195939556767544e-05,
      "loss": 0.964,
      "step": 3581
    },
    {
      "epoch": 0.44680054883372833,
      "grad_norm": 0.0,
      "learning_rate": 1.2191997877352239e-05,
      "loss": 0.9915,
      "step": 3582
    },
    {
      "epoch": 0.44692528377198454,
      "grad_norm": 0.0,
      "learning_rate": 1.2188055840147621e-05,
      "loss": 0.9708,
      "step": 3583
    },
    {
      "epoch": 0.44705001871024075,
      "grad_norm": 0.0,
      "learning_rate": 1.218411344579713e-05,
      "loss": 0.9064,
      "step": 3584
    },
    {
      "epoch": 0.44717475364849696,
      "grad_norm": 0.0,
      "learning_rate": 1.218017069494426e-05,
      "loss": 0.9405,
      "step": 3585
    },
    {
      "epoch": 0.44729948858675317,
      "grad_norm": 0.0,
      "learning_rate": 1.2176227588232575e-05,
      "loss": 0.9254,
      "step": 3586
    },
    {
      "epoch": 0.4474242235250094,
      "grad_norm": 0.0,
      "learning_rate": 1.217228412630568e-05,
      "loss": 0.9418,
      "step": 3587
    },
    {
      "epoch": 0.44754895846326553,
      "grad_norm": 0.0,
      "learning_rate": 1.2168340309807257e-05,
      "loss": 1.0066,
      "step": 3588
    },
    {
      "epoch": 0.44767369340152174,
      "grad_norm": 0.0,
      "learning_rate": 1.2164396139381029e-05,
      "loss": 0.991,
      "step": 3589
    },
    {
      "epoch": 0.44779842833977795,
      "grad_norm": 0.0,
      "learning_rate": 1.2160451615670793e-05,
      "loss": 0.9328,
      "step": 3590
    },
    {
      "epoch": 0.44792316327803416,
      "grad_norm": 0.0,
      "learning_rate": 1.2156506739320381e-05,
      "loss": 0.9746,
      "step": 3591
    },
    {
      "epoch": 0.4480478982162904,
      "grad_norm": 0.0,
      "learning_rate": 1.2152561510973709e-05,
      "loss": 0.9838,
      "step": 3592
    },
    {
      "epoch": 0.4481726331545466,
      "grad_norm": 0.0,
      "learning_rate": 1.214861593127473e-05,
      "loss": 0.936,
      "step": 3593
    },
    {
      "epoch": 0.4482973680928028,
      "grad_norm": 0.0,
      "learning_rate": 1.2144670000867462e-05,
      "loss": 1.0212,
      "step": 3594
    },
    {
      "epoch": 0.448422103031059,
      "grad_norm": 0.0,
      "learning_rate": 1.2140723720395989e-05,
      "loss": 0.977,
      "step": 3595
    },
    {
      "epoch": 0.4485468379693152,
      "grad_norm": 0.0,
      "learning_rate": 1.2136777090504434e-05,
      "loss": 0.9165,
      "step": 3596
    },
    {
      "epoch": 0.4486715729075714,
      "grad_norm": 0.0,
      "learning_rate": 1.2132830111836991e-05,
      "loss": 0.9736,
      "step": 3597
    },
    {
      "epoch": 0.44879630784582764,
      "grad_norm": 0.0,
      "learning_rate": 1.2128882785037905e-05,
      "loss": 0.935,
      "step": 3598
    },
    {
      "epoch": 0.44892104278408385,
      "grad_norm": 0.0,
      "learning_rate": 1.2124935110751481e-05,
      "loss": 0.9958,
      "step": 3599
    },
    {
      "epoch": 0.44904577772234,
      "grad_norm": 0.0,
      "learning_rate": 1.2120987089622077e-05,
      "loss": 0.9337,
      "step": 3600
    },
    {
      "epoch": 0.4491705126605962,
      "grad_norm": 0.0,
      "learning_rate": 1.211703872229411e-05,
      "loss": 0.9409,
      "step": 3601
    },
    {
      "epoch": 0.4492952475988524,
      "grad_norm": 0.0,
      "learning_rate": 1.2113090009412054e-05,
      "loss": 0.9656,
      "step": 3602
    },
    {
      "epoch": 0.44941998253710863,
      "grad_norm": 0.0,
      "learning_rate": 1.2109140951620435e-05,
      "loss": 0.9994,
      "step": 3603
    },
    {
      "epoch": 0.44954471747536484,
      "grad_norm": 0.0,
      "learning_rate": 1.2105191549563842e-05,
      "loss": 0.9808,
      "step": 3604
    },
    {
      "epoch": 0.44966945241362105,
      "grad_norm": 0.0,
      "learning_rate": 1.2101241803886921e-05,
      "loss": 0.9782,
      "step": 3605
    },
    {
      "epoch": 0.44979418735187726,
      "grad_norm": 0.0,
      "learning_rate": 1.209729171523436e-05,
      "loss": 1.0256,
      "step": 3606
    },
    {
      "epoch": 0.44991892229013347,
      "grad_norm": 0.0,
      "learning_rate": 1.2093341284250922e-05,
      "loss": 0.9873,
      "step": 3607
    },
    {
      "epoch": 0.4500436572283897,
      "grad_norm": 0.0,
      "learning_rate": 1.2089390511581411e-05,
      "loss": 0.934,
      "step": 3608
    },
    {
      "epoch": 0.4501683921666459,
      "grad_norm": 0.0,
      "learning_rate": 1.2085439397870699e-05,
      "loss": 0.976,
      "step": 3609
    },
    {
      "epoch": 0.4502931271049021,
      "grad_norm": 0.0,
      "learning_rate": 1.20814879437637e-05,
      "loss": 0.9288,
      "step": 3610
    },
    {
      "epoch": 0.4504178620431583,
      "grad_norm": 0.0,
      "learning_rate": 1.2077536149905395e-05,
      "loss": 0.9451,
      "step": 3611
    },
    {
      "epoch": 0.45054259698141447,
      "grad_norm": 0.0,
      "learning_rate": 1.2073584016940822e-05,
      "loss": 0.9981,
      "step": 3612
    },
    {
      "epoch": 0.4506673319196707,
      "grad_norm": 0.0,
      "learning_rate": 1.2069631545515056e-05,
      "loss": 0.9219,
      "step": 3613
    },
    {
      "epoch": 0.4507920668579269,
      "grad_norm": 0.0,
      "learning_rate": 1.2065678736273247e-05,
      "loss": 0.9685,
      "step": 3614
    },
    {
      "epoch": 0.4509168017961831,
      "grad_norm": 0.0,
      "learning_rate": 1.2061725589860595e-05,
      "loss": 0.9645,
      "step": 3615
    },
    {
      "epoch": 0.4510415367344393,
      "grad_norm": 0.0,
      "learning_rate": 1.205777210692235e-05,
      "loss": 1.002,
      "step": 3616
    },
    {
      "epoch": 0.4511662716726955,
      "grad_norm": 0.0,
      "learning_rate": 1.205381828810382e-05,
      "loss": 0.999,
      "step": 3617
    },
    {
      "epoch": 0.45129100661095173,
      "grad_norm": 0.0,
      "learning_rate": 1.2049864134050372e-05,
      "loss": 1.0243,
      "step": 3618
    },
    {
      "epoch": 0.45141574154920794,
      "grad_norm": 0.0,
      "learning_rate": 1.2045909645407419e-05,
      "loss": 1.0007,
      "step": 3619
    },
    {
      "epoch": 0.45154047648746415,
      "grad_norm": 0.0,
      "learning_rate": 1.2041954822820434e-05,
      "loss": 1.0168,
      "step": 3620
    },
    {
      "epoch": 0.45166521142572036,
      "grad_norm": 0.0,
      "learning_rate": 1.2037999666934943e-05,
      "loss": 0.9692,
      "step": 3621
    },
    {
      "epoch": 0.45178994636397657,
      "grad_norm": 0.0,
      "learning_rate": 1.2034044178396534e-05,
      "loss": 0.9963,
      "step": 3622
    },
    {
      "epoch": 0.4519146813022328,
      "grad_norm": 0.0,
      "learning_rate": 1.2030088357850831e-05,
      "loss": 0.9316,
      "step": 3623
    },
    {
      "epoch": 0.45203941624048893,
      "grad_norm": 0.0,
      "learning_rate": 1.2026132205943534e-05,
      "loss": 0.9618,
      "step": 3624
    },
    {
      "epoch": 0.45216415117874514,
      "grad_norm": 0.0,
      "learning_rate": 1.2022175723320382e-05,
      "loss": 0.9683,
      "step": 3625
    },
    {
      "epoch": 0.45228888611700135,
      "grad_norm": 0.0,
      "learning_rate": 1.201821891062717e-05,
      "loss": 0.9712,
      "step": 3626
    },
    {
      "epoch": 0.45241362105525756,
      "grad_norm": 0.0,
      "learning_rate": 1.2014261768509757e-05,
      "loss": 1.0183,
      "step": 3627
    },
    {
      "epoch": 0.4525383559935138,
      "grad_norm": 0.0,
      "learning_rate": 1.2010304297614042e-05,
      "loss": 0.9932,
      "step": 3628
    },
    {
      "epoch": 0.45266309093177,
      "grad_norm": 0.0,
      "learning_rate": 1.2006346498585988e-05,
      "loss": 1.0142,
      "step": 3629
    },
    {
      "epoch": 0.4527878258700262,
      "grad_norm": 0.0,
      "learning_rate": 1.2002388372071603e-05,
      "loss": 0.9171,
      "step": 3630
    },
    {
      "epoch": 0.4529125608082824,
      "grad_norm": 0.0,
      "learning_rate": 1.1998429918716954e-05,
      "loss": 0.9815,
      "step": 3631
    },
    {
      "epoch": 0.4530372957465386,
      "grad_norm": 0.0,
      "learning_rate": 1.1994471139168166e-05,
      "loss": 0.9453,
      "step": 3632
    },
    {
      "epoch": 0.4531620306847948,
      "grad_norm": 0.0,
      "learning_rate": 1.1990512034071407e-05,
      "loss": 0.9519,
      "step": 3633
    },
    {
      "epoch": 0.45328676562305104,
      "grad_norm": 0.0,
      "learning_rate": 1.19865526040729e-05,
      "loss": 0.9518,
      "step": 3634
    },
    {
      "epoch": 0.45341150056130725,
      "grad_norm": 0.0,
      "learning_rate": 1.1982592849818933e-05,
      "loss": 1.004,
      "step": 3635
    },
    {
      "epoch": 0.4535362354995634,
      "grad_norm": 0.0,
      "learning_rate": 1.197863277195583e-05,
      "loss": 0.9694,
      "step": 3636
    },
    {
      "epoch": 0.4536609704378196,
      "grad_norm": 0.0,
      "learning_rate": 1.1974672371129977e-05,
      "loss": 0.9898,
      "step": 3637
    },
    {
      "epoch": 0.4537857053760758,
      "grad_norm": 0.0,
      "learning_rate": 1.1970711647987812e-05,
      "loss": 0.9935,
      "step": 3638
    },
    {
      "epoch": 0.45391044031433203,
      "grad_norm": 0.0,
      "learning_rate": 1.1966750603175825e-05,
      "loss": 0.9232,
      "step": 3639
    },
    {
      "epoch": 0.45403517525258824,
      "grad_norm": 0.0,
      "learning_rate": 1.1962789237340555e-05,
      "loss": 0.9883,
      "step": 3640
    },
    {
      "epoch": 0.45415991019084445,
      "grad_norm": 0.0,
      "learning_rate": 1.1958827551128604e-05,
      "loss": 0.9457,
      "step": 3641
    },
    {
      "epoch": 0.45428464512910066,
      "grad_norm": 0.0,
      "learning_rate": 1.1954865545186616e-05,
      "loss": 0.9598,
      "step": 3642
    },
    {
      "epoch": 0.45440938006735687,
      "grad_norm": 0.0,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 1.0123,
      "step": 3643
    },
    {
      "epoch": 0.4545341150056131,
      "grad_norm": 0.0,
      "learning_rate": 1.1946940576699367e-05,
      "loss": 0.9619,
      "step": 3644
    },
    {
      "epoch": 0.4546588499438693,
      "grad_norm": 0.0,
      "learning_rate": 1.1942977615447666e-05,
      "loss": 0.9281,
      "step": 3645
    },
    {
      "epoch": 0.4547835848821255,
      "grad_norm": 0.0,
      "learning_rate": 1.1939014337053036e-05,
      "loss": 1.0266,
      "step": 3646
    },
    {
      "epoch": 0.4549083198203817,
      "grad_norm": 0.0,
      "learning_rate": 1.1935050742162384e-05,
      "loss": 0.9428,
      "step": 3647
    },
    {
      "epoch": 0.45503305475863787,
      "grad_norm": 0.0,
      "learning_rate": 1.1931086831422667e-05,
      "loss": 0.9763,
      "step": 3648
    },
    {
      "epoch": 0.4551577896968941,
      "grad_norm": 0.0,
      "learning_rate": 1.1927122605480899e-05,
      "loss": 0.9202,
      "step": 3649
    },
    {
      "epoch": 0.4552825246351503,
      "grad_norm": 0.0,
      "learning_rate": 1.1923158064984137e-05,
      "loss": 0.9502,
      "step": 3650
    },
    {
      "epoch": 0.4554072595734065,
      "grad_norm": 0.0,
      "learning_rate": 1.1919193210579497e-05,
      "loss": 0.9717,
      "step": 3651
    },
    {
      "epoch": 0.4555319945116627,
      "grad_norm": 0.0,
      "learning_rate": 1.1915228042914144e-05,
      "loss": 0.9259,
      "step": 3652
    },
    {
      "epoch": 0.4556567294499189,
      "grad_norm": 0.0,
      "learning_rate": 1.1911262562635291e-05,
      "loss": 0.9759,
      "step": 3653
    },
    {
      "epoch": 0.45578146438817513,
      "grad_norm": 0.0,
      "learning_rate": 1.1907296770390204e-05,
      "loss": 0.9554,
      "step": 3654
    },
    {
      "epoch": 0.45590619932643134,
      "grad_norm": 0.0,
      "learning_rate": 1.1903330666826202e-05,
      "loss": 0.987,
      "step": 3655
    },
    {
      "epoch": 0.45603093426468755,
      "grad_norm": 0.0,
      "learning_rate": 1.1899364252590652e-05,
      "loss": 0.9398,
      "step": 3656
    },
    {
      "epoch": 0.45615566920294376,
      "grad_norm": 0.0,
      "learning_rate": 1.1895397528330975e-05,
      "loss": 0.9202,
      "step": 3657
    },
    {
      "epoch": 0.45628040414119997,
      "grad_norm": 0.0,
      "learning_rate": 1.1891430494694637e-05,
      "loss": 0.9873,
      "step": 3658
    },
    {
      "epoch": 0.4564051390794562,
      "grad_norm": 0.0,
      "learning_rate": 1.1887463152329162e-05,
      "loss": 0.9719,
      "step": 3659
    },
    {
      "epoch": 0.45652987401771233,
      "grad_norm": 0.0,
      "learning_rate": 1.1883495501882114e-05,
      "loss": 0.9658,
      "step": 3660
    },
    {
      "epoch": 0.45665460895596854,
      "grad_norm": 0.0,
      "learning_rate": 1.187952754400112e-05,
      "loss": 0.9444,
      "step": 3661
    },
    {
      "epoch": 0.45677934389422475,
      "grad_norm": 0.0,
      "learning_rate": 1.1875559279333847e-05,
      "loss": 0.9569,
      "step": 3662
    },
    {
      "epoch": 0.45690407883248096,
      "grad_norm": 0.0,
      "learning_rate": 1.187159070852802e-05,
      "loss": 0.9787,
      "step": 3663
    },
    {
      "epoch": 0.4570288137707372,
      "grad_norm": 0.0,
      "learning_rate": 1.1867621832231401e-05,
      "loss": 0.939,
      "step": 3664
    },
    {
      "epoch": 0.4571535487089934,
      "grad_norm": 0.0,
      "learning_rate": 1.1863652651091824e-05,
      "loss": 0.9601,
      "step": 3665
    },
    {
      "epoch": 0.4572782836472496,
      "grad_norm": 0.0,
      "learning_rate": 1.1859683165757146e-05,
      "loss": 0.9617,
      "step": 3666
    },
    {
      "epoch": 0.4574030185855058,
      "grad_norm": 0.0,
      "learning_rate": 1.1855713376875295e-05,
      "loss": 1.0165,
      "step": 3667
    },
    {
      "epoch": 0.457527753523762,
      "grad_norm": 0.0,
      "learning_rate": 1.1851743285094239e-05,
      "loss": 0.9373,
      "step": 3668
    },
    {
      "epoch": 0.4576524884620182,
      "grad_norm": 0.0,
      "learning_rate": 1.1847772891061996e-05,
      "loss": 0.9735,
      "step": 3669
    },
    {
      "epoch": 0.45777722340027444,
      "grad_norm": 0.0,
      "learning_rate": 1.1843802195426634e-05,
      "loss": 0.9725,
      "step": 3670
    },
    {
      "epoch": 0.45790195833853065,
      "grad_norm": 0.0,
      "learning_rate": 1.1839831198836274e-05,
      "loss": 0.938,
      "step": 3671
    },
    {
      "epoch": 0.4580266932767868,
      "grad_norm": 0.0,
      "learning_rate": 1.1835859901939079e-05,
      "loss": 0.9826,
      "step": 3672
    },
    {
      "epoch": 0.458151428215043,
      "grad_norm": 0.0,
      "learning_rate": 1.1831888305383268e-05,
      "loss": 1.0022,
      "step": 3673
    },
    {
      "epoch": 0.4582761631532992,
      "grad_norm": 0.0,
      "learning_rate": 1.18279164098171e-05,
      "loss": 0.9562,
      "step": 3674
    },
    {
      "epoch": 0.45840089809155543,
      "grad_norm": 0.0,
      "learning_rate": 1.1823944215888899e-05,
      "loss": 0.985,
      "step": 3675
    },
    {
      "epoch": 0.45852563302981164,
      "grad_norm": 0.0,
      "learning_rate": 1.1819971724247017e-05,
      "loss": 0.9706,
      "step": 3676
    },
    {
      "epoch": 0.45865036796806785,
      "grad_norm": 0.0,
      "learning_rate": 1.1815998935539865e-05,
      "loss": 0.9604,
      "step": 3677
    },
    {
      "epoch": 0.45877510290632406,
      "grad_norm": 0.0,
      "learning_rate": 1.1812025850415912e-05,
      "loss": 0.8965,
      "step": 3678
    },
    {
      "epoch": 0.45889983784458027,
      "grad_norm": 0.0,
      "learning_rate": 1.1808052469523654e-05,
      "loss": 1.0053,
      "step": 3679
    },
    {
      "epoch": 0.4590245727828365,
      "grad_norm": 0.0,
      "learning_rate": 1.1804078793511655e-05,
      "loss": 0.9182,
      "step": 3680
    },
    {
      "epoch": 0.4591493077210927,
      "grad_norm": 0.0,
      "learning_rate": 1.1800104823028515e-05,
      "loss": 0.9636,
      "step": 3681
    },
    {
      "epoch": 0.4592740426593489,
      "grad_norm": 0.0,
      "learning_rate": 1.1796130558722889e-05,
      "loss": 0.9524,
      "step": 3682
    },
    {
      "epoch": 0.4593987775976051,
      "grad_norm": 0.0,
      "learning_rate": 1.1792156001243473e-05,
      "loss": 0.9885,
      "step": 3683
    },
    {
      "epoch": 0.45952351253586127,
      "grad_norm": 0.0,
      "learning_rate": 1.1788181151239018e-05,
      "loss": 0.992,
      "step": 3684
    },
    {
      "epoch": 0.4596482474741175,
      "grad_norm": 0.0,
      "learning_rate": 1.1784206009358323e-05,
      "loss": 0.999,
      "step": 3685
    },
    {
      "epoch": 0.4597729824123737,
      "grad_norm": 0.0,
      "learning_rate": 1.1780230576250227e-05,
      "loss": 0.9668,
      "step": 3686
    },
    {
      "epoch": 0.4598977173506299,
      "grad_norm": 0.0,
      "learning_rate": 1.1776254852563619e-05,
      "loss": 0.9723,
      "step": 3687
    },
    {
      "epoch": 0.4600224522888861,
      "grad_norm": 0.0,
      "learning_rate": 1.1772278838947442e-05,
      "loss": 0.9348,
      "step": 3688
    },
    {
      "epoch": 0.4601471872271423,
      "grad_norm": 0.0,
      "learning_rate": 1.1768302536050675e-05,
      "loss": 0.9465,
      "step": 3689
    },
    {
      "epoch": 0.46027192216539853,
      "grad_norm": 0.0,
      "learning_rate": 1.1764325944522359e-05,
      "loss": 1.0134,
      "step": 3690
    },
    {
      "epoch": 0.46039665710365474,
      "grad_norm": 0.0,
      "learning_rate": 1.1760349065011572e-05,
      "loss": 0.9797,
      "step": 3691
    },
    {
      "epoch": 0.46052139204191095,
      "grad_norm": 0.0,
      "learning_rate": 1.1756371898167437e-05,
      "loss": 0.9678,
      "step": 3692
    },
    {
      "epoch": 0.46064612698016716,
      "grad_norm": 0.0,
      "learning_rate": 1.1752394444639132e-05,
      "loss": 1.049,
      "step": 3693
    },
    {
      "epoch": 0.46077086191842337,
      "grad_norm": 0.0,
      "learning_rate": 1.1748416705075873e-05,
      "loss": 0.9399,
      "step": 3694
    },
    {
      "epoch": 0.4608955968566796,
      "grad_norm": 0.0,
      "learning_rate": 1.1744438680126936e-05,
      "loss": 0.9358,
      "step": 3695
    },
    {
      "epoch": 0.46102033179493573,
      "grad_norm": 0.0,
      "learning_rate": 1.1740460370441624e-05,
      "loss": 0.9816,
      "step": 3696
    },
    {
      "epoch": 0.46114506673319194,
      "grad_norm": 0.0,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.9783,
      "step": 3697
    },
    {
      "epoch": 0.46126980167144815,
      "grad_norm": 0.0,
      "learning_rate": 1.1732502899459384e-05,
      "loss": 0.9702,
      "step": 3698
    },
    {
      "epoch": 0.46139453660970436,
      "grad_norm": 0.0,
      "learning_rate": 1.1728523739461312e-05,
      "loss": 0.9767,
      "step": 3699
    },
    {
      "epoch": 0.4615192715479606,
      "grad_norm": 0.0,
      "learning_rate": 1.1724544297324592e-05,
      "loss": 0.9547,
      "step": 3700
    },
    {
      "epoch": 0.4616440064862168,
      "grad_norm": 0.0,
      "learning_rate": 1.1720564573698765e-05,
      "loss": 0.9849,
      "step": 3701
    },
    {
      "epoch": 0.461768741424473,
      "grad_norm": 0.0,
      "learning_rate": 1.1716584569233425e-05,
      "loss": 0.9256,
      "step": 3702
    },
    {
      "epoch": 0.4618934763627292,
      "grad_norm": 0.0,
      "learning_rate": 1.1712604284578211e-05,
      "loss": 0.9105,
      "step": 3703
    },
    {
      "epoch": 0.4620182113009854,
      "grad_norm": 0.0,
      "learning_rate": 1.1708623720382798e-05,
      "loss": 0.9767,
      "step": 3704
    },
    {
      "epoch": 0.4621429462392416,
      "grad_norm": 0.0,
      "learning_rate": 1.1704642877296926e-05,
      "loss": 0.992,
      "step": 3705
    },
    {
      "epoch": 0.46226768117749784,
      "grad_norm": 0.0,
      "learning_rate": 1.1700661755970357e-05,
      "loss": 0.9844,
      "step": 3706
    },
    {
      "epoch": 0.46239241611575405,
      "grad_norm": 0.0,
      "learning_rate": 1.1696680357052918e-05,
      "loss": 0.9491,
      "step": 3707
    },
    {
      "epoch": 0.46251715105401026,
      "grad_norm": 0.0,
      "learning_rate": 1.1692698681194474e-05,
      "loss": 0.985,
      "step": 3708
    },
    {
      "epoch": 0.4626418859922664,
      "grad_norm": 0.0,
      "learning_rate": 1.1688716729044931e-05,
      "loss": 0.9671,
      "step": 3709
    },
    {
      "epoch": 0.4627666209305226,
      "grad_norm": 0.0,
      "learning_rate": 1.1684734501254242e-05,
      "loss": 0.9645,
      "step": 3710
    },
    {
      "epoch": 0.46289135586877883,
      "grad_norm": 0.0,
      "learning_rate": 1.1680751998472415e-05,
      "loss": 0.9833,
      "step": 3711
    },
    {
      "epoch": 0.46301609080703504,
      "grad_norm": 0.0,
      "learning_rate": 1.1676769221349489e-05,
      "loss": 0.9284,
      "step": 3712
    },
    {
      "epoch": 0.46314082574529125,
      "grad_norm": 0.0,
      "learning_rate": 1.1672786170535553e-05,
      "loss": 0.9516,
      "step": 3713
    },
    {
      "epoch": 0.46326556068354746,
      "grad_norm": 0.0,
      "learning_rate": 1.1668802846680747e-05,
      "loss": 0.971,
      "step": 3714
    },
    {
      "epoch": 0.46339029562180367,
      "grad_norm": 0.0,
      "learning_rate": 1.1664819250435246e-05,
      "loss": 0.9167,
      "step": 3715
    },
    {
      "epoch": 0.4635150305600599,
      "grad_norm": 0.0,
      "learning_rate": 1.166083538244927e-05,
      "loss": 0.9503,
      "step": 3716
    },
    {
      "epoch": 0.4636397654983161,
      "grad_norm": 0.0,
      "learning_rate": 1.165685124337309e-05,
      "loss": 0.9572,
      "step": 3717
    },
    {
      "epoch": 0.4637645004365723,
      "grad_norm": 0.0,
      "learning_rate": 1.1652866833857021e-05,
      "loss": 0.9256,
      "step": 3718
    },
    {
      "epoch": 0.4638892353748285,
      "grad_norm": 0.0,
      "learning_rate": 1.1648882154551416e-05,
      "loss": 0.9467,
      "step": 3719
    },
    {
      "epoch": 0.4640139703130847,
      "grad_norm": 0.0,
      "learning_rate": 1.1644897206106673e-05,
      "loss": 0.94,
      "step": 3720
    },
    {
      "epoch": 0.4641387052513409,
      "grad_norm": 0.0,
      "learning_rate": 1.1640911989173242e-05,
      "loss": 0.9616,
      "step": 3721
    },
    {
      "epoch": 0.4642634401895971,
      "grad_norm": 0.0,
      "learning_rate": 1.1636926504401605e-05,
      "loss": 0.9638,
      "step": 3722
    },
    {
      "epoch": 0.4643881751278533,
      "grad_norm": 0.0,
      "learning_rate": 1.1632940752442294e-05,
      "loss": 1.0578,
      "step": 3723
    },
    {
      "epoch": 0.4645129100661095,
      "grad_norm": 0.0,
      "learning_rate": 1.162895473394589e-05,
      "loss": 0.9832,
      "step": 3724
    },
    {
      "epoch": 0.4646376450043657,
      "grad_norm": 0.0,
      "learning_rate": 1.1624968449563005e-05,
      "loss": 0.9844,
      "step": 3725
    },
    {
      "epoch": 0.46476237994262193,
      "grad_norm": 0.0,
      "learning_rate": 1.1620981899944304e-05,
      "loss": 0.9672,
      "step": 3726
    },
    {
      "epoch": 0.46488711488087814,
      "grad_norm": 0.0,
      "learning_rate": 1.1616995085740492e-05,
      "loss": 0.9397,
      "step": 3727
    },
    {
      "epoch": 0.46501184981913435,
      "grad_norm": 0.0,
      "learning_rate": 1.161300800760232e-05,
      "loss": 0.9585,
      "step": 3728
    },
    {
      "epoch": 0.46513658475739056,
      "grad_norm": 0.0,
      "learning_rate": 1.1609020666180574e-05,
      "loss": 0.9757,
      "step": 3729
    },
    {
      "epoch": 0.46526131969564677,
      "grad_norm": 0.0,
      "learning_rate": 1.1605033062126092e-05,
      "loss": 0.9645,
      "step": 3730
    },
    {
      "epoch": 0.465386054633903,
      "grad_norm": 0.0,
      "learning_rate": 1.1601045196089758e-05,
      "loss": 0.9578,
      "step": 3731
    },
    {
      "epoch": 0.4655107895721592,
      "grad_norm": 0.0,
      "learning_rate": 1.1597057068722478e-05,
      "loss": 0.9343,
      "step": 3732
    },
    {
      "epoch": 0.46563552451041534,
      "grad_norm": 0.0,
      "learning_rate": 1.1593068680675227e-05,
      "loss": 0.9254,
      "step": 3733
    },
    {
      "epoch": 0.46576025944867155,
      "grad_norm": 0.0,
      "learning_rate": 1.1589080032599007e-05,
      "loss": 0.9349,
      "step": 3734
    },
    {
      "epoch": 0.46588499438692776,
      "grad_norm": 0.0,
      "learning_rate": 1.1585091125144863e-05,
      "loss": 1.006,
      "step": 3735
    },
    {
      "epoch": 0.466009729325184,
      "grad_norm": 0.0,
      "learning_rate": 1.1581101958963887e-05,
      "loss": 0.9077,
      "step": 3736
    },
    {
      "epoch": 0.4661344642634402,
      "grad_norm": 0.0,
      "learning_rate": 1.1577112534707213e-05,
      "loss": 0.9565,
      "step": 3737
    },
    {
      "epoch": 0.4662591992016964,
      "grad_norm": 0.0,
      "learning_rate": 1.1573122853026015e-05,
      "loss": 0.95,
      "step": 3738
    },
    {
      "epoch": 0.4663839341399526,
      "grad_norm": 0.0,
      "learning_rate": 1.1569132914571509e-05,
      "loss": 0.9482,
      "step": 3739
    },
    {
      "epoch": 0.4665086690782088,
      "grad_norm": 0.0,
      "learning_rate": 1.1565142719994953e-05,
      "loss": 0.9021,
      "step": 3740
    },
    {
      "epoch": 0.466633404016465,
      "grad_norm": 0.0,
      "learning_rate": 1.156115226994765e-05,
      "loss": 0.9783,
      "step": 3741
    },
    {
      "epoch": 0.46675813895472124,
      "grad_norm": 0.0,
      "learning_rate": 1.155716156508094e-05,
      "loss": 0.9683,
      "step": 3742
    },
    {
      "epoch": 0.46688287389297745,
      "grad_norm": 0.0,
      "learning_rate": 1.1553170606046201e-05,
      "loss": 0.949,
      "step": 3743
    },
    {
      "epoch": 0.46700760883123366,
      "grad_norm": 0.0,
      "learning_rate": 1.1549179393494872e-05,
      "loss": 0.9667,
      "step": 3744
    },
    {
      "epoch": 0.4671323437694898,
      "grad_norm": 0.0,
      "learning_rate": 1.1545187928078407e-05,
      "loss": 0.9398,
      "step": 3745
    },
    {
      "epoch": 0.467257078707746,
      "grad_norm": 0.0,
      "learning_rate": 1.1541196210448317e-05,
      "loss": 0.9789,
      "step": 3746
    },
    {
      "epoch": 0.46738181364600223,
      "grad_norm": 0.0,
      "learning_rate": 1.1537204241256155e-05,
      "loss": 0.9683,
      "step": 3747
    },
    {
      "epoch": 0.46750654858425844,
      "grad_norm": 0.0,
      "learning_rate": 1.1533212021153509e-05,
      "loss": 0.9174,
      "step": 3748
    },
    {
      "epoch": 0.46763128352251465,
      "grad_norm": 0.0,
      "learning_rate": 1.1529219550792005e-05,
      "loss": 1.0049,
      "step": 3749
    },
    {
      "epoch": 0.46775601846077086,
      "grad_norm": 0.0,
      "learning_rate": 1.1525226830823322e-05,
      "loss": 0.954,
      "step": 3750
    },
    {
      "epoch": 0.46788075339902707,
      "grad_norm": 0.0,
      "learning_rate": 1.1521233861899168e-05,
      "loss": 0.931,
      "step": 3751
    },
    {
      "epoch": 0.4680054883372833,
      "grad_norm": 0.0,
      "learning_rate": 1.1517240644671297e-05,
      "loss": 0.9671,
      "step": 3752
    },
    {
      "epoch": 0.4681302232755395,
      "grad_norm": 0.0,
      "learning_rate": 1.1513247179791505e-05,
      "loss": 0.9868,
      "step": 3753
    },
    {
      "epoch": 0.4682549582137957,
      "grad_norm": 0.0,
      "learning_rate": 1.1509253467911625e-05,
      "loss": 0.9877,
      "step": 3754
    },
    {
      "epoch": 0.4683796931520519,
      "grad_norm": 0.0,
      "learning_rate": 1.150525950968353e-05,
      "loss": 0.9152,
      "step": 3755
    },
    {
      "epoch": 0.4685044280903081,
      "grad_norm": 0.0,
      "learning_rate": 1.1501265305759136e-05,
      "loss": 0.9465,
      "step": 3756
    },
    {
      "epoch": 0.4686291630285643,
      "grad_norm": 0.0,
      "learning_rate": 1.14972708567904e-05,
      "loss": 0.955,
      "step": 3757
    },
    {
      "epoch": 0.4687538979668205,
      "grad_norm": 0.0,
      "learning_rate": 1.149327616342931e-05,
      "loss": 1.0019,
      "step": 3758
    },
    {
      "epoch": 0.4688786329050767,
      "grad_norm": 0.0,
      "learning_rate": 1.1489281226327906e-05,
      "loss": 1.0068,
      "step": 3759
    },
    {
      "epoch": 0.4690033678433329,
      "grad_norm": 0.0,
      "learning_rate": 1.1485286046138259e-05,
      "loss": 0.932,
      "step": 3760
    },
    {
      "epoch": 0.4691281027815891,
      "grad_norm": 0.0,
      "learning_rate": 1.1481290623512491e-05,
      "loss": 0.9817,
      "step": 3761
    },
    {
      "epoch": 0.46925283771984533,
      "grad_norm": 0.0,
      "learning_rate": 1.1477294959102744e-05,
      "loss": 0.9425,
      "step": 3762
    },
    {
      "epoch": 0.46937757265810154,
      "grad_norm": 0.0,
      "learning_rate": 1.1473299053561222e-05,
      "loss": 0.9206,
      "step": 3763
    },
    {
      "epoch": 0.46950230759635775,
      "grad_norm": 0.0,
      "learning_rate": 1.1469302907540152e-05,
      "loss": 0.9481,
      "step": 3764
    },
    {
      "epoch": 0.46962704253461396,
      "grad_norm": 0.0,
      "learning_rate": 1.1465306521691805e-05,
      "loss": 0.9327,
      "step": 3765
    },
    {
      "epoch": 0.46975177747287017,
      "grad_norm": 0.0,
      "learning_rate": 1.1461309896668493e-05,
      "loss": 0.9412,
      "step": 3766
    },
    {
      "epoch": 0.4698765124111264,
      "grad_norm": 0.0,
      "learning_rate": 1.1457313033122568e-05,
      "loss": 0.9642,
      "step": 3767
    },
    {
      "epoch": 0.4700012473493826,
      "grad_norm": 0.0,
      "learning_rate": 1.1453315931706417e-05,
      "loss": 0.9847,
      "step": 3768
    },
    {
      "epoch": 0.47012598228763874,
      "grad_norm": 0.0,
      "learning_rate": 1.1449318593072468e-05,
      "loss": 0.9211,
      "step": 3769
    },
    {
      "epoch": 0.47025071722589495,
      "grad_norm": 0.0,
      "learning_rate": 1.1445321017873187e-05,
      "loss": 0.9341,
      "step": 3770
    },
    {
      "epoch": 0.47037545216415116,
      "grad_norm": 0.0,
      "learning_rate": 1.1441323206761084e-05,
      "loss": 0.9648,
      "step": 3771
    },
    {
      "epoch": 0.4705001871024074,
      "grad_norm": 0.0,
      "learning_rate": 1.1437325160388692e-05,
      "loss": 0.9826,
      "step": 3772
    },
    {
      "epoch": 0.4706249220406636,
      "grad_norm": 0.0,
      "learning_rate": 1.1433326879408599e-05,
      "loss": 0.979,
      "step": 3773
    },
    {
      "epoch": 0.4707496569789198,
      "grad_norm": 0.0,
      "learning_rate": 1.1429328364473431e-05,
      "loss": 1.0074,
      "step": 3774
    },
    {
      "epoch": 0.470874391917176,
      "grad_norm": 0.0,
      "learning_rate": 1.1425329616235836e-05,
      "loss": 0.9325,
      "step": 3775
    },
    {
      "epoch": 0.4709991268554322,
      "grad_norm": 0.0,
      "learning_rate": 1.1421330635348518e-05,
      "loss": 0.9631,
      "step": 3776
    },
    {
      "epoch": 0.4711238617936884,
      "grad_norm": 0.0,
      "learning_rate": 1.1417331422464206e-05,
      "loss": 0.9208,
      "step": 3777
    },
    {
      "epoch": 0.47124859673194464,
      "grad_norm": 0.0,
      "learning_rate": 1.1413331978235677e-05,
      "loss": 0.9405,
      "step": 3778
    },
    {
      "epoch": 0.47137333167020085,
      "grad_norm": 0.0,
      "learning_rate": 1.140933230331574e-05,
      "loss": 0.9559,
      "step": 3779
    },
    {
      "epoch": 0.47149806660845706,
      "grad_norm": 0.0,
      "learning_rate": 1.1405332398357239e-05,
      "loss": 0.9563,
      "step": 3780
    },
    {
      "epoch": 0.4716228015467132,
      "grad_norm": 0.0,
      "learning_rate": 1.1401332264013063e-05,
      "loss": 0.9993,
      "step": 3781
    },
    {
      "epoch": 0.4717475364849694,
      "grad_norm": 0.0,
      "learning_rate": 1.1397331900936137e-05,
      "loss": 0.9478,
      "step": 3782
    },
    {
      "epoch": 0.47187227142322563,
      "grad_norm": 0.0,
      "learning_rate": 1.1393331309779417e-05,
      "loss": 0.9361,
      "step": 3783
    },
    {
      "epoch": 0.47199700636148184,
      "grad_norm": 0.0,
      "learning_rate": 1.13893304911959e-05,
      "loss": 0.927,
      "step": 3784
    },
    {
      "epoch": 0.47212174129973805,
      "grad_norm": 0.0,
      "learning_rate": 1.1385329445838623e-05,
      "loss": 0.97,
      "step": 3785
    },
    {
      "epoch": 0.47224647623799426,
      "grad_norm": 0.0,
      "learning_rate": 1.1381328174360655e-05,
      "loss": 1.0228,
      "step": 3786
    },
    {
      "epoch": 0.47237121117625047,
      "grad_norm": 0.0,
      "learning_rate": 1.1377326677415108e-05,
      "loss": 0.9361,
      "step": 3787
    },
    {
      "epoch": 0.4724959461145067,
      "grad_norm": 0.0,
      "learning_rate": 1.1373324955655127e-05,
      "loss": 0.9198,
      "step": 3788
    },
    {
      "epoch": 0.4726206810527629,
      "grad_norm": 0.0,
      "learning_rate": 1.1369323009733888e-05,
      "loss": 0.9594,
      "step": 3789
    },
    {
      "epoch": 0.4727454159910191,
      "grad_norm": 0.0,
      "learning_rate": 1.1365320840304615e-05,
      "loss": 0.951,
      "step": 3790
    },
    {
      "epoch": 0.4728701509292753,
      "grad_norm": 0.0,
      "learning_rate": 1.1361318448020563e-05,
      "loss": 1.0014,
      "step": 3791
    },
    {
      "epoch": 0.4729948858675315,
      "grad_norm": 0.0,
      "learning_rate": 1.1357315833535016e-05,
      "loss": 0.9829,
      "step": 3792
    },
    {
      "epoch": 0.4731196208057877,
      "grad_norm": 0.0,
      "learning_rate": 1.1353312997501313e-05,
      "loss": 0.9905,
      "step": 3793
    },
    {
      "epoch": 0.4732443557440439,
      "grad_norm": 0.0,
      "learning_rate": 1.134930994057281e-05,
      "loss": 0.9971,
      "step": 3794
    },
    {
      "epoch": 0.4733690906823001,
      "grad_norm": 0.0,
      "learning_rate": 1.1345306663402909e-05,
      "loss": 0.9543,
      "step": 3795
    },
    {
      "epoch": 0.4734938256205563,
      "grad_norm": 0.0,
      "learning_rate": 1.1341303166645043e-05,
      "loss": 0.9691,
      "step": 3796
    },
    {
      "epoch": 0.4736185605588125,
      "grad_norm": 0.0,
      "learning_rate": 1.133729945095269e-05,
      "loss": 0.9722,
      "step": 3797
    },
    {
      "epoch": 0.47374329549706873,
      "grad_norm": 0.0,
      "learning_rate": 1.133329551697935e-05,
      "loss": 0.9076,
      "step": 3798
    },
    {
      "epoch": 0.47386803043532494,
      "grad_norm": 0.0,
      "learning_rate": 1.1329291365378574e-05,
      "loss": 0.9719,
      "step": 3799
    },
    {
      "epoch": 0.47399276537358115,
      "grad_norm": 0.0,
      "learning_rate": 1.1325286996803932e-05,
      "loss": 0.9598,
      "step": 3800
    },
    {
      "epoch": 0.47411750031183736,
      "grad_norm": 0.0,
      "learning_rate": 1.1321282411909044e-05,
      "loss": 0.9925,
      "step": 3801
    },
    {
      "epoch": 0.47424223525009357,
      "grad_norm": 0.0,
      "learning_rate": 1.1317277611347554e-05,
      "loss": 0.9266,
      "step": 3802
    },
    {
      "epoch": 0.4743669701883498,
      "grad_norm": 0.0,
      "learning_rate": 1.131327259577315e-05,
      "loss": 0.9891,
      "step": 3803
    },
    {
      "epoch": 0.474491705126606,
      "grad_norm": 0.0,
      "learning_rate": 1.1309267365839556e-05,
      "loss": 0.9592,
      "step": 3804
    },
    {
      "epoch": 0.47461644006486214,
      "grad_norm": 0.0,
      "learning_rate": 1.130526192220052e-05,
      "loss": 0.9899,
      "step": 3805
    },
    {
      "epoch": 0.47474117500311835,
      "grad_norm": 0.0,
      "learning_rate": 1.1301256265509827e-05,
      "loss": 0.9372,
      "step": 3806
    },
    {
      "epoch": 0.47486590994137456,
      "grad_norm": 0.0,
      "learning_rate": 1.1297250396421314e-05,
      "loss": 0.9671,
      "step": 3807
    },
    {
      "epoch": 0.4749906448796308,
      "grad_norm": 0.0,
      "learning_rate": 1.1293244315588827e-05,
      "loss": 0.941,
      "step": 3808
    },
    {
      "epoch": 0.475115379817887,
      "grad_norm": 0.0,
      "learning_rate": 1.1289238023666267e-05,
      "loss": 0.9071,
      "step": 3809
    },
    {
      "epoch": 0.4752401147561432,
      "grad_norm": 0.0,
      "learning_rate": 1.1285231521307562e-05,
      "loss": 0.9689,
      "step": 3810
    },
    {
      "epoch": 0.4753648496943994,
      "grad_norm": 0.0,
      "learning_rate": 1.1281224809166673e-05,
      "loss": 1.0019,
      "step": 3811
    },
    {
      "epoch": 0.4754895846326556,
      "grad_norm": 0.0,
      "learning_rate": 1.1277217887897592e-05,
      "loss": 1.0149,
      "step": 3812
    },
    {
      "epoch": 0.4756143195709118,
      "grad_norm": 0.0,
      "learning_rate": 1.1273210758154354e-05,
      "loss": 0.9078,
      "step": 3813
    },
    {
      "epoch": 0.47573905450916804,
      "grad_norm": 0.0,
      "learning_rate": 1.1269203420591024e-05,
      "loss": 0.9556,
      "step": 3814
    },
    {
      "epoch": 0.47586378944742425,
      "grad_norm": 0.0,
      "learning_rate": 1.1265195875861698e-05,
      "loss": 0.9222,
      "step": 3815
    },
    {
      "epoch": 0.47598852438568046,
      "grad_norm": 0.0,
      "learning_rate": 1.1261188124620512e-05,
      "loss": 0.9685,
      "step": 3816
    },
    {
      "epoch": 0.4761132593239366,
      "grad_norm": 0.0,
      "learning_rate": 1.125718016752163e-05,
      "loss": 0.9776,
      "step": 3817
    },
    {
      "epoch": 0.4762379942621928,
      "grad_norm": 0.0,
      "learning_rate": 1.1253172005219249e-05,
      "loss": 0.931,
      "step": 3818
    },
    {
      "epoch": 0.47636272920044903,
      "grad_norm": 0.0,
      "learning_rate": 1.1249163638367603e-05,
      "loss": 0.8946,
      "step": 3819
    },
    {
      "epoch": 0.47648746413870524,
      "grad_norm": 0.0,
      "learning_rate": 1.1245155067620963e-05,
      "loss": 0.9534,
      "step": 3820
    },
    {
      "epoch": 0.47661219907696145,
      "grad_norm": 0.0,
      "learning_rate": 1.1241146293633625e-05,
      "loss": 0.923,
      "step": 3821
    },
    {
      "epoch": 0.47673693401521766,
      "grad_norm": 0.0,
      "learning_rate": 1.1237137317059923e-05,
      "loss": 0.9775,
      "step": 3822
    },
    {
      "epoch": 0.47686166895347387,
      "grad_norm": 0.0,
      "learning_rate": 1.1233128138554222e-05,
      "loss": 0.9928,
      "step": 3823
    },
    {
      "epoch": 0.4769864038917301,
      "grad_norm": 0.0,
      "learning_rate": 1.1229118758770923e-05,
      "loss": 0.8817,
      "step": 3824
    },
    {
      "epoch": 0.4771111388299863,
      "grad_norm": 0.0,
      "learning_rate": 1.1225109178364456e-05,
      "loss": 0.947,
      "step": 3825
    },
    {
      "epoch": 0.4772358737682425,
      "grad_norm": 0.0,
      "learning_rate": 1.1221099397989283e-05,
      "loss": 0.9872,
      "step": 3826
    },
    {
      "epoch": 0.4773606087064987,
      "grad_norm": 0.0,
      "learning_rate": 1.1217089418299914e-05,
      "loss": 0.9364,
      "step": 3827
    },
    {
      "epoch": 0.4774853436447549,
      "grad_norm": 0.0,
      "learning_rate": 1.1213079239950861e-05,
      "loss": 0.954,
      "step": 3828
    },
    {
      "epoch": 0.4776100785830111,
      "grad_norm": 0.0,
      "learning_rate": 1.1209068863596695e-05,
      "loss": 0.9241,
      "step": 3829
    },
    {
      "epoch": 0.4777348135212673,
      "grad_norm": 0.0,
      "learning_rate": 1.1205058289892012e-05,
      "loss": 0.969,
      "step": 3830
    },
    {
      "epoch": 0.4778595484595235,
      "grad_norm": 0.0,
      "learning_rate": 1.1201047519491434e-05,
      "loss": 0.9923,
      "step": 3831
    },
    {
      "epoch": 0.4779842833977797,
      "grad_norm": 0.0,
      "learning_rate": 1.1197036553049626e-05,
      "loss": 0.9319,
      "step": 3832
    },
    {
      "epoch": 0.4781090183360359,
      "grad_norm": 0.0,
      "learning_rate": 1.1193025391221275e-05,
      "loss": 0.907,
      "step": 3833
    },
    {
      "epoch": 0.47823375327429213,
      "grad_norm": 0.0,
      "learning_rate": 1.1189014034661106e-05,
      "loss": 0.9779,
      "step": 3834
    },
    {
      "epoch": 0.47835848821254834,
      "grad_norm": 0.0,
      "learning_rate": 1.1185002484023867e-05,
      "loss": 0.9296,
      "step": 3835
    },
    {
      "epoch": 0.47848322315080455,
      "grad_norm": 0.0,
      "learning_rate": 1.1180990739964348e-05,
      "loss": 0.9629,
      "step": 3836
    },
    {
      "epoch": 0.47860795808906076,
      "grad_norm": 0.0,
      "learning_rate": 1.1176978803137372e-05,
      "loss": 0.9963,
      "step": 3837
    },
    {
      "epoch": 0.47873269302731697,
      "grad_norm": 0.0,
      "learning_rate": 1.1172966674197785e-05,
      "loss": 0.9625,
      "step": 3838
    },
    {
      "epoch": 0.4788574279655732,
      "grad_norm": 0.0,
      "learning_rate": 1.1168954353800463e-05,
      "loss": 0.9598,
      "step": 3839
    },
    {
      "epoch": 0.4789821629038294,
      "grad_norm": 0.0,
      "learning_rate": 1.1164941842600323e-05,
      "loss": 0.9092,
      "step": 3840
    },
    {
      "epoch": 0.47910689784208554,
      "grad_norm": 0.0,
      "learning_rate": 1.1160929141252303e-05,
      "loss": 0.9655,
      "step": 3841
    },
    {
      "epoch": 0.47923163278034175,
      "grad_norm": 0.0,
      "learning_rate": 1.1156916250411383e-05,
      "loss": 0.9672,
      "step": 3842
    },
    {
      "epoch": 0.47935636771859796,
      "grad_norm": 0.0,
      "learning_rate": 1.1152903170732565e-05,
      "loss": 0.9531,
      "step": 3843
    },
    {
      "epoch": 0.4794811026568542,
      "grad_norm": 0.0,
      "learning_rate": 1.1148889902870888e-05,
      "loss": 0.9435,
      "step": 3844
    },
    {
      "epoch": 0.4796058375951104,
      "grad_norm": 0.0,
      "learning_rate": 1.1144876447481409e-05,
      "loss": 0.9408,
      "step": 3845
    },
    {
      "epoch": 0.4797305725333666,
      "grad_norm": 0.0,
      "learning_rate": 1.1140862805219235e-05,
      "loss": 0.9843,
      "step": 3846
    },
    {
      "epoch": 0.4798553074716228,
      "grad_norm": 0.0,
      "learning_rate": 1.1136848976739492e-05,
      "loss": 0.9744,
      "step": 3847
    },
    {
      "epoch": 0.479980042409879,
      "grad_norm": 0.0,
      "learning_rate": 1.1132834962697335e-05,
      "loss": 0.9308,
      "step": 3848
    },
    {
      "epoch": 0.4801047773481352,
      "grad_norm": 0.0,
      "learning_rate": 1.1128820763747949e-05,
      "loss": 0.9222,
      "step": 3849
    },
    {
      "epoch": 0.48022951228639144,
      "grad_norm": 0.0,
      "learning_rate": 1.1124806380546564e-05,
      "loss": 0.9672,
      "step": 3850
    },
    {
      "epoch": 0.48035424722464765,
      "grad_norm": 0.0,
      "learning_rate": 1.1120791813748419e-05,
      "loss": 0.9977,
      "step": 3851
    },
    {
      "epoch": 0.48047898216290386,
      "grad_norm": 0.0,
      "learning_rate": 1.1116777064008793e-05,
      "loss": 0.991,
      "step": 3852
    },
    {
      "epoch": 0.48060371710116,
      "grad_norm": 0.0,
      "learning_rate": 1.1112762131983e-05,
      "loss": 0.921,
      "step": 3853
    },
    {
      "epoch": 0.4807284520394162,
      "grad_norm": 0.0,
      "learning_rate": 1.1108747018326369e-05,
      "loss": 0.9436,
      "step": 3854
    },
    {
      "epoch": 0.48085318697767243,
      "grad_norm": 0.0,
      "learning_rate": 1.1104731723694278e-05,
      "loss": 0.9733,
      "step": 3855
    },
    {
      "epoch": 0.48097792191592864,
      "grad_norm": 0.0,
      "learning_rate": 1.1100716248742119e-05,
      "loss": 0.9286,
      "step": 3856
    },
    {
      "epoch": 0.48110265685418485,
      "grad_norm": 0.0,
      "learning_rate": 1.1096700594125318e-05,
      "loss": 0.9382,
      "step": 3857
    },
    {
      "epoch": 0.48122739179244106,
      "grad_norm": 0.0,
      "learning_rate": 1.1092684760499332e-05,
      "loss": 1.0201,
      "step": 3858
    },
    {
      "epoch": 0.48135212673069727,
      "grad_norm": 0.0,
      "learning_rate": 1.1088668748519646e-05,
      "loss": 0.9911,
      "step": 3859
    },
    {
      "epoch": 0.4814768616689535,
      "grad_norm": 0.0,
      "learning_rate": 1.1084652558841777e-05,
      "loss": 0.9733,
      "step": 3860
    },
    {
      "epoch": 0.4816015966072097,
      "grad_norm": 0.0,
      "learning_rate": 1.1080636192121267e-05,
      "loss": 0.9855,
      "step": 3861
    },
    {
      "epoch": 0.4817263315454659,
      "grad_norm": 0.0,
      "learning_rate": 1.1076619649013687e-05,
      "loss": 0.9667,
      "step": 3862
    },
    {
      "epoch": 0.4818510664837221,
      "grad_norm": 0.0,
      "learning_rate": 1.1072602930174639e-05,
      "loss": 0.9884,
      "step": 3863
    },
    {
      "epoch": 0.4819758014219783,
      "grad_norm": 0.0,
      "learning_rate": 1.1068586036259752e-05,
      "loss": 0.9103,
      "step": 3864
    },
    {
      "epoch": 0.4821005363602345,
      "grad_norm": 0.0,
      "learning_rate": 1.1064568967924683e-05,
      "loss": 0.9322,
      "step": 3865
    },
    {
      "epoch": 0.4822252712984907,
      "grad_norm": 0.0,
      "learning_rate": 1.1060551725825124e-05,
      "loss": 0.9297,
      "step": 3866
    },
    {
      "epoch": 0.4823500062367469,
      "grad_norm": 0.0,
      "learning_rate": 1.1056534310616787e-05,
      "loss": 0.9706,
      "step": 3867
    },
    {
      "epoch": 0.4824747411750031,
      "grad_norm": 0.0,
      "learning_rate": 1.1052516722955412e-05,
      "loss": 1.004,
      "step": 3868
    },
    {
      "epoch": 0.4825994761132593,
      "grad_norm": 0.0,
      "learning_rate": 1.1048498963496776e-05,
      "loss": 0.9647,
      "step": 3869
    },
    {
      "epoch": 0.48272421105151553,
      "grad_norm": 0.0,
      "learning_rate": 1.104448103289668e-05,
      "loss": 0.9769,
      "step": 3870
    },
    {
      "epoch": 0.48284894598977174,
      "grad_norm": 0.0,
      "learning_rate": 1.1040462931810942e-05,
      "loss": 0.9903,
      "step": 3871
    },
    {
      "epoch": 0.48297368092802795,
      "grad_norm": 0.0,
      "learning_rate": 1.103644466089543e-05,
      "loss": 0.945,
      "step": 3872
    },
    {
      "epoch": 0.48309841586628416,
      "grad_norm": 0.0,
      "learning_rate": 1.1032426220806018e-05,
      "loss": 0.9402,
      "step": 3873
    },
    {
      "epoch": 0.48322315080454037,
      "grad_norm": 0.0,
      "learning_rate": 1.1028407612198618e-05,
      "loss": 0.9069,
      "step": 3874
    },
    {
      "epoch": 0.4833478857427966,
      "grad_norm": 0.0,
      "learning_rate": 1.102438883572917e-05,
      "loss": 0.9577,
      "step": 3875
    },
    {
      "epoch": 0.4834726206810528,
      "grad_norm": 0.0,
      "learning_rate": 1.102036989205364e-05,
      "loss": 0.8837,
      "step": 3876
    },
    {
      "epoch": 0.48359735561930894,
      "grad_norm": 0.0,
      "learning_rate": 1.101635078182802e-05,
      "loss": 0.9674,
      "step": 3877
    },
    {
      "epoch": 0.48372209055756515,
      "grad_norm": 0.0,
      "learning_rate": 1.1012331505708332e-05,
      "loss": 0.9578,
      "step": 3878
    },
    {
      "epoch": 0.48384682549582136,
      "grad_norm": 0.0,
      "learning_rate": 1.100831206435062e-05,
      "loss": 1.0,
      "step": 3879
    },
    {
      "epoch": 0.4839715604340776,
      "grad_norm": 0.0,
      "learning_rate": 1.1004292458410959e-05,
      "loss": 0.9657,
      "step": 3880
    },
    {
      "epoch": 0.4840962953723338,
      "grad_norm": 0.0,
      "learning_rate": 1.100027268854545e-05,
      "loss": 0.9706,
      "step": 3881
    },
    {
      "epoch": 0.48422103031059,
      "grad_norm": 0.0,
      "learning_rate": 1.0996252755410224e-05,
      "loss": 1.008,
      "step": 3882
    },
    {
      "epoch": 0.4843457652488462,
      "grad_norm": 0.0,
      "learning_rate": 1.0992232659661433e-05,
      "loss": 0.9431,
      "step": 3883
    },
    {
      "epoch": 0.4844705001871024,
      "grad_norm": 0.0,
      "learning_rate": 1.0988212401955259e-05,
      "loss": 0.9376,
      "step": 3884
    },
    {
      "epoch": 0.4845952351253586,
      "grad_norm": 0.0,
      "learning_rate": 1.0984191982947902e-05,
      "loss": 0.9706,
      "step": 3885
    },
    {
      "epoch": 0.48471997006361484,
      "grad_norm": 0.0,
      "learning_rate": 1.098017140329561e-05,
      "loss": 0.9596,
      "step": 3886
    },
    {
      "epoch": 0.48484470500187105,
      "grad_norm": 0.0,
      "learning_rate": 1.0976150663654627e-05,
      "loss": 0.913,
      "step": 3887
    },
    {
      "epoch": 0.48496943994012726,
      "grad_norm": 0.0,
      "learning_rate": 1.0972129764681252e-05,
      "loss": 1.0045,
      "step": 3888
    },
    {
      "epoch": 0.4850941748783834,
      "grad_norm": 0.0,
      "learning_rate": 1.0968108707031792e-05,
      "loss": 0.9479,
      "step": 3889
    },
    {
      "epoch": 0.4852189098166396,
      "grad_norm": 0.0,
      "learning_rate": 1.0964087491362588e-05,
      "loss": 0.9772,
      "step": 3890
    },
    {
      "epoch": 0.48534364475489583,
      "grad_norm": 0.0,
      "learning_rate": 1.0960066118329995e-05,
      "loss": 0.9601,
      "step": 3891
    },
    {
      "epoch": 0.48546837969315204,
      "grad_norm": 0.0,
      "learning_rate": 1.0956044588590408e-05,
      "loss": 0.945,
      "step": 3892
    },
    {
      "epoch": 0.48559311463140825,
      "grad_norm": 0.0,
      "learning_rate": 1.0952022902800248e-05,
      "loss": 0.9052,
      "step": 3893
    },
    {
      "epoch": 0.48571784956966446,
      "grad_norm": 0.0,
      "learning_rate": 1.0948001061615945e-05,
      "loss": 0.96,
      "step": 3894
    },
    {
      "epoch": 0.48584258450792067,
      "grad_norm": 0.0,
      "learning_rate": 1.0943979065693974e-05,
      "loss": 0.9605,
      "step": 3895
    },
    {
      "epoch": 0.4859673194461769,
      "grad_norm": 0.0,
      "learning_rate": 1.0939956915690819e-05,
      "loss": 0.9764,
      "step": 3896
    },
    {
      "epoch": 0.4860920543844331,
      "grad_norm": 0.0,
      "learning_rate": 1.0935934612263002e-05,
      "loss": 0.9297,
      "step": 3897
    },
    {
      "epoch": 0.4862167893226893,
      "grad_norm": 0.0,
      "learning_rate": 1.0931912156067058e-05,
      "loss": 0.9226,
      "step": 3898
    },
    {
      "epoch": 0.4863415242609455,
      "grad_norm": 0.0,
      "learning_rate": 1.0927889547759555e-05,
      "loss": 0.9625,
      "step": 3899
    },
    {
      "epoch": 0.4864662591992017,
      "grad_norm": 0.0,
      "learning_rate": 1.0923866787997093e-05,
      "loss": 0.968,
      "step": 3900
    },
    {
      "epoch": 0.4865909941374579,
      "grad_norm": 0.0,
      "learning_rate": 1.0919843877436276e-05,
      "loss": 0.9646,
      "step": 3901
    },
    {
      "epoch": 0.4867157290757141,
      "grad_norm": 0.0,
      "learning_rate": 1.0915820816733748e-05,
      "loss": 0.9583,
      "step": 3902
    },
    {
      "epoch": 0.4868404640139703,
      "grad_norm": 0.0,
      "learning_rate": 1.091179760654618e-05,
      "loss": 0.979,
      "step": 3903
    },
    {
      "epoch": 0.4869651989522265,
      "grad_norm": 0.0,
      "learning_rate": 1.0907774247530252e-05,
      "loss": 0.9392,
      "step": 3904
    },
    {
      "epoch": 0.4870899338904827,
      "grad_norm": 0.0,
      "learning_rate": 1.0903750740342682e-05,
      "loss": 0.9846,
      "step": 3905
    },
    {
      "epoch": 0.48721466882873893,
      "grad_norm": 0.0,
      "learning_rate": 1.0899727085640209e-05,
      "loss": 0.9757,
      "step": 3906
    },
    {
      "epoch": 0.48733940376699514,
      "grad_norm": 0.0,
      "learning_rate": 1.0895703284079595e-05,
      "loss": 0.9683,
      "step": 3907
    },
    {
      "epoch": 0.48746413870525135,
      "grad_norm": 0.0,
      "learning_rate": 1.0891679336317622e-05,
      "loss": 0.9979,
      "step": 3908
    },
    {
      "epoch": 0.48758887364350756,
      "grad_norm": 0.0,
      "learning_rate": 1.08876552430111e-05,
      "loss": 0.9209,
      "step": 3909
    },
    {
      "epoch": 0.48771360858176377,
      "grad_norm": 0.0,
      "learning_rate": 1.088363100481687e-05,
      "loss": 0.9516,
      "step": 3910
    },
    {
      "epoch": 0.48783834352002,
      "grad_norm": 0.0,
      "learning_rate": 1.0879606622391781e-05,
      "loss": 0.9694,
      "step": 3911
    },
    {
      "epoch": 0.4879630784582762,
      "grad_norm": 0.0,
      "learning_rate": 1.087558209639272e-05,
      "loss": 0.9694,
      "step": 3912
    },
    {
      "epoch": 0.48808781339653234,
      "grad_norm": 0.0,
      "learning_rate": 1.0871557427476585e-05,
      "loss": 0.9634,
      "step": 3913
    },
    {
      "epoch": 0.48821254833478855,
      "grad_norm": 0.0,
      "learning_rate": 1.0867532616300305e-05,
      "loss": 0.9592,
      "step": 3914
    },
    {
      "epoch": 0.48833728327304476,
      "grad_norm": 0.0,
      "learning_rate": 1.0863507663520831e-05,
      "loss": 0.9771,
      "step": 3915
    },
    {
      "epoch": 0.488462018211301,
      "grad_norm": 0.0,
      "learning_rate": 1.085948256979514e-05,
      "loss": 0.9707,
      "step": 3916
    },
    {
      "epoch": 0.4885867531495572,
      "grad_norm": 0.0,
      "learning_rate": 1.0855457335780224e-05,
      "loss": 0.962,
      "step": 3917
    },
    {
      "epoch": 0.4887114880878134,
      "grad_norm": 0.0,
      "learning_rate": 1.0851431962133106e-05,
      "loss": 0.9139,
      "step": 3918
    },
    {
      "epoch": 0.4888362230260696,
      "grad_norm": 0.0,
      "learning_rate": 1.0847406449510827e-05,
      "loss": 0.9126,
      "step": 3919
    },
    {
      "epoch": 0.4889609579643258,
      "grad_norm": 0.0,
      "learning_rate": 1.0843380798570452e-05,
      "loss": 0.9019,
      "step": 3920
    },
    {
      "epoch": 0.489085692902582,
      "grad_norm": 0.0,
      "learning_rate": 1.0839355009969068e-05,
      "loss": 0.9677,
      "step": 3921
    },
    {
      "epoch": 0.48921042784083824,
      "grad_norm": 0.0,
      "learning_rate": 1.0835329084363787e-05,
      "loss": 0.9314,
      "step": 3922
    },
    {
      "epoch": 0.48933516277909445,
      "grad_norm": 0.0,
      "learning_rate": 1.0831303022411745e-05,
      "loss": 0.9767,
      "step": 3923
    },
    {
      "epoch": 0.48945989771735066,
      "grad_norm": 0.0,
      "learning_rate": 1.082727682477009e-05,
      "loss": 0.947,
      "step": 3924
    },
    {
      "epoch": 0.4895846326556068,
      "grad_norm": 0.0,
      "learning_rate": 1.0823250492096e-05,
      "loss": 0.9161,
      "step": 3925
    },
    {
      "epoch": 0.489709367593863,
      "grad_norm": 0.0,
      "learning_rate": 1.0819224025046677e-05,
      "loss": 0.9562,
      "step": 3926
    },
    {
      "epoch": 0.48983410253211923,
      "grad_norm": 0.0,
      "learning_rate": 1.081519742427934e-05,
      "loss": 0.9412,
      "step": 3927
    },
    {
      "epoch": 0.48995883747037544,
      "grad_norm": 0.0,
      "learning_rate": 1.0811170690451232e-05,
      "loss": 0.9313,
      "step": 3928
    },
    {
      "epoch": 0.49008357240863165,
      "grad_norm": 0.0,
      "learning_rate": 1.0807143824219622e-05,
      "loss": 0.9777,
      "step": 3929
    },
    {
      "epoch": 0.49020830734688786,
      "grad_norm": 0.0,
      "learning_rate": 1.0803116826241789e-05,
      "loss": 0.9613,
      "step": 3930
    },
    {
      "epoch": 0.49033304228514407,
      "grad_norm": 0.0,
      "learning_rate": 1.0799089697175041e-05,
      "loss": 0.9558,
      "step": 3931
    },
    {
      "epoch": 0.4904577772234003,
      "grad_norm": 0.0,
      "learning_rate": 1.0795062437676712e-05,
      "loss": 0.9099,
      "step": 3932
    },
    {
      "epoch": 0.4905825121616565,
      "grad_norm": 0.0,
      "learning_rate": 1.0791035048404148e-05,
      "loss": 0.9569,
      "step": 3933
    },
    {
      "epoch": 0.4907072470999127,
      "grad_norm": 0.0,
      "learning_rate": 1.0787007530014723e-05,
      "loss": 0.9577,
      "step": 3934
    },
    {
      "epoch": 0.4908319820381689,
      "grad_norm": 0.0,
      "learning_rate": 1.0782979883165832e-05,
      "loss": 0.9624,
      "step": 3935
    },
    {
      "epoch": 0.4909567169764251,
      "grad_norm": 0.0,
      "learning_rate": 1.0778952108514882e-05,
      "loss": 0.9895,
      "step": 3936
    },
    {
      "epoch": 0.4910814519146813,
      "grad_norm": 0.0,
      "learning_rate": 1.077492420671931e-05,
      "loss": 0.9889,
      "step": 3937
    },
    {
      "epoch": 0.4912061868529375,
      "grad_norm": 0.0,
      "learning_rate": 1.0770896178436572e-05,
      "loss": 0.9694,
      "step": 3938
    },
    {
      "epoch": 0.4913309217911937,
      "grad_norm": 0.0,
      "learning_rate": 1.0766868024324146e-05,
      "loss": 1.0258,
      "step": 3939
    },
    {
      "epoch": 0.4914556567294499,
      "grad_norm": 0.0,
      "learning_rate": 1.0762839745039526e-05,
      "loss": 0.9964,
      "step": 3940
    },
    {
      "epoch": 0.4915803916677061,
      "grad_norm": 0.0,
      "learning_rate": 1.0758811341240226e-05,
      "loss": 1.0452,
      "step": 3941
    },
    {
      "epoch": 0.49170512660596233,
      "grad_norm": 0.0,
      "learning_rate": 1.0754782813583784e-05,
      "loss": 0.9506,
      "step": 3942
    },
    {
      "epoch": 0.49182986154421854,
      "grad_norm": 0.0,
      "learning_rate": 1.0750754162727763e-05,
      "loss": 0.9921,
      "step": 3943
    },
    {
      "epoch": 0.49195459648247475,
      "grad_norm": 0.0,
      "learning_rate": 1.0746725389329733e-05,
      "loss": 0.951,
      "step": 3944
    },
    {
      "epoch": 0.49207933142073096,
      "grad_norm": 0.0,
      "learning_rate": 1.0742696494047295e-05,
      "loss": 0.9118,
      "step": 3945
    },
    {
      "epoch": 0.49220406635898717,
      "grad_norm": 0.0,
      "learning_rate": 1.0738667477538072e-05,
      "loss": 0.946,
      "step": 3946
    },
    {
      "epoch": 0.4923288012972434,
      "grad_norm": 0.0,
      "learning_rate": 1.0734638340459688e-05,
      "loss": 0.9108,
      "step": 3947
    },
    {
      "epoch": 0.4924535362354996,
      "grad_norm": 0.0,
      "learning_rate": 1.0730609083469808e-05,
      "loss": 0.9887,
      "step": 3948
    },
    {
      "epoch": 0.49257827117375574,
      "grad_norm": 0.0,
      "learning_rate": 1.0726579707226108e-05,
      "loss": 0.9034,
      "step": 3949
    },
    {
      "epoch": 0.49270300611201195,
      "grad_norm": 0.0,
      "learning_rate": 1.0722550212386282e-05,
      "loss": 0.9181,
      "step": 3950
    },
    {
      "epoch": 0.49282774105026816,
      "grad_norm": 0.0,
      "learning_rate": 1.0718520599608046e-05,
      "loss": 0.9475,
      "step": 3951
    },
    {
      "epoch": 0.4929524759885244,
      "grad_norm": 0.0,
      "learning_rate": 1.0714490869549134e-05,
      "loss": 0.9278,
      "step": 3952
    },
    {
      "epoch": 0.4930772109267806,
      "grad_norm": 0.0,
      "learning_rate": 1.0710461022867303e-05,
      "loss": 0.9341,
      "step": 3953
    },
    {
      "epoch": 0.4932019458650368,
      "grad_norm": 0.0,
      "learning_rate": 1.0706431060220318e-05,
      "loss": 0.9401,
      "step": 3954
    },
    {
      "epoch": 0.493326680803293,
      "grad_norm": 0.0,
      "learning_rate": 1.0702400982265977e-05,
      "loss": 0.935,
      "step": 3955
    },
    {
      "epoch": 0.4934514157415492,
      "grad_norm": 0.0,
      "learning_rate": 1.069837078966209e-05,
      "loss": 1.0162,
      "step": 3956
    },
    {
      "epoch": 0.4935761506798054,
      "grad_norm": 0.0,
      "learning_rate": 1.0694340483066483e-05,
      "loss": 0.9545,
      "step": 3957
    },
    {
      "epoch": 0.49370088561806164,
      "grad_norm": 0.0,
      "learning_rate": 1.0690310063137003e-05,
      "loss": 0.9531,
      "step": 3958
    },
    {
      "epoch": 0.49382562055631785,
      "grad_norm": 0.0,
      "learning_rate": 1.068627953053152e-05,
      "loss": 0.9886,
      "step": 3959
    },
    {
      "epoch": 0.49395035549457406,
      "grad_norm": 0.0,
      "learning_rate": 1.0682248885907916e-05,
      "loss": 1.0141,
      "step": 3960
    },
    {
      "epoch": 0.4940750904328302,
      "grad_norm": 0.0,
      "learning_rate": 1.0678218129924093e-05,
      "loss": 0.9634,
      "step": 3961
    },
    {
      "epoch": 0.4941998253710864,
      "grad_norm": 0.0,
      "learning_rate": 1.0674187263237977e-05,
      "loss": 0.9422,
      "step": 3962
    },
    {
      "epoch": 0.49432456030934263,
      "grad_norm": 0.0,
      "learning_rate": 1.0670156286507503e-05,
      "loss": 0.9833,
      "step": 3963
    },
    {
      "epoch": 0.49444929524759884,
      "grad_norm": 0.0,
      "learning_rate": 1.0666125200390627e-05,
      "loss": 0.9321,
      "step": 3964
    },
    {
      "epoch": 0.49457403018585505,
      "grad_norm": 0.0,
      "learning_rate": 1.0662094005545322e-05,
      "loss": 0.9767,
      "step": 3965
    },
    {
      "epoch": 0.49469876512411126,
      "grad_norm": 0.0,
      "learning_rate": 1.065806270262959e-05,
      "loss": 0.998,
      "step": 3966
    },
    {
      "epoch": 0.49482350006236747,
      "grad_norm": 0.0,
      "learning_rate": 1.0654031292301432e-05,
      "loss": 1.0105,
      "step": 3967
    },
    {
      "epoch": 0.4949482350006237,
      "grad_norm": 0.0,
      "learning_rate": 1.064999977521888e-05,
      "loss": 0.9239,
      "step": 3968
    },
    {
      "epoch": 0.4950729699388799,
      "grad_norm": 0.0,
      "learning_rate": 1.064596815203998e-05,
      "loss": 0.9348,
      "step": 3969
    },
    {
      "epoch": 0.4951977048771361,
      "grad_norm": 0.0,
      "learning_rate": 1.0641936423422789e-05,
      "loss": 0.9702,
      "step": 3970
    },
    {
      "epoch": 0.4953224398153923,
      "grad_norm": 0.0,
      "learning_rate": 1.0637904590025392e-05,
      "loss": 0.9978,
      "step": 3971
    },
    {
      "epoch": 0.4954471747536485,
      "grad_norm": 0.0,
      "learning_rate": 1.0633872652505885e-05,
      "loss": 1.0193,
      "step": 3972
    },
    {
      "epoch": 0.4955719096919047,
      "grad_norm": 0.0,
      "learning_rate": 1.0629840611522382e-05,
      "loss": 0.9318,
      "step": 3973
    },
    {
      "epoch": 0.4956966446301609,
      "grad_norm": 0.0,
      "learning_rate": 1.0625808467733012e-05,
      "loss": 0.9812,
      "step": 3974
    },
    {
      "epoch": 0.4958213795684171,
      "grad_norm": 0.0,
      "learning_rate": 1.0621776221795923e-05,
      "loss": 0.956,
      "step": 3975
    },
    {
      "epoch": 0.4959461145066733,
      "grad_norm": 0.0,
      "learning_rate": 1.0617743874369282e-05,
      "loss": 0.9142,
      "step": 3976
    },
    {
      "epoch": 0.4960708494449295,
      "grad_norm": 0.0,
      "learning_rate": 1.0613711426111264e-05,
      "loss": 0.9664,
      "step": 3977
    },
    {
      "epoch": 0.49619558438318573,
      "grad_norm": 0.0,
      "learning_rate": 1.0609678877680069e-05,
      "loss": 0.9619,
      "step": 3978
    },
    {
      "epoch": 0.49632031932144194,
      "grad_norm": 0.0,
      "learning_rate": 1.0605646229733917e-05,
      "loss": 0.9916,
      "step": 3979
    },
    {
      "epoch": 0.49644505425969815,
      "grad_norm": 0.0,
      "learning_rate": 1.0601613482931028e-05,
      "loss": 0.9648,
      "step": 3980
    },
    {
      "epoch": 0.49656978919795436,
      "grad_norm": 0.0,
      "learning_rate": 1.0597580637929652e-05,
      "loss": 1.0033,
      "step": 3981
    },
    {
      "epoch": 0.49669452413621057,
      "grad_norm": 0.0,
      "learning_rate": 1.0593547695388054e-05,
      "loss": 0.9568,
      "step": 3982
    },
    {
      "epoch": 0.4968192590744668,
      "grad_norm": 0.0,
      "learning_rate": 1.0589514655964506e-05,
      "loss": 0.9721,
      "step": 3983
    },
    {
      "epoch": 0.496943994012723,
      "grad_norm": 0.0,
      "learning_rate": 1.0585481520317305e-05,
      "loss": 1.0029,
      "step": 3984
    },
    {
      "epoch": 0.49706872895097914,
      "grad_norm": 0.0,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 1.0147,
      "step": 3985
    },
    {
      "epoch": 0.49719346388923535,
      "grad_norm": 0.0,
      "learning_rate": 1.0577414962985198e-05,
      "loss": 0.9656,
      "step": 3986
    },
    {
      "epoch": 0.49731819882749156,
      "grad_norm": 0.0,
      "learning_rate": 1.0573381542616954e-05,
      "loss": 0.8953,
      "step": 3987
    },
    {
      "epoch": 0.4974429337657478,
      "grad_norm": 0.0,
      "learning_rate": 1.0569348028658388e-05,
      "loss": 0.9929,
      "step": 3988
    },
    {
      "epoch": 0.497567668704004,
      "grad_norm": 0.0,
      "learning_rate": 1.0565314421767871e-05,
      "loss": 0.9874,
      "step": 3989
    },
    {
      "epoch": 0.4976924036422602,
      "grad_norm": 0.0,
      "learning_rate": 1.056128072260379e-05,
      "loss": 0.9307,
      "step": 3990
    },
    {
      "epoch": 0.4978171385805164,
      "grad_norm": 0.0,
      "learning_rate": 1.0557246931824545e-05,
      "loss": 1.0005,
      "step": 3991
    },
    {
      "epoch": 0.4979418735187726,
      "grad_norm": 0.0,
      "learning_rate": 1.0553213050088553e-05,
      "loss": 0.9684,
      "step": 3992
    },
    {
      "epoch": 0.4980666084570288,
      "grad_norm": 0.0,
      "learning_rate": 1.0549179078054243e-05,
      "loss": 0.9474,
      "step": 3993
    },
    {
      "epoch": 0.49819134339528504,
      "grad_norm": 0.0,
      "learning_rate": 1.0545145016380065e-05,
      "loss": 0.9187,
      "step": 3994
    },
    {
      "epoch": 0.49831607833354125,
      "grad_norm": 0.0,
      "learning_rate": 1.0541110865724475e-05,
      "loss": 0.9652,
      "step": 3995
    },
    {
      "epoch": 0.49844081327179746,
      "grad_norm": 0.0,
      "learning_rate": 1.0537076626745953e-05,
      "loss": 0.9892,
      "step": 3996
    },
    {
      "epoch": 0.4985655482100536,
      "grad_norm": 0.0,
      "learning_rate": 1.0533042300102984e-05,
      "loss": 0.9161,
      "step": 3997
    },
    {
      "epoch": 0.4986902831483098,
      "grad_norm": 0.0,
      "learning_rate": 1.0529007886454075e-05,
      "loss": 0.9359,
      "step": 3998
    },
    {
      "epoch": 0.49881501808656603,
      "grad_norm": 0.0,
      "learning_rate": 1.0524973386457742e-05,
      "loss": 0.9745,
      "step": 3999
    },
    {
      "epoch": 0.49893975302482224,
      "grad_norm": 0.0,
      "learning_rate": 1.0520938800772516e-05,
      "loss": 1.0079,
      "step": 4000
    },
    {
      "epoch": 0.49906448796307845,
      "grad_norm": 0.0,
      "learning_rate": 1.0516904130056946e-05,
      "loss": 0.9459,
      "step": 4001
    },
    {
      "epoch": 0.49918922290133466,
      "grad_norm": 0.0,
      "learning_rate": 1.0512869374969593e-05,
      "loss": 0.9739,
      "step": 4002
    },
    {
      "epoch": 0.49931395783959087,
      "grad_norm": 0.0,
      "learning_rate": 1.0508834536169028e-05,
      "loss": 0.9257,
      "step": 4003
    },
    {
      "epoch": 0.4994386927778471,
      "grad_norm": 0.0,
      "learning_rate": 1.0504799614313837e-05,
      "loss": 0.9673,
      "step": 4004
    },
    {
      "epoch": 0.4995634277161033,
      "grad_norm": 0.0,
      "learning_rate": 1.0500764610062627e-05,
      "loss": 0.998,
      "step": 4005
    },
    {
      "epoch": 0.4996881626543595,
      "grad_norm": 0.0,
      "learning_rate": 1.0496729524074002e-05,
      "loss": 0.9912,
      "step": 4006
    },
    {
      "epoch": 0.4998128975926157,
      "grad_norm": 0.0,
      "learning_rate": 1.0492694357006598e-05,
      "loss": 0.9683,
      "step": 4007
    },
    {
      "epoch": 0.4999376325308719,
      "grad_norm": 0.0,
      "learning_rate": 1.0488659109519059e-05,
      "loss": 0.9466,
      "step": 4008
    },
    {
      "epoch": 0.5000623674691281,
      "grad_norm": 0.0,
      "learning_rate": 1.0484623782270032e-05,
      "loss": 0.9679,
      "step": 4009
    },
    {
      "epoch": 0.5001871024073843,
      "grad_norm": 0.0,
      "learning_rate": 1.0480588375918185e-05,
      "loss": 0.951,
      "step": 4010
    },
    {
      "epoch": 0.5003118373456406,
      "grad_norm": 0.0,
      "learning_rate": 1.0476552891122197e-05,
      "loss": 0.9913,
      "step": 4011
    },
    {
      "epoch": 0.5004365722838967,
      "grad_norm": 0.0,
      "learning_rate": 1.047251732854077e-05,
      "loss": 0.9463,
      "step": 4012
    },
    {
      "epoch": 0.500561307222153,
      "grad_norm": 0.0,
      "learning_rate": 1.04684816888326e-05,
      "loss": 0.963,
      "step": 4013
    },
    {
      "epoch": 0.5006860421604091,
      "grad_norm": 0.0,
      "learning_rate": 1.0464445972656408e-05,
      "loss": 0.9677,
      "step": 4014
    },
    {
      "epoch": 0.5008107770986653,
      "grad_norm": 0.0,
      "learning_rate": 1.0460410180670921e-05,
      "loss": 0.9878,
      "step": 4015
    },
    {
      "epoch": 0.5009355120369215,
      "grad_norm": 0.0,
      "learning_rate": 1.0456374313534891e-05,
      "loss": 0.9436,
      "step": 4016
    },
    {
      "epoch": 0.5010602469751777,
      "grad_norm": 0.0,
      "learning_rate": 1.0452338371907065e-05,
      "loss": 0.905,
      "step": 4017
    },
    {
      "epoch": 0.501184981913434,
      "grad_norm": 0.0,
      "learning_rate": 1.0448302356446214e-05,
      "loss": 0.9537,
      "step": 4018
    },
    {
      "epoch": 0.5013097168516901,
      "grad_norm": 0.0,
      "learning_rate": 1.0444266267811118e-05,
      "loss": 0.9721,
      "step": 4019
    },
    {
      "epoch": 0.5014344517899464,
      "grad_norm": 0.0,
      "learning_rate": 1.0440230106660565e-05,
      "loss": 0.9332,
      "step": 4020
    },
    {
      "epoch": 0.5015591867282025,
      "grad_norm": 0.0,
      "learning_rate": 1.0436193873653362e-05,
      "loss": 0.9367,
      "step": 4021
    },
    {
      "epoch": 0.5016839216664588,
      "grad_norm": 0.0,
      "learning_rate": 1.0432157569448323e-05,
      "loss": 0.9608,
      "step": 4022
    },
    {
      "epoch": 0.501808656604715,
      "grad_norm": 0.0,
      "learning_rate": 1.042812119470427e-05,
      "loss": 0.9861,
      "step": 4023
    },
    {
      "epoch": 0.5019333915429712,
      "grad_norm": 0.0,
      "learning_rate": 1.0424084750080049e-05,
      "loss": 0.9829,
      "step": 4024
    },
    {
      "epoch": 0.5020581264812274,
      "grad_norm": 0.0,
      "learning_rate": 1.0420048236234506e-05,
      "loss": 0.9917,
      "step": 4025
    },
    {
      "epoch": 0.5021828614194837,
      "grad_norm": 0.0,
      "learning_rate": 1.04160116538265e-05,
      "loss": 0.9988,
      "step": 4026
    },
    {
      "epoch": 0.5023075963577398,
      "grad_norm": 0.0,
      "learning_rate": 1.0411975003514905e-05,
      "loss": 0.9876,
      "step": 4027
    },
    {
      "epoch": 0.502432331295996,
      "grad_norm": 0.0,
      "learning_rate": 1.04079382859586e-05,
      "loss": 0.9532,
      "step": 4028
    },
    {
      "epoch": 0.5025570662342522,
      "grad_norm": 0.0,
      "learning_rate": 1.0403901501816489e-05,
      "loss": 0.9775,
      "step": 4029
    },
    {
      "epoch": 0.5026818011725084,
      "grad_norm": 0.0,
      "learning_rate": 1.0399864651747467e-05,
      "loss": 0.9604,
      "step": 4030
    },
    {
      "epoch": 0.5028065361107646,
      "grad_norm": 0.0,
      "learning_rate": 1.0395827736410453e-05,
      "loss": 0.9102,
      "step": 4031
    },
    {
      "epoch": 0.5029312710490208,
      "grad_norm": 0.0,
      "learning_rate": 1.0391790756464377e-05,
      "loss": 0.9514,
      "step": 4032
    },
    {
      "epoch": 0.5030560059872771,
      "grad_norm": 0.0,
      "learning_rate": 1.038775371256817e-05,
      "loss": 0.9995,
      "step": 4033
    },
    {
      "epoch": 0.5031807409255332,
      "grad_norm": 0.0,
      "learning_rate": 1.038371660538078e-05,
      "loss": 0.9175,
      "step": 4034
    },
    {
      "epoch": 0.5033054758637895,
      "grad_norm": 0.0,
      "learning_rate": 1.037967943556117e-05,
      "loss": 1.0132,
      "step": 4035
    },
    {
      "epoch": 0.5034302108020456,
      "grad_norm": 0.0,
      "learning_rate": 1.0375642203768303e-05,
      "loss": 0.9797,
      "step": 4036
    },
    {
      "epoch": 0.5035549457403019,
      "grad_norm": 0.0,
      "learning_rate": 1.0371604910661158e-05,
      "loss": 0.9376,
      "step": 4037
    },
    {
      "epoch": 0.5036796806785581,
      "grad_norm": 0.0,
      "learning_rate": 1.0367567556898723e-05,
      "loss": 0.9818,
      "step": 4038
    },
    {
      "epoch": 0.5038044156168142,
      "grad_norm": 0.0,
      "learning_rate": 1.036353014314e-05,
      "loss": 0.9486,
      "step": 4039
    },
    {
      "epoch": 0.5039291505550705,
      "grad_norm": 0.0,
      "learning_rate": 1.0359492670043988e-05,
      "loss": 0.9894,
      "step": 4040
    },
    {
      "epoch": 0.5040538854933266,
      "grad_norm": 0.0,
      "learning_rate": 1.035545513826971e-05,
      "loss": 0.9947,
      "step": 4041
    },
    {
      "epoch": 0.5041786204315829,
      "grad_norm": 0.0,
      "learning_rate": 1.0351417548476199e-05,
      "loss": 0.9606,
      "step": 4042
    },
    {
      "epoch": 0.5043033553698391,
      "grad_norm": 0.0,
      "learning_rate": 1.0347379901322478e-05,
      "loss": 0.9357,
      "step": 4043
    },
    {
      "epoch": 0.5044280903080953,
      "grad_norm": 0.0,
      "learning_rate": 1.0343342197467601e-05,
      "loss": 1.0328,
      "step": 4044
    },
    {
      "epoch": 0.5045528252463515,
      "grad_norm": 0.0,
      "learning_rate": 1.0339304437570626e-05,
      "loss": 0.9422,
      "step": 4045
    },
    {
      "epoch": 0.5046775601846077,
      "grad_norm": 0.0,
      "learning_rate": 1.0335266622290608e-05,
      "loss": 1.0152,
      "step": 4046
    },
    {
      "epoch": 0.5048022951228639,
      "grad_norm": 0.0,
      "learning_rate": 1.0331228752286624e-05,
      "loss": 0.9187,
      "step": 4047
    },
    {
      "epoch": 0.5049270300611202,
      "grad_norm": 0.0,
      "learning_rate": 1.0327190828217763e-05,
      "loss": 0.9345,
      "step": 4048
    },
    {
      "epoch": 0.5050517649993763,
      "grad_norm": 0.0,
      "learning_rate": 1.0323152850743107e-05,
      "loss": 0.9496,
      "step": 4049
    },
    {
      "epoch": 0.5051764999376326,
      "grad_norm": 0.0,
      "learning_rate": 1.031911482052176e-05,
      "loss": 0.8995,
      "step": 4050
    },
    {
      "epoch": 0.5053012348758887,
      "grad_norm": 0.0,
      "learning_rate": 1.0315076738212829e-05,
      "loss": 0.9326,
      "step": 4051
    },
    {
      "epoch": 0.5054259698141449,
      "grad_norm": 0.0,
      "learning_rate": 1.0311038604475431e-05,
      "loss": 0.9429,
      "step": 4052
    },
    {
      "epoch": 0.5055507047524012,
      "grad_norm": 0.0,
      "learning_rate": 1.0307000419968693e-05,
      "loss": 0.9597,
      "step": 4053
    },
    {
      "epoch": 0.5056754396906573,
      "grad_norm": 0.0,
      "learning_rate": 1.0302962185351746e-05,
      "loss": 0.9326,
      "step": 4054
    },
    {
      "epoch": 0.5058001746289136,
      "grad_norm": 0.0,
      "learning_rate": 1.0298923901283736e-05,
      "loss": 0.9667,
      "step": 4055
    },
    {
      "epoch": 0.5059249095671697,
      "grad_norm": 0.0,
      "learning_rate": 1.0294885568423805e-05,
      "loss": 1.0133,
      "step": 4056
    },
    {
      "epoch": 0.506049644505426,
      "grad_norm": 0.0,
      "learning_rate": 1.0290847187431115e-05,
      "loss": 0.942,
      "step": 4057
    },
    {
      "epoch": 0.5061743794436822,
      "grad_norm": 0.0,
      "learning_rate": 1.0286808758964834e-05,
      "loss": 0.9353,
      "step": 4058
    },
    {
      "epoch": 0.5062991143819384,
      "grad_norm": 0.0,
      "learning_rate": 1.0282770283684133e-05,
      "loss": 0.9309,
      "step": 4059
    },
    {
      "epoch": 0.5064238493201946,
      "grad_norm": 0.0,
      "learning_rate": 1.0278731762248194e-05,
      "loss": 0.9854,
      "step": 4060
    },
    {
      "epoch": 0.5065485842584508,
      "grad_norm": 0.0,
      "learning_rate": 1.0274693195316202e-05,
      "loss": 0.9602,
      "step": 4061
    },
    {
      "epoch": 0.506673319196707,
      "grad_norm": 0.0,
      "learning_rate": 1.0270654583547358e-05,
      "loss": 0.9669,
      "step": 4062
    },
    {
      "epoch": 0.5067980541349631,
      "grad_norm": 0.0,
      "learning_rate": 1.0266615927600858e-05,
      "loss": 0.9417,
      "step": 4063
    },
    {
      "epoch": 0.5069227890732194,
      "grad_norm": 0.0,
      "learning_rate": 1.0262577228135921e-05,
      "loss": 0.9351,
      "step": 4064
    },
    {
      "epoch": 0.5070475240114756,
      "grad_norm": 0.0,
      "learning_rate": 1.0258538485811765e-05,
      "loss": 0.9261,
      "step": 4065
    },
    {
      "epoch": 0.5071722589497318,
      "grad_norm": 0.0,
      "learning_rate": 1.0254499701287604e-05,
      "loss": 0.9323,
      "step": 4066
    },
    {
      "epoch": 0.507296993887988,
      "grad_norm": 0.0,
      "learning_rate": 1.0250460875222676e-05,
      "loss": 0.9645,
      "step": 4067
    },
    {
      "epoch": 0.5074217288262443,
      "grad_norm": 0.0,
      "learning_rate": 1.0246422008276223e-05,
      "loss": 0.9214,
      "step": 4068
    },
    {
      "epoch": 0.5075464637645004,
      "grad_norm": 0.0,
      "learning_rate": 1.0242383101107483e-05,
      "loss": 0.9917,
      "step": 4069
    },
    {
      "epoch": 0.5076711987027567,
      "grad_norm": 0.0,
      "learning_rate": 1.0238344154375712e-05,
      "loss": 0.9562,
      "step": 4070
    },
    {
      "epoch": 0.5077959336410128,
      "grad_norm": 0.0,
      "learning_rate": 1.0234305168740164e-05,
      "loss": 0.9778,
      "step": 4071
    },
    {
      "epoch": 0.5079206685792691,
      "grad_norm": 0.0,
      "learning_rate": 1.0230266144860108e-05,
      "loss": 0.9324,
      "step": 4072
    },
    {
      "epoch": 0.5080454035175253,
      "grad_norm": 0.0,
      "learning_rate": 1.0226227083394811e-05,
      "loss": 0.9689,
      "step": 4073
    },
    {
      "epoch": 0.5081701384557815,
      "grad_norm": 0.0,
      "learning_rate": 1.0222187985003552e-05,
      "loss": 0.9327,
      "step": 4074
    },
    {
      "epoch": 0.5082948733940377,
      "grad_norm": 0.0,
      "learning_rate": 1.0218148850345613e-05,
      "loss": 0.975,
      "step": 4075
    },
    {
      "epoch": 0.5084196083322938,
      "grad_norm": 0.0,
      "learning_rate": 1.0214109680080282e-05,
      "loss": 0.9818,
      "step": 4076
    },
    {
      "epoch": 0.5085443432705501,
      "grad_norm": 0.0,
      "learning_rate": 1.021007047486685e-05,
      "loss": 0.9176,
      "step": 4077
    },
    {
      "epoch": 0.5086690782088062,
      "grad_norm": 0.0,
      "learning_rate": 1.0206031235364627e-05,
      "loss": 0.9742,
      "step": 4078
    },
    {
      "epoch": 0.5087938131470625,
      "grad_norm": 0.0,
      "learning_rate": 1.0201991962232906e-05,
      "loss": 0.9557,
      "step": 4079
    },
    {
      "epoch": 0.5089185480853187,
      "grad_norm": 0.0,
      "learning_rate": 1.0197952656131008e-05,
      "loss": 0.9299,
      "step": 4080
    },
    {
      "epoch": 0.5090432830235749,
      "grad_norm": 0.0,
      "learning_rate": 1.0193913317718245e-05,
      "loss": 0.951,
      "step": 4081
    },
    {
      "epoch": 0.5091680179618311,
      "grad_norm": 0.0,
      "learning_rate": 1.0189873947653941e-05,
      "loss": 0.9084,
      "step": 4082
    },
    {
      "epoch": 0.5092927529000874,
      "grad_norm": 0.0,
      "learning_rate": 1.0185834546597421e-05,
      "loss": 0.9881,
      "step": 4083
    },
    {
      "epoch": 0.5094174878383435,
      "grad_norm": 0.0,
      "learning_rate": 1.0181795115208017e-05,
      "loss": 0.9773,
      "step": 4084
    },
    {
      "epoch": 0.5095422227765998,
      "grad_norm": 0.0,
      "learning_rate": 1.0177755654145071e-05,
      "loss": 0.9482,
      "step": 4085
    },
    {
      "epoch": 0.5096669577148559,
      "grad_norm": 0.0,
      "learning_rate": 1.0173716164067918e-05,
      "loss": 0.9301,
      "step": 4086
    },
    {
      "epoch": 0.5097916926531121,
      "grad_norm": 0.0,
      "learning_rate": 1.0169676645635908e-05,
      "loss": 0.9737,
      "step": 4087
    },
    {
      "epoch": 0.5099164275913683,
      "grad_norm": 0.0,
      "learning_rate": 1.0165637099508395e-05,
      "loss": 0.9956,
      "step": 4088
    },
    {
      "epoch": 0.5100411625296245,
      "grad_norm": 0.0,
      "learning_rate": 1.0161597526344728e-05,
      "loss": 0.9757,
      "step": 4089
    },
    {
      "epoch": 0.5101658974678808,
      "grad_norm": 0.0,
      "learning_rate": 1.015755792680427e-05,
      "loss": 0.9886,
      "step": 4090
    },
    {
      "epoch": 0.5102906324061369,
      "grad_norm": 0.0,
      "learning_rate": 1.015351830154639e-05,
      "loss": 1.0141,
      "step": 4091
    },
    {
      "epoch": 0.5104153673443932,
      "grad_norm": 0.0,
      "learning_rate": 1.0149478651230451e-05,
      "loss": 0.9088,
      "step": 4092
    },
    {
      "epoch": 0.5105401022826493,
      "grad_norm": 0.0,
      "learning_rate": 1.014543897651583e-05,
      "loss": 0.9907,
      "step": 4093
    },
    {
      "epoch": 0.5106648372209056,
      "grad_norm": 0.0,
      "learning_rate": 1.0141399278061897e-05,
      "loss": 1.0239,
      "step": 4094
    },
    {
      "epoch": 0.5107895721591618,
      "grad_norm": 0.0,
      "learning_rate": 1.0137359556528041e-05,
      "loss": 0.9018,
      "step": 4095
    },
    {
      "epoch": 0.510914307097418,
      "grad_norm": 0.0,
      "learning_rate": 1.0133319812573642e-05,
      "loss": 0.9534,
      "step": 4096
    },
    {
      "epoch": 0.5110390420356742,
      "grad_norm": 0.0,
      "learning_rate": 1.0129280046858085e-05,
      "loss": 0.9542,
      "step": 4097
    },
    {
      "epoch": 0.5111637769739305,
      "grad_norm": 0.0,
      "learning_rate": 1.0125240260040772e-05,
      "loss": 0.9765,
      "step": 4098
    },
    {
      "epoch": 0.5112885119121866,
      "grad_norm": 0.0,
      "learning_rate": 1.0121200452781086e-05,
      "loss": 0.9776,
      "step": 4099
    },
    {
      "epoch": 0.5114132468504428,
      "grad_norm": 0.0,
      "learning_rate": 1.0117160625738432e-05,
      "loss": 0.9888,
      "step": 4100
    },
    {
      "epoch": 0.511537981788699,
      "grad_norm": 0.0,
      "learning_rate": 1.011312077957221e-05,
      "loss": 0.9556,
      "step": 4101
    },
    {
      "epoch": 0.5116627167269552,
      "grad_norm": 0.0,
      "learning_rate": 1.0109080914941825e-05,
      "loss": 0.9636,
      "step": 4102
    },
    {
      "epoch": 0.5117874516652114,
      "grad_norm": 0.0,
      "learning_rate": 1.0105041032506685e-05,
      "loss": 0.9788,
      "step": 4103
    },
    {
      "epoch": 0.5119121866034676,
      "grad_norm": 0.0,
      "learning_rate": 1.01010011329262e-05,
      "loss": 0.923,
      "step": 4104
    },
    {
      "epoch": 0.5120369215417239,
      "grad_norm": 0.0,
      "learning_rate": 1.0096961216859786e-05,
      "loss": 0.9314,
      "step": 4105
    },
    {
      "epoch": 0.51216165647998,
      "grad_norm": 0.0,
      "learning_rate": 1.0092921284966854e-05,
      "loss": 0.9461,
      "step": 4106
    },
    {
      "epoch": 0.5122863914182363,
      "grad_norm": 0.0,
      "learning_rate": 1.0088881337906825e-05,
      "loss": 0.9782,
      "step": 4107
    },
    {
      "epoch": 0.5124111263564924,
      "grad_norm": 0.0,
      "learning_rate": 1.0084841376339128e-05,
      "loss": 0.94,
      "step": 4108
    },
    {
      "epoch": 0.5125358612947487,
      "grad_norm": 0.0,
      "learning_rate": 1.0080801400923172e-05,
      "loss": 0.9533,
      "step": 4109
    },
    {
      "epoch": 0.5126605962330049,
      "grad_norm": 0.0,
      "learning_rate": 1.0076761412318395e-05,
      "loss": 0.9891,
      "step": 4110
    },
    {
      "epoch": 0.512785331171261,
      "grad_norm": 0.0,
      "learning_rate": 1.007272141118422e-05,
      "loss": 0.9464,
      "step": 4111
    },
    {
      "epoch": 0.5129100661095173,
      "grad_norm": 0.0,
      "learning_rate": 1.0068681398180074e-05,
      "loss": 0.948,
      "step": 4112
    },
    {
      "epoch": 0.5130348010477734,
      "grad_norm": 0.0,
      "learning_rate": 1.0064641373965394e-05,
      "loss": 0.9286,
      "step": 4113
    },
    {
      "epoch": 0.5131595359860297,
      "grad_norm": 0.0,
      "learning_rate": 1.0060601339199613e-05,
      "loss": 0.9929,
      "step": 4114
    },
    {
      "epoch": 0.5132842709242859,
      "grad_norm": 0.0,
      "learning_rate": 1.0056561294542166e-05,
      "loss": 0.9634,
      "step": 4115
    },
    {
      "epoch": 0.5134090058625421,
      "grad_norm": 0.0,
      "learning_rate": 1.0052521240652489e-05,
      "loss": 0.9715,
      "step": 4116
    },
    {
      "epoch": 0.5135337408007983,
      "grad_norm": 0.0,
      "learning_rate": 1.004848117819002e-05,
      "loss": 0.9341,
      "step": 4117
    },
    {
      "epoch": 0.5136584757390545,
      "grad_norm": 0.0,
      "learning_rate": 1.0044441107814203e-05,
      "loss": 0.986,
      "step": 4118
    },
    {
      "epoch": 0.5137832106773107,
      "grad_norm": 0.0,
      "learning_rate": 1.0040401030184474e-05,
      "loss": 0.9272,
      "step": 4119
    },
    {
      "epoch": 0.513907945615567,
      "grad_norm": 0.0,
      "learning_rate": 1.003636094596028e-05,
      "loss": 0.9437,
      "step": 4120
    },
    {
      "epoch": 0.5140326805538231,
      "grad_norm": 0.0,
      "learning_rate": 1.0032320855801063e-05,
      "loss": 0.9101,
      "step": 4121
    },
    {
      "epoch": 0.5141574154920794,
      "grad_norm": 0.0,
      "learning_rate": 1.002828076036627e-05,
      "loss": 0.9389,
      "step": 4122
    },
    {
      "epoch": 0.5142821504303355,
      "grad_norm": 0.0,
      "learning_rate": 1.0024240660315341e-05,
      "loss": 0.9317,
      "step": 4123
    },
    {
      "epoch": 0.5144068853685917,
      "grad_norm": 0.0,
      "learning_rate": 1.0020200556307727e-05,
      "loss": 0.938,
      "step": 4124
    },
    {
      "epoch": 0.514531620306848,
      "grad_norm": 0.0,
      "learning_rate": 1.001616044900287e-05,
      "loss": 0.9914,
      "step": 4125
    },
    {
      "epoch": 0.5146563552451041,
      "grad_norm": 0.0,
      "learning_rate": 1.0012120339060221e-05,
      "loss": 1.0269,
      "step": 4126
    },
    {
      "epoch": 0.5147810901833604,
      "grad_norm": 0.0,
      "learning_rate": 1.0008080227139229e-05,
      "loss": 0.9137,
      "step": 4127
    },
    {
      "epoch": 0.5149058251216165,
      "grad_norm": 0.0,
      "learning_rate": 1.000404011389934e-05,
      "loss": 0.9118,
      "step": 4128
    },
    {
      "epoch": 0.5150305600598728,
      "grad_norm": 0.0,
      "learning_rate": 1e-05,
      "loss": 0.9367,
      "step": 4129
    },
    {
      "epoch": 0.515155294998129,
      "grad_norm": 0.0,
      "learning_rate": 9.995959886100665e-06,
      "loss": 0.9434,
      "step": 4130
    },
    {
      "epoch": 0.5152800299363852,
      "grad_norm": 0.0,
      "learning_rate": 9.991919772860775e-06,
      "loss": 0.9181,
      "step": 4131
    },
    {
      "epoch": 0.5154047648746414,
      "grad_norm": 0.0,
      "learning_rate": 9.987879660939782e-06,
      "loss": 0.9477,
      "step": 4132
    },
    {
      "epoch": 0.5155294998128976,
      "grad_norm": 0.0,
      "learning_rate": 9.98383955099713e-06,
      "loss": 0.9584,
      "step": 4133
    },
    {
      "epoch": 0.5156542347511538,
      "grad_norm": 0.0,
      "learning_rate": 9.979799443692276e-06,
      "loss": 0.9012,
      "step": 4134
    },
    {
      "epoch": 0.51577896968941,
      "grad_norm": 0.0,
      "learning_rate": 9.975759339684665e-06,
      "loss": 0.9617,
      "step": 4135
    },
    {
      "epoch": 0.5159037046276662,
      "grad_norm": 0.0,
      "learning_rate": 9.971719239633734e-06,
      "loss": 0.9176,
      "step": 4136
    },
    {
      "epoch": 0.5160284395659224,
      "grad_norm": 0.0,
      "learning_rate": 9.96767914419894e-06,
      "loss": 0.9773,
      "step": 4137
    },
    {
      "epoch": 0.5161531745041786,
      "grad_norm": 0.0,
      "learning_rate": 9.963639054039722e-06,
      "loss": 1.006,
      "step": 4138
    },
    {
      "epoch": 0.5162779094424348,
      "grad_norm": 0.0,
      "learning_rate": 9.959598969815528e-06,
      "loss": 0.8844,
      "step": 4139
    },
    {
      "epoch": 0.5164026443806911,
      "grad_norm": 0.0,
      "learning_rate": 9.9555588921858e-06,
      "loss": 0.9162,
      "step": 4140
    },
    {
      "epoch": 0.5165273793189472,
      "grad_norm": 0.0,
      "learning_rate": 9.951518821809984e-06,
      "loss": 0.9851,
      "step": 4141
    },
    {
      "epoch": 0.5166521142572035,
      "grad_norm": 0.0,
      "learning_rate": 9.947478759347518e-06,
      "loss": 0.9109,
      "step": 4142
    },
    {
      "epoch": 0.5167768491954596,
      "grad_norm": 0.0,
      "learning_rate": 9.943438705457837e-06,
      "loss": 0.9232,
      "step": 4143
    },
    {
      "epoch": 0.5169015841337159,
      "grad_norm": 0.0,
      "learning_rate": 9.93939866080039e-06,
      "loss": 0.9559,
      "step": 4144
    },
    {
      "epoch": 0.517026319071972,
      "grad_norm": 0.0,
      "learning_rate": 9.935358626034607e-06,
      "loss": 0.9676,
      "step": 4145
    },
    {
      "epoch": 0.5171510540102283,
      "grad_norm": 0.0,
      "learning_rate": 9.93131860181993e-06,
      "loss": 0.8908,
      "step": 4146
    },
    {
      "epoch": 0.5172757889484845,
      "grad_norm": 0.0,
      "learning_rate": 9.927278588815786e-06,
      "loss": 0.9467,
      "step": 4147
    },
    {
      "epoch": 0.5174005238867406,
      "grad_norm": 0.0,
      "learning_rate": 9.923238587681609e-06,
      "loss": 0.9621,
      "step": 4148
    },
    {
      "epoch": 0.5175252588249969,
      "grad_norm": 0.0,
      "learning_rate": 9.91919859907683e-06,
      "loss": 0.9231,
      "step": 4149
    },
    {
      "epoch": 0.517649993763253,
      "grad_norm": 0.0,
      "learning_rate": 9.915158623660877e-06,
      "loss": 0.9393,
      "step": 4150
    },
    {
      "epoch": 0.5177747287015093,
      "grad_norm": 0.0,
      "learning_rate": 9.911118662093176e-06,
      "loss": 0.9805,
      "step": 4151
    },
    {
      "epoch": 0.5178994636397655,
      "grad_norm": 0.0,
      "learning_rate": 9.90707871503315e-06,
      "loss": 0.9437,
      "step": 4152
    },
    {
      "epoch": 0.5180241985780217,
      "grad_norm": 0.0,
      "learning_rate": 9.903038783140216e-06,
      "loss": 0.9659,
      "step": 4153
    },
    {
      "epoch": 0.5181489335162779,
      "grad_norm": 0.0,
      "learning_rate": 9.898998867073802e-06,
      "loss": 0.9528,
      "step": 4154
    },
    {
      "epoch": 0.5182736684545342,
      "grad_norm": 0.0,
      "learning_rate": 9.894958967493317e-06,
      "loss": 0.9584,
      "step": 4155
    },
    {
      "epoch": 0.5183984033927903,
      "grad_norm": 0.0,
      "learning_rate": 9.890919085058179e-06,
      "loss": 1.0016,
      "step": 4156
    },
    {
      "epoch": 0.5185231383310466,
      "grad_norm": 0.0,
      "learning_rate": 9.886879220427791e-06,
      "loss": 0.9524,
      "step": 4157
    },
    {
      "epoch": 0.5186478732693027,
      "grad_norm": 0.0,
      "learning_rate": 9.882839374261572e-06,
      "loss": 0.9842,
      "step": 4158
    },
    {
      "epoch": 0.5187726082075589,
      "grad_norm": 0.0,
      "learning_rate": 9.878799547218919e-06,
      "loss": 0.9752,
      "step": 4159
    },
    {
      "epoch": 0.5188973431458151,
      "grad_norm": 0.0,
      "learning_rate": 9.874759739959231e-06,
      "loss": 0.9757,
      "step": 4160
    },
    {
      "epoch": 0.5190220780840713,
      "grad_norm": 0.0,
      "learning_rate": 9.870719953141917e-06,
      "loss": 1.0116,
      "step": 4161
    },
    {
      "epoch": 0.5191468130223276,
      "grad_norm": 0.0,
      "learning_rate": 9.86668018742636e-06,
      "loss": 0.9933,
      "step": 4162
    },
    {
      "epoch": 0.5192715479605837,
      "grad_norm": 0.0,
      "learning_rate": 9.862640443471962e-06,
      "loss": 0.9573,
      "step": 4163
    },
    {
      "epoch": 0.51939628289884,
      "grad_norm": 0.0,
      "learning_rate": 9.858600721938105e-06,
      "loss": 0.9645,
      "step": 4164
    },
    {
      "epoch": 0.5195210178370961,
      "grad_norm": 0.0,
      "learning_rate": 9.854561023484174e-06,
      "loss": 0.9069,
      "step": 4165
    },
    {
      "epoch": 0.5196457527753524,
      "grad_norm": 0.0,
      "learning_rate": 9.850521348769552e-06,
      "loss": 0.9161,
      "step": 4166
    },
    {
      "epoch": 0.5197704877136086,
      "grad_norm": 0.0,
      "learning_rate": 9.846481698453613e-06,
      "loss": 0.8889,
      "step": 4167
    },
    {
      "epoch": 0.5198952226518648,
      "grad_norm": 0.0,
      "learning_rate": 9.842442073195733e-06,
      "loss": 0.9003,
      "step": 4168
    },
    {
      "epoch": 0.520019957590121,
      "grad_norm": 0.0,
      "learning_rate": 9.838402473655277e-06,
      "loss": 0.9444,
      "step": 4169
    },
    {
      "epoch": 0.5201446925283773,
      "grad_norm": 0.0,
      "learning_rate": 9.834362900491609e-06,
      "loss": 0.9693,
      "step": 4170
    },
    {
      "epoch": 0.5202694274666334,
      "grad_norm": 0.0,
      "learning_rate": 9.830323354364097e-06,
      "loss": 0.9667,
      "step": 4171
    },
    {
      "epoch": 0.5203941624048896,
      "grad_norm": 0.0,
      "learning_rate": 9.826283835932084e-06,
      "loss": 0.9453,
      "step": 4172
    },
    {
      "epoch": 0.5205188973431458,
      "grad_norm": 0.0,
      "learning_rate": 9.822244345854932e-06,
      "loss": 0.9356,
      "step": 4173
    },
    {
      "epoch": 0.520643632281402,
      "grad_norm": 0.0,
      "learning_rate": 9.818204884791983e-06,
      "loss": 0.9728,
      "step": 4174
    },
    {
      "epoch": 0.5207683672196582,
      "grad_norm": 0.0,
      "learning_rate": 9.814165453402582e-06,
      "loss": 0.9554,
      "step": 4175
    },
    {
      "epoch": 0.5208931021579144,
      "grad_norm": 0.0,
      "learning_rate": 9.810126052346062e-06,
      "loss": 0.9124,
      "step": 4176
    },
    {
      "epoch": 0.5210178370961707,
      "grad_norm": 0.0,
      "learning_rate": 9.806086682281759e-06,
      "loss": 0.9654,
      "step": 4177
    },
    {
      "epoch": 0.5211425720344268,
      "grad_norm": 0.0,
      "learning_rate": 9.802047343868997e-06,
      "loss": 0.9393,
      "step": 4178
    },
    {
      "epoch": 0.5212673069726831,
      "grad_norm": 0.0,
      "learning_rate": 9.798008037767094e-06,
      "loss": 0.9665,
      "step": 4179
    },
    {
      "epoch": 0.5213920419109392,
      "grad_norm": 0.0,
      "learning_rate": 9.793968764635378e-06,
      "loss": 0.9501,
      "step": 4180
    },
    {
      "epoch": 0.5215167768491955,
      "grad_norm": 0.0,
      "learning_rate": 9.789929525133153e-06,
      "loss": 0.9615,
      "step": 4181
    },
    {
      "epoch": 0.5216415117874517,
      "grad_norm": 0.0,
      "learning_rate": 9.785890319919721e-06,
      "loss": 0.9723,
      "step": 4182
    },
    {
      "epoch": 0.5217662467257078,
      "grad_norm": 0.0,
      "learning_rate": 9.78185114965439e-06,
      "loss": 0.9455,
      "step": 4183
    },
    {
      "epoch": 0.5218909816639641,
      "grad_norm": 0.0,
      "learning_rate": 9.77781201499645e-06,
      "loss": 0.9198,
      "step": 4184
    },
    {
      "epoch": 0.5220157166022202,
      "grad_norm": 0.0,
      "learning_rate": 9.77377291660519e-06,
      "loss": 0.948,
      "step": 4185
    },
    {
      "epoch": 0.5221404515404765,
      "grad_norm": 0.0,
      "learning_rate": 9.769733855139895e-06,
      "loss": 0.9923,
      "step": 4186
    },
    {
      "epoch": 0.5222651864787327,
      "grad_norm": 0.0,
      "learning_rate": 9.765694831259839e-06,
      "loss": 0.9496,
      "step": 4187
    },
    {
      "epoch": 0.5223899214169889,
      "grad_norm": 0.0,
      "learning_rate": 9.761655845624293e-06,
      "loss": 0.9239,
      "step": 4188
    },
    {
      "epoch": 0.5225146563552451,
      "grad_norm": 0.0,
      "learning_rate": 9.757616898892517e-06,
      "loss": 0.9177,
      "step": 4189
    },
    {
      "epoch": 0.5226393912935013,
      "grad_norm": 0.0,
      "learning_rate": 9.75357799172378e-06,
      "loss": 0.9767,
      "step": 4190
    },
    {
      "epoch": 0.5227641262317575,
      "grad_norm": 0.0,
      "learning_rate": 9.749539124777323e-06,
      "loss": 0.959,
      "step": 4191
    },
    {
      "epoch": 0.5228888611700138,
      "grad_norm": 0.0,
      "learning_rate": 9.7455002987124e-06,
      "loss": 0.9712,
      "step": 4192
    },
    {
      "epoch": 0.5230135961082699,
      "grad_norm": 0.0,
      "learning_rate": 9.741461514188242e-06,
      "loss": 0.9816,
      "step": 4193
    },
    {
      "epoch": 0.5231383310465262,
      "grad_norm": 0.0,
      "learning_rate": 9.73742277186408e-06,
      "loss": 0.9554,
      "step": 4194
    },
    {
      "epoch": 0.5232630659847823,
      "grad_norm": 0.0,
      "learning_rate": 9.733384072399145e-06,
      "loss": 0.9452,
      "step": 4195
    },
    {
      "epoch": 0.5233878009230385,
      "grad_norm": 0.0,
      "learning_rate": 9.729345416452644e-06,
      "loss": 0.9395,
      "step": 4196
    },
    {
      "epoch": 0.5235125358612948,
      "grad_norm": 0.0,
      "learning_rate": 9.725306804683803e-06,
      "loss": 0.9764,
      "step": 4197
    },
    {
      "epoch": 0.5236372707995509,
      "grad_norm": 0.0,
      "learning_rate": 9.721268237751813e-06,
      "loss": 0.9182,
      "step": 4198
    },
    {
      "epoch": 0.5237620057378072,
      "grad_norm": 0.0,
      "learning_rate": 9.717229716315868e-06,
      "loss": 0.9716,
      "step": 4199
    },
    {
      "epoch": 0.5238867406760633,
      "grad_norm": 0.0,
      "learning_rate": 9.713191241035169e-06,
      "loss": 1.0092,
      "step": 4200
    },
    {
      "epoch": 0.5240114756143196,
      "grad_norm": 0.0,
      "learning_rate": 9.709152812568886e-06,
      "loss": 0.9284,
      "step": 4201
    },
    {
      "epoch": 0.5241362105525758,
      "grad_norm": 0.0,
      "learning_rate": 9.705114431576198e-06,
      "loss": 0.9245,
      "step": 4202
    },
    {
      "epoch": 0.524260945490832,
      "grad_norm": 0.0,
      "learning_rate": 9.70107609871627e-06,
      "loss": 0.9536,
      "step": 4203
    },
    {
      "epoch": 0.5243856804290882,
      "grad_norm": 0.0,
      "learning_rate": 9.697037814648255e-06,
      "loss": 0.9801,
      "step": 4204
    },
    {
      "epoch": 0.5245104153673444,
      "grad_norm": 0.0,
      "learning_rate": 9.69299958003131e-06,
      "loss": 0.9647,
      "step": 4205
    },
    {
      "epoch": 0.5246351503056006,
      "grad_norm": 0.0,
      "learning_rate": 9.688961395524569e-06,
      "loss": 0.9629,
      "step": 4206
    },
    {
      "epoch": 0.5247598852438567,
      "grad_norm": 0.0,
      "learning_rate": 9.684923261787175e-06,
      "loss": 0.9328,
      "step": 4207
    },
    {
      "epoch": 0.524884620182113,
      "grad_norm": 0.0,
      "learning_rate": 9.68088517947824e-06,
      "loss": 0.9976,
      "step": 4208
    },
    {
      "epoch": 0.5250093551203692,
      "grad_norm": 0.0,
      "learning_rate": 9.676847149256894e-06,
      "loss": 0.9986,
      "step": 4209
    },
    {
      "epoch": 0.5251340900586254,
      "grad_norm": 0.0,
      "learning_rate": 9.67280917178224e-06,
      "loss": 0.9834,
      "step": 4210
    },
    {
      "epoch": 0.5252588249968816,
      "grad_norm": 0.0,
      "learning_rate": 9.668771247713376e-06,
      "loss": 0.959,
      "step": 4211
    },
    {
      "epoch": 0.5253835599351379,
      "grad_norm": 0.0,
      "learning_rate": 9.664733377709395e-06,
      "loss": 0.9622,
      "step": 4212
    },
    {
      "epoch": 0.525508294873394,
      "grad_norm": 0.0,
      "learning_rate": 9.660695562429379e-06,
      "loss": 0.9562,
      "step": 4213
    },
    {
      "epoch": 0.5256330298116503,
      "grad_norm": 0.0,
      "learning_rate": 9.6566578025324e-06,
      "loss": 0.9359,
      "step": 4214
    },
    {
      "epoch": 0.5257577647499064,
      "grad_norm": 0.0,
      "learning_rate": 9.652620098677527e-06,
      "loss": 0.8896,
      "step": 4215
    },
    {
      "epoch": 0.5258824996881627,
      "grad_norm": 0.0,
      "learning_rate": 9.648582451523805e-06,
      "loss": 1.0019,
      "step": 4216
    },
    {
      "epoch": 0.5260072346264189,
      "grad_norm": 0.0,
      "learning_rate": 9.644544861730293e-06,
      "loss": 0.984,
      "step": 4217
    },
    {
      "epoch": 0.5261319695646751,
      "grad_norm": 0.0,
      "learning_rate": 9.640507329956012e-06,
      "loss": 0.9314,
      "step": 4218
    },
    {
      "epoch": 0.5262567045029313,
      "grad_norm": 0.0,
      "learning_rate": 9.636469856860005e-06,
      "loss": 0.9177,
      "step": 4219
    },
    {
      "epoch": 0.5263814394411874,
      "grad_norm": 0.0,
      "learning_rate": 9.63243244310128e-06,
      "loss": 0.9817,
      "step": 4220
    },
    {
      "epoch": 0.5265061743794437,
      "grad_norm": 0.0,
      "learning_rate": 9.628395089338844e-06,
      "loss": 0.9281,
      "step": 4221
    },
    {
      "epoch": 0.5266309093176998,
      "grad_norm": 0.0,
      "learning_rate": 9.624357796231702e-06,
      "loss": 0.9234,
      "step": 4222
    },
    {
      "epoch": 0.5267556442559561,
      "grad_norm": 0.0,
      "learning_rate": 9.620320564438834e-06,
      "loss": 0.9901,
      "step": 4223
    },
    {
      "epoch": 0.5268803791942123,
      "grad_norm": 0.0,
      "learning_rate": 9.616283394619223e-06,
      "loss": 0.9427,
      "step": 4224
    },
    {
      "epoch": 0.5270051141324685,
      "grad_norm": 0.0,
      "learning_rate": 9.612246287431832e-06,
      "loss": 0.9371,
      "step": 4225
    },
    {
      "epoch": 0.5271298490707247,
      "grad_norm": 0.0,
      "learning_rate": 9.608209243535627e-06,
      "loss": 0.9661,
      "step": 4226
    },
    {
      "epoch": 0.527254584008981,
      "grad_norm": 0.0,
      "learning_rate": 9.60417226358955e-06,
      "loss": 0.9249,
      "step": 4227
    },
    {
      "epoch": 0.5273793189472371,
      "grad_norm": 0.0,
      "learning_rate": 9.600135348252535e-06,
      "loss": 0.9566,
      "step": 4228
    },
    {
      "epoch": 0.5275040538854934,
      "grad_norm": 0.0,
      "learning_rate": 9.596098498183515e-06,
      "loss": 0.9437,
      "step": 4229
    },
    {
      "epoch": 0.5276287888237495,
      "grad_norm": 0.0,
      "learning_rate": 9.5920617140414e-06,
      "loss": 0.913,
      "step": 4230
    },
    {
      "epoch": 0.5277535237620057,
      "grad_norm": 0.0,
      "learning_rate": 9.588024996485098e-06,
      "loss": 0.9389,
      "step": 4231
    },
    {
      "epoch": 0.527878258700262,
      "grad_norm": 0.0,
      "learning_rate": 9.583988346173504e-06,
      "loss": 1.0094,
      "step": 4232
    },
    {
      "epoch": 0.5280029936385181,
      "grad_norm": 0.0,
      "learning_rate": 9.579951763765498e-06,
      "loss": 0.9672,
      "step": 4233
    },
    {
      "epoch": 0.5281277285767744,
      "grad_norm": 0.0,
      "learning_rate": 9.575915249919955e-06,
      "loss": 0.9574,
      "step": 4234
    },
    {
      "epoch": 0.5282524635150305,
      "grad_norm": 0.0,
      "learning_rate": 9.57187880529573e-06,
      "loss": 0.905,
      "step": 4235
    },
    {
      "epoch": 0.5283771984532868,
      "grad_norm": 0.0,
      "learning_rate": 9.56784243055168e-06,
      "loss": 0.9948,
      "step": 4236
    },
    {
      "epoch": 0.5285019333915429,
      "grad_norm": 0.0,
      "learning_rate": 9.563806126346643e-06,
      "loss": 0.9607,
      "step": 4237
    },
    {
      "epoch": 0.5286266683297992,
      "grad_norm": 0.0,
      "learning_rate": 9.559769893339437e-06,
      "loss": 0.9477,
      "step": 4238
    },
    {
      "epoch": 0.5287514032680554,
      "grad_norm": 0.0,
      "learning_rate": 9.555733732188887e-06,
      "loss": 0.9118,
      "step": 4239
    },
    {
      "epoch": 0.5288761382063116,
      "grad_norm": 0.0,
      "learning_rate": 9.551697643553787e-06,
      "loss": 0.9585,
      "step": 4240
    },
    {
      "epoch": 0.5290008731445678,
      "grad_norm": 0.0,
      "learning_rate": 9.547661628092938e-06,
      "loss": 0.9267,
      "step": 4241
    },
    {
      "epoch": 0.529125608082824,
      "grad_norm": 0.0,
      "learning_rate": 9.54362568646511e-06,
      "loss": 0.9643,
      "step": 4242
    },
    {
      "epoch": 0.5292503430210802,
      "grad_norm": 0.0,
      "learning_rate": 9.53958981932908e-06,
      "loss": 0.9768,
      "step": 4243
    },
    {
      "epoch": 0.5293750779593364,
      "grad_norm": 0.0,
      "learning_rate": 9.535554027343599e-06,
      "loss": 0.9475,
      "step": 4244
    },
    {
      "epoch": 0.5294998128975926,
      "grad_norm": 0.0,
      "learning_rate": 9.531518311167402e-06,
      "loss": 0.9265,
      "step": 4245
    },
    {
      "epoch": 0.5296245478358488,
      "grad_norm": 0.0,
      "learning_rate": 9.527482671459233e-06,
      "loss": 0.974,
      "step": 4246
    },
    {
      "epoch": 0.529749282774105,
      "grad_norm": 0.0,
      "learning_rate": 9.523447108877801e-06,
      "loss": 0.9532,
      "step": 4247
    },
    {
      "epoch": 0.5298740177123612,
      "grad_norm": 0.0,
      "learning_rate": 9.519411624081818e-06,
      "loss": 0.9601,
      "step": 4248
    },
    {
      "epoch": 0.5299987526506175,
      "grad_norm": 0.0,
      "learning_rate": 9.515376217729973e-06,
      "loss": 0.9278,
      "step": 4249
    },
    {
      "epoch": 0.5301234875888736,
      "grad_norm": 0.0,
      "learning_rate": 9.511340890480944e-06,
      "loss": 0.923,
      "step": 4250
    },
    {
      "epoch": 0.5302482225271299,
      "grad_norm": 0.0,
      "learning_rate": 9.507305642993403e-06,
      "loss": 0.9708,
      "step": 4251
    },
    {
      "epoch": 0.530372957465386,
      "grad_norm": 0.0,
      "learning_rate": 9.503270475925997e-06,
      "loss": 0.9271,
      "step": 4252
    },
    {
      "epoch": 0.5304976924036423,
      "grad_norm": 0.0,
      "learning_rate": 9.499235389937378e-06,
      "loss": 0.9529,
      "step": 4253
    },
    {
      "epoch": 0.5306224273418985,
      "grad_norm": 0.0,
      "learning_rate": 9.495200385686168e-06,
      "loss": 0.9676,
      "step": 4254
    },
    {
      "epoch": 0.5307471622801546,
      "grad_norm": 0.0,
      "learning_rate": 9.491165463830975e-06,
      "loss": 0.9228,
      "step": 4255
    },
    {
      "epoch": 0.5308718972184109,
      "grad_norm": 0.0,
      "learning_rate": 9.48713062503041e-06,
      "loss": 0.9553,
      "step": 4256
    },
    {
      "epoch": 0.530996632156667,
      "grad_norm": 0.0,
      "learning_rate": 9.483095869943056e-06,
      "loss": 0.9783,
      "step": 4257
    },
    {
      "epoch": 0.5311213670949233,
      "grad_norm": 0.0,
      "learning_rate": 9.479061199227486e-06,
      "loss": 0.9528,
      "step": 4258
    },
    {
      "epoch": 0.5312461020331795,
      "grad_norm": 0.0,
      "learning_rate": 9.475026613542264e-06,
      "loss": 0.9341,
      "step": 4259
    },
    {
      "epoch": 0.5313708369714357,
      "grad_norm": 0.0,
      "learning_rate": 9.47099211354593e-06,
      "loss": 0.9928,
      "step": 4260
    },
    {
      "epoch": 0.5314955719096919,
      "grad_norm": 0.0,
      "learning_rate": 9.46695769989702e-06,
      "loss": 0.947,
      "step": 4261
    },
    {
      "epoch": 0.5316203068479481,
      "grad_norm": 0.0,
      "learning_rate": 9.462923373254049e-06,
      "loss": 1.0086,
      "step": 4262
    },
    {
      "epoch": 0.5317450417862043,
      "grad_norm": 0.0,
      "learning_rate": 9.458889134275527e-06,
      "loss": 0.9385,
      "step": 4263
    },
    {
      "epoch": 0.5318697767244606,
      "grad_norm": 0.0,
      "learning_rate": 9.454854983619936e-06,
      "loss": 0.9242,
      "step": 4264
    },
    {
      "epoch": 0.5319945116627167,
      "grad_norm": 0.0,
      "learning_rate": 9.450820921945758e-06,
      "loss": 0.9537,
      "step": 4265
    },
    {
      "epoch": 0.532119246600973,
      "grad_norm": 0.0,
      "learning_rate": 9.44678694991145e-06,
      "loss": 0.948,
      "step": 4266
    },
    {
      "epoch": 0.5322439815392291,
      "grad_norm": 0.0,
      "learning_rate": 9.442753068175458e-06,
      "loss": 0.9236,
      "step": 4267
    },
    {
      "epoch": 0.5323687164774853,
      "grad_norm": 0.0,
      "learning_rate": 9.438719277396213e-06,
      "loss": 0.9757,
      "step": 4268
    },
    {
      "epoch": 0.5324934514157416,
      "grad_norm": 0.0,
      "learning_rate": 9.43468557823213e-06,
      "loss": 0.9265,
      "step": 4269
    },
    {
      "epoch": 0.5326181863539977,
      "grad_norm": 0.0,
      "learning_rate": 9.430651971341616e-06,
      "loss": 0.9465,
      "step": 4270
    },
    {
      "epoch": 0.532742921292254,
      "grad_norm": 0.0,
      "learning_rate": 9.426618457383051e-06,
      "loss": 0.9057,
      "step": 4271
    },
    {
      "epoch": 0.5328676562305101,
      "grad_norm": 0.0,
      "learning_rate": 9.422585037014805e-06,
      "loss": 0.9399,
      "step": 4272
    },
    {
      "epoch": 0.5329923911687664,
      "grad_norm": 0.0,
      "learning_rate": 9.418551710895243e-06,
      "loss": 1.0215,
      "step": 4273
    },
    {
      "epoch": 0.5331171261070226,
      "grad_norm": 0.0,
      "learning_rate": 9.414518479682697e-06,
      "loss": 1.0108,
      "step": 4274
    },
    {
      "epoch": 0.5332418610452788,
      "grad_norm": 0.0,
      "learning_rate": 9.410485344035498e-06,
      "loss": 0.9492,
      "step": 4275
    },
    {
      "epoch": 0.533366595983535,
      "grad_norm": 0.0,
      "learning_rate": 9.406452304611951e-06,
      "loss": 0.9306,
      "step": 4276
    },
    {
      "epoch": 0.5334913309217912,
      "grad_norm": 0.0,
      "learning_rate": 9.40241936207035e-06,
      "loss": 0.9009,
      "step": 4277
    },
    {
      "epoch": 0.5336160658600474,
      "grad_norm": 0.0,
      "learning_rate": 9.398386517068975e-06,
      "loss": 0.966,
      "step": 4278
    },
    {
      "epoch": 0.5337408007983035,
      "grad_norm": 0.0,
      "learning_rate": 9.394353770266084e-06,
      "loss": 0.9885,
      "step": 4279
    },
    {
      "epoch": 0.5338655357365598,
      "grad_norm": 0.0,
      "learning_rate": 9.390321122319933e-06,
      "loss": 0.9444,
      "step": 4280
    },
    {
      "epoch": 0.533990270674816,
      "grad_norm": 0.0,
      "learning_rate": 9.386288573888736e-06,
      "loss": 0.9523,
      "step": 4281
    },
    {
      "epoch": 0.5341150056130722,
      "grad_norm": 0.0,
      "learning_rate": 9.382256125630722e-06,
      "loss": 0.9578,
      "step": 4282
    },
    {
      "epoch": 0.5342397405513284,
      "grad_norm": 0.0,
      "learning_rate": 9.37822377820408e-06,
      "loss": 0.9626,
      "step": 4283
    },
    {
      "epoch": 0.5343644754895847,
      "grad_norm": 0.0,
      "learning_rate": 9.374191532266991e-06,
      "loss": 0.9524,
      "step": 4284
    },
    {
      "epoch": 0.5344892104278408,
      "grad_norm": 0.0,
      "learning_rate": 9.370159388477623e-06,
      "loss": 0.9489,
      "step": 4285
    },
    {
      "epoch": 0.5346139453660971,
      "grad_norm": 0.0,
      "learning_rate": 9.366127347494117e-06,
      "loss": 0.9202,
      "step": 4286
    },
    {
      "epoch": 0.5347386803043532,
      "grad_norm": 0.0,
      "learning_rate": 9.362095409974612e-06,
      "loss": 0.9252,
      "step": 4287
    },
    {
      "epoch": 0.5348634152426095,
      "grad_norm": 0.0,
      "learning_rate": 9.358063576577216e-06,
      "loss": 0.9499,
      "step": 4288
    },
    {
      "epoch": 0.5349881501808657,
      "grad_norm": 0.0,
      "learning_rate": 9.354031847960023e-06,
      "loss": 0.9346,
      "step": 4289
    },
    {
      "epoch": 0.5351128851191219,
      "grad_norm": 0.0,
      "learning_rate": 9.350000224781125e-06,
      "loss": 0.9101,
      "step": 4290
    },
    {
      "epoch": 0.5352376200573781,
      "grad_norm": 0.0,
      "learning_rate": 9.34596870769857e-06,
      "loss": 0.9602,
      "step": 4291
    },
    {
      "epoch": 0.5353623549956342,
      "grad_norm": 0.0,
      "learning_rate": 9.341937297370414e-06,
      "loss": 0.9028,
      "step": 4292
    },
    {
      "epoch": 0.5354870899338905,
      "grad_norm": 0.0,
      "learning_rate": 9.33790599445468e-06,
      "loss": 0.9448,
      "step": 4293
    },
    {
      "epoch": 0.5356118248721466,
      "grad_norm": 0.0,
      "learning_rate": 9.333874799609378e-06,
      "loss": 0.9334,
      "step": 4294
    },
    {
      "epoch": 0.5357365598104029,
      "grad_norm": 0.0,
      "learning_rate": 9.329843713492502e-06,
      "loss": 0.9123,
      "step": 4295
    },
    {
      "epoch": 0.5358612947486591,
      "grad_norm": 0.0,
      "learning_rate": 9.325812736762025e-06,
      "loss": 0.9288,
      "step": 4296
    },
    {
      "epoch": 0.5359860296869153,
      "grad_norm": 0.0,
      "learning_rate": 9.321781870075908e-06,
      "loss": 0.9348,
      "step": 4297
    },
    {
      "epoch": 0.5361107646251715,
      "grad_norm": 0.0,
      "learning_rate": 9.317751114092084e-06,
      "loss": 0.9612,
      "step": 4298
    },
    {
      "epoch": 0.5362354995634278,
      "grad_norm": 0.0,
      "learning_rate": 9.313720469468482e-06,
      "loss": 0.9932,
      "step": 4299
    },
    {
      "epoch": 0.5363602345016839,
      "grad_norm": 0.0,
      "learning_rate": 9.309689936863002e-06,
      "loss": 0.9686,
      "step": 4300
    },
    {
      "epoch": 0.5364849694399402,
      "grad_norm": 0.0,
      "learning_rate": 9.305659516933519e-06,
      "loss": 0.9814,
      "step": 4301
    },
    {
      "epoch": 0.5366097043781963,
      "grad_norm": 0.0,
      "learning_rate": 9.301629210337914e-06,
      "loss": 0.9477,
      "step": 4302
    },
    {
      "epoch": 0.5367344393164525,
      "grad_norm": 0.0,
      "learning_rate": 9.297599017734023e-06,
      "loss": 0.9815,
      "step": 4303
    },
    {
      "epoch": 0.5368591742547087,
      "grad_norm": 0.0,
      "learning_rate": 9.293568939779685e-06,
      "loss": 0.9638,
      "step": 4304
    },
    {
      "epoch": 0.5369839091929649,
      "grad_norm": 0.0,
      "learning_rate": 9.289538977132702e-06,
      "loss": 0.9263,
      "step": 4305
    },
    {
      "epoch": 0.5371086441312212,
      "grad_norm": 0.0,
      "learning_rate": 9.285509130450869e-06,
      "loss": 0.9509,
      "step": 4306
    },
    {
      "epoch": 0.5372333790694773,
      "grad_norm": 0.0,
      "learning_rate": 9.281479400391958e-06,
      "loss": 0.9564,
      "step": 4307
    },
    {
      "epoch": 0.5373581140077336,
      "grad_norm": 0.0,
      "learning_rate": 9.27744978761372e-06,
      "loss": 0.9209,
      "step": 4308
    },
    {
      "epoch": 0.5374828489459897,
      "grad_norm": 0.0,
      "learning_rate": 9.273420292773895e-06,
      "loss": 0.9428,
      "step": 4309
    },
    {
      "epoch": 0.537607583884246,
      "grad_norm": 0.0,
      "learning_rate": 9.269390916530197e-06,
      "loss": 0.9859,
      "step": 4310
    },
    {
      "epoch": 0.5377323188225022,
      "grad_norm": 0.0,
      "learning_rate": 9.265361659540315e-06,
      "loss": 0.9895,
      "step": 4311
    },
    {
      "epoch": 0.5378570537607584,
      "grad_norm": 0.0,
      "learning_rate": 9.261332522461934e-06,
      "loss": 0.8889,
      "step": 4312
    },
    {
      "epoch": 0.5379817886990146,
      "grad_norm": 0.0,
      "learning_rate": 9.257303505952705e-06,
      "loss": 0.9487,
      "step": 4313
    },
    {
      "epoch": 0.5381065236372709,
      "grad_norm": 0.0,
      "learning_rate": 9.253274610670269e-06,
      "loss": 0.9161,
      "step": 4314
    },
    {
      "epoch": 0.538231258575527,
      "grad_norm": 0.0,
      "learning_rate": 9.249245837272237e-06,
      "loss": 0.9442,
      "step": 4315
    },
    {
      "epoch": 0.5383559935137832,
      "grad_norm": 0.0,
      "learning_rate": 9.245217186416218e-06,
      "loss": 0.9533,
      "step": 4316
    },
    {
      "epoch": 0.5384807284520394,
      "grad_norm": 0.0,
      "learning_rate": 9.241188658759779e-06,
      "loss": 0.9418,
      "step": 4317
    },
    {
      "epoch": 0.5386054633902956,
      "grad_norm": 0.0,
      "learning_rate": 9.237160254960477e-06,
      "loss": 0.9193,
      "step": 4318
    },
    {
      "epoch": 0.5387301983285518,
      "grad_norm": 0.0,
      "learning_rate": 9.233131975675857e-06,
      "loss": 0.9479,
      "step": 4319
    },
    {
      "epoch": 0.538854933266808,
      "grad_norm": 0.0,
      "learning_rate": 9.229103821563428e-06,
      "loss": 0.9608,
      "step": 4320
    },
    {
      "epoch": 0.5389796682050643,
      "grad_norm": 0.0,
      "learning_rate": 9.225075793280693e-06,
      "loss": 0.9468,
      "step": 4321
    },
    {
      "epoch": 0.5391044031433204,
      "grad_norm": 0.0,
      "learning_rate": 9.221047891485122e-06,
      "loss": 0.9361,
      "step": 4322
    },
    {
      "epoch": 0.5392291380815767,
      "grad_norm": 0.0,
      "learning_rate": 9.217020116834173e-06,
      "loss": 0.9632,
      "step": 4323
    },
    {
      "epoch": 0.5393538730198328,
      "grad_norm": 0.0,
      "learning_rate": 9.212992469985278e-06,
      "loss": 0.9288,
      "step": 4324
    },
    {
      "epoch": 0.5394786079580891,
      "grad_norm": 0.0,
      "learning_rate": 9.208964951595852e-06,
      "loss": 0.944,
      "step": 4325
    },
    {
      "epoch": 0.5396033428963453,
      "grad_norm": 0.0,
      "learning_rate": 9.204937562323293e-06,
      "loss": 0.9607,
      "step": 4326
    },
    {
      "epoch": 0.5397280778346015,
      "grad_norm": 0.0,
      "learning_rate": 9.200910302824964e-06,
      "loss": 0.9942,
      "step": 4327
    },
    {
      "epoch": 0.5398528127728577,
      "grad_norm": 0.0,
      "learning_rate": 9.196883173758215e-06,
      "loss": 0.959,
      "step": 4328
    },
    {
      "epoch": 0.5399775477111138,
      "grad_norm": 0.0,
      "learning_rate": 9.192856175780383e-06,
      "loss": 0.9679,
      "step": 4329
    },
    {
      "epoch": 0.5401022826493701,
      "grad_norm": 0.0,
      "learning_rate": 9.188829309548768e-06,
      "loss": 0.9429,
      "step": 4330
    },
    {
      "epoch": 0.5402270175876263,
      "grad_norm": 0.0,
      "learning_rate": 9.184802575720664e-06,
      "loss": 0.9162,
      "step": 4331
    },
    {
      "epoch": 0.5403517525258825,
      "grad_norm": 0.0,
      "learning_rate": 9.180775974953325e-06,
      "loss": 0.9678,
      "step": 4332
    },
    {
      "epoch": 0.5404764874641387,
      "grad_norm": 0.0,
      "learning_rate": 9.176749507904002e-06,
      "loss": 0.9248,
      "step": 4333
    },
    {
      "epoch": 0.5406012224023949,
      "grad_norm": 0.0,
      "learning_rate": 9.172723175229915e-06,
      "loss": 0.9713,
      "step": 4334
    },
    {
      "epoch": 0.5407259573406511,
      "grad_norm": 0.0,
      "learning_rate": 9.168696977588256e-06,
      "loss": 0.9537,
      "step": 4335
    },
    {
      "epoch": 0.5408506922789074,
      "grad_norm": 0.0,
      "learning_rate": 9.164670915636214e-06,
      "loss": 0.9766,
      "step": 4336
    },
    {
      "epoch": 0.5409754272171635,
      "grad_norm": 0.0,
      "learning_rate": 9.160644990030932e-06,
      "loss": 0.9872,
      "step": 4337
    },
    {
      "epoch": 0.5411001621554198,
      "grad_norm": 0.0,
      "learning_rate": 9.156619201429551e-06,
      "loss": 0.9363,
      "step": 4338
    },
    {
      "epoch": 0.5412248970936759,
      "grad_norm": 0.0,
      "learning_rate": 9.152593550489178e-06,
      "loss": 0.8979,
      "step": 4339
    },
    {
      "epoch": 0.5413496320319321,
      "grad_norm": 0.0,
      "learning_rate": 9.148568037866898e-06,
      "loss": 0.9812,
      "step": 4340
    },
    {
      "epoch": 0.5414743669701884,
      "grad_norm": 0.0,
      "learning_rate": 9.14454266421978e-06,
      "loss": 0.9382,
      "step": 4341
    },
    {
      "epoch": 0.5415991019084445,
      "grad_norm": 0.0,
      "learning_rate": 9.140517430204864e-06,
      "loss": 0.9991,
      "step": 4342
    },
    {
      "epoch": 0.5417238368467008,
      "grad_norm": 0.0,
      "learning_rate": 9.136492336479172e-06,
      "loss": 0.9449,
      "step": 4343
    },
    {
      "epoch": 0.5418485717849569,
      "grad_norm": 0.0,
      "learning_rate": 9.132467383699702e-06,
      "loss": 0.9698,
      "step": 4344
    },
    {
      "epoch": 0.5419733067232132,
      "grad_norm": 0.0,
      "learning_rate": 9.128442572523418e-06,
      "loss": 0.9457,
      "step": 4345
    },
    {
      "epoch": 0.5420980416614694,
      "grad_norm": 0.0,
      "learning_rate": 9.124417903607286e-06,
      "loss": 0.9315,
      "step": 4346
    },
    {
      "epoch": 0.5422227765997256,
      "grad_norm": 0.0,
      "learning_rate": 9.120393377608219e-06,
      "loss": 0.9493,
      "step": 4347
    },
    {
      "epoch": 0.5423475115379818,
      "grad_norm": 0.0,
      "learning_rate": 9.116368995183132e-06,
      "loss": 0.9763,
      "step": 4348
    },
    {
      "epoch": 0.542472246476238,
      "grad_norm": 0.0,
      "learning_rate": 9.1123447569889e-06,
      "loss": 1.0093,
      "step": 4349
    },
    {
      "epoch": 0.5425969814144942,
      "grad_norm": 0.0,
      "learning_rate": 9.108320663682381e-06,
      "loss": 0.912,
      "step": 4350
    },
    {
      "epoch": 0.5427217163527505,
      "grad_norm": 0.0,
      "learning_rate": 9.10429671592041e-06,
      "loss": 0.9321,
      "step": 4351
    },
    {
      "epoch": 0.5428464512910066,
      "grad_norm": 0.0,
      "learning_rate": 9.100272914359793e-06,
      "loss": 0.9247,
      "step": 4352
    },
    {
      "epoch": 0.5429711862292628,
      "grad_norm": 0.0,
      "learning_rate": 9.096249259657322e-06,
      "loss": 0.8941,
      "step": 4353
    },
    {
      "epoch": 0.543095921167519,
      "grad_norm": 0.0,
      "learning_rate": 9.09222575246975e-06,
      "loss": 0.9674,
      "step": 4354
    },
    {
      "epoch": 0.5432206561057752,
      "grad_norm": 0.0,
      "learning_rate": 9.088202393453823e-06,
      "loss": 0.9368,
      "step": 4355
    },
    {
      "epoch": 0.5433453910440315,
      "grad_norm": 0.0,
      "learning_rate": 9.084179183266254e-06,
      "loss": 0.9525,
      "step": 4356
    },
    {
      "epoch": 0.5434701259822876,
      "grad_norm": 0.0,
      "learning_rate": 9.080156122563726e-06,
      "loss": 0.919,
      "step": 4357
    },
    {
      "epoch": 0.5435948609205439,
      "grad_norm": 0.0,
      "learning_rate": 9.07613321200291e-06,
      "loss": 0.9183,
      "step": 4358
    },
    {
      "epoch": 0.5437195958588,
      "grad_norm": 0.0,
      "learning_rate": 9.072110452240445e-06,
      "loss": 0.8995,
      "step": 4359
    },
    {
      "epoch": 0.5438443307970563,
      "grad_norm": 0.0,
      "learning_rate": 9.068087843932945e-06,
      "loss": 0.9379,
      "step": 4360
    },
    {
      "epoch": 0.5439690657353125,
      "grad_norm": 0.0,
      "learning_rate": 9.064065387737005e-06,
      "loss": 0.8884,
      "step": 4361
    },
    {
      "epoch": 0.5440938006735687,
      "grad_norm": 0.0,
      "learning_rate": 9.060043084309184e-06,
      "loss": 0.9546,
      "step": 4362
    },
    {
      "epoch": 0.5442185356118249,
      "grad_norm": 0.0,
      "learning_rate": 9.05602093430603e-06,
      "loss": 0.9132,
      "step": 4363
    },
    {
      "epoch": 0.544343270550081,
      "grad_norm": 0.0,
      "learning_rate": 9.051998938384055e-06,
      "loss": 0.925,
      "step": 4364
    },
    {
      "epoch": 0.5444680054883373,
      "grad_norm": 0.0,
      "learning_rate": 9.047977097199756e-06,
      "loss": 0.8782,
      "step": 4365
    },
    {
      "epoch": 0.5445927404265934,
      "grad_norm": 0.0,
      "learning_rate": 9.04395541140959e-06,
      "loss": 0.886,
      "step": 4366
    },
    {
      "epoch": 0.5447174753648497,
      "grad_norm": 0.0,
      "learning_rate": 9.039933881670008e-06,
      "loss": 0.9511,
      "step": 4367
    },
    {
      "epoch": 0.5448422103031059,
      "grad_norm": 0.0,
      "learning_rate": 9.035912508637419e-06,
      "loss": 0.9265,
      "step": 4368
    },
    {
      "epoch": 0.5449669452413621,
      "grad_norm": 0.0,
      "learning_rate": 9.03189129296821e-06,
      "loss": 0.9664,
      "step": 4369
    },
    {
      "epoch": 0.5450916801796183,
      "grad_norm": 0.0,
      "learning_rate": 9.027870235318751e-06,
      "loss": 0.9657,
      "step": 4370
    },
    {
      "epoch": 0.5452164151178746,
      "grad_norm": 0.0,
      "learning_rate": 9.023849336345371e-06,
      "loss": 0.9913,
      "step": 4371
    },
    {
      "epoch": 0.5453411500561307,
      "grad_norm": 0.0,
      "learning_rate": 9.019828596704394e-06,
      "loss": 0.9457,
      "step": 4372
    },
    {
      "epoch": 0.545465884994387,
      "grad_norm": 0.0,
      "learning_rate": 9.015808017052103e-06,
      "loss": 0.9243,
      "step": 4373
    },
    {
      "epoch": 0.5455906199326431,
      "grad_norm": 0.0,
      "learning_rate": 9.011787598044746e-06,
      "loss": 0.9233,
      "step": 4374
    },
    {
      "epoch": 0.5457153548708994,
      "grad_norm": 0.0,
      "learning_rate": 9.007767340338572e-06,
      "loss": 0.9859,
      "step": 4375
    },
    {
      "epoch": 0.5458400898091555,
      "grad_norm": 0.0,
      "learning_rate": 9.003747244589778e-06,
      "loss": 0.9579,
      "step": 4376
    },
    {
      "epoch": 0.5459648247474117,
      "grad_norm": 0.0,
      "learning_rate": 8.999727311454552e-06,
      "loss": 0.9429,
      "step": 4377
    },
    {
      "epoch": 0.546089559685668,
      "grad_norm": 0.0,
      "learning_rate": 8.995707541589045e-06,
      "loss": 0.9537,
      "step": 4378
    },
    {
      "epoch": 0.5462142946239241,
      "grad_norm": 0.0,
      "learning_rate": 8.991687935649385e-06,
      "loss": 0.946,
      "step": 4379
    },
    {
      "epoch": 0.5463390295621804,
      "grad_norm": 0.0,
      "learning_rate": 8.987668494291673e-06,
      "loss": 0.9217,
      "step": 4380
    },
    {
      "epoch": 0.5464637645004365,
      "grad_norm": 0.0,
      "learning_rate": 8.983649218171981e-06,
      "loss": 0.9821,
      "step": 4381
    },
    {
      "epoch": 0.5465884994386928,
      "grad_norm": 0.0,
      "learning_rate": 8.979630107946362e-06,
      "loss": 0.9184,
      "step": 4382
    },
    {
      "epoch": 0.546713234376949,
      "grad_norm": 0.0,
      "learning_rate": 8.975611164270832e-06,
      "loss": 0.9465,
      "step": 4383
    },
    {
      "epoch": 0.5468379693152052,
      "grad_norm": 0.0,
      "learning_rate": 8.971592387801384e-06,
      "loss": 0.9021,
      "step": 4384
    },
    {
      "epoch": 0.5469627042534614,
      "grad_norm": 0.0,
      "learning_rate": 8.967573779193987e-06,
      "loss": 0.9368,
      "step": 4385
    },
    {
      "epoch": 0.5470874391917177,
      "grad_norm": 0.0,
      "learning_rate": 8.963555339104574e-06,
      "loss": 0.9154,
      "step": 4386
    },
    {
      "epoch": 0.5472121741299738,
      "grad_norm": 0.0,
      "learning_rate": 8.95953706818906e-06,
      "loss": 0.8944,
      "step": 4387
    },
    {
      "epoch": 0.54733690906823,
      "grad_norm": 0.0,
      "learning_rate": 8.955518967103324e-06,
      "loss": 0.9611,
      "step": 4388
    },
    {
      "epoch": 0.5474616440064862,
      "grad_norm": 0.0,
      "learning_rate": 8.951501036503225e-06,
      "loss": 0.9338,
      "step": 4389
    },
    {
      "epoch": 0.5475863789447424,
      "grad_norm": 0.0,
      "learning_rate": 8.947483277044593e-06,
      "loss": 0.9219,
      "step": 4390
    },
    {
      "epoch": 0.5477111138829986,
      "grad_norm": 0.0,
      "learning_rate": 8.943465689383215e-06,
      "loss": 0.9944,
      "step": 4391
    },
    {
      "epoch": 0.5478358488212548,
      "grad_norm": 0.0,
      "learning_rate": 8.93944827417488e-06,
      "loss": 0.953,
      "step": 4392
    },
    {
      "epoch": 0.5479605837595111,
      "grad_norm": 0.0,
      "learning_rate": 8.935431032075317e-06,
      "loss": 0.9475,
      "step": 4393
    },
    {
      "epoch": 0.5480853186977672,
      "grad_norm": 0.0,
      "learning_rate": 8.931413963740251e-06,
      "loss": 0.9708,
      "step": 4394
    },
    {
      "epoch": 0.5482100536360235,
      "grad_norm": 0.0,
      "learning_rate": 8.927397069825366e-06,
      "loss": 0.982,
      "step": 4395
    },
    {
      "epoch": 0.5483347885742796,
      "grad_norm": 0.0,
      "learning_rate": 8.923380350986317e-06,
      "loss": 0.9641,
      "step": 4396
    },
    {
      "epoch": 0.5484595235125359,
      "grad_norm": 0.0,
      "learning_rate": 8.919363807878738e-06,
      "loss": 0.907,
      "step": 4397
    },
    {
      "epoch": 0.5485842584507921,
      "grad_norm": 0.0,
      "learning_rate": 8.915347441158223e-06,
      "loss": 0.9182,
      "step": 4398
    },
    {
      "epoch": 0.5487089933890483,
      "grad_norm": 0.0,
      "learning_rate": 8.911331251480357e-06,
      "loss": 0.976,
      "step": 4399
    },
    {
      "epoch": 0.5488337283273045,
      "grad_norm": 0.0,
      "learning_rate": 8.90731523950067e-06,
      "loss": 0.9562,
      "step": 4400
    },
    {
      "epoch": 0.5489584632655606,
      "grad_norm": 0.0,
      "learning_rate": 8.903299405874685e-06,
      "loss": 0.9573,
      "step": 4401
    },
    {
      "epoch": 0.5490831982038169,
      "grad_norm": 0.0,
      "learning_rate": 8.899283751257885e-06,
      "loss": 0.9833,
      "step": 4402
    },
    {
      "epoch": 0.5492079331420731,
      "grad_norm": 0.0,
      "learning_rate": 8.895268276305724e-06,
      "loss": 0.927,
      "step": 4403
    },
    {
      "epoch": 0.5493326680803293,
      "grad_norm": 0.0,
      "learning_rate": 8.891252981673633e-06,
      "loss": 0.9845,
      "step": 4404
    },
    {
      "epoch": 0.5494574030185855,
      "grad_norm": 0.0,
      "learning_rate": 8.887237868017004e-06,
      "loss": 0.964,
      "step": 4405
    },
    {
      "epoch": 0.5495821379568417,
      "grad_norm": 0.0,
      "learning_rate": 8.88322293599121e-06,
      "loss": 0.9446,
      "step": 4406
    },
    {
      "epoch": 0.5497068728950979,
      "grad_norm": 0.0,
      "learning_rate": 8.879208186251586e-06,
      "loss": 0.8941,
      "step": 4407
    },
    {
      "epoch": 0.5498316078333542,
      "grad_norm": 0.0,
      "learning_rate": 8.875193619453438e-06,
      "loss": 0.9326,
      "step": 4408
    },
    {
      "epoch": 0.5499563427716103,
      "grad_norm": 0.0,
      "learning_rate": 8.871179236252053e-06,
      "loss": 0.923,
      "step": 4409
    },
    {
      "epoch": 0.5500810777098666,
      "grad_norm": 0.0,
      "learning_rate": 8.867165037302667e-06,
      "loss": 0.9323,
      "step": 4410
    },
    {
      "epoch": 0.5502058126481227,
      "grad_norm": 0.0,
      "learning_rate": 8.863151023260512e-06,
      "loss": 0.9415,
      "step": 4411
    },
    {
      "epoch": 0.5503305475863789,
      "grad_norm": 0.0,
      "learning_rate": 8.859137194780768e-06,
      "loss": 0.9488,
      "step": 4412
    },
    {
      "epoch": 0.5504552825246352,
      "grad_norm": 0.0,
      "learning_rate": 8.855123552518593e-06,
      "loss": 0.9297,
      "step": 4413
    },
    {
      "epoch": 0.5505800174628913,
      "grad_norm": 0.0,
      "learning_rate": 8.851110097129119e-06,
      "loss": 0.948,
      "step": 4414
    },
    {
      "epoch": 0.5507047524011476,
      "grad_norm": 0.0,
      "learning_rate": 8.847096829267437e-06,
      "loss": 0.9055,
      "step": 4415
    },
    {
      "epoch": 0.5508294873394037,
      "grad_norm": 0.0,
      "learning_rate": 8.84308374958862e-06,
      "loss": 0.9479,
      "step": 4416
    },
    {
      "epoch": 0.55095422227766,
      "grad_norm": 0.0,
      "learning_rate": 8.839070858747697e-06,
      "loss": 0.9504,
      "step": 4417
    },
    {
      "epoch": 0.5510789572159162,
      "grad_norm": 0.0,
      "learning_rate": 8.83505815739968e-06,
      "loss": 0.9267,
      "step": 4418
    },
    {
      "epoch": 0.5512036921541724,
      "grad_norm": 0.0,
      "learning_rate": 8.831045646199544e-06,
      "loss": 0.9939,
      "step": 4419
    },
    {
      "epoch": 0.5513284270924286,
      "grad_norm": 0.0,
      "learning_rate": 8.827033325802218e-06,
      "loss": 0.8871,
      "step": 4420
    },
    {
      "epoch": 0.5514531620306848,
      "grad_norm": 0.0,
      "learning_rate": 8.82302119686263e-06,
      "loss": 0.8951,
      "step": 4421
    },
    {
      "epoch": 0.551577896968941,
      "grad_norm": 0.0,
      "learning_rate": 8.819009260035652e-06,
      "loss": 0.9508,
      "step": 4422
    },
    {
      "epoch": 0.5517026319071973,
      "grad_norm": 0.0,
      "learning_rate": 8.814997515976137e-06,
      "loss": 0.961,
      "step": 4423
    },
    {
      "epoch": 0.5518273668454534,
      "grad_norm": 0.0,
      "learning_rate": 8.810985965338901e-06,
      "loss": 0.9696,
      "step": 4424
    },
    {
      "epoch": 0.5519521017837096,
      "grad_norm": 0.0,
      "learning_rate": 8.806974608778728e-06,
      "loss": 0.9518,
      "step": 4425
    },
    {
      "epoch": 0.5520768367219658,
      "grad_norm": 0.0,
      "learning_rate": 8.802963446950378e-06,
      "loss": 0.8907,
      "step": 4426
    },
    {
      "epoch": 0.552201571660222,
      "grad_norm": 0.0,
      "learning_rate": 8.798952480508564e-06,
      "loss": 0.9745,
      "step": 4427
    },
    {
      "epoch": 0.5523263065984783,
      "grad_norm": 0.0,
      "learning_rate": 8.794941710107991e-06,
      "loss": 0.9348,
      "step": 4428
    },
    {
      "epoch": 0.5524510415367344,
      "grad_norm": 0.0,
      "learning_rate": 8.79093113640331e-06,
      "loss": 0.9979,
      "step": 4429
    },
    {
      "epoch": 0.5525757764749907,
      "grad_norm": 0.0,
      "learning_rate": 8.786920760049142e-06,
      "loss": 0.9131,
      "step": 4430
    },
    {
      "epoch": 0.5527005114132468,
      "grad_norm": 0.0,
      "learning_rate": 8.782910581700093e-06,
      "loss": 0.9649,
      "step": 4431
    },
    {
      "epoch": 0.5528252463515031,
      "grad_norm": 0.0,
      "learning_rate": 8.778900602010715e-06,
      "loss": 0.972,
      "step": 4432
    },
    {
      "epoch": 0.5529499812897593,
      "grad_norm": 0.0,
      "learning_rate": 8.774890821635548e-06,
      "loss": 0.9317,
      "step": 4433
    },
    {
      "epoch": 0.5530747162280155,
      "grad_norm": 0.0,
      "learning_rate": 8.770881241229079e-06,
      "loss": 0.9288,
      "step": 4434
    },
    {
      "epoch": 0.5531994511662717,
      "grad_norm": 0.0,
      "learning_rate": 8.76687186144578e-06,
      "loss": 0.9626,
      "step": 4435
    },
    {
      "epoch": 0.5533241861045278,
      "grad_norm": 0.0,
      "learning_rate": 8.762862682940082e-06,
      "loss": 0.8851,
      "step": 4436
    },
    {
      "epoch": 0.5534489210427841,
      "grad_norm": 0.0,
      "learning_rate": 8.758853706366375e-06,
      "loss": 0.9554,
      "step": 4437
    },
    {
      "epoch": 0.5535736559810402,
      "grad_norm": 0.0,
      "learning_rate": 8.754844932379038e-06,
      "loss": 0.9778,
      "step": 4438
    },
    {
      "epoch": 0.5536983909192965,
      "grad_norm": 0.0,
      "learning_rate": 8.750836361632398e-06,
      "loss": 0.9265,
      "step": 4439
    },
    {
      "epoch": 0.5538231258575527,
      "grad_norm": 0.0,
      "learning_rate": 8.746827994780755e-06,
      "loss": 0.9776,
      "step": 4440
    },
    {
      "epoch": 0.5539478607958089,
      "grad_norm": 0.0,
      "learning_rate": 8.742819832478376e-06,
      "loss": 0.9061,
      "step": 4441
    },
    {
      "epoch": 0.5540725957340651,
      "grad_norm": 0.0,
      "learning_rate": 8.738811875379491e-06,
      "loss": 0.9463,
      "step": 4442
    },
    {
      "epoch": 0.5541973306723214,
      "grad_norm": 0.0,
      "learning_rate": 8.734804124138303e-06,
      "loss": 0.9334,
      "step": 4443
    },
    {
      "epoch": 0.5543220656105775,
      "grad_norm": 0.0,
      "learning_rate": 8.730796579408976e-06,
      "loss": 0.9494,
      "step": 4444
    },
    {
      "epoch": 0.5544468005488338,
      "grad_norm": 0.0,
      "learning_rate": 8.72678924184565e-06,
      "loss": 0.9605,
      "step": 4445
    },
    {
      "epoch": 0.5545715354870899,
      "grad_norm": 0.0,
      "learning_rate": 8.722782112102413e-06,
      "loss": 0.9458,
      "step": 4446
    },
    {
      "epoch": 0.5546962704253462,
      "grad_norm": 0.0,
      "learning_rate": 8.71877519083333e-06,
      "loss": 0.9312,
      "step": 4447
    },
    {
      "epoch": 0.5548210053636023,
      "grad_norm": 0.0,
      "learning_rate": 8.714768478692441e-06,
      "loss": 0.9468,
      "step": 4448
    },
    {
      "epoch": 0.5549457403018585,
      "grad_norm": 0.0,
      "learning_rate": 8.710761976333734e-06,
      "loss": 0.9237,
      "step": 4449
    },
    {
      "epoch": 0.5550704752401148,
      "grad_norm": 0.0,
      "learning_rate": 8.706755684411176e-06,
      "loss": 0.9448,
      "step": 4450
    },
    {
      "epoch": 0.5551952101783709,
      "grad_norm": 0.0,
      "learning_rate": 8.702749603578691e-06,
      "loss": 0.8897,
      "step": 4451
    },
    {
      "epoch": 0.5553199451166272,
      "grad_norm": 0.0,
      "learning_rate": 8.698743734490174e-06,
      "loss": 0.9253,
      "step": 4452
    },
    {
      "epoch": 0.5554446800548833,
      "grad_norm": 0.0,
      "learning_rate": 8.694738077799487e-06,
      "loss": 0.8929,
      "step": 4453
    },
    {
      "epoch": 0.5555694149931396,
      "grad_norm": 0.0,
      "learning_rate": 8.690732634160446e-06,
      "loss": 0.9496,
      "step": 4454
    },
    {
      "epoch": 0.5556941499313958,
      "grad_norm": 0.0,
      "learning_rate": 8.686727404226852e-06,
      "loss": 0.9638,
      "step": 4455
    },
    {
      "epoch": 0.555818884869652,
      "grad_norm": 0.0,
      "learning_rate": 8.682722388652446e-06,
      "loss": 0.9502,
      "step": 4456
    },
    {
      "epoch": 0.5559436198079082,
      "grad_norm": 0.0,
      "learning_rate": 8.678717588090959e-06,
      "loss": 0.9354,
      "step": 4457
    },
    {
      "epoch": 0.5560683547461645,
      "grad_norm": 0.0,
      "learning_rate": 8.674713003196071e-06,
      "loss": 0.9446,
      "step": 4458
    },
    {
      "epoch": 0.5561930896844206,
      "grad_norm": 0.0,
      "learning_rate": 8.67070863462143e-06,
      "loss": 0.9843,
      "step": 4459
    },
    {
      "epoch": 0.5563178246226768,
      "grad_norm": 0.0,
      "learning_rate": 8.666704483020652e-06,
      "loss": 0.936,
      "step": 4460
    },
    {
      "epoch": 0.556442559560933,
      "grad_norm": 0.0,
      "learning_rate": 8.662700549047313e-06,
      "loss": 0.9872,
      "step": 4461
    },
    {
      "epoch": 0.5565672944991892,
      "grad_norm": 0.0,
      "learning_rate": 8.658696833354959e-06,
      "loss": 0.9621,
      "step": 4462
    },
    {
      "epoch": 0.5566920294374454,
      "grad_norm": 0.0,
      "learning_rate": 8.654693336597096e-06,
      "loss": 0.9365,
      "step": 4463
    },
    {
      "epoch": 0.5568167643757016,
      "grad_norm": 0.0,
      "learning_rate": 8.650690059427192e-06,
      "loss": 0.9683,
      "step": 4464
    },
    {
      "epoch": 0.5569414993139579,
      "grad_norm": 0.0,
      "learning_rate": 8.646687002498692e-06,
      "loss": 0.9937,
      "step": 4465
    },
    {
      "epoch": 0.557066234252214,
      "grad_norm": 0.0,
      "learning_rate": 8.642684166464984e-06,
      "loss": 0.8877,
      "step": 4466
    },
    {
      "epoch": 0.5571909691904703,
      "grad_norm": 0.0,
      "learning_rate": 8.638681551979442e-06,
      "loss": 0.9705,
      "step": 4467
    },
    {
      "epoch": 0.5573157041287264,
      "grad_norm": 0.0,
      "learning_rate": 8.634679159695387e-06,
      "loss": 0.9457,
      "step": 4468
    },
    {
      "epoch": 0.5574404390669827,
      "grad_norm": 0.0,
      "learning_rate": 8.630676990266115e-06,
      "loss": 0.9968,
      "step": 4469
    },
    {
      "epoch": 0.5575651740052389,
      "grad_norm": 0.0,
      "learning_rate": 8.626675044344878e-06,
      "loss": 0.9441,
      "step": 4470
    },
    {
      "epoch": 0.5576899089434951,
      "grad_norm": 0.0,
      "learning_rate": 8.622673322584894e-06,
      "loss": 0.911,
      "step": 4471
    },
    {
      "epoch": 0.5578146438817513,
      "grad_norm": 0.0,
      "learning_rate": 8.618671825639348e-06,
      "loss": 0.9293,
      "step": 4472
    },
    {
      "epoch": 0.5579393788200074,
      "grad_norm": 0.0,
      "learning_rate": 8.614670554161377e-06,
      "loss": 0.9506,
      "step": 4473
    },
    {
      "epoch": 0.5580641137582637,
      "grad_norm": 0.0,
      "learning_rate": 8.610669508804101e-06,
      "loss": 0.9925,
      "step": 4474
    },
    {
      "epoch": 0.5581888486965199,
      "grad_norm": 0.0,
      "learning_rate": 8.606668690220588e-06,
      "loss": 0.9176,
      "step": 4475
    },
    {
      "epoch": 0.5583135836347761,
      "grad_norm": 0.0,
      "learning_rate": 8.602668099063865e-06,
      "loss": 0.9493,
      "step": 4476
    },
    {
      "epoch": 0.5584383185730323,
      "grad_norm": 0.0,
      "learning_rate": 8.598667735986939e-06,
      "loss": 1.0171,
      "step": 4477
    },
    {
      "epoch": 0.5585630535112885,
      "grad_norm": 0.0,
      "learning_rate": 8.594667601642763e-06,
      "loss": 0.935,
      "step": 4478
    },
    {
      "epoch": 0.5586877884495447,
      "grad_norm": 0.0,
      "learning_rate": 8.590667696684266e-06,
      "loss": 0.9462,
      "step": 4479
    },
    {
      "epoch": 0.558812523387801,
      "grad_norm": 0.0,
      "learning_rate": 8.586668021764328e-06,
      "loss": 0.9436,
      "step": 4480
    },
    {
      "epoch": 0.5589372583260571,
      "grad_norm": 0.0,
      "learning_rate": 8.582668577535797e-06,
      "loss": 0.9308,
      "step": 4481
    },
    {
      "epoch": 0.5590619932643134,
      "grad_norm": 0.0,
      "learning_rate": 8.578669364651489e-06,
      "loss": 0.9745,
      "step": 4482
    },
    {
      "epoch": 0.5591867282025695,
      "grad_norm": 0.0,
      "learning_rate": 8.574670383764166e-06,
      "loss": 0.9036,
      "step": 4483
    },
    {
      "epoch": 0.5593114631408257,
      "grad_norm": 0.0,
      "learning_rate": 8.570671635526572e-06,
      "loss": 0.9576,
      "step": 4484
    },
    {
      "epoch": 0.559436198079082,
      "grad_norm": 0.0,
      "learning_rate": 8.5666731205914e-06,
      "loss": 0.9263,
      "step": 4485
    },
    {
      "epoch": 0.5595609330173381,
      "grad_norm": 0.0,
      "learning_rate": 8.562674839611311e-06,
      "loss": 0.9935,
      "step": 4486
    },
    {
      "epoch": 0.5596856679555944,
      "grad_norm": 0.0,
      "learning_rate": 8.558676793238922e-06,
      "loss": 0.9481,
      "step": 4487
    },
    {
      "epoch": 0.5598104028938505,
      "grad_norm": 0.0,
      "learning_rate": 8.554678982126814e-06,
      "loss": 0.951,
      "step": 4488
    },
    {
      "epoch": 0.5599351378321068,
      "grad_norm": 0.0,
      "learning_rate": 8.550681406927534e-06,
      "loss": 0.9467,
      "step": 4489
    },
    {
      "epoch": 0.560059872770363,
      "grad_norm": 0.0,
      "learning_rate": 8.546684068293583e-06,
      "loss": 0.925,
      "step": 4490
    },
    {
      "epoch": 0.5601846077086192,
      "grad_norm": 0.0,
      "learning_rate": 8.542686966877435e-06,
      "loss": 0.9433,
      "step": 4491
    },
    {
      "epoch": 0.5603093426468754,
      "grad_norm": 0.0,
      "learning_rate": 8.53869010333151e-06,
      "loss": 0.9449,
      "step": 4492
    },
    {
      "epoch": 0.5604340775851316,
      "grad_norm": 0.0,
      "learning_rate": 8.534693478308196e-06,
      "loss": 0.883,
      "step": 4493
    },
    {
      "epoch": 0.5605588125233878,
      "grad_norm": 0.0,
      "learning_rate": 8.530697092459851e-06,
      "loss": 0.9615,
      "step": 4494
    },
    {
      "epoch": 0.5606835474616441,
      "grad_norm": 0.0,
      "learning_rate": 8.52670094643878e-06,
      "loss": 0.956,
      "step": 4495
    },
    {
      "epoch": 0.5608082823999002,
      "grad_norm": 0.0,
      "learning_rate": 8.522705040897257e-06,
      "loss": 0.9292,
      "step": 4496
    },
    {
      "epoch": 0.5609330173381564,
      "grad_norm": 0.0,
      "learning_rate": 8.518709376487515e-06,
      "loss": 0.9648,
      "step": 4497
    },
    {
      "epoch": 0.5610577522764126,
      "grad_norm": 0.0,
      "learning_rate": 8.514713953861743e-06,
      "loss": 0.9091,
      "step": 4498
    },
    {
      "epoch": 0.5611824872146688,
      "grad_norm": 0.0,
      "learning_rate": 8.510718773672099e-06,
      "loss": 0.9552,
      "step": 4499
    },
    {
      "epoch": 0.5613072221529251,
      "grad_norm": 0.0,
      "learning_rate": 8.506723836570692e-06,
      "loss": 0.9333,
      "step": 4500
    },
    {
      "epoch": 0.5614319570911812,
      "grad_norm": 0.0,
      "learning_rate": 8.502729143209605e-06,
      "loss": 0.9448,
      "step": 4501
    },
    {
      "epoch": 0.5615566920294375,
      "grad_norm": 0.0,
      "learning_rate": 8.498734694240869e-06,
      "loss": 0.9329,
      "step": 4502
    },
    {
      "epoch": 0.5616814269676936,
      "grad_norm": 0.0,
      "learning_rate": 8.494740490316471e-06,
      "loss": 0.9264,
      "step": 4503
    },
    {
      "epoch": 0.5618061619059499,
      "grad_norm": 0.0,
      "learning_rate": 8.490746532088378e-06,
      "loss": 0.8917,
      "step": 4504
    },
    {
      "epoch": 0.561930896844206,
      "grad_norm": 0.0,
      "learning_rate": 8.486752820208496e-06,
      "loss": 0.9819,
      "step": 4505
    },
    {
      "epoch": 0.5620556317824623,
      "grad_norm": 0.0,
      "learning_rate": 8.482759355328704e-06,
      "loss": 0.9836,
      "step": 4506
    },
    {
      "epoch": 0.5621803667207185,
      "grad_norm": 0.0,
      "learning_rate": 8.478766138100834e-06,
      "loss": 0.9706,
      "step": 4507
    },
    {
      "epoch": 0.5623051016589746,
      "grad_norm": 0.0,
      "learning_rate": 8.474773169176683e-06,
      "loss": 0.9146,
      "step": 4508
    },
    {
      "epoch": 0.5624298365972309,
      "grad_norm": 0.0,
      "learning_rate": 8.470780449208e-06,
      "loss": 0.9721,
      "step": 4509
    },
    {
      "epoch": 0.562554571535487,
      "grad_norm": 0.0,
      "learning_rate": 8.466787978846493e-06,
      "loss": 0.8869,
      "step": 4510
    },
    {
      "epoch": 0.5626793064737433,
      "grad_norm": 0.0,
      "learning_rate": 8.462795758743846e-06,
      "loss": 0.9053,
      "step": 4511
    },
    {
      "epoch": 0.5628040414119995,
      "grad_norm": 0.0,
      "learning_rate": 8.458803789551683e-06,
      "loss": 0.9066,
      "step": 4512
    },
    {
      "epoch": 0.5629287763502557,
      "grad_norm": 0.0,
      "learning_rate": 8.454812071921597e-06,
      "loss": 0.9282,
      "step": 4513
    },
    {
      "epoch": 0.5630535112885119,
      "grad_norm": 0.0,
      "learning_rate": 8.450820606505133e-06,
      "loss": 0.8972,
      "step": 4514
    },
    {
      "epoch": 0.5631782462267682,
      "grad_norm": 0.0,
      "learning_rate": 8.4468293939538e-06,
      "loss": 0.9553,
      "step": 4515
    },
    {
      "epoch": 0.5633029811650243,
      "grad_norm": 0.0,
      "learning_rate": 8.442838434919066e-06,
      "loss": 0.9291,
      "step": 4516
    },
    {
      "epoch": 0.5634277161032806,
      "grad_norm": 0.0,
      "learning_rate": 8.438847730052352e-06,
      "loss": 0.9078,
      "step": 4517
    },
    {
      "epoch": 0.5635524510415367,
      "grad_norm": 0.0,
      "learning_rate": 8.43485728000505e-06,
      "loss": 0.9388,
      "step": 4518
    },
    {
      "epoch": 0.563677185979793,
      "grad_norm": 0.0,
      "learning_rate": 8.430867085428496e-06,
      "loss": 0.8979,
      "step": 4519
    },
    {
      "epoch": 0.5638019209180491,
      "grad_norm": 0.0,
      "learning_rate": 8.426877146973987e-06,
      "loss": 0.9298,
      "step": 4520
    },
    {
      "epoch": 0.5639266558563053,
      "grad_norm": 0.0,
      "learning_rate": 8.422887465292789e-06,
      "loss": 0.8972,
      "step": 4521
    },
    {
      "epoch": 0.5640513907945616,
      "grad_norm": 0.0,
      "learning_rate": 8.418898041036115e-06,
      "loss": 0.9743,
      "step": 4522
    },
    {
      "epoch": 0.5641761257328177,
      "grad_norm": 0.0,
      "learning_rate": 8.414908874855142e-06,
      "loss": 1.0071,
      "step": 4523
    },
    {
      "epoch": 0.564300860671074,
      "grad_norm": 0.0,
      "learning_rate": 8.410919967400996e-06,
      "loss": 0.9045,
      "step": 4524
    },
    {
      "epoch": 0.5644255956093301,
      "grad_norm": 0.0,
      "learning_rate": 8.406931319324776e-06,
      "loss": 0.8844,
      "step": 4525
    },
    {
      "epoch": 0.5645503305475864,
      "grad_norm": 0.0,
      "learning_rate": 8.402942931277525e-06,
      "loss": 0.9738,
      "step": 4526
    },
    {
      "epoch": 0.5646750654858426,
      "grad_norm": 0.0,
      "learning_rate": 8.398954803910245e-06,
      "loss": 0.9846,
      "step": 4527
    },
    {
      "epoch": 0.5647998004240988,
      "grad_norm": 0.0,
      "learning_rate": 8.39496693787391e-06,
      "loss": 0.8857,
      "step": 4528
    },
    {
      "epoch": 0.564924535362355,
      "grad_norm": 0.0,
      "learning_rate": 8.390979333819427e-06,
      "loss": 0.9429,
      "step": 4529
    },
    {
      "epoch": 0.5650492703006113,
      "grad_norm": 0.0,
      "learning_rate": 8.386991992397684e-06,
      "loss": 0.9242,
      "step": 4530
    },
    {
      "epoch": 0.5651740052388674,
      "grad_norm": 0.0,
      "learning_rate": 8.383004914259511e-06,
      "loss": 0.9515,
      "step": 4531
    },
    {
      "epoch": 0.5652987401771236,
      "grad_norm": 0.0,
      "learning_rate": 8.379018100055697e-06,
      "loss": 0.93,
      "step": 4532
    },
    {
      "epoch": 0.5654234751153798,
      "grad_norm": 0.0,
      "learning_rate": 8.375031550437e-06,
      "loss": 0.883,
      "step": 4533
    },
    {
      "epoch": 0.565548210053636,
      "grad_norm": 0.0,
      "learning_rate": 8.371045266054114e-06,
      "loss": 0.9259,
      "step": 4534
    },
    {
      "epoch": 0.5656729449918922,
      "grad_norm": 0.0,
      "learning_rate": 8.367059247557708e-06,
      "loss": 0.9507,
      "step": 4535
    },
    {
      "epoch": 0.5657976799301484,
      "grad_norm": 0.0,
      "learning_rate": 8.3630734955984e-06,
      "loss": 0.9372,
      "step": 4536
    },
    {
      "epoch": 0.5659224148684047,
      "grad_norm": 0.0,
      "learning_rate": 8.35908801082676e-06,
      "loss": 0.9851,
      "step": 4537
    },
    {
      "epoch": 0.5660471498066608,
      "grad_norm": 0.0,
      "learning_rate": 8.35510279389333e-06,
      "loss": 0.9031,
      "step": 4538
    },
    {
      "epoch": 0.5661718847449171,
      "grad_norm": 0.0,
      "learning_rate": 8.351117845448586e-06,
      "loss": 1.0056,
      "step": 4539
    },
    {
      "epoch": 0.5662966196831732,
      "grad_norm": 0.0,
      "learning_rate": 8.34713316614298e-06,
      "loss": 0.904,
      "step": 4540
    },
    {
      "epoch": 0.5664213546214295,
      "grad_norm": 0.0,
      "learning_rate": 8.34314875662691e-06,
      "loss": 0.9614,
      "step": 4541
    },
    {
      "epoch": 0.5665460895596857,
      "grad_norm": 0.0,
      "learning_rate": 8.339164617550732e-06,
      "loss": 0.9234,
      "step": 4542
    },
    {
      "epoch": 0.5666708244979419,
      "grad_norm": 0.0,
      "learning_rate": 8.335180749564759e-06,
      "loss": 0.944,
      "step": 4543
    },
    {
      "epoch": 0.5667955594361981,
      "grad_norm": 0.0,
      "learning_rate": 8.331197153319256e-06,
      "loss": 0.9401,
      "step": 4544
    },
    {
      "epoch": 0.5669202943744542,
      "grad_norm": 0.0,
      "learning_rate": 8.327213829464448e-06,
      "loss": 0.9148,
      "step": 4545
    },
    {
      "epoch": 0.5670450293127105,
      "grad_norm": 0.0,
      "learning_rate": 8.323230778650511e-06,
      "loss": 0.9448,
      "step": 4546
    },
    {
      "epoch": 0.5671697642509667,
      "grad_norm": 0.0,
      "learning_rate": 8.319248001527586e-06,
      "loss": 0.924,
      "step": 4547
    },
    {
      "epoch": 0.5672944991892229,
      "grad_norm": 0.0,
      "learning_rate": 8.315265498745762e-06,
      "loss": 0.9891,
      "step": 4548
    },
    {
      "epoch": 0.5674192341274791,
      "grad_norm": 0.0,
      "learning_rate": 8.311283270955072e-06,
      "loss": 0.9398,
      "step": 4549
    },
    {
      "epoch": 0.5675439690657353,
      "grad_norm": 0.0,
      "learning_rate": 8.307301318805531e-06,
      "loss": 1.0083,
      "step": 4550
    },
    {
      "epoch": 0.5676687040039915,
      "grad_norm": 0.0,
      "learning_rate": 8.303319642947083e-06,
      "loss": 0.9258,
      "step": 4551
    },
    {
      "epoch": 0.5677934389422478,
      "grad_norm": 0.0,
      "learning_rate": 8.299338244029646e-06,
      "loss": 0.9249,
      "step": 4552
    },
    {
      "epoch": 0.5679181738805039,
      "grad_norm": 0.0,
      "learning_rate": 8.295357122703081e-06,
      "loss": 0.9261,
      "step": 4553
    },
    {
      "epoch": 0.5680429088187602,
      "grad_norm": 0.0,
      "learning_rate": 8.291376279617203e-06,
      "loss": 0.9784,
      "step": 4554
    },
    {
      "epoch": 0.5681676437570163,
      "grad_norm": 0.0,
      "learning_rate": 8.287395715421796e-06,
      "loss": 0.9206,
      "step": 4555
    },
    {
      "epoch": 0.5682923786952725,
      "grad_norm": 0.0,
      "learning_rate": 8.283415430766576e-06,
      "loss": 0.9945,
      "step": 4556
    },
    {
      "epoch": 0.5684171136335288,
      "grad_norm": 0.0,
      "learning_rate": 8.279435426301238e-06,
      "loss": 0.9407,
      "step": 4557
    },
    {
      "epoch": 0.5685418485717849,
      "grad_norm": 0.0,
      "learning_rate": 8.27545570267541e-06,
      "loss": 1.0006,
      "step": 4558
    },
    {
      "epoch": 0.5686665835100412,
      "grad_norm": 0.0,
      "learning_rate": 8.27147626053869e-06,
      "loss": 0.9517,
      "step": 4559
    },
    {
      "epoch": 0.5687913184482973,
      "grad_norm": 0.0,
      "learning_rate": 8.26749710054062e-06,
      "loss": 1.0103,
      "step": 4560
    },
    {
      "epoch": 0.5689160533865536,
      "grad_norm": 0.0,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.9287,
      "step": 4561
    },
    {
      "epoch": 0.5690407883248098,
      "grad_norm": 0.0,
      "learning_rate": 8.25953962955838e-06,
      "loss": 0.9389,
      "step": 4562
    },
    {
      "epoch": 0.569165523263066,
      "grad_norm": 0.0,
      "learning_rate": 8.255561319873066e-06,
      "loss": 0.9494,
      "step": 4563
    },
    {
      "epoch": 0.5692902582013222,
      "grad_norm": 0.0,
      "learning_rate": 8.251583294924129e-06,
      "loss": 0.9696,
      "step": 4564
    },
    {
      "epoch": 0.5694149931395784,
      "grad_norm": 0.0,
      "learning_rate": 8.247605555360873e-06,
      "loss": 0.9305,
      "step": 4565
    },
    {
      "epoch": 0.5695397280778346,
      "grad_norm": 0.0,
      "learning_rate": 8.243628101832564e-06,
      "loss": 0.9147,
      "step": 4566
    },
    {
      "epoch": 0.5696644630160909,
      "grad_norm": 0.0,
      "learning_rate": 8.239650934988432e-06,
      "loss": 0.9635,
      "step": 4567
    },
    {
      "epoch": 0.569789197954347,
      "grad_norm": 0.0,
      "learning_rate": 8.235674055477641e-06,
      "loss": 0.8758,
      "step": 4568
    },
    {
      "epoch": 0.5699139328926032,
      "grad_norm": 0.0,
      "learning_rate": 8.231697463949327e-06,
      "loss": 0.9161,
      "step": 4569
    },
    {
      "epoch": 0.5700386678308594,
      "grad_norm": 0.0,
      "learning_rate": 8.227721161052564e-06,
      "loss": 0.9156,
      "step": 4570
    },
    {
      "epoch": 0.5701634027691156,
      "grad_norm": 0.0,
      "learning_rate": 8.223745147436385e-06,
      "loss": 0.9769,
      "step": 4571
    },
    {
      "epoch": 0.5702881377073719,
      "grad_norm": 0.0,
      "learning_rate": 8.21976942374978e-06,
      "loss": 0.9342,
      "step": 4572
    },
    {
      "epoch": 0.570412872645628,
      "grad_norm": 0.0,
      "learning_rate": 8.215793990641679e-06,
      "loss": 0.9666,
      "step": 4573
    },
    {
      "epoch": 0.5705376075838843,
      "grad_norm": 0.0,
      "learning_rate": 8.211818848760983e-06,
      "loss": 0.9431,
      "step": 4574
    },
    {
      "epoch": 0.5706623425221404,
      "grad_norm": 0.0,
      "learning_rate": 8.207843998756527e-06,
      "loss": 0.9115,
      "step": 4575
    },
    {
      "epoch": 0.5707870774603967,
      "grad_norm": 0.0,
      "learning_rate": 8.203869441277114e-06,
      "loss": 0.9266,
      "step": 4576
    },
    {
      "epoch": 0.5709118123986529,
      "grad_norm": 0.0,
      "learning_rate": 8.199895176971489e-06,
      "loss": 1.0261,
      "step": 4577
    },
    {
      "epoch": 0.5710365473369091,
      "grad_norm": 0.0,
      "learning_rate": 8.195921206488348e-06,
      "loss": 0.9282,
      "step": 4578
    },
    {
      "epoch": 0.5711612822751653,
      "grad_norm": 0.0,
      "learning_rate": 8.191947530476349e-06,
      "loss": 0.889,
      "step": 4579
    },
    {
      "epoch": 0.5712860172134214,
      "grad_norm": 0.0,
      "learning_rate": 8.187974149584093e-06,
      "loss": 0.9417,
      "step": 4580
    },
    {
      "epoch": 0.5714107521516777,
      "grad_norm": 0.0,
      "learning_rate": 8.184001064460137e-06,
      "loss": 0.9101,
      "step": 4581
    },
    {
      "epoch": 0.5715354870899338,
      "grad_norm": 0.0,
      "learning_rate": 8.180028275752988e-06,
      "loss": 0.9237,
      "step": 4582
    },
    {
      "epoch": 0.5716602220281901,
      "grad_norm": 0.0,
      "learning_rate": 8.176055784111103e-06,
      "loss": 1.0013,
      "step": 4583
    },
    {
      "epoch": 0.5717849569664463,
      "grad_norm": 0.0,
      "learning_rate": 8.172083590182902e-06,
      "loss": 0.9881,
      "step": 4584
    },
    {
      "epoch": 0.5719096919047025,
      "grad_norm": 0.0,
      "learning_rate": 8.168111694616733e-06,
      "loss": 0.9656,
      "step": 4585
    },
    {
      "epoch": 0.5720344268429587,
      "grad_norm": 0.0,
      "learning_rate": 8.164140098060923e-06,
      "loss": 0.9267,
      "step": 4586
    },
    {
      "epoch": 0.572159161781215,
      "grad_norm": 0.0,
      "learning_rate": 8.160168801163729e-06,
      "loss": 0.9723,
      "step": 4587
    },
    {
      "epoch": 0.5722838967194711,
      "grad_norm": 0.0,
      "learning_rate": 8.156197804573368e-06,
      "loss": 0.9097,
      "step": 4588
    },
    {
      "epoch": 0.5724086316577274,
      "grad_norm": 0.0,
      "learning_rate": 8.152227108938009e-06,
      "loss": 0.9404,
      "step": 4589
    },
    {
      "epoch": 0.5725333665959835,
      "grad_norm": 0.0,
      "learning_rate": 8.148256714905764e-06,
      "loss": 0.9327,
      "step": 4590
    },
    {
      "epoch": 0.5726581015342398,
      "grad_norm": 0.0,
      "learning_rate": 8.144286623124708e-06,
      "loss": 0.899,
      "step": 4591
    },
    {
      "epoch": 0.572782836472496,
      "grad_norm": 0.0,
      "learning_rate": 8.140316834242854e-06,
      "loss": 0.9143,
      "step": 4592
    },
    {
      "epoch": 0.5729075714107521,
      "grad_norm": 0.0,
      "learning_rate": 8.13634734890818e-06,
      "loss": 0.9937,
      "step": 4593
    },
    {
      "epoch": 0.5730323063490084,
      "grad_norm": 0.0,
      "learning_rate": 8.1323781677686e-06,
      "loss": 0.9675,
      "step": 4594
    },
    {
      "epoch": 0.5731570412872645,
      "grad_norm": 0.0,
      "learning_rate": 8.128409291471984e-06,
      "loss": 0.9542,
      "step": 4595
    },
    {
      "epoch": 0.5732817762255208,
      "grad_norm": 0.0,
      "learning_rate": 8.124440720666156e-06,
      "loss": 0.9324,
      "step": 4596
    },
    {
      "epoch": 0.5734065111637769,
      "grad_norm": 0.0,
      "learning_rate": 8.120472455998882e-06,
      "loss": 0.9923,
      "step": 4597
    },
    {
      "epoch": 0.5735312461020332,
      "grad_norm": 0.0,
      "learning_rate": 8.11650449811789e-06,
      "loss": 0.9463,
      "step": 4598
    },
    {
      "epoch": 0.5736559810402894,
      "grad_norm": 0.0,
      "learning_rate": 8.112536847670844e-06,
      "loss": 0.9743,
      "step": 4599
    },
    {
      "epoch": 0.5737807159785456,
      "grad_norm": 0.0,
      "learning_rate": 8.108569505305366e-06,
      "loss": 0.9724,
      "step": 4600
    },
    {
      "epoch": 0.5739054509168018,
      "grad_norm": 0.0,
      "learning_rate": 8.10460247166903e-06,
      "loss": 0.8949,
      "step": 4601
    },
    {
      "epoch": 0.574030185855058,
      "grad_norm": 0.0,
      "learning_rate": 8.100635747409347e-06,
      "loss": 0.9439,
      "step": 4602
    },
    {
      "epoch": 0.5741549207933142,
      "grad_norm": 0.0,
      "learning_rate": 8.0966693331738e-06,
      "loss": 0.9413,
      "step": 4603
    },
    {
      "epoch": 0.5742796557315704,
      "grad_norm": 0.0,
      "learning_rate": 8.0927032296098e-06,
      "loss": 0.9312,
      "step": 4604
    },
    {
      "epoch": 0.5744043906698266,
      "grad_norm": 0.0,
      "learning_rate": 8.088737437364712e-06,
      "loss": 0.9263,
      "step": 4605
    },
    {
      "epoch": 0.5745291256080828,
      "grad_norm": 0.0,
      "learning_rate": 8.08477195708586e-06,
      "loss": 0.9498,
      "step": 4606
    },
    {
      "epoch": 0.574653860546339,
      "grad_norm": 0.0,
      "learning_rate": 8.080806789420504e-06,
      "loss": 0.931,
      "step": 4607
    },
    {
      "epoch": 0.5747785954845952,
      "grad_norm": 0.0,
      "learning_rate": 8.076841935015866e-06,
      "loss": 0.9985,
      "step": 4608
    },
    {
      "epoch": 0.5749033304228515,
      "grad_norm": 0.0,
      "learning_rate": 8.072877394519103e-06,
      "loss": 0.9688,
      "step": 4609
    },
    {
      "epoch": 0.5750280653611076,
      "grad_norm": 0.0,
      "learning_rate": 8.068913168577336e-06,
      "loss": 0.9332,
      "step": 4610
    },
    {
      "epoch": 0.5751528002993639,
      "grad_norm": 0.0,
      "learning_rate": 8.064949257837622e-06,
      "loss": 0.9158,
      "step": 4611
    },
    {
      "epoch": 0.57527753523762,
      "grad_norm": 0.0,
      "learning_rate": 8.060985662946967e-06,
      "loss": 0.8925,
      "step": 4612
    },
    {
      "epoch": 0.5754022701758763,
      "grad_norm": 0.0,
      "learning_rate": 8.057022384552336e-06,
      "loss": 0.8866,
      "step": 4613
    },
    {
      "epoch": 0.5755270051141325,
      "grad_norm": 0.0,
      "learning_rate": 8.053059423300635e-06,
      "loss": 0.9282,
      "step": 4614
    },
    {
      "epoch": 0.5756517400523887,
      "grad_norm": 0.0,
      "learning_rate": 8.04909677983872e-06,
      "loss": 0.9357,
      "step": 4615
    },
    {
      "epoch": 0.5757764749906449,
      "grad_norm": 0.0,
      "learning_rate": 8.04513445481339e-06,
      "loss": 0.8948,
      "step": 4616
    },
    {
      "epoch": 0.575901209928901,
      "grad_norm": 0.0,
      "learning_rate": 8.041172448871399e-06,
      "loss": 0.8618,
      "step": 4617
    },
    {
      "epoch": 0.5760259448671573,
      "grad_norm": 0.0,
      "learning_rate": 8.037210762659447e-06,
      "loss": 0.94,
      "step": 4618
    },
    {
      "epoch": 0.5761506798054135,
      "grad_norm": 0.0,
      "learning_rate": 8.033249396824177e-06,
      "loss": 0.9382,
      "step": 4619
    },
    {
      "epoch": 0.5762754147436697,
      "grad_norm": 0.0,
      "learning_rate": 8.02928835201219e-06,
      "loss": 0.9283,
      "step": 4620
    },
    {
      "epoch": 0.5764001496819259,
      "grad_norm": 0.0,
      "learning_rate": 8.025327628870028e-06,
      "loss": 0.957,
      "step": 4621
    },
    {
      "epoch": 0.5765248846201821,
      "grad_norm": 0.0,
      "learning_rate": 8.021367228044172e-06,
      "loss": 0.9328,
      "step": 4622
    },
    {
      "epoch": 0.5766496195584383,
      "grad_norm": 0.0,
      "learning_rate": 8.01740715018107e-06,
      "loss": 1.008,
      "step": 4623
    },
    {
      "epoch": 0.5767743544966946,
      "grad_norm": 0.0,
      "learning_rate": 8.0134473959271e-06,
      "loss": 0.9907,
      "step": 4624
    },
    {
      "epoch": 0.5768990894349507,
      "grad_norm": 0.0,
      "learning_rate": 8.009487965928597e-06,
      "loss": 0.9268,
      "step": 4625
    },
    {
      "epoch": 0.577023824373207,
      "grad_norm": 0.0,
      "learning_rate": 8.005528860831835e-06,
      "loss": 0.8837,
      "step": 4626
    },
    {
      "epoch": 0.5771485593114631,
      "grad_norm": 0.0,
      "learning_rate": 8.001570081283048e-06,
      "loss": 0.9497,
      "step": 4627
    },
    {
      "epoch": 0.5772732942497193,
      "grad_norm": 0.0,
      "learning_rate": 7.997611627928403e-06,
      "loss": 0.994,
      "step": 4628
    },
    {
      "epoch": 0.5773980291879756,
      "grad_norm": 0.0,
      "learning_rate": 7.993653501414015e-06,
      "loss": 0.9672,
      "step": 4629
    },
    {
      "epoch": 0.5775227641262317,
      "grad_norm": 0.0,
      "learning_rate": 7.989695702385961e-06,
      "loss": 0.9156,
      "step": 4630
    },
    {
      "epoch": 0.577647499064488,
      "grad_norm": 0.0,
      "learning_rate": 7.985738231490245e-06,
      "loss": 0.9143,
      "step": 4631
    },
    {
      "epoch": 0.5777722340027441,
      "grad_norm": 0.0,
      "learning_rate": 7.981781089372832e-06,
      "loss": 0.9194,
      "step": 4632
    },
    {
      "epoch": 0.5778969689410004,
      "grad_norm": 0.0,
      "learning_rate": 7.977824276679623e-06,
      "loss": 0.9778,
      "step": 4633
    },
    {
      "epoch": 0.5780217038792566,
      "grad_norm": 0.0,
      "learning_rate": 7.973867794056469e-06,
      "loss": 0.9647,
      "step": 4634
    },
    {
      "epoch": 0.5781464388175128,
      "grad_norm": 0.0,
      "learning_rate": 7.969911642149172e-06,
      "loss": 0.8971,
      "step": 4635
    },
    {
      "epoch": 0.578271173755769,
      "grad_norm": 0.0,
      "learning_rate": 7.96595582160347e-06,
      "loss": 0.923,
      "step": 4636
    },
    {
      "epoch": 0.5783959086940252,
      "grad_norm": 0.0,
      "learning_rate": 7.962000333065058e-06,
      "loss": 1.0058,
      "step": 4637
    },
    {
      "epoch": 0.5785206436322814,
      "grad_norm": 0.0,
      "learning_rate": 7.95804517717957e-06,
      "loss": 0.918,
      "step": 4638
    },
    {
      "epoch": 0.5786453785705377,
      "grad_norm": 0.0,
      "learning_rate": 7.954090354592583e-06,
      "loss": 0.9627,
      "step": 4639
    },
    {
      "epoch": 0.5787701135087938,
      "grad_norm": 0.0,
      "learning_rate": 7.95013586594963e-06,
      "loss": 0.909,
      "step": 4640
    },
    {
      "epoch": 0.57889484844705,
      "grad_norm": 0.0,
      "learning_rate": 7.946181711896181e-06,
      "loss": 0.9402,
      "step": 4641
    },
    {
      "epoch": 0.5790195833853062,
      "grad_norm": 0.0,
      "learning_rate": 7.942227893077652e-06,
      "loss": 0.9594,
      "step": 4642
    },
    {
      "epoch": 0.5791443183235624,
      "grad_norm": 0.0,
      "learning_rate": 7.938274410139406e-06,
      "loss": 0.9242,
      "step": 4643
    },
    {
      "epoch": 0.5792690532618187,
      "grad_norm": 0.0,
      "learning_rate": 7.934321263726754e-06,
      "loss": 0.8933,
      "step": 4644
    },
    {
      "epoch": 0.5793937882000748,
      "grad_norm": 0.0,
      "learning_rate": 7.930368454484949e-06,
      "loss": 0.9271,
      "step": 4645
    },
    {
      "epoch": 0.5795185231383311,
      "grad_norm": 0.0,
      "learning_rate": 7.926415983059181e-06,
      "loss": 0.9599,
      "step": 4646
    },
    {
      "epoch": 0.5796432580765872,
      "grad_norm": 0.0,
      "learning_rate": 7.922463850094607e-06,
      "loss": 0.9298,
      "step": 4647
    },
    {
      "epoch": 0.5797679930148435,
      "grad_norm": 0.0,
      "learning_rate": 7.918512056236301e-06,
      "loss": 0.9141,
      "step": 4648
    },
    {
      "epoch": 0.5798927279530997,
      "grad_norm": 0.0,
      "learning_rate": 7.914560602129304e-06,
      "loss": 0.9201,
      "step": 4649
    },
    {
      "epoch": 0.5800174628913559,
      "grad_norm": 0.0,
      "learning_rate": 7.91060948841859e-06,
      "loss": 0.9645,
      "step": 4650
    },
    {
      "epoch": 0.5801421978296121,
      "grad_norm": 0.0,
      "learning_rate": 7.90665871574908e-06,
      "loss": 0.9375,
      "step": 4651
    },
    {
      "epoch": 0.5802669327678682,
      "grad_norm": 0.0,
      "learning_rate": 7.902708284765643e-06,
      "loss": 0.9472,
      "step": 4652
    },
    {
      "epoch": 0.5803916677061245,
      "grad_norm": 0.0,
      "learning_rate": 7.898758196113084e-06,
      "loss": 0.9759,
      "step": 4653
    },
    {
      "epoch": 0.5805164026443806,
      "grad_norm": 0.0,
      "learning_rate": 7.89480845043616e-06,
      "loss": 1.0059,
      "step": 4654
    },
    {
      "epoch": 0.5806411375826369,
      "grad_norm": 0.0,
      "learning_rate": 7.89085904837957e-06,
      "loss": 0.9356,
      "step": 4655
    },
    {
      "epoch": 0.5807658725208931,
      "grad_norm": 0.0,
      "learning_rate": 7.886909990587949e-06,
      "loss": 0.9149,
      "step": 4656
    },
    {
      "epoch": 0.5808906074591493,
      "grad_norm": 0.0,
      "learning_rate": 7.882961277705897e-06,
      "loss": 0.933,
      "step": 4657
    },
    {
      "epoch": 0.5810153423974055,
      "grad_norm": 0.0,
      "learning_rate": 7.879012910377925e-06,
      "loss": 0.9452,
      "step": 4658
    },
    {
      "epoch": 0.5811400773356618,
      "grad_norm": 0.0,
      "learning_rate": 7.875064889248522e-06,
      "loss": 1.0153,
      "step": 4659
    },
    {
      "epoch": 0.5812648122739179,
      "grad_norm": 0.0,
      "learning_rate": 7.871117214962096e-06,
      "loss": 0.9634,
      "step": 4660
    },
    {
      "epoch": 0.5813895472121742,
      "grad_norm": 0.0,
      "learning_rate": 7.86716988816301e-06,
      "loss": 0.9606,
      "step": 4661
    },
    {
      "epoch": 0.5815142821504303,
      "grad_norm": 0.0,
      "learning_rate": 7.863222909495569e-06,
      "loss": 0.8842,
      "step": 4662
    },
    {
      "epoch": 0.5816390170886866,
      "grad_norm": 0.0,
      "learning_rate": 7.859276279604014e-06,
      "loss": 0.9313,
      "step": 4663
    },
    {
      "epoch": 0.5817637520269427,
      "grad_norm": 0.0,
      "learning_rate": 7.85532999913254e-06,
      "loss": 0.9228,
      "step": 4664
    },
    {
      "epoch": 0.5818884869651989,
      "grad_norm": 0.0,
      "learning_rate": 7.851384068725272e-06,
      "loss": 0.9249,
      "step": 4665
    },
    {
      "epoch": 0.5820132219034552,
      "grad_norm": 0.0,
      "learning_rate": 7.847438489026296e-06,
      "loss": 0.9179,
      "step": 4666
    },
    {
      "epoch": 0.5821379568417113,
      "grad_norm": 0.0,
      "learning_rate": 7.843493260679624e-06,
      "loss": 0.9277,
      "step": 4667
    },
    {
      "epoch": 0.5822626917799676,
      "grad_norm": 0.0,
      "learning_rate": 7.839548384329212e-06,
      "loss": 0.9675,
      "step": 4668
    },
    {
      "epoch": 0.5823874267182237,
      "grad_norm": 0.0,
      "learning_rate": 7.835603860618973e-06,
      "loss": 0.9612,
      "step": 4669
    },
    {
      "epoch": 0.58251216165648,
      "grad_norm": 0.0,
      "learning_rate": 7.831659690192745e-06,
      "loss": 0.9259,
      "step": 4670
    },
    {
      "epoch": 0.5826368965947362,
      "grad_norm": 0.0,
      "learning_rate": 7.827715873694321e-06,
      "loss": 0.9353,
      "step": 4671
    },
    {
      "epoch": 0.5827616315329924,
      "grad_norm": 0.0,
      "learning_rate": 7.823772411767432e-06,
      "loss": 0.9581,
      "step": 4672
    },
    {
      "epoch": 0.5828863664712486,
      "grad_norm": 0.0,
      "learning_rate": 7.819829305055743e-06,
      "loss": 0.9377,
      "step": 4673
    },
    {
      "epoch": 0.5830111014095049,
      "grad_norm": 0.0,
      "learning_rate": 7.815886554202876e-06,
      "loss": 0.9015,
      "step": 4674
    },
    {
      "epoch": 0.583135836347761,
      "grad_norm": 0.0,
      "learning_rate": 7.81194415985238e-06,
      "loss": 0.9159,
      "step": 4675
    },
    {
      "epoch": 0.5832605712860172,
      "grad_norm": 0.0,
      "learning_rate": 7.808002122647763e-06,
      "loss": 0.9816,
      "step": 4676
    },
    {
      "epoch": 0.5833853062242734,
      "grad_norm": 0.0,
      "learning_rate": 7.804060443232456e-06,
      "loss": 0.9556,
      "step": 4677
    },
    {
      "epoch": 0.5835100411625296,
      "grad_norm": 0.0,
      "learning_rate": 7.800119122249847e-06,
      "loss": 0.9114,
      "step": 4678
    },
    {
      "epoch": 0.5836347761007858,
      "grad_norm": 0.0,
      "learning_rate": 7.796178160343255e-06,
      "loss": 0.9775,
      "step": 4679
    },
    {
      "epoch": 0.583759511039042,
      "grad_norm": 0.0,
      "learning_rate": 7.792237558155944e-06,
      "loss": 0.9019,
      "step": 4680
    },
    {
      "epoch": 0.5838842459772983,
      "grad_norm": 0.0,
      "learning_rate": 7.788297316331123e-06,
      "loss": 0.9315,
      "step": 4681
    },
    {
      "epoch": 0.5840089809155544,
      "grad_norm": 0.0,
      "learning_rate": 7.784357435511933e-06,
      "loss": 0.9299,
      "step": 4682
    },
    {
      "epoch": 0.5841337158538107,
      "grad_norm": 0.0,
      "learning_rate": 7.780417916341471e-06,
      "loss": 0.9309,
      "step": 4683
    },
    {
      "epoch": 0.5842584507920668,
      "grad_norm": 0.0,
      "learning_rate": 7.77647875946276e-06,
      "loss": 0.9086,
      "step": 4684
    },
    {
      "epoch": 0.5843831857303231,
      "grad_norm": 0.0,
      "learning_rate": 7.772539965518764e-06,
      "loss": 0.8944,
      "step": 4685
    },
    {
      "epoch": 0.5845079206685793,
      "grad_norm": 0.0,
      "learning_rate": 7.768601535152404e-06,
      "loss": 0.9276,
      "step": 4686
    },
    {
      "epoch": 0.5846326556068355,
      "grad_norm": 0.0,
      "learning_rate": 7.764663469006526e-06,
      "loss": 0.9469,
      "step": 4687
    },
    {
      "epoch": 0.5847573905450917,
      "grad_norm": 0.0,
      "learning_rate": 7.760725767723923e-06,
      "loss": 0.9541,
      "step": 4688
    },
    {
      "epoch": 0.5848821254833478,
      "grad_norm": 0.0,
      "learning_rate": 7.756788431947327e-06,
      "loss": 0.9893,
      "step": 4689
    },
    {
      "epoch": 0.5850068604216041,
      "grad_norm": 0.0,
      "learning_rate": 7.752851462319408e-06,
      "loss": 0.9395,
      "step": 4690
    },
    {
      "epoch": 0.5851315953598603,
      "grad_norm": 0.0,
      "learning_rate": 7.748914859482781e-06,
      "loss": 0.9483,
      "step": 4691
    },
    {
      "epoch": 0.5852563302981165,
      "grad_norm": 0.0,
      "learning_rate": 7.744978624079995e-06,
      "loss": 0.9725,
      "step": 4692
    },
    {
      "epoch": 0.5853810652363727,
      "grad_norm": 0.0,
      "learning_rate": 7.741042756753554e-06,
      "loss": 0.9644,
      "step": 4693
    },
    {
      "epoch": 0.5855058001746289,
      "grad_norm": 0.0,
      "learning_rate": 7.737107258145878e-06,
      "loss": 0.9289,
      "step": 4694
    },
    {
      "epoch": 0.5856305351128851,
      "grad_norm": 0.0,
      "learning_rate": 7.733172128899345e-06,
      "loss": 0.929,
      "step": 4695
    },
    {
      "epoch": 0.5857552700511414,
      "grad_norm": 0.0,
      "learning_rate": 7.72923736965627e-06,
      "loss": 0.9928,
      "step": 4696
    },
    {
      "epoch": 0.5858800049893975,
      "grad_norm": 0.0,
      "learning_rate": 7.725302981058899e-06,
      "loss": 0.9213,
      "step": 4697
    },
    {
      "epoch": 0.5860047399276538,
      "grad_norm": 0.0,
      "learning_rate": 7.721368963749426e-06,
      "loss": 0.9271,
      "step": 4698
    },
    {
      "epoch": 0.5861294748659099,
      "grad_norm": 0.0,
      "learning_rate": 7.717435318369984e-06,
      "loss": 0.9005,
      "step": 4699
    },
    {
      "epoch": 0.5862542098041661,
      "grad_norm": 0.0,
      "learning_rate": 7.713502045562641e-06,
      "loss": 0.8722,
      "step": 4700
    },
    {
      "epoch": 0.5863789447424224,
      "grad_norm": 0.0,
      "learning_rate": 7.709569145969406e-06,
      "loss": 0.9903,
      "step": 4701
    },
    {
      "epoch": 0.5865036796806785,
      "grad_norm": 0.0,
      "learning_rate": 7.705636620232225e-06,
      "loss": 0.9416,
      "step": 4702
    },
    {
      "epoch": 0.5866284146189348,
      "grad_norm": 0.0,
      "learning_rate": 7.701704468992994e-06,
      "loss": 0.8964,
      "step": 4703
    },
    {
      "epoch": 0.5867531495571909,
      "grad_norm": 0.0,
      "learning_rate": 7.697772692893526e-06,
      "loss": 0.9391,
      "step": 4704
    },
    {
      "epoch": 0.5868778844954472,
      "grad_norm": 0.0,
      "learning_rate": 7.6938412925756e-06,
      "loss": 0.9144,
      "step": 4705
    },
    {
      "epoch": 0.5870026194337034,
      "grad_norm": 0.0,
      "learning_rate": 7.68991026868091e-06,
      "loss": 0.9476,
      "step": 4706
    },
    {
      "epoch": 0.5871273543719596,
      "grad_norm": 0.0,
      "learning_rate": 7.685979621851102e-06,
      "loss": 0.9307,
      "step": 4707
    },
    {
      "epoch": 0.5872520893102158,
      "grad_norm": 0.0,
      "learning_rate": 7.682049352727756e-06,
      "loss": 0.9625,
      "step": 4708
    },
    {
      "epoch": 0.587376824248472,
      "grad_norm": 0.0,
      "learning_rate": 7.67811946195239e-06,
      "loss": 0.9771,
      "step": 4709
    },
    {
      "epoch": 0.5875015591867282,
      "grad_norm": 0.0,
      "learning_rate": 7.674189950166463e-06,
      "loss": 0.9542,
      "step": 4710
    },
    {
      "epoch": 0.5876262941249845,
      "grad_norm": 0.0,
      "learning_rate": 7.670260818011365e-06,
      "loss": 0.9627,
      "step": 4711
    },
    {
      "epoch": 0.5877510290632406,
      "grad_norm": 0.0,
      "learning_rate": 7.666332066128439e-06,
      "loss": 0.995,
      "step": 4712
    },
    {
      "epoch": 0.5878757640014968,
      "grad_norm": 0.0,
      "learning_rate": 7.662403695158953e-06,
      "loss": 0.9042,
      "step": 4713
    },
    {
      "epoch": 0.588000498939753,
      "grad_norm": 0.0,
      "learning_rate": 7.658475705744109e-06,
      "loss": 0.9011,
      "step": 4714
    },
    {
      "epoch": 0.5881252338780092,
      "grad_norm": 0.0,
      "learning_rate": 7.654548098525062e-06,
      "loss": 0.9518,
      "step": 4715
    },
    {
      "epoch": 0.5882499688162655,
      "grad_norm": 0.0,
      "learning_rate": 7.650620874142893e-06,
      "loss": 0.9821,
      "step": 4716
    },
    {
      "epoch": 0.5883747037545216,
      "grad_norm": 0.0,
      "learning_rate": 7.646694033238626e-06,
      "loss": 0.9489,
      "step": 4717
    },
    {
      "epoch": 0.5884994386927779,
      "grad_norm": 0.0,
      "learning_rate": 7.64276757645322e-06,
      "loss": 0.9408,
      "step": 4718
    },
    {
      "epoch": 0.588624173631034,
      "grad_norm": 0.0,
      "learning_rate": 7.638841504427567e-06,
      "loss": 0.9407,
      "step": 4719
    },
    {
      "epoch": 0.5887489085692903,
      "grad_norm": 0.0,
      "learning_rate": 7.634915817802509e-06,
      "loss": 0.9064,
      "step": 4720
    },
    {
      "epoch": 0.5888736435075465,
      "grad_norm": 0.0,
      "learning_rate": 7.630990517218809e-06,
      "loss": 0.9285,
      "step": 4721
    },
    {
      "epoch": 0.5889983784458027,
      "grad_norm": 0.0,
      "learning_rate": 7.627065603317183e-06,
      "loss": 0.9517,
      "step": 4722
    },
    {
      "epoch": 0.5891231133840589,
      "grad_norm": 0.0,
      "learning_rate": 7.623141076738271e-06,
      "loss": 0.9319,
      "step": 4723
    },
    {
      "epoch": 0.589247848322315,
      "grad_norm": 0.0,
      "learning_rate": 7.619216938122653e-06,
      "loss": 0.9037,
      "step": 4724
    },
    {
      "epoch": 0.5893725832605713,
      "grad_norm": 0.0,
      "learning_rate": 7.615293188110854e-06,
      "loss": 0.9823,
      "step": 4725
    },
    {
      "epoch": 0.5894973181988274,
      "grad_norm": 0.0,
      "learning_rate": 7.6113698273433215e-06,
      "loss": 0.9251,
      "step": 4726
    },
    {
      "epoch": 0.5896220531370837,
      "grad_norm": 0.0,
      "learning_rate": 7.607446856460452e-06,
      "loss": 0.9244,
      "step": 4727
    },
    {
      "epoch": 0.5897467880753399,
      "grad_norm": 0.0,
      "learning_rate": 7.603524276102568e-06,
      "loss": 0.9536,
      "step": 4728
    },
    {
      "epoch": 0.5898715230135961,
      "grad_norm": 0.0,
      "learning_rate": 7.599602086909943e-06,
      "loss": 0.9361,
      "step": 4729
    },
    {
      "epoch": 0.5899962579518523,
      "grad_norm": 0.0,
      "learning_rate": 7.595680289522769e-06,
      "loss": 0.9646,
      "step": 4730
    },
    {
      "epoch": 0.5901209928901086,
      "grad_norm": 0.0,
      "learning_rate": 7.591758884581179e-06,
      "loss": 0.962,
      "step": 4731
    },
    {
      "epoch": 0.5902457278283647,
      "grad_norm": 0.0,
      "learning_rate": 7.5878378727252565e-06,
      "loss": 0.8979,
      "step": 4732
    },
    {
      "epoch": 0.590370462766621,
      "grad_norm": 0.0,
      "learning_rate": 7.5839172545950005e-06,
      "loss": 0.914,
      "step": 4733
    },
    {
      "epoch": 0.5904951977048771,
      "grad_norm": 0.0,
      "learning_rate": 7.57999703083036e-06,
      "loss": 0.9272,
      "step": 4734
    },
    {
      "epoch": 0.5906199326431334,
      "grad_norm": 0.0,
      "learning_rate": 7.576077202071211e-06,
      "loss": 0.9396,
      "step": 4735
    },
    {
      "epoch": 0.5907446675813895,
      "grad_norm": 0.0,
      "learning_rate": 7.5721577689573685e-06,
      "loss": 0.9641,
      "step": 4736
    },
    {
      "epoch": 0.5908694025196457,
      "grad_norm": 0.0,
      "learning_rate": 7.568238732128586e-06,
      "loss": 0.9195,
      "step": 4737
    },
    {
      "epoch": 0.590994137457902,
      "grad_norm": 0.0,
      "learning_rate": 7.564320092224542e-06,
      "loss": 0.9502,
      "step": 4738
    },
    {
      "epoch": 0.5911188723961581,
      "grad_norm": 0.0,
      "learning_rate": 7.5604018498848656e-06,
      "loss": 0.924,
      "step": 4739
    },
    {
      "epoch": 0.5912436073344144,
      "grad_norm": 0.0,
      "learning_rate": 7.556484005749112e-06,
      "loss": 0.9665,
      "step": 4740
    },
    {
      "epoch": 0.5913683422726705,
      "grad_norm": 0.0,
      "learning_rate": 7.552566560456762e-06,
      "loss": 0.9344,
      "step": 4741
    },
    {
      "epoch": 0.5914930772109268,
      "grad_norm": 0.0,
      "learning_rate": 7.548649514647255e-06,
      "loss": 0.8884,
      "step": 4742
    },
    {
      "epoch": 0.591617812149183,
      "grad_norm": 0.0,
      "learning_rate": 7.544732868959942e-06,
      "loss": 0.9017,
      "step": 4743
    },
    {
      "epoch": 0.5917425470874392,
      "grad_norm": 0.0,
      "learning_rate": 7.5408166240341244e-06,
      "loss": 0.9187,
      "step": 4744
    },
    {
      "epoch": 0.5918672820256954,
      "grad_norm": 0.0,
      "learning_rate": 7.536900780509029e-06,
      "loss": 0.9477,
      "step": 4745
    },
    {
      "epoch": 0.5919920169639517,
      "grad_norm": 0.0,
      "learning_rate": 7.5329853390238186e-06,
      "loss": 0.9121,
      "step": 4746
    },
    {
      "epoch": 0.5921167519022078,
      "grad_norm": 0.0,
      "learning_rate": 7.529070300217596e-06,
      "loss": 0.9281,
      "step": 4747
    },
    {
      "epoch": 0.592241486840464,
      "grad_norm": 0.0,
      "learning_rate": 7.525155664729387e-06,
      "loss": 0.8883,
      "step": 4748
    },
    {
      "epoch": 0.5923662217787202,
      "grad_norm": 0.0,
      "learning_rate": 7.52124143319817e-06,
      "loss": 0.9194,
      "step": 4749
    },
    {
      "epoch": 0.5924909567169764,
      "grad_norm": 0.0,
      "learning_rate": 7.5173276062628364e-06,
      "loss": 0.9566,
      "step": 4750
    },
    {
      "epoch": 0.5926156916552326,
      "grad_norm": 0.0,
      "learning_rate": 7.513414184562229e-06,
      "loss": 0.961,
      "step": 4751
    },
    {
      "epoch": 0.5927404265934888,
      "grad_norm": 0.0,
      "learning_rate": 7.509501168735113e-06,
      "loss": 0.944,
      "step": 4752
    },
    {
      "epoch": 0.5928651615317451,
      "grad_norm": 0.0,
      "learning_rate": 7.505588559420188e-06,
      "loss": 0.9604,
      "step": 4753
    },
    {
      "epoch": 0.5929898964700012,
      "grad_norm": 0.0,
      "learning_rate": 7.501676357256097e-06,
      "loss": 0.9708,
      "step": 4754
    },
    {
      "epoch": 0.5931146314082575,
      "grad_norm": 0.0,
      "learning_rate": 7.497764562881404e-06,
      "loss": 0.9405,
      "step": 4755
    },
    {
      "epoch": 0.5932393663465136,
      "grad_norm": 0.0,
      "learning_rate": 7.493853176934618e-06,
      "loss": 0.9511,
      "step": 4756
    },
    {
      "epoch": 0.5933641012847699,
      "grad_norm": 0.0,
      "learning_rate": 7.4899422000541735e-06,
      "loss": 0.9095,
      "step": 4757
    },
    {
      "epoch": 0.5934888362230261,
      "grad_norm": 0.0,
      "learning_rate": 7.4860316328784345e-06,
      "loss": 0.9567,
      "step": 4758
    },
    {
      "epoch": 0.5936135711612823,
      "grad_norm": 0.0,
      "learning_rate": 7.482121476045716e-06,
      "loss": 0.9022,
      "step": 4759
    },
    {
      "epoch": 0.5937383060995385,
      "grad_norm": 0.0,
      "learning_rate": 7.478211730194245e-06,
      "loss": 0.9579,
      "step": 4760
    },
    {
      "epoch": 0.5938630410377946,
      "grad_norm": 0.0,
      "learning_rate": 7.474302395962196e-06,
      "loss": 0.9282,
      "step": 4761
    },
    {
      "epoch": 0.5939877759760509,
      "grad_norm": 0.0,
      "learning_rate": 7.4703934739876675e-06,
      "loss": 0.9562,
      "step": 4762
    },
    {
      "epoch": 0.5941125109143071,
      "grad_norm": 0.0,
      "learning_rate": 7.466484964908693e-06,
      "loss": 0.9097,
      "step": 4763
    },
    {
      "epoch": 0.5942372458525633,
      "grad_norm": 0.0,
      "learning_rate": 7.462576869363245e-06,
      "loss": 0.9525,
      "step": 4764
    },
    {
      "epoch": 0.5943619807908195,
      "grad_norm": 0.0,
      "learning_rate": 7.458669187989214e-06,
      "loss": 0.9353,
      "step": 4765
    },
    {
      "epoch": 0.5944867157290757,
      "grad_norm": 0.0,
      "learning_rate": 7.454761921424446e-06,
      "loss": 0.9091,
      "step": 4766
    },
    {
      "epoch": 0.5946114506673319,
      "grad_norm": 0.0,
      "learning_rate": 7.45085507030669e-06,
      "loss": 0.9644,
      "step": 4767
    },
    {
      "epoch": 0.5947361856055882,
      "grad_norm": 0.0,
      "learning_rate": 7.446948635273655e-06,
      "loss": 0.9305,
      "step": 4768
    },
    {
      "epoch": 0.5948609205438443,
      "grad_norm": 0.0,
      "learning_rate": 7.443042616962966e-06,
      "loss": 0.9682,
      "step": 4769
    },
    {
      "epoch": 0.5949856554821006,
      "grad_norm": 0.0,
      "learning_rate": 7.439137016012179e-06,
      "loss": 0.9628,
      "step": 4770
    },
    {
      "epoch": 0.5951103904203567,
      "grad_norm": 0.0,
      "learning_rate": 7.435231833058791e-06,
      "loss": 0.907,
      "step": 4771
    },
    {
      "epoch": 0.5952351253586129,
      "grad_norm": 0.0,
      "learning_rate": 7.431327068740225e-06,
      "loss": 0.9153,
      "step": 4772
    },
    {
      "epoch": 0.5953598602968692,
      "grad_norm": 0.0,
      "learning_rate": 7.427422723693839e-06,
      "loss": 0.9276,
      "step": 4773
    },
    {
      "epoch": 0.5954845952351253,
      "grad_norm": 0.0,
      "learning_rate": 7.423518798556919e-06,
      "loss": 0.9286,
      "step": 4774
    },
    {
      "epoch": 0.5956093301733816,
      "grad_norm": 0.0,
      "learning_rate": 7.4196152939666784e-06,
      "loss": 0.9413,
      "step": 4775
    },
    {
      "epoch": 0.5957340651116377,
      "grad_norm": 0.0,
      "learning_rate": 7.415712210560279e-06,
      "loss": 0.9484,
      "step": 4776
    },
    {
      "epoch": 0.595858800049894,
      "grad_norm": 0.0,
      "learning_rate": 7.411809548974792e-06,
      "loss": 0.9648,
      "step": 4777
    },
    {
      "epoch": 0.5959835349881502,
      "grad_norm": 0.0,
      "learning_rate": 7.407907309847238e-06,
      "loss": 0.8932,
      "step": 4778
    },
    {
      "epoch": 0.5961082699264064,
      "grad_norm": 0.0,
      "learning_rate": 7.404005493814557e-06,
      "loss": 0.9511,
      "step": 4779
    },
    {
      "epoch": 0.5962330048646626,
      "grad_norm": 0.0,
      "learning_rate": 7.400104101513622e-06,
      "loss": 0.9646,
      "step": 4780
    },
    {
      "epoch": 0.5963577398029188,
      "grad_norm": 0.0,
      "learning_rate": 7.396203133581243e-06,
      "loss": 0.9328,
      "step": 4781
    },
    {
      "epoch": 0.596482474741175,
      "grad_norm": 0.0,
      "learning_rate": 7.392302590654152e-06,
      "loss": 0.9681,
      "step": 4782
    },
    {
      "epoch": 0.5966072096794313,
      "grad_norm": 0.0,
      "learning_rate": 7.3884024733690195e-06,
      "loss": 0.9561,
      "step": 4783
    },
    {
      "epoch": 0.5967319446176874,
      "grad_norm": 0.0,
      "learning_rate": 7.384502782362437e-06,
      "loss": 0.9627,
      "step": 4784
    },
    {
      "epoch": 0.5968566795559436,
      "grad_norm": 0.0,
      "learning_rate": 7.380603518270942e-06,
      "loss": 0.9849,
      "step": 4785
    },
    {
      "epoch": 0.5969814144941998,
      "grad_norm": 0.0,
      "learning_rate": 7.376704681730988e-06,
      "loss": 0.9413,
      "step": 4786
    },
    {
      "epoch": 0.597106149432456,
      "grad_norm": 0.0,
      "learning_rate": 7.372806273378958e-06,
      "loss": 0.9,
      "step": 4787
    },
    {
      "epoch": 0.5972308843707123,
      "grad_norm": 0.0,
      "learning_rate": 7.368908293851178e-06,
      "loss": 0.8921,
      "step": 4788
    },
    {
      "epoch": 0.5973556193089684,
      "grad_norm": 0.0,
      "learning_rate": 7.365010743783894e-06,
      "loss": 0.9255,
      "step": 4789
    },
    {
      "epoch": 0.5974803542472247,
      "grad_norm": 0.0,
      "learning_rate": 7.361113623813284e-06,
      "loss": 0.9246,
      "step": 4790
    },
    {
      "epoch": 0.5976050891854808,
      "grad_norm": 0.0,
      "learning_rate": 7.3572169345754575e-06,
      "loss": 0.9047,
      "step": 4791
    },
    {
      "epoch": 0.5977298241237371,
      "grad_norm": 0.0,
      "learning_rate": 7.3533206767064505e-06,
      "loss": 0.9936,
      "step": 4792
    },
    {
      "epoch": 0.5978545590619933,
      "grad_norm": 0.0,
      "learning_rate": 7.3494248508422316e-06,
      "loss": 0.9209,
      "step": 4793
    },
    {
      "epoch": 0.5979792940002495,
      "grad_norm": 0.0,
      "learning_rate": 7.345529457618694e-06,
      "loss": 0.9068,
      "step": 4794
    },
    {
      "epoch": 0.5981040289385057,
      "grad_norm": 0.0,
      "learning_rate": 7.341634497671673e-06,
      "loss": 0.936,
      "step": 4795
    },
    {
      "epoch": 0.5982287638767618,
      "grad_norm": 0.0,
      "learning_rate": 7.33773997163692e-06,
      "loss": 0.9284,
      "step": 4796
    },
    {
      "epoch": 0.5983534988150181,
      "grad_norm": 0.0,
      "learning_rate": 7.333845880150112e-06,
      "loss": 0.9727,
      "step": 4797
    },
    {
      "epoch": 0.5984782337532742,
      "grad_norm": 0.0,
      "learning_rate": 7.3299522238468755e-06,
      "loss": 0.9345,
      "step": 4798
    },
    {
      "epoch": 0.5986029686915305,
      "grad_norm": 0.0,
      "learning_rate": 7.326059003362744e-06,
      "loss": 0.911,
      "step": 4799
    },
    {
      "epoch": 0.5987277036297867,
      "grad_norm": 0.0,
      "learning_rate": 7.322166219333197e-06,
      "loss": 0.974,
      "step": 4800
    },
    {
      "epoch": 0.5988524385680429,
      "grad_norm": 0.0,
      "learning_rate": 7.3182738723936255e-06,
      "loss": 0.9205,
      "step": 4801
    },
    {
      "epoch": 0.5989771735062991,
      "grad_norm": 0.0,
      "learning_rate": 7.31438196317937e-06,
      "loss": 0.9153,
      "step": 4802
    },
    {
      "epoch": 0.5991019084445554,
      "grad_norm": 0.0,
      "learning_rate": 7.31049049232568e-06,
      "loss": 0.9929,
      "step": 4803
    },
    {
      "epoch": 0.5992266433828115,
      "grad_norm": 0.0,
      "learning_rate": 7.306599460467741e-06,
      "loss": 0.8853,
      "step": 4804
    },
    {
      "epoch": 0.5993513783210678,
      "grad_norm": 0.0,
      "learning_rate": 7.3027088682406725e-06,
      "loss": 0.958,
      "step": 4805
    },
    {
      "epoch": 0.5994761132593239,
      "grad_norm": 0.0,
      "learning_rate": 7.298818716279513e-06,
      "loss": 0.9751,
      "step": 4806
    },
    {
      "epoch": 0.5996008481975802,
      "grad_norm": 0.0,
      "learning_rate": 7.294929005219238e-06,
      "loss": 0.9006,
      "step": 4807
    },
    {
      "epoch": 0.5997255831358363,
      "grad_norm": 0.0,
      "learning_rate": 7.291039735694745e-06,
      "loss": 0.8898,
      "step": 4808
    },
    {
      "epoch": 0.5998503180740925,
      "grad_norm": 0.0,
      "learning_rate": 7.287150908340857e-06,
      "loss": 0.9368,
      "step": 4809
    },
    {
      "epoch": 0.5999750530123488,
      "grad_norm": 0.0,
      "learning_rate": 7.2832625237923315e-06,
      "loss": 0.9734,
      "step": 4810
    },
    {
      "epoch": 0.6000997879506049,
      "grad_norm": 0.0,
      "learning_rate": 7.279374582683849e-06,
      "loss": 0.9431,
      "step": 4811
    },
    {
      "epoch": 0.6002245228888612,
      "grad_norm": 0.0,
      "learning_rate": 7.275487085650026e-06,
      "loss": 0.9536,
      "step": 4812
    },
    {
      "epoch": 0.6003492578271173,
      "grad_norm": 0.0,
      "learning_rate": 7.271600033325393e-06,
      "loss": 0.9155,
      "step": 4813
    },
    {
      "epoch": 0.6004739927653736,
      "grad_norm": 0.0,
      "learning_rate": 7.267713426344414e-06,
      "loss": 0.9114,
      "step": 4814
    },
    {
      "epoch": 0.6005987277036298,
      "grad_norm": 0.0,
      "learning_rate": 7.263827265341488e-06,
      "loss": 0.9632,
      "step": 4815
    },
    {
      "epoch": 0.600723462641886,
      "grad_norm": 0.0,
      "learning_rate": 7.259941550950928e-06,
      "loss": 0.9554,
      "step": 4816
    },
    {
      "epoch": 0.6008481975801422,
      "grad_norm": 0.0,
      "learning_rate": 7.256056283806987e-06,
      "loss": 0.9105,
      "step": 4817
    },
    {
      "epoch": 0.6009729325183985,
      "grad_norm": 0.0,
      "learning_rate": 7.252171464543831e-06,
      "loss": 0.9427,
      "step": 4818
    },
    {
      "epoch": 0.6010976674566546,
      "grad_norm": 0.0,
      "learning_rate": 7.248287093795568e-06,
      "loss": 0.9522,
      "step": 4819
    },
    {
      "epoch": 0.6012224023949108,
      "grad_norm": 0.0,
      "learning_rate": 7.244403172196222e-06,
      "loss": 0.9322,
      "step": 4820
    },
    {
      "epoch": 0.601347137333167,
      "grad_norm": 0.0,
      "learning_rate": 7.24051970037974e-06,
      "loss": 0.9279,
      "step": 4821
    },
    {
      "epoch": 0.6014718722714232,
      "grad_norm": 0.0,
      "learning_rate": 7.236636678980018e-06,
      "loss": 0.9295,
      "step": 4822
    },
    {
      "epoch": 0.6015966072096794,
      "grad_norm": 0.0,
      "learning_rate": 7.232754108630846e-06,
      "loss": 0.9248,
      "step": 4823
    },
    {
      "epoch": 0.6017213421479356,
      "grad_norm": 0.0,
      "learning_rate": 7.2288719899659685e-06,
      "loss": 0.948,
      "step": 4824
    },
    {
      "epoch": 0.6018460770861919,
      "grad_norm": 0.0,
      "learning_rate": 7.224990323619045e-06,
      "loss": 0.9259,
      "step": 4825
    },
    {
      "epoch": 0.601970812024448,
      "grad_norm": 0.0,
      "learning_rate": 7.221109110223654e-06,
      "loss": 0.9213,
      "step": 4826
    },
    {
      "epoch": 0.6020955469627043,
      "grad_norm": 0.0,
      "learning_rate": 7.217228350413315e-06,
      "loss": 0.8995,
      "step": 4827
    },
    {
      "epoch": 0.6022202819009604,
      "grad_norm": 0.0,
      "learning_rate": 7.213348044821459e-06,
      "loss": 0.9426,
      "step": 4828
    },
    {
      "epoch": 0.6023450168392167,
      "grad_norm": 0.0,
      "learning_rate": 7.2094681940814556e-06,
      "loss": 0.8974,
      "step": 4829
    },
    {
      "epoch": 0.6024697517774729,
      "grad_norm": 0.0,
      "learning_rate": 7.205588798826593e-06,
      "loss": 0.8937,
      "step": 4830
    },
    {
      "epoch": 0.6025944867157291,
      "grad_norm": 0.0,
      "learning_rate": 7.201709859690081e-06,
      "loss": 0.9783,
      "step": 4831
    },
    {
      "epoch": 0.6027192216539853,
      "grad_norm": 0.0,
      "learning_rate": 7.197831377305069e-06,
      "loss": 0.8707,
      "step": 4832
    },
    {
      "epoch": 0.6028439565922414,
      "grad_norm": 0.0,
      "learning_rate": 7.193953352304612e-06,
      "loss": 0.9677,
      "step": 4833
    },
    {
      "epoch": 0.6029686915304977,
      "grad_norm": 0.0,
      "learning_rate": 7.190075785321713e-06,
      "loss": 0.9711,
      "step": 4834
    },
    {
      "epoch": 0.6030934264687539,
      "grad_norm": 0.0,
      "learning_rate": 7.186198676989283e-06,
      "loss": 0.9465,
      "step": 4835
    },
    {
      "epoch": 0.6032181614070101,
      "grad_norm": 0.0,
      "learning_rate": 7.182322027940165e-06,
      "loss": 0.9211,
      "step": 4836
    },
    {
      "epoch": 0.6033428963452663,
      "grad_norm": 0.0,
      "learning_rate": 7.178445838807124e-06,
      "loss": 0.9605,
      "step": 4837
    },
    {
      "epoch": 0.6034676312835225,
      "grad_norm": 0.0,
      "learning_rate": 7.174570110222854e-06,
      "loss": 0.9639,
      "step": 4838
    },
    {
      "epoch": 0.6035923662217787,
      "grad_norm": 0.0,
      "learning_rate": 7.170694842819971e-06,
      "loss": 0.9258,
      "step": 4839
    },
    {
      "epoch": 0.603717101160035,
      "grad_norm": 0.0,
      "learning_rate": 7.1668200372310124e-06,
      "loss": 0.9354,
      "step": 4840
    },
    {
      "epoch": 0.6038418360982911,
      "grad_norm": 0.0,
      "learning_rate": 7.162945694088451e-06,
      "loss": 0.9293,
      "step": 4841
    },
    {
      "epoch": 0.6039665710365474,
      "grad_norm": 0.0,
      "learning_rate": 7.159071814024675e-06,
      "loss": 0.962,
      "step": 4842
    },
    {
      "epoch": 0.6040913059748035,
      "grad_norm": 0.0,
      "learning_rate": 7.155198397671997e-06,
      "loss": 0.9664,
      "step": 4843
    },
    {
      "epoch": 0.6042160409130597,
      "grad_norm": 0.0,
      "learning_rate": 7.151325445662658e-06,
      "loss": 0.9542,
      "step": 4844
    },
    {
      "epoch": 0.604340775851316,
      "grad_norm": 0.0,
      "learning_rate": 7.14745295862882e-06,
      "loss": 0.9042,
      "step": 4845
    },
    {
      "epoch": 0.6044655107895721,
      "grad_norm": 0.0,
      "learning_rate": 7.1435809372025725e-06,
      "loss": 0.9372,
      "step": 4846
    },
    {
      "epoch": 0.6045902457278284,
      "grad_norm": 0.0,
      "learning_rate": 7.139709382015927e-06,
      "loss": 0.9321,
      "step": 4847
    },
    {
      "epoch": 0.6047149806660845,
      "grad_norm": 0.0,
      "learning_rate": 7.135838293700814e-06,
      "loss": 0.9448,
      "step": 4848
    },
    {
      "epoch": 0.6048397156043408,
      "grad_norm": 0.0,
      "learning_rate": 7.131967672889101e-06,
      "loss": 0.9125,
      "step": 4849
    },
    {
      "epoch": 0.604964450542597,
      "grad_norm": 0.0,
      "learning_rate": 7.12809752021256e-06,
      "loss": 0.9559,
      "step": 4850
    },
    {
      "epoch": 0.6050891854808532,
      "grad_norm": 0.0,
      "learning_rate": 7.12422783630291e-06,
      "loss": 0.9519,
      "step": 4851
    },
    {
      "epoch": 0.6052139204191094,
      "grad_norm": 0.0,
      "learning_rate": 7.120358621791772e-06,
      "loss": 0.9341,
      "step": 4852
    },
    {
      "epoch": 0.6053386553573656,
      "grad_norm": 0.0,
      "learning_rate": 7.116489877310705e-06,
      "loss": 0.9655,
      "step": 4853
    },
    {
      "epoch": 0.6054633902956218,
      "grad_norm": 0.0,
      "learning_rate": 7.1126216034911826e-06,
      "loss": 0.9352,
      "step": 4854
    },
    {
      "epoch": 0.6055881252338781,
      "grad_norm": 0.0,
      "learning_rate": 7.1087538009646025e-06,
      "loss": 0.918,
      "step": 4855
    },
    {
      "epoch": 0.6057128601721342,
      "grad_norm": 0.0,
      "learning_rate": 7.104886470362292e-06,
      "loss": 0.8742,
      "step": 4856
    },
    {
      "epoch": 0.6058375951103904,
      "grad_norm": 0.0,
      "learning_rate": 7.101019612315492e-06,
      "loss": 0.9289,
      "step": 4857
    },
    {
      "epoch": 0.6059623300486466,
      "grad_norm": 0.0,
      "learning_rate": 7.097153227455379e-06,
      "loss": 0.9275,
      "step": 4858
    },
    {
      "epoch": 0.6060870649869028,
      "grad_norm": 0.0,
      "learning_rate": 7.0932873164130375e-06,
      "loss": 0.9231,
      "step": 4859
    },
    {
      "epoch": 0.6062117999251591,
      "grad_norm": 0.0,
      "learning_rate": 7.08942187981948e-06,
      "loss": 0.9262,
      "step": 4860
    },
    {
      "epoch": 0.6063365348634152,
      "grad_norm": 0.0,
      "learning_rate": 7.085556918305652e-06,
      "loss": 0.9575,
      "step": 4861
    },
    {
      "epoch": 0.6064612698016715,
      "grad_norm": 0.0,
      "learning_rate": 7.081692432502404e-06,
      "loss": 0.9789,
      "step": 4862
    },
    {
      "epoch": 0.6065860047399276,
      "grad_norm": 0.0,
      "learning_rate": 7.077828423040525e-06,
      "loss": 0.9399,
      "step": 4863
    },
    {
      "epoch": 0.6067107396781839,
      "grad_norm": 0.0,
      "learning_rate": 7.073964890550713e-06,
      "loss": 0.9227,
      "step": 4864
    },
    {
      "epoch": 0.60683547461644,
      "grad_norm": 0.0,
      "learning_rate": 7.070101835663594e-06,
      "loss": 0.8746,
      "step": 4865
    },
    {
      "epoch": 0.6069602095546963,
      "grad_norm": 0.0,
      "learning_rate": 7.066239259009722e-06,
      "loss": 0.9236,
      "step": 4866
    },
    {
      "epoch": 0.6070849444929525,
      "grad_norm": 0.0,
      "learning_rate": 7.062377161219556e-06,
      "loss": 0.9346,
      "step": 4867
    },
    {
      "epoch": 0.6072096794312086,
      "grad_norm": 0.0,
      "learning_rate": 7.058515542923499e-06,
      "loss": 0.9244,
      "step": 4868
    },
    {
      "epoch": 0.6073344143694649,
      "grad_norm": 0.0,
      "learning_rate": 7.054654404751858e-06,
      "loss": 0.893,
      "step": 4869
    },
    {
      "epoch": 0.607459149307721,
      "grad_norm": 0.0,
      "learning_rate": 7.050793747334872e-06,
      "loss": 0.9436,
      "step": 4870
    },
    {
      "epoch": 0.6075838842459773,
      "grad_norm": 0.0,
      "learning_rate": 7.046933571302697e-06,
      "loss": 0.9022,
      "step": 4871
    },
    {
      "epoch": 0.6077086191842335,
      "grad_norm": 0.0,
      "learning_rate": 7.043073877285405e-06,
      "loss": 0.8585,
      "step": 4872
    },
    {
      "epoch": 0.6078333541224897,
      "grad_norm": 0.0,
      "learning_rate": 7.039214665913004e-06,
      "loss": 0.9234,
      "step": 4873
    },
    {
      "epoch": 0.6079580890607459,
      "grad_norm": 0.0,
      "learning_rate": 7.035355937815407e-06,
      "loss": 0.893,
      "step": 4874
    },
    {
      "epoch": 0.6080828239990022,
      "grad_norm": 0.0,
      "learning_rate": 7.031497693622461e-06,
      "loss": 0.9628,
      "step": 4875
    },
    {
      "epoch": 0.6082075589372583,
      "grad_norm": 0.0,
      "learning_rate": 7.027639933963928e-06,
      "loss": 0.9274,
      "step": 4876
    },
    {
      "epoch": 0.6083322938755146,
      "grad_norm": 0.0,
      "learning_rate": 7.023782659469484e-06,
      "loss": 0.8855,
      "step": 4877
    },
    {
      "epoch": 0.6084570288137707,
      "grad_norm": 0.0,
      "learning_rate": 7.019925870768747e-06,
      "loss": 0.9314,
      "step": 4878
    },
    {
      "epoch": 0.608581763752027,
      "grad_norm": 0.0,
      "learning_rate": 7.016069568491233e-06,
      "loss": 0.9338,
      "step": 4879
    },
    {
      "epoch": 0.6087064986902831,
      "grad_norm": 0.0,
      "learning_rate": 7.01221375326639e-06,
      "loss": 0.8651,
      "step": 4880
    },
    {
      "epoch": 0.6088312336285393,
      "grad_norm": 0.0,
      "learning_rate": 7.008358425723586e-06,
      "loss": 0.9435,
      "step": 4881
    },
    {
      "epoch": 0.6089559685667956,
      "grad_norm": 0.0,
      "learning_rate": 7.004503586492103e-06,
      "loss": 0.9161,
      "step": 4882
    },
    {
      "epoch": 0.6090807035050517,
      "grad_norm": 0.0,
      "learning_rate": 7.000649236201153e-06,
      "loss": 0.9309,
      "step": 4883
    },
    {
      "epoch": 0.609205438443308,
      "grad_norm": 0.0,
      "learning_rate": 6.996795375479856e-06,
      "loss": 0.887,
      "step": 4884
    },
    {
      "epoch": 0.6093301733815641,
      "grad_norm": 0.0,
      "learning_rate": 6.992942004957271e-06,
      "loss": 0.9493,
      "step": 4885
    },
    {
      "epoch": 0.6094549083198204,
      "grad_norm": 0.0,
      "learning_rate": 6.9890891252623515e-06,
      "loss": 0.9022,
      "step": 4886
    },
    {
      "epoch": 0.6095796432580766,
      "grad_norm": 0.0,
      "learning_rate": 6.985236737023997e-06,
      "loss": 0.9341,
      "step": 4887
    },
    {
      "epoch": 0.6097043781963328,
      "grad_norm": 0.0,
      "learning_rate": 6.981384840871007e-06,
      "loss": 0.948,
      "step": 4888
    },
    {
      "epoch": 0.609829113134589,
      "grad_norm": 0.0,
      "learning_rate": 6.977533437432109e-06,
      "loss": 0.9197,
      "step": 4889
    },
    {
      "epoch": 0.6099538480728453,
      "grad_norm": 0.0,
      "learning_rate": 6.973682527335951e-06,
      "loss": 0.9252,
      "step": 4890
    },
    {
      "epoch": 0.6100785830111014,
      "grad_norm": 0.0,
      "learning_rate": 6.9698321112110966e-06,
      "loss": 0.9661,
      "step": 4891
    },
    {
      "epoch": 0.6102033179493576,
      "grad_norm": 0.0,
      "learning_rate": 6.965982189686033e-06,
      "loss": 0.9431,
      "step": 4892
    },
    {
      "epoch": 0.6103280528876138,
      "grad_norm": 0.0,
      "learning_rate": 6.962132763389162e-06,
      "loss": 0.9091,
      "step": 4893
    },
    {
      "epoch": 0.61045278782587,
      "grad_norm": 0.0,
      "learning_rate": 6.958283832948807e-06,
      "loss": 0.9472,
      "step": 4894
    },
    {
      "epoch": 0.6105775227641262,
      "grad_norm": 0.0,
      "learning_rate": 6.954435398993215e-06,
      "loss": 0.9437,
      "step": 4895
    },
    {
      "epoch": 0.6107022577023824,
      "grad_norm": 0.0,
      "learning_rate": 6.950587462150539e-06,
      "loss": 0.8938,
      "step": 4896
    },
    {
      "epoch": 0.6108269926406387,
      "grad_norm": 0.0,
      "learning_rate": 6.94674002304887e-06,
      "loss": 0.918,
      "step": 4897
    },
    {
      "epoch": 0.6109517275788948,
      "grad_norm": 0.0,
      "learning_rate": 6.9428930823162e-06,
      "loss": 0.8929,
      "step": 4898
    },
    {
      "epoch": 0.6110764625171511,
      "grad_norm": 0.0,
      "learning_rate": 6.939046640580447e-06,
      "loss": 0.9633,
      "step": 4899
    },
    {
      "epoch": 0.6112011974554072,
      "grad_norm": 0.0,
      "learning_rate": 6.935200698469448e-06,
      "loss": 0.9242,
      "step": 4900
    },
    {
      "epoch": 0.6113259323936635,
      "grad_norm": 0.0,
      "learning_rate": 6.931355256610959e-06,
      "loss": 0.9742,
      "step": 4901
    },
    {
      "epoch": 0.6114506673319197,
      "grad_norm": 0.0,
      "learning_rate": 6.9275103156326536e-06,
      "loss": 0.9537,
      "step": 4902
    },
    {
      "epoch": 0.6115754022701759,
      "grad_norm": 0.0,
      "learning_rate": 6.923665876162118e-06,
      "loss": 0.9479,
      "step": 4903
    },
    {
      "epoch": 0.6117001372084321,
      "grad_norm": 0.0,
      "learning_rate": 6.9198219388268685e-06,
      "loss": 0.9,
      "step": 4904
    },
    {
      "epoch": 0.6118248721466882,
      "grad_norm": 0.0,
      "learning_rate": 6.915978504254331e-06,
      "loss": 0.912,
      "step": 4905
    },
    {
      "epoch": 0.6119496070849445,
      "grad_norm": 0.0,
      "learning_rate": 6.912135573071845e-06,
      "loss": 0.9201,
      "step": 4906
    },
    {
      "epoch": 0.6120743420232007,
      "grad_norm": 0.0,
      "learning_rate": 6.9082931459066814e-06,
      "loss": 0.9103,
      "step": 4907
    },
    {
      "epoch": 0.6121990769614569,
      "grad_norm": 0.0,
      "learning_rate": 6.904451223386015e-06,
      "loss": 0.925,
      "step": 4908
    },
    {
      "epoch": 0.6123238118997131,
      "grad_norm": 0.0,
      "learning_rate": 6.90060980613695e-06,
      "loss": 0.8952,
      "step": 4909
    },
    {
      "epoch": 0.6124485468379693,
      "grad_norm": 0.0,
      "learning_rate": 6.8967688947864995e-06,
      "loss": 0.9373,
      "step": 4910
    },
    {
      "epoch": 0.6125732817762255,
      "grad_norm": 0.0,
      "learning_rate": 6.892928489961596e-06,
      "loss": 0.9037,
      "step": 4911
    },
    {
      "epoch": 0.6126980167144818,
      "grad_norm": 0.0,
      "learning_rate": 6.889088592289092e-06,
      "loss": 0.9313,
      "step": 4912
    },
    {
      "epoch": 0.6128227516527379,
      "grad_norm": 0.0,
      "learning_rate": 6.885249202395754e-06,
      "loss": 0.8975,
      "step": 4913
    },
    {
      "epoch": 0.6129474865909942,
      "grad_norm": 0.0,
      "learning_rate": 6.881410320908271e-06,
      "loss": 0.9545,
      "step": 4914
    },
    {
      "epoch": 0.6130722215292503,
      "grad_norm": 0.0,
      "learning_rate": 6.877571948453245e-06,
      "loss": 0.9639,
      "step": 4915
    },
    {
      "epoch": 0.6131969564675065,
      "grad_norm": 0.0,
      "learning_rate": 6.873734085657188e-06,
      "loss": 0.8952,
      "step": 4916
    },
    {
      "epoch": 0.6133216914057628,
      "grad_norm": 0.0,
      "learning_rate": 6.8698967331465435e-06,
      "loss": 0.8847,
      "step": 4917
    },
    {
      "epoch": 0.6134464263440189,
      "grad_norm": 0.0,
      "learning_rate": 6.866059891547661e-06,
      "loss": 0.9739,
      "step": 4918
    },
    {
      "epoch": 0.6135711612822752,
      "grad_norm": 0.0,
      "learning_rate": 6.862223561486811e-06,
      "loss": 0.9775,
      "step": 4919
    },
    {
      "epoch": 0.6136958962205313,
      "grad_norm": 0.0,
      "learning_rate": 6.858387743590175e-06,
      "loss": 0.9101,
      "step": 4920
    },
    {
      "epoch": 0.6138206311587876,
      "grad_norm": 0.0,
      "learning_rate": 6.854552438483866e-06,
      "loss": 0.9588,
      "step": 4921
    },
    {
      "epoch": 0.6139453660970438,
      "grad_norm": 0.0,
      "learning_rate": 6.850717646793892e-06,
      "loss": 0.9568,
      "step": 4922
    },
    {
      "epoch": 0.6140701010353,
      "grad_norm": 0.0,
      "learning_rate": 6.846883369146187e-06,
      "loss": 0.9252,
      "step": 4923
    },
    {
      "epoch": 0.6141948359735562,
      "grad_norm": 0.0,
      "learning_rate": 6.84304960616661e-06,
      "loss": 0.9554,
      "step": 4924
    },
    {
      "epoch": 0.6143195709118124,
      "grad_norm": 0.0,
      "learning_rate": 6.839216358480922e-06,
      "loss": 0.9088,
      "step": 4925
    },
    {
      "epoch": 0.6144443058500686,
      "grad_norm": 0.0,
      "learning_rate": 6.835383626714809e-06,
      "loss": 0.9207,
      "step": 4926
    },
    {
      "epoch": 0.6145690407883249,
      "grad_norm": 0.0,
      "learning_rate": 6.831551411493868e-06,
      "loss": 0.9293,
      "step": 4927
    },
    {
      "epoch": 0.614693775726581,
      "grad_norm": 0.0,
      "learning_rate": 6.82771971344361e-06,
      "loss": 0.9668,
      "step": 4928
    },
    {
      "epoch": 0.6148185106648372,
      "grad_norm": 0.0,
      "learning_rate": 6.823888533189469e-06,
      "loss": 0.9651,
      "step": 4929
    },
    {
      "epoch": 0.6149432456030934,
      "grad_norm": 0.0,
      "learning_rate": 6.820057871356786e-06,
      "loss": 0.8876,
      "step": 4930
    },
    {
      "epoch": 0.6150679805413496,
      "grad_norm": 0.0,
      "learning_rate": 6.81622772857083e-06,
      "loss": 0.9076,
      "step": 4931
    },
    {
      "epoch": 0.6151927154796059,
      "grad_norm": 0.0,
      "learning_rate": 6.812398105456769e-06,
      "loss": 0.9202,
      "step": 4932
    },
    {
      "epoch": 0.615317450417862,
      "grad_norm": 0.0,
      "learning_rate": 6.8085690026396915e-06,
      "loss": 0.9332,
      "step": 4933
    },
    {
      "epoch": 0.6154421853561183,
      "grad_norm": 0.0,
      "learning_rate": 6.804740420744613e-06,
      "loss": 0.928,
      "step": 4934
    },
    {
      "epoch": 0.6155669202943744,
      "grad_norm": 0.0,
      "learning_rate": 6.800912360396447e-06,
      "loss": 0.9712,
      "step": 4935
    },
    {
      "epoch": 0.6156916552326307,
      "grad_norm": 0.0,
      "learning_rate": 6.797084822220036e-06,
      "loss": 0.8845,
      "step": 4936
    },
    {
      "epoch": 0.6158163901708869,
      "grad_norm": 0.0,
      "learning_rate": 6.793257806840123e-06,
      "loss": 0.9291,
      "step": 4937
    },
    {
      "epoch": 0.6159411251091431,
      "grad_norm": 0.0,
      "learning_rate": 6.7894313148813814e-06,
      "loss": 0.9268,
      "step": 4938
    },
    {
      "epoch": 0.6160658600473993,
      "grad_norm": 0.0,
      "learning_rate": 6.785605346968387e-06,
      "loss": 0.9133,
      "step": 4939
    },
    {
      "epoch": 0.6161905949856555,
      "grad_norm": 0.0,
      "learning_rate": 6.781779903725629e-06,
      "loss": 0.8899,
      "step": 4940
    },
    {
      "epoch": 0.6163153299239117,
      "grad_norm": 0.0,
      "learning_rate": 6.777954985777529e-06,
      "loss": 0.9168,
      "step": 4941
    },
    {
      "epoch": 0.6164400648621678,
      "grad_norm": 0.0,
      "learning_rate": 6.7741305937483955e-06,
      "loss": 0.8869,
      "step": 4942
    },
    {
      "epoch": 0.6165647998004241,
      "grad_norm": 0.0,
      "learning_rate": 6.770306728262478e-06,
      "loss": 0.9354,
      "step": 4943
    },
    {
      "epoch": 0.6166895347386803,
      "grad_norm": 0.0,
      "learning_rate": 6.766483389943921e-06,
      "loss": 0.9184,
      "step": 4944
    },
    {
      "epoch": 0.6168142696769365,
      "grad_norm": 0.0,
      "learning_rate": 6.762660579416791e-06,
      "loss": 0.8831,
      "step": 4945
    },
    {
      "epoch": 0.6169390046151927,
      "grad_norm": 0.0,
      "learning_rate": 6.758838297305068e-06,
      "loss": 0.8932,
      "step": 4946
    },
    {
      "epoch": 0.617063739553449,
      "grad_norm": 0.0,
      "learning_rate": 6.755016544232642e-06,
      "loss": 0.9566,
      "step": 4947
    },
    {
      "epoch": 0.6171884744917051,
      "grad_norm": 0.0,
      "learning_rate": 6.751195320823325e-06,
      "loss": 0.9034,
      "step": 4948
    },
    {
      "epoch": 0.6173132094299614,
      "grad_norm": 0.0,
      "learning_rate": 6.747374627700832e-06,
      "loss": 0.9234,
      "step": 4949
    },
    {
      "epoch": 0.6174379443682175,
      "grad_norm": 0.0,
      "learning_rate": 6.743554465488795e-06,
      "loss": 0.9355,
      "step": 4950
    },
    {
      "epoch": 0.6175626793064738,
      "grad_norm": 0.0,
      "learning_rate": 6.73973483481077e-06,
      "loss": 0.9749,
      "step": 4951
    },
    {
      "epoch": 0.61768741424473,
      "grad_norm": 0.0,
      "learning_rate": 6.7359157362902025e-06,
      "loss": 0.9079,
      "step": 4952
    },
    {
      "epoch": 0.6178121491829861,
      "grad_norm": 0.0,
      "learning_rate": 6.73209717055048e-06,
      "loss": 0.8748,
      "step": 4953
    },
    {
      "epoch": 0.6179368841212424,
      "grad_norm": 0.0,
      "learning_rate": 6.72827913821488e-06,
      "loss": 0.928,
      "step": 4954
    },
    {
      "epoch": 0.6180616190594985,
      "grad_norm": 0.0,
      "learning_rate": 6.7244616399066065e-06,
      "loss": 0.9665,
      "step": 4955
    },
    {
      "epoch": 0.6181863539977548,
      "grad_norm": 0.0,
      "learning_rate": 6.720644676248769e-06,
      "loss": 0.9132,
      "step": 4956
    },
    {
      "epoch": 0.6183110889360109,
      "grad_norm": 0.0,
      "learning_rate": 6.716828247864391e-06,
      "loss": 0.9665,
      "step": 4957
    },
    {
      "epoch": 0.6184358238742672,
      "grad_norm": 0.0,
      "learning_rate": 6.713012355376413e-06,
      "loss": 1.0286,
      "step": 4958
    },
    {
      "epoch": 0.6185605588125234,
      "grad_norm": 0.0,
      "learning_rate": 6.7091969994076785e-06,
      "loss": 0.9791,
      "step": 4959
    },
    {
      "epoch": 0.6186852937507796,
      "grad_norm": 0.0,
      "learning_rate": 6.705382180580959e-06,
      "loss": 0.9402,
      "step": 4960
    },
    {
      "epoch": 0.6188100286890358,
      "grad_norm": 0.0,
      "learning_rate": 6.701567899518924e-06,
      "loss": 0.9588,
      "step": 4961
    },
    {
      "epoch": 0.618934763627292,
      "grad_norm": 0.0,
      "learning_rate": 6.697754156844159e-06,
      "loss": 0.9154,
      "step": 4962
    },
    {
      "epoch": 0.6190594985655482,
      "grad_norm": 0.0,
      "learning_rate": 6.693940953179166e-06,
      "loss": 0.9103,
      "step": 4963
    },
    {
      "epoch": 0.6191842335038045,
      "grad_norm": 0.0,
      "learning_rate": 6.690128289146354e-06,
      "loss": 0.9243,
      "step": 4964
    },
    {
      "epoch": 0.6193089684420606,
      "grad_norm": 0.0,
      "learning_rate": 6.6863161653680484e-06,
      "loss": 0.9256,
      "step": 4965
    },
    {
      "epoch": 0.6194337033803168,
      "grad_norm": 0.0,
      "learning_rate": 6.682504582466482e-06,
      "loss": 0.9859,
      "step": 4966
    },
    {
      "epoch": 0.619558438318573,
      "grad_norm": 0.0,
      "learning_rate": 6.678693541063799e-06,
      "loss": 0.9362,
      "step": 4967
    },
    {
      "epoch": 0.6196831732568292,
      "grad_norm": 0.0,
      "learning_rate": 6.674883041782062e-06,
      "loss": 0.9577,
      "step": 4968
    },
    {
      "epoch": 0.6198079081950855,
      "grad_norm": 0.0,
      "learning_rate": 6.671073085243234e-06,
      "loss": 0.8799,
      "step": 4969
    },
    {
      "epoch": 0.6199326431333416,
      "grad_norm": 0.0,
      "learning_rate": 6.667263672069204e-06,
      "loss": 0.9681,
      "step": 4970
    },
    {
      "epoch": 0.6200573780715979,
      "grad_norm": 0.0,
      "learning_rate": 6.6634548028817596e-06,
      "loss": 0.8977,
      "step": 4971
    },
    {
      "epoch": 0.620182113009854,
      "grad_norm": 0.0,
      "learning_rate": 6.659646478302606e-06,
      "loss": 0.895,
      "step": 4972
    },
    {
      "epoch": 0.6203068479481103,
      "grad_norm": 0.0,
      "learning_rate": 6.655838698953359e-06,
      "loss": 0.9323,
      "step": 4973
    },
    {
      "epoch": 0.6204315828863665,
      "grad_norm": 0.0,
      "learning_rate": 6.652031465455537e-06,
      "loss": 0.904,
      "step": 4974
    },
    {
      "epoch": 0.6205563178246227,
      "grad_norm": 0.0,
      "learning_rate": 6.648224778430585e-06,
      "loss": 0.9049,
      "step": 4975
    },
    {
      "epoch": 0.6206810527628789,
      "grad_norm": 0.0,
      "learning_rate": 6.644418638499845e-06,
      "loss": 0.9146,
      "step": 4976
    },
    {
      "epoch": 0.620805787701135,
      "grad_norm": 0.0,
      "learning_rate": 6.640613046284581e-06,
      "loss": 0.9123,
      "step": 4977
    },
    {
      "epoch": 0.6209305226393913,
      "grad_norm": 0.0,
      "learning_rate": 6.636808002405955e-06,
      "loss": 0.9104,
      "step": 4978
    },
    {
      "epoch": 0.6210552575776475,
      "grad_norm": 0.0,
      "learning_rate": 6.633003507485045e-06,
      "loss": 0.9613,
      "step": 4979
    },
    {
      "epoch": 0.6211799925159037,
      "grad_norm": 0.0,
      "learning_rate": 6.6291995621428494e-06,
      "loss": 0.9188,
      "step": 4980
    },
    {
      "epoch": 0.6213047274541599,
      "grad_norm": 0.0,
      "learning_rate": 6.6253961670002595e-06,
      "loss": 0.9263,
      "step": 4981
    },
    {
      "epoch": 0.6214294623924161,
      "grad_norm": 0.0,
      "learning_rate": 6.62159332267809e-06,
      "loss": 0.9655,
      "step": 4982
    },
    {
      "epoch": 0.6215541973306723,
      "grad_norm": 0.0,
      "learning_rate": 6.6177910297970605e-06,
      "loss": 0.9359,
      "step": 4983
    },
    {
      "epoch": 0.6216789322689286,
      "grad_norm": 0.0,
      "learning_rate": 6.613989288977798e-06,
      "loss": 0.9601,
      "step": 4984
    },
    {
      "epoch": 0.6218036672071847,
      "grad_norm": 0.0,
      "learning_rate": 6.610188100840844e-06,
      "loss": 0.891,
      "step": 4985
    },
    {
      "epoch": 0.621928402145441,
      "grad_norm": 0.0,
      "learning_rate": 6.6063874660066475e-06,
      "loss": 0.9459,
      "step": 4986
    },
    {
      "epoch": 0.6220531370836971,
      "grad_norm": 0.0,
      "learning_rate": 6.602587385095571e-06,
      "loss": 0.9721,
      "step": 4987
    },
    {
      "epoch": 0.6221778720219534,
      "grad_norm": 0.0,
      "learning_rate": 6.598787858727885e-06,
      "loss": 0.9332,
      "step": 4988
    },
    {
      "epoch": 0.6223026069602096,
      "grad_norm": 0.0,
      "learning_rate": 6.594988887523759e-06,
      "loss": 0.9301,
      "step": 4989
    },
    {
      "epoch": 0.6224273418984657,
      "grad_norm": 0.0,
      "learning_rate": 6.591190472103288e-06,
      "loss": 0.9148,
      "step": 4990
    },
    {
      "epoch": 0.622552076836722,
      "grad_norm": 0.0,
      "learning_rate": 6.587392613086468e-06,
      "loss": 0.9478,
      "step": 4991
    },
    {
      "epoch": 0.6226768117749781,
      "grad_norm": 0.0,
      "learning_rate": 6.5835953110932045e-06,
      "loss": 0.951,
      "step": 4992
    },
    {
      "epoch": 0.6228015467132344,
      "grad_norm": 0.0,
      "learning_rate": 6.579798566743314e-06,
      "loss": 0.903,
      "step": 4993
    },
    {
      "epoch": 0.6229262816514906,
      "grad_norm": 0.0,
      "learning_rate": 6.576002380656521e-06,
      "loss": 0.8939,
      "step": 4994
    },
    {
      "epoch": 0.6230510165897468,
      "grad_norm": 0.0,
      "learning_rate": 6.572206753452457e-06,
      "loss": 0.8804,
      "step": 4995
    },
    {
      "epoch": 0.623175751528003,
      "grad_norm": 0.0,
      "learning_rate": 6.568411685750662e-06,
      "loss": 0.9856,
      "step": 4996
    },
    {
      "epoch": 0.6233004864662592,
      "grad_norm": 0.0,
      "learning_rate": 6.564617178170593e-06,
      "loss": 0.8853,
      "step": 4997
    },
    {
      "epoch": 0.6234252214045154,
      "grad_norm": 0.0,
      "learning_rate": 6.560823231331606e-06,
      "loss": 0.8701,
      "step": 4998
    },
    {
      "epoch": 0.6235499563427717,
      "grad_norm": 0.0,
      "learning_rate": 6.557029845852969e-06,
      "loss": 0.9239,
      "step": 4999
    },
    {
      "epoch": 0.6236746912810278,
      "grad_norm": 0.0,
      "learning_rate": 6.553237022353857e-06,
      "loss": 0.9321,
      "step": 5000
    },
    {
      "epoch": 0.623799426219284,
      "grad_norm": 0.0,
      "learning_rate": 6.549444761453355e-06,
      "loss": 0.9135,
      "step": 5001
    },
    {
      "epoch": 0.6239241611575402,
      "grad_norm": 0.0,
      "learning_rate": 6.545653063770458e-06,
      "loss": 0.9688,
      "step": 5002
    },
    {
      "epoch": 0.6240488960957964,
      "grad_norm": 0.0,
      "learning_rate": 6.541861929924061e-06,
      "loss": 0.9161,
      "step": 5003
    },
    {
      "epoch": 0.6241736310340527,
      "grad_norm": 0.0,
      "learning_rate": 6.538071360532981e-06,
      "loss": 0.9542,
      "step": 5004
    },
    {
      "epoch": 0.6242983659723088,
      "grad_norm": 0.0,
      "learning_rate": 6.534281356215928e-06,
      "loss": 0.8754,
      "step": 5005
    },
    {
      "epoch": 0.6244231009105651,
      "grad_norm": 0.0,
      "learning_rate": 6.5304919175915235e-06,
      "loss": 0.8912,
      "step": 5006
    },
    {
      "epoch": 0.6245478358488212,
      "grad_norm": 0.0,
      "learning_rate": 6.526703045278308e-06,
      "loss": 0.8878,
      "step": 5007
    },
    {
      "epoch": 0.6246725707870775,
      "grad_norm": 0.0,
      "learning_rate": 6.522914739894716e-06,
      "loss": 0.9181,
      "step": 5008
    },
    {
      "epoch": 0.6247973057253337,
      "grad_norm": 0.0,
      "learning_rate": 6.519127002059096e-06,
      "loss": 0.9243,
      "step": 5009
    },
    {
      "epoch": 0.6249220406635899,
      "grad_norm": 0.0,
      "learning_rate": 6.5153398323897e-06,
      "loss": 0.9472,
      "step": 5010
    },
    {
      "epoch": 0.6250467756018461,
      "grad_norm": 0.0,
      "learning_rate": 6.5115532315046935e-06,
      "loss": 0.9212,
      "step": 5011
    },
    {
      "epoch": 0.6251715105401023,
      "grad_norm": 0.0,
      "learning_rate": 6.5077672000221425e-06,
      "loss": 0.9141,
      "step": 5012
    },
    {
      "epoch": 0.6252962454783585,
      "grad_norm": 0.0,
      "learning_rate": 6.503981738560019e-06,
      "loss": 0.9173,
      "step": 5013
    },
    {
      "epoch": 0.6254209804166146,
      "grad_norm": 0.0,
      "learning_rate": 6.500196847736217e-06,
      "loss": 0.9319,
      "step": 5014
    },
    {
      "epoch": 0.6255457153548709,
      "grad_norm": 0.0,
      "learning_rate": 6.496412528168512e-06,
      "loss": 0.9467,
      "step": 5015
    },
    {
      "epoch": 0.6256704502931271,
      "grad_norm": 0.0,
      "learning_rate": 6.4926287804746105e-06,
      "loss": 0.8847,
      "step": 5016
    },
    {
      "epoch": 0.6257951852313833,
      "grad_norm": 0.0,
      "learning_rate": 6.488845605272114e-06,
      "loss": 0.9483,
      "step": 5017
    },
    {
      "epoch": 0.6259199201696395,
      "grad_norm": 0.0,
      "learning_rate": 6.485063003178526e-06,
      "loss": 0.9396,
      "step": 5018
    },
    {
      "epoch": 0.6260446551078958,
      "grad_norm": 0.0,
      "learning_rate": 6.481280974811273e-06,
      "loss": 0.9546,
      "step": 5019
    },
    {
      "epoch": 0.6261693900461519,
      "grad_norm": 0.0,
      "learning_rate": 6.4774995207876654e-06,
      "loss": 0.873,
      "step": 5020
    },
    {
      "epoch": 0.6262941249844082,
      "grad_norm": 0.0,
      "learning_rate": 6.473718641724942e-06,
      "loss": 0.9057,
      "step": 5021
    },
    {
      "epoch": 0.6264188599226643,
      "grad_norm": 0.0,
      "learning_rate": 6.4699383382402326e-06,
      "loss": 0.9534,
      "step": 5022
    },
    {
      "epoch": 0.6265435948609206,
      "grad_norm": 0.0,
      "learning_rate": 6.466158610950575e-06,
      "loss": 0.8971,
      "step": 5023
    },
    {
      "epoch": 0.6266683297991767,
      "grad_norm": 0.0,
      "learning_rate": 6.462379460472925e-06,
      "loss": 0.9439,
      "step": 5024
    },
    {
      "epoch": 0.6267930647374329,
      "grad_norm": 0.0,
      "learning_rate": 6.4586008874241245e-06,
      "loss": 0.9471,
      "step": 5025
    },
    {
      "epoch": 0.6269177996756892,
      "grad_norm": 0.0,
      "learning_rate": 6.454822892420941e-06,
      "loss": 0.8951,
      "step": 5026
    },
    {
      "epoch": 0.6270425346139453,
      "grad_norm": 0.0,
      "learning_rate": 6.4510454760800335e-06,
      "loss": 0.9465,
      "step": 5027
    },
    {
      "epoch": 0.6271672695522016,
      "grad_norm": 0.0,
      "learning_rate": 6.4472686390179735e-06,
      "loss": 0.9077,
      "step": 5028
    },
    {
      "epoch": 0.6272920044904577,
      "grad_norm": 0.0,
      "learning_rate": 6.443492381851237e-06,
      "loss": 0.9832,
      "step": 5029
    },
    {
      "epoch": 0.627416739428714,
      "grad_norm": 0.0,
      "learning_rate": 6.4397167051962e-06,
      "loss": 0.9208,
      "step": 5030
    },
    {
      "epoch": 0.6275414743669702,
      "grad_norm": 0.0,
      "learning_rate": 6.435941609669154e-06,
      "loss": 0.9231,
      "step": 5031
    },
    {
      "epoch": 0.6276662093052264,
      "grad_norm": 0.0,
      "learning_rate": 6.432167095886282e-06,
      "loss": 0.9102,
      "step": 5032
    },
    {
      "epoch": 0.6277909442434826,
      "grad_norm": 0.0,
      "learning_rate": 6.428393164463689e-06,
      "loss": 0.9666,
      "step": 5033
    },
    {
      "epoch": 0.6279156791817389,
      "grad_norm": 0.0,
      "learning_rate": 6.424619816017372e-06,
      "loss": 0.9727,
      "step": 5034
    },
    {
      "epoch": 0.628040414119995,
      "grad_norm": 0.0,
      "learning_rate": 6.42084705116323e-06,
      "loss": 0.908,
      "step": 5035
    },
    {
      "epoch": 0.6281651490582513,
      "grad_norm": 0.0,
      "learning_rate": 6.417074870517085e-06,
      "loss": 0.9528,
      "step": 5036
    },
    {
      "epoch": 0.6282898839965074,
      "grad_norm": 0.0,
      "learning_rate": 6.413303274694642e-06,
      "loss": 0.9708,
      "step": 5037
    },
    {
      "epoch": 0.6284146189347636,
      "grad_norm": 0.0,
      "learning_rate": 6.409532264311529e-06,
      "loss": 0.9735,
      "step": 5038
    },
    {
      "epoch": 0.6285393538730198,
      "grad_norm": 0.0,
      "learning_rate": 6.4057618399832645e-06,
      "loss": 0.9287,
      "step": 5039
    },
    {
      "epoch": 0.628664088811276,
      "grad_norm": 0.0,
      "learning_rate": 6.4019920023252745e-06,
      "loss": 0.8874,
      "step": 5040
    },
    {
      "epoch": 0.6287888237495323,
      "grad_norm": 0.0,
      "learning_rate": 6.3982227519528986e-06,
      "loss": 0.9003,
      "step": 5041
    },
    {
      "epoch": 0.6289135586877884,
      "grad_norm": 0.0,
      "learning_rate": 6.394454089481366e-06,
      "loss": 0.896,
      "step": 5042
    },
    {
      "epoch": 0.6290382936260447,
      "grad_norm": 0.0,
      "learning_rate": 6.390686015525825e-06,
      "loss": 0.9412,
      "step": 5043
    },
    {
      "epoch": 0.6291630285643008,
      "grad_norm": 0.0,
      "learning_rate": 6.3869185307013135e-06,
      "loss": 0.905,
      "step": 5044
    },
    {
      "epoch": 0.6292877635025571,
      "grad_norm": 0.0,
      "learning_rate": 6.383151635622786e-06,
      "loss": 0.9483,
      "step": 5045
    },
    {
      "epoch": 0.6294124984408133,
      "grad_norm": 0.0,
      "learning_rate": 6.379385330905092e-06,
      "loss": 0.9469,
      "step": 5046
    },
    {
      "epoch": 0.6295372333790695,
      "grad_norm": 0.0,
      "learning_rate": 6.375619617162985e-06,
      "loss": 0.9561,
      "step": 5047
    },
    {
      "epoch": 0.6296619683173257,
      "grad_norm": 0.0,
      "learning_rate": 6.371854495011128e-06,
      "loss": 0.9955,
      "step": 5048
    },
    {
      "epoch": 0.6297867032555818,
      "grad_norm": 0.0,
      "learning_rate": 6.368089965064079e-06,
      "loss": 0.9334,
      "step": 5049
    },
    {
      "epoch": 0.6299114381938381,
      "grad_norm": 0.0,
      "learning_rate": 6.364326027936314e-06,
      "loss": 0.9345,
      "step": 5050
    },
    {
      "epoch": 0.6300361731320943,
      "grad_norm": 0.0,
      "learning_rate": 6.360562684242193e-06,
      "loss": 0.9667,
      "step": 5051
    },
    {
      "epoch": 0.6301609080703505,
      "grad_norm": 0.0,
      "learning_rate": 6.3567999345959875e-06,
      "loss": 0.9397,
      "step": 5052
    },
    {
      "epoch": 0.6302856430086067,
      "grad_norm": 0.0,
      "learning_rate": 6.353037779611882e-06,
      "loss": 0.9596,
      "step": 5053
    },
    {
      "epoch": 0.6304103779468629,
      "grad_norm": 0.0,
      "learning_rate": 6.349276219903948e-06,
      "loss": 0.9251,
      "step": 5054
    },
    {
      "epoch": 0.6305351128851191,
      "grad_norm": 0.0,
      "learning_rate": 6.345515256086172e-06,
      "loss": 0.926,
      "step": 5055
    },
    {
      "epoch": 0.6306598478233754,
      "grad_norm": 0.0,
      "learning_rate": 6.3417548887724354e-06,
      "loss": 0.9722,
      "step": 5056
    },
    {
      "epoch": 0.6307845827616315,
      "grad_norm": 0.0,
      "learning_rate": 6.337995118576521e-06,
      "loss": 0.9059,
      "step": 5057
    },
    {
      "epoch": 0.6309093176998878,
      "grad_norm": 0.0,
      "learning_rate": 6.334235946112126e-06,
      "loss": 0.9539,
      "step": 5058
    },
    {
      "epoch": 0.6310340526381439,
      "grad_norm": 0.0,
      "learning_rate": 6.330477371992831e-06,
      "loss": 0.9538,
      "step": 5059
    },
    {
      "epoch": 0.6311587875764002,
      "grad_norm": 0.0,
      "learning_rate": 6.326719396832146e-06,
      "loss": 0.9118,
      "step": 5060
    },
    {
      "epoch": 0.6312835225146564,
      "grad_norm": 0.0,
      "learning_rate": 6.3229620212434504e-06,
      "loss": 0.9366,
      "step": 5061
    },
    {
      "epoch": 0.6314082574529125,
      "grad_norm": 0.0,
      "learning_rate": 6.319205245840057e-06,
      "loss": 0.8726,
      "step": 5062
    },
    {
      "epoch": 0.6315329923911688,
      "grad_norm": 0.0,
      "learning_rate": 6.315449071235158e-06,
      "loss": 0.908,
      "step": 5063
    },
    {
      "epoch": 0.6316577273294249,
      "grad_norm": 0.0,
      "learning_rate": 6.311693498041855e-06,
      "loss": 0.884,
      "step": 5064
    },
    {
      "epoch": 0.6317824622676812,
      "grad_norm": 0.0,
      "learning_rate": 6.3079385268731575e-06,
      "loss": 0.9652,
      "step": 5065
    },
    {
      "epoch": 0.6319071972059374,
      "grad_norm": 0.0,
      "learning_rate": 6.304184158341966e-06,
      "loss": 0.9518,
      "step": 5066
    },
    {
      "epoch": 0.6320319321441936,
      "grad_norm": 0.0,
      "learning_rate": 6.300430393061094e-06,
      "loss": 0.9371,
      "step": 5067
    },
    {
      "epoch": 0.6321566670824498,
      "grad_norm": 0.0,
      "learning_rate": 6.2966772316432464e-06,
      "loss": 0.9126,
      "step": 5068
    },
    {
      "epoch": 0.632281402020706,
      "grad_norm": 0.0,
      "learning_rate": 6.292924674701032e-06,
      "loss": 0.9027,
      "step": 5069
    },
    {
      "epoch": 0.6324061369589622,
      "grad_norm": 0.0,
      "learning_rate": 6.289172722846971e-06,
      "loss": 0.9255,
      "step": 5070
    },
    {
      "epoch": 0.6325308718972185,
      "grad_norm": 0.0,
      "learning_rate": 6.2854213766934655e-06,
      "loss": 0.93,
      "step": 5071
    },
    {
      "epoch": 0.6326556068354746,
      "grad_norm": 0.0,
      "learning_rate": 6.281670636852839e-06,
      "loss": 0.9091,
      "step": 5072
    },
    {
      "epoch": 0.6327803417737308,
      "grad_norm": 0.0,
      "learning_rate": 6.277920503937303e-06,
      "loss": 0.8966,
      "step": 5073
    },
    {
      "epoch": 0.632905076711987,
      "grad_norm": 0.0,
      "learning_rate": 6.274170978558971e-06,
      "loss": 1.0057,
      "step": 5074
    },
    {
      "epoch": 0.6330298116502432,
      "grad_norm": 0.0,
      "learning_rate": 6.270422061329866e-06,
      "loss": 0.9432,
      "step": 5075
    },
    {
      "epoch": 0.6331545465884995,
      "grad_norm": 0.0,
      "learning_rate": 6.2666737528619e-06,
      "loss": 0.9481,
      "step": 5076
    },
    {
      "epoch": 0.6332792815267556,
      "grad_norm": 0.0,
      "learning_rate": 6.262926053766898e-06,
      "loss": 0.9518,
      "step": 5077
    },
    {
      "epoch": 0.6334040164650119,
      "grad_norm": 0.0,
      "learning_rate": 6.25917896465657e-06,
      "loss": 0.8952,
      "step": 5078
    },
    {
      "epoch": 0.633528751403268,
      "grad_norm": 0.0,
      "learning_rate": 6.255432486142543e-06,
      "loss": 0.9361,
      "step": 5079
    },
    {
      "epoch": 0.6336534863415243,
      "grad_norm": 0.0,
      "learning_rate": 6.251686618836336e-06,
      "loss": 0.8998,
      "step": 5080
    },
    {
      "epoch": 0.6337782212797805,
      "grad_norm": 0.0,
      "learning_rate": 6.247941363349365e-06,
      "loss": 0.9123,
      "step": 5081
    },
    {
      "epoch": 0.6339029562180367,
      "grad_norm": 0.0,
      "learning_rate": 6.244196720292954e-06,
      "loss": 0.9105,
      "step": 5082
    },
    {
      "epoch": 0.6340276911562929,
      "grad_norm": 0.0,
      "learning_rate": 6.2404526902783205e-06,
      "loss": 0.9111,
      "step": 5083
    },
    {
      "epoch": 0.6341524260945491,
      "grad_norm": 0.0,
      "learning_rate": 6.236709273916585e-06,
      "loss": 0.9133,
      "step": 5084
    },
    {
      "epoch": 0.6342771610328053,
      "grad_norm": 0.0,
      "learning_rate": 6.232966471818771e-06,
      "loss": 0.8972,
      "step": 5085
    },
    {
      "epoch": 0.6344018959710614,
      "grad_norm": 0.0,
      "learning_rate": 6.229224284595792e-06,
      "loss": 0.9144,
      "step": 5086
    },
    {
      "epoch": 0.6345266309093177,
      "grad_norm": 0.0,
      "learning_rate": 6.225482712858473e-06,
      "loss": 0.9683,
      "step": 5087
    },
    {
      "epoch": 0.6346513658475739,
      "grad_norm": 0.0,
      "learning_rate": 6.2217417572175255e-06,
      "loss": 0.9336,
      "step": 5088
    },
    {
      "epoch": 0.6347761007858301,
      "grad_norm": 0.0,
      "learning_rate": 6.218001418283577e-06,
      "loss": 0.9216,
      "step": 5089
    },
    {
      "epoch": 0.6349008357240863,
      "grad_norm": 0.0,
      "learning_rate": 6.214261696667139e-06,
      "loss": 0.9474,
      "step": 5090
    },
    {
      "epoch": 0.6350255706623426,
      "grad_norm": 0.0,
      "learning_rate": 6.2105225929786296e-06,
      "loss": 0.8929,
      "step": 5091
    },
    {
      "epoch": 0.6351503056005987,
      "grad_norm": 0.0,
      "learning_rate": 6.206784107828367e-06,
      "loss": 0.916,
      "step": 5092
    },
    {
      "epoch": 0.635275040538855,
      "grad_norm": 0.0,
      "learning_rate": 6.203046241826563e-06,
      "loss": 0.8898,
      "step": 5093
    },
    {
      "epoch": 0.6353997754771111,
      "grad_norm": 0.0,
      "learning_rate": 6.199308995583335e-06,
      "loss": 0.8908,
      "step": 5094
    },
    {
      "epoch": 0.6355245104153674,
      "grad_norm": 0.0,
      "learning_rate": 6.195572369708689e-06,
      "loss": 0.9297,
      "step": 5095
    },
    {
      "epoch": 0.6356492453536235,
      "grad_norm": 0.0,
      "learning_rate": 6.191836364812548e-06,
      "loss": 0.9091,
      "step": 5096
    },
    {
      "epoch": 0.6357739802918797,
      "grad_norm": 0.0,
      "learning_rate": 6.188100981504712e-06,
      "loss": 0.9056,
      "step": 5097
    },
    {
      "epoch": 0.635898715230136,
      "grad_norm": 0.0,
      "learning_rate": 6.1843662203948905e-06,
      "loss": 0.9548,
      "step": 5098
    },
    {
      "epoch": 0.6360234501683921,
      "grad_norm": 0.0,
      "learning_rate": 6.180632082092699e-06,
      "loss": 0.9287,
      "step": 5099
    },
    {
      "epoch": 0.6361481851066484,
      "grad_norm": 0.0,
      "learning_rate": 6.176898567207633e-06,
      "loss": 0.9186,
      "step": 5100
    },
    {
      "epoch": 0.6362729200449045,
      "grad_norm": 0.0,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.9331,
      "step": 5101
    },
    {
      "epoch": 0.6363976549831608,
      "grad_norm": 0.0,
      "learning_rate": 6.16943341012641e-06,
      "loss": 0.8877,
      "step": 5102
    },
    {
      "epoch": 0.636522389921417,
      "grad_norm": 0.0,
      "learning_rate": 6.165701769148748e-06,
      "loss": 0.9774,
      "step": 5103
    },
    {
      "epoch": 0.6366471248596732,
      "grad_norm": 0.0,
      "learning_rate": 6.161970754025224e-06,
      "loss": 0.9027,
      "step": 5104
    },
    {
      "epoch": 0.6367718597979294,
      "grad_norm": 0.0,
      "learning_rate": 6.158240365364823e-06,
      "loss": 0.9683,
      "step": 5105
    },
    {
      "epoch": 0.6368965947361857,
      "grad_norm": 0.0,
      "learning_rate": 6.154510603776451e-06,
      "loss": 0.9379,
      "step": 5106
    },
    {
      "epoch": 0.6370213296744418,
      "grad_norm": 0.0,
      "learning_rate": 6.150781469868892e-06,
      "loss": 0.9247,
      "step": 5107
    },
    {
      "epoch": 0.6371460646126981,
      "grad_norm": 0.0,
      "learning_rate": 6.147052964250831e-06,
      "loss": 0.9442,
      "step": 5108
    },
    {
      "epoch": 0.6372707995509542,
      "grad_norm": 0.0,
      "learning_rate": 6.143325087530863e-06,
      "loss": 0.9259,
      "step": 5109
    },
    {
      "epoch": 0.6373955344892104,
      "grad_norm": 0.0,
      "learning_rate": 6.139597840317464e-06,
      "loss": 0.9298,
      "step": 5110
    },
    {
      "epoch": 0.6375202694274666,
      "grad_norm": 0.0,
      "learning_rate": 6.13587122321902e-06,
      "loss": 0.943,
      "step": 5111
    },
    {
      "epoch": 0.6376450043657228,
      "grad_norm": 0.0,
      "learning_rate": 6.1321452368438035e-06,
      "loss": 0.9514,
      "step": 5112
    },
    {
      "epoch": 0.6377697393039791,
      "grad_norm": 0.0,
      "learning_rate": 6.1284198817999964e-06,
      "loss": 0.9485,
      "step": 5113
    },
    {
      "epoch": 0.6378944742422352,
      "grad_norm": 0.0,
      "learning_rate": 6.1246951586956635e-06,
      "loss": 0.9311,
      "step": 5114
    },
    {
      "epoch": 0.6380192091804915,
      "grad_norm": 0.0,
      "learning_rate": 6.1209710681387745e-06,
      "loss": 0.9241,
      "step": 5115
    },
    {
      "epoch": 0.6381439441187476,
      "grad_norm": 0.0,
      "learning_rate": 6.117247610737202e-06,
      "loss": 0.9782,
      "step": 5116
    },
    {
      "epoch": 0.6382686790570039,
      "grad_norm": 0.0,
      "learning_rate": 6.113524787098699e-06,
      "loss": 0.9135,
      "step": 5117
    },
    {
      "epoch": 0.6383934139952601,
      "grad_norm": 0.0,
      "learning_rate": 6.109802597830932e-06,
      "loss": 0.9416,
      "step": 5118
    },
    {
      "epoch": 0.6385181489335163,
      "grad_norm": 0.0,
      "learning_rate": 6.106081043541452e-06,
      "loss": 0.9339,
      "step": 5119
    },
    {
      "epoch": 0.6386428838717725,
      "grad_norm": 0.0,
      "learning_rate": 6.102360124837707e-06,
      "loss": 0.9342,
      "step": 5120
    },
    {
      "epoch": 0.6387676188100286,
      "grad_norm": 0.0,
      "learning_rate": 6.098639842327052e-06,
      "loss": 0.9523,
      "step": 5121
    },
    {
      "epoch": 0.6388923537482849,
      "grad_norm": 0.0,
      "learning_rate": 6.094920196616726e-06,
      "loss": 0.9707,
      "step": 5122
    },
    {
      "epoch": 0.6390170886865411,
      "grad_norm": 0.0,
      "learning_rate": 6.091201188313871e-06,
      "loss": 0.929,
      "step": 5123
    },
    {
      "epoch": 0.6391418236247973,
      "grad_norm": 0.0,
      "learning_rate": 6.087482818025523e-06,
      "loss": 0.8739,
      "step": 5124
    },
    {
      "epoch": 0.6392665585630535,
      "grad_norm": 0.0,
      "learning_rate": 6.083765086358611e-06,
      "loss": 0.911,
      "step": 5125
    },
    {
      "epoch": 0.6393912935013097,
      "grad_norm": 0.0,
      "learning_rate": 6.080047993919966e-06,
      "loss": 0.9157,
      "step": 5126
    },
    {
      "epoch": 0.6395160284395659,
      "grad_norm": 0.0,
      "learning_rate": 6.076331541316308e-06,
      "loss": 0.8973,
      "step": 5127
    },
    {
      "epoch": 0.6396407633778222,
      "grad_norm": 0.0,
      "learning_rate": 6.072615729154261e-06,
      "loss": 0.9747,
      "step": 5128
    },
    {
      "epoch": 0.6397654983160783,
      "grad_norm": 0.0,
      "learning_rate": 6.0689005580403325e-06,
      "loss": 0.9183,
      "step": 5129
    },
    {
      "epoch": 0.6398902332543346,
      "grad_norm": 0.0,
      "learning_rate": 6.065186028580938e-06,
      "loss": 0.9644,
      "step": 5130
    },
    {
      "epoch": 0.6400149681925907,
      "grad_norm": 0.0,
      "learning_rate": 6.0614721413823794e-06,
      "loss": 0.9236,
      "step": 5131
    },
    {
      "epoch": 0.640139703130847,
      "grad_norm": 0.0,
      "learning_rate": 6.0577588970508534e-06,
      "loss": 0.9156,
      "step": 5132
    },
    {
      "epoch": 0.6402644380691032,
      "grad_norm": 0.0,
      "learning_rate": 6.054046296192464e-06,
      "loss": 0.9521,
      "step": 5133
    },
    {
      "epoch": 0.6403891730073593,
      "grad_norm": 0.0,
      "learning_rate": 6.050334339413191e-06,
      "loss": 0.9435,
      "step": 5134
    },
    {
      "epoch": 0.6405139079456156,
      "grad_norm": 0.0,
      "learning_rate": 6.046623027318928e-06,
      "loss": 0.943,
      "step": 5135
    },
    {
      "epoch": 0.6406386428838717,
      "grad_norm": 0.0,
      "learning_rate": 6.042912360515452e-06,
      "loss": 0.8812,
      "step": 5136
    },
    {
      "epoch": 0.640763377822128,
      "grad_norm": 0.0,
      "learning_rate": 6.039202339608432e-06,
      "loss": 0.939,
      "step": 5137
    },
    {
      "epoch": 0.6408881127603842,
      "grad_norm": 0.0,
      "learning_rate": 6.0354929652034445e-06,
      "loss": 0.9625,
      "step": 5138
    },
    {
      "epoch": 0.6410128476986404,
      "grad_norm": 0.0,
      "learning_rate": 6.031784237905949e-06,
      "loss": 0.9457,
      "step": 5139
    },
    {
      "epoch": 0.6411375826368966,
      "grad_norm": 0.0,
      "learning_rate": 6.028076158321304e-06,
      "loss": 0.9555,
      "step": 5140
    },
    {
      "epoch": 0.6412623175751528,
      "grad_norm": 0.0,
      "learning_rate": 6.024368727054763e-06,
      "loss": 0.9364,
      "step": 5141
    },
    {
      "epoch": 0.641387052513409,
      "grad_norm": 0.0,
      "learning_rate": 6.020661944711466e-06,
      "loss": 0.903,
      "step": 5142
    },
    {
      "epoch": 0.6415117874516653,
      "grad_norm": 0.0,
      "learning_rate": 6.016955811896464e-06,
      "loss": 0.9763,
      "step": 5143
    },
    {
      "epoch": 0.6416365223899214,
      "grad_norm": 0.0,
      "learning_rate": 6.013250329214681e-06,
      "loss": 0.9002,
      "step": 5144
    },
    {
      "epoch": 0.6417612573281776,
      "grad_norm": 0.0,
      "learning_rate": 6.009545497270952e-06,
      "loss": 0.9636,
      "step": 5145
    },
    {
      "epoch": 0.6418859922664338,
      "grad_norm": 0.0,
      "learning_rate": 6.005841316669996e-06,
      "loss": 0.9091,
      "step": 5146
    },
    {
      "epoch": 0.64201072720469,
      "grad_norm": 0.0,
      "learning_rate": 6.00213778801643e-06,
      "loss": 0.9238,
      "step": 5147
    },
    {
      "epoch": 0.6421354621429463,
      "grad_norm": 0.0,
      "learning_rate": 5.9984349119147635e-06,
      "loss": 0.93,
      "step": 5148
    },
    {
      "epoch": 0.6422601970812024,
      "grad_norm": 0.0,
      "learning_rate": 5.994732688969396e-06,
      "loss": 0.9465,
      "step": 5149
    },
    {
      "epoch": 0.6423849320194587,
      "grad_norm": 0.0,
      "learning_rate": 5.991031119784627e-06,
      "loss": 0.9534,
      "step": 5150
    },
    {
      "epoch": 0.6425096669577148,
      "grad_norm": 0.0,
      "learning_rate": 5.987330204964641e-06,
      "loss": 0.8806,
      "step": 5151
    },
    {
      "epoch": 0.6426344018959711,
      "grad_norm": 0.0,
      "learning_rate": 5.983629945113529e-06,
      "loss": 0.9138,
      "step": 5152
    },
    {
      "epoch": 0.6427591368342273,
      "grad_norm": 0.0,
      "learning_rate": 5.979930340835265e-06,
      "loss": 0.9194,
      "step": 5153
    },
    {
      "epoch": 0.6428838717724835,
      "grad_norm": 0.0,
      "learning_rate": 5.976231392733707e-06,
      "loss": 0.8898,
      "step": 5154
    },
    {
      "epoch": 0.6430086067107397,
      "grad_norm": 0.0,
      "learning_rate": 5.97253310141263e-06,
      "loss": 0.9098,
      "step": 5155
    },
    {
      "epoch": 0.6431333416489959,
      "grad_norm": 0.0,
      "learning_rate": 5.968835467475679e-06,
      "loss": 0.9265,
      "step": 5156
    },
    {
      "epoch": 0.6432580765872521,
      "grad_norm": 0.0,
      "learning_rate": 5.965138491526408e-06,
      "loss": 0.9285,
      "step": 5157
    },
    {
      "epoch": 0.6433828115255082,
      "grad_norm": 0.0,
      "learning_rate": 5.961442174168253e-06,
      "loss": 0.8771,
      "step": 5158
    },
    {
      "epoch": 0.6435075464637645,
      "grad_norm": 0.0,
      "learning_rate": 5.957746516004543e-06,
      "loss": 0.9386,
      "step": 5159
    },
    {
      "epoch": 0.6436322814020207,
      "grad_norm": 0.0,
      "learning_rate": 5.954051517638511e-06,
      "loss": 0.9123,
      "step": 5160
    },
    {
      "epoch": 0.6437570163402769,
      "grad_norm": 0.0,
      "learning_rate": 5.950357179673264e-06,
      "loss": 0.976,
      "step": 5161
    },
    {
      "epoch": 0.6438817512785331,
      "grad_norm": 0.0,
      "learning_rate": 5.94666350271182e-06,
      "loss": 0.9599,
      "step": 5162
    },
    {
      "epoch": 0.6440064862167894,
      "grad_norm": 0.0,
      "learning_rate": 5.942970487357073e-06,
      "loss": 0.9413,
      "step": 5163
    },
    {
      "epoch": 0.6441312211550455,
      "grad_norm": 0.0,
      "learning_rate": 5.939278134211824e-06,
      "loss": 0.935,
      "step": 5164
    },
    {
      "epoch": 0.6442559560933018,
      "grad_norm": 0.0,
      "learning_rate": 5.935586443878754e-06,
      "loss": 0.905,
      "step": 5165
    },
    {
      "epoch": 0.6443806910315579,
      "grad_norm": 0.0,
      "learning_rate": 5.931895416960436e-06,
      "loss": 0.9522,
      "step": 5166
    },
    {
      "epoch": 0.6445054259698142,
      "grad_norm": 0.0,
      "learning_rate": 5.928205054059345e-06,
      "loss": 0.9306,
      "step": 5167
    },
    {
      "epoch": 0.6446301609080703,
      "grad_norm": 0.0,
      "learning_rate": 5.924515355777833e-06,
      "loss": 0.9586,
      "step": 5168
    },
    {
      "epoch": 0.6447548958463265,
      "grad_norm": 0.0,
      "learning_rate": 5.920826322718165e-06,
      "loss": 0.9439,
      "step": 5169
    },
    {
      "epoch": 0.6448796307845828,
      "grad_norm": 0.0,
      "learning_rate": 5.917137955482474e-06,
      "loss": 0.9056,
      "step": 5170
    },
    {
      "epoch": 0.6450043657228389,
      "grad_norm": 0.0,
      "learning_rate": 5.9134502546727944e-06,
      "loss": 0.9332,
      "step": 5171
    },
    {
      "epoch": 0.6451291006610952,
      "grad_norm": 0.0,
      "learning_rate": 5.909763220891057e-06,
      "loss": 0.8942,
      "step": 5172
    },
    {
      "epoch": 0.6452538355993513,
      "grad_norm": 0.0,
      "learning_rate": 5.9060768547390746e-06,
      "loss": 0.9971,
      "step": 5173
    },
    {
      "epoch": 0.6453785705376076,
      "grad_norm": 0.0,
      "learning_rate": 5.902391156818558e-06,
      "loss": 0.9386,
      "step": 5174
    },
    {
      "epoch": 0.6455033054758638,
      "grad_norm": 0.0,
      "learning_rate": 5.898706127731106e-06,
      "loss": 0.9404,
      "step": 5175
    },
    {
      "epoch": 0.64562804041412,
      "grad_norm": 0.0,
      "learning_rate": 5.895021768078204e-06,
      "loss": 0.9385,
      "step": 5176
    },
    {
      "epoch": 0.6457527753523762,
      "grad_norm": 0.0,
      "learning_rate": 5.891338078461238e-06,
      "loss": 0.9364,
      "step": 5177
    },
    {
      "epoch": 0.6458775102906325,
      "grad_norm": 0.0,
      "learning_rate": 5.887655059481473e-06,
      "loss": 0.9091,
      "step": 5178
    },
    {
      "epoch": 0.6460022452288886,
      "grad_norm": 0.0,
      "learning_rate": 5.8839727117400795e-06,
      "loss": 0.9343,
      "step": 5179
    },
    {
      "epoch": 0.6461269801671449,
      "grad_norm": 0.0,
      "learning_rate": 5.880291035838097e-06,
      "loss": 0.9518,
      "step": 5180
    },
    {
      "epoch": 0.646251715105401,
      "grad_norm": 0.0,
      "learning_rate": 5.87661003237648e-06,
      "loss": 0.9435,
      "step": 5181
    },
    {
      "epoch": 0.6463764500436572,
      "grad_norm": 0.0,
      "learning_rate": 5.872929701956054e-06,
      "loss": 0.9032,
      "step": 5182
    },
    {
      "epoch": 0.6465011849819134,
      "grad_norm": 0.0,
      "learning_rate": 5.869250045177542e-06,
      "loss": 0.9465,
      "step": 5183
    },
    {
      "epoch": 0.6466259199201696,
      "grad_norm": 0.0,
      "learning_rate": 5.865571062641562e-06,
      "loss": 0.9252,
      "step": 5184
    },
    {
      "epoch": 0.6467506548584259,
      "grad_norm": 0.0,
      "learning_rate": 5.8618927549486095e-06,
      "loss": 0.9348,
      "step": 5185
    },
    {
      "epoch": 0.646875389796682,
      "grad_norm": 0.0,
      "learning_rate": 5.858215122699082e-06,
      "loss": 0.8962,
      "step": 5186
    },
    {
      "epoch": 0.6470001247349383,
      "grad_norm": 0.0,
      "learning_rate": 5.854538166493262e-06,
      "loss": 0.9022,
      "step": 5187
    },
    {
      "epoch": 0.6471248596731944,
      "grad_norm": 0.0,
      "learning_rate": 5.850861886931314e-06,
      "loss": 0.8805,
      "step": 5188
    },
    {
      "epoch": 0.6472495946114507,
      "grad_norm": 0.0,
      "learning_rate": 5.847186284613312e-06,
      "loss": 0.9351,
      "step": 5189
    },
    {
      "epoch": 0.6473743295497069,
      "grad_norm": 0.0,
      "learning_rate": 5.8435113601391935e-06,
      "loss": 0.9432,
      "step": 5190
    },
    {
      "epoch": 0.6474990644879631,
      "grad_norm": 0.0,
      "learning_rate": 5.839837114108811e-06,
      "loss": 0.9227,
      "step": 5191
    },
    {
      "epoch": 0.6476237994262193,
      "grad_norm": 0.0,
      "learning_rate": 5.836163547121887e-06,
      "loss": 0.9532,
      "step": 5192
    },
    {
      "epoch": 0.6477485343644754,
      "grad_norm": 0.0,
      "learning_rate": 5.832490659778041e-06,
      "loss": 0.9274,
      "step": 5193
    },
    {
      "epoch": 0.6478732693027317,
      "grad_norm": 0.0,
      "learning_rate": 5.828818452676786e-06,
      "loss": 0.9436,
      "step": 5194
    },
    {
      "epoch": 0.6479980042409879,
      "grad_norm": 0.0,
      "learning_rate": 5.825146926417504e-06,
      "loss": 0.8971,
      "step": 5195
    },
    {
      "epoch": 0.6481227391792441,
      "grad_norm": 0.0,
      "learning_rate": 5.821476081599502e-06,
      "loss": 0.9939,
      "step": 5196
    },
    {
      "epoch": 0.6482474741175003,
      "grad_norm": 0.0,
      "learning_rate": 5.817805918821937e-06,
      "loss": 0.8779,
      "step": 5197
    },
    {
      "epoch": 0.6483722090557565,
      "grad_norm": 0.0,
      "learning_rate": 5.814136438683879e-06,
      "loss": 0.9384,
      "step": 5198
    },
    {
      "epoch": 0.6484969439940127,
      "grad_norm": 0.0,
      "learning_rate": 5.810467641784285e-06,
      "loss": 0.935,
      "step": 5199
    },
    {
      "epoch": 0.648621678932269,
      "grad_norm": 0.0,
      "learning_rate": 5.806799528721985e-06,
      "loss": 0.9694,
      "step": 5200
    },
    {
      "epoch": 0.6487464138705251,
      "grad_norm": 0.0,
      "learning_rate": 5.80313210009571e-06,
      "loss": 0.9065,
      "step": 5201
    },
    {
      "epoch": 0.6488711488087814,
      "grad_norm": 0.0,
      "learning_rate": 5.7994653565040806e-06,
      "loss": 0.9526,
      "step": 5202
    },
    {
      "epoch": 0.6489958837470375,
      "grad_norm": 0.0,
      "learning_rate": 5.795799298545599e-06,
      "loss": 0.8874,
      "step": 5203
    },
    {
      "epoch": 0.6491206186852938,
      "grad_norm": 0.0,
      "learning_rate": 5.792133926818664e-06,
      "loss": 0.9618,
      "step": 5204
    },
    {
      "epoch": 0.64924535362355,
      "grad_norm": 0.0,
      "learning_rate": 5.788469241921546e-06,
      "loss": 0.9105,
      "step": 5205
    },
    {
      "epoch": 0.6493700885618061,
      "grad_norm": 0.0,
      "learning_rate": 5.7848052444524215e-06,
      "loss": 0.9262,
      "step": 5206
    },
    {
      "epoch": 0.6494948235000624,
      "grad_norm": 0.0,
      "learning_rate": 5.781141935009344e-06,
      "loss": 0.9298,
      "step": 5207
    },
    {
      "epoch": 0.6496195584383185,
      "grad_norm": 0.0,
      "learning_rate": 5.777479314190266e-06,
      "loss": 0.9188,
      "step": 5208
    },
    {
      "epoch": 0.6497442933765748,
      "grad_norm": 0.0,
      "learning_rate": 5.773817382593008e-06,
      "loss": 0.9482,
      "step": 5209
    },
    {
      "epoch": 0.649869028314831,
      "grad_norm": 0.0,
      "learning_rate": 5.770156140815293e-06,
      "loss": 0.9303,
      "step": 5210
    },
    {
      "epoch": 0.6499937632530872,
      "grad_norm": 0.0,
      "learning_rate": 5.7664955894547345e-06,
      "loss": 0.9316,
      "step": 5211
    },
    {
      "epoch": 0.6501184981913434,
      "grad_norm": 0.0,
      "learning_rate": 5.762835729108815e-06,
      "loss": 0.9449,
      "step": 5212
    },
    {
      "epoch": 0.6502432331295996,
      "grad_norm": 0.0,
      "learning_rate": 5.75917656037493e-06,
      "loss": 0.9176,
      "step": 5213
    },
    {
      "epoch": 0.6503679680678558,
      "grad_norm": 0.0,
      "learning_rate": 5.755518083850338e-06,
      "loss": 0.9307,
      "step": 5214
    },
    {
      "epoch": 0.6504927030061121,
      "grad_norm": 0.0,
      "learning_rate": 5.751860300132195e-06,
      "loss": 0.9102,
      "step": 5215
    },
    {
      "epoch": 0.6506174379443682,
      "grad_norm": 0.0,
      "learning_rate": 5.7482032098175515e-06,
      "loss": 0.9304,
      "step": 5216
    },
    {
      "epoch": 0.6507421728826244,
      "grad_norm": 0.0,
      "learning_rate": 5.744546813503327e-06,
      "loss": 0.9393,
      "step": 5217
    },
    {
      "epoch": 0.6508669078208806,
      "grad_norm": 0.0,
      "learning_rate": 5.740891111786342e-06,
      "loss": 0.926,
      "step": 5218
    },
    {
      "epoch": 0.6509916427591368,
      "grad_norm": 0.0,
      "learning_rate": 5.7372361052633e-06,
      "loss": 0.8736,
      "step": 5219
    },
    {
      "epoch": 0.6511163776973931,
      "grad_norm": 0.0,
      "learning_rate": 5.733581794530793e-06,
      "loss": 0.933,
      "step": 5220
    },
    {
      "epoch": 0.6512411126356492,
      "grad_norm": 0.0,
      "learning_rate": 5.729928180185288e-06,
      "loss": 0.9216,
      "step": 5221
    },
    {
      "epoch": 0.6513658475739055,
      "grad_norm": 0.0,
      "learning_rate": 5.7262752628231525e-06,
      "loss": 0.8986,
      "step": 5222
    },
    {
      "epoch": 0.6514905825121616,
      "grad_norm": 0.0,
      "learning_rate": 5.722623043040634e-06,
      "loss": 0.8854,
      "step": 5223
    },
    {
      "epoch": 0.6516153174504179,
      "grad_norm": 0.0,
      "learning_rate": 5.718971521433867e-06,
      "loss": 0.8735,
      "step": 5224
    },
    {
      "epoch": 0.651740052388674,
      "grad_norm": 0.0,
      "learning_rate": 5.715320698598875e-06,
      "loss": 0.9413,
      "step": 5225
    },
    {
      "epoch": 0.6518647873269303,
      "grad_norm": 0.0,
      "learning_rate": 5.711670575131558e-06,
      "loss": 0.9172,
      "step": 5226
    },
    {
      "epoch": 0.6519895222651865,
      "grad_norm": 0.0,
      "learning_rate": 5.708021151627712e-06,
      "loss": 0.942,
      "step": 5227
    },
    {
      "epoch": 0.6521142572034427,
      "grad_norm": 0.0,
      "learning_rate": 5.704372428683016e-06,
      "loss": 0.9743,
      "step": 5228
    },
    {
      "epoch": 0.6522389921416989,
      "grad_norm": 0.0,
      "learning_rate": 5.700724406893025e-06,
      "loss": 0.9002,
      "step": 5229
    },
    {
      "epoch": 0.652363727079955,
      "grad_norm": 0.0,
      "learning_rate": 5.697077086853205e-06,
      "loss": 0.9444,
      "step": 5230
    },
    {
      "epoch": 0.6524884620182113,
      "grad_norm": 0.0,
      "learning_rate": 5.693430469158878e-06,
      "loss": 0.8815,
      "step": 5231
    },
    {
      "epoch": 0.6526131969564675,
      "grad_norm": 0.0,
      "learning_rate": 5.689784554405258e-06,
      "loss": 0.9128,
      "step": 5232
    },
    {
      "epoch": 0.6527379318947237,
      "grad_norm": 0.0,
      "learning_rate": 5.686139343187468e-06,
      "loss": 0.8808,
      "step": 5233
    },
    {
      "epoch": 0.6528626668329799,
      "grad_norm": 0.0,
      "learning_rate": 5.682494836100485e-06,
      "loss": 0.9132,
      "step": 5234
    },
    {
      "epoch": 0.6529874017712362,
      "grad_norm": 0.0,
      "learning_rate": 5.6788510337391875e-06,
      "loss": 0.861,
      "step": 5235
    },
    {
      "epoch": 0.6531121367094923,
      "grad_norm": 0.0,
      "learning_rate": 5.675207936698337e-06,
      "loss": 0.9151,
      "step": 5236
    },
    {
      "epoch": 0.6532368716477486,
      "grad_norm": 0.0,
      "learning_rate": 5.671565545572584e-06,
      "loss": 0.9229,
      "step": 5237
    },
    {
      "epoch": 0.6533616065860047,
      "grad_norm": 0.0,
      "learning_rate": 5.667923860956448e-06,
      "loss": 0.9392,
      "step": 5238
    },
    {
      "epoch": 0.653486341524261,
      "grad_norm": 0.0,
      "learning_rate": 5.66428288344435e-06,
      "loss": 0.925,
      "step": 5239
    },
    {
      "epoch": 0.6536110764625171,
      "grad_norm": 0.0,
      "learning_rate": 5.660642613630587e-06,
      "loss": 0.9286,
      "step": 5240
    },
    {
      "epoch": 0.6537358114007733,
      "grad_norm": 0.0,
      "learning_rate": 5.657003052109344e-06,
      "loss": 0.9113,
      "step": 5241
    },
    {
      "epoch": 0.6538605463390296,
      "grad_norm": 0.0,
      "learning_rate": 5.6533641994746936e-06,
      "loss": 0.9329,
      "step": 5242
    },
    {
      "epoch": 0.6539852812772857,
      "grad_norm": 0.0,
      "learning_rate": 5.649726056320579e-06,
      "loss": 0.9539,
      "step": 5243
    },
    {
      "epoch": 0.654110016215542,
      "grad_norm": 0.0,
      "learning_rate": 5.6460886232408445e-06,
      "loss": 0.8781,
      "step": 5244
    },
    {
      "epoch": 0.6542347511537981,
      "grad_norm": 0.0,
      "learning_rate": 5.642451900829209e-06,
      "loss": 0.8897,
      "step": 5245
    },
    {
      "epoch": 0.6543594860920544,
      "grad_norm": 0.0,
      "learning_rate": 5.63881588967927e-06,
      "loss": 0.9473,
      "step": 5246
    },
    {
      "epoch": 0.6544842210303106,
      "grad_norm": 0.0,
      "learning_rate": 5.63518059038453e-06,
      "loss": 0.9048,
      "step": 5247
    },
    {
      "epoch": 0.6546089559685668,
      "grad_norm": 0.0,
      "learning_rate": 5.631546003538355e-06,
      "loss": 0.9013,
      "step": 5248
    },
    {
      "epoch": 0.654733690906823,
      "grad_norm": 0.0,
      "learning_rate": 5.627912129733992e-06,
      "loss": 0.9169,
      "step": 5249
    },
    {
      "epoch": 0.6548584258450793,
      "grad_norm": 0.0,
      "learning_rate": 5.624278969564597e-06,
      "loss": 0.8939,
      "step": 5250
    },
    {
      "epoch": 0.6549831607833354,
      "grad_norm": 0.0,
      "learning_rate": 5.620646523623181e-06,
      "loss": 0.9679,
      "step": 5251
    },
    {
      "epoch": 0.6551078957215917,
      "grad_norm": 0.0,
      "learning_rate": 5.617014792502657e-06,
      "loss": 0.8899,
      "step": 5252
    },
    {
      "epoch": 0.6552326306598478,
      "grad_norm": 0.0,
      "learning_rate": 5.613383776795811e-06,
      "loss": 0.9479,
      "step": 5253
    },
    {
      "epoch": 0.655357365598104,
      "grad_norm": 0.0,
      "learning_rate": 5.609753477095324e-06,
      "loss": 0.9063,
      "step": 5254
    },
    {
      "epoch": 0.6554821005363602,
      "grad_norm": 0.0,
      "learning_rate": 5.606123893993742e-06,
      "loss": 0.9154,
      "step": 5255
    },
    {
      "epoch": 0.6556068354746164,
      "grad_norm": 0.0,
      "learning_rate": 5.602495028083511e-06,
      "loss": 0.9453,
      "step": 5256
    },
    {
      "epoch": 0.6557315704128727,
      "grad_norm": 0.0,
      "learning_rate": 5.598866879956955e-06,
      "loss": 0.9258,
      "step": 5257
    },
    {
      "epoch": 0.6558563053511288,
      "grad_norm": 0.0,
      "learning_rate": 5.5952394502062694e-06,
      "loss": 0.9059,
      "step": 5258
    },
    {
      "epoch": 0.6559810402893851,
      "grad_norm": 0.0,
      "learning_rate": 5.591612739423555e-06,
      "loss": 0.9085,
      "step": 5259
    },
    {
      "epoch": 0.6561057752276412,
      "grad_norm": 0.0,
      "learning_rate": 5.587986748200773e-06,
      "loss": 0.8987,
      "step": 5260
    },
    {
      "epoch": 0.6562305101658975,
      "grad_norm": 0.0,
      "learning_rate": 5.584361477129779e-06,
      "loss": 0.9408,
      "step": 5261
    },
    {
      "epoch": 0.6563552451041537,
      "grad_norm": 0.0,
      "learning_rate": 5.580736926802315e-06,
      "loss": 0.8975,
      "step": 5262
    },
    {
      "epoch": 0.6564799800424099,
      "grad_norm": 0.0,
      "learning_rate": 5.5771130978099896e-06,
      "loss": 0.8984,
      "step": 5263
    },
    {
      "epoch": 0.6566047149806661,
      "grad_norm": 0.0,
      "learning_rate": 5.573489990744305e-06,
      "loss": 0.933,
      "step": 5264
    },
    {
      "epoch": 0.6567294499189222,
      "grad_norm": 0.0,
      "learning_rate": 5.569867606196652e-06,
      "loss": 0.9423,
      "step": 5265
    },
    {
      "epoch": 0.6568541848571785,
      "grad_norm": 0.0,
      "learning_rate": 5.566245944758279e-06,
      "loss": 0.9248,
      "step": 5266
    },
    {
      "epoch": 0.6569789197954347,
      "grad_norm": 0.0,
      "learning_rate": 5.562625007020351e-06,
      "loss": 0.9762,
      "step": 5267
    },
    {
      "epoch": 0.6571036547336909,
      "grad_norm": 0.0,
      "learning_rate": 5.559004793573883e-06,
      "loss": 0.9146,
      "step": 5268
    },
    {
      "epoch": 0.6572283896719471,
      "grad_norm": 0.0,
      "learning_rate": 5.555385305009791e-06,
      "loss": 0.94,
      "step": 5269
    },
    {
      "epoch": 0.6573531246102033,
      "grad_norm": 0.0,
      "learning_rate": 5.5517665419188635e-06,
      "loss": 0.9202,
      "step": 5270
    },
    {
      "epoch": 0.6574778595484595,
      "grad_norm": 0.0,
      "learning_rate": 5.548148504891779e-06,
      "loss": 0.9087,
      "step": 5271
    },
    {
      "epoch": 0.6576025944867158,
      "grad_norm": 0.0,
      "learning_rate": 5.5445311945190875e-06,
      "loss": 0.9177,
      "step": 5272
    },
    {
      "epoch": 0.6577273294249719,
      "grad_norm": 0.0,
      "learning_rate": 5.540914611391226e-06,
      "loss": 0.9765,
      "step": 5273
    },
    {
      "epoch": 0.6578520643632282,
      "grad_norm": 0.0,
      "learning_rate": 5.5372987560985155e-06,
      "loss": 0.9277,
      "step": 5274
    },
    {
      "epoch": 0.6579767993014843,
      "grad_norm": 0.0,
      "learning_rate": 5.533683629231146e-06,
      "loss": 0.858,
      "step": 5275
    },
    {
      "epoch": 0.6581015342397406,
      "grad_norm": 0.0,
      "learning_rate": 5.5300692313792095e-06,
      "loss": 0.947,
      "step": 5276
    },
    {
      "epoch": 0.6582262691779968,
      "grad_norm": 0.0,
      "learning_rate": 5.5264555631326625e-06,
      "loss": 0.9492,
      "step": 5277
    },
    {
      "epoch": 0.6583510041162529,
      "grad_norm": 0.0,
      "learning_rate": 5.522842625081335e-06,
      "loss": 0.8912,
      "step": 5278
    },
    {
      "epoch": 0.6584757390545092,
      "grad_norm": 0.0,
      "learning_rate": 5.5192304178149695e-06,
      "loss": 0.8956,
      "step": 5279
    },
    {
      "epoch": 0.6586004739927653,
      "grad_norm": 0.0,
      "learning_rate": 5.515618941923154e-06,
      "loss": 0.9439,
      "step": 5280
    },
    {
      "epoch": 0.6587252089310216,
      "grad_norm": 0.0,
      "learning_rate": 5.512008197995379e-06,
      "loss": 0.9176,
      "step": 5281
    },
    {
      "epoch": 0.6588499438692778,
      "grad_norm": 0.0,
      "learning_rate": 5.508398186621011e-06,
      "loss": 0.9071,
      "step": 5282
    },
    {
      "epoch": 0.658974678807534,
      "grad_norm": 0.0,
      "learning_rate": 5.504788908389287e-06,
      "loss": 0.9169,
      "step": 5283
    },
    {
      "epoch": 0.6590994137457902,
      "grad_norm": 0.0,
      "learning_rate": 5.501180363889338e-06,
      "loss": 0.922,
      "step": 5284
    },
    {
      "epoch": 0.6592241486840464,
      "grad_norm": 0.0,
      "learning_rate": 5.497572553710167e-06,
      "loss": 0.9315,
      "step": 5285
    },
    {
      "epoch": 0.6593488836223026,
      "grad_norm": 0.0,
      "learning_rate": 5.493965478440662e-06,
      "loss": 0.9318,
      "step": 5286
    },
    {
      "epoch": 0.6594736185605589,
      "grad_norm": 0.0,
      "learning_rate": 5.490359138669588e-06,
      "loss": 0.909,
      "step": 5287
    },
    {
      "epoch": 0.659598353498815,
      "grad_norm": 0.0,
      "learning_rate": 5.486753534985591e-06,
      "loss": 0.9249,
      "step": 5288
    },
    {
      "epoch": 0.6597230884370712,
      "grad_norm": 0.0,
      "learning_rate": 5.483148667977194e-06,
      "loss": 0.9413,
      "step": 5289
    },
    {
      "epoch": 0.6598478233753274,
      "grad_norm": 0.0,
      "learning_rate": 5.479544538232804e-06,
      "loss": 0.9541,
      "step": 5290
    },
    {
      "epoch": 0.6599725583135836,
      "grad_norm": 0.0,
      "learning_rate": 5.4759411463407075e-06,
      "loss": 0.9458,
      "step": 5291
    },
    {
      "epoch": 0.6600972932518399,
      "grad_norm": 0.0,
      "learning_rate": 5.47233849288906e-06,
      "loss": 0.9066,
      "step": 5292
    },
    {
      "epoch": 0.660222028190096,
      "grad_norm": 0.0,
      "learning_rate": 5.468736578465918e-06,
      "loss": 0.9135,
      "step": 5293
    },
    {
      "epoch": 0.6603467631283523,
      "grad_norm": 0.0,
      "learning_rate": 5.465135403659201e-06,
      "loss": 0.9516,
      "step": 5294
    },
    {
      "epoch": 0.6604714980666084,
      "grad_norm": 0.0,
      "learning_rate": 5.461534969056701e-06,
      "loss": 0.9331,
      "step": 5295
    },
    {
      "epoch": 0.6605962330048647,
      "grad_norm": 0.0,
      "learning_rate": 5.457935275246116e-06,
      "loss": 0.8872,
      "step": 5296
    },
    {
      "epoch": 0.6607209679431209,
      "grad_norm": 0.0,
      "learning_rate": 5.454336322814995e-06,
      "loss": 0.9489,
      "step": 5297
    },
    {
      "epoch": 0.6608457028813771,
      "grad_norm": 0.0,
      "learning_rate": 5.450738112350782e-06,
      "loss": 0.9075,
      "step": 5298
    },
    {
      "epoch": 0.6609704378196333,
      "grad_norm": 0.0,
      "learning_rate": 5.447140644440798e-06,
      "loss": 0.9418,
      "step": 5299
    },
    {
      "epoch": 0.6610951727578895,
      "grad_norm": 0.0,
      "learning_rate": 5.443543919672236e-06,
      "loss": 0.9279,
      "step": 5300
    },
    {
      "epoch": 0.6612199076961457,
      "grad_norm": 0.0,
      "learning_rate": 5.4399479386321714e-06,
      "loss": 0.9027,
      "step": 5301
    },
    {
      "epoch": 0.6613446426344018,
      "grad_norm": 0.0,
      "learning_rate": 5.436352701907563e-06,
      "loss": 0.966,
      "step": 5302
    },
    {
      "epoch": 0.6614693775726581,
      "grad_norm": 0.0,
      "learning_rate": 5.432758210085247e-06,
      "loss": 0.8905,
      "step": 5303
    },
    {
      "epoch": 0.6615941125109143,
      "grad_norm": 0.0,
      "learning_rate": 5.429164463751921e-06,
      "loss": 0.8639,
      "step": 5304
    },
    {
      "epoch": 0.6617188474491705,
      "grad_norm": 0.0,
      "learning_rate": 5.4255714634941934e-06,
      "loss": 0.9463,
      "step": 5305
    },
    {
      "epoch": 0.6618435823874267,
      "grad_norm": 0.0,
      "learning_rate": 5.42197920989852e-06,
      "loss": 0.9027,
      "step": 5306
    },
    {
      "epoch": 0.661968317325683,
      "grad_norm": 0.0,
      "learning_rate": 5.418387703551251e-06,
      "loss": 0.9243,
      "step": 5307
    },
    {
      "epoch": 0.6620930522639391,
      "grad_norm": 0.0,
      "learning_rate": 5.414796945038614e-06,
      "loss": 0.9335,
      "step": 5308
    },
    {
      "epoch": 0.6622177872021954,
      "grad_norm": 0.0,
      "learning_rate": 5.411206934946703e-06,
      "loss": 0.9131,
      "step": 5309
    },
    {
      "epoch": 0.6623425221404515,
      "grad_norm": 0.0,
      "learning_rate": 5.407617673861505e-06,
      "loss": 0.9224,
      "step": 5310
    },
    {
      "epoch": 0.6624672570787078,
      "grad_norm": 0.0,
      "learning_rate": 5.404029162368878e-06,
      "loss": 0.9101,
      "step": 5311
    },
    {
      "epoch": 0.662591992016964,
      "grad_norm": 0.0,
      "learning_rate": 5.400441401054548e-06,
      "loss": 0.9368,
      "step": 5312
    },
    {
      "epoch": 0.6627167269552201,
      "grad_norm": 0.0,
      "learning_rate": 5.3968543905041446e-06,
      "loss": 0.9087,
      "step": 5313
    },
    {
      "epoch": 0.6628414618934764,
      "grad_norm": 0.0,
      "learning_rate": 5.393268131303144e-06,
      "loss": 0.9178,
      "step": 5314
    },
    {
      "epoch": 0.6629661968317325,
      "grad_norm": 0.0,
      "learning_rate": 5.38968262403692e-06,
      "loss": 0.9019,
      "step": 5315
    },
    {
      "epoch": 0.6630909317699888,
      "grad_norm": 0.0,
      "learning_rate": 5.386097869290722e-06,
      "loss": 0.9364,
      "step": 5316
    },
    {
      "epoch": 0.6632156667082449,
      "grad_norm": 0.0,
      "learning_rate": 5.382513867649663e-06,
      "loss": 0.9122,
      "step": 5317
    },
    {
      "epoch": 0.6633404016465012,
      "grad_norm": 0.0,
      "learning_rate": 5.378930619698747e-06,
      "loss": 1.0189,
      "step": 5318
    },
    {
      "epoch": 0.6634651365847574,
      "grad_norm": 0.0,
      "learning_rate": 5.37534812602285e-06,
      "loss": 0.9303,
      "step": 5319
    },
    {
      "epoch": 0.6635898715230136,
      "grad_norm": 0.0,
      "learning_rate": 5.3717663872067295e-06,
      "loss": 0.9296,
      "step": 5320
    },
    {
      "epoch": 0.6637146064612698,
      "grad_norm": 0.0,
      "learning_rate": 5.3681854038350044e-06,
      "loss": 0.8876,
      "step": 5321
    },
    {
      "epoch": 0.663839341399526,
      "grad_norm": 0.0,
      "learning_rate": 5.364605176492196e-06,
      "loss": 0.9211,
      "step": 5322
    },
    {
      "epoch": 0.6639640763377822,
      "grad_norm": 0.0,
      "learning_rate": 5.3610257057626765e-06,
      "loss": 0.9373,
      "step": 5323
    },
    {
      "epoch": 0.6640888112760385,
      "grad_norm": 0.0,
      "learning_rate": 5.3574469922307105e-06,
      "loss": 0.8926,
      "step": 5324
    },
    {
      "epoch": 0.6642135462142946,
      "grad_norm": 0.0,
      "learning_rate": 5.353869036480437e-06,
      "loss": 0.9649,
      "step": 5325
    },
    {
      "epoch": 0.6643382811525508,
      "grad_norm": 0.0,
      "learning_rate": 5.35029183909586e-06,
      "loss": 0.8905,
      "step": 5326
    },
    {
      "epoch": 0.664463016090807,
      "grad_norm": 0.0,
      "learning_rate": 5.346715400660875e-06,
      "loss": 0.8629,
      "step": 5327
    },
    {
      "epoch": 0.6645877510290632,
      "grad_norm": 0.0,
      "learning_rate": 5.343139721759248e-06,
      "loss": 0.9118,
      "step": 5328
    },
    {
      "epoch": 0.6647124859673195,
      "grad_norm": 0.0,
      "learning_rate": 5.339564802974615e-06,
      "loss": 0.9056,
      "step": 5329
    },
    {
      "epoch": 0.6648372209055756,
      "grad_norm": 0.0,
      "learning_rate": 5.335990644890493e-06,
      "loss": 0.884,
      "step": 5330
    },
    {
      "epoch": 0.6649619558438319,
      "grad_norm": 0.0,
      "learning_rate": 5.332417248090278e-06,
      "loss": 0.9268,
      "step": 5331
    },
    {
      "epoch": 0.665086690782088,
      "grad_norm": 0.0,
      "learning_rate": 5.328844613157236e-06,
      "loss": 0.8844,
      "step": 5332
    },
    {
      "epoch": 0.6652114257203443,
      "grad_norm": 0.0,
      "learning_rate": 5.325272740674517e-06,
      "loss": 0.9096,
      "step": 5333
    },
    {
      "epoch": 0.6653361606586005,
      "grad_norm": 0.0,
      "learning_rate": 5.321701631225133e-06,
      "loss": 0.9037,
      "step": 5334
    },
    {
      "epoch": 0.6654608955968567,
      "grad_norm": 0.0,
      "learning_rate": 5.318131285391981e-06,
      "loss": 0.9876,
      "step": 5335
    },
    {
      "epoch": 0.6655856305351129,
      "grad_norm": 0.0,
      "learning_rate": 5.314561703757832e-06,
      "loss": 0.9016,
      "step": 5336
    },
    {
      "epoch": 0.665710365473369,
      "grad_norm": 0.0,
      "learning_rate": 5.310992886905337e-06,
      "loss": 0.9746,
      "step": 5337
    },
    {
      "epoch": 0.6658351004116253,
      "grad_norm": 0.0,
      "learning_rate": 5.307424835417003e-06,
      "loss": 0.9462,
      "step": 5338
    },
    {
      "epoch": 0.6659598353498815,
      "grad_norm": 0.0,
      "learning_rate": 5.3038575498752444e-06,
      "loss": 0.9368,
      "step": 5339
    },
    {
      "epoch": 0.6660845702881377,
      "grad_norm": 0.0,
      "learning_rate": 5.300291030862322e-06,
      "loss": 0.9147,
      "step": 5340
    },
    {
      "epoch": 0.6662093052263939,
      "grad_norm": 0.0,
      "learning_rate": 5.296725278960375e-06,
      "loss": 0.9397,
      "step": 5341
    },
    {
      "epoch": 0.6663340401646501,
      "grad_norm": 0.0,
      "learning_rate": 5.293160294751438e-06,
      "loss": 0.9542,
      "step": 5342
    },
    {
      "epoch": 0.6664587751029063,
      "grad_norm": 0.0,
      "learning_rate": 5.289596078817396e-06,
      "loss": 0.9134,
      "step": 5343
    },
    {
      "epoch": 0.6665835100411626,
      "grad_norm": 0.0,
      "learning_rate": 5.286032631740023e-06,
      "loss": 0.9215,
      "step": 5344
    },
    {
      "epoch": 0.6667082449794187,
      "grad_norm": 0.0,
      "learning_rate": 5.282469954100968e-06,
      "loss": 0.9446,
      "step": 5345
    },
    {
      "epoch": 0.666832979917675,
      "grad_norm": 0.0,
      "learning_rate": 5.278908046481739e-06,
      "loss": 0.9566,
      "step": 5346
    },
    {
      "epoch": 0.6669577148559311,
      "grad_norm": 0.0,
      "learning_rate": 5.275346909463737e-06,
      "loss": 0.9377,
      "step": 5347
    },
    {
      "epoch": 0.6670824497941874,
      "grad_norm": 0.0,
      "learning_rate": 5.271786543628225e-06,
      "loss": 0.8979,
      "step": 5348
    },
    {
      "epoch": 0.6672071847324436,
      "grad_norm": 0.0,
      "learning_rate": 5.268226949556347e-06,
      "loss": 0.8913,
      "step": 5349
    },
    {
      "epoch": 0.6673319196706997,
      "grad_norm": 0.0,
      "learning_rate": 5.264668127829122e-06,
      "loss": 0.97,
      "step": 5350
    },
    {
      "epoch": 0.667456654608956,
      "grad_norm": 0.0,
      "learning_rate": 5.261110079027433e-06,
      "loss": 0.8922,
      "step": 5351
    },
    {
      "epoch": 0.6675813895472121,
      "grad_norm": 0.0,
      "learning_rate": 5.257552803732044e-06,
      "loss": 0.9172,
      "step": 5352
    },
    {
      "epoch": 0.6677061244854684,
      "grad_norm": 0.0,
      "learning_rate": 5.253996302523596e-06,
      "loss": 0.9109,
      "step": 5353
    },
    {
      "epoch": 0.6678308594237246,
      "grad_norm": 0.0,
      "learning_rate": 5.250440575982601e-06,
      "loss": 0.9607,
      "step": 5354
    },
    {
      "epoch": 0.6679555943619808,
      "grad_norm": 0.0,
      "learning_rate": 5.246885624689434e-06,
      "loss": 0.9237,
      "step": 5355
    },
    {
      "epoch": 0.668080329300237,
      "grad_norm": 0.0,
      "learning_rate": 5.243331449224359e-06,
      "loss": 0.911,
      "step": 5356
    },
    {
      "epoch": 0.6682050642384932,
      "grad_norm": 0.0,
      "learning_rate": 5.239778050167511e-06,
      "loss": 0.9371,
      "step": 5357
    },
    {
      "epoch": 0.6683297991767494,
      "grad_norm": 0.0,
      "learning_rate": 5.236225428098882e-06,
      "loss": 0.8861,
      "step": 5358
    },
    {
      "epoch": 0.6684545341150057,
      "grad_norm": 0.0,
      "learning_rate": 5.232673583598365e-06,
      "loss": 0.8691,
      "step": 5359
    },
    {
      "epoch": 0.6685792690532618,
      "grad_norm": 0.0,
      "learning_rate": 5.229122517245698e-06,
      "loss": 0.9249,
      "step": 5360
    },
    {
      "epoch": 0.668704003991518,
      "grad_norm": 0.0,
      "learning_rate": 5.2255722296205104e-06,
      "loss": 0.9056,
      "step": 5361
    },
    {
      "epoch": 0.6688287389297742,
      "grad_norm": 0.0,
      "learning_rate": 5.2220227213023e-06,
      "loss": 0.8859,
      "step": 5362
    },
    {
      "epoch": 0.6689534738680304,
      "grad_norm": 0.0,
      "learning_rate": 5.218473992870432e-06,
      "loss": 0.9037,
      "step": 5363
    },
    {
      "epoch": 0.6690782088062867,
      "grad_norm": 0.0,
      "learning_rate": 5.214926044904147e-06,
      "loss": 0.9441,
      "step": 5364
    },
    {
      "epoch": 0.6692029437445428,
      "grad_norm": 0.0,
      "learning_rate": 5.211378877982565e-06,
      "loss": 0.9552,
      "step": 5365
    },
    {
      "epoch": 0.6693276786827991,
      "grad_norm": 0.0,
      "learning_rate": 5.207832492684674e-06,
      "loss": 0.8454,
      "step": 5366
    },
    {
      "epoch": 0.6694524136210552,
      "grad_norm": 0.0,
      "learning_rate": 5.204286889589325e-06,
      "loss": 0.9179,
      "step": 5367
    },
    {
      "epoch": 0.6695771485593115,
      "grad_norm": 0.0,
      "learning_rate": 5.200742069275255e-06,
      "loss": 0.9022,
      "step": 5368
    },
    {
      "epoch": 0.6697018834975677,
      "grad_norm": 0.0,
      "learning_rate": 5.197198032321069e-06,
      "loss": 0.8773,
      "step": 5369
    },
    {
      "epoch": 0.6698266184358239,
      "grad_norm": 0.0,
      "learning_rate": 5.193654779305241e-06,
      "loss": 0.9066,
      "step": 5370
    },
    {
      "epoch": 0.6699513533740801,
      "grad_norm": 0.0,
      "learning_rate": 5.190112310806126e-06,
      "loss": 0.9315,
      "step": 5371
    },
    {
      "epoch": 0.6700760883123363,
      "grad_norm": 0.0,
      "learning_rate": 5.186570627401932e-06,
      "loss": 0.9305,
      "step": 5372
    },
    {
      "epoch": 0.6702008232505925,
      "grad_norm": 0.0,
      "learning_rate": 5.18302972967076e-06,
      "loss": 0.9117,
      "step": 5373
    },
    {
      "epoch": 0.6703255581888486,
      "grad_norm": 0.0,
      "learning_rate": 5.179489618190574e-06,
      "loss": 0.8818,
      "step": 5374
    },
    {
      "epoch": 0.6704502931271049,
      "grad_norm": 0.0,
      "learning_rate": 5.175950293539198e-06,
      "loss": 0.8803,
      "step": 5375
    },
    {
      "epoch": 0.6705750280653611,
      "grad_norm": 0.0,
      "learning_rate": 5.172411756294358e-06,
      "loss": 0.9047,
      "step": 5376
    },
    {
      "epoch": 0.6706997630036173,
      "grad_norm": 0.0,
      "learning_rate": 5.168874007033615e-06,
      "loss": 0.9762,
      "step": 5377
    },
    {
      "epoch": 0.6708244979418735,
      "grad_norm": 0.0,
      "learning_rate": 5.165337046334429e-06,
      "loss": 0.8949,
      "step": 5378
    },
    {
      "epoch": 0.6709492328801298,
      "grad_norm": 0.0,
      "learning_rate": 5.16180087477412e-06,
      "loss": 0.9083,
      "step": 5379
    },
    {
      "epoch": 0.6710739678183859,
      "grad_norm": 0.0,
      "learning_rate": 5.1582654929298745e-06,
      "loss": 0.9533,
      "step": 5380
    },
    {
      "epoch": 0.6711987027566422,
      "grad_norm": 0.0,
      "learning_rate": 5.154730901378762e-06,
      "loss": 0.8954,
      "step": 5381
    },
    {
      "epoch": 0.6713234376948983,
      "grad_norm": 0.0,
      "learning_rate": 5.151197100697714e-06,
      "loss": 0.896,
      "step": 5382
    },
    {
      "epoch": 0.6714481726331546,
      "grad_norm": 0.0,
      "learning_rate": 5.14766409146354e-06,
      "loss": 0.945,
      "step": 5383
    },
    {
      "epoch": 0.6715729075714107,
      "grad_norm": 0.0,
      "learning_rate": 5.14413187425291e-06,
      "loss": 0.9058,
      "step": 5384
    },
    {
      "epoch": 0.6716976425096669,
      "grad_norm": 0.0,
      "learning_rate": 5.1406004496423735e-06,
      "loss": 0.9362,
      "step": 5385
    },
    {
      "epoch": 0.6718223774479232,
      "grad_norm": 0.0,
      "learning_rate": 5.137069818208352e-06,
      "loss": 0.9307,
      "step": 5386
    },
    {
      "epoch": 0.6719471123861793,
      "grad_norm": 0.0,
      "learning_rate": 5.133539980527123e-06,
      "loss": 0.988,
      "step": 5387
    },
    {
      "epoch": 0.6720718473244356,
      "grad_norm": 0.0,
      "learning_rate": 5.130010937174858e-06,
      "loss": 0.9175,
      "step": 5388
    },
    {
      "epoch": 0.6721965822626917,
      "grad_norm": 0.0,
      "learning_rate": 5.1264826887275774e-06,
      "loss": 0.9412,
      "step": 5389
    },
    {
      "epoch": 0.672321317200948,
      "grad_norm": 0.0,
      "learning_rate": 5.122955235761182e-06,
      "loss": 0.9463,
      "step": 5390
    },
    {
      "epoch": 0.6724460521392042,
      "grad_norm": 0.0,
      "learning_rate": 5.119428578851447e-06,
      "loss": 0.8866,
      "step": 5391
    },
    {
      "epoch": 0.6725707870774604,
      "grad_norm": 0.0,
      "learning_rate": 5.115902718574001e-06,
      "loss": 0.9389,
      "step": 5392
    },
    {
      "epoch": 0.6726955220157166,
      "grad_norm": 0.0,
      "learning_rate": 5.112377655504359e-06,
      "loss": 0.8928,
      "step": 5393
    },
    {
      "epoch": 0.6728202569539729,
      "grad_norm": 0.0,
      "learning_rate": 5.1088533902179e-06,
      "loss": 0.951,
      "step": 5394
    },
    {
      "epoch": 0.672944991892229,
      "grad_norm": 0.0,
      "learning_rate": 5.105329923289872e-06,
      "loss": 0.9529,
      "step": 5395
    },
    {
      "epoch": 0.6730697268304853,
      "grad_norm": 0.0,
      "learning_rate": 5.1018072552954e-06,
      "loss": 0.9002,
      "step": 5396
    },
    {
      "epoch": 0.6731944617687414,
      "grad_norm": 0.0,
      "learning_rate": 5.098285386809463e-06,
      "loss": 0.9162,
      "step": 5397
    },
    {
      "epoch": 0.6733191967069976,
      "grad_norm": 0.0,
      "learning_rate": 5.094764318406921e-06,
      "loss": 0.9081,
      "step": 5398
    },
    {
      "epoch": 0.6734439316452538,
      "grad_norm": 0.0,
      "learning_rate": 5.0912440506625026e-06,
      "loss": 0.9509,
      "step": 5399
    },
    {
      "epoch": 0.67356866658351,
      "grad_norm": 0.0,
      "learning_rate": 5.087724584150808e-06,
      "loss": 0.937,
      "step": 5400
    },
    {
      "epoch": 0.6736934015217663,
      "grad_norm": 0.0,
      "learning_rate": 5.084205919446295e-06,
      "loss": 0.8796,
      "step": 5401
    },
    {
      "epoch": 0.6738181364600224,
      "grad_norm": 0.0,
      "learning_rate": 5.080688057123304e-06,
      "loss": 0.9696,
      "step": 5402
    },
    {
      "epoch": 0.6739428713982787,
      "grad_norm": 0.0,
      "learning_rate": 5.0771709977560404e-06,
      "loss": 0.906,
      "step": 5403
    },
    {
      "epoch": 0.6740676063365348,
      "grad_norm": 0.0,
      "learning_rate": 5.073654741918565e-06,
      "loss": 0.915,
      "step": 5404
    },
    {
      "epoch": 0.6741923412747911,
      "grad_norm": 0.0,
      "learning_rate": 5.070139290184837e-06,
      "loss": 0.9657,
      "step": 5405
    },
    {
      "epoch": 0.6743170762130473,
      "grad_norm": 0.0,
      "learning_rate": 5.066624643128653e-06,
      "loss": 0.9045,
      "step": 5406
    },
    {
      "epoch": 0.6744418111513035,
      "grad_norm": 0.0,
      "learning_rate": 5.063110801323697e-06,
      "loss": 0.9155,
      "step": 5407
    },
    {
      "epoch": 0.6745665460895597,
      "grad_norm": 0.0,
      "learning_rate": 5.059597765343521e-06,
      "loss": 0.9356,
      "step": 5408
    },
    {
      "epoch": 0.6746912810278158,
      "grad_norm": 0.0,
      "learning_rate": 5.056085535761532e-06,
      "loss": 0.913,
      "step": 5409
    },
    {
      "epoch": 0.6748160159660721,
      "grad_norm": 0.0,
      "learning_rate": 5.052574113151021e-06,
      "loss": 0.9559,
      "step": 5410
    },
    {
      "epoch": 0.6749407509043283,
      "grad_norm": 0.0,
      "learning_rate": 5.049063498085136e-06,
      "loss": 0.9126,
      "step": 5411
    },
    {
      "epoch": 0.6750654858425845,
      "grad_norm": 0.0,
      "learning_rate": 5.045553691136906e-06,
      "loss": 0.9169,
      "step": 5412
    },
    {
      "epoch": 0.6751902207808407,
      "grad_norm": 0.0,
      "learning_rate": 5.042044692879211e-06,
      "loss": 0.9102,
      "step": 5413
    },
    {
      "epoch": 0.6753149557190969,
      "grad_norm": 0.0,
      "learning_rate": 5.03853650388481e-06,
      "loss": 0.9376,
      "step": 5414
    },
    {
      "epoch": 0.6754396906573531,
      "grad_norm": 0.0,
      "learning_rate": 5.035029124726329e-06,
      "loss": 0.905,
      "step": 5415
    },
    {
      "epoch": 0.6755644255956094,
      "grad_norm": 0.0,
      "learning_rate": 5.031522555976263e-06,
      "loss": 0.9026,
      "step": 5416
    },
    {
      "epoch": 0.6756891605338655,
      "grad_norm": 0.0,
      "learning_rate": 5.0280167982069735e-06,
      "loss": 0.8835,
      "step": 5417
    },
    {
      "epoch": 0.6758138954721218,
      "grad_norm": 0.0,
      "learning_rate": 5.024511851990681e-06,
      "loss": 0.9192,
      "step": 5418
    },
    {
      "epoch": 0.6759386304103779,
      "grad_norm": 0.0,
      "learning_rate": 5.021007717899486e-06,
      "loss": 0.941,
      "step": 5419
    },
    {
      "epoch": 0.6760633653486342,
      "grad_norm": 0.0,
      "learning_rate": 5.017504396505355e-06,
      "loss": 0.9614,
      "step": 5420
    },
    {
      "epoch": 0.6761881002868904,
      "grad_norm": 0.0,
      "learning_rate": 5.0140018883801076e-06,
      "loss": 0.8937,
      "step": 5421
    },
    {
      "epoch": 0.6763128352251465,
      "grad_norm": 0.0,
      "learning_rate": 5.010500194095456e-06,
      "loss": 0.917,
      "step": 5422
    },
    {
      "epoch": 0.6764375701634028,
      "grad_norm": 0.0,
      "learning_rate": 5.006999314222951e-06,
      "loss": 0.9186,
      "step": 5423
    },
    {
      "epoch": 0.6765623051016589,
      "grad_norm": 0.0,
      "learning_rate": 5.003499249334034e-06,
      "loss": 0.9096,
      "step": 5424
    },
    {
      "epoch": 0.6766870400399152,
      "grad_norm": 0.0,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.9124,
      "step": 5425
    },
    {
      "epoch": 0.6768117749781714,
      "grad_norm": 0.0,
      "learning_rate": 4.996501566792017e-06,
      "loss": 0.9346,
      "step": 5426
    },
    {
      "epoch": 0.6769365099164276,
      "grad_norm": 0.0,
      "learning_rate": 4.993003950281114e-06,
      "loss": 0.9265,
      "step": 5427
    },
    {
      "epoch": 0.6770612448546838,
      "grad_norm": 0.0,
      "learning_rate": 4.98950715103819e-06,
      "loss": 0.8493,
      "step": 5428
    },
    {
      "epoch": 0.67718597979294,
      "grad_norm": 0.0,
      "learning_rate": 4.986011169634017e-06,
      "loss": 0.8918,
      "step": 5429
    },
    {
      "epoch": 0.6773107147311962,
      "grad_norm": 0.0,
      "learning_rate": 4.982516006639221e-06,
      "loss": 0.8991,
      "step": 5430
    },
    {
      "epoch": 0.6774354496694525,
      "grad_norm": 0.0,
      "learning_rate": 4.9790216626243015e-06,
      "loss": 0.9461,
      "step": 5431
    },
    {
      "epoch": 0.6775601846077086,
      "grad_norm": 0.0,
      "learning_rate": 4.975528138159629e-06,
      "loss": 0.9028,
      "step": 5432
    },
    {
      "epoch": 0.6776849195459648,
      "grad_norm": 0.0,
      "learning_rate": 4.972035433815422e-06,
      "loss": 0.958,
      "step": 5433
    },
    {
      "epoch": 0.677809654484221,
      "grad_norm": 0.0,
      "learning_rate": 4.968543550161795e-06,
      "loss": 0.9383,
      "step": 5434
    },
    {
      "epoch": 0.6779343894224772,
      "grad_norm": 0.0,
      "learning_rate": 4.965052487768699e-06,
      "loss": 0.879,
      "step": 5435
    },
    {
      "epoch": 0.6780591243607335,
      "grad_norm": 0.0,
      "learning_rate": 4.961562247205968e-06,
      "loss": 0.8807,
      "step": 5436
    },
    {
      "epoch": 0.6781838592989896,
      "grad_norm": 0.0,
      "learning_rate": 4.958072829043299e-06,
      "loss": 0.9858,
      "step": 5437
    },
    {
      "epoch": 0.6783085942372459,
      "grad_norm": 0.0,
      "learning_rate": 4.954584233850247e-06,
      "loss": 0.9021,
      "step": 5438
    },
    {
      "epoch": 0.678433329175502,
      "grad_norm": 0.0,
      "learning_rate": 4.951096462196243e-06,
      "loss": 0.9266,
      "step": 5439
    },
    {
      "epoch": 0.6785580641137583,
      "grad_norm": 0.0,
      "learning_rate": 4.9476095146505775e-06,
      "loss": 0.9106,
      "step": 5440
    },
    {
      "epoch": 0.6786827990520145,
      "grad_norm": 0.0,
      "learning_rate": 4.944123391782411e-06,
      "loss": 0.9066,
      "step": 5441
    },
    {
      "epoch": 0.6788075339902707,
      "grad_norm": 0.0,
      "learning_rate": 4.940638094160767e-06,
      "loss": 0.9169,
      "step": 5442
    },
    {
      "epoch": 0.6789322689285269,
      "grad_norm": 0.0,
      "learning_rate": 4.9371536223545295e-06,
      "loss": 0.9207,
      "step": 5443
    },
    {
      "epoch": 0.6790570038667831,
      "grad_norm": 0.0,
      "learning_rate": 4.933669976932452e-06,
      "loss": 0.9296,
      "step": 5444
    },
    {
      "epoch": 0.6791817388050393,
      "grad_norm": 0.0,
      "learning_rate": 4.930187158463158e-06,
      "loss": 0.8947,
      "step": 5445
    },
    {
      "epoch": 0.6793064737432954,
      "grad_norm": 0.0,
      "learning_rate": 4.926705167515132e-06,
      "loss": 1.0408,
      "step": 5446
    },
    {
      "epoch": 0.6794312086815517,
      "grad_norm": 0.0,
      "learning_rate": 4.923224004656716e-06,
      "loss": 0.9057,
      "step": 5447
    },
    {
      "epoch": 0.6795559436198079,
      "grad_norm": 0.0,
      "learning_rate": 4.919743670456127e-06,
      "loss": 0.9436,
      "step": 5448
    },
    {
      "epoch": 0.6796806785580641,
      "grad_norm": 0.0,
      "learning_rate": 4.916264165481448e-06,
      "loss": 0.9554,
      "step": 5449
    },
    {
      "epoch": 0.6798054134963203,
      "grad_norm": 0.0,
      "learning_rate": 4.912785490300609e-06,
      "loss": 0.9242,
      "step": 5450
    },
    {
      "epoch": 0.6799301484345766,
      "grad_norm": 0.0,
      "learning_rate": 4.9093076454814335e-06,
      "loss": 0.8832,
      "step": 5451
    },
    {
      "epoch": 0.6800548833728327,
      "grad_norm": 0.0,
      "learning_rate": 4.9058306315915826e-06,
      "loss": 0.9139,
      "step": 5452
    },
    {
      "epoch": 0.680179618311089,
      "grad_norm": 0.0,
      "learning_rate": 4.902354449198596e-06,
      "loss": 0.9475,
      "step": 5453
    },
    {
      "epoch": 0.6803043532493451,
      "grad_norm": 0.0,
      "learning_rate": 4.8988790988698774e-06,
      "loss": 0.905,
      "step": 5454
    },
    {
      "epoch": 0.6804290881876014,
      "grad_norm": 0.0,
      "learning_rate": 4.895404581172686e-06,
      "loss": 0.9953,
      "step": 5455
    },
    {
      "epoch": 0.6805538231258575,
      "grad_norm": 0.0,
      "learning_rate": 4.891930896674152e-06,
      "loss": 0.9483,
      "step": 5456
    },
    {
      "epoch": 0.6806785580641137,
      "grad_norm": 0.0,
      "learning_rate": 4.888458045941269e-06,
      "loss": 0.9538,
      "step": 5457
    },
    {
      "epoch": 0.68080329300237,
      "grad_norm": 0.0,
      "learning_rate": 4.884986029540898e-06,
      "loss": 0.933,
      "step": 5458
    },
    {
      "epoch": 0.6809280279406261,
      "grad_norm": 0.0,
      "learning_rate": 4.881514848039752e-06,
      "loss": 0.9078,
      "step": 5459
    },
    {
      "epoch": 0.6810527628788824,
      "grad_norm": 0.0,
      "learning_rate": 4.878044502004419e-06,
      "loss": 0.9232,
      "step": 5460
    },
    {
      "epoch": 0.6811774978171385,
      "grad_norm": 0.0,
      "learning_rate": 4.874574992001348e-06,
      "loss": 0.8729,
      "step": 5461
    },
    {
      "epoch": 0.6813022327553948,
      "grad_norm": 0.0,
      "learning_rate": 4.871106318596849e-06,
      "loss": 0.917,
      "step": 5462
    },
    {
      "epoch": 0.681426967693651,
      "grad_norm": 0.0,
      "learning_rate": 4.867638482357101e-06,
      "loss": 0.9011,
      "step": 5463
    },
    {
      "epoch": 0.6815517026319072,
      "grad_norm": 0.0,
      "learning_rate": 4.864171483848136e-06,
      "loss": 0.9809,
      "step": 5464
    },
    {
      "epoch": 0.6816764375701634,
      "grad_norm": 0.0,
      "learning_rate": 4.860705323635857e-06,
      "loss": 0.9579,
      "step": 5465
    },
    {
      "epoch": 0.6818011725084197,
      "grad_norm": 0.0,
      "learning_rate": 4.8572400022860345e-06,
      "loss": 0.9287,
      "step": 5466
    },
    {
      "epoch": 0.6819259074466758,
      "grad_norm": 0.0,
      "learning_rate": 4.8537755203642845e-06,
      "loss": 0.9135,
      "step": 5467
    },
    {
      "epoch": 0.6820506423849321,
      "grad_norm": 0.0,
      "learning_rate": 4.850311878436113e-06,
      "loss": 0.9654,
      "step": 5468
    },
    {
      "epoch": 0.6821753773231882,
      "grad_norm": 0.0,
      "learning_rate": 4.846849077066864e-06,
      "loss": 0.9123,
      "step": 5469
    },
    {
      "epoch": 0.6823001122614444,
      "grad_norm": 0.0,
      "learning_rate": 4.843387116821749e-06,
      "loss": 0.9296,
      "step": 5470
    },
    {
      "epoch": 0.6824248471997006,
      "grad_norm": 0.0,
      "learning_rate": 4.839925998265863e-06,
      "loss": 0.9041,
      "step": 5471
    },
    {
      "epoch": 0.6825495821379568,
      "grad_norm": 0.0,
      "learning_rate": 4.836465721964134e-06,
      "loss": 0.8676,
      "step": 5472
    },
    {
      "epoch": 0.6826743170762131,
      "grad_norm": 0.0,
      "learning_rate": 4.8330062884813714e-06,
      "loss": 0.8845,
      "step": 5473
    },
    {
      "epoch": 0.6827990520144692,
      "grad_norm": 0.0,
      "learning_rate": 4.829547698382246e-06,
      "loss": 0.9179,
      "step": 5474
    },
    {
      "epoch": 0.6829237869527255,
      "grad_norm": 0.0,
      "learning_rate": 4.8260899522312775e-06,
      "loss": 0.9164,
      "step": 5475
    },
    {
      "epoch": 0.6830485218909816,
      "grad_norm": 0.0,
      "learning_rate": 4.822633050592863e-06,
      "loss": 0.9421,
      "step": 5476
    },
    {
      "epoch": 0.6831732568292379,
      "grad_norm": 0.0,
      "learning_rate": 4.819176994031255e-06,
      "loss": 0.9228,
      "step": 5477
    },
    {
      "epoch": 0.6832979917674941,
      "grad_norm": 0.0,
      "learning_rate": 4.815721783110569e-06,
      "loss": 0.9095,
      "step": 5478
    },
    {
      "epoch": 0.6834227267057503,
      "grad_norm": 0.0,
      "learning_rate": 4.812267418394784e-06,
      "loss": 0.8997,
      "step": 5479
    },
    {
      "epoch": 0.6835474616440065,
      "grad_norm": 0.0,
      "learning_rate": 4.808813900447742e-06,
      "loss": 0.923,
      "step": 5480
    },
    {
      "epoch": 0.6836721965822626,
      "grad_norm": 0.0,
      "learning_rate": 4.805361229833136e-06,
      "loss": 0.9399,
      "step": 5481
    },
    {
      "epoch": 0.6837969315205189,
      "grad_norm": 0.0,
      "learning_rate": 4.801909407114535e-06,
      "loss": 0.9168,
      "step": 5482
    },
    {
      "epoch": 0.6839216664587751,
      "grad_norm": 0.0,
      "learning_rate": 4.798458432855364e-06,
      "loss": 0.9344,
      "step": 5483
    },
    {
      "epoch": 0.6840464013970313,
      "grad_norm": 0.0,
      "learning_rate": 4.795008307618904e-06,
      "loss": 0.8857,
      "step": 5484
    },
    {
      "epoch": 0.6841711363352875,
      "grad_norm": 0.0,
      "learning_rate": 4.791559031968304e-06,
      "loss": 0.9197,
      "step": 5485
    },
    {
      "epoch": 0.6842958712735437,
      "grad_norm": 0.0,
      "learning_rate": 4.7881106064665785e-06,
      "loss": 0.9332,
      "step": 5486
    },
    {
      "epoch": 0.6844206062117999,
      "grad_norm": 0.0,
      "learning_rate": 4.784663031676585e-06,
      "loss": 0.8623,
      "step": 5487
    },
    {
      "epoch": 0.6845453411500562,
      "grad_norm": 0.0,
      "learning_rate": 4.781216308161072e-06,
      "loss": 0.9097,
      "step": 5488
    },
    {
      "epoch": 0.6846700760883123,
      "grad_norm": 0.0,
      "learning_rate": 4.7777704364826175e-06,
      "loss": 0.9346,
      "step": 5489
    },
    {
      "epoch": 0.6847948110265686,
      "grad_norm": 0.0,
      "learning_rate": 4.774325417203678e-06,
      "loss": 0.9114,
      "step": 5490
    },
    {
      "epoch": 0.6849195459648247,
      "grad_norm": 0.0,
      "learning_rate": 4.770881250886574e-06,
      "loss": 0.9565,
      "step": 5491
    },
    {
      "epoch": 0.685044280903081,
      "grad_norm": 0.0,
      "learning_rate": 4.767437938093472e-06,
      "loss": 0.9139,
      "step": 5492
    },
    {
      "epoch": 0.6851690158413372,
      "grad_norm": 0.0,
      "learning_rate": 4.7639954793864095e-06,
      "loss": 0.9135,
      "step": 5493
    },
    {
      "epoch": 0.6852937507795933,
      "grad_norm": 0.0,
      "learning_rate": 4.760553875327284e-06,
      "loss": 0.9071,
      "step": 5494
    },
    {
      "epoch": 0.6854184857178496,
      "grad_norm": 0.0,
      "learning_rate": 4.7571131264778555e-06,
      "loss": 0.8622,
      "step": 5495
    },
    {
      "epoch": 0.6855432206561057,
      "grad_norm": 0.0,
      "learning_rate": 4.753673233399731e-06,
      "loss": 0.911,
      "step": 5496
    },
    {
      "epoch": 0.685667955594362,
      "grad_norm": 0.0,
      "learning_rate": 4.7502341966544e-06,
      "loss": 0.9356,
      "step": 5497
    },
    {
      "epoch": 0.6857926905326182,
      "grad_norm": 0.0,
      "learning_rate": 4.7467960168031905e-06,
      "loss": 0.9471,
      "step": 5498
    },
    {
      "epoch": 0.6859174254708744,
      "grad_norm": 0.0,
      "learning_rate": 4.743358694407303e-06,
      "loss": 0.8808,
      "step": 5499
    },
    {
      "epoch": 0.6860421604091306,
      "grad_norm": 0.0,
      "learning_rate": 4.7399222300278e-06,
      "loss": 0.8683,
      "step": 5500
    },
    {
      "epoch": 0.6861668953473868,
      "grad_norm": 0.0,
      "learning_rate": 4.736486624225591e-06,
      "loss": 0.9099,
      "step": 5501
    },
    {
      "epoch": 0.686291630285643,
      "grad_norm": 0.0,
      "learning_rate": 4.733051877561458e-06,
      "loss": 0.9519,
      "step": 5502
    },
    {
      "epoch": 0.6864163652238993,
      "grad_norm": 0.0,
      "learning_rate": 4.729617990596041e-06,
      "loss": 0.9428,
      "step": 5503
    },
    {
      "epoch": 0.6865411001621554,
      "grad_norm": 0.0,
      "learning_rate": 4.726184963889826e-06,
      "loss": 0.9136,
      "step": 5504
    },
    {
      "epoch": 0.6866658351004116,
      "grad_norm": 0.0,
      "learning_rate": 4.722752798003184e-06,
      "loss": 0.9558,
      "step": 5505
    },
    {
      "epoch": 0.6867905700386678,
      "grad_norm": 0.0,
      "learning_rate": 4.7193214934963204e-06,
      "loss": 0.9434,
      "step": 5506
    },
    {
      "epoch": 0.686915304976924,
      "grad_norm": 0.0,
      "learning_rate": 4.715891050929316e-06,
      "loss": 0.9415,
      "step": 5507
    },
    {
      "epoch": 0.6870400399151803,
      "grad_norm": 0.0,
      "learning_rate": 4.7124614708621055e-06,
      "loss": 0.9083,
      "step": 5508
    },
    {
      "epoch": 0.6871647748534364,
      "grad_norm": 0.0,
      "learning_rate": 4.7090327538544765e-06,
      "loss": 0.9236,
      "step": 5509
    },
    {
      "epoch": 0.6872895097916927,
      "grad_norm": 0.0,
      "learning_rate": 4.705604900466088e-06,
      "loss": 0.9359,
      "step": 5510
    },
    {
      "epoch": 0.6874142447299488,
      "grad_norm": 0.0,
      "learning_rate": 4.702177911256452e-06,
      "loss": 0.897,
      "step": 5511
    },
    {
      "epoch": 0.6875389796682051,
      "grad_norm": 0.0,
      "learning_rate": 4.6987517867849395e-06,
      "loss": 0.9395,
      "step": 5512
    },
    {
      "epoch": 0.6876637146064613,
      "grad_norm": 0.0,
      "learning_rate": 4.6953265276107725e-06,
      "loss": 0.9198,
      "step": 5513
    },
    {
      "epoch": 0.6877884495447175,
      "grad_norm": 0.0,
      "learning_rate": 4.691902134293054e-06,
      "loss": 0.9157,
      "step": 5514
    },
    {
      "epoch": 0.6879131844829737,
      "grad_norm": 0.0,
      "learning_rate": 4.688478607390723e-06,
      "loss": 0.8705,
      "step": 5515
    },
    {
      "epoch": 0.6880379194212299,
      "grad_norm": 0.0,
      "learning_rate": 4.685055947462579e-06,
      "loss": 0.9505,
      "step": 5516
    },
    {
      "epoch": 0.6881626543594861,
      "grad_norm": 0.0,
      "learning_rate": 4.681634155067301e-06,
      "loss": 0.9031,
      "step": 5517
    },
    {
      "epoch": 0.6882873892977422,
      "grad_norm": 0.0,
      "learning_rate": 4.6782132307634e-06,
      "loss": 0.8416,
      "step": 5518
    },
    {
      "epoch": 0.6884121242359985,
      "grad_norm": 0.0,
      "learning_rate": 4.674793175109261e-06,
      "loss": 0.9211,
      "step": 5519
    },
    {
      "epoch": 0.6885368591742547,
      "grad_norm": 0.0,
      "learning_rate": 4.6713739886631285e-06,
      "loss": 0.9324,
      "step": 5520
    },
    {
      "epoch": 0.6886615941125109,
      "grad_norm": 0.0,
      "learning_rate": 4.66795567198309e-06,
      "loss": 0.9477,
      "step": 5521
    },
    {
      "epoch": 0.6887863290507671,
      "grad_norm": 0.0,
      "learning_rate": 4.664538225627107e-06,
      "loss": 0.8675,
      "step": 5522
    },
    {
      "epoch": 0.6889110639890234,
      "grad_norm": 0.0,
      "learning_rate": 4.66112165015299e-06,
      "loss": 0.8696,
      "step": 5523
    },
    {
      "epoch": 0.6890357989272795,
      "grad_norm": 0.0,
      "learning_rate": 4.657705946118414e-06,
      "loss": 0.9055,
      "step": 5524
    },
    {
      "epoch": 0.6891605338655358,
      "grad_norm": 0.0,
      "learning_rate": 4.654291114080908e-06,
      "loss": 0.8742,
      "step": 5525
    },
    {
      "epoch": 0.6892852688037919,
      "grad_norm": 0.0,
      "learning_rate": 4.650877154597853e-06,
      "loss": 0.9267,
      "step": 5526
    },
    {
      "epoch": 0.6894100037420482,
      "grad_norm": 0.0,
      "learning_rate": 4.6474640682265e-06,
      "loss": 0.9828,
      "step": 5527
    },
    {
      "epoch": 0.6895347386803043,
      "grad_norm": 0.0,
      "learning_rate": 4.644051855523945e-06,
      "loss": 0.8863,
      "step": 5528
    },
    {
      "epoch": 0.6896594736185605,
      "grad_norm": 0.0,
      "learning_rate": 4.640640517047154e-06,
      "loss": 0.8695,
      "step": 5529
    },
    {
      "epoch": 0.6897842085568168,
      "grad_norm": 0.0,
      "learning_rate": 4.6372300533529314e-06,
      "loss": 0.9166,
      "step": 5530
    },
    {
      "epoch": 0.6899089434950729,
      "grad_norm": 0.0,
      "learning_rate": 4.633820464997967e-06,
      "loss": 0.8954,
      "step": 5531
    },
    {
      "epoch": 0.6900336784333292,
      "grad_norm": 0.0,
      "learning_rate": 4.630411752538784e-06,
      "loss": 0.9244,
      "step": 5532
    },
    {
      "epoch": 0.6901584133715853,
      "grad_norm": 0.0,
      "learning_rate": 4.627003916531761e-06,
      "loss": 0.9462,
      "step": 5533
    },
    {
      "epoch": 0.6902831483098416,
      "grad_norm": 0.0,
      "learning_rate": 4.623596957533159e-06,
      "loss": 0.9393,
      "step": 5534
    },
    {
      "epoch": 0.6904078832480978,
      "grad_norm": 0.0,
      "learning_rate": 4.620190876099068e-06,
      "loss": 0.888,
      "step": 5535
    },
    {
      "epoch": 0.690532618186354,
      "grad_norm": 0.0,
      "learning_rate": 4.6167856727854495e-06,
      "loss": 0.8911,
      "step": 5536
    },
    {
      "epoch": 0.6906573531246102,
      "grad_norm": 0.0,
      "learning_rate": 4.613381348148125e-06,
      "loss": 0.8869,
      "step": 5537
    },
    {
      "epoch": 0.6907820880628665,
      "grad_norm": 0.0,
      "learning_rate": 4.609977902742755e-06,
      "loss": 0.9454,
      "step": 5538
    },
    {
      "epoch": 0.6909068230011226,
      "grad_norm": 0.0,
      "learning_rate": 4.606575337124874e-06,
      "loss": 0.8968,
      "step": 5539
    },
    {
      "epoch": 0.6910315579393789,
      "grad_norm": 0.0,
      "learning_rate": 4.603173651849865e-06,
      "loss": 0.9262,
      "step": 5540
    },
    {
      "epoch": 0.691156292877635,
      "grad_norm": 0.0,
      "learning_rate": 4.599772847472973e-06,
      "loss": 0.9412,
      "step": 5541
    },
    {
      "epoch": 0.6912810278158912,
      "grad_norm": 0.0,
      "learning_rate": 4.5963729245492875e-06,
      "loss": 0.9253,
      "step": 5542
    },
    {
      "epoch": 0.6914057627541474,
      "grad_norm": 0.0,
      "learning_rate": 4.592973883633765e-06,
      "loss": 0.9316,
      "step": 5543
    },
    {
      "epoch": 0.6915304976924036,
      "grad_norm": 0.0,
      "learning_rate": 4.589575725281216e-06,
      "loss": 0.9238,
      "step": 5544
    },
    {
      "epoch": 0.6916552326306599,
      "grad_norm": 0.0,
      "learning_rate": 4.586178450046304e-06,
      "loss": 0.8824,
      "step": 5545
    },
    {
      "epoch": 0.691779967568916,
      "grad_norm": 0.0,
      "learning_rate": 4.582782058483554e-06,
      "loss": 0.8531,
      "step": 5546
    },
    {
      "epoch": 0.6919047025071723,
      "grad_norm": 0.0,
      "learning_rate": 4.579386551147336e-06,
      "loss": 0.9489,
      "step": 5547
    },
    {
      "epoch": 0.6920294374454284,
      "grad_norm": 0.0,
      "learning_rate": 4.575991928591885e-06,
      "loss": 1.0001,
      "step": 5548
    },
    {
      "epoch": 0.6921541723836847,
      "grad_norm": 0.0,
      "learning_rate": 4.572598191371294e-06,
      "loss": 0.9202,
      "step": 5549
    },
    {
      "epoch": 0.6922789073219409,
      "grad_norm": 0.0,
      "learning_rate": 4.569205340039495e-06,
      "loss": 0.9127,
      "step": 5550
    },
    {
      "epoch": 0.6924036422601971,
      "grad_norm": 0.0,
      "learning_rate": 4.565813375150302e-06,
      "loss": 0.9101,
      "step": 5551
    },
    {
      "epoch": 0.6925283771984533,
      "grad_norm": 0.0,
      "learning_rate": 4.562422297257356e-06,
      "loss": 0.9489,
      "step": 5552
    },
    {
      "epoch": 0.6926531121367095,
      "grad_norm": 0.0,
      "learning_rate": 4.559032106914173e-06,
      "loss": 0.8652,
      "step": 5553
    },
    {
      "epoch": 0.6927778470749657,
      "grad_norm": 0.0,
      "learning_rate": 4.5556428046741195e-06,
      "loss": 0.8766,
      "step": 5554
    },
    {
      "epoch": 0.6929025820132219,
      "grad_norm": 0.0,
      "learning_rate": 4.552254391090408e-06,
      "loss": 0.8917,
      "step": 5555
    },
    {
      "epoch": 0.6930273169514781,
      "grad_norm": 0.0,
      "learning_rate": 4.548866866716118e-06,
      "loss": 0.971,
      "step": 5556
    },
    {
      "epoch": 0.6931520518897343,
      "grad_norm": 0.0,
      "learning_rate": 4.545480232104176e-06,
      "loss": 0.9532,
      "step": 5557
    },
    {
      "epoch": 0.6932767868279905,
      "grad_norm": 0.0,
      "learning_rate": 4.542094487807372e-06,
      "loss": 0.8886,
      "step": 5558
    },
    {
      "epoch": 0.6934015217662467,
      "grad_norm": 0.0,
      "learning_rate": 4.538709634378335e-06,
      "loss": 0.9351,
      "step": 5559
    },
    {
      "epoch": 0.693526256704503,
      "grad_norm": 0.0,
      "learning_rate": 4.535325672369567e-06,
      "loss": 0.9153,
      "step": 5560
    },
    {
      "epoch": 0.6936509916427591,
      "grad_norm": 0.0,
      "learning_rate": 4.531942602333416e-06,
      "loss": 0.9519,
      "step": 5561
    },
    {
      "epoch": 0.6937757265810154,
      "grad_norm": 0.0,
      "learning_rate": 4.528560424822072e-06,
      "loss": 0.8818,
      "step": 5562
    },
    {
      "epoch": 0.6939004615192715,
      "grad_norm": 0.0,
      "learning_rate": 4.52517914038761e-06,
      "loss": 0.8959,
      "step": 5563
    },
    {
      "epoch": 0.6940251964575278,
      "grad_norm": 0.0,
      "learning_rate": 4.521798749581927e-06,
      "loss": 0.8509,
      "step": 5564
    },
    {
      "epoch": 0.694149931395784,
      "grad_norm": 0.0,
      "learning_rate": 4.518419252956794e-06,
      "loss": 0.9566,
      "step": 5565
    },
    {
      "epoch": 0.6942746663340401,
      "grad_norm": 0.0,
      "learning_rate": 4.5150406510638326e-06,
      "loss": 0.9149,
      "step": 5566
    },
    {
      "epoch": 0.6943994012722964,
      "grad_norm": 0.0,
      "learning_rate": 4.511662944454507e-06,
      "loss": 0.9174,
      "step": 5567
    },
    {
      "epoch": 0.6945241362105525,
      "grad_norm": 0.0,
      "learning_rate": 4.508286133680149e-06,
      "loss": 0.8963,
      "step": 5568
    },
    {
      "epoch": 0.6946488711488088,
      "grad_norm": 0.0,
      "learning_rate": 4.504910219291941e-06,
      "loss": 0.9363,
      "step": 5569
    },
    {
      "epoch": 0.694773606087065,
      "grad_norm": 0.0,
      "learning_rate": 4.501535201840913e-06,
      "loss": 0.8914,
      "step": 5570
    },
    {
      "epoch": 0.6948983410253212,
      "grad_norm": 0.0,
      "learning_rate": 4.498161081877961e-06,
      "loss": 0.895,
      "step": 5571
    },
    {
      "epoch": 0.6950230759635774,
      "grad_norm": 0.0,
      "learning_rate": 4.494787859953817e-06,
      "loss": 0.942,
      "step": 5572
    },
    {
      "epoch": 0.6951478109018336,
      "grad_norm": 0.0,
      "learning_rate": 4.491415536619081e-06,
      "loss": 0.9435,
      "step": 5573
    },
    {
      "epoch": 0.6952725458400898,
      "grad_norm": 0.0,
      "learning_rate": 4.488044112424198e-06,
      "loss": 0.9259,
      "step": 5574
    },
    {
      "epoch": 0.6953972807783461,
      "grad_norm": 0.0,
      "learning_rate": 4.484673587919477e-06,
      "loss": 0.9498,
      "step": 5575
    },
    {
      "epoch": 0.6955220157166022,
      "grad_norm": 0.0,
      "learning_rate": 4.481303963655061e-06,
      "loss": 0.9352,
      "step": 5576
    },
    {
      "epoch": 0.6956467506548585,
      "grad_norm": 0.0,
      "learning_rate": 4.477935240180966e-06,
      "loss": 0.9119,
      "step": 5577
    },
    {
      "epoch": 0.6957714855931146,
      "grad_norm": 0.0,
      "learning_rate": 4.474567418047053e-06,
      "loss": 0.9193,
      "step": 5578
    },
    {
      "epoch": 0.6958962205313708,
      "grad_norm": 0.0,
      "learning_rate": 4.471200497803025e-06,
      "loss": 0.9344,
      "step": 5579
    },
    {
      "epoch": 0.6960209554696271,
      "grad_norm": 0.0,
      "learning_rate": 4.467834479998464e-06,
      "loss": 0.8643,
      "step": 5580
    },
    {
      "epoch": 0.6961456904078832,
      "grad_norm": 0.0,
      "learning_rate": 4.464469365182778e-06,
      "loss": 0.9312,
      "step": 5581
    },
    {
      "epoch": 0.6962704253461395,
      "grad_norm": 0.0,
      "learning_rate": 4.461105153905241e-06,
      "loss": 0.945,
      "step": 5582
    },
    {
      "epoch": 0.6963951602843956,
      "grad_norm": 0.0,
      "learning_rate": 4.457741846714981e-06,
      "loss": 0.875,
      "step": 5583
    },
    {
      "epoch": 0.6965198952226519,
      "grad_norm": 0.0,
      "learning_rate": 4.454379444160966e-06,
      "loss": 0.9396,
      "step": 5584
    },
    {
      "epoch": 0.696644630160908,
      "grad_norm": 0.0,
      "learning_rate": 4.4510179467920325e-06,
      "loss": 0.941,
      "step": 5585
    },
    {
      "epoch": 0.6967693650991643,
      "grad_norm": 0.0,
      "learning_rate": 4.447657355156858e-06,
      "loss": 0.9618,
      "step": 5586
    },
    {
      "epoch": 0.6968941000374205,
      "grad_norm": 0.0,
      "learning_rate": 4.444297669803981e-06,
      "loss": 0.9212,
      "step": 5587
    },
    {
      "epoch": 0.6970188349756767,
      "grad_norm": 0.0,
      "learning_rate": 4.440938891281779e-06,
      "loss": 0.941,
      "step": 5588
    },
    {
      "epoch": 0.6971435699139329,
      "grad_norm": 0.0,
      "learning_rate": 4.437581020138493e-06,
      "loss": 0.9181,
      "step": 5589
    },
    {
      "epoch": 0.697268304852189,
      "grad_norm": 0.0,
      "learning_rate": 4.434224056922212e-06,
      "loss": 0.8808,
      "step": 5590
    },
    {
      "epoch": 0.6973930397904453,
      "grad_norm": 0.0,
      "learning_rate": 4.430868002180877e-06,
      "loss": 0.9068,
      "step": 5591
    },
    {
      "epoch": 0.6975177747287015,
      "grad_norm": 0.0,
      "learning_rate": 4.427512856462285e-06,
      "loss": 0.9411,
      "step": 5592
    },
    {
      "epoch": 0.6976425096669577,
      "grad_norm": 0.0,
      "learning_rate": 4.424158620314073e-06,
      "loss": 0.9059,
      "step": 5593
    },
    {
      "epoch": 0.6977672446052139,
      "grad_norm": 0.0,
      "learning_rate": 4.42080529428374e-06,
      "loss": 0.9142,
      "step": 5594
    },
    {
      "epoch": 0.6978919795434702,
      "grad_norm": 0.0,
      "learning_rate": 4.417452878918638e-06,
      "loss": 0.9088,
      "step": 5595
    },
    {
      "epoch": 0.6980167144817263,
      "grad_norm": 0.0,
      "learning_rate": 4.414101374765953e-06,
      "loss": 0.8844,
      "step": 5596
    },
    {
      "epoch": 0.6981414494199826,
      "grad_norm": 0.0,
      "learning_rate": 4.410750782372751e-06,
      "loss": 0.9195,
      "step": 5597
    },
    {
      "epoch": 0.6982661843582387,
      "grad_norm": 0.0,
      "learning_rate": 4.407401102285923e-06,
      "loss": 0.9731,
      "step": 5598
    },
    {
      "epoch": 0.698390919296495,
      "grad_norm": 0.0,
      "learning_rate": 4.4040523350522224e-06,
      "loss": 0.9417,
      "step": 5599
    },
    {
      "epoch": 0.6985156542347511,
      "grad_norm": 0.0,
      "learning_rate": 4.400704481218259e-06,
      "loss": 0.8774,
      "step": 5600
    },
    {
      "epoch": 0.6986403891730074,
      "grad_norm": 0.0,
      "learning_rate": 4.3973575413304755e-06,
      "loss": 0.9091,
      "step": 5601
    },
    {
      "epoch": 0.6987651241112636,
      "grad_norm": 0.0,
      "learning_rate": 4.394011515935186e-06,
      "loss": 0.9036,
      "step": 5602
    },
    {
      "epoch": 0.6988898590495197,
      "grad_norm": 0.0,
      "learning_rate": 4.390666405578541e-06,
      "loss": 0.9043,
      "step": 5603
    },
    {
      "epoch": 0.699014593987776,
      "grad_norm": 0.0,
      "learning_rate": 4.3873222108065535e-06,
      "loss": 0.9204,
      "step": 5604
    },
    {
      "epoch": 0.6991393289260321,
      "grad_norm": 0.0,
      "learning_rate": 4.3839789321650724e-06,
      "loss": 0.9142,
      "step": 5605
    },
    {
      "epoch": 0.6992640638642884,
      "grad_norm": 0.0,
      "learning_rate": 4.380636570199809e-06,
      "loss": 0.9274,
      "step": 5606
    },
    {
      "epoch": 0.6993887988025446,
      "grad_norm": 0.0,
      "learning_rate": 4.37729512545632e-06,
      "loss": 0.9323,
      "step": 5607
    },
    {
      "epoch": 0.6995135337408008,
      "grad_norm": 0.0,
      "learning_rate": 4.373954598480015e-06,
      "loss": 0.9895,
      "step": 5608
    },
    {
      "epoch": 0.699638268679057,
      "grad_norm": 0.0,
      "learning_rate": 4.3706149898161546e-06,
      "loss": 0.9682,
      "step": 5609
    },
    {
      "epoch": 0.6997630036173133,
      "grad_norm": 0.0,
      "learning_rate": 4.36727630000984e-06,
      "loss": 0.9194,
      "step": 5610
    },
    {
      "epoch": 0.6998877385555694,
      "grad_norm": 0.0,
      "learning_rate": 4.363938529606034e-06,
      "loss": 0.9412,
      "step": 5611
    },
    {
      "epoch": 0.7000124734938257,
      "grad_norm": 0.0,
      "learning_rate": 4.360601679149546e-06,
      "loss": 0.9467,
      "step": 5612
    },
    {
      "epoch": 0.7001372084320818,
      "grad_norm": 0.0,
      "learning_rate": 4.357265749185027e-06,
      "loss": 0.8696,
      "step": 5613
    },
    {
      "epoch": 0.700261943370338,
      "grad_norm": 0.0,
      "learning_rate": 4.353930740256997e-06,
      "loss": 0.8962,
      "step": 5614
    },
    {
      "epoch": 0.7003866783085942,
      "grad_norm": 0.0,
      "learning_rate": 4.350596652909803e-06,
      "loss": 0.9595,
      "step": 5615
    },
    {
      "epoch": 0.7005114132468504,
      "grad_norm": 0.0,
      "learning_rate": 4.347263487687656e-06,
      "loss": 0.9271,
      "step": 5616
    },
    {
      "epoch": 0.7006361481851067,
      "grad_norm": 0.0,
      "learning_rate": 4.343931245134616e-06,
      "loss": 0.9462,
      "step": 5617
    },
    {
      "epoch": 0.7007608831233628,
      "grad_norm": 0.0,
      "learning_rate": 4.340599925794582e-06,
      "loss": 0.9014,
      "step": 5618
    },
    {
      "epoch": 0.7008856180616191,
      "grad_norm": 0.0,
      "learning_rate": 4.337269530211313e-06,
      "loss": 0.9099,
      "step": 5619
    },
    {
      "epoch": 0.7010103529998752,
      "grad_norm": 0.0,
      "learning_rate": 4.333940058928413e-06,
      "loss": 0.9377,
      "step": 5620
    },
    {
      "epoch": 0.7011350879381315,
      "grad_norm": 0.0,
      "learning_rate": 4.33061151248934e-06,
      "loss": 0.8721,
      "step": 5621
    },
    {
      "epoch": 0.7012598228763877,
      "grad_norm": 0.0,
      "learning_rate": 4.3272838914373885e-06,
      "loss": 0.9101,
      "step": 5622
    },
    {
      "epoch": 0.7013845578146439,
      "grad_norm": 0.0,
      "learning_rate": 4.323957196315714e-06,
      "loss": 0.8572,
      "step": 5623
    },
    {
      "epoch": 0.7015092927529001,
      "grad_norm": 0.0,
      "learning_rate": 4.320631427667321e-06,
      "loss": 0.9495,
      "step": 5624
    },
    {
      "epoch": 0.7016340276911563,
      "grad_norm": 0.0,
      "learning_rate": 4.317306586035049e-06,
      "loss": 0.9214,
      "step": 5625
    },
    {
      "epoch": 0.7017587626294125,
      "grad_norm": 0.0,
      "learning_rate": 4.313982671961609e-06,
      "loss": 0.9333,
      "step": 5626
    },
    {
      "epoch": 0.7018834975676687,
      "grad_norm": 0.0,
      "learning_rate": 4.3106596859895355e-06,
      "loss": 0.9292,
      "step": 5627
    },
    {
      "epoch": 0.7020082325059249,
      "grad_norm": 0.0,
      "learning_rate": 4.30733762866123e-06,
      "loss": 0.9032,
      "step": 5628
    },
    {
      "epoch": 0.7021329674441811,
      "grad_norm": 0.0,
      "learning_rate": 4.304016500518937e-06,
      "loss": 0.9651,
      "step": 5629
    },
    {
      "epoch": 0.7022577023824373,
      "grad_norm": 0.0,
      "learning_rate": 4.300696302104745e-06,
      "loss": 0.9431,
      "step": 5630
    },
    {
      "epoch": 0.7023824373206935,
      "grad_norm": 0.0,
      "learning_rate": 4.297377033960593e-06,
      "loss": 0.8985,
      "step": 5631
    },
    {
      "epoch": 0.7025071722589498,
      "grad_norm": 0.0,
      "learning_rate": 4.294058696628272e-06,
      "loss": 0.9211,
      "step": 5632
    },
    {
      "epoch": 0.7026319071972059,
      "grad_norm": 0.0,
      "learning_rate": 4.2907412906494175e-06,
      "loss": 0.9403,
      "step": 5633
    },
    {
      "epoch": 0.7027566421354622,
      "grad_norm": 0.0,
      "learning_rate": 4.287424816565517e-06,
      "loss": 0.8723,
      "step": 5634
    },
    {
      "epoch": 0.7028813770737183,
      "grad_norm": 0.0,
      "learning_rate": 4.284109274917897e-06,
      "loss": 0.9252,
      "step": 5635
    },
    {
      "epoch": 0.7030061120119746,
      "grad_norm": 0.0,
      "learning_rate": 4.28079466624774e-06,
      "loss": 0.9429,
      "step": 5636
    },
    {
      "epoch": 0.7031308469502308,
      "grad_norm": 0.0,
      "learning_rate": 4.277480991096073e-06,
      "loss": 0.9297,
      "step": 5637
    },
    {
      "epoch": 0.7032555818884869,
      "grad_norm": 0.0,
      "learning_rate": 4.274168250003776e-06,
      "loss": 0.8649,
      "step": 5638
    },
    {
      "epoch": 0.7033803168267432,
      "grad_norm": 0.0,
      "learning_rate": 4.270856443511565e-06,
      "loss": 0.9004,
      "step": 5639
    },
    {
      "epoch": 0.7035050517649993,
      "grad_norm": 0.0,
      "learning_rate": 4.267545572160014e-06,
      "loss": 0.904,
      "step": 5640
    },
    {
      "epoch": 0.7036297867032556,
      "grad_norm": 0.0,
      "learning_rate": 4.264235636489542e-06,
      "loss": 0.9086,
      "step": 5641
    },
    {
      "epoch": 0.7037545216415118,
      "grad_norm": 0.0,
      "learning_rate": 4.260926637040405e-06,
      "loss": 0.927,
      "step": 5642
    },
    {
      "epoch": 0.703879256579768,
      "grad_norm": 0.0,
      "learning_rate": 4.2576185743527294e-06,
      "loss": 0.8241,
      "step": 5643
    },
    {
      "epoch": 0.7040039915180242,
      "grad_norm": 0.0,
      "learning_rate": 4.254311448966468e-06,
      "loss": 0.8581,
      "step": 5644
    },
    {
      "epoch": 0.7041287264562804,
      "grad_norm": 0.0,
      "learning_rate": 4.251005261421418e-06,
      "loss": 0.8891,
      "step": 5645
    },
    {
      "epoch": 0.7042534613945366,
      "grad_norm": 0.0,
      "learning_rate": 4.247700012257249e-06,
      "loss": 0.9239,
      "step": 5646
    },
    {
      "epoch": 0.7043781963327929,
      "grad_norm": 0.0,
      "learning_rate": 4.244395702013448e-06,
      "loss": 0.9391,
      "step": 5647
    },
    {
      "epoch": 0.704502931271049,
      "grad_norm": 0.0,
      "learning_rate": 4.2410923312293665e-06,
      "loss": 0.9307,
      "step": 5648
    },
    {
      "epoch": 0.7046276662093053,
      "grad_norm": 0.0,
      "learning_rate": 4.237789900444197e-06,
      "loss": 0.9276,
      "step": 5649
    },
    {
      "epoch": 0.7047524011475614,
      "grad_norm": 0.0,
      "learning_rate": 4.234488410196985e-06,
      "loss": 0.9418,
      "step": 5650
    },
    {
      "epoch": 0.7048771360858176,
      "grad_norm": 0.0,
      "learning_rate": 4.231187861026608e-06,
      "loss": 0.9104,
      "step": 5651
    },
    {
      "epoch": 0.7050018710240739,
      "grad_norm": 0.0,
      "learning_rate": 4.2278882534718035e-06,
      "loss": 0.9075,
      "step": 5652
    },
    {
      "epoch": 0.70512660596233,
      "grad_norm": 0.0,
      "learning_rate": 4.22458958807115e-06,
      "loss": 0.8972,
      "step": 5653
    },
    {
      "epoch": 0.7052513409005863,
      "grad_norm": 0.0,
      "learning_rate": 4.221291865363073e-06,
      "loss": 0.9006,
      "step": 5654
    },
    {
      "epoch": 0.7053760758388424,
      "grad_norm": 0.0,
      "learning_rate": 4.2179950858858465e-06,
      "loss": 0.924,
      "step": 5655
    },
    {
      "epoch": 0.7055008107770987,
      "grad_norm": 0.0,
      "learning_rate": 4.214699250177583e-06,
      "loss": 0.898,
      "step": 5656
    },
    {
      "epoch": 0.7056255457153549,
      "grad_norm": 0.0,
      "learning_rate": 4.211404358776249e-06,
      "loss": 0.9549,
      "step": 5657
    },
    {
      "epoch": 0.7057502806536111,
      "grad_norm": 0.0,
      "learning_rate": 4.208110412219656e-06,
      "loss": 0.8477,
      "step": 5658
    },
    {
      "epoch": 0.7058750155918673,
      "grad_norm": 0.0,
      "learning_rate": 4.20481741104545e-06,
      "loss": 0.9342,
      "step": 5659
    },
    {
      "epoch": 0.7059997505301235,
      "grad_norm": 0.0,
      "learning_rate": 4.201525355791144e-06,
      "loss": 0.899,
      "step": 5660
    },
    {
      "epoch": 0.7061244854683797,
      "grad_norm": 0.0,
      "learning_rate": 4.198234246994079e-06,
      "loss": 0.9104,
      "step": 5661
    },
    {
      "epoch": 0.7062492204066358,
      "grad_norm": 0.0,
      "learning_rate": 4.194944085191441e-06,
      "loss": 0.9207,
      "step": 5662
    },
    {
      "epoch": 0.7063739553448921,
      "grad_norm": 0.0,
      "learning_rate": 4.1916548709202784e-06,
      "loss": 0.8626,
      "step": 5663
    },
    {
      "epoch": 0.7064986902831483,
      "grad_norm": 0.0,
      "learning_rate": 4.1883666047174665e-06,
      "loss": 0.9062,
      "step": 5664
    },
    {
      "epoch": 0.7066234252214045,
      "grad_norm": 0.0,
      "learning_rate": 4.185079287119733e-06,
      "loss": 0.9054,
      "step": 5665
    },
    {
      "epoch": 0.7067481601596607,
      "grad_norm": 0.0,
      "learning_rate": 4.181792918663655e-06,
      "loss": 0.9361,
      "step": 5666
    },
    {
      "epoch": 0.706872895097917,
      "grad_norm": 0.0,
      "learning_rate": 4.178507499885651e-06,
      "loss": 0.9133,
      "step": 5667
    },
    {
      "epoch": 0.7069976300361731,
      "grad_norm": 0.0,
      "learning_rate": 4.17522303132198e-06,
      "loss": 0.9161,
      "step": 5668
    },
    {
      "epoch": 0.7071223649744294,
      "grad_norm": 0.0,
      "learning_rate": 4.171939513508751e-06,
      "loss": 0.9463,
      "step": 5669
    },
    {
      "epoch": 0.7072470999126855,
      "grad_norm": 0.0,
      "learning_rate": 4.168656946981922e-06,
      "loss": 0.8916,
      "step": 5670
    },
    {
      "epoch": 0.7073718348509418,
      "grad_norm": 0.0,
      "learning_rate": 4.1653753322772795e-06,
      "loss": 0.9137,
      "step": 5671
    },
    {
      "epoch": 0.707496569789198,
      "grad_norm": 0.0,
      "learning_rate": 4.162094669930479e-06,
      "loss": 0.946,
      "step": 5672
    },
    {
      "epoch": 0.7076213047274542,
      "grad_norm": 0.0,
      "learning_rate": 4.158814960476999e-06,
      "loss": 0.9205,
      "step": 5673
    },
    {
      "epoch": 0.7077460396657104,
      "grad_norm": 0.0,
      "learning_rate": 4.1555362044521715e-06,
      "loss": 0.9582,
      "step": 5674
    },
    {
      "epoch": 0.7078707746039665,
      "grad_norm": 0.0,
      "learning_rate": 4.152258402391178e-06,
      "loss": 0.9258,
      "step": 5675
    },
    {
      "epoch": 0.7079955095422228,
      "grad_norm": 0.0,
      "learning_rate": 4.14898155482903e-06,
      "loss": 0.9239,
      "step": 5676
    },
    {
      "epoch": 0.7081202444804789,
      "grad_norm": 0.0,
      "learning_rate": 4.1457056623005954e-06,
      "loss": 0.8917,
      "step": 5677
    },
    {
      "epoch": 0.7082449794187352,
      "grad_norm": 0.0,
      "learning_rate": 4.142430725340587e-06,
      "loss": 0.9817,
      "step": 5678
    },
    {
      "epoch": 0.7083697143569914,
      "grad_norm": 0.0,
      "learning_rate": 4.139156744483545e-06,
      "loss": 0.8832,
      "step": 5679
    },
    {
      "epoch": 0.7084944492952476,
      "grad_norm": 0.0,
      "learning_rate": 4.13588372026388e-06,
      "loss": 0.8772,
      "step": 5680
    },
    {
      "epoch": 0.7086191842335038,
      "grad_norm": 0.0,
      "learning_rate": 4.132611653215822e-06,
      "loss": 0.9318,
      "step": 5681
    },
    {
      "epoch": 0.70874391917176,
      "grad_norm": 0.0,
      "learning_rate": 4.129340543873459e-06,
      "loss": 0.8697,
      "step": 5682
    },
    {
      "epoch": 0.7088686541100162,
      "grad_norm": 0.0,
      "learning_rate": 4.126070392770718e-06,
      "loss": 0.9466,
      "step": 5683
    },
    {
      "epoch": 0.7089933890482725,
      "grad_norm": 0.0,
      "learning_rate": 4.122801200441372e-06,
      "loss": 0.9006,
      "step": 5684
    },
    {
      "epoch": 0.7091181239865286,
      "grad_norm": 0.0,
      "learning_rate": 4.119532967419031e-06,
      "loss": 0.9406,
      "step": 5685
    },
    {
      "epoch": 0.7092428589247848,
      "grad_norm": 0.0,
      "learning_rate": 4.116265694237155e-06,
      "loss": 0.9361,
      "step": 5686
    },
    {
      "epoch": 0.709367593863041,
      "grad_norm": 0.0,
      "learning_rate": 4.112999381429049e-06,
      "loss": 0.8742,
      "step": 5687
    },
    {
      "epoch": 0.7094923288012972,
      "grad_norm": 0.0,
      "learning_rate": 4.1097340295278475e-06,
      "loss": 0.9531,
      "step": 5688
    },
    {
      "epoch": 0.7096170637395535,
      "grad_norm": 0.0,
      "learning_rate": 4.106469639066552e-06,
      "loss": 0.8974,
      "step": 5689
    },
    {
      "epoch": 0.7097417986778096,
      "grad_norm": 0.0,
      "learning_rate": 4.103206210577983e-06,
      "loss": 0.9572,
      "step": 5690
    },
    {
      "epoch": 0.7098665336160659,
      "grad_norm": 0.0,
      "learning_rate": 4.0999437445948185e-06,
      "loss": 0.8795,
      "step": 5691
    },
    {
      "epoch": 0.709991268554322,
      "grad_norm": 0.0,
      "learning_rate": 4.096682241649578e-06,
      "loss": 0.8974,
      "step": 5692
    },
    {
      "epoch": 0.7101160034925783,
      "grad_norm": 0.0,
      "learning_rate": 4.093421702274614e-06,
      "loss": 0.9446,
      "step": 5693
    },
    {
      "epoch": 0.7102407384308345,
      "grad_norm": 0.0,
      "learning_rate": 4.090162127002131e-06,
      "loss": 0.9555,
      "step": 5694
    },
    {
      "epoch": 0.7103654733690907,
      "grad_norm": 0.0,
      "learning_rate": 4.086903516364179e-06,
      "loss": 0.9725,
      "step": 5695
    },
    {
      "epoch": 0.7104902083073469,
      "grad_norm": 0.0,
      "learning_rate": 4.083645870892639e-06,
      "loss": 0.8969,
      "step": 5696
    },
    {
      "epoch": 0.7106149432456031,
      "grad_norm": 0.0,
      "learning_rate": 4.080389191119241e-06,
      "loss": 0.898,
      "step": 5697
    },
    {
      "epoch": 0.7107396781838593,
      "grad_norm": 0.0,
      "learning_rate": 4.077133477575562e-06,
      "loss": 0.935,
      "step": 5698
    },
    {
      "epoch": 0.7108644131221155,
      "grad_norm": 0.0,
      "learning_rate": 4.073878730793012e-06,
      "loss": 0.8836,
      "step": 5699
    },
    {
      "epoch": 0.7109891480603717,
      "grad_norm": 0.0,
      "learning_rate": 4.0706249513028516e-06,
      "loss": 0.9319,
      "step": 5700
    },
    {
      "epoch": 0.7111138829986279,
      "grad_norm": 0.0,
      "learning_rate": 4.067372139636181e-06,
      "loss": 0.934,
      "step": 5701
    },
    {
      "epoch": 0.7112386179368841,
      "grad_norm": 0.0,
      "learning_rate": 4.064120296323935e-06,
      "loss": 0.8865,
      "step": 5702
    },
    {
      "epoch": 0.7113633528751403,
      "grad_norm": 0.0,
      "learning_rate": 4.060869421896897e-06,
      "loss": 0.9401,
      "step": 5703
    },
    {
      "epoch": 0.7114880878133966,
      "grad_norm": 0.0,
      "learning_rate": 4.057619516885699e-06,
      "loss": 0.8878,
      "step": 5704
    },
    {
      "epoch": 0.7116128227516527,
      "grad_norm": 0.0,
      "learning_rate": 4.054370581820796e-06,
      "loss": 0.9371,
      "step": 5705
    },
    {
      "epoch": 0.711737557689909,
      "grad_norm": 0.0,
      "learning_rate": 4.051122617232508e-06,
      "loss": 0.9084,
      "step": 5706
    },
    {
      "epoch": 0.7118622926281651,
      "grad_norm": 0.0,
      "learning_rate": 4.047875623650981e-06,
      "loss": 0.9181,
      "step": 5707
    },
    {
      "epoch": 0.7119870275664214,
      "grad_norm": 0.0,
      "learning_rate": 4.0446296016061955e-06,
      "loss": 0.8848,
      "step": 5708
    },
    {
      "epoch": 0.7121117625046776,
      "grad_norm": 0.0,
      "learning_rate": 4.041384551628002e-06,
      "loss": 0.9695,
      "step": 5709
    },
    {
      "epoch": 0.7122364974429337,
      "grad_norm": 0.0,
      "learning_rate": 4.038140474246061e-06,
      "loss": 0.8751,
      "step": 5710
    },
    {
      "epoch": 0.71236123238119,
      "grad_norm": 0.0,
      "learning_rate": 4.034897369989891e-06,
      "loss": 0.9076,
      "step": 5711
    },
    {
      "epoch": 0.7124859673194461,
      "grad_norm": 0.0,
      "learning_rate": 4.031655239388854e-06,
      "loss": 0.9706,
      "step": 5712
    },
    {
      "epoch": 0.7126107022577024,
      "grad_norm": 0.0,
      "learning_rate": 4.028414082972141e-06,
      "loss": 0.8907,
      "step": 5713
    },
    {
      "epoch": 0.7127354371959586,
      "grad_norm": 0.0,
      "learning_rate": 4.025173901268791e-06,
      "loss": 0.9124,
      "step": 5714
    },
    {
      "epoch": 0.7128601721342148,
      "grad_norm": 0.0,
      "learning_rate": 4.021934694807685e-06,
      "loss": 0.9622,
      "step": 5715
    },
    {
      "epoch": 0.712984907072471,
      "grad_norm": 0.0,
      "learning_rate": 4.018696464117543e-06,
      "loss": 0.9129,
      "step": 5716
    },
    {
      "epoch": 0.7131096420107272,
      "grad_norm": 0.0,
      "learning_rate": 4.015459209726929e-06,
      "loss": 0.9443,
      "step": 5717
    },
    {
      "epoch": 0.7132343769489834,
      "grad_norm": 0.0,
      "learning_rate": 4.012222932164239e-06,
      "loss": 0.886,
      "step": 5718
    },
    {
      "epoch": 0.7133591118872397,
      "grad_norm": 0.0,
      "learning_rate": 4.008987631957717e-06,
      "loss": 0.9211,
      "step": 5719
    },
    {
      "epoch": 0.7134838468254958,
      "grad_norm": 0.0,
      "learning_rate": 4.005753309635446e-06,
      "loss": 0.8903,
      "step": 5720
    },
    {
      "epoch": 0.7136085817637521,
      "grad_norm": 0.0,
      "learning_rate": 4.002519965725352e-06,
      "loss": 0.9058,
      "step": 5721
    },
    {
      "epoch": 0.7137333167020082,
      "grad_norm": 0.0,
      "learning_rate": 3.999287600755192e-06,
      "loss": 0.9001,
      "step": 5722
    },
    {
      "epoch": 0.7138580516402644,
      "grad_norm": 0.0,
      "learning_rate": 3.996056215252573e-06,
      "loss": 0.8837,
      "step": 5723
    },
    {
      "epoch": 0.7139827865785207,
      "grad_norm": 0.0,
      "learning_rate": 3.992825809744941e-06,
      "loss": 0.9504,
      "step": 5724
    },
    {
      "epoch": 0.7141075215167768,
      "grad_norm": 0.0,
      "learning_rate": 3.989596384759571e-06,
      "loss": 0.9359,
      "step": 5725
    },
    {
      "epoch": 0.7142322564550331,
      "grad_norm": 0.0,
      "learning_rate": 3.986367940823598e-06,
      "loss": 0.9203,
      "step": 5726
    },
    {
      "epoch": 0.7143569913932892,
      "grad_norm": 0.0,
      "learning_rate": 3.983140478463977e-06,
      "loss": 0.8978,
      "step": 5727
    },
    {
      "epoch": 0.7144817263315455,
      "grad_norm": 0.0,
      "learning_rate": 3.979913998207514e-06,
      "loss": 0.8955,
      "step": 5728
    },
    {
      "epoch": 0.7146064612698017,
      "grad_norm": 0.0,
      "learning_rate": 3.976688500580857e-06,
      "loss": 0.8813,
      "step": 5729
    },
    {
      "epoch": 0.7147311962080579,
      "grad_norm": 0.0,
      "learning_rate": 3.973463986110478e-06,
      "loss": 0.8805,
      "step": 5730
    },
    {
      "epoch": 0.7148559311463141,
      "grad_norm": 0.0,
      "learning_rate": 3.970240455322705e-06,
      "loss": 0.9185,
      "step": 5731
    },
    {
      "epoch": 0.7149806660845703,
      "grad_norm": 0.0,
      "learning_rate": 3.9670179087437e-06,
      "loss": 0.9674,
      "step": 5732
    },
    {
      "epoch": 0.7151054010228265,
      "grad_norm": 0.0,
      "learning_rate": 3.963796346899465e-06,
      "loss": 0.9504,
      "step": 5733
    },
    {
      "epoch": 0.7152301359610826,
      "grad_norm": 0.0,
      "learning_rate": 3.9605757703158355e-06,
      "loss": 0.9665,
      "step": 5734
    },
    {
      "epoch": 0.7153548708993389,
      "grad_norm": 0.0,
      "learning_rate": 3.957356179518493e-06,
      "loss": 0.9022,
      "step": 5735
    },
    {
      "epoch": 0.7154796058375951,
      "grad_norm": 0.0,
      "learning_rate": 3.954137575032956e-06,
      "loss": 0.9096,
      "step": 5736
    },
    {
      "epoch": 0.7156043407758513,
      "grad_norm": 0.0,
      "learning_rate": 3.950919957384583e-06,
      "loss": 0.9055,
      "step": 5737
    },
    {
      "epoch": 0.7157290757141075,
      "grad_norm": 0.0,
      "learning_rate": 3.947703327098572e-06,
      "loss": 0.9328,
      "step": 5738
    },
    {
      "epoch": 0.7158538106523638,
      "grad_norm": 0.0,
      "learning_rate": 3.944487684699952e-06,
      "loss": 0.9042,
      "step": 5739
    },
    {
      "epoch": 0.7159785455906199,
      "grad_norm": 0.0,
      "learning_rate": 3.9412730307136e-06,
      "loss": 0.8821,
      "step": 5740
    },
    {
      "epoch": 0.7161032805288762,
      "grad_norm": 0.0,
      "learning_rate": 3.938059365664234e-06,
      "loss": 0.931,
      "step": 5741
    },
    {
      "epoch": 0.7162280154671323,
      "grad_norm": 0.0,
      "learning_rate": 3.934846690076393e-06,
      "loss": 0.9324,
      "step": 5742
    },
    {
      "epoch": 0.7163527504053886,
      "grad_norm": 0.0,
      "learning_rate": 3.93163500447448e-06,
      "loss": 0.9245,
      "step": 5743
    },
    {
      "epoch": 0.7164774853436447,
      "grad_norm": 0.0,
      "learning_rate": 3.9284243093827136e-06,
      "loss": 0.9189,
      "step": 5744
    },
    {
      "epoch": 0.716602220281901,
      "grad_norm": 0.0,
      "learning_rate": 3.925214605325164e-06,
      "loss": 0.9071,
      "step": 5745
    },
    {
      "epoch": 0.7167269552201572,
      "grad_norm": 0.0,
      "learning_rate": 3.922005892825737e-06,
      "loss": 0.8847,
      "step": 5746
    },
    {
      "epoch": 0.7168516901584133,
      "grad_norm": 0.0,
      "learning_rate": 3.918798172408173e-06,
      "loss": 0.9196,
      "step": 5747
    },
    {
      "epoch": 0.7169764250966696,
      "grad_norm": 0.0,
      "learning_rate": 3.9155914445960504e-06,
      "loss": 0.8998,
      "step": 5748
    },
    {
      "epoch": 0.7171011600349257,
      "grad_norm": 0.0,
      "learning_rate": 3.912385709912794e-06,
      "loss": 0.945,
      "step": 5749
    },
    {
      "epoch": 0.717225894973182,
      "grad_norm": 0.0,
      "learning_rate": 3.90918096888166e-06,
      "loss": 0.9155,
      "step": 5750
    },
    {
      "epoch": 0.7173506299114382,
      "grad_norm": 0.0,
      "learning_rate": 3.905977222025736e-06,
      "loss": 0.9098,
      "step": 5751
    },
    {
      "epoch": 0.7174753648496944,
      "grad_norm": 0.0,
      "learning_rate": 3.9027744698679595e-06,
      "loss": 0.9427,
      "step": 5752
    },
    {
      "epoch": 0.7176000997879506,
      "grad_norm": 0.0,
      "learning_rate": 3.8995727129311035e-06,
      "loss": 0.9382,
      "step": 5753
    },
    {
      "epoch": 0.7177248347262069,
      "grad_norm": 0.0,
      "learning_rate": 3.896371951737764e-06,
      "loss": 0.9258,
      "step": 5754
    },
    {
      "epoch": 0.717849569664463,
      "grad_norm": 0.0,
      "learning_rate": 3.893172186810401e-06,
      "loss": 0.902,
      "step": 5755
    },
    {
      "epoch": 0.7179743046027193,
      "grad_norm": 0.0,
      "learning_rate": 3.889973418671287e-06,
      "loss": 0.9654,
      "step": 5756
    },
    {
      "epoch": 0.7180990395409754,
      "grad_norm": 0.0,
      "learning_rate": 3.886775647842542e-06,
      "loss": 0.8619,
      "step": 5757
    },
    {
      "epoch": 0.7182237744792316,
      "grad_norm": 0.0,
      "learning_rate": 3.88357887484613e-06,
      "loss": 0.9294,
      "step": 5758
    },
    {
      "epoch": 0.7183485094174878,
      "grad_norm": 0.0,
      "learning_rate": 3.880383100203835e-06,
      "loss": 0.8533,
      "step": 5759
    },
    {
      "epoch": 0.718473244355744,
      "grad_norm": 0.0,
      "learning_rate": 3.877188324437293e-06,
      "loss": 0.9064,
      "step": 5760
    },
    {
      "epoch": 0.7185979792940003,
      "grad_norm": 0.0,
      "learning_rate": 3.873994548067972e-06,
      "loss": 0.9381,
      "step": 5761
    },
    {
      "epoch": 0.7187227142322564,
      "grad_norm": 0.0,
      "learning_rate": 3.870801771617176e-06,
      "loss": 0.8826,
      "step": 5762
    },
    {
      "epoch": 0.7188474491705127,
      "grad_norm": 0.0,
      "learning_rate": 3.867609995606051e-06,
      "loss": 0.9063,
      "step": 5763
    },
    {
      "epoch": 0.7189721841087688,
      "grad_norm": 0.0,
      "learning_rate": 3.864419220555568e-06,
      "loss": 0.932,
      "step": 5764
    },
    {
      "epoch": 0.7190969190470251,
      "grad_norm": 0.0,
      "learning_rate": 3.861229446986545e-06,
      "loss": 0.9086,
      "step": 5765
    },
    {
      "epoch": 0.7192216539852813,
      "grad_norm": 0.0,
      "learning_rate": 3.858040675419633e-06,
      "loss": 0.9069,
      "step": 5766
    },
    {
      "epoch": 0.7193463889235375,
      "grad_norm": 0.0,
      "learning_rate": 3.854852906375326e-06,
      "loss": 0.9414,
      "step": 5767
    },
    {
      "epoch": 0.7194711238617937,
      "grad_norm": 0.0,
      "learning_rate": 3.851666140373937e-06,
      "loss": 0.8929,
      "step": 5768
    },
    {
      "epoch": 0.71959585880005,
      "grad_norm": 0.0,
      "learning_rate": 3.848480377935635e-06,
      "loss": 0.9668,
      "step": 5769
    },
    {
      "epoch": 0.7197205937383061,
      "grad_norm": 0.0,
      "learning_rate": 3.845295619580417e-06,
      "loss": 0.9099,
      "step": 5770
    },
    {
      "epoch": 0.7198453286765623,
      "grad_norm": 0.0,
      "learning_rate": 3.842111865828105e-06,
      "loss": 0.8917,
      "step": 5771
    },
    {
      "epoch": 0.7199700636148185,
      "grad_norm": 0.0,
      "learning_rate": 3.838929117198385e-06,
      "loss": 0.9152,
      "step": 5772
    },
    {
      "epoch": 0.7200947985530747,
      "grad_norm": 0.0,
      "learning_rate": 3.835747374210747e-06,
      "loss": 0.9275,
      "step": 5773
    },
    {
      "epoch": 0.7202195334913309,
      "grad_norm": 0.0,
      "learning_rate": 3.83256663738454e-06,
      "loss": 0.9287,
      "step": 5774
    },
    {
      "epoch": 0.7203442684295871,
      "grad_norm": 0.0,
      "learning_rate": 3.829386907238939e-06,
      "loss": 0.9437,
      "step": 5775
    },
    {
      "epoch": 0.7204690033678434,
      "grad_norm": 0.0,
      "learning_rate": 3.826208184292952e-06,
      "loss": 0.9384,
      "step": 5776
    },
    {
      "epoch": 0.7205937383060995,
      "grad_norm": 0.0,
      "learning_rate": 3.823030469065431e-06,
      "loss": 0.9429,
      "step": 5777
    },
    {
      "epoch": 0.7207184732443558,
      "grad_norm": 0.0,
      "learning_rate": 3.819853762075057e-06,
      "loss": 0.8827,
      "step": 5778
    },
    {
      "epoch": 0.7208432081826119,
      "grad_norm": 0.0,
      "learning_rate": 3.816678063840353e-06,
      "loss": 0.9009,
      "step": 5779
    },
    {
      "epoch": 0.7209679431208682,
      "grad_norm": 0.0,
      "learning_rate": 3.8135033748796667e-06,
      "loss": 0.8927,
      "step": 5780
    },
    {
      "epoch": 0.7210926780591244,
      "grad_norm": 0.0,
      "learning_rate": 3.81032969571119e-06,
      "loss": 0.8921,
      "step": 5781
    },
    {
      "epoch": 0.7212174129973805,
      "grad_norm": 0.0,
      "learning_rate": 3.8071570268529477e-06,
      "loss": 0.8967,
      "step": 5782
    },
    {
      "epoch": 0.7213421479356368,
      "grad_norm": 0.0,
      "learning_rate": 3.803985368822799e-06,
      "loss": 0.9297,
      "step": 5783
    },
    {
      "epoch": 0.7214668828738929,
      "grad_norm": 0.0,
      "learning_rate": 3.800814722138443e-06,
      "loss": 0.912,
      "step": 5784
    },
    {
      "epoch": 0.7215916178121492,
      "grad_norm": 0.0,
      "learning_rate": 3.797645087317401e-06,
      "loss": 0.8987,
      "step": 5785
    },
    {
      "epoch": 0.7217163527504054,
      "grad_norm": 0.0,
      "learning_rate": 3.7944764648770404e-06,
      "loss": 0.9313,
      "step": 5786
    },
    {
      "epoch": 0.7218410876886616,
      "grad_norm": 0.0,
      "learning_rate": 3.791308855334566e-06,
      "loss": 0.89,
      "step": 5787
    },
    {
      "epoch": 0.7219658226269178,
      "grad_norm": 0.0,
      "learning_rate": 3.788142259206998e-06,
      "loss": 0.9337,
      "step": 5788
    },
    {
      "epoch": 0.722090557565174,
      "grad_norm": 0.0,
      "learning_rate": 3.7849766770112217e-06,
      "loss": 0.8874,
      "step": 5789
    },
    {
      "epoch": 0.7222152925034302,
      "grad_norm": 0.0,
      "learning_rate": 3.781812109263928e-06,
      "loss": 0.8844,
      "step": 5790
    },
    {
      "epoch": 0.7223400274416865,
      "grad_norm": 0.0,
      "learning_rate": 3.7786485564816577e-06,
      "loss": 0.8732,
      "step": 5791
    },
    {
      "epoch": 0.7224647623799426,
      "grad_norm": 0.0,
      "learning_rate": 3.775486019180785e-06,
      "loss": 0.9625,
      "step": 5792
    },
    {
      "epoch": 0.7225894973181989,
      "grad_norm": 0.0,
      "learning_rate": 3.772324497877511e-06,
      "loss": 0.933,
      "step": 5793
    },
    {
      "epoch": 0.722714232256455,
      "grad_norm": 0.0,
      "learning_rate": 3.7691639930878767e-06,
      "loss": 0.879,
      "step": 5794
    },
    {
      "epoch": 0.7228389671947112,
      "grad_norm": 0.0,
      "learning_rate": 3.766004505327757e-06,
      "loss": 0.9386,
      "step": 5795
    },
    {
      "epoch": 0.7229637021329675,
      "grad_norm": 0.0,
      "learning_rate": 3.7628460351128638e-06,
      "loss": 0.8814,
      "step": 5796
    },
    {
      "epoch": 0.7230884370712236,
      "grad_norm": 0.0,
      "learning_rate": 3.7596885829587325e-06,
      "loss": 0.9613,
      "step": 5797
    },
    {
      "epoch": 0.7232131720094799,
      "grad_norm": 0.0,
      "learning_rate": 3.7565321493807417e-06,
      "loss": 0.963,
      "step": 5798
    },
    {
      "epoch": 0.723337906947736,
      "grad_norm": 0.0,
      "learning_rate": 3.7533767348941043e-06,
      "loss": 0.9169,
      "step": 5799
    },
    {
      "epoch": 0.7234626418859923,
      "grad_norm": 0.0,
      "learning_rate": 3.750222340013854e-06,
      "loss": 0.8727,
      "step": 5800
    },
    {
      "epoch": 0.7235873768242485,
      "grad_norm": 0.0,
      "learning_rate": 3.74706896525488e-06,
      "loss": 0.9267,
      "step": 5801
    },
    {
      "epoch": 0.7237121117625047,
      "grad_norm": 0.0,
      "learning_rate": 3.7439166111318827e-06,
      "loss": 0.9467,
      "step": 5802
    },
    {
      "epoch": 0.7238368467007609,
      "grad_norm": 0.0,
      "learning_rate": 3.7407652781594094e-06,
      "loss": 0.8974,
      "step": 5803
    },
    {
      "epoch": 0.7239615816390171,
      "grad_norm": 0.0,
      "learning_rate": 3.7376149668518414e-06,
      "loss": 0.8569,
      "step": 5804
    },
    {
      "epoch": 0.7240863165772733,
      "grad_norm": 0.0,
      "learning_rate": 3.73446567772338e-06,
      "loss": 0.9718,
      "step": 5805
    },
    {
      "epoch": 0.7242110515155294,
      "grad_norm": 0.0,
      "learning_rate": 3.7313174112880745e-06,
      "loss": 0.9381,
      "step": 5806
    },
    {
      "epoch": 0.7243357864537857,
      "grad_norm": 0.0,
      "learning_rate": 3.728170168059799e-06,
      "loss": 0.9196,
      "step": 5807
    },
    {
      "epoch": 0.7244605213920419,
      "grad_norm": 0.0,
      "learning_rate": 3.725023948552264e-06,
      "loss": 0.9117,
      "step": 5808
    },
    {
      "epoch": 0.7245852563302981,
      "grad_norm": 0.0,
      "learning_rate": 3.7218787532790167e-06,
      "loss": 0.9157,
      "step": 5809
    },
    {
      "epoch": 0.7247099912685543,
      "grad_norm": 0.0,
      "learning_rate": 3.7187345827534226e-06,
      "loss": 0.9196,
      "step": 5810
    },
    {
      "epoch": 0.7248347262068106,
      "grad_norm": 0.0,
      "learning_rate": 3.7155914374886946e-06,
      "loss": 0.896,
      "step": 5811
    },
    {
      "epoch": 0.7249594611450667,
      "grad_norm": 0.0,
      "learning_rate": 3.7124493179978737e-06,
      "loss": 0.9526,
      "step": 5812
    },
    {
      "epoch": 0.725084196083323,
      "grad_norm": 0.0,
      "learning_rate": 3.709308224793835e-06,
      "loss": 0.9277,
      "step": 5813
    },
    {
      "epoch": 0.7252089310215791,
      "grad_norm": 0.0,
      "learning_rate": 3.7061681583892782e-06,
      "loss": 0.9171,
      "step": 5814
    },
    {
      "epoch": 0.7253336659598354,
      "grad_norm": 0.0,
      "learning_rate": 3.703029119296745e-06,
      "loss": 0.8983,
      "step": 5815
    },
    {
      "epoch": 0.7254584008980915,
      "grad_norm": 0.0,
      "learning_rate": 3.699891108028608e-06,
      "loss": 0.894,
      "step": 5816
    },
    {
      "epoch": 0.7255831358363478,
      "grad_norm": 0.0,
      "learning_rate": 3.6967541250970605e-06,
      "loss": 0.9496,
      "step": 5817
    },
    {
      "epoch": 0.725707870774604,
      "grad_norm": 0.0,
      "learning_rate": 3.693618171014152e-06,
      "loss": 0.9334,
      "step": 5818
    },
    {
      "epoch": 0.7258326057128601,
      "grad_norm": 0.0,
      "learning_rate": 3.690483246291735e-06,
      "loss": 0.9242,
      "step": 5819
    },
    {
      "epoch": 0.7259573406511164,
      "grad_norm": 0.0,
      "learning_rate": 3.687349351441516e-06,
      "loss": 0.9334,
      "step": 5820
    },
    {
      "epoch": 0.7260820755893725,
      "grad_norm": 0.0,
      "learning_rate": 3.6842164869750265e-06,
      "loss": 0.8922,
      "step": 5821
    },
    {
      "epoch": 0.7262068105276288,
      "grad_norm": 0.0,
      "learning_rate": 3.681084653403624e-06,
      "loss": 0.9357,
      "step": 5822
    },
    {
      "epoch": 0.726331545465885,
      "grad_norm": 0.0,
      "learning_rate": 3.677953851238504e-06,
      "loss": 0.8529,
      "step": 5823
    },
    {
      "epoch": 0.7264562804041412,
      "grad_norm": 0.0,
      "learning_rate": 3.674824080990693e-06,
      "loss": 0.9246,
      "step": 5824
    },
    {
      "epoch": 0.7265810153423974,
      "grad_norm": 0.0,
      "learning_rate": 3.6716953431710525e-06,
      "loss": 0.9153,
      "step": 5825
    },
    {
      "epoch": 0.7267057502806537,
      "grad_norm": 0.0,
      "learning_rate": 3.6685676382902647e-06,
      "loss": 0.9429,
      "step": 5826
    },
    {
      "epoch": 0.7268304852189098,
      "grad_norm": 0.0,
      "learning_rate": 3.665440966858852e-06,
      "loss": 0.9112,
      "step": 5827
    },
    {
      "epoch": 0.7269552201571661,
      "grad_norm": 0.0,
      "learning_rate": 3.662315329387167e-06,
      "loss": 0.9228,
      "step": 5828
    },
    {
      "epoch": 0.7270799550954222,
      "grad_norm": 0.0,
      "learning_rate": 3.6591907263853933e-06,
      "loss": 0.8718,
      "step": 5829
    },
    {
      "epoch": 0.7272046900336784,
      "grad_norm": 0.0,
      "learning_rate": 3.6560671583635467e-06,
      "loss": 0.8943,
      "step": 5830
    },
    {
      "epoch": 0.7273294249719346,
      "grad_norm": 0.0,
      "learning_rate": 3.652944625831466e-06,
      "loss": 0.868,
      "step": 5831
    },
    {
      "epoch": 0.7274541599101908,
      "grad_norm": 0.0,
      "learning_rate": 3.6498231292988317e-06,
      "loss": 0.9149,
      "step": 5832
    },
    {
      "epoch": 0.7275788948484471,
      "grad_norm": 0.0,
      "learning_rate": 3.646702669275152e-06,
      "loss": 0.8946,
      "step": 5833
    },
    {
      "epoch": 0.7277036297867032,
      "grad_norm": 0.0,
      "learning_rate": 3.6435832462697564e-06,
      "loss": 0.9112,
      "step": 5834
    },
    {
      "epoch": 0.7278283647249595,
      "grad_norm": 0.0,
      "learning_rate": 3.6404648607918247e-06,
      "loss": 0.9356,
      "step": 5835
    },
    {
      "epoch": 0.7279530996632156,
      "grad_norm": 0.0,
      "learning_rate": 3.637347513350352e-06,
      "loss": 0.8688,
      "step": 5836
    },
    {
      "epoch": 0.7280778346014719,
      "grad_norm": 0.0,
      "learning_rate": 3.6342312044541596e-06,
      "loss": 0.9469,
      "step": 5837
    },
    {
      "epoch": 0.7282025695397281,
      "grad_norm": 0.0,
      "learning_rate": 3.631115934611922e-06,
      "loss": 0.9533,
      "step": 5838
    },
    {
      "epoch": 0.7283273044779843,
      "grad_norm": 0.0,
      "learning_rate": 3.628001704332118e-06,
      "loss": 0.9127,
      "step": 5839
    },
    {
      "epoch": 0.7284520394162405,
      "grad_norm": 0.0,
      "learning_rate": 3.6248885141230737e-06,
      "loss": 0.9276,
      "step": 5840
    },
    {
      "epoch": 0.7285767743544967,
      "grad_norm": 0.0,
      "learning_rate": 3.6217763644929393e-06,
      "loss": 0.9196,
      "step": 5841
    },
    {
      "epoch": 0.7287015092927529,
      "grad_norm": 0.0,
      "learning_rate": 3.6186652559497006e-06,
      "loss": 0.9473,
      "step": 5842
    },
    {
      "epoch": 0.7288262442310091,
      "grad_norm": 0.0,
      "learning_rate": 3.615555189001161e-06,
      "loss": 0.9433,
      "step": 5843
    },
    {
      "epoch": 0.7289509791692653,
      "grad_norm": 0.0,
      "learning_rate": 3.6124461641549648e-06,
      "loss": 0.9552,
      "step": 5844
    },
    {
      "epoch": 0.7290757141075215,
      "grad_norm": 0.0,
      "learning_rate": 3.6093381819185837e-06,
      "loss": 0.8897,
      "step": 5845
    },
    {
      "epoch": 0.7292004490457777,
      "grad_norm": 0.0,
      "learning_rate": 3.6062312427993195e-06,
      "loss": 0.9629,
      "step": 5846
    },
    {
      "epoch": 0.7293251839840339,
      "grad_norm": 0.0,
      "learning_rate": 3.603125347304306e-06,
      "loss": 0.9053,
      "step": 5847
    },
    {
      "epoch": 0.7294499189222902,
      "grad_norm": 0.0,
      "learning_rate": 3.600020495940496e-06,
      "loss": 0.9402,
      "step": 5848
    },
    {
      "epoch": 0.7295746538605463,
      "grad_norm": 0.0,
      "learning_rate": 3.5969166892146844e-06,
      "loss": 0.9029,
      "step": 5849
    },
    {
      "epoch": 0.7296993887988026,
      "grad_norm": 0.0,
      "learning_rate": 3.593813927633494e-06,
      "loss": 0.9139,
      "step": 5850
    },
    {
      "epoch": 0.7298241237370587,
      "grad_norm": 0.0,
      "learning_rate": 3.5907122117033653e-06,
      "loss": 0.8779,
      "step": 5851
    },
    {
      "epoch": 0.729948858675315,
      "grad_norm": 0.0,
      "learning_rate": 3.58761154193058e-06,
      "loss": 0.9361,
      "step": 5852
    },
    {
      "epoch": 0.7300735936135712,
      "grad_norm": 0.0,
      "learning_rate": 3.5845119188212518e-06,
      "loss": 0.9064,
      "step": 5853
    },
    {
      "epoch": 0.7301983285518273,
      "grad_norm": 0.0,
      "learning_rate": 3.5814133428813035e-06,
      "loss": 0.9485,
      "step": 5854
    },
    {
      "epoch": 0.7303230634900836,
      "grad_norm": 0.0,
      "learning_rate": 3.5783158146165177e-06,
      "loss": 0.9257,
      "step": 5855
    },
    {
      "epoch": 0.7304477984283397,
      "grad_norm": 0.0,
      "learning_rate": 3.575219334532476e-06,
      "loss": 0.8881,
      "step": 5856
    },
    {
      "epoch": 0.730572533366596,
      "grad_norm": 0.0,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 0.927,
      "step": 5857
    },
    {
      "epoch": 0.7306972683048522,
      "grad_norm": 0.0,
      "learning_rate": 3.5690295209281634e-06,
      "loss": 0.903,
      "step": 5858
    },
    {
      "epoch": 0.7308220032431084,
      "grad_norm": 0.0,
      "learning_rate": 3.5659361884182285e-06,
      "loss": 0.9706,
      "step": 5859
    },
    {
      "epoch": 0.7309467381813646,
      "grad_norm": 0.0,
      "learning_rate": 3.5628439061097063e-06,
      "loss": 0.9311,
      "step": 5860
    },
    {
      "epoch": 0.7310714731196208,
      "grad_norm": 0.0,
      "learning_rate": 3.5597526745073387e-06,
      "loss": 0.9701,
      "step": 5861
    },
    {
      "epoch": 0.731196208057877,
      "grad_norm": 0.0,
      "learning_rate": 3.556662494115695e-06,
      "loss": 0.9137,
      "step": 5862
    },
    {
      "epoch": 0.7313209429961333,
      "grad_norm": 0.0,
      "learning_rate": 3.553573365439161e-06,
      "loss": 0.9547,
      "step": 5863
    },
    {
      "epoch": 0.7314456779343894,
      "grad_norm": 0.0,
      "learning_rate": 3.550485288981975e-06,
      "loss": 0.9812,
      "step": 5864
    },
    {
      "epoch": 0.7315704128726457,
      "grad_norm": 0.0,
      "learning_rate": 3.5473982652481773e-06,
      "loss": 0.9164,
      "step": 5865
    },
    {
      "epoch": 0.7316951478109018,
      "grad_norm": 0.0,
      "learning_rate": 3.544312294741652e-06,
      "loss": 0.9206,
      "step": 5866
    },
    {
      "epoch": 0.731819882749158,
      "grad_norm": 0.0,
      "learning_rate": 3.541227377966111e-06,
      "loss": 0.9516,
      "step": 5867
    },
    {
      "epoch": 0.7319446176874143,
      "grad_norm": 0.0,
      "learning_rate": 3.538143515425083e-06,
      "loss": 0.8635,
      "step": 5868
    },
    {
      "epoch": 0.7320693526256704,
      "grad_norm": 0.0,
      "learning_rate": 3.5350607076219358e-06,
      "loss": 0.9196,
      "step": 5869
    },
    {
      "epoch": 0.7321940875639267,
      "grad_norm": 0.0,
      "learning_rate": 3.531978955059864e-06,
      "loss": 0.8977,
      "step": 5870
    },
    {
      "epoch": 0.7323188225021828,
      "grad_norm": 0.0,
      "learning_rate": 3.5288982582418783e-06,
      "loss": 0.9292,
      "step": 5871
    },
    {
      "epoch": 0.7324435574404391,
      "grad_norm": 0.0,
      "learning_rate": 3.5258186176708385e-06,
      "loss": 0.9451,
      "step": 5872
    },
    {
      "epoch": 0.7325682923786953,
      "grad_norm": 0.0,
      "learning_rate": 3.522740033849411e-06,
      "loss": 0.9188,
      "step": 5873
    },
    {
      "epoch": 0.7326930273169515,
      "grad_norm": 0.0,
      "learning_rate": 3.5196625072801007e-06,
      "loss": 0.895,
      "step": 5874
    },
    {
      "epoch": 0.7328177622552077,
      "grad_norm": 0.0,
      "learning_rate": 3.5165860384652374e-06,
      "loss": 0.8268,
      "step": 5875
    },
    {
      "epoch": 0.7329424971934639,
      "grad_norm": 0.0,
      "learning_rate": 3.513510627906982e-06,
      "loss": 0.9149,
      "step": 5876
    },
    {
      "epoch": 0.7330672321317201,
      "grad_norm": 0.0,
      "learning_rate": 3.510436276107312e-06,
      "loss": 0.874,
      "step": 5877
    },
    {
      "epoch": 0.7331919670699762,
      "grad_norm": 0.0,
      "learning_rate": 3.5073629835680433e-06,
      "loss": 0.877,
      "step": 5878
    },
    {
      "epoch": 0.7333167020082325,
      "grad_norm": 0.0,
      "learning_rate": 3.5042907507908177e-06,
      "loss": 0.8525,
      "step": 5879
    },
    {
      "epoch": 0.7334414369464887,
      "grad_norm": 0.0,
      "learning_rate": 3.5012195782770896e-06,
      "loss": 0.9396,
      "step": 5880
    },
    {
      "epoch": 0.7335661718847449,
      "grad_norm": 0.0,
      "learning_rate": 3.4981494665281666e-06,
      "loss": 0.936,
      "step": 5881
    },
    {
      "epoch": 0.7336909068230011,
      "grad_norm": 0.0,
      "learning_rate": 3.4950804160451624e-06,
      "loss": 0.9376,
      "step": 5882
    },
    {
      "epoch": 0.7338156417612574,
      "grad_norm": 0.0,
      "learning_rate": 3.4920124273290148e-06,
      "loss": 0.8693,
      "step": 5883
    },
    {
      "epoch": 0.7339403766995135,
      "grad_norm": 0.0,
      "learning_rate": 3.4889455008805107e-06,
      "loss": 0.9365,
      "step": 5884
    },
    {
      "epoch": 0.7340651116377698,
      "grad_norm": 0.0,
      "learning_rate": 3.4858796372002412e-06,
      "loss": 0.9389,
      "step": 5885
    },
    {
      "epoch": 0.7341898465760259,
      "grad_norm": 0.0,
      "learning_rate": 3.482814836788635e-06,
      "loss": 0.9116,
      "step": 5886
    },
    {
      "epoch": 0.7343145815142822,
      "grad_norm": 0.0,
      "learning_rate": 3.4797511001459474e-06,
      "loss": 0.8644,
      "step": 5887
    },
    {
      "epoch": 0.7344393164525383,
      "grad_norm": 0.0,
      "learning_rate": 3.4766884277722524e-06,
      "loss": 0.8721,
      "step": 5888
    },
    {
      "epoch": 0.7345640513907946,
      "grad_norm": 0.0,
      "learning_rate": 3.473626820167457e-06,
      "loss": 0.8809,
      "step": 5889
    },
    {
      "epoch": 0.7346887863290508,
      "grad_norm": 0.0,
      "learning_rate": 3.470566277831293e-06,
      "loss": 0.9303,
      "step": 5890
    },
    {
      "epoch": 0.7348135212673069,
      "grad_norm": 0.0,
      "learning_rate": 3.4675068012633195e-06,
      "loss": 0.9267,
      "step": 5891
    },
    {
      "epoch": 0.7349382562055632,
      "grad_norm": 0.0,
      "learning_rate": 3.464448390962919e-06,
      "loss": 0.8901,
      "step": 5892
    },
    {
      "epoch": 0.7350629911438193,
      "grad_norm": 0.0,
      "learning_rate": 3.4613910474293045e-06,
      "loss": 0.8936,
      "step": 5893
    },
    {
      "epoch": 0.7351877260820756,
      "grad_norm": 0.0,
      "learning_rate": 3.4583347711615055e-06,
      "loss": 0.8997,
      "step": 5894
    },
    {
      "epoch": 0.7353124610203318,
      "grad_norm": 0.0,
      "learning_rate": 3.4552795626583855e-06,
      "loss": 0.8789,
      "step": 5895
    },
    {
      "epoch": 0.735437195958588,
      "grad_norm": 0.0,
      "learning_rate": 3.452225422418637e-06,
      "loss": 0.9725,
      "step": 5896
    },
    {
      "epoch": 0.7355619308968442,
      "grad_norm": 0.0,
      "learning_rate": 3.4491723509407603e-06,
      "loss": 0.9517,
      "step": 5897
    },
    {
      "epoch": 0.7356866658351005,
      "grad_norm": 0.0,
      "learning_rate": 3.4461203487231087e-06,
      "loss": 0.9172,
      "step": 5898
    },
    {
      "epoch": 0.7358114007733566,
      "grad_norm": 0.0,
      "learning_rate": 3.4430694162638378e-06,
      "loss": 0.8449,
      "step": 5899
    },
    {
      "epoch": 0.7359361357116129,
      "grad_norm": 0.0,
      "learning_rate": 3.4400195540609314e-06,
      "loss": 0.9364,
      "step": 5900
    },
    {
      "epoch": 0.736060870649869,
      "grad_norm": 0.0,
      "learning_rate": 3.4369707626122163e-06,
      "loss": 0.8798,
      "step": 5901
    },
    {
      "epoch": 0.7361856055881252,
      "grad_norm": 0.0,
      "learning_rate": 3.4339230424153225e-06,
      "loss": 0.9294,
      "step": 5902
    },
    {
      "epoch": 0.7363103405263814,
      "grad_norm": 0.0,
      "learning_rate": 3.430876393967718e-06,
      "loss": 0.9681,
      "step": 5903
    },
    {
      "epoch": 0.7364350754646376,
      "grad_norm": 0.0,
      "learning_rate": 3.4278308177666964e-06,
      "loss": 0.928,
      "step": 5904
    },
    {
      "epoch": 0.7365598104028939,
      "grad_norm": 0.0,
      "learning_rate": 3.424786314309365e-06,
      "loss": 0.9659,
      "step": 5905
    },
    {
      "epoch": 0.73668454534115,
      "grad_norm": 0.0,
      "learning_rate": 3.4217428840926683e-06,
      "loss": 0.877,
      "step": 5906
    },
    {
      "epoch": 0.7368092802794063,
      "grad_norm": 0.0,
      "learning_rate": 3.418700527613369e-06,
      "loss": 0.9108,
      "step": 5907
    },
    {
      "epoch": 0.7369340152176624,
      "grad_norm": 0.0,
      "learning_rate": 3.4156592453680604e-06,
      "loss": 0.9057,
      "step": 5908
    },
    {
      "epoch": 0.7370587501559187,
      "grad_norm": 0.0,
      "learning_rate": 3.4126190378531465e-06,
      "loss": 0.9127,
      "step": 5909
    },
    {
      "epoch": 0.7371834850941749,
      "grad_norm": 0.0,
      "learning_rate": 3.4095799055648793e-06,
      "loss": 0.9212,
      "step": 5910
    },
    {
      "epoch": 0.7373082200324311,
      "grad_norm": 0.0,
      "learning_rate": 3.4065418489993118e-06,
      "loss": 0.9149,
      "step": 5911
    },
    {
      "epoch": 0.7374329549706873,
      "grad_norm": 0.0,
      "learning_rate": 3.4035048686523344e-06,
      "loss": 0.8771,
      "step": 5912
    },
    {
      "epoch": 0.7375576899089435,
      "grad_norm": 0.0,
      "learning_rate": 3.4004689650196633e-06,
      "loss": 0.9328,
      "step": 5913
    },
    {
      "epoch": 0.7376824248471997,
      "grad_norm": 0.0,
      "learning_rate": 3.3974341385968256e-06,
      "loss": 0.9066,
      "step": 5914
    },
    {
      "epoch": 0.7378071597854559,
      "grad_norm": 0.0,
      "learning_rate": 3.3944003898791877e-06,
      "loss": 0.9112,
      "step": 5915
    },
    {
      "epoch": 0.7379318947237121,
      "grad_norm": 0.0,
      "learning_rate": 3.3913677193619356e-06,
      "loss": 0.922,
      "step": 5916
    },
    {
      "epoch": 0.7380566296619683,
      "grad_norm": 0.0,
      "learning_rate": 3.388336127540067e-06,
      "loss": 0.95,
      "step": 5917
    },
    {
      "epoch": 0.7381813646002245,
      "grad_norm": 0.0,
      "learning_rate": 3.3853056149084283e-06,
      "loss": 0.9196,
      "step": 5918
    },
    {
      "epoch": 0.7383060995384807,
      "grad_norm": 0.0,
      "learning_rate": 3.382276181961666e-06,
      "loss": 0.9222,
      "step": 5919
    },
    {
      "epoch": 0.738430834476737,
      "grad_norm": 0.0,
      "learning_rate": 3.3792478291942623e-06,
      "loss": 0.9095,
      "step": 5920
    },
    {
      "epoch": 0.7385555694149931,
      "grad_norm": 0.0,
      "learning_rate": 3.3762205571005236e-06,
      "loss": 0.9135,
      "step": 5921
    },
    {
      "epoch": 0.7386803043532494,
      "grad_norm": 0.0,
      "learning_rate": 3.3731943661745725e-06,
      "loss": 0.9087,
      "step": 5922
    },
    {
      "epoch": 0.7388050392915055,
      "grad_norm": 0.0,
      "learning_rate": 3.3701692569103607e-06,
      "loss": 0.8772,
      "step": 5923
    },
    {
      "epoch": 0.7389297742297618,
      "grad_norm": 0.0,
      "learning_rate": 3.3671452298016637e-06,
      "loss": 0.9226,
      "step": 5924
    },
    {
      "epoch": 0.739054509168018,
      "grad_norm": 0.0,
      "learning_rate": 3.364122285342082e-06,
      "loss": 0.9108,
      "step": 5925
    },
    {
      "epoch": 0.7391792441062741,
      "grad_norm": 0.0,
      "learning_rate": 3.3611004240250257e-06,
      "loss": 0.9229,
      "step": 5926
    },
    {
      "epoch": 0.7393039790445304,
      "grad_norm": 0.0,
      "learning_rate": 3.358079646343753e-06,
      "loss": 0.9232,
      "step": 5927
    },
    {
      "epoch": 0.7394287139827865,
      "grad_norm": 0.0,
      "learning_rate": 3.355059952791323e-06,
      "loss": 0.8677,
      "step": 5928
    },
    {
      "epoch": 0.7395534489210428,
      "grad_norm": 0.0,
      "learning_rate": 3.3520413438606215e-06,
      "loss": 0.8614,
      "step": 5929
    },
    {
      "epoch": 0.739678183859299,
      "grad_norm": 0.0,
      "learning_rate": 3.3490238200443727e-06,
      "loss": 0.9178,
      "step": 5930
    },
    {
      "epoch": 0.7398029187975552,
      "grad_norm": 0.0,
      "learning_rate": 3.3460073818351023e-06,
      "loss": 0.9405,
      "step": 5931
    },
    {
      "epoch": 0.7399276537358114,
      "grad_norm": 0.0,
      "learning_rate": 3.342992029725175e-06,
      "loss": 0.8983,
      "step": 5932
    },
    {
      "epoch": 0.7400523886740676,
      "grad_norm": 0.0,
      "learning_rate": 3.339977764206773e-06,
      "loss": 0.9192,
      "step": 5933
    },
    {
      "epoch": 0.7401771236123238,
      "grad_norm": 0.0,
      "learning_rate": 3.3369645857718958e-06,
      "loss": 0.8924,
      "step": 5934
    },
    {
      "epoch": 0.7403018585505801,
      "grad_norm": 0.0,
      "learning_rate": 3.3339524949123714e-06,
      "loss": 0.9372,
      "step": 5935
    },
    {
      "epoch": 0.7404265934888362,
      "grad_norm": 0.0,
      "learning_rate": 3.33094149211985e-06,
      "loss": 0.8959,
      "step": 5936
    },
    {
      "epoch": 0.7405513284270925,
      "grad_norm": 0.0,
      "learning_rate": 3.3279315778858034e-06,
      "loss": 0.8889,
      "step": 5937
    },
    {
      "epoch": 0.7406760633653486,
      "grad_norm": 0.0,
      "learning_rate": 3.324922752701528e-06,
      "loss": 0.8865,
      "step": 5938
    },
    {
      "epoch": 0.7408007983036048,
      "grad_norm": 0.0,
      "learning_rate": 3.321915017058135e-06,
      "loss": 0.9008,
      "step": 5939
    },
    {
      "epoch": 0.7409255332418611,
      "grad_norm": 0.0,
      "learning_rate": 3.318908371446563e-06,
      "loss": 0.8971,
      "step": 5940
    },
    {
      "epoch": 0.7410502681801172,
      "grad_norm": 0.0,
      "learning_rate": 3.3159028163575757e-06,
      "loss": 0.9115,
      "step": 5941
    },
    {
      "epoch": 0.7411750031183735,
      "grad_norm": 0.0,
      "learning_rate": 3.312898352281756e-06,
      "loss": 0.9124,
      "step": 5942
    },
    {
      "epoch": 0.7412997380566296,
      "grad_norm": 0.0,
      "learning_rate": 3.3098949797095e-06,
      "loss": 0.9044,
      "step": 5943
    },
    {
      "epoch": 0.7414244729948859,
      "grad_norm": 0.0,
      "learning_rate": 3.306892699131047e-06,
      "loss": 0.8948,
      "step": 5944
    },
    {
      "epoch": 0.741549207933142,
      "grad_norm": 0.0,
      "learning_rate": 3.3038915110364377e-06,
      "loss": 0.9039,
      "step": 5945
    },
    {
      "epoch": 0.7416739428713983,
      "grad_norm": 0.0,
      "learning_rate": 3.300891415915535e-06,
      "loss": 0.9319,
      "step": 5946
    },
    {
      "epoch": 0.7417986778096545,
      "grad_norm": 0.0,
      "learning_rate": 3.2978924142580427e-06,
      "loss": 0.9598,
      "step": 5947
    },
    {
      "epoch": 0.7419234127479107,
      "grad_norm": 0.0,
      "learning_rate": 3.2948945065534655e-06,
      "loss": 0.8994,
      "step": 5948
    },
    {
      "epoch": 0.7420481476861669,
      "grad_norm": 0.0,
      "learning_rate": 3.2918976932911385e-06,
      "loss": 0.8925,
      "step": 5949
    },
    {
      "epoch": 0.742172882624423,
      "grad_norm": 0.0,
      "learning_rate": 3.2889019749602213e-06,
      "loss": 0.9115,
      "step": 5950
    },
    {
      "epoch": 0.7422976175626793,
      "grad_norm": 0.0,
      "learning_rate": 3.2859073520496854e-06,
      "loss": 0.8979,
      "step": 5951
    },
    {
      "epoch": 0.7424223525009355,
      "grad_norm": 0.0,
      "learning_rate": 3.2829138250483304e-06,
      "loss": 0.948,
      "step": 5952
    },
    {
      "epoch": 0.7425470874391917,
      "grad_norm": 0.0,
      "learning_rate": 3.279921394444776e-06,
      "loss": 0.9125,
      "step": 5953
    },
    {
      "epoch": 0.7426718223774479,
      "grad_norm": 0.0,
      "learning_rate": 3.2769300607274625e-06,
      "loss": 0.9079,
      "step": 5954
    },
    {
      "epoch": 0.7427965573157042,
      "grad_norm": 0.0,
      "learning_rate": 3.273939824384653e-06,
      "loss": 0.9546,
      "step": 5955
    },
    {
      "epoch": 0.7429212922539603,
      "grad_norm": 0.0,
      "learning_rate": 3.2709506859044248e-06,
      "loss": 0.9178,
      "step": 5956
    },
    {
      "epoch": 0.7430460271922166,
      "grad_norm": 0.0,
      "learning_rate": 3.267962645774684e-06,
      "loss": 0.9348,
      "step": 5957
    },
    {
      "epoch": 0.7431707621304727,
      "grad_norm": 0.0,
      "learning_rate": 3.264975704483151e-06,
      "loss": 0.9374,
      "step": 5958
    },
    {
      "epoch": 0.743295497068729,
      "grad_norm": 0.0,
      "learning_rate": 3.2619898625173763e-06,
      "loss": 0.8804,
      "step": 5959
    },
    {
      "epoch": 0.7434202320069851,
      "grad_norm": 0.0,
      "learning_rate": 3.2590051203647176e-06,
      "loss": 0.9169,
      "step": 5960
    },
    {
      "epoch": 0.7435449669452414,
      "grad_norm": 0.0,
      "learning_rate": 3.256021478512362e-06,
      "loss": 0.8955,
      "step": 5961
    },
    {
      "epoch": 0.7436697018834976,
      "grad_norm": 0.0,
      "learning_rate": 3.253038937447319e-06,
      "loss": 0.9449,
      "step": 5962
    },
    {
      "epoch": 0.7437944368217537,
      "grad_norm": 0.0,
      "learning_rate": 3.250057497656406e-06,
      "loss": 0.851,
      "step": 5963
    },
    {
      "epoch": 0.74391917176001,
      "grad_norm": 0.0,
      "learning_rate": 3.2470771596262797e-06,
      "loss": 0.9054,
      "step": 5964
    },
    {
      "epoch": 0.7440439066982661,
      "grad_norm": 0.0,
      "learning_rate": 3.2440979238433977e-06,
      "loss": 0.9556,
      "step": 5965
    },
    {
      "epoch": 0.7441686416365224,
      "grad_norm": 0.0,
      "learning_rate": 3.2411197907940504e-06,
      "loss": 0.889,
      "step": 5966
    },
    {
      "epoch": 0.7442933765747786,
      "grad_norm": 0.0,
      "learning_rate": 3.238142760964348e-06,
      "loss": 0.9103,
      "step": 5967
    },
    {
      "epoch": 0.7444181115130348,
      "grad_norm": 0.0,
      "learning_rate": 3.2351668348402077e-06,
      "loss": 0.9121,
      "step": 5968
    },
    {
      "epoch": 0.744542846451291,
      "grad_norm": 0.0,
      "learning_rate": 3.2321920129073815e-06,
      "loss": 0.9369,
      "step": 5969
    },
    {
      "epoch": 0.7446675813895473,
      "grad_norm": 0.0,
      "learning_rate": 3.229218295651433e-06,
      "loss": 0.9098,
      "step": 5970
    },
    {
      "epoch": 0.7447923163278034,
      "grad_norm": 0.0,
      "learning_rate": 3.2262456835577525e-06,
      "loss": 0.9007,
      "step": 5971
    },
    {
      "epoch": 0.7449170512660597,
      "grad_norm": 0.0,
      "learning_rate": 3.2232741771115394e-06,
      "loss": 0.917,
      "step": 5972
    },
    {
      "epoch": 0.7450417862043158,
      "grad_norm": 0.0,
      "learning_rate": 3.220303776797821e-06,
      "loss": 0.9223,
      "step": 5973
    },
    {
      "epoch": 0.745166521142572,
      "grad_norm": 0.0,
      "learning_rate": 3.217334483101441e-06,
      "loss": 0.9149,
      "step": 5974
    },
    {
      "epoch": 0.7452912560808282,
      "grad_norm": 0.0,
      "learning_rate": 3.2143662965070634e-06,
      "loss": 0.8904,
      "step": 5975
    },
    {
      "epoch": 0.7454159910190844,
      "grad_norm": 0.0,
      "learning_rate": 3.2113992174991738e-06,
      "loss": 0.8728,
      "step": 5976
    },
    {
      "epoch": 0.7455407259573407,
      "grad_norm": 0.0,
      "learning_rate": 3.2084332465620692e-06,
      "loss": 0.8937,
      "step": 5977
    },
    {
      "epoch": 0.7456654608955968,
      "grad_norm": 0.0,
      "learning_rate": 3.2054683841798727e-06,
      "loss": 0.9415,
      "step": 5978
    },
    {
      "epoch": 0.7457901958338531,
      "grad_norm": 0.0,
      "learning_rate": 3.2025046308365284e-06,
      "loss": 0.8695,
      "step": 5979
    },
    {
      "epoch": 0.7459149307721092,
      "grad_norm": 0.0,
      "learning_rate": 3.1995419870157853e-06,
      "loss": 0.9321,
      "step": 5980
    },
    {
      "epoch": 0.7460396657103655,
      "grad_norm": 0.0,
      "learning_rate": 3.1965804532012357e-06,
      "loss": 0.8844,
      "step": 5981
    },
    {
      "epoch": 0.7461644006486217,
      "grad_norm": 0.0,
      "learning_rate": 3.1936200298762653e-06,
      "loss": 0.9502,
      "step": 5982
    },
    {
      "epoch": 0.7462891355868779,
      "grad_norm": 0.0,
      "learning_rate": 3.1906607175240943e-06,
      "loss": 0.9208,
      "step": 5983
    },
    {
      "epoch": 0.7464138705251341,
      "grad_norm": 0.0,
      "learning_rate": 3.187702516627761e-06,
      "loss": 0.9551,
      "step": 5984
    },
    {
      "epoch": 0.7465386054633903,
      "grad_norm": 0.0,
      "learning_rate": 3.1847454276701097e-06,
      "loss": 0.8954,
      "step": 5985
    },
    {
      "epoch": 0.7466633404016465,
      "grad_norm": 0.0,
      "learning_rate": 3.1817894511338152e-06,
      "loss": 0.8946,
      "step": 5986
    },
    {
      "epoch": 0.7467880753399027,
      "grad_norm": 0.0,
      "learning_rate": 3.1788345875013704e-06,
      "loss": 0.9102,
      "step": 5987
    },
    {
      "epoch": 0.7469128102781589,
      "grad_norm": 0.0,
      "learning_rate": 3.175880837255084e-06,
      "loss": 0.9249,
      "step": 5988
    },
    {
      "epoch": 0.7470375452164151,
      "grad_norm": 0.0,
      "learning_rate": 3.1729282008770767e-06,
      "loss": 0.8915,
      "step": 5989
    },
    {
      "epoch": 0.7471622801546713,
      "grad_norm": 0.0,
      "learning_rate": 3.169976678849296e-06,
      "loss": 0.9789,
      "step": 5990
    },
    {
      "epoch": 0.7472870150929275,
      "grad_norm": 0.0,
      "learning_rate": 3.1670262716535092e-06,
      "loss": 0.889,
      "step": 5991
    },
    {
      "epoch": 0.7474117500311838,
      "grad_norm": 0.0,
      "learning_rate": 3.1640769797712865e-06,
      "loss": 0.8803,
      "step": 5992
    },
    {
      "epoch": 0.7475364849694399,
      "grad_norm": 0.0,
      "learning_rate": 3.1611288036840393e-06,
      "loss": 0.8917,
      "step": 5993
    },
    {
      "epoch": 0.7476612199076962,
      "grad_norm": 0.0,
      "learning_rate": 3.1581817438729755e-06,
      "loss": 0.9205,
      "step": 5994
    },
    {
      "epoch": 0.7477859548459523,
      "grad_norm": 0.0,
      "learning_rate": 3.155235800819132e-06,
      "loss": 0.8391,
      "step": 5995
    },
    {
      "epoch": 0.7479106897842086,
      "grad_norm": 0.0,
      "learning_rate": 3.152290975003366e-06,
      "loss": 0.9229,
      "step": 5996
    },
    {
      "epoch": 0.7480354247224648,
      "grad_norm": 0.0,
      "learning_rate": 3.1493472669063373e-06,
      "loss": 0.9357,
      "step": 5997
    },
    {
      "epoch": 0.7481601596607209,
      "grad_norm": 0.0,
      "learning_rate": 3.1464046770085398e-06,
      "loss": 0.9685,
      "step": 5998
    },
    {
      "epoch": 0.7482848945989772,
      "grad_norm": 0.0,
      "learning_rate": 3.143463205790277e-06,
      "loss": 0.9484,
      "step": 5999
    },
    {
      "epoch": 0.7484096295372333,
      "grad_norm": 0.0,
      "learning_rate": 3.1405228537316713e-06,
      "loss": 0.9347,
      "step": 6000
    },
    {
      "epoch": 0.7485343644754896,
      "grad_norm": 0.0,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 0.9302,
      "step": 6001
    },
    {
      "epoch": 0.7486590994137458,
      "grad_norm": 0.0,
      "learning_rate": 3.1346455090130103e-06,
      "loss": 0.9319,
      "step": 6002
    },
    {
      "epoch": 0.748783834352002,
      "grad_norm": 0.0,
      "learning_rate": 3.131708517312283e-06,
      "loss": 0.893,
      "step": 6003
    },
    {
      "epoch": 0.7489085692902582,
      "grad_norm": 0.0,
      "learning_rate": 3.1287726466898737e-06,
      "loss": 0.9326,
      "step": 6004
    },
    {
      "epoch": 0.7490333042285144,
      "grad_norm": 0.0,
      "learning_rate": 3.1258378976249938e-06,
      "loss": 0.917,
      "step": 6005
    },
    {
      "epoch": 0.7491580391667706,
      "grad_norm": 0.0,
      "learning_rate": 3.122904270596663e-06,
      "loss": 0.8947,
      "step": 6006
    },
    {
      "epoch": 0.7492827741050269,
      "grad_norm": 0.0,
      "learning_rate": 3.119971766083726e-06,
      "loss": 0.8854,
      "step": 6007
    },
    {
      "epoch": 0.749407509043283,
      "grad_norm": 0.0,
      "learning_rate": 3.1170403845648443e-06,
      "loss": 0.8804,
      "step": 6008
    },
    {
      "epoch": 0.7495322439815393,
      "grad_norm": 0.0,
      "learning_rate": 3.1141101265184838e-06,
      "loss": 0.9266,
      "step": 6009
    },
    {
      "epoch": 0.7496569789197954,
      "grad_norm": 0.0,
      "learning_rate": 3.11118099242295e-06,
      "loss": 0.8957,
      "step": 6010
    },
    {
      "epoch": 0.7497817138580516,
      "grad_norm": 0.0,
      "learning_rate": 3.1082529827563435e-06,
      "loss": 0.9164,
      "step": 6011
    },
    {
      "epoch": 0.7499064487963079,
      "grad_norm": 0.0,
      "learning_rate": 3.1053260979965837e-06,
      "loss": 0.8879,
      "step": 6012
    },
    {
      "epoch": 0.750031183734564,
      "grad_norm": 0.0,
      "learning_rate": 3.102400338621425e-06,
      "loss": 0.864,
      "step": 6013
    },
    {
      "epoch": 0.7501559186728203,
      "grad_norm": 0.0,
      "learning_rate": 3.099475705108417e-06,
      "loss": 0.9273,
      "step": 6014
    },
    {
      "epoch": 0.7502806536110764,
      "grad_norm": 0.0,
      "learning_rate": 3.096552197934933e-06,
      "loss": 0.9336,
      "step": 6015
    },
    {
      "epoch": 0.7504053885493327,
      "grad_norm": 0.0,
      "learning_rate": 3.0936298175781664e-06,
      "loss": 0.9161,
      "step": 6016
    },
    {
      "epoch": 0.7505301234875889,
      "grad_norm": 0.0,
      "learning_rate": 3.090708564515125e-06,
      "loss": 0.9161,
      "step": 6017
    },
    {
      "epoch": 0.7506548584258451,
      "grad_norm": 0.0,
      "learning_rate": 3.0877884392226244e-06,
      "loss": 0.9252,
      "step": 6018
    },
    {
      "epoch": 0.7507795933641013,
      "grad_norm": 0.0,
      "learning_rate": 3.0848694421773075e-06,
      "loss": 0.9238,
      "step": 6019
    },
    {
      "epoch": 0.7509043283023575,
      "grad_norm": 0.0,
      "learning_rate": 3.081951573855626e-06,
      "loss": 0.8933,
      "step": 6020
    },
    {
      "epoch": 0.7510290632406137,
      "grad_norm": 0.0,
      "learning_rate": 3.0790348347338507e-06,
      "loss": 0.9152,
      "step": 6021
    },
    {
      "epoch": 0.7511537981788698,
      "grad_norm": 0.0,
      "learning_rate": 3.076119225288071e-06,
      "loss": 0.9422,
      "step": 6022
    },
    {
      "epoch": 0.7512785331171261,
      "grad_norm": 0.0,
      "learning_rate": 3.0732047459941793e-06,
      "loss": 0.8952,
      "step": 6023
    },
    {
      "epoch": 0.7514032680553823,
      "grad_norm": 0.0,
      "learning_rate": 3.070291397327898e-06,
      "loss": 0.9463,
      "step": 6024
    },
    {
      "epoch": 0.7515280029936385,
      "grad_norm": 0.0,
      "learning_rate": 3.0673791797647602e-06,
      "loss": 0.9185,
      "step": 6025
    },
    {
      "epoch": 0.7516527379318947,
      "grad_norm": 0.0,
      "learning_rate": 3.064468093780104e-06,
      "loss": 0.8995,
      "step": 6026
    },
    {
      "epoch": 0.751777472870151,
      "grad_norm": 0.0,
      "learning_rate": 3.0615581398491056e-06,
      "loss": 0.8675,
      "step": 6027
    },
    {
      "epoch": 0.7519022078084071,
      "grad_norm": 0.0,
      "learning_rate": 3.058649318446736e-06,
      "loss": 0.9198,
      "step": 6028
    },
    {
      "epoch": 0.7520269427466634,
      "grad_norm": 0.0,
      "learning_rate": 3.0557416300477805e-06,
      "loss": 0.8518,
      "step": 6029
    },
    {
      "epoch": 0.7521516776849195,
      "grad_norm": 0.0,
      "learning_rate": 3.052835075126862e-06,
      "loss": 0.8997,
      "step": 6030
    },
    {
      "epoch": 0.7522764126231758,
      "grad_norm": 0.0,
      "learning_rate": 3.0499296541583935e-06,
      "loss": 0.8863,
      "step": 6031
    },
    {
      "epoch": 0.752401147561432,
      "grad_norm": 0.0,
      "learning_rate": 3.047025367616615e-06,
      "loss": 0.9342,
      "step": 6032
    },
    {
      "epoch": 0.7525258824996882,
      "grad_norm": 0.0,
      "learning_rate": 3.04412221597558e-06,
      "loss": 0.9076,
      "step": 6033
    },
    {
      "epoch": 0.7526506174379444,
      "grad_norm": 0.0,
      "learning_rate": 3.0412201997091596e-06,
      "loss": 0.8985,
      "step": 6034
    },
    {
      "epoch": 0.7527753523762005,
      "grad_norm": 0.0,
      "learning_rate": 3.0383193192910287e-06,
      "loss": 0.8797,
      "step": 6035
    },
    {
      "epoch": 0.7529000873144568,
      "grad_norm": 0.0,
      "learning_rate": 3.035419575194689e-06,
      "loss": 0.9514,
      "step": 6036
    },
    {
      "epoch": 0.7530248222527129,
      "grad_norm": 0.0,
      "learning_rate": 3.032520967893453e-06,
      "loss": 0.8931,
      "step": 6037
    },
    {
      "epoch": 0.7531495571909692,
      "grad_norm": 0.0,
      "learning_rate": 3.029623497860438e-06,
      "loss": 0.9457,
      "step": 6038
    },
    {
      "epoch": 0.7532742921292254,
      "grad_norm": 0.0,
      "learning_rate": 3.0267271655685958e-06,
      "loss": 0.9256,
      "step": 6039
    },
    {
      "epoch": 0.7533990270674816,
      "grad_norm": 0.0,
      "learning_rate": 3.0238319714906737e-06,
      "loss": 0.9153,
      "step": 6040
    },
    {
      "epoch": 0.7535237620057378,
      "grad_norm": 0.0,
      "learning_rate": 3.0209379160992414e-06,
      "loss": 0.9165,
      "step": 6041
    },
    {
      "epoch": 0.753648496943994,
      "grad_norm": 0.0,
      "learning_rate": 3.0180449998666862e-06,
      "loss": 0.899,
      "step": 6042
    },
    {
      "epoch": 0.7537732318822502,
      "grad_norm": 0.0,
      "learning_rate": 3.015153223265197e-06,
      "loss": 0.913,
      "step": 6043
    },
    {
      "epoch": 0.7538979668205065,
      "grad_norm": 0.0,
      "learning_rate": 3.012262586766789e-06,
      "loss": 0.8598,
      "step": 6044
    },
    {
      "epoch": 0.7540227017587626,
      "grad_norm": 0.0,
      "learning_rate": 3.0093730908432894e-06,
      "loss": 0.8747,
      "step": 6045
    },
    {
      "epoch": 0.7541474366970188,
      "grad_norm": 0.0,
      "learning_rate": 3.0064847359663284e-06,
      "loss": 0.8655,
      "step": 6046
    },
    {
      "epoch": 0.754272171635275,
      "grad_norm": 0.0,
      "learning_rate": 3.00359752260737e-06,
      "loss": 0.863,
      "step": 6047
    },
    {
      "epoch": 0.7543969065735312,
      "grad_norm": 0.0,
      "learning_rate": 3.000711451237671e-06,
      "loss": 0.924,
      "step": 6048
    },
    {
      "epoch": 0.7545216415117875,
      "grad_norm": 0.0,
      "learning_rate": 2.9978265223283152e-06,
      "loss": 0.9125,
      "step": 6049
    },
    {
      "epoch": 0.7546463764500436,
      "grad_norm": 0.0,
      "learning_rate": 2.994942736350194e-06,
      "loss": 0.9602,
      "step": 6050
    },
    {
      "epoch": 0.7547711113882999,
      "grad_norm": 0.0,
      "learning_rate": 2.9920600937740175e-06,
      "loss": 0.9065,
      "step": 6051
    },
    {
      "epoch": 0.754895846326556,
      "grad_norm": 0.0,
      "learning_rate": 2.9891785950703e-06,
      "loss": 0.9802,
      "step": 6052
    },
    {
      "epoch": 0.7550205812648123,
      "grad_norm": 0.0,
      "learning_rate": 2.986298240709378e-06,
      "loss": 0.9815,
      "step": 6053
    },
    {
      "epoch": 0.7551453162030685,
      "grad_norm": 0.0,
      "learning_rate": 2.983419031161401e-06,
      "loss": 0.9187,
      "step": 6054
    },
    {
      "epoch": 0.7552700511413247,
      "grad_norm": 0.0,
      "learning_rate": 2.980540966896317e-06,
      "loss": 0.9525,
      "step": 6055
    },
    {
      "epoch": 0.7553947860795809,
      "grad_norm": 0.0,
      "learning_rate": 2.9776640483839146e-06,
      "loss": 0.9127,
      "step": 6056
    },
    {
      "epoch": 0.7555195210178371,
      "grad_norm": 0.0,
      "learning_rate": 2.9747882760937684e-06,
      "loss": 0.8579,
      "step": 6057
    },
    {
      "epoch": 0.7556442559560933,
      "grad_norm": 0.0,
      "learning_rate": 2.9719136504952793e-06,
      "loss": 0.9292,
      "step": 6058
    },
    {
      "epoch": 0.7557689908943495,
      "grad_norm": 0.0,
      "learning_rate": 2.9690401720576634e-06,
      "loss": 0.8996,
      "step": 6059
    },
    {
      "epoch": 0.7558937258326057,
      "grad_norm": 0.0,
      "learning_rate": 2.966167841249937e-06,
      "loss": 0.9068,
      "step": 6060
    },
    {
      "epoch": 0.7560184607708619,
      "grad_norm": 0.0,
      "learning_rate": 2.9632966585409415e-06,
      "loss": 0.8952,
      "step": 6061
    },
    {
      "epoch": 0.7561431957091181,
      "grad_norm": 0.0,
      "learning_rate": 2.9604266243993287e-06,
      "loss": 0.9394,
      "step": 6062
    },
    {
      "epoch": 0.7562679306473743,
      "grad_norm": 0.0,
      "learning_rate": 2.957557739293554e-06,
      "loss": 0.9033,
      "step": 6063
    },
    {
      "epoch": 0.7563926655856306,
      "grad_norm": 0.0,
      "learning_rate": 2.9546900036918956e-06,
      "loss": 0.9265,
      "step": 6064
    },
    {
      "epoch": 0.7565174005238867,
      "grad_norm": 0.0,
      "learning_rate": 2.9518234180624393e-06,
      "loss": 0.9282,
      "step": 6065
    },
    {
      "epoch": 0.756642135462143,
      "grad_norm": 0.0,
      "learning_rate": 2.948957982873085e-06,
      "loss": 0.899,
      "step": 6066
    },
    {
      "epoch": 0.7567668704003991,
      "grad_norm": 0.0,
      "learning_rate": 2.9460936985915444e-06,
      "loss": 0.9529,
      "step": 6067
    },
    {
      "epoch": 0.7568916053386554,
      "grad_norm": 0.0,
      "learning_rate": 2.9432305656853423e-06,
      "loss": 0.9377,
      "step": 6068
    },
    {
      "epoch": 0.7570163402769116,
      "grad_norm": 0.0,
      "learning_rate": 2.940368584621809e-06,
      "loss": 0.9526,
      "step": 6069
    },
    {
      "epoch": 0.7571410752151677,
      "grad_norm": 0.0,
      "learning_rate": 2.9375077558680954e-06,
      "loss": 0.946,
      "step": 6070
    },
    {
      "epoch": 0.757265810153424,
      "grad_norm": 0.0,
      "learning_rate": 2.9346480798911636e-06,
      "loss": 0.9173,
      "step": 6071
    },
    {
      "epoch": 0.7573905450916801,
      "grad_norm": 0.0,
      "learning_rate": 2.931789557157775e-06,
      "loss": 0.8712,
      "step": 6072
    },
    {
      "epoch": 0.7575152800299364,
      "grad_norm": 0.0,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.8973,
      "step": 6073
    },
    {
      "epoch": 0.7576400149681926,
      "grad_norm": 0.0,
      "learning_rate": 2.9260759732878042e-06,
      "loss": 0.8909,
      "step": 6074
    },
    {
      "epoch": 0.7577647499064488,
      "grad_norm": 0.0,
      "learning_rate": 2.9232209130838094e-06,
      "loss": 0.9203,
      "step": 6075
    },
    {
      "epoch": 0.757889484844705,
      "grad_norm": 0.0,
      "learning_rate": 2.9203670079885725e-06,
      "loss": 0.9193,
      "step": 6076
    },
    {
      "epoch": 0.7580142197829612,
      "grad_norm": 0.0,
      "learning_rate": 2.917514258467914e-06,
      "loss": 0.9116,
      "step": 6077
    },
    {
      "epoch": 0.7581389547212174,
      "grad_norm": 0.0,
      "learning_rate": 2.9146626649874764e-06,
      "loss": 0.9091,
      "step": 6078
    },
    {
      "epoch": 0.7582636896594737,
      "grad_norm": 0.0,
      "learning_rate": 2.9118122280127158e-06,
      "loss": 0.8933,
      "step": 6079
    },
    {
      "epoch": 0.7583884245977298,
      "grad_norm": 0.0,
      "learning_rate": 2.9089629480088886e-06,
      "loss": 0.9698,
      "step": 6080
    },
    {
      "epoch": 0.7585131595359861,
      "grad_norm": 0.0,
      "learning_rate": 2.906114825441072e-06,
      "loss": 0.8538,
      "step": 6081
    },
    {
      "epoch": 0.7586378944742422,
      "grad_norm": 0.0,
      "learning_rate": 2.9032678607741526e-06,
      "loss": 0.9152,
      "step": 6082
    },
    {
      "epoch": 0.7587626294124984,
      "grad_norm": 0.0,
      "learning_rate": 2.9004220544728243e-06,
      "loss": 0.9145,
      "step": 6083
    },
    {
      "epoch": 0.7588873643507547,
      "grad_norm": 0.0,
      "learning_rate": 2.897577407001597e-06,
      "loss": 0.9393,
      "step": 6084
    },
    {
      "epoch": 0.7590120992890108,
      "grad_norm": 0.0,
      "learning_rate": 2.8947339188247916e-06,
      "loss": 0.9195,
      "step": 6085
    },
    {
      "epoch": 0.7591368342272671,
      "grad_norm": 0.0,
      "learning_rate": 2.8918915904065305e-06,
      "loss": 0.9069,
      "step": 6086
    },
    {
      "epoch": 0.7592615691655232,
      "grad_norm": 0.0,
      "learning_rate": 2.8890504222107553e-06,
      "loss": 0.8874,
      "step": 6087
    },
    {
      "epoch": 0.7593863041037795,
      "grad_norm": 0.0,
      "learning_rate": 2.8862104147012203e-06,
      "loss": 0.9347,
      "step": 6088
    },
    {
      "epoch": 0.7595110390420357,
      "grad_norm": 0.0,
      "learning_rate": 2.8833715683414796e-06,
      "loss": 0.9676,
      "step": 6089
    },
    {
      "epoch": 0.7596357739802919,
      "grad_norm": 0.0,
      "learning_rate": 2.8805338835949093e-06,
      "loss": 0.9172,
      "step": 6090
    },
    {
      "epoch": 0.7597605089185481,
      "grad_norm": 0.0,
      "learning_rate": 2.877697360924693e-06,
      "loss": 0.9044,
      "step": 6091
    },
    {
      "epoch": 0.7598852438568043,
      "grad_norm": 0.0,
      "learning_rate": 2.874862000793812e-06,
      "loss": 0.9041,
      "step": 6092
    },
    {
      "epoch": 0.7600099787950605,
      "grad_norm": 0.0,
      "learning_rate": 2.872027803665084e-06,
      "loss": 0.8964,
      "step": 6093
    },
    {
      "epoch": 0.7601347137333166,
      "grad_norm": 0.0,
      "learning_rate": 2.8691947700011103e-06,
      "loss": 0.8687,
      "step": 6094
    },
    {
      "epoch": 0.7602594486715729,
      "grad_norm": 0.0,
      "learning_rate": 2.8663629002643157e-06,
      "loss": 0.9108,
      "step": 6095
    },
    {
      "epoch": 0.7603841836098291,
      "grad_norm": 0.0,
      "learning_rate": 2.863532194916937e-06,
      "loss": 0.9251,
      "step": 6096
    },
    {
      "epoch": 0.7605089185480853,
      "grad_norm": 0.0,
      "learning_rate": 2.8607026544210115e-06,
      "loss": 0.9137,
      "step": 6097
    },
    {
      "epoch": 0.7606336534863415,
      "grad_norm": 0.0,
      "learning_rate": 2.857874279238393e-06,
      "loss": 0.9575,
      "step": 6098
    },
    {
      "epoch": 0.7607583884245978,
      "grad_norm": 0.0,
      "learning_rate": 2.855047069830743e-06,
      "loss": 0.8949,
      "step": 6099
    },
    {
      "epoch": 0.7608831233628539,
      "grad_norm": 0.0,
      "learning_rate": 2.8522210266595386e-06,
      "loss": 0.8936,
      "step": 6100
    },
    {
      "epoch": 0.7610078583011102,
      "grad_norm": 0.0,
      "learning_rate": 2.849396150186051e-06,
      "loss": 0.9086,
      "step": 6101
    },
    {
      "epoch": 0.7611325932393663,
      "grad_norm": 0.0,
      "learning_rate": 2.8465724408713834e-06,
      "loss": 0.8997,
      "step": 6102
    },
    {
      "epoch": 0.7612573281776226,
      "grad_norm": 0.0,
      "learning_rate": 2.8437498991764266e-06,
      "loss": 0.9041,
      "step": 6103
    },
    {
      "epoch": 0.7613820631158787,
      "grad_norm": 0.0,
      "learning_rate": 2.8409285255618948e-06,
      "loss": 0.8882,
      "step": 6104
    },
    {
      "epoch": 0.761506798054135,
      "grad_norm": 0.0,
      "learning_rate": 2.83810832048831e-06,
      "loss": 0.8993,
      "step": 6105
    },
    {
      "epoch": 0.7616315329923912,
      "grad_norm": 0.0,
      "learning_rate": 2.835289284415994e-06,
      "loss": 0.9242,
      "step": 6106
    },
    {
      "epoch": 0.7617562679306473,
      "grad_norm": 0.0,
      "learning_rate": 2.8324714178050895e-06,
      "loss": 0.9274,
      "step": 6107
    },
    {
      "epoch": 0.7618810028689036,
      "grad_norm": 0.0,
      "learning_rate": 2.8296547211155435e-06,
      "loss": 0.8718,
      "step": 6108
    },
    {
      "epoch": 0.7620057378071597,
      "grad_norm": 0.0,
      "learning_rate": 2.826839194807105e-06,
      "loss": 0.9256,
      "step": 6109
    },
    {
      "epoch": 0.762130472745416,
      "grad_norm": 0.0,
      "learning_rate": 2.8240248393393498e-06,
      "loss": 0.9099,
      "step": 6110
    },
    {
      "epoch": 0.7622552076836722,
      "grad_norm": 0.0,
      "learning_rate": 2.821211655171644e-06,
      "loss": 0.885,
      "step": 6111
    },
    {
      "epoch": 0.7623799426219284,
      "grad_norm": 0.0,
      "learning_rate": 2.818399642763171e-06,
      "loss": 0.9123,
      "step": 6112
    },
    {
      "epoch": 0.7625046775601846,
      "grad_norm": 0.0,
      "learning_rate": 2.815588802572927e-06,
      "loss": 0.9242,
      "step": 6113
    },
    {
      "epoch": 0.7626294124984409,
      "grad_norm": 0.0,
      "learning_rate": 2.8127791350597067e-06,
      "loss": 0.975,
      "step": 6114
    },
    {
      "epoch": 0.762754147436697,
      "grad_norm": 0.0,
      "learning_rate": 2.809970640682119e-06,
      "loss": 0.9072,
      "step": 6115
    },
    {
      "epoch": 0.7628788823749533,
      "grad_norm": 0.0,
      "learning_rate": 2.8071633198985825e-06,
      "loss": 0.8814,
      "step": 6116
    },
    {
      "epoch": 0.7630036173132094,
      "grad_norm": 0.0,
      "learning_rate": 2.8043571731673256e-06,
      "loss": 0.9143,
      "step": 6117
    },
    {
      "epoch": 0.7631283522514656,
      "grad_norm": 0.0,
      "learning_rate": 2.8015522009463736e-06,
      "loss": 0.9149,
      "step": 6118
    },
    {
      "epoch": 0.7632530871897218,
      "grad_norm": 0.0,
      "learning_rate": 2.7987484036935797e-06,
      "loss": 0.9046,
      "step": 6119
    },
    {
      "epoch": 0.763377822127978,
      "grad_norm": 0.0,
      "learning_rate": 2.795945781866589e-06,
      "loss": 0.9209,
      "step": 6120
    },
    {
      "epoch": 0.7635025570662343,
      "grad_norm": 0.0,
      "learning_rate": 2.793144335922854e-06,
      "loss": 0.9731,
      "step": 6121
    },
    {
      "epoch": 0.7636272920044904,
      "grad_norm": 0.0,
      "learning_rate": 2.790344066319652e-06,
      "loss": 0.9288,
      "step": 6122
    },
    {
      "epoch": 0.7637520269427467,
      "grad_norm": 0.0,
      "learning_rate": 2.7875449735140516e-06,
      "loss": 0.9335,
      "step": 6123
    },
    {
      "epoch": 0.7638767618810028,
      "grad_norm": 0.0,
      "learning_rate": 2.7847470579629356e-06,
      "loss": 0.9211,
      "step": 6124
    },
    {
      "epoch": 0.7640014968192591,
      "grad_norm": 0.0,
      "learning_rate": 2.7819503201229993e-06,
      "loss": 0.9087,
      "step": 6125
    },
    {
      "epoch": 0.7641262317575153,
      "grad_norm": 0.0,
      "learning_rate": 2.779154760450732e-06,
      "loss": 0.9081,
      "step": 6126
    },
    {
      "epoch": 0.7642509666957715,
      "grad_norm": 0.0,
      "learning_rate": 2.776360379402445e-06,
      "loss": 0.9482,
      "step": 6127
    },
    {
      "epoch": 0.7643757016340277,
      "grad_norm": 0.0,
      "learning_rate": 2.7735671774342508e-06,
      "loss": 0.9323,
      "step": 6128
    },
    {
      "epoch": 0.764500436572284,
      "grad_norm": 0.0,
      "learning_rate": 2.770775155002071e-06,
      "loss": 0.9283,
      "step": 6129
    },
    {
      "epoch": 0.7646251715105401,
      "grad_norm": 0.0,
      "learning_rate": 2.767984312561637e-06,
      "loss": 0.9036,
      "step": 6130
    },
    {
      "epoch": 0.7647499064487963,
      "grad_norm": 0.0,
      "learning_rate": 2.7651946505684778e-06,
      "loss": 0.9386,
      "step": 6131
    },
    {
      "epoch": 0.7648746413870525,
      "grad_norm": 0.0,
      "learning_rate": 2.7624061694779403e-06,
      "loss": 0.9301,
      "step": 6132
    },
    {
      "epoch": 0.7649993763253087,
      "grad_norm": 0.0,
      "learning_rate": 2.7596188697451742e-06,
      "loss": 0.9133,
      "step": 6133
    },
    {
      "epoch": 0.7651241112635649,
      "grad_norm": 0.0,
      "learning_rate": 2.7568327518251405e-06,
      "loss": 0.9144,
      "step": 6134
    },
    {
      "epoch": 0.7652488462018211,
      "grad_norm": 0.0,
      "learning_rate": 2.754047816172595e-06,
      "loss": 0.8939,
      "step": 6135
    },
    {
      "epoch": 0.7653735811400774,
      "grad_norm": 0.0,
      "learning_rate": 2.751264063242122e-06,
      "loss": 0.8747,
      "step": 6136
    },
    {
      "epoch": 0.7654983160783335,
      "grad_norm": 0.0,
      "learning_rate": 2.7484814934880923e-06,
      "loss": 0.9412,
      "step": 6137
    },
    {
      "epoch": 0.7656230510165898,
      "grad_norm": 0.0,
      "learning_rate": 2.7457001073646874e-06,
      "loss": 0.9515,
      "step": 6138
    },
    {
      "epoch": 0.7657477859548459,
      "grad_norm": 0.0,
      "learning_rate": 2.7429199053259105e-06,
      "loss": 0.9561,
      "step": 6139
    },
    {
      "epoch": 0.7658725208931022,
      "grad_norm": 0.0,
      "learning_rate": 2.7401408878255518e-06,
      "loss": 0.8855,
      "step": 6140
    },
    {
      "epoch": 0.7659972558313584,
      "grad_norm": 0.0,
      "learning_rate": 2.7373630553172204e-06,
      "loss": 0.9259,
      "step": 6141
    },
    {
      "epoch": 0.7661219907696145,
      "grad_norm": 0.0,
      "learning_rate": 2.7345864082543317e-06,
      "loss": 0.8887,
      "step": 6142
    },
    {
      "epoch": 0.7662467257078708,
      "grad_norm": 0.0,
      "learning_rate": 2.7318109470900987e-06,
      "loss": 0.9155,
      "step": 6143
    },
    {
      "epoch": 0.7663714606461269,
      "grad_norm": 0.0,
      "learning_rate": 2.7290366722775486e-06,
      "loss": 0.8714,
      "step": 6144
    },
    {
      "epoch": 0.7664961955843832,
      "grad_norm": 0.0,
      "learning_rate": 2.726263584269513e-06,
      "loss": 0.9241,
      "step": 6145
    },
    {
      "epoch": 0.7666209305226394,
      "grad_norm": 0.0,
      "learning_rate": 2.7234916835186343e-06,
      "loss": 0.9237,
      "step": 6146
    },
    {
      "epoch": 0.7667456654608956,
      "grad_norm": 0.0,
      "learning_rate": 2.7207209704773485e-06,
      "loss": 0.9035,
      "step": 6147
    },
    {
      "epoch": 0.7668704003991518,
      "grad_norm": 0.0,
      "learning_rate": 2.7179514455979104e-06,
      "loss": 0.8991,
      "step": 6148
    },
    {
      "epoch": 0.766995135337408,
      "grad_norm": 0.0,
      "learning_rate": 2.7151831093323756e-06,
      "loss": 0.9149,
      "step": 6149
    },
    {
      "epoch": 0.7671198702756642,
      "grad_norm": 0.0,
      "learning_rate": 2.7124159621326074e-06,
      "loss": 0.9639,
      "step": 6150
    },
    {
      "epoch": 0.7672446052139205,
      "grad_norm": 0.0,
      "learning_rate": 2.709650004450275e-06,
      "loss": 0.8806,
      "step": 6151
    },
    {
      "epoch": 0.7673693401521766,
      "grad_norm": 0.0,
      "learning_rate": 2.706885236736848e-06,
      "loss": 0.9032,
      "step": 6152
    },
    {
      "epoch": 0.7674940750904329,
      "grad_norm": 0.0,
      "learning_rate": 2.7041216594436093e-06,
      "loss": 0.9161,
      "step": 6153
    },
    {
      "epoch": 0.767618810028689,
      "grad_norm": 0.0,
      "learning_rate": 2.7013592730216464e-06,
      "loss": 0.9119,
      "step": 6154
    },
    {
      "epoch": 0.7677435449669452,
      "grad_norm": 0.0,
      "learning_rate": 2.698598077921841e-06,
      "loss": 0.9148,
      "step": 6155
    },
    {
      "epoch": 0.7678682799052015,
      "grad_norm": 0.0,
      "learning_rate": 2.695838074594904e-06,
      "loss": 0.8967,
      "step": 6156
    },
    {
      "epoch": 0.7679930148434576,
      "grad_norm": 0.0,
      "learning_rate": 2.693079263491326e-06,
      "loss": 0.9111,
      "step": 6157
    },
    {
      "epoch": 0.7681177497817139,
      "grad_norm": 0.0,
      "learning_rate": 2.6903216450614187e-06,
      "loss": 0.9449,
      "step": 6158
    },
    {
      "epoch": 0.76824248471997,
      "grad_norm": 0.0,
      "learning_rate": 2.6875652197552993e-06,
      "loss": 0.895,
      "step": 6159
    },
    {
      "epoch": 0.7683672196582263,
      "grad_norm": 0.0,
      "learning_rate": 2.6848099880228774e-06,
      "loss": 0.9567,
      "step": 6160
    },
    {
      "epoch": 0.7684919545964825,
      "grad_norm": 0.0,
      "learning_rate": 2.6820559503138797e-06,
      "loss": 0.8631,
      "step": 6161
    },
    {
      "epoch": 0.7686166895347387,
      "grad_norm": 0.0,
      "learning_rate": 2.6793031070778363e-06,
      "loss": 0.909,
      "step": 6162
    },
    {
      "epoch": 0.7687414244729949,
      "grad_norm": 0.0,
      "learning_rate": 2.6765514587640815e-06,
      "loss": 0.8608,
      "step": 6163
    },
    {
      "epoch": 0.7688661594112511,
      "grad_norm": 0.0,
      "learning_rate": 2.67380100582175e-06,
      "loss": 0.9229,
      "step": 6164
    },
    {
      "epoch": 0.7689908943495073,
      "grad_norm": 0.0,
      "learning_rate": 2.6710517486997856e-06,
      "loss": 0.9197,
      "step": 6165
    },
    {
      "epoch": 0.7691156292877634,
      "grad_norm": 0.0,
      "learning_rate": 2.6683036878469413e-06,
      "loss": 0.8847,
      "step": 6166
    },
    {
      "epoch": 0.7692403642260197,
      "grad_norm": 0.0,
      "learning_rate": 2.6655568237117602e-06,
      "loss": 0.8928,
      "step": 6167
    },
    {
      "epoch": 0.7693650991642759,
      "grad_norm": 0.0,
      "learning_rate": 2.6628111567426107e-06,
      "loss": 0.9262,
      "step": 6168
    },
    {
      "epoch": 0.7694898341025321,
      "grad_norm": 0.0,
      "learning_rate": 2.660066687387648e-06,
      "loss": 0.9048,
      "step": 6169
    },
    {
      "epoch": 0.7696145690407883,
      "grad_norm": 0.0,
      "learning_rate": 2.6573234160948404e-06,
      "loss": 0.8483,
      "step": 6170
    },
    {
      "epoch": 0.7697393039790446,
      "grad_norm": 0.0,
      "learning_rate": 2.6545813433119615e-06,
      "loss": 0.9287,
      "step": 6171
    },
    {
      "epoch": 0.7698640389173007,
      "grad_norm": 0.0,
      "learning_rate": 2.651840469486582e-06,
      "loss": 0.9552,
      "step": 6172
    },
    {
      "epoch": 0.769988773855557,
      "grad_norm": 0.0,
      "learning_rate": 2.649100795066084e-06,
      "loss": 0.9319,
      "step": 6173
    },
    {
      "epoch": 0.7701135087938131,
      "grad_norm": 0.0,
      "learning_rate": 2.646362320497651e-06,
      "loss": 0.916,
      "step": 6174
    },
    {
      "epoch": 0.7702382437320694,
      "grad_norm": 0.0,
      "learning_rate": 2.6436250462282707e-06,
      "loss": 0.9052,
      "step": 6175
    },
    {
      "epoch": 0.7703629786703255,
      "grad_norm": 0.0,
      "learning_rate": 2.6408889727047405e-06,
      "loss": 0.8774,
      "step": 6176
    },
    {
      "epoch": 0.7704877136085818,
      "grad_norm": 0.0,
      "learning_rate": 2.6381541003736486e-06,
      "loss": 0.9288,
      "step": 6177
    },
    {
      "epoch": 0.770612448546838,
      "grad_norm": 0.0,
      "learning_rate": 2.635420429681398e-06,
      "loss": 0.9005,
      "step": 6178
    },
    {
      "epoch": 0.7707371834850941,
      "grad_norm": 0.0,
      "learning_rate": 2.632687961074194e-06,
      "loss": 0.9641,
      "step": 6179
    },
    {
      "epoch": 0.7708619184233504,
      "grad_norm": 0.0,
      "learning_rate": 2.629956694998046e-06,
      "loss": 0.9061,
      "step": 6180
    },
    {
      "epoch": 0.7709866533616065,
      "grad_norm": 0.0,
      "learning_rate": 2.6272266318987606e-06,
      "loss": 0.8961,
      "step": 6181
    },
    {
      "epoch": 0.7711113882998628,
      "grad_norm": 0.0,
      "learning_rate": 2.6244977722219557e-06,
      "loss": 0.9203,
      "step": 6182
    },
    {
      "epoch": 0.771236123238119,
      "grad_norm": 0.0,
      "learning_rate": 2.6217701164130514e-06,
      "loss": 0.8929,
      "step": 6183
    },
    {
      "epoch": 0.7713608581763752,
      "grad_norm": 0.0,
      "learning_rate": 2.6190436649172637e-06,
      "loss": 0.891,
      "step": 6184
    },
    {
      "epoch": 0.7714855931146314,
      "grad_norm": 0.0,
      "learning_rate": 2.6163184181796285e-06,
      "loss": 0.8932,
      "step": 6185
    },
    {
      "epoch": 0.7716103280528877,
      "grad_norm": 0.0,
      "learning_rate": 2.613594376644966e-06,
      "loss": 0.9136,
      "step": 6186
    },
    {
      "epoch": 0.7717350629911438,
      "grad_norm": 0.0,
      "learning_rate": 2.610871540757911e-06,
      "loss": 0.9202,
      "step": 6187
    },
    {
      "epoch": 0.7718597979294001,
      "grad_norm": 0.0,
      "learning_rate": 2.608149910962903e-06,
      "loss": 0.8953,
      "step": 6188
    },
    {
      "epoch": 0.7719845328676562,
      "grad_norm": 0.0,
      "learning_rate": 2.6054294877041732e-06,
      "loss": 0.9181,
      "step": 6189
    },
    {
      "epoch": 0.7721092678059125,
      "grad_norm": 0.0,
      "learning_rate": 2.602710271425767e-06,
      "loss": 0.9215,
      "step": 6190
    },
    {
      "epoch": 0.7722340027441686,
      "grad_norm": 0.0,
      "learning_rate": 2.599992262571529e-06,
      "loss": 0.8957,
      "step": 6191
    },
    {
      "epoch": 0.7723587376824248,
      "grad_norm": 0.0,
      "learning_rate": 2.5972754615851116e-06,
      "loss": 0.9027,
      "step": 6192
    },
    {
      "epoch": 0.7724834726206811,
      "grad_norm": 0.0,
      "learning_rate": 2.594559868909956e-06,
      "loss": 0.8852,
      "step": 6193
    },
    {
      "epoch": 0.7726082075589372,
      "grad_norm": 0.0,
      "learning_rate": 2.5918454849893202e-06,
      "loss": 0.9313,
      "step": 6194
    },
    {
      "epoch": 0.7727329424971935,
      "grad_norm": 0.0,
      "learning_rate": 2.5891323102662593e-06,
      "loss": 0.9337,
      "step": 6195
    },
    {
      "epoch": 0.7728576774354496,
      "grad_norm": 0.0,
      "learning_rate": 2.5864203451836333e-06,
      "loss": 0.9046,
      "step": 6196
    },
    {
      "epoch": 0.7729824123737059,
      "grad_norm": 0.0,
      "learning_rate": 2.583709590184105e-06,
      "loss": 0.9276,
      "step": 6197
    },
    {
      "epoch": 0.7731071473119621,
      "grad_norm": 0.0,
      "learning_rate": 2.5810000457101312e-06,
      "loss": 0.8807,
      "step": 6198
    },
    {
      "epoch": 0.7732318822502183,
      "grad_norm": 0.0,
      "learning_rate": 2.578291712203983e-06,
      "loss": 0.8911,
      "step": 6199
    },
    {
      "epoch": 0.7733566171884745,
      "grad_norm": 0.0,
      "learning_rate": 2.5755845901077305e-06,
      "loss": 0.8253,
      "step": 6200
    },
    {
      "epoch": 0.7734813521267307,
      "grad_norm": 0.0,
      "learning_rate": 2.5728786798632344e-06,
      "loss": 0.9254,
      "step": 6201
    },
    {
      "epoch": 0.7736060870649869,
      "grad_norm": 0.0,
      "learning_rate": 2.5701739819121807e-06,
      "loss": 0.9247,
      "step": 6202
    },
    {
      "epoch": 0.7737308220032431,
      "grad_norm": 0.0,
      "learning_rate": 2.5674704966960384e-06,
      "loss": 0.8779,
      "step": 6203
    },
    {
      "epoch": 0.7738555569414993,
      "grad_norm": 0.0,
      "learning_rate": 2.5647682246560778e-06,
      "loss": 0.8673,
      "step": 6204
    },
    {
      "epoch": 0.7739802918797555,
      "grad_norm": 0.0,
      "learning_rate": 2.5620671662333907e-06,
      "loss": 0.9657,
      "step": 6205
    },
    {
      "epoch": 0.7741050268180117,
      "grad_norm": 0.0,
      "learning_rate": 2.5593673218688474e-06,
      "loss": 0.8751,
      "step": 6206
    },
    {
      "epoch": 0.7742297617562679,
      "grad_norm": 0.0,
      "learning_rate": 2.556668692003135e-06,
      "loss": 0.9229,
      "step": 6207
    },
    {
      "epoch": 0.7743544966945242,
      "grad_norm": 0.0,
      "learning_rate": 2.5539712770767377e-06,
      "loss": 0.934,
      "step": 6208
    },
    {
      "epoch": 0.7744792316327803,
      "grad_norm": 0.0,
      "learning_rate": 2.5512750775299434e-06,
      "loss": 0.9228,
      "step": 6209
    },
    {
      "epoch": 0.7746039665710366,
      "grad_norm": 0.0,
      "learning_rate": 2.548580093802835e-06,
      "loss": 0.9161,
      "step": 6210
    },
    {
      "epoch": 0.7747287015092927,
      "grad_norm": 0.0,
      "learning_rate": 2.545886326335305e-06,
      "loss": 0.8768,
      "step": 6211
    },
    {
      "epoch": 0.774853436447549,
      "grad_norm": 0.0,
      "learning_rate": 2.5431937755670433e-06,
      "loss": 0.9018,
      "step": 6212
    },
    {
      "epoch": 0.7749781713858052,
      "grad_norm": 0.0,
      "learning_rate": 2.540502441937542e-06,
      "loss": 0.9616,
      "step": 6213
    },
    {
      "epoch": 0.7751029063240614,
      "grad_norm": 0.0,
      "learning_rate": 2.5378123258861e-06,
      "loss": 0.9065,
      "step": 6214
    },
    {
      "epoch": 0.7752276412623176,
      "grad_norm": 0.0,
      "learning_rate": 2.5351234278518022e-06,
      "loss": 0.908,
      "step": 6215
    },
    {
      "epoch": 0.7753523762005737,
      "grad_norm": 0.0,
      "learning_rate": 2.5324357482735495e-06,
      "loss": 0.9281,
      "step": 6216
    },
    {
      "epoch": 0.77547711113883,
      "grad_norm": 0.0,
      "learning_rate": 2.529749287590042e-06,
      "loss": 0.9102,
      "step": 6217
    },
    {
      "epoch": 0.7756018460770862,
      "grad_norm": 0.0,
      "learning_rate": 2.52706404623977e-06,
      "loss": 0.9195,
      "step": 6218
    },
    {
      "epoch": 0.7757265810153424,
      "grad_norm": 0.0,
      "learning_rate": 2.5243800246610418e-06,
      "loss": 0.8649,
      "step": 6219
    },
    {
      "epoch": 0.7758513159535986,
      "grad_norm": 0.0,
      "learning_rate": 2.5216972232919533e-06,
      "loss": 0.9044,
      "step": 6220
    },
    {
      "epoch": 0.7759760508918548,
      "grad_norm": 0.0,
      "learning_rate": 2.5190156425703993e-06,
      "loss": 0.9296,
      "step": 6221
    },
    {
      "epoch": 0.776100785830111,
      "grad_norm": 0.0,
      "learning_rate": 2.5163352829340925e-06,
      "loss": 0.8633,
      "step": 6222
    },
    {
      "epoch": 0.7762255207683673,
      "grad_norm": 0.0,
      "learning_rate": 2.5136561448205267e-06,
      "loss": 0.9154,
      "step": 6223
    },
    {
      "epoch": 0.7763502557066234,
      "grad_norm": 0.0,
      "learning_rate": 2.510978228667008e-06,
      "loss": 0.899,
      "step": 6224
    },
    {
      "epoch": 0.7764749906448797,
      "grad_norm": 0.0,
      "learning_rate": 2.50830153491064e-06,
      "loss": 0.9052,
      "step": 6225
    },
    {
      "epoch": 0.7765997255831358,
      "grad_norm": 0.0,
      "learning_rate": 2.5056260639883278e-06,
      "loss": 0.9156,
      "step": 6226
    },
    {
      "epoch": 0.776724460521392,
      "grad_norm": 0.0,
      "learning_rate": 2.5029518163367715e-06,
      "loss": 0.9228,
      "step": 6227
    },
    {
      "epoch": 0.7768491954596483,
      "grad_norm": 0.0,
      "learning_rate": 2.5002787923924797e-06,
      "loss": 0.8699,
      "step": 6228
    },
    {
      "epoch": 0.7769739303979044,
      "grad_norm": 0.0,
      "learning_rate": 2.497606992591758e-06,
      "loss": 0.9029,
      "step": 6229
    },
    {
      "epoch": 0.7770986653361607,
      "grad_norm": 0.0,
      "learning_rate": 2.4949364173707026e-06,
      "loss": 0.872,
      "step": 6230
    },
    {
      "epoch": 0.7772234002744168,
      "grad_norm": 0.0,
      "learning_rate": 2.4922670671652315e-06,
      "loss": 0.8817,
      "step": 6231
    },
    {
      "epoch": 0.7773481352126731,
      "grad_norm": 0.0,
      "learning_rate": 2.4895989424110423e-06,
      "loss": 0.9069,
      "step": 6232
    },
    {
      "epoch": 0.7774728701509293,
      "grad_norm": 0.0,
      "learning_rate": 2.4869320435436407e-06,
      "loss": 0.9023,
      "step": 6233
    },
    {
      "epoch": 0.7775976050891855,
      "grad_norm": 0.0,
      "learning_rate": 2.484266370998336e-06,
      "loss": 0.9081,
      "step": 6234
    },
    {
      "epoch": 0.7777223400274417,
      "grad_norm": 0.0,
      "learning_rate": 2.4816019252102274e-06,
      "loss": 0.9325,
      "step": 6235
    },
    {
      "epoch": 0.7778470749656979,
      "grad_norm": 0.0,
      "learning_rate": 2.4789387066142222e-06,
      "loss": 0.8939,
      "step": 6236
    },
    {
      "epoch": 0.7779718099039541,
      "grad_norm": 0.0,
      "learning_rate": 2.4762767156450285e-06,
      "loss": 0.9042,
      "step": 6237
    },
    {
      "epoch": 0.7780965448422104,
      "grad_norm": 0.0,
      "learning_rate": 2.4736159527371395e-06,
      "loss": 0.8853,
      "step": 6238
    },
    {
      "epoch": 0.7782212797804665,
      "grad_norm": 0.0,
      "learning_rate": 2.4709564183248734e-06,
      "loss": 0.9045,
      "step": 6239
    },
    {
      "epoch": 0.7783460147187227,
      "grad_norm": 0.0,
      "learning_rate": 2.4682981128423223e-06,
      "loss": 0.9094,
      "step": 6240
    },
    {
      "epoch": 0.7784707496569789,
      "grad_norm": 0.0,
      "learning_rate": 2.4656410367233928e-06,
      "loss": 0.8991,
      "step": 6241
    },
    {
      "epoch": 0.7785954845952351,
      "grad_norm": 0.0,
      "learning_rate": 2.4629851904017852e-06,
      "loss": 0.8825,
      "step": 6242
    },
    {
      "epoch": 0.7787202195334914,
      "grad_norm": 0.0,
      "learning_rate": 2.460330574311006e-06,
      "loss": 0.8867,
      "step": 6243
    },
    {
      "epoch": 0.7788449544717475,
      "grad_norm": 0.0,
      "learning_rate": 2.4576771888843478e-06,
      "loss": 0.8724,
      "step": 6244
    },
    {
      "epoch": 0.7789696894100038,
      "grad_norm": 0.0,
      "learning_rate": 2.4550250345549122e-06,
      "loss": 0.8971,
      "step": 6245
    },
    {
      "epoch": 0.7790944243482599,
      "grad_norm": 0.0,
      "learning_rate": 2.4523741117556023e-06,
      "loss": 0.9172,
      "step": 6246
    },
    {
      "epoch": 0.7792191592865162,
      "grad_norm": 0.0,
      "learning_rate": 2.4497244209191062e-06,
      "loss": 0.8729,
      "step": 6247
    },
    {
      "epoch": 0.7793438942247723,
      "grad_norm": 0.0,
      "learning_rate": 2.4470759624779317e-06,
      "loss": 0.8811,
      "step": 6248
    },
    {
      "epoch": 0.7794686291630286,
      "grad_norm": 0.0,
      "learning_rate": 2.4444287368643682e-06,
      "loss": 0.9145,
      "step": 6249
    },
    {
      "epoch": 0.7795933641012848,
      "grad_norm": 0.0,
      "learning_rate": 2.4417827445105035e-06,
      "loss": 0.9145,
      "step": 6250
    },
    {
      "epoch": 0.7797180990395409,
      "grad_norm": 0.0,
      "learning_rate": 2.439137985848242e-06,
      "loss": 0.8923,
      "step": 6251
    },
    {
      "epoch": 0.7798428339777972,
      "grad_norm": 0.0,
      "learning_rate": 2.436494461309268e-06,
      "loss": 0.963,
      "step": 6252
    },
    {
      "epoch": 0.7799675689160533,
      "grad_norm": 0.0,
      "learning_rate": 2.4338521713250717e-06,
      "loss": 0.8685,
      "step": 6253
    },
    {
      "epoch": 0.7800923038543096,
      "grad_norm": 0.0,
      "learning_rate": 2.431211116326946e-06,
      "loss": 0.9113,
      "step": 6254
    },
    {
      "epoch": 0.7802170387925658,
      "grad_norm": 0.0,
      "learning_rate": 2.428571296745971e-06,
      "loss": 0.8842,
      "step": 6255
    },
    {
      "epoch": 0.780341773730822,
      "grad_norm": 0.0,
      "learning_rate": 2.425932713013035e-06,
      "loss": 0.8641,
      "step": 6256
    },
    {
      "epoch": 0.7804665086690782,
      "grad_norm": 0.0,
      "learning_rate": 2.423295365558821e-06,
      "loss": 0.8839,
      "step": 6257
    },
    {
      "epoch": 0.7805912436073345,
      "grad_norm": 0.0,
      "learning_rate": 2.420659254813812e-06,
      "loss": 0.9302,
      "step": 6258
    },
    {
      "epoch": 0.7807159785455906,
      "grad_norm": 0.0,
      "learning_rate": 2.418024381208286e-06,
      "loss": 0.9567,
      "step": 6259
    },
    {
      "epoch": 0.7808407134838469,
      "grad_norm": 0.0,
      "learning_rate": 2.4153907451723246e-06,
      "loss": 0.9469,
      "step": 6260
    },
    {
      "epoch": 0.780965448422103,
      "grad_norm": 0.0,
      "learning_rate": 2.4127583471357974e-06,
      "loss": 0.909,
      "step": 6261
    },
    {
      "epoch": 0.7810901833603593,
      "grad_norm": 0.0,
      "learning_rate": 2.4101271875283818e-06,
      "loss": 0.9432,
      "step": 6262
    },
    {
      "epoch": 0.7812149182986154,
      "grad_norm": 0.0,
      "learning_rate": 2.4074972667795505e-06,
      "loss": 0.9092,
      "step": 6263
    },
    {
      "epoch": 0.7813396532368716,
      "grad_norm": 0.0,
      "learning_rate": 2.4048685853185673e-06,
      "loss": 0.9353,
      "step": 6264
    },
    {
      "epoch": 0.7814643881751279,
      "grad_norm": 0.0,
      "learning_rate": 2.4022411435745076e-06,
      "loss": 0.8972,
      "step": 6265
    },
    {
      "epoch": 0.781589123113384,
      "grad_norm": 0.0,
      "learning_rate": 2.3996149419762315e-06,
      "loss": 0.8765,
      "step": 6266
    },
    {
      "epoch": 0.7817138580516403,
      "grad_norm": 0.0,
      "learning_rate": 2.3969899809523955e-06,
      "loss": 0.8577,
      "step": 6267
    },
    {
      "epoch": 0.7818385929898964,
      "grad_norm": 0.0,
      "learning_rate": 2.3943662609314723e-06,
      "loss": 0.9306,
      "step": 6268
    },
    {
      "epoch": 0.7819633279281527,
      "grad_norm": 0.0,
      "learning_rate": 2.391743782341709e-06,
      "loss": 0.9337,
      "step": 6269
    },
    {
      "epoch": 0.7820880628664089,
      "grad_norm": 0.0,
      "learning_rate": 2.389122545611163e-06,
      "loss": 0.9105,
      "step": 6270
    },
    {
      "epoch": 0.7822127978046651,
      "grad_norm": 0.0,
      "learning_rate": 2.3865025511676896e-06,
      "loss": 0.8821,
      "step": 6271
    },
    {
      "epoch": 0.7823375327429213,
      "grad_norm": 0.0,
      "learning_rate": 2.383883799438933e-06,
      "loss": 0.8425,
      "step": 6272
    },
    {
      "epoch": 0.7824622676811775,
      "grad_norm": 0.0,
      "learning_rate": 2.3812662908523407e-06,
      "loss": 0.8852,
      "step": 6273
    },
    {
      "epoch": 0.7825870026194337,
      "grad_norm": 0.0,
      "learning_rate": 2.3786500258351576e-06,
      "loss": 0.9224,
      "step": 6274
    },
    {
      "epoch": 0.7827117375576899,
      "grad_norm": 0.0,
      "learning_rate": 2.3760350048144253e-06,
      "loss": 0.8413,
      "step": 6275
    },
    {
      "epoch": 0.7828364724959461,
      "grad_norm": 0.0,
      "learning_rate": 2.373421228216974e-06,
      "loss": 0.8523,
      "step": 6276
    },
    {
      "epoch": 0.7829612074342023,
      "grad_norm": 0.0,
      "learning_rate": 2.37080869646945e-06,
      "loss": 0.964,
      "step": 6277
    },
    {
      "epoch": 0.7830859423724585,
      "grad_norm": 0.0,
      "learning_rate": 2.368197409998273e-06,
      "loss": 0.8717,
      "step": 6278
    },
    {
      "epoch": 0.7832106773107147,
      "grad_norm": 0.0,
      "learning_rate": 2.365587369229676e-06,
      "loss": 0.9056,
      "step": 6279
    },
    {
      "epoch": 0.783335412248971,
      "grad_norm": 0.0,
      "learning_rate": 2.362978574589686e-06,
      "loss": 0.8971,
      "step": 6280
    },
    {
      "epoch": 0.7834601471872271,
      "grad_norm": 0.0,
      "learning_rate": 2.360371026504118e-06,
      "loss": 0.9048,
      "step": 6281
    },
    {
      "epoch": 0.7835848821254834,
      "grad_norm": 0.0,
      "learning_rate": 2.357764725398591e-06,
      "loss": 0.9292,
      "step": 6282
    },
    {
      "epoch": 0.7837096170637395,
      "grad_norm": 0.0,
      "learning_rate": 2.355159671698526e-06,
      "loss": 0.9525,
      "step": 6283
    },
    {
      "epoch": 0.7838343520019958,
      "grad_norm": 0.0,
      "learning_rate": 2.3525558658291203e-06,
      "loss": 0.9014,
      "step": 6284
    },
    {
      "epoch": 0.783959086940252,
      "grad_norm": 0.0,
      "learning_rate": 2.3499533082153937e-06,
      "loss": 0.9452,
      "step": 6285
    },
    {
      "epoch": 0.7840838218785082,
      "grad_norm": 0.0,
      "learning_rate": 2.3473519992821427e-06,
      "loss": 0.9125,
      "step": 6286
    },
    {
      "epoch": 0.7842085568167644,
      "grad_norm": 0.0,
      "learning_rate": 2.3447519394539654e-06,
      "loss": 0.9,
      "step": 6287
    },
    {
      "epoch": 0.7843332917550205,
      "grad_norm": 0.0,
      "learning_rate": 2.3421531291552636e-06,
      "loss": 0.8562,
      "step": 6288
    },
    {
      "epoch": 0.7844580266932768,
      "grad_norm": 0.0,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.9,
      "step": 6289
    },
    {
      "epoch": 0.784582761631533,
      "grad_norm": 0.0,
      "learning_rate": 2.336959258842828e-06,
      "loss": 0.9028,
      "step": 6290
    },
    {
      "epoch": 0.7847074965697892,
      "grad_norm": 0.0,
      "learning_rate": 2.3343641996768686e-06,
      "loss": 0.9134,
      "step": 6291
    },
    {
      "epoch": 0.7848322315080454,
      "grad_norm": 0.0,
      "learning_rate": 2.331770391735925e-06,
      "loss": 0.9179,
      "step": 6292
    },
    {
      "epoch": 0.7849569664463016,
      "grad_norm": 0.0,
      "learning_rate": 2.3291778354433614e-06,
      "loss": 0.8765,
      "step": 6293
    },
    {
      "epoch": 0.7850817013845578,
      "grad_norm": 0.0,
      "learning_rate": 2.3265865312223612e-06,
      "loss": 0.8591,
      "step": 6294
    },
    {
      "epoch": 0.7852064363228141,
      "grad_norm": 0.0,
      "learning_rate": 2.3239964794958825e-06,
      "loss": 0.9264,
      "step": 6295
    },
    {
      "epoch": 0.7853311712610702,
      "grad_norm": 0.0,
      "learning_rate": 2.321407680686689e-06,
      "loss": 0.9015,
      "step": 6296
    },
    {
      "epoch": 0.7854559061993265,
      "grad_norm": 0.0,
      "learning_rate": 2.318820135217341e-06,
      "loss": 0.9097,
      "step": 6297
    },
    {
      "epoch": 0.7855806411375826,
      "grad_norm": 0.0,
      "learning_rate": 2.316233843510186e-06,
      "loss": 0.9086,
      "step": 6298
    },
    {
      "epoch": 0.7857053760758388,
      "grad_norm": 0.0,
      "learning_rate": 2.313648805987374e-06,
      "loss": 0.8724,
      "step": 6299
    },
    {
      "epoch": 0.7858301110140951,
      "grad_norm": 0.0,
      "learning_rate": 2.311065023070852e-06,
      "loss": 0.9049,
      "step": 6300
    },
    {
      "epoch": 0.7859548459523512,
      "grad_norm": 0.0,
      "learning_rate": 2.3084824951823515e-06,
      "loss": 0.9201,
      "step": 6301
    },
    {
      "epoch": 0.7860795808906075,
      "grad_norm": 0.0,
      "learning_rate": 2.305901222743411e-06,
      "loss": 0.9356,
      "step": 6302
    },
    {
      "epoch": 0.7862043158288636,
      "grad_norm": 0.0,
      "learning_rate": 2.3033212061753562e-06,
      "loss": 0.875,
      "step": 6303
    },
    {
      "epoch": 0.7863290507671199,
      "grad_norm": 0.0,
      "learning_rate": 2.3007424458993143e-06,
      "loss": 0.8912,
      "step": 6304
    },
    {
      "epoch": 0.786453785705376,
      "grad_norm": 0.0,
      "learning_rate": 2.298164942336205e-06,
      "loss": 0.8785,
      "step": 6305
    },
    {
      "epoch": 0.7865785206436323,
      "grad_norm": 0.0,
      "learning_rate": 2.2955886959067363e-06,
      "loss": 0.9298,
      "step": 6306
    },
    {
      "epoch": 0.7867032555818885,
      "grad_norm": 0.0,
      "learning_rate": 2.2930137070314196e-06,
      "loss": 0.9677,
      "step": 6307
    },
    {
      "epoch": 0.7868279905201447,
      "grad_norm": 0.0,
      "learning_rate": 2.2904399761305573e-06,
      "loss": 0.8513,
      "step": 6308
    },
    {
      "epoch": 0.7869527254584009,
      "grad_norm": 0.0,
      "learning_rate": 2.2878675036242515e-06,
      "loss": 0.8998,
      "step": 6309
    },
    {
      "epoch": 0.7870774603966572,
      "grad_norm": 0.0,
      "learning_rate": 2.285296289932385e-06,
      "loss": 0.8901,
      "step": 6310
    },
    {
      "epoch": 0.7872021953349133,
      "grad_norm": 0.0,
      "learning_rate": 2.2827263354746553e-06,
      "loss": 0.9237,
      "step": 6311
    },
    {
      "epoch": 0.7873269302731695,
      "grad_norm": 0.0,
      "learning_rate": 2.2801576406705404e-06,
      "loss": 0.8631,
      "step": 6312
    },
    {
      "epoch": 0.7874516652114257,
      "grad_norm": 0.0,
      "learning_rate": 2.2775902059393087e-06,
      "loss": 0.9339,
      "step": 6313
    },
    {
      "epoch": 0.7875764001496819,
      "grad_norm": 0.0,
      "learning_rate": 2.2750240317000417e-06,
      "loss": 0.8978,
      "step": 6314
    },
    {
      "epoch": 0.7877011350879382,
      "grad_norm": 0.0,
      "learning_rate": 2.272459118371595e-06,
      "loss": 0.9348,
      "step": 6315
    },
    {
      "epoch": 0.7878258700261943,
      "grad_norm": 0.0,
      "learning_rate": 2.26989546637263e-06,
      "loss": 0.8874,
      "step": 6316
    },
    {
      "epoch": 0.7879506049644506,
      "grad_norm": 0.0,
      "learning_rate": 2.267333076121604e-06,
      "loss": 0.9024,
      "step": 6317
    },
    {
      "epoch": 0.7880753399027067,
      "grad_norm": 0.0,
      "learning_rate": 2.264771948036756e-06,
      "loss": 0.9054,
      "step": 6318
    },
    {
      "epoch": 0.788200074840963,
      "grad_norm": 0.0,
      "learning_rate": 2.262212082536129e-06,
      "loss": 0.9462,
      "step": 6319
    },
    {
      "epoch": 0.7883248097792191,
      "grad_norm": 0.0,
      "learning_rate": 2.2596534800375604e-06,
      "loss": 0.9321,
      "step": 6320
    },
    {
      "epoch": 0.7884495447174754,
      "grad_norm": 0.0,
      "learning_rate": 2.2570961409586756e-06,
      "loss": 0.9075,
      "step": 6321
    },
    {
      "epoch": 0.7885742796557316,
      "grad_norm": 0.0,
      "learning_rate": 2.2545400657169013e-06,
      "loss": 0.9127,
      "step": 6322
    },
    {
      "epoch": 0.7886990145939877,
      "grad_norm": 0.0,
      "learning_rate": 2.2519852547294473e-06,
      "loss": 0.9158,
      "step": 6323
    },
    {
      "epoch": 0.788823749532244,
      "grad_norm": 0.0,
      "learning_rate": 2.2494317084133265e-06,
      "loss": 0.8903,
      "step": 6324
    },
    {
      "epoch": 0.7889484844705001,
      "grad_norm": 0.0,
      "learning_rate": 2.246879427185341e-06,
      "loss": 0.9308,
      "step": 6325
    },
    {
      "epoch": 0.7890732194087564,
      "grad_norm": 0.0,
      "learning_rate": 2.2443284114620913e-06,
      "loss": 0.8902,
      "step": 6326
    },
    {
      "epoch": 0.7891979543470126,
      "grad_norm": 0.0,
      "learning_rate": 2.2417786616599613e-06,
      "loss": 0.8977,
      "step": 6327
    },
    {
      "epoch": 0.7893226892852688,
      "grad_norm": 0.0,
      "learning_rate": 2.2392301781951374e-06,
      "loss": 0.891,
      "step": 6328
    },
    {
      "epoch": 0.789447424223525,
      "grad_norm": 0.0,
      "learning_rate": 2.2366829614835995e-06,
      "loss": 0.8803,
      "step": 6329
    },
    {
      "epoch": 0.7895721591617813,
      "grad_norm": 0.0,
      "learning_rate": 2.234137011941108e-06,
      "loss": 0.885,
      "step": 6330
    },
    {
      "epoch": 0.7896968941000374,
      "grad_norm": 0.0,
      "learning_rate": 2.2315923299832388e-06,
      "loss": 0.9197,
      "step": 6331
    },
    {
      "epoch": 0.7898216290382937,
      "grad_norm": 0.0,
      "learning_rate": 2.2290489160253383e-06,
      "loss": 0.8835,
      "step": 6332
    },
    {
      "epoch": 0.7899463639765498,
      "grad_norm": 0.0,
      "learning_rate": 2.2265067704825593e-06,
      "loss": 0.9336,
      "step": 6333
    },
    {
      "epoch": 0.7900710989148061,
      "grad_norm": 0.0,
      "learning_rate": 2.223965893769847e-06,
      "loss": 0.889,
      "step": 6334
    },
    {
      "epoch": 0.7901958338530622,
      "grad_norm": 0.0,
      "learning_rate": 2.221426286301931e-06,
      "loss": 0.8873,
      "step": 6335
    },
    {
      "epoch": 0.7903205687913184,
      "grad_norm": 0.0,
      "learning_rate": 2.218887948493341e-06,
      "loss": 0.9146,
      "step": 6336
    },
    {
      "epoch": 0.7904453037295747,
      "grad_norm": 0.0,
      "learning_rate": 2.2163508807584e-06,
      "loss": 0.8991,
      "step": 6337
    },
    {
      "epoch": 0.7905700386678308,
      "grad_norm": 0.0,
      "learning_rate": 2.2138150835112217e-06,
      "loss": 0.8985,
      "step": 6338
    },
    {
      "epoch": 0.7906947736060871,
      "grad_norm": 0.0,
      "learning_rate": 2.211280557165708e-06,
      "loss": 0.9184,
      "step": 6339
    },
    {
      "epoch": 0.7908195085443432,
      "grad_norm": 0.0,
      "learning_rate": 2.20874730213556e-06,
      "loss": 0.8619,
      "step": 6340
    },
    {
      "epoch": 0.7909442434825995,
      "grad_norm": 0.0,
      "learning_rate": 2.206215318834268e-06,
      "loss": 0.8868,
      "step": 6341
    },
    {
      "epoch": 0.7910689784208557,
      "grad_norm": 0.0,
      "learning_rate": 2.203684607675116e-06,
      "loss": 0.9263,
      "step": 6342
    },
    {
      "epoch": 0.7911937133591119,
      "grad_norm": 0.0,
      "learning_rate": 2.201155169071184e-06,
      "loss": 0.9139,
      "step": 6343
    },
    {
      "epoch": 0.7913184482973681,
      "grad_norm": 0.0,
      "learning_rate": 2.198627003435332e-06,
      "loss": 0.9152,
      "step": 6344
    },
    {
      "epoch": 0.7914431832356243,
      "grad_norm": 0.0,
      "learning_rate": 2.196100111180225e-06,
      "loss": 0.9046,
      "step": 6345
    },
    {
      "epoch": 0.7915679181738805,
      "grad_norm": 0.0,
      "learning_rate": 2.193574492718319e-06,
      "loss": 0.9297,
      "step": 6346
    },
    {
      "epoch": 0.7916926531121367,
      "grad_norm": 0.0,
      "learning_rate": 2.1910501484618473e-06,
      "loss": 0.9004,
      "step": 6347
    },
    {
      "epoch": 0.7918173880503929,
      "grad_norm": 0.0,
      "learning_rate": 2.188527078822861e-06,
      "loss": 0.9782,
      "step": 6348
    },
    {
      "epoch": 0.7919421229886491,
      "grad_norm": 0.0,
      "learning_rate": 2.1860052842131775e-06,
      "loss": 0.8592,
      "step": 6349
    },
    {
      "epoch": 0.7920668579269053,
      "grad_norm": 0.0,
      "learning_rate": 2.1834847650444223e-06,
      "loss": 0.9235,
      "step": 6350
    },
    {
      "epoch": 0.7921915928651615,
      "grad_norm": 0.0,
      "learning_rate": 2.180965521728009e-06,
      "loss": 0.8872,
      "step": 6351
    },
    {
      "epoch": 0.7923163278034178,
      "grad_norm": 0.0,
      "learning_rate": 2.178447554675136e-06,
      "loss": 0.9169,
      "step": 6352
    },
    {
      "epoch": 0.7924410627416739,
      "grad_norm": 0.0,
      "learning_rate": 2.1759308642968024e-06,
      "loss": 0.8938,
      "step": 6353
    },
    {
      "epoch": 0.7925657976799302,
      "grad_norm": 0.0,
      "learning_rate": 2.1734154510037965e-06,
      "loss": 0.8601,
      "step": 6354
    },
    {
      "epoch": 0.7926905326181863,
      "grad_norm": 0.0,
      "learning_rate": 2.170901315206698e-06,
      "loss": 0.9305,
      "step": 6355
    },
    {
      "epoch": 0.7928152675564426,
      "grad_norm": 0.0,
      "learning_rate": 2.1683884573158732e-06,
      "loss": 0.8768,
      "step": 6356
    },
    {
      "epoch": 0.7929400024946988,
      "grad_norm": 0.0,
      "learning_rate": 2.165876877741485e-06,
      "loss": 0.9486,
      "step": 6357
    },
    {
      "epoch": 0.793064737432955,
      "grad_norm": 0.0,
      "learning_rate": 2.163366576893491e-06,
      "loss": 0.8676,
      "step": 6358
    },
    {
      "epoch": 0.7931894723712112,
      "grad_norm": 0.0,
      "learning_rate": 2.1608575551816257e-06,
      "loss": 0.9228,
      "step": 6359
    },
    {
      "epoch": 0.7933142073094673,
      "grad_norm": 0.0,
      "learning_rate": 2.158349813015438e-06,
      "loss": 0.8627,
      "step": 6360
    },
    {
      "epoch": 0.7934389422477236,
      "grad_norm": 0.0,
      "learning_rate": 2.155843350804243e-06,
      "loss": 0.9327,
      "step": 6361
    },
    {
      "epoch": 0.7935636771859798,
      "grad_norm": 0.0,
      "learning_rate": 2.153338168957165e-06,
      "loss": 0.8848,
      "step": 6362
    },
    {
      "epoch": 0.793688412124236,
      "grad_norm": 0.0,
      "learning_rate": 2.150834267883113e-06,
      "loss": 0.9029,
      "step": 6363
    },
    {
      "epoch": 0.7938131470624922,
      "grad_norm": 0.0,
      "learning_rate": 2.1483316479907824e-06,
      "loss": 0.9516,
      "step": 6364
    },
    {
      "epoch": 0.7939378820007484,
      "grad_norm": 0.0,
      "learning_rate": 2.1458303096886657e-06,
      "loss": 0.9197,
      "step": 6365
    },
    {
      "epoch": 0.7940626169390046,
      "grad_norm": 0.0,
      "learning_rate": 2.1433302533850453e-06,
      "loss": 0.9197,
      "step": 6366
    },
    {
      "epoch": 0.7941873518772609,
      "grad_norm": 0.0,
      "learning_rate": 2.1408314794879925e-06,
      "loss": 0.9042,
      "step": 6367
    },
    {
      "epoch": 0.794312086815517,
      "grad_norm": 0.0,
      "learning_rate": 2.1383339884053743e-06,
      "loss": 0.9341,
      "step": 6368
    },
    {
      "epoch": 0.7944368217537733,
      "grad_norm": 0.0,
      "learning_rate": 2.135837780544838e-06,
      "loss": 0.8793,
      "step": 6369
    },
    {
      "epoch": 0.7945615566920294,
      "grad_norm": 0.0,
      "learning_rate": 2.1333428563138304e-06,
      "loss": 0.8937,
      "step": 6370
    },
    {
      "epoch": 0.7946862916302856,
      "grad_norm": 0.0,
      "learning_rate": 2.1308492161195848e-06,
      "loss": 0.8805,
      "step": 6371
    },
    {
      "epoch": 0.7948110265685419,
      "grad_norm": 0.0,
      "learning_rate": 2.1283568603691306e-06,
      "loss": 0.9824,
      "step": 6372
    },
    {
      "epoch": 0.794935761506798,
      "grad_norm": 0.0,
      "learning_rate": 2.125865789469278e-06,
      "loss": 0.9077,
      "step": 6373
    },
    {
      "epoch": 0.7950604964450543,
      "grad_norm": 0.0,
      "learning_rate": 2.1233760038266336e-06,
      "loss": 0.8802,
      "step": 6374
    },
    {
      "epoch": 0.7951852313833104,
      "grad_norm": 0.0,
      "learning_rate": 2.120887503847596e-06,
      "loss": 0.9124,
      "step": 6375
    },
    {
      "epoch": 0.7953099663215667,
      "grad_norm": 0.0,
      "learning_rate": 2.118400289938345e-06,
      "loss": 0.8788,
      "step": 6376
    },
    {
      "epoch": 0.7954347012598229,
      "grad_norm": 0.0,
      "learning_rate": 2.1159143625048652e-06,
      "loss": 0.9083,
      "step": 6377
    },
    {
      "epoch": 0.7955594361980791,
      "grad_norm": 0.0,
      "learning_rate": 2.1134297219529155e-06,
      "loss": 0.8723,
      "step": 6378
    },
    {
      "epoch": 0.7956841711363353,
      "grad_norm": 0.0,
      "learning_rate": 2.110946368688055e-06,
      "loss": 0.9095,
      "step": 6379
    },
    {
      "epoch": 0.7958089060745915,
      "grad_norm": 0.0,
      "learning_rate": 2.1084643031156306e-06,
      "loss": 0.9159,
      "step": 6380
    },
    {
      "epoch": 0.7959336410128477,
      "grad_norm": 0.0,
      "learning_rate": 2.1059835256407746e-06,
      "loss": 0.923,
      "step": 6381
    },
    {
      "epoch": 0.796058375951104,
      "grad_norm": 0.0,
      "learning_rate": 2.1035040366684145e-06,
      "loss": 0.9278,
      "step": 6382
    },
    {
      "epoch": 0.7961831108893601,
      "grad_norm": 0.0,
      "learning_rate": 2.1010258366032645e-06,
      "loss": 0.8936,
      "step": 6383
    },
    {
      "epoch": 0.7963078458276163,
      "grad_norm": 0.0,
      "learning_rate": 2.0985489258498326e-06,
      "loss": 0.9301,
      "step": 6384
    },
    {
      "epoch": 0.7964325807658725,
      "grad_norm": 0.0,
      "learning_rate": 2.0960733048124082e-06,
      "loss": 0.8908,
      "step": 6385
    },
    {
      "epoch": 0.7965573157041287,
      "grad_norm": 0.0,
      "learning_rate": 2.093598973895078e-06,
      "loss": 0.9054,
      "step": 6386
    },
    {
      "epoch": 0.796682050642385,
      "grad_norm": 0.0,
      "learning_rate": 2.091125933501713e-06,
      "loss": 0.8583,
      "step": 6387
    },
    {
      "epoch": 0.7968067855806411,
      "grad_norm": 0.0,
      "learning_rate": 2.0886541840359776e-06,
      "loss": 0.8937,
      "step": 6388
    },
    {
      "epoch": 0.7969315205188974,
      "grad_norm": 0.0,
      "learning_rate": 2.0861837259013262e-06,
      "loss": 0.9192,
      "step": 6389
    },
    {
      "epoch": 0.7970562554571535,
      "grad_norm": 0.0,
      "learning_rate": 2.0837145595009946e-06,
      "loss": 0.9333,
      "step": 6390
    },
    {
      "epoch": 0.7971809903954098,
      "grad_norm": 0.0,
      "learning_rate": 2.0812466852380153e-06,
      "loss": 0.9235,
      "step": 6391
    },
    {
      "epoch": 0.797305725333666,
      "grad_norm": 0.0,
      "learning_rate": 2.07878010351521e-06,
      "loss": 0.9168,
      "step": 6392
    },
    {
      "epoch": 0.7974304602719222,
      "grad_norm": 0.0,
      "learning_rate": 2.0763148147351785e-06,
      "loss": 0.895,
      "step": 6393
    },
    {
      "epoch": 0.7975551952101784,
      "grad_norm": 0.0,
      "learning_rate": 2.0738508193003305e-06,
      "loss": 0.9073,
      "step": 6394
    },
    {
      "epoch": 0.7976799301484345,
      "grad_norm": 0.0,
      "learning_rate": 2.0713881176128438e-06,
      "loss": 0.8956,
      "step": 6395
    },
    {
      "epoch": 0.7978046650866908,
      "grad_norm": 0.0,
      "learning_rate": 2.0689267100746947e-06,
      "loss": 0.9083,
      "step": 6396
    },
    {
      "epoch": 0.7979294000249469,
      "grad_norm": 0.0,
      "learning_rate": 2.0664665970876496e-06,
      "loss": 0.862,
      "step": 6397
    },
    {
      "epoch": 0.7980541349632032,
      "grad_norm": 0.0,
      "learning_rate": 2.0640077790532576e-06,
      "loss": 0.8689,
      "step": 6398
    },
    {
      "epoch": 0.7981788699014594,
      "grad_norm": 0.0,
      "learning_rate": 2.061550256372861e-06,
      "loss": 0.9156,
      "step": 6399
    },
    {
      "epoch": 0.7983036048397156,
      "grad_norm": 0.0,
      "learning_rate": 2.0590940294475893e-06,
      "loss": 0.8484,
      "step": 6400
    },
    {
      "epoch": 0.7984283397779718,
      "grad_norm": 0.0,
      "learning_rate": 2.0566390986783646e-06,
      "loss": 0.9556,
      "step": 6401
    },
    {
      "epoch": 0.798553074716228,
      "grad_norm": 0.0,
      "learning_rate": 2.0541854644658867e-06,
      "loss": 0.9226,
      "step": 6402
    },
    {
      "epoch": 0.7986778096544842,
      "grad_norm": 0.0,
      "learning_rate": 2.0517331272106543e-06,
      "loss": 0.9098,
      "step": 6403
    },
    {
      "epoch": 0.7988025445927405,
      "grad_norm": 0.0,
      "learning_rate": 2.0492820873129528e-06,
      "loss": 0.9269,
      "step": 6404
    },
    {
      "epoch": 0.7989272795309966,
      "grad_norm": 0.0,
      "learning_rate": 2.0468323451728446e-06,
      "loss": 0.9249,
      "step": 6405
    },
    {
      "epoch": 0.7990520144692529,
      "grad_norm": 0.0,
      "learning_rate": 2.0443839011902023e-06,
      "loss": 0.8758,
      "step": 6406
    },
    {
      "epoch": 0.799176749407509,
      "grad_norm": 0.0,
      "learning_rate": 2.0419367557646653e-06,
      "loss": 0.9091,
      "step": 6407
    },
    {
      "epoch": 0.7993014843457652,
      "grad_norm": 0.0,
      "learning_rate": 2.039490909295669e-06,
      "loss": 0.9286,
      "step": 6408
    },
    {
      "epoch": 0.7994262192840215,
      "grad_norm": 0.0,
      "learning_rate": 2.0370463621824445e-06,
      "loss": 0.8951,
      "step": 6409
    },
    {
      "epoch": 0.7995509542222776,
      "grad_norm": 0.0,
      "learning_rate": 2.0346031148239954e-06,
      "loss": 0.8587,
      "step": 6410
    },
    {
      "epoch": 0.7996756891605339,
      "grad_norm": 0.0,
      "learning_rate": 2.0321611676191244e-06,
      "loss": 0.9462,
      "step": 6411
    },
    {
      "epoch": 0.79980042409879,
      "grad_norm": 0.0,
      "learning_rate": 2.0297205209664194e-06,
      "loss": 0.908,
      "step": 6412
    },
    {
      "epoch": 0.7999251590370463,
      "grad_norm": 0.0,
      "learning_rate": 2.027281175264254e-06,
      "loss": 0.9498,
      "step": 6413
    },
    {
      "epoch": 0.8000498939753025,
      "grad_norm": 0.0,
      "learning_rate": 2.024843130910795e-06,
      "loss": 0.9254,
      "step": 6414
    },
    {
      "epoch": 0.8001746289135587,
      "grad_norm": 0.0,
      "learning_rate": 2.0224063883039868e-06,
      "loss": 0.8991,
      "step": 6415
    },
    {
      "epoch": 0.8002993638518149,
      "grad_norm": 0.0,
      "learning_rate": 2.0199709478415707e-06,
      "loss": 0.8747,
      "step": 6416
    },
    {
      "epoch": 0.8004240987900711,
      "grad_norm": 0.0,
      "learning_rate": 2.01753680992107e-06,
      "loss": 0.9521,
      "step": 6417
    },
    {
      "epoch": 0.8005488337283273,
      "grad_norm": 0.0,
      "learning_rate": 2.0151039749398027e-06,
      "loss": 0.9352,
      "step": 6418
    },
    {
      "epoch": 0.8006735686665835,
      "grad_norm": 0.0,
      "learning_rate": 2.012672443294861e-06,
      "loss": 0.8717,
      "step": 6419
    },
    {
      "epoch": 0.8007983036048397,
      "grad_norm": 0.0,
      "learning_rate": 2.010242215383137e-06,
      "loss": 0.8619,
      "step": 6420
    },
    {
      "epoch": 0.8009230385430959,
      "grad_norm": 0.0,
      "learning_rate": 2.007813291601306e-06,
      "loss": 0.8839,
      "step": 6421
    },
    {
      "epoch": 0.8010477734813521,
      "grad_norm": 0.0,
      "learning_rate": 2.005385672345822e-06,
      "loss": 0.8976,
      "step": 6422
    },
    {
      "epoch": 0.8011725084196083,
      "grad_norm": 0.0,
      "learning_rate": 2.002959358012946e-06,
      "loss": 0.9293,
      "step": 6423
    },
    {
      "epoch": 0.8012972433578646,
      "grad_norm": 0.0,
      "learning_rate": 2.0005343489987038e-06,
      "loss": 0.8669,
      "step": 6424
    },
    {
      "epoch": 0.8014219782961207,
      "grad_norm": 0.0,
      "learning_rate": 1.9981106456989207e-06,
      "loss": 0.9532,
      "step": 6425
    },
    {
      "epoch": 0.801546713234377,
      "grad_norm": 0.0,
      "learning_rate": 1.99568824850921e-06,
      "loss": 0.8662,
      "step": 6426
    },
    {
      "epoch": 0.8016714481726331,
      "grad_norm": 0.0,
      "learning_rate": 1.993267157824962e-06,
      "loss": 0.8719,
      "step": 6427
    },
    {
      "epoch": 0.8017961831108894,
      "grad_norm": 0.0,
      "learning_rate": 1.9908473740413612e-06,
      "loss": 0.8969,
      "step": 6428
    },
    {
      "epoch": 0.8019209180491456,
      "grad_norm": 0.0,
      "learning_rate": 1.9884288975533806e-06,
      "loss": 0.8931,
      "step": 6429
    },
    {
      "epoch": 0.8020456529874018,
      "grad_norm": 0.0,
      "learning_rate": 1.9860117287557756e-06,
      "loss": 0.901,
      "step": 6430
    },
    {
      "epoch": 0.802170387925658,
      "grad_norm": 0.0,
      "learning_rate": 1.9835958680430854e-06,
      "loss": 0.9038,
      "step": 6431
    },
    {
      "epoch": 0.8022951228639141,
      "grad_norm": 0.0,
      "learning_rate": 1.9811813158096427e-06,
      "loss": 0.906,
      "step": 6432
    },
    {
      "epoch": 0.8024198578021704,
      "grad_norm": 0.0,
      "learning_rate": 1.9787680724495617e-06,
      "loss": 0.8885,
      "step": 6433
    },
    {
      "epoch": 0.8025445927404266,
      "grad_norm": 0.0,
      "learning_rate": 1.9763561383567463e-06,
      "loss": 0.9115,
      "step": 6434
    },
    {
      "epoch": 0.8026693276786828,
      "grad_norm": 0.0,
      "learning_rate": 1.9739455139248864e-06,
      "loss": 0.9305,
      "step": 6435
    },
    {
      "epoch": 0.802794062616939,
      "grad_norm": 0.0,
      "learning_rate": 1.971536199547451e-06,
      "loss": 0.8825,
      "step": 6436
    },
    {
      "epoch": 0.8029187975551952,
      "grad_norm": 0.0,
      "learning_rate": 1.9691281956177054e-06,
      "loss": 0.925,
      "step": 6437
    },
    {
      "epoch": 0.8030435324934514,
      "grad_norm": 0.0,
      "learning_rate": 1.966721502528697e-06,
      "loss": 0.9341,
      "step": 6438
    },
    {
      "epoch": 0.8031682674317077,
      "grad_norm": 0.0,
      "learning_rate": 1.964316120673252e-06,
      "loss": 0.8979,
      "step": 6439
    },
    {
      "epoch": 0.8032930023699638,
      "grad_norm": 0.0,
      "learning_rate": 1.9619120504440004e-06,
      "loss": 0.9135,
      "step": 6440
    },
    {
      "epoch": 0.8034177373082201,
      "grad_norm": 0.0,
      "learning_rate": 1.9595092922333414e-06,
      "loss": 0.8712,
      "step": 6441
    },
    {
      "epoch": 0.8035424722464762,
      "grad_norm": 0.0,
      "learning_rate": 1.957107846433459e-06,
      "loss": 0.8463,
      "step": 6442
    },
    {
      "epoch": 0.8036672071847324,
      "grad_norm": 0.0,
      "learning_rate": 1.9547077134363423e-06,
      "loss": 0.8886,
      "step": 6443
    },
    {
      "epoch": 0.8037919421229887,
      "grad_norm": 0.0,
      "learning_rate": 1.9523088936337453e-06,
      "loss": 0.9117,
      "step": 6444
    },
    {
      "epoch": 0.8039166770612448,
      "grad_norm": 0.0,
      "learning_rate": 1.949911387417217e-06,
      "loss": 0.9733,
      "step": 6445
    },
    {
      "epoch": 0.8040414119995011,
      "grad_norm": 0.0,
      "learning_rate": 1.9475151951780956e-06,
      "loss": 0.9316,
      "step": 6446
    },
    {
      "epoch": 0.8041661469377572,
      "grad_norm": 0.0,
      "learning_rate": 1.945120317307493e-06,
      "loss": 0.8964,
      "step": 6447
    },
    {
      "epoch": 0.8042908818760135,
      "grad_norm": 0.0,
      "learning_rate": 1.9427267541963167e-06,
      "loss": 0.9312,
      "step": 6448
    },
    {
      "epoch": 0.8044156168142697,
      "grad_norm": 0.0,
      "learning_rate": 1.9403345062352574e-06,
      "loss": 0.9094,
      "step": 6449
    },
    {
      "epoch": 0.8045403517525259,
      "grad_norm": 0.0,
      "learning_rate": 1.937943573814789e-06,
      "loss": 0.9164,
      "step": 6450
    },
    {
      "epoch": 0.8046650866907821,
      "grad_norm": 0.0,
      "learning_rate": 1.9355539573251737e-06,
      "loss": 0.8984,
      "step": 6451
    },
    {
      "epoch": 0.8047898216290383,
      "grad_norm": 0.0,
      "learning_rate": 1.9331656571564582e-06,
      "loss": 0.9022,
      "step": 6452
    },
    {
      "epoch": 0.8049145565672945,
      "grad_norm": 0.0,
      "learning_rate": 1.930778673698468e-06,
      "loss": 0.945,
      "step": 6453
    },
    {
      "epoch": 0.8050392915055508,
      "grad_norm": 0.0,
      "learning_rate": 1.928393007340823e-06,
      "loss": 0.8845,
      "step": 6454
    },
    {
      "epoch": 0.8051640264438069,
      "grad_norm": 0.0,
      "learning_rate": 1.9260086584729253e-06,
      "loss": 0.8993,
      "step": 6455
    },
    {
      "epoch": 0.8052887613820631,
      "grad_norm": 0.0,
      "learning_rate": 1.923625627483956e-06,
      "loss": 0.891,
      "step": 6456
    },
    {
      "epoch": 0.8054134963203193,
      "grad_norm": 0.0,
      "learning_rate": 1.9212439147628893e-06,
      "loss": 0.9146,
      "step": 6457
    },
    {
      "epoch": 0.8055382312585755,
      "grad_norm": 0.0,
      "learning_rate": 1.9188635206984818e-06,
      "loss": 0.9154,
      "step": 6458
    },
    {
      "epoch": 0.8056629661968318,
      "grad_norm": 0.0,
      "learning_rate": 1.9164844456792675e-06,
      "loss": 0.911,
      "step": 6459
    },
    {
      "epoch": 0.8057877011350879,
      "grad_norm": 0.0,
      "learning_rate": 1.914106690093581e-06,
      "loss": 0.9082,
      "step": 6460
    },
    {
      "epoch": 0.8059124360733442,
      "grad_norm": 0.0,
      "learning_rate": 1.9117302543295235e-06,
      "loss": 0.8648,
      "step": 6461
    },
    {
      "epoch": 0.8060371710116003,
      "grad_norm": 0.0,
      "learning_rate": 1.909355138774994e-06,
      "loss": 0.9046,
      "step": 6462
    },
    {
      "epoch": 0.8061619059498566,
      "grad_norm": 0.0,
      "learning_rate": 1.906981343817671e-06,
      "loss": 0.8904,
      "step": 6463
    },
    {
      "epoch": 0.8062866408881127,
      "grad_norm": 0.0,
      "learning_rate": 1.904608869845015e-06,
      "loss": 0.9428,
      "step": 6464
    },
    {
      "epoch": 0.806411375826369,
      "grad_norm": 0.0,
      "learning_rate": 1.9022377172442753e-06,
      "loss": 0.9265,
      "step": 6465
    },
    {
      "epoch": 0.8065361107646252,
      "grad_norm": 0.0,
      "learning_rate": 1.8998678864024823e-06,
      "loss": 0.907,
      "step": 6466
    },
    {
      "epoch": 0.8066608457028813,
      "grad_norm": 0.0,
      "learning_rate": 1.8974993777064566e-06,
      "loss": 0.9047,
      "step": 6467
    },
    {
      "epoch": 0.8067855806411376,
      "grad_norm": 0.0,
      "learning_rate": 1.89513219154279e-06,
      "loss": 0.8617,
      "step": 6468
    },
    {
      "epoch": 0.8069103155793937,
      "grad_norm": 0.0,
      "learning_rate": 1.8927663282978781e-06,
      "loss": 0.9459,
      "step": 6469
    },
    {
      "epoch": 0.80703505051765,
      "grad_norm": 0.0,
      "learning_rate": 1.8904017883578806e-06,
      "loss": 0.9275,
      "step": 6470
    },
    {
      "epoch": 0.8071597854559062,
      "grad_norm": 0.0,
      "learning_rate": 1.8880385721087534e-06,
      "loss": 0.9517,
      "step": 6471
    },
    {
      "epoch": 0.8072845203941624,
      "grad_norm": 0.0,
      "learning_rate": 1.8856766799362346e-06,
      "loss": 0.9355,
      "step": 6472
    },
    {
      "epoch": 0.8074092553324186,
      "grad_norm": 0.0,
      "learning_rate": 1.8833161122258404e-06,
      "loss": 0.9395,
      "step": 6473
    },
    {
      "epoch": 0.8075339902706749,
      "grad_norm": 0.0,
      "learning_rate": 1.8809568693628765e-06,
      "loss": 0.9226,
      "step": 6474
    },
    {
      "epoch": 0.807658725208931,
      "grad_norm": 0.0,
      "learning_rate": 1.8785989517324355e-06,
      "loss": 0.9188,
      "step": 6475
    },
    {
      "epoch": 0.8077834601471873,
      "grad_norm": 0.0,
      "learning_rate": 1.876242359719378e-06,
      "loss": 0.8872,
      "step": 6476
    },
    {
      "epoch": 0.8079081950854434,
      "grad_norm": 0.0,
      "learning_rate": 1.873887093708372e-06,
      "loss": 0.9373,
      "step": 6477
    },
    {
      "epoch": 0.8080329300236997,
      "grad_norm": 0.0,
      "learning_rate": 1.8715331540838488e-06,
      "loss": 0.9574,
      "step": 6478
    },
    {
      "epoch": 0.8081576649619558,
      "grad_norm": 0.0,
      "learning_rate": 1.8691805412300311e-06,
      "loss": 0.9381,
      "step": 6479
    },
    {
      "epoch": 0.808282399900212,
      "grad_norm": 0.0,
      "learning_rate": 1.8668292555309287e-06,
      "loss": 0.9056,
      "step": 6480
    },
    {
      "epoch": 0.8084071348384683,
      "grad_norm": 0.0,
      "learning_rate": 1.8644792973703252e-06,
      "loss": 0.9025,
      "step": 6481
    },
    {
      "epoch": 0.8085318697767244,
      "grad_norm": 0.0,
      "learning_rate": 1.8621306671317963e-06,
      "loss": 0.8387,
      "step": 6482
    },
    {
      "epoch": 0.8086566047149807,
      "grad_norm": 0.0,
      "learning_rate": 1.8597833651986962e-06,
      "loss": 0.877,
      "step": 6483
    },
    {
      "epoch": 0.8087813396532368,
      "grad_norm": 0.0,
      "learning_rate": 1.8574373919541678e-06,
      "loss": 0.9417,
      "step": 6484
    },
    {
      "epoch": 0.8089060745914931,
      "grad_norm": 0.0,
      "learning_rate": 1.8550927477811254e-06,
      "loss": 0.9219,
      "step": 6485
    },
    {
      "epoch": 0.8090308095297493,
      "grad_norm": 0.0,
      "learning_rate": 1.852749433062282e-06,
      "loss": 0.881,
      "step": 6486
    },
    {
      "epoch": 0.8091555444680055,
      "grad_norm": 0.0,
      "learning_rate": 1.8504074481801237e-06,
      "loss": 0.8944,
      "step": 6487
    },
    {
      "epoch": 0.8092802794062617,
      "grad_norm": 0.0,
      "learning_rate": 1.8480667935169138e-06,
      "loss": 0.8822,
      "step": 6488
    },
    {
      "epoch": 0.809405014344518,
      "grad_norm": 0.0,
      "learning_rate": 1.8457274694547178e-06,
      "loss": 0.8701,
      "step": 6489
    },
    {
      "epoch": 0.8095297492827741,
      "grad_norm": 0.0,
      "learning_rate": 1.8433894763753634e-06,
      "loss": 0.8582,
      "step": 6490
    },
    {
      "epoch": 0.8096544842210303,
      "grad_norm": 0.0,
      "learning_rate": 1.8410528146604746e-06,
      "loss": 0.9473,
      "step": 6491
    },
    {
      "epoch": 0.8097792191592865,
      "grad_norm": 0.0,
      "learning_rate": 1.8387174846914546e-06,
      "loss": 0.9225,
      "step": 6492
    },
    {
      "epoch": 0.8099039540975427,
      "grad_norm": 0.0,
      "learning_rate": 1.8363834868494833e-06,
      "loss": 0.908,
      "step": 6493
    },
    {
      "epoch": 0.8100286890357989,
      "grad_norm": 0.0,
      "learning_rate": 1.8340508215155307e-06,
      "loss": 0.9427,
      "step": 6494
    },
    {
      "epoch": 0.8101534239740551,
      "grad_norm": 0.0,
      "learning_rate": 1.831719489070346e-06,
      "loss": 0.9387,
      "step": 6495
    },
    {
      "epoch": 0.8102781589123114,
      "grad_norm": 0.0,
      "learning_rate": 1.829389489894462e-06,
      "loss": 0.9216,
      "step": 6496
    },
    {
      "epoch": 0.8104028938505675,
      "grad_norm": 0.0,
      "learning_rate": 1.8270608243681953e-06,
      "loss": 0.945,
      "step": 6497
    },
    {
      "epoch": 0.8105276287888238,
      "grad_norm": 0.0,
      "learning_rate": 1.8247334928716386e-06,
      "loss": 0.8993,
      "step": 6498
    },
    {
      "epoch": 0.8106523637270799,
      "grad_norm": 0.0,
      "learning_rate": 1.8224074957846728e-06,
      "loss": 0.935,
      "step": 6499
    },
    {
      "epoch": 0.8107770986653362,
      "grad_norm": 0.0,
      "learning_rate": 1.8200828334869603e-06,
      "loss": 0.941,
      "step": 6500
    },
    {
      "epoch": 0.8109018336035924,
      "grad_norm": 0.0,
      "learning_rate": 1.8177595063579466e-06,
      "loss": 0.9066,
      "step": 6501
    },
    {
      "epoch": 0.8110265685418486,
      "grad_norm": 0.0,
      "learning_rate": 1.8154375147768488e-06,
      "loss": 0.91,
      "step": 6502
    },
    {
      "epoch": 0.8111513034801048,
      "grad_norm": 0.0,
      "learning_rate": 1.8131168591226854e-06,
      "loss": 0.8908,
      "step": 6503
    },
    {
      "epoch": 0.8112760384183609,
      "grad_norm": 0.0,
      "learning_rate": 1.8107975397742406e-06,
      "loss": 0.9649,
      "step": 6504
    },
    {
      "epoch": 0.8114007733566172,
      "grad_norm": 0.0,
      "learning_rate": 1.808479557110081e-06,
      "loss": 0.9223,
      "step": 6505
    },
    {
      "epoch": 0.8115255082948734,
      "grad_norm": 0.0,
      "learning_rate": 1.80616291150857e-06,
      "loss": 0.8908,
      "step": 6506
    },
    {
      "epoch": 0.8116502432331296,
      "grad_norm": 0.0,
      "learning_rate": 1.8038476033478347e-06,
      "loss": 0.892,
      "step": 6507
    },
    {
      "epoch": 0.8117749781713858,
      "grad_norm": 0.0,
      "learning_rate": 1.8015336330057943e-06,
      "loss": 0.9625,
      "step": 6508
    },
    {
      "epoch": 0.811899713109642,
      "grad_norm": 0.0,
      "learning_rate": 1.7992210008601507e-06,
      "loss": 0.9075,
      "step": 6509
    },
    {
      "epoch": 0.8120244480478982,
      "grad_norm": 0.0,
      "learning_rate": 1.7969097072883767e-06,
      "loss": 0.8933,
      "step": 6510
    },
    {
      "epoch": 0.8121491829861545,
      "grad_norm": 0.0,
      "learning_rate": 1.7945997526677371e-06,
      "loss": 0.8724,
      "step": 6511
    },
    {
      "epoch": 0.8122739179244106,
      "grad_norm": 0.0,
      "learning_rate": 1.7922911373752749e-06,
      "loss": 0.9189,
      "step": 6512
    },
    {
      "epoch": 0.8123986528626669,
      "grad_norm": 0.0,
      "learning_rate": 1.7899838617878163e-06,
      "loss": 0.894,
      "step": 6513
    },
    {
      "epoch": 0.812523387800923,
      "grad_norm": 0.0,
      "learning_rate": 1.7876779262819633e-06,
      "loss": 0.8762,
      "step": 6514
    },
    {
      "epoch": 0.8126481227391792,
      "grad_norm": 0.0,
      "learning_rate": 1.7853733312341036e-06,
      "loss": 0.926,
      "step": 6515
    },
    {
      "epoch": 0.8127728576774355,
      "grad_norm": 0.0,
      "learning_rate": 1.7830700770204058e-06,
      "loss": 0.8767,
      "step": 6516
    },
    {
      "epoch": 0.8128975926156916,
      "grad_norm": 0.0,
      "learning_rate": 1.7807681640168194e-06,
      "loss": 0.907,
      "step": 6517
    },
    {
      "epoch": 0.8130223275539479,
      "grad_norm": 0.0,
      "learning_rate": 1.778467592599077e-06,
      "loss": 0.9304,
      "step": 6518
    },
    {
      "epoch": 0.813147062492204,
      "grad_norm": 0.0,
      "learning_rate": 1.7761683631426841e-06,
      "loss": 0.9228,
      "step": 6519
    },
    {
      "epoch": 0.8132717974304603,
      "grad_norm": 0.0,
      "learning_rate": 1.7738704760229352e-06,
      "loss": 0.9448,
      "step": 6520
    },
    {
      "epoch": 0.8133965323687165,
      "grad_norm": 0.0,
      "learning_rate": 1.7715739316149083e-06,
      "loss": 0.8559,
      "step": 6521
    },
    {
      "epoch": 0.8135212673069727,
      "grad_norm": 0.0,
      "learning_rate": 1.7692787302934466e-06,
      "loss": 0.8561,
      "step": 6522
    },
    {
      "epoch": 0.8136460022452289,
      "grad_norm": 0.0,
      "learning_rate": 1.7669848724331984e-06,
      "loss": 0.8639,
      "step": 6523
    },
    {
      "epoch": 0.8137707371834851,
      "grad_norm": 0.0,
      "learning_rate": 1.7646923584085685e-06,
      "loss": 0.856,
      "step": 6524
    },
    {
      "epoch": 0.8138954721217413,
      "grad_norm": 0.0,
      "learning_rate": 1.7624011885937565e-06,
      "loss": 0.9371,
      "step": 6525
    },
    {
      "epoch": 0.8140202070599976,
      "grad_norm": 0.0,
      "learning_rate": 1.760111363362741e-06,
      "loss": 0.8945,
      "step": 6526
    },
    {
      "epoch": 0.8141449419982537,
      "grad_norm": 0.0,
      "learning_rate": 1.7578228830892762e-06,
      "loss": 0.9665,
      "step": 6527
    },
    {
      "epoch": 0.8142696769365099,
      "grad_norm": 0.0,
      "learning_rate": 1.7555357481469003e-06,
      "loss": 0.9126,
      "step": 6528
    },
    {
      "epoch": 0.8143944118747661,
      "grad_norm": 0.0,
      "learning_rate": 1.7532499589089324e-06,
      "loss": 0.9447,
      "step": 6529
    },
    {
      "epoch": 0.8145191468130223,
      "grad_norm": 0.0,
      "learning_rate": 1.7509655157484728e-06,
      "loss": 0.8812,
      "step": 6530
    },
    {
      "epoch": 0.8146438817512786,
      "grad_norm": 0.0,
      "learning_rate": 1.7486824190383966e-06,
      "loss": 0.9351,
      "step": 6531
    },
    {
      "epoch": 0.8147686166895347,
      "grad_norm": 0.0,
      "learning_rate": 1.7464006691513624e-06,
      "loss": 0.9038,
      "step": 6532
    },
    {
      "epoch": 0.814893351627791,
      "grad_norm": 0.0,
      "learning_rate": 1.7441202664598144e-06,
      "loss": 0.9117,
      "step": 6533
    },
    {
      "epoch": 0.8150180865660471,
      "grad_norm": 0.0,
      "learning_rate": 1.7418412113359639e-06,
      "loss": 0.8912,
      "step": 6534
    },
    {
      "epoch": 0.8151428215043034,
      "grad_norm": 0.0,
      "learning_rate": 1.7395635041518188e-06,
      "loss": 0.8758,
      "step": 6535
    },
    {
      "epoch": 0.8152675564425595,
      "grad_norm": 0.0,
      "learning_rate": 1.737287145279153e-06,
      "loss": 0.9183,
      "step": 6536
    },
    {
      "epoch": 0.8153922913808158,
      "grad_norm": 0.0,
      "learning_rate": 1.7350121350895254e-06,
      "loss": 0.9329,
      "step": 6537
    },
    {
      "epoch": 0.815517026319072,
      "grad_norm": 0.0,
      "learning_rate": 1.7327384739542807e-06,
      "loss": 0.9356,
      "step": 6538
    },
    {
      "epoch": 0.8156417612573281,
      "grad_norm": 0.0,
      "learning_rate": 1.7304661622445296e-06,
      "loss": 0.8731,
      "step": 6539
    },
    {
      "epoch": 0.8157664961955844,
      "grad_norm": 0.0,
      "learning_rate": 1.7281952003311753e-06,
      "loss": 0.8812,
      "step": 6540
    },
    {
      "epoch": 0.8158912311338405,
      "grad_norm": 0.0,
      "learning_rate": 1.7259255885848946e-06,
      "loss": 0.8555,
      "step": 6541
    },
    {
      "epoch": 0.8160159660720968,
      "grad_norm": 0.0,
      "learning_rate": 1.723657327376147e-06,
      "loss": 0.93,
      "step": 6542
    },
    {
      "epoch": 0.816140701010353,
      "grad_norm": 0.0,
      "learning_rate": 1.7213904170751706e-06,
      "loss": 0.8464,
      "step": 6543
    },
    {
      "epoch": 0.8162654359486092,
      "grad_norm": 0.0,
      "learning_rate": 1.7191248580519792e-06,
      "loss": 0.85,
      "step": 6544
    },
    {
      "epoch": 0.8163901708868654,
      "grad_norm": 0.0,
      "learning_rate": 1.7168606506763696e-06,
      "loss": 0.8824,
      "step": 6545
    },
    {
      "epoch": 0.8165149058251217,
      "grad_norm": 0.0,
      "learning_rate": 1.7145977953179194e-06,
      "loss": 0.8896,
      "step": 6546
    },
    {
      "epoch": 0.8166396407633778,
      "grad_norm": 0.0,
      "learning_rate": 1.7123362923459853e-06,
      "loss": 0.8715,
      "step": 6547
    },
    {
      "epoch": 0.8167643757016341,
      "grad_norm": 0.0,
      "learning_rate": 1.7100761421296962e-06,
      "loss": 0.9,
      "step": 6548
    },
    {
      "epoch": 0.8168891106398902,
      "grad_norm": 0.0,
      "learning_rate": 1.7078173450379688e-06,
      "loss": 0.9338,
      "step": 6549
    },
    {
      "epoch": 0.8170138455781465,
      "grad_norm": 0.0,
      "learning_rate": 1.7055599014394974e-06,
      "loss": 0.8864,
      "step": 6550
    },
    {
      "epoch": 0.8171385805164026,
      "grad_norm": 0.0,
      "learning_rate": 1.7033038117027466e-06,
      "loss": 0.8806,
      "step": 6551
    },
    {
      "epoch": 0.8172633154546588,
      "grad_norm": 0.0,
      "learning_rate": 1.7010490761959774e-06,
      "loss": 0.8677,
      "step": 6552
    },
    {
      "epoch": 0.8173880503929151,
      "grad_norm": 0.0,
      "learning_rate": 1.698795695287212e-06,
      "loss": 0.856,
      "step": 6553
    },
    {
      "epoch": 0.8175127853311712,
      "grad_norm": 0.0,
      "learning_rate": 1.6965436693442616e-06,
      "loss": 0.9074,
      "step": 6554
    },
    {
      "epoch": 0.8176375202694275,
      "grad_norm": 0.0,
      "learning_rate": 1.6942929987347156e-06,
      "loss": 0.9331,
      "step": 6555
    },
    {
      "epoch": 0.8177622552076836,
      "grad_norm": 0.0,
      "learning_rate": 1.6920436838259357e-06,
      "loss": 0.9754,
      "step": 6556
    },
    {
      "epoch": 0.8178869901459399,
      "grad_norm": 0.0,
      "learning_rate": 1.6897957249850682e-06,
      "loss": 0.9548,
      "step": 6557
    },
    {
      "epoch": 0.8180117250841961,
      "grad_norm": 0.0,
      "learning_rate": 1.6875491225790385e-06,
      "loss": 0.8601,
      "step": 6558
    },
    {
      "epoch": 0.8181364600224523,
      "grad_norm": 0.0,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.8761,
      "step": 6559
    },
    {
      "epoch": 0.8182611949607085,
      "grad_norm": 0.0,
      "learning_rate": 1.6830599885380783e-06,
      "loss": 0.8649,
      "step": 6560
    },
    {
      "epoch": 0.8183859298989647,
      "grad_norm": 0.0,
      "learning_rate": 1.680817457635885e-06,
      "loss": 0.9318,
      "step": 6561
    },
    {
      "epoch": 0.8185106648372209,
      "grad_norm": 0.0,
      "learning_rate": 1.678576284634008e-06,
      "loss": 0.9044,
      "step": 6562
    },
    {
      "epoch": 0.8186353997754771,
      "grad_norm": 0.0,
      "learning_rate": 1.6763364698982632e-06,
      "loss": 0.9205,
      "step": 6563
    },
    {
      "epoch": 0.8187601347137333,
      "grad_norm": 0.0,
      "learning_rate": 1.6740980137942474e-06,
      "loss": 0.8993,
      "step": 6564
    },
    {
      "epoch": 0.8188848696519895,
      "grad_norm": 0.0,
      "learning_rate": 1.671860916687329e-06,
      "loss": 0.9478,
      "step": 6565
    },
    {
      "epoch": 0.8190096045902457,
      "grad_norm": 0.0,
      "learning_rate": 1.66962517894266e-06,
      "loss": 0.9393,
      "step": 6566
    },
    {
      "epoch": 0.8191343395285019,
      "grad_norm": 0.0,
      "learning_rate": 1.6673908009251705e-06,
      "loss": 0.9056,
      "step": 6567
    },
    {
      "epoch": 0.8192590744667582,
      "grad_norm": 0.0,
      "learning_rate": 1.6651577829995625e-06,
      "loss": 0.9448,
      "step": 6568
    },
    {
      "epoch": 0.8193838094050143,
      "grad_norm": 0.0,
      "learning_rate": 1.6629261255303286e-06,
      "loss": 0.8742,
      "step": 6569
    },
    {
      "epoch": 0.8195085443432706,
      "grad_norm": 0.0,
      "learning_rate": 1.6606958288817255e-06,
      "loss": 0.8881,
      "step": 6570
    },
    {
      "epoch": 0.8196332792815267,
      "grad_norm": 0.0,
      "learning_rate": 1.6584668934177962e-06,
      "loss": 0.8992,
      "step": 6571
    },
    {
      "epoch": 0.819758014219783,
      "grad_norm": 0.0,
      "learning_rate": 1.6562393195023608e-06,
      "loss": 0.9399,
      "step": 6572
    },
    {
      "epoch": 0.8198827491580392,
      "grad_norm": 0.0,
      "learning_rate": 1.6540131074990106e-06,
      "loss": 0.8726,
      "step": 6573
    },
    {
      "epoch": 0.8200074840962954,
      "grad_norm": 0.0,
      "learning_rate": 1.651788257771123e-06,
      "loss": 0.9444,
      "step": 6574
    },
    {
      "epoch": 0.8201322190345516,
      "grad_norm": 0.0,
      "learning_rate": 1.6495647706818486e-06,
      "loss": 0.9274,
      "step": 6575
    },
    {
      "epoch": 0.8202569539728077,
      "grad_norm": 0.0,
      "learning_rate": 1.6473426465941189e-06,
      "loss": 0.9487,
      "step": 6576
    },
    {
      "epoch": 0.820381688911064,
      "grad_norm": 0.0,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 0.8645,
      "step": 6577
    },
    {
      "epoch": 0.8205064238493202,
      "grad_norm": 0.0,
      "learning_rate": 1.6429024888738887e-06,
      "loss": 0.8868,
      "step": 6578
    },
    {
      "epoch": 0.8206311587875764,
      "grad_norm": 0.0,
      "learning_rate": 1.640684455966135e-06,
      "loss": 0.8931,
      "step": 6579
    },
    {
      "epoch": 0.8207558937258326,
      "grad_norm": 0.0,
      "learning_rate": 1.6384677875094145e-06,
      "loss": 0.9173,
      "step": 6580
    },
    {
      "epoch": 0.8208806286640888,
      "grad_norm": 0.0,
      "learning_rate": 1.636252483865548e-06,
      "loss": 0.9416,
      "step": 6581
    },
    {
      "epoch": 0.821005363602345,
      "grad_norm": 0.0,
      "learning_rate": 1.6340385453961216e-06,
      "loss": 0.893,
      "step": 6582
    },
    {
      "epoch": 0.8211300985406013,
      "grad_norm": 0.0,
      "learning_rate": 1.6318259724625096e-06,
      "loss": 0.9111,
      "step": 6583
    },
    {
      "epoch": 0.8212548334788574,
      "grad_norm": 0.0,
      "learning_rate": 1.629614765425861e-06,
      "loss": 0.9261,
      "step": 6584
    },
    {
      "epoch": 0.8213795684171137,
      "grad_norm": 0.0,
      "learning_rate": 1.6274049246470958e-06,
      "loss": 0.8913,
      "step": 6585
    },
    {
      "epoch": 0.8215043033553698,
      "grad_norm": 0.0,
      "learning_rate": 1.6251964504869221e-06,
      "loss": 0.9258,
      "step": 6586
    },
    {
      "epoch": 0.821629038293626,
      "grad_norm": 0.0,
      "learning_rate": 1.622989343305813e-06,
      "loss": 0.9209,
      "step": 6587
    },
    {
      "epoch": 0.8217537732318823,
      "grad_norm": 0.0,
      "learning_rate": 1.6207836034640268e-06,
      "loss": 0.9005,
      "step": 6588
    },
    {
      "epoch": 0.8218785081701384,
      "grad_norm": 0.0,
      "learning_rate": 1.6185792313215977e-06,
      "loss": 0.9078,
      "step": 6589
    },
    {
      "epoch": 0.8220032431083947,
      "grad_norm": 0.0,
      "learning_rate": 1.6163762272383298e-06,
      "loss": 0.8895,
      "step": 6590
    },
    {
      "epoch": 0.8221279780466508,
      "grad_norm": 0.0,
      "learning_rate": 1.6141745915738127e-06,
      "loss": 0.8694,
      "step": 6591
    },
    {
      "epoch": 0.8222527129849071,
      "grad_norm": 0.0,
      "learning_rate": 1.6119743246874064e-06,
      "loss": 0.9034,
      "step": 6592
    },
    {
      "epoch": 0.8223774479231633,
      "grad_norm": 0.0,
      "learning_rate": 1.6097754269382538e-06,
      "loss": 0.9086,
      "step": 6593
    },
    {
      "epoch": 0.8225021828614195,
      "grad_norm": 0.0,
      "learning_rate": 1.6075778986852663e-06,
      "loss": 0.9251,
      "step": 6594
    },
    {
      "epoch": 0.8226269177996757,
      "grad_norm": 0.0,
      "learning_rate": 1.6053817402871363e-06,
      "loss": 0.904,
      "step": 6595
    },
    {
      "epoch": 0.8227516527379319,
      "grad_norm": 0.0,
      "learning_rate": 1.6031869521023369e-06,
      "loss": 0.8865,
      "step": 6596
    },
    {
      "epoch": 0.8228763876761881,
      "grad_norm": 0.0,
      "learning_rate": 1.6009935344891036e-06,
      "loss": 0.8542,
      "step": 6597
    },
    {
      "epoch": 0.8230011226144444,
      "grad_norm": 0.0,
      "learning_rate": 1.598801487805469e-06,
      "loss": 0.8959,
      "step": 6598
    },
    {
      "epoch": 0.8231258575527005,
      "grad_norm": 0.0,
      "learning_rate": 1.5966108124092217e-06,
      "loss": 0.8864,
      "step": 6599
    },
    {
      "epoch": 0.8232505924909567,
      "grad_norm": 0.0,
      "learning_rate": 1.5944215086579384e-06,
      "loss": 0.9262,
      "step": 6600
    },
    {
      "epoch": 0.8233753274292129,
      "grad_norm": 0.0,
      "learning_rate": 1.5922335769089703e-06,
      "loss": 0.8385,
      "step": 6601
    },
    {
      "epoch": 0.8235000623674691,
      "grad_norm": 0.0,
      "learning_rate": 1.5900470175194383e-06,
      "loss": 0.899,
      "step": 6602
    },
    {
      "epoch": 0.8236247973057254,
      "grad_norm": 0.0,
      "learning_rate": 1.5878618308462479e-06,
      "loss": 0.9328,
      "step": 6603
    },
    {
      "epoch": 0.8237495322439815,
      "grad_norm": 0.0,
      "learning_rate": 1.585678017246075e-06,
      "loss": 0.8837,
      "step": 6604
    },
    {
      "epoch": 0.8238742671822378,
      "grad_norm": 0.0,
      "learning_rate": 1.5834955770753723e-06,
      "loss": 0.9123,
      "step": 6605
    },
    {
      "epoch": 0.8239990021204939,
      "grad_norm": 0.0,
      "learning_rate": 1.5813145106903737e-06,
      "loss": 0.9178,
      "step": 6606
    },
    {
      "epoch": 0.8241237370587502,
      "grad_norm": 0.0,
      "learning_rate": 1.579134818447079e-06,
      "loss": 0.9158,
      "step": 6607
    },
    {
      "epoch": 0.8242484719970063,
      "grad_norm": 0.0,
      "learning_rate": 1.5769565007012699e-06,
      "loss": 0.9061,
      "step": 6608
    },
    {
      "epoch": 0.8243732069352626,
      "grad_norm": 0.0,
      "learning_rate": 1.5747795578085046e-06,
      "loss": 0.9077,
      "step": 6609
    },
    {
      "epoch": 0.8244979418735188,
      "grad_norm": 0.0,
      "learning_rate": 1.5726039901241152e-06,
      "loss": 0.8893,
      "step": 6610
    },
    {
      "epoch": 0.8246226768117749,
      "grad_norm": 0.0,
      "learning_rate": 1.5704297980032068e-06,
      "loss": 0.9387,
      "step": 6611
    },
    {
      "epoch": 0.8247474117500312,
      "grad_norm": 0.0,
      "learning_rate": 1.5682569818006632e-06,
      "loss": 0.8946,
      "step": 6612
    },
    {
      "epoch": 0.8248721466882873,
      "grad_norm": 0.0,
      "learning_rate": 1.566085541871145e-06,
      "loss": 0.899,
      "step": 6613
    },
    {
      "epoch": 0.8249968816265436,
      "grad_norm": 0.0,
      "learning_rate": 1.5639154785690801e-06,
      "loss": 0.9553,
      "step": 6614
    },
    {
      "epoch": 0.8251216165647998,
      "grad_norm": 0.0,
      "learning_rate": 1.5617467922486863e-06,
      "loss": 0.8938,
      "step": 6615
    },
    {
      "epoch": 0.825246351503056,
      "grad_norm": 0.0,
      "learning_rate": 1.5595794832639421e-06,
      "loss": 0.8875,
      "step": 6616
    },
    {
      "epoch": 0.8253710864413122,
      "grad_norm": 0.0,
      "learning_rate": 1.557413551968604e-06,
      "loss": 0.9238,
      "step": 6617
    },
    {
      "epoch": 0.8254958213795685,
      "grad_norm": 0.0,
      "learning_rate": 1.5552489987162155e-06,
      "loss": 0.8887,
      "step": 6618
    },
    {
      "epoch": 0.8256205563178246,
      "grad_norm": 0.0,
      "learning_rate": 1.5530858238600789e-06,
      "loss": 0.9367,
      "step": 6619
    },
    {
      "epoch": 0.8257452912560809,
      "grad_norm": 0.0,
      "learning_rate": 1.5509240277532821e-06,
      "loss": 0.8723,
      "step": 6620
    },
    {
      "epoch": 0.825870026194337,
      "grad_norm": 0.0,
      "learning_rate": 1.5487636107486825e-06,
      "loss": 0.931,
      "step": 6621
    },
    {
      "epoch": 0.8259947611325933,
      "grad_norm": 0.0,
      "learning_rate": 1.5466045731989199e-06,
      "loss": 0.8833,
      "step": 6622
    },
    {
      "epoch": 0.8261194960708494,
      "grad_norm": 0.0,
      "learning_rate": 1.5444469154563968e-06,
      "loss": 0.8777,
      "step": 6623
    },
    {
      "epoch": 0.8262442310091056,
      "grad_norm": 0.0,
      "learning_rate": 1.5422906378733006e-06,
      "loss": 0.99,
      "step": 6624
    },
    {
      "epoch": 0.8263689659473619,
      "grad_norm": 0.0,
      "learning_rate": 1.5401357408015893e-06,
      "loss": 0.9183,
      "step": 6625
    },
    {
      "epoch": 0.826493700885618,
      "grad_norm": 0.0,
      "learning_rate": 1.5379822245929976e-06,
      "loss": 0.8479,
      "step": 6626
    },
    {
      "epoch": 0.8266184358238743,
      "grad_norm": 0.0,
      "learning_rate": 1.5358300895990352e-06,
      "loss": 0.9229,
      "step": 6627
    },
    {
      "epoch": 0.8267431707621304,
      "grad_norm": 0.0,
      "learning_rate": 1.53367933617098e-06,
      "loss": 0.8665,
      "step": 6628
    },
    {
      "epoch": 0.8268679057003867,
      "grad_norm": 0.0,
      "learning_rate": 1.5315299646598924e-06,
      "loss": 0.9153,
      "step": 6629
    },
    {
      "epoch": 0.8269926406386429,
      "grad_norm": 0.0,
      "learning_rate": 1.5293819754166062e-06,
      "loss": 0.9358,
      "step": 6630
    },
    {
      "epoch": 0.8271173755768991,
      "grad_norm": 0.0,
      "learning_rate": 1.5272353687917197e-06,
      "loss": 0.9014,
      "step": 6631
    },
    {
      "epoch": 0.8272421105151553,
      "grad_norm": 0.0,
      "learning_rate": 1.5250901451356226e-06,
      "loss": 0.9012,
      "step": 6632
    },
    {
      "epoch": 0.8273668454534115,
      "grad_norm": 0.0,
      "learning_rate": 1.522946304798465e-06,
      "loss": 0.8695,
      "step": 6633
    },
    {
      "epoch": 0.8274915803916677,
      "grad_norm": 0.0,
      "learning_rate": 1.5208038481301713e-06,
      "loss": 0.9328,
      "step": 6634
    },
    {
      "epoch": 0.8276163153299239,
      "grad_norm": 0.0,
      "learning_rate": 1.518662775480454e-06,
      "loss": 0.9222,
      "step": 6635
    },
    {
      "epoch": 0.8277410502681801,
      "grad_norm": 0.0,
      "learning_rate": 1.5165230871987823e-06,
      "loss": 0.9225,
      "step": 6636
    },
    {
      "epoch": 0.8278657852064363,
      "grad_norm": 0.0,
      "learning_rate": 1.5143847836344105e-06,
      "loss": 0.893,
      "step": 6637
    },
    {
      "epoch": 0.8279905201446925,
      "grad_norm": 0.0,
      "learning_rate": 1.5122478651363626e-06,
      "loss": 0.9188,
      "step": 6638
    },
    {
      "epoch": 0.8281152550829487,
      "grad_norm": 0.0,
      "learning_rate": 1.5101123320534406e-06,
      "loss": 0.9218,
      "step": 6639
    },
    {
      "epoch": 0.828239990021205,
      "grad_norm": 0.0,
      "learning_rate": 1.5079781847342122e-06,
      "loss": 0.8453,
      "step": 6640
    },
    {
      "epoch": 0.8283647249594611,
      "grad_norm": 0.0,
      "learning_rate": 1.505845423527027e-06,
      "loss": 0.8703,
      "step": 6641
    },
    {
      "epoch": 0.8284894598977174,
      "grad_norm": 0.0,
      "learning_rate": 1.5037140487800073e-06,
      "loss": 0.9197,
      "step": 6642
    },
    {
      "epoch": 0.8286141948359735,
      "grad_norm": 0.0,
      "learning_rate": 1.5015840608410403e-06,
      "loss": 0.9075,
      "step": 6643
    },
    {
      "epoch": 0.8287389297742298,
      "grad_norm": 0.0,
      "learning_rate": 1.4994554600578036e-06,
      "loss": 0.9277,
      "step": 6644
    },
    {
      "epoch": 0.828863664712486,
      "grad_norm": 0.0,
      "learning_rate": 1.4973282467777306e-06,
      "loss": 0.8771,
      "step": 6645
    },
    {
      "epoch": 0.8289883996507422,
      "grad_norm": 0.0,
      "learning_rate": 1.4952024213480387e-06,
      "loss": 0.8954,
      "step": 6646
    },
    {
      "epoch": 0.8291131345889984,
      "grad_norm": 0.0,
      "learning_rate": 1.4930779841157184e-06,
      "loss": 0.8889,
      "step": 6647
    },
    {
      "epoch": 0.8292378695272545,
      "grad_norm": 0.0,
      "learning_rate": 1.4909549354275277e-06,
      "loss": 0.8935,
      "step": 6648
    },
    {
      "epoch": 0.8293626044655108,
      "grad_norm": 0.0,
      "learning_rate": 1.4888332756300027e-06,
      "loss": 0.9418,
      "step": 6649
    },
    {
      "epoch": 0.829487339403767,
      "grad_norm": 0.0,
      "learning_rate": 1.4867130050694556e-06,
      "loss": 0.867,
      "step": 6650
    },
    {
      "epoch": 0.8296120743420232,
      "grad_norm": 0.0,
      "learning_rate": 1.4845941240919603e-06,
      "loss": 0.8927,
      "step": 6651
    },
    {
      "epoch": 0.8297368092802794,
      "grad_norm": 0.0,
      "learning_rate": 1.4824766330433793e-06,
      "loss": 0.9163,
      "step": 6652
    },
    {
      "epoch": 0.8298615442185356,
      "grad_norm": 0.0,
      "learning_rate": 1.480360532269336e-06,
      "loss": 0.9276,
      "step": 6653
    },
    {
      "epoch": 0.8299862791567918,
      "grad_norm": 0.0,
      "learning_rate": 1.478245822115233e-06,
      "loss": 0.9527,
      "step": 6654
    },
    {
      "epoch": 0.8301110140950481,
      "grad_norm": 0.0,
      "learning_rate": 1.4761325029262442e-06,
      "loss": 0.8961,
      "step": 6655
    },
    {
      "epoch": 0.8302357490333042,
      "grad_norm": 0.0,
      "learning_rate": 1.474020575047318e-06,
      "loss": 0.9274,
      "step": 6656
    },
    {
      "epoch": 0.8303604839715605,
      "grad_norm": 0.0,
      "learning_rate": 1.47191003882317e-06,
      "loss": 0.9028,
      "step": 6657
    },
    {
      "epoch": 0.8304852189098166,
      "grad_norm": 0.0,
      "learning_rate": 1.4698008945982966e-06,
      "loss": 0.9205,
      "step": 6658
    },
    {
      "epoch": 0.8306099538480728,
      "grad_norm": 0.0,
      "learning_rate": 1.4676931427169639e-06,
      "loss": 0.9456,
      "step": 6659
    },
    {
      "epoch": 0.8307346887863291,
      "grad_norm": 0.0,
      "learning_rate": 1.465586783523204e-06,
      "loss": 0.9437,
      "step": 6660
    },
    {
      "epoch": 0.8308594237245852,
      "grad_norm": 0.0,
      "learning_rate": 1.4634818173608378e-06,
      "loss": 0.9067,
      "step": 6661
    },
    {
      "epoch": 0.8309841586628415,
      "grad_norm": 0.0,
      "learning_rate": 1.4613782445734404e-06,
      "loss": 0.8968,
      "step": 6662
    },
    {
      "epoch": 0.8311088936010976,
      "grad_norm": 0.0,
      "learning_rate": 1.4592760655043713e-06,
      "loss": 0.9218,
      "step": 6663
    },
    {
      "epoch": 0.8312336285393539,
      "grad_norm": 0.0,
      "learning_rate": 1.457175280496761e-06,
      "loss": 0.9335,
      "step": 6664
    },
    {
      "epoch": 0.83135836347761,
      "grad_norm": 0.0,
      "learning_rate": 1.4550758898935047e-06,
      "loss": 0.9082,
      "step": 6665
    },
    {
      "epoch": 0.8314830984158663,
      "grad_norm": 0.0,
      "learning_rate": 1.4529778940372806e-06,
      "loss": 0.9298,
      "step": 6666
    },
    {
      "epoch": 0.8316078333541225,
      "grad_norm": 0.0,
      "learning_rate": 1.4508812932705364e-06,
      "loss": 0.8999,
      "step": 6667
    },
    {
      "epoch": 0.8317325682923787,
      "grad_norm": 0.0,
      "learning_rate": 1.4487860879354842e-06,
      "loss": 0.8592,
      "step": 6668
    },
    {
      "epoch": 0.8318573032306349,
      "grad_norm": 0.0,
      "learning_rate": 1.4466922783741167e-06,
      "loss": 0.9199,
      "step": 6669
    },
    {
      "epoch": 0.8319820381688912,
      "grad_norm": 0.0,
      "learning_rate": 1.4445998649281978e-06,
      "loss": 0.8868,
      "step": 6670
    },
    {
      "epoch": 0.8321067731071473,
      "grad_norm": 0.0,
      "learning_rate": 1.4425088479392612e-06,
      "loss": 0.8762,
      "step": 6671
    },
    {
      "epoch": 0.8322315080454035,
      "grad_norm": 0.0,
      "learning_rate": 1.440419227748614e-06,
      "loss": 0.8749,
      "step": 6672
    },
    {
      "epoch": 0.8323562429836597,
      "grad_norm": 0.0,
      "learning_rate": 1.4383310046973365e-06,
      "loss": 0.9303,
      "step": 6673
    },
    {
      "epoch": 0.8324809779219159,
      "grad_norm": 0.0,
      "learning_rate": 1.4362441791262748e-06,
      "loss": 0.9118,
      "step": 6674
    },
    {
      "epoch": 0.8326057128601722,
      "grad_norm": 0.0,
      "learning_rate": 1.4341587513760546e-06,
      "loss": 0.9268,
      "step": 6675
    },
    {
      "epoch": 0.8327304477984283,
      "grad_norm": 0.0,
      "learning_rate": 1.4320747217870722e-06,
      "loss": 0.9025,
      "step": 6676
    },
    {
      "epoch": 0.8328551827366846,
      "grad_norm": 0.0,
      "learning_rate": 1.4299920906994868e-06,
      "loss": 0.8946,
      "step": 6677
    },
    {
      "epoch": 0.8329799176749407,
      "grad_norm": 0.0,
      "learning_rate": 1.4279108584532453e-06,
      "loss": 0.8903,
      "step": 6678
    },
    {
      "epoch": 0.833104652613197,
      "grad_norm": 0.0,
      "learning_rate": 1.4258310253880537e-06,
      "loss": 0.9156,
      "step": 6679
    },
    {
      "epoch": 0.8332293875514531,
      "grad_norm": 0.0,
      "learning_rate": 1.423752591843387e-06,
      "loss": 0.8789,
      "step": 6680
    },
    {
      "epoch": 0.8333541224897094,
      "grad_norm": 0.0,
      "learning_rate": 1.4216755581585073e-06,
      "loss": 0.897,
      "step": 6681
    },
    {
      "epoch": 0.8334788574279656,
      "grad_norm": 0.0,
      "learning_rate": 1.4195999246724335e-06,
      "loss": 0.8808,
      "step": 6682
    },
    {
      "epoch": 0.8336035923662217,
      "grad_norm": 0.0,
      "learning_rate": 1.4175256917239621e-06,
      "loss": 0.9183,
      "step": 6683
    },
    {
      "epoch": 0.833728327304478,
      "grad_norm": 0.0,
      "learning_rate": 1.4154528596516625e-06,
      "loss": 0.8744,
      "step": 6684
    },
    {
      "epoch": 0.8338530622427341,
      "grad_norm": 0.0,
      "learning_rate": 1.4133814287938707e-06,
      "loss": 0.8933,
      "step": 6685
    },
    {
      "epoch": 0.8339777971809904,
      "grad_norm": 0.0,
      "learning_rate": 1.4113113994886962e-06,
      "loss": 0.8338,
      "step": 6686
    },
    {
      "epoch": 0.8341025321192466,
      "grad_norm": 0.0,
      "learning_rate": 1.4092427720740209e-06,
      "loss": 0.8709,
      "step": 6687
    },
    {
      "epoch": 0.8342272670575028,
      "grad_norm": 0.0,
      "learning_rate": 1.407175546887497e-06,
      "loss": 0.9199,
      "step": 6688
    },
    {
      "epoch": 0.834352001995759,
      "grad_norm": 0.0,
      "learning_rate": 1.40510972426655e-06,
      "loss": 0.9417,
      "step": 6689
    },
    {
      "epoch": 0.8344767369340153,
      "grad_norm": 0.0,
      "learning_rate": 1.4030453045483693e-06,
      "loss": 0.9061,
      "step": 6690
    },
    {
      "epoch": 0.8346014718722714,
      "grad_norm": 0.0,
      "learning_rate": 1.4009822880699231e-06,
      "loss": 0.8893,
      "step": 6691
    },
    {
      "epoch": 0.8347262068105277,
      "grad_norm": 0.0,
      "learning_rate": 1.3989206751679484e-06,
      "loss": 0.9022,
      "step": 6692
    },
    {
      "epoch": 0.8348509417487838,
      "grad_norm": 0.0,
      "learning_rate": 1.3968604661789532e-06,
      "loss": 0.9107,
      "step": 6693
    },
    {
      "epoch": 0.8349756766870401,
      "grad_norm": 0.0,
      "learning_rate": 1.3948016614392113e-06,
      "loss": 0.883,
      "step": 6694
    },
    {
      "epoch": 0.8351004116252962,
      "grad_norm": 0.0,
      "learning_rate": 1.3927442612847741e-06,
      "loss": 0.8859,
      "step": 6695
    },
    {
      "epoch": 0.8352251465635524,
      "grad_norm": 0.0,
      "learning_rate": 1.390688266051463e-06,
      "loss": 0.862,
      "step": 6696
    },
    {
      "epoch": 0.8353498815018087,
      "grad_norm": 0.0,
      "learning_rate": 1.388633676074862e-06,
      "loss": 0.9394,
      "step": 6697
    },
    {
      "epoch": 0.8354746164400648,
      "grad_norm": 0.0,
      "learning_rate": 1.3865804916903414e-06,
      "loss": 0.9308,
      "step": 6698
    },
    {
      "epoch": 0.8355993513783211,
      "grad_norm": 0.0,
      "learning_rate": 1.3845287132330254e-06,
      "loss": 0.8698,
      "step": 6699
    },
    {
      "epoch": 0.8357240863165772,
      "grad_norm": 0.0,
      "learning_rate": 1.3824783410378183e-06,
      "loss": 0.8889,
      "step": 6700
    },
    {
      "epoch": 0.8358488212548335,
      "grad_norm": 0.0,
      "learning_rate": 1.3804293754393948e-06,
      "loss": 0.8907,
      "step": 6701
    },
    {
      "epoch": 0.8359735561930897,
      "grad_norm": 0.0,
      "learning_rate": 1.3783818167721918e-06,
      "loss": 0.9426,
      "step": 6702
    },
    {
      "epoch": 0.8360982911313459,
      "grad_norm": 0.0,
      "learning_rate": 1.3763356653704274e-06,
      "loss": 0.9249,
      "step": 6703
    },
    {
      "epoch": 0.8362230260696021,
      "grad_norm": 0.0,
      "learning_rate": 1.3742909215680833e-06,
      "loss": 0.9236,
      "step": 6704
    },
    {
      "epoch": 0.8363477610078583,
      "grad_norm": 0.0,
      "learning_rate": 1.372247585698916e-06,
      "loss": 0.8859,
      "step": 6705
    },
    {
      "epoch": 0.8364724959461145,
      "grad_norm": 0.0,
      "learning_rate": 1.370205658096445e-06,
      "loss": 0.8651,
      "step": 6706
    },
    {
      "epoch": 0.8365972308843707,
      "grad_norm": 0.0,
      "learning_rate": 1.3681651390939666e-06,
      "loss": 0.9227,
      "step": 6707
    },
    {
      "epoch": 0.8367219658226269,
      "grad_norm": 0.0,
      "learning_rate": 1.366126029024545e-06,
      "loss": 0.9209,
      "step": 6708
    },
    {
      "epoch": 0.8368467007608831,
      "grad_norm": 0.0,
      "learning_rate": 1.3640883282210127e-06,
      "loss": 0.904,
      "step": 6709
    },
    {
      "epoch": 0.8369714356991393,
      "grad_norm": 0.0,
      "learning_rate": 1.3620520370159795e-06,
      "loss": 0.9049,
      "step": 6710
    },
    {
      "epoch": 0.8370961706373955,
      "grad_norm": 0.0,
      "learning_rate": 1.3600171557418117e-06,
      "loss": 0.8741,
      "step": 6711
    },
    {
      "epoch": 0.8372209055756518,
      "grad_norm": 0.0,
      "learning_rate": 1.357983684730657e-06,
      "loss": 0.8992,
      "step": 6712
    },
    {
      "epoch": 0.8373456405139079,
      "grad_norm": 0.0,
      "learning_rate": 1.3559516243144323e-06,
      "loss": 0.901,
      "step": 6713
    },
    {
      "epoch": 0.8374703754521642,
      "grad_norm": 0.0,
      "learning_rate": 1.3539209748248116e-06,
      "loss": 0.9285,
      "step": 6714
    },
    {
      "epoch": 0.8375951103904203,
      "grad_norm": 0.0,
      "learning_rate": 1.3518917365932594e-06,
      "loss": 0.8731,
      "step": 6715
    },
    {
      "epoch": 0.8377198453286766,
      "grad_norm": 0.0,
      "learning_rate": 1.3498639099509915e-06,
      "loss": 0.9369,
      "step": 6716
    },
    {
      "epoch": 0.8378445802669328,
      "grad_norm": 0.0,
      "learning_rate": 1.347837495229002e-06,
      "loss": 0.8943,
      "step": 6717
    },
    {
      "epoch": 0.837969315205189,
      "grad_norm": 0.0,
      "learning_rate": 1.3458124927580552e-06,
      "loss": 0.8758,
      "step": 6718
    },
    {
      "epoch": 0.8380940501434452,
      "grad_norm": 0.0,
      "learning_rate": 1.343788902868678e-06,
      "loss": 0.8891,
      "step": 6719
    },
    {
      "epoch": 0.8382187850817013,
      "grad_norm": 0.0,
      "learning_rate": 1.341766725891175e-06,
      "loss": 0.9218,
      "step": 6720
    },
    {
      "epoch": 0.8383435200199576,
      "grad_norm": 0.0,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.9277,
      "step": 6721
    },
    {
      "epoch": 0.8384682549582138,
      "grad_norm": 0.0,
      "learning_rate": 1.3377266119918375e-06,
      "loss": 0.8897,
      "step": 6722
    },
    {
      "epoch": 0.83859298989647,
      "grad_norm": 0.0,
      "learning_rate": 1.3357086757294512e-06,
      "loss": 0.8947,
      "step": 6723
    },
    {
      "epoch": 0.8387177248347262,
      "grad_norm": 0.0,
      "learning_rate": 1.3336921536978343e-06,
      "loss": 0.8988,
      "step": 6724
    },
    {
      "epoch": 0.8388424597729824,
      "grad_norm": 0.0,
      "learning_rate": 1.3316770462261365e-06,
      "loss": 0.9402,
      "step": 6725
    },
    {
      "epoch": 0.8389671947112386,
      "grad_norm": 0.0,
      "learning_rate": 1.3296633536432668e-06,
      "loss": 0.8712,
      "step": 6726
    },
    {
      "epoch": 0.8390919296494949,
      "grad_norm": 0.0,
      "learning_rate": 1.3276510762779215e-06,
      "loss": 0.8683,
      "step": 6727
    },
    {
      "epoch": 0.839216664587751,
      "grad_norm": 0.0,
      "learning_rate": 1.3256402144585456e-06,
      "loss": 0.8919,
      "step": 6728
    },
    {
      "epoch": 0.8393413995260073,
      "grad_norm": 0.0,
      "learning_rate": 1.3236307685133675e-06,
      "loss": 0.911,
      "step": 6729
    },
    {
      "epoch": 0.8394661344642634,
      "grad_norm": 0.0,
      "learning_rate": 1.3216227387703795e-06,
      "loss": 0.8931,
      "step": 6730
    },
    {
      "epoch": 0.8395908694025196,
      "grad_norm": 0.0,
      "learning_rate": 1.3196161255573392e-06,
      "loss": 0.9012,
      "step": 6731
    },
    {
      "epoch": 0.8397156043407759,
      "grad_norm": 0.0,
      "learning_rate": 1.3176109292017781e-06,
      "loss": 0.8697,
      "step": 6732
    },
    {
      "epoch": 0.839840339279032,
      "grad_norm": 0.0,
      "learning_rate": 1.3156071500309953e-06,
      "loss": 0.8986,
      "step": 6733
    },
    {
      "epoch": 0.8399650742172883,
      "grad_norm": 0.0,
      "learning_rate": 1.3136047883720582e-06,
      "loss": 0.9494,
      "step": 6734
    },
    {
      "epoch": 0.8400898091555444,
      "grad_norm": 0.0,
      "learning_rate": 1.3116038445518043e-06,
      "loss": 0.8832,
      "step": 6735
    },
    {
      "epoch": 0.8402145440938007,
      "grad_norm": 0.0,
      "learning_rate": 1.3096043188968333e-06,
      "loss": 0.923,
      "step": 6736
    },
    {
      "epoch": 0.8403392790320569,
      "grad_norm": 0.0,
      "learning_rate": 1.307606211733522e-06,
      "loss": 0.876,
      "step": 6737
    },
    {
      "epoch": 0.8404640139703131,
      "grad_norm": 0.0,
      "learning_rate": 1.3056095233880105e-06,
      "loss": 0.8987,
      "step": 6738
    },
    {
      "epoch": 0.8405887489085693,
      "grad_norm": 0.0,
      "learning_rate": 1.3036142541862119e-06,
      "loss": 0.8867,
      "step": 6739
    },
    {
      "epoch": 0.8407134838468255,
      "grad_norm": 0.0,
      "learning_rate": 1.3016204044537995e-06,
      "loss": 0.8826,
      "step": 6740
    },
    {
      "epoch": 0.8408382187850817,
      "grad_norm": 0.0,
      "learning_rate": 1.2996279745162211e-06,
      "loss": 0.8844,
      "step": 6741
    },
    {
      "epoch": 0.840962953723338,
      "grad_norm": 0.0,
      "learning_rate": 1.2976369646986954e-06,
      "loss": 0.9209,
      "step": 6742
    },
    {
      "epoch": 0.8410876886615941,
      "grad_norm": 0.0,
      "learning_rate": 1.2956473753261978e-06,
      "loss": 0.9671,
      "step": 6743
    },
    {
      "epoch": 0.8412124235998503,
      "grad_norm": 0.0,
      "learning_rate": 1.2936592067234888e-06,
      "loss": 0.8605,
      "step": 6744
    },
    {
      "epoch": 0.8413371585381065,
      "grad_norm": 0.0,
      "learning_rate": 1.2916724592150799e-06,
      "loss": 0.9455,
      "step": 6745
    },
    {
      "epoch": 0.8414618934763627,
      "grad_norm": 0.0,
      "learning_rate": 1.2896871331252615e-06,
      "loss": 0.9076,
      "step": 6746
    },
    {
      "epoch": 0.841586628414619,
      "grad_norm": 0.0,
      "learning_rate": 1.287703228778091e-06,
      "loss": 0.914,
      "step": 6747
    },
    {
      "epoch": 0.8417113633528751,
      "grad_norm": 0.0,
      "learning_rate": 1.2857207464973876e-06,
      "loss": 0.889,
      "step": 6748
    },
    {
      "epoch": 0.8418360982911314,
      "grad_norm": 0.0,
      "learning_rate": 1.283739686606743e-06,
      "loss": 0.8356,
      "step": 6749
    },
    {
      "epoch": 0.8419608332293875,
      "grad_norm": 0.0,
      "learning_rate": 1.2817600494295179e-06,
      "loss": 0.9667,
      "step": 6750
    },
    {
      "epoch": 0.8420855681676438,
      "grad_norm": 0.0,
      "learning_rate": 1.2797818352888413e-06,
      "loss": 0.8597,
      "step": 6751
    },
    {
      "epoch": 0.8422103031059,
      "grad_norm": 0.0,
      "learning_rate": 1.2778050445076018e-06,
      "loss": 0.8927,
      "step": 6752
    },
    {
      "epoch": 0.8423350380441562,
      "grad_norm": 0.0,
      "learning_rate": 1.2758296774084633e-06,
      "loss": 0.9371,
      "step": 6753
    },
    {
      "epoch": 0.8424597729824124,
      "grad_norm": 0.0,
      "learning_rate": 1.273855734313858e-06,
      "loss": 0.9225,
      "step": 6754
    },
    {
      "epoch": 0.8425845079206685,
      "grad_norm": 0.0,
      "learning_rate": 1.2718832155459803e-06,
      "loss": 0.8781,
      "step": 6755
    },
    {
      "epoch": 0.8427092428589248,
      "grad_norm": 0.0,
      "learning_rate": 1.2699121214267995e-06,
      "loss": 0.8744,
      "step": 6756
    },
    {
      "epoch": 0.8428339777971809,
      "grad_norm": 0.0,
      "learning_rate": 1.2679424522780426e-06,
      "loss": 0.887,
      "step": 6757
    },
    {
      "epoch": 0.8429587127354372,
      "grad_norm": 0.0,
      "learning_rate": 1.2659742084212111e-06,
      "loss": 0.9444,
      "step": 6758
    },
    {
      "epoch": 0.8430834476736934,
      "grad_norm": 0.0,
      "learning_rate": 1.2640073901775752e-06,
      "loss": 0.8816,
      "step": 6759
    },
    {
      "epoch": 0.8432081826119496,
      "grad_norm": 0.0,
      "learning_rate": 1.262041997868162e-06,
      "loss": 0.8959,
      "step": 6760
    },
    {
      "epoch": 0.8433329175502058,
      "grad_norm": 0.0,
      "learning_rate": 1.2600780318137806e-06,
      "loss": 0.9314,
      "step": 6761
    },
    {
      "epoch": 0.843457652488462,
      "grad_norm": 0.0,
      "learning_rate": 1.2581154923349958e-06,
      "loss": 0.8796,
      "step": 6762
    },
    {
      "epoch": 0.8435823874267182,
      "grad_norm": 0.0,
      "learning_rate": 1.2561543797521436e-06,
      "loss": 0.9398,
      "step": 6763
    },
    {
      "epoch": 0.8437071223649745,
      "grad_norm": 0.0,
      "learning_rate": 1.2541946943853289e-06,
      "loss": 0.9099,
      "step": 6764
    },
    {
      "epoch": 0.8438318573032306,
      "grad_norm": 0.0,
      "learning_rate": 1.2522364365544204e-06,
      "loss": 0.874,
      "step": 6765
    },
    {
      "epoch": 0.8439565922414869,
      "grad_norm": 0.0,
      "learning_rate": 1.2502796065790534e-06,
      "loss": 0.8297,
      "step": 6766
    },
    {
      "epoch": 0.844081327179743,
      "grad_norm": 0.0,
      "learning_rate": 1.2483242047786347e-06,
      "loss": 0.8874,
      "step": 6767
    },
    {
      "epoch": 0.8442060621179992,
      "grad_norm": 0.0,
      "learning_rate": 1.2463702314723369e-06,
      "loss": 0.8626,
      "step": 6768
    },
    {
      "epoch": 0.8443307970562555,
      "grad_norm": 0.0,
      "learning_rate": 1.2444176869790925e-06,
      "loss": 0.9193,
      "step": 6769
    },
    {
      "epoch": 0.8444555319945116,
      "grad_norm": 0.0,
      "learning_rate": 1.2424665716176087e-06,
      "loss": 0.8973,
      "step": 6770
    },
    {
      "epoch": 0.8445802669327679,
      "grad_norm": 0.0,
      "learning_rate": 1.2405168857063587e-06,
      "loss": 0.9439,
      "step": 6771
    },
    {
      "epoch": 0.844705001871024,
      "grad_norm": 0.0,
      "learning_rate": 1.2385686295635735e-06,
      "loss": 0.935,
      "step": 6772
    },
    {
      "epoch": 0.8448297368092803,
      "grad_norm": 0.0,
      "learning_rate": 1.2366218035072674e-06,
      "loss": 0.9056,
      "step": 6773
    },
    {
      "epoch": 0.8449544717475365,
      "grad_norm": 0.0,
      "learning_rate": 1.234676407855203e-06,
      "loss": 0.9071,
      "step": 6774
    },
    {
      "epoch": 0.8450792066857927,
      "grad_norm": 0.0,
      "learning_rate": 1.2327324429249232e-06,
      "loss": 0.8939,
      "step": 6775
    },
    {
      "epoch": 0.8452039416240489,
      "grad_norm": 0.0,
      "learning_rate": 1.2307899090337317e-06,
      "loss": 0.9218,
      "step": 6776
    },
    {
      "epoch": 0.8453286765623051,
      "grad_norm": 0.0,
      "learning_rate": 1.2288488064986947e-06,
      "loss": 0.8809,
      "step": 6777
    },
    {
      "epoch": 0.8454534115005613,
      "grad_norm": 0.0,
      "learning_rate": 1.2269091356366525e-06,
      "loss": 0.9348,
      "step": 6778
    },
    {
      "epoch": 0.8455781464388175,
      "grad_norm": 0.0,
      "learning_rate": 1.2249708967642082e-06,
      "loss": 0.9314,
      "step": 6779
    },
    {
      "epoch": 0.8457028813770737,
      "grad_norm": 0.0,
      "learning_rate": 1.22303409019773e-06,
      "loss": 0.9333,
      "step": 6780
    },
    {
      "epoch": 0.8458276163153299,
      "grad_norm": 0.0,
      "learning_rate": 1.2210987162533571e-06,
      "loss": 0.9153,
      "step": 6781
    },
    {
      "epoch": 0.8459523512535861,
      "grad_norm": 0.0,
      "learning_rate": 1.219164775246986e-06,
      "loss": 0.895,
      "step": 6782
    },
    {
      "epoch": 0.8460770861918423,
      "grad_norm": 0.0,
      "learning_rate": 1.2172322674942872e-06,
      "loss": 0.9179,
      "step": 6783
    },
    {
      "epoch": 0.8462018211300986,
      "grad_norm": 0.0,
      "learning_rate": 1.215301193310695e-06,
      "loss": 0.9268,
      "step": 6784
    },
    {
      "epoch": 0.8463265560683547,
      "grad_norm": 0.0,
      "learning_rate": 1.213371553011411e-06,
      "loss": 0.9118,
      "step": 6785
    },
    {
      "epoch": 0.846451291006611,
      "grad_norm": 0.0,
      "learning_rate": 1.2114433469113963e-06,
      "loss": 0.9224,
      "step": 6786
    },
    {
      "epoch": 0.8465760259448671,
      "grad_norm": 0.0,
      "learning_rate": 1.2095165753253868e-06,
      "loss": 0.8994,
      "step": 6787
    },
    {
      "epoch": 0.8467007608831234,
      "grad_norm": 0.0,
      "learning_rate": 1.20759123856788e-06,
      "loss": 0.8803,
      "step": 6788
    },
    {
      "epoch": 0.8468254958213796,
      "grad_norm": 0.0,
      "learning_rate": 1.2056673369531357e-06,
      "loss": 0.9247,
      "step": 6789
    },
    {
      "epoch": 0.8469502307596358,
      "grad_norm": 0.0,
      "learning_rate": 1.2037448707951894e-06,
      "loss": 0.9125,
      "step": 6790
    },
    {
      "epoch": 0.847074965697892,
      "grad_norm": 0.0,
      "learning_rate": 1.2018238404078298e-06,
      "loss": 0.91,
      "step": 6791
    },
    {
      "epoch": 0.8471997006361481,
      "grad_norm": 0.0,
      "learning_rate": 1.1999042461046207e-06,
      "loss": 0.8867,
      "step": 6792
    },
    {
      "epoch": 0.8473244355744044,
      "grad_norm": 0.0,
      "learning_rate": 1.1979860881988903e-06,
      "loss": 0.9302,
      "step": 6793
    },
    {
      "epoch": 0.8474491705126606,
      "grad_norm": 0.0,
      "learning_rate": 1.1960693670037248e-06,
      "loss": 0.9102,
      "step": 6794
    },
    {
      "epoch": 0.8475739054509168,
      "grad_norm": 0.0,
      "learning_rate": 1.1941540828319852e-06,
      "loss": 0.9282,
      "step": 6795
    },
    {
      "epoch": 0.847698640389173,
      "grad_norm": 0.0,
      "learning_rate": 1.1922402359962936e-06,
      "loss": 0.8542,
      "step": 6796
    },
    {
      "epoch": 0.8478233753274292,
      "grad_norm": 0.0,
      "learning_rate": 1.190327826809039e-06,
      "loss": 0.9462,
      "step": 6797
    },
    {
      "epoch": 0.8479481102656854,
      "grad_norm": 0.0,
      "learning_rate": 1.1884168555823728e-06,
      "loss": 0.9517,
      "step": 6798
    },
    {
      "epoch": 0.8480728452039417,
      "grad_norm": 0.0,
      "learning_rate": 1.186507322628213e-06,
      "loss": 0.891,
      "step": 6799
    },
    {
      "epoch": 0.8481975801421978,
      "grad_norm": 0.0,
      "learning_rate": 1.1845992282582464e-06,
      "loss": 0.9184,
      "step": 6800
    },
    {
      "epoch": 0.8483223150804541,
      "grad_norm": 0.0,
      "learning_rate": 1.18269257278392e-06,
      "loss": 0.912,
      "step": 6801
    },
    {
      "epoch": 0.8484470500187102,
      "grad_norm": 0.0,
      "learning_rate": 1.1807873565164507e-06,
      "loss": 0.8901,
      "step": 6802
    },
    {
      "epoch": 0.8485717849569665,
      "grad_norm": 0.0,
      "learning_rate": 1.1788835797668142e-06,
      "loss": 0.8955,
      "step": 6803
    },
    {
      "epoch": 0.8486965198952227,
      "grad_norm": 0.0,
      "learning_rate": 1.176981242845756e-06,
      "loss": 0.9812,
      "step": 6804
    },
    {
      "epoch": 0.8488212548334788,
      "grad_norm": 0.0,
      "learning_rate": 1.1750803460637895e-06,
      "loss": 0.9305,
      "step": 6805
    },
    {
      "epoch": 0.8489459897717351,
      "grad_norm": 0.0,
      "learning_rate": 1.1731808897311803e-06,
      "loss": 0.8985,
      "step": 6806
    },
    {
      "epoch": 0.8490707247099912,
      "grad_norm": 0.0,
      "learning_rate": 1.1712828741579763e-06,
      "loss": 0.9302,
      "step": 6807
    },
    {
      "epoch": 0.8491954596482475,
      "grad_norm": 0.0,
      "learning_rate": 1.1693862996539795e-06,
      "loss": 0.8647,
      "step": 6808
    },
    {
      "epoch": 0.8493201945865037,
      "grad_norm": 0.0,
      "learning_rate": 1.1674911665287514e-06,
      "loss": 0.9543,
      "step": 6809
    },
    {
      "epoch": 0.8494449295247599,
      "grad_norm": 0.0,
      "learning_rate": 1.1655974750916366e-06,
      "loss": 0.849,
      "step": 6810
    },
    {
      "epoch": 0.8495696644630161,
      "grad_norm": 0.0,
      "learning_rate": 1.1637052256517245e-06,
      "loss": 0.905,
      "step": 6811
    },
    {
      "epoch": 0.8496943994012723,
      "grad_norm": 0.0,
      "learning_rate": 1.1618144185178825e-06,
      "loss": 0.8634,
      "step": 6812
    },
    {
      "epoch": 0.8498191343395285,
      "grad_norm": 0.0,
      "learning_rate": 1.1599250539987362e-06,
      "loss": 0.9291,
      "step": 6813
    },
    {
      "epoch": 0.8499438692777848,
      "grad_norm": 0.0,
      "learning_rate": 1.1580371324026807e-06,
      "loss": 0.8566,
      "step": 6814
    },
    {
      "epoch": 0.8500686042160409,
      "grad_norm": 0.0,
      "learning_rate": 1.1561506540378674e-06,
      "loss": 0.8892,
      "step": 6815
    },
    {
      "epoch": 0.8501933391542971,
      "grad_norm": 0.0,
      "learning_rate": 1.1542656192122192e-06,
      "loss": 0.9006,
      "step": 6816
    },
    {
      "epoch": 0.8503180740925533,
      "grad_norm": 0.0,
      "learning_rate": 1.152382028233422e-06,
      "loss": 0.9244,
      "step": 6817
    },
    {
      "epoch": 0.8504428090308095,
      "grad_norm": 0.0,
      "learning_rate": 1.1504998814089252e-06,
      "loss": 0.9657,
      "step": 6818
    },
    {
      "epoch": 0.8505675439690658,
      "grad_norm": 0.0,
      "learning_rate": 1.1486191790459445e-06,
      "loss": 0.907,
      "step": 6819
    },
    {
      "epoch": 0.8506922789073219,
      "grad_norm": 0.0,
      "learning_rate": 1.146739921451453e-06,
      "loss": 0.8903,
      "step": 6820
    },
    {
      "epoch": 0.8508170138455782,
      "grad_norm": 0.0,
      "learning_rate": 1.1448621089321964e-06,
      "loss": 0.9136,
      "step": 6821
    },
    {
      "epoch": 0.8509417487838343,
      "grad_norm": 0.0,
      "learning_rate": 1.1429857417946822e-06,
      "loss": 0.9131,
      "step": 6822
    },
    {
      "epoch": 0.8510664837220906,
      "grad_norm": 0.0,
      "learning_rate": 1.141110820345177e-06,
      "loss": 0.9215,
      "step": 6823
    },
    {
      "epoch": 0.8511912186603467,
      "grad_norm": 0.0,
      "learning_rate": 1.1392373448897164e-06,
      "loss": 0.9151,
      "step": 6824
    },
    {
      "epoch": 0.851315953598603,
      "grad_norm": 0.0,
      "learning_rate": 1.1373653157341035e-06,
      "loss": 0.8913,
      "step": 6825
    },
    {
      "epoch": 0.8514406885368592,
      "grad_norm": 0.0,
      "learning_rate": 1.1354947331838917e-06,
      "loss": 0.8728,
      "step": 6826
    },
    {
      "epoch": 0.8515654234751154,
      "grad_norm": 0.0,
      "learning_rate": 1.1336255975444165e-06,
      "loss": 0.8782,
      "step": 6827
    },
    {
      "epoch": 0.8516901584133716,
      "grad_norm": 0.0,
      "learning_rate": 1.131757909120762e-06,
      "loss": 0.9105,
      "step": 6828
    },
    {
      "epoch": 0.8518148933516277,
      "grad_norm": 0.0,
      "learning_rate": 1.129891668217783e-06,
      "loss": 0.8742,
      "step": 6829
    },
    {
      "epoch": 0.851939628289884,
      "grad_norm": 0.0,
      "learning_rate": 1.128026875140099e-06,
      "loss": 0.9327,
      "step": 6830
    },
    {
      "epoch": 0.8520643632281402,
      "grad_norm": 0.0,
      "learning_rate": 1.1261635301920914e-06,
      "loss": 0.8867,
      "step": 6831
    },
    {
      "epoch": 0.8521890981663964,
      "grad_norm": 0.0,
      "learning_rate": 1.1243016336779021e-06,
      "loss": 0.9148,
      "step": 6832
    },
    {
      "epoch": 0.8523138331046526,
      "grad_norm": 0.0,
      "learning_rate": 1.1224411859014417e-06,
      "loss": 0.9064,
      "step": 6833
    },
    {
      "epoch": 0.8524385680429089,
      "grad_norm": 0.0,
      "learning_rate": 1.1205821871663825e-06,
      "loss": 0.9092,
      "step": 6834
    },
    {
      "epoch": 0.852563302981165,
      "grad_norm": 0.0,
      "learning_rate": 1.1187246377761563e-06,
      "loss": 0.9145,
      "step": 6835
    },
    {
      "epoch": 0.8526880379194213,
      "grad_norm": 0.0,
      "learning_rate": 1.1168685380339684e-06,
      "loss": 0.898,
      "step": 6836
    },
    {
      "epoch": 0.8528127728576774,
      "grad_norm": 0.0,
      "learning_rate": 1.1150138882427763e-06,
      "loss": 0.9444,
      "step": 6837
    },
    {
      "epoch": 0.8529375077959337,
      "grad_norm": 0.0,
      "learning_rate": 1.1131606887053058e-06,
      "loss": 0.9304,
      "step": 6838
    },
    {
      "epoch": 0.8530622427341898,
      "grad_norm": 0.0,
      "learning_rate": 1.1113089397240495e-06,
      "loss": 0.8863,
      "step": 6839
    },
    {
      "epoch": 0.853186977672446,
      "grad_norm": 0.0,
      "learning_rate": 1.1094586416012555e-06,
      "loss": 0.9343,
      "step": 6840
    },
    {
      "epoch": 0.8533117126107023,
      "grad_norm": 0.0,
      "learning_rate": 1.1076097946389398e-06,
      "loss": 0.8755,
      "step": 6841
    },
    {
      "epoch": 0.8534364475489584,
      "grad_norm": 0.0,
      "learning_rate": 1.1057623991388833e-06,
      "loss": 0.9597,
      "step": 6842
    },
    {
      "epoch": 0.8535611824872147,
      "grad_norm": 0.0,
      "learning_rate": 1.1039164554026217e-06,
      "loss": 0.9074,
      "step": 6843
    },
    {
      "epoch": 0.8536859174254708,
      "grad_norm": 0.0,
      "learning_rate": 1.102071963731468e-06,
      "loss": 0.9128,
      "step": 6844
    },
    {
      "epoch": 0.8538106523637271,
      "grad_norm": 0.0,
      "learning_rate": 1.100228924426483e-06,
      "loss": 0.9204,
      "step": 6845
    },
    {
      "epoch": 0.8539353873019833,
      "grad_norm": 0.0,
      "learning_rate": 1.098387337788499e-06,
      "loss": 0.9075,
      "step": 6846
    },
    {
      "epoch": 0.8540601222402395,
      "grad_norm": 0.0,
      "learning_rate": 1.0965472041181102e-06,
      "loss": 0.8952,
      "step": 6847
    },
    {
      "epoch": 0.8541848571784957,
      "grad_norm": 0.0,
      "learning_rate": 1.094708523715675e-06,
      "loss": 0.8625,
      "step": 6848
    },
    {
      "epoch": 0.854309592116752,
      "grad_norm": 0.0,
      "learning_rate": 1.092871296881307e-06,
      "loss": 0.883,
      "step": 6849
    },
    {
      "epoch": 0.8544343270550081,
      "grad_norm": 0.0,
      "learning_rate": 1.0910355239148917e-06,
      "loss": 0.8766,
      "step": 6850
    },
    {
      "epoch": 0.8545590619932644,
      "grad_norm": 0.0,
      "learning_rate": 1.089201205116074e-06,
      "loss": 0.902,
      "step": 6851
    },
    {
      "epoch": 0.8546837969315205,
      "grad_norm": 0.0,
      "learning_rate": 1.087368340784256e-06,
      "loss": 0.9283,
      "step": 6852
    },
    {
      "epoch": 0.8548085318697767,
      "grad_norm": 0.0,
      "learning_rate": 1.0855369312186148e-06,
      "loss": 0.8959,
      "step": 6853
    },
    {
      "epoch": 0.8549332668080329,
      "grad_norm": 0.0,
      "learning_rate": 1.0837069767180785e-06,
      "loss": 0.895,
      "step": 6854
    },
    {
      "epoch": 0.8550580017462891,
      "grad_norm": 0.0,
      "learning_rate": 1.0818784775813385e-06,
      "loss": 0.9043,
      "step": 6855
    },
    {
      "epoch": 0.8551827366845454,
      "grad_norm": 0.0,
      "learning_rate": 1.0800514341068592e-06,
      "loss": 0.8901,
      "step": 6856
    },
    {
      "epoch": 0.8553074716228015,
      "grad_norm": 0.0,
      "learning_rate": 1.078225846592854e-06,
      "loss": 0.8736,
      "step": 6857
    },
    {
      "epoch": 0.8554322065610578,
      "grad_norm": 0.0,
      "learning_rate": 1.0764017153373086e-06,
      "loss": 0.915,
      "step": 6858
    },
    {
      "epoch": 0.8555569414993139,
      "grad_norm": 0.0,
      "learning_rate": 1.0745790406379675e-06,
      "loss": 0.8981,
      "step": 6859
    },
    {
      "epoch": 0.8556816764375702,
      "grad_norm": 0.0,
      "learning_rate": 1.072757822792333e-06,
      "loss": 0.9034,
      "step": 6860
    },
    {
      "epoch": 0.8558064113758264,
      "grad_norm": 0.0,
      "learning_rate": 1.0709380620976762e-06,
      "loss": 0.8834,
      "step": 6861
    },
    {
      "epoch": 0.8559311463140826,
      "grad_norm": 0.0,
      "learning_rate": 1.0691197588510284e-06,
      "loss": 0.8909,
      "step": 6862
    },
    {
      "epoch": 0.8560558812523388,
      "grad_norm": 0.0,
      "learning_rate": 1.067302913349182e-06,
      "loss": 0.9016,
      "step": 6863
    },
    {
      "epoch": 0.8561806161905949,
      "grad_norm": 0.0,
      "learning_rate": 1.0654875258886921e-06,
      "loss": 0.9068,
      "step": 6864
    },
    {
      "epoch": 0.8563053511288512,
      "grad_norm": 0.0,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 0.8573,
      "step": 6865
    },
    {
      "epoch": 0.8564300860671074,
      "grad_norm": 0.0,
      "learning_rate": 1.0618611262768152e-06,
      "loss": 0.8961,
      "step": 6866
    },
    {
      "epoch": 0.8565548210053636,
      "grad_norm": 0.0,
      "learning_rate": 1.0600501147173447e-06,
      "loss": 0.9083,
      "step": 6867
    },
    {
      "epoch": 0.8566795559436198,
      "grad_norm": 0.0,
      "learning_rate": 1.0582405623830726e-06,
      "loss": 0.9376,
      "step": 6868
    },
    {
      "epoch": 0.856804290881876,
      "grad_norm": 0.0,
      "learning_rate": 1.0564324695693572e-06,
      "loss": 0.8748,
      "step": 6869
    },
    {
      "epoch": 0.8569290258201322,
      "grad_norm": 0.0,
      "learning_rate": 1.0546258365713323e-06,
      "loss": 0.9256,
      "step": 6870
    },
    {
      "epoch": 0.8570537607583885,
      "grad_norm": 0.0,
      "learning_rate": 1.0528206636838834e-06,
      "loss": 0.8314,
      "step": 6871
    },
    {
      "epoch": 0.8571784956966446,
      "grad_norm": 0.0,
      "learning_rate": 1.0510169512016544e-06,
      "loss": 0.9014,
      "step": 6872
    },
    {
      "epoch": 0.8573032306349009,
      "grad_norm": 0.0,
      "learning_rate": 1.0492146994190667e-06,
      "loss": 0.9138,
      "step": 6873
    },
    {
      "epoch": 0.857427965573157,
      "grad_norm": 0.0,
      "learning_rate": 1.0474139086302848e-06,
      "loss": 0.8274,
      "step": 6874
    },
    {
      "epoch": 0.8575527005114133,
      "grad_norm": 0.0,
      "learning_rate": 1.0456145791292471e-06,
      "loss": 0.9244,
      "step": 6875
    },
    {
      "epoch": 0.8576774354496695,
      "grad_norm": 0.0,
      "learning_rate": 1.0438167112096508e-06,
      "loss": 0.9079,
      "step": 6876
    },
    {
      "epoch": 0.8578021703879256,
      "grad_norm": 0.0,
      "learning_rate": 1.042020305164949e-06,
      "loss": 0.8605,
      "step": 6877
    },
    {
      "epoch": 0.8579269053261819,
      "grad_norm": 0.0,
      "learning_rate": 1.040225361288364e-06,
      "loss": 0.8412,
      "step": 6878
    },
    {
      "epoch": 0.858051640264438,
      "grad_norm": 0.0,
      "learning_rate": 1.0384318798728732e-06,
      "loss": 0.8897,
      "step": 6879
    },
    {
      "epoch": 0.8581763752026943,
      "grad_norm": 0.0,
      "learning_rate": 1.0366398612112227e-06,
      "loss": 0.9044,
      "step": 6880
    },
    {
      "epoch": 0.8583011101409505,
      "grad_norm": 0.0,
      "learning_rate": 1.0348493055959064e-06,
      "loss": 0.8626,
      "step": 6881
    },
    {
      "epoch": 0.8584258450792067,
      "grad_norm": 0.0,
      "learning_rate": 1.0330602133191992e-06,
      "loss": 0.9261,
      "step": 6882
    },
    {
      "epoch": 0.8585505800174629,
      "grad_norm": 0.0,
      "learning_rate": 1.0312725846731174e-06,
      "loss": 0.8448,
      "step": 6883
    },
    {
      "epoch": 0.8586753149557191,
      "grad_norm": 0.0,
      "learning_rate": 1.0294864199494503e-06,
      "loss": 0.93,
      "step": 6884
    },
    {
      "epoch": 0.8588000498939753,
      "grad_norm": 0.0,
      "learning_rate": 1.0277017194397466e-06,
      "loss": 0.9158,
      "step": 6885
    },
    {
      "epoch": 0.8589247848322316,
      "grad_norm": 0.0,
      "learning_rate": 1.0259184834353108e-06,
      "loss": 0.8561,
      "step": 6886
    },
    {
      "epoch": 0.8590495197704877,
      "grad_norm": 0.0,
      "learning_rate": 1.0241367122272139e-06,
      "loss": 0.9068,
      "step": 6887
    },
    {
      "epoch": 0.8591742547087439,
      "grad_norm": 0.0,
      "learning_rate": 1.0223564061062884e-06,
      "loss": 0.8963,
      "step": 6888
    },
    {
      "epoch": 0.8592989896470001,
      "grad_norm": 0.0,
      "learning_rate": 1.0205775653631178e-06,
      "loss": 0.9327,
      "step": 6889
    },
    {
      "epoch": 0.8594237245852563,
      "grad_norm": 0.0,
      "learning_rate": 1.018800190288063e-06,
      "loss": 0.9215,
      "step": 6890
    },
    {
      "epoch": 0.8595484595235126,
      "grad_norm": 0.0,
      "learning_rate": 1.0170242811712295e-06,
      "loss": 0.8622,
      "step": 6891
    },
    {
      "epoch": 0.8596731944617687,
      "grad_norm": 0.0,
      "learning_rate": 1.0152498383024922e-06,
      "loss": 0.8792,
      "step": 6892
    },
    {
      "epoch": 0.859797929400025,
      "grad_norm": 0.0,
      "learning_rate": 1.0134768619714885e-06,
      "loss": 0.9126,
      "step": 6893
    },
    {
      "epoch": 0.8599226643382811,
      "grad_norm": 0.0,
      "learning_rate": 1.0117053524676078e-06,
      "loss": 0.9288,
      "step": 6894
    },
    {
      "epoch": 0.8600473992765374,
      "grad_norm": 0.0,
      "learning_rate": 1.009935310080007e-06,
      "loss": 0.9472,
      "step": 6895
    },
    {
      "epoch": 0.8601721342147935,
      "grad_norm": 0.0,
      "learning_rate": 1.0081667350976022e-06,
      "loss": 0.864,
      "step": 6896
    },
    {
      "epoch": 0.8602968691530498,
      "grad_norm": 0.0,
      "learning_rate": 1.0063996278090704e-06,
      "loss": 0.9097,
      "step": 6897
    },
    {
      "epoch": 0.860421604091306,
      "grad_norm": 0.0,
      "learning_rate": 1.0046339885028434e-06,
      "loss": 0.8675,
      "step": 6898
    },
    {
      "epoch": 0.8605463390295622,
      "grad_norm": 0.0,
      "learning_rate": 1.0028698174671248e-06,
      "loss": 0.8729,
      "step": 6899
    },
    {
      "epoch": 0.8606710739678184,
      "grad_norm": 0.0,
      "learning_rate": 1.0011071149898688e-06,
      "loss": 0.8943,
      "step": 6900
    },
    {
      "epoch": 0.8607958089060745,
      "grad_norm": 0.0,
      "learning_rate": 9.993458813587885e-07,
      "loss": 0.9296,
      "step": 6901
    },
    {
      "epoch": 0.8609205438443308,
      "grad_norm": 0.0,
      "learning_rate": 9.975861168613688e-07,
      "loss": 0.924,
      "step": 6902
    },
    {
      "epoch": 0.861045278782587,
      "grad_norm": 0.0,
      "learning_rate": 9.958278217848428e-07,
      "loss": 0.8943,
      "step": 6903
    },
    {
      "epoch": 0.8611700137208432,
      "grad_norm": 0.0,
      "learning_rate": 9.94070996416211e-07,
      "loss": 0.9084,
      "step": 6904
    },
    {
      "epoch": 0.8612947486590994,
      "grad_norm": 0.0,
      "learning_rate": 9.92315641042232e-07,
      "loss": 0.8974,
      "step": 6905
    },
    {
      "epoch": 0.8614194835973557,
      "grad_norm": 0.0,
      "learning_rate": 9.905617559494208e-07,
      "loss": 0.8789,
      "step": 6906
    },
    {
      "epoch": 0.8615442185356118,
      "grad_norm": 0.0,
      "learning_rate": 9.888093414240574e-07,
      "loss": 0.9129,
      "step": 6907
    },
    {
      "epoch": 0.8616689534738681,
      "grad_norm": 0.0,
      "learning_rate": 9.870583977521807e-07,
      "loss": 0.8843,
      "step": 6908
    },
    {
      "epoch": 0.8617936884121242,
      "grad_norm": 0.0,
      "learning_rate": 9.85308925219588e-07,
      "loss": 0.9082,
      "step": 6909
    },
    {
      "epoch": 0.8619184233503805,
      "grad_norm": 0.0,
      "learning_rate": 9.835609241118404e-07,
      "loss": 0.9198,
      "step": 6910
    },
    {
      "epoch": 0.8620431582886366,
      "grad_norm": 0.0,
      "learning_rate": 9.818143947142512e-07,
      "loss": 0.8787,
      "step": 6911
    },
    {
      "epoch": 0.8621678932268928,
      "grad_norm": 0.0,
      "learning_rate": 9.800693373118998e-07,
      "loss": 0.942,
      "step": 6912
    },
    {
      "epoch": 0.8622926281651491,
      "grad_norm": 0.0,
      "learning_rate": 9.783257521896228e-07,
      "loss": 0.9106,
      "step": 6913
    },
    {
      "epoch": 0.8624173631034052,
      "grad_norm": 0.0,
      "learning_rate": 9.765836396320216e-07,
      "loss": 0.8806,
      "step": 6914
    },
    {
      "epoch": 0.8625420980416615,
      "grad_norm": 0.0,
      "learning_rate": 9.748429999234444e-07,
      "loss": 0.9062,
      "step": 6915
    },
    {
      "epoch": 0.8626668329799176,
      "grad_norm": 0.0,
      "learning_rate": 9.73103833348017e-07,
      "loss": 0.851,
      "step": 6916
    },
    {
      "epoch": 0.8627915679181739,
      "grad_norm": 0.0,
      "learning_rate": 9.713661401896101e-07,
      "loss": 0.9581,
      "step": 6917
    },
    {
      "epoch": 0.8629163028564301,
      "grad_norm": 0.0,
      "learning_rate": 9.696299207318571e-07,
      "loss": 0.8685,
      "step": 6918
    },
    {
      "epoch": 0.8630410377946863,
      "grad_norm": 0.0,
      "learning_rate": 9.678951752581577e-07,
      "loss": 0.8783,
      "step": 6919
    },
    {
      "epoch": 0.8631657727329425,
      "grad_norm": 0.0,
      "learning_rate": 9.661619040516623e-07,
      "loss": 0.9038,
      "step": 6920
    },
    {
      "epoch": 0.8632905076711987,
      "grad_norm": 0.0,
      "learning_rate": 9.644301073952866e-07,
      "loss": 0.9334,
      "step": 6921
    },
    {
      "epoch": 0.8634152426094549,
      "grad_norm": 0.0,
      "learning_rate": 9.626997855717045e-07,
      "loss": 0.9104,
      "step": 6922
    },
    {
      "epoch": 0.8635399775477112,
      "grad_norm": 0.0,
      "learning_rate": 9.609709388633447e-07,
      "loss": 0.8893,
      "step": 6923
    },
    {
      "epoch": 0.8636647124859673,
      "grad_norm": 0.0,
      "learning_rate": 9.592435675523991e-07,
      "loss": 0.9286,
      "step": 6924
    },
    {
      "epoch": 0.8637894474242235,
      "grad_norm": 0.0,
      "learning_rate": 9.575176719208211e-07,
      "loss": 0.8489,
      "step": 6925
    },
    {
      "epoch": 0.8639141823624797,
      "grad_norm": 0.0,
      "learning_rate": 9.557932522503176e-07,
      "loss": 0.8833,
      "step": 6926
    },
    {
      "epoch": 0.8640389173007359,
      "grad_norm": 0.0,
      "learning_rate": 9.540703088223614e-07,
      "loss": 0.9026,
      "step": 6927
    },
    {
      "epoch": 0.8641636522389922,
      "grad_norm": 0.0,
      "learning_rate": 9.523488419181737e-07,
      "loss": 0.9375,
      "step": 6928
    },
    {
      "epoch": 0.8642883871772483,
      "grad_norm": 0.0,
      "learning_rate": 9.506288518187468e-07,
      "loss": 0.8773,
      "step": 6929
    },
    {
      "epoch": 0.8644131221155046,
      "grad_norm": 0.0,
      "learning_rate": 9.489103388048238e-07,
      "loss": 0.8814,
      "step": 6930
    },
    {
      "epoch": 0.8645378570537607,
      "grad_norm": 0.0,
      "learning_rate": 9.471933031569125e-07,
      "loss": 0.8873,
      "step": 6931
    },
    {
      "epoch": 0.864662591992017,
      "grad_norm": 0.0,
      "learning_rate": 9.45477745155271e-07,
      "loss": 0.86,
      "step": 6932
    },
    {
      "epoch": 0.8647873269302732,
      "grad_norm": 0.0,
      "learning_rate": 9.437636650799264e-07,
      "loss": 0.9326,
      "step": 6933
    },
    {
      "epoch": 0.8649120618685294,
      "grad_norm": 0.0,
      "learning_rate": 9.420510632106594e-07,
      "loss": 0.9104,
      "step": 6934
    },
    {
      "epoch": 0.8650367968067856,
      "grad_norm": 0.0,
      "learning_rate": 9.403399398270041e-07,
      "loss": 0.8711,
      "step": 6935
    },
    {
      "epoch": 0.8651615317450417,
      "grad_norm": 0.0,
      "learning_rate": 9.386302952082671e-07,
      "loss": 0.9023,
      "step": 6936
    },
    {
      "epoch": 0.865286266683298,
      "grad_norm": 0.0,
      "learning_rate": 9.369221296335007e-07,
      "loss": 0.9024,
      "step": 6937
    },
    {
      "epoch": 0.8654110016215542,
      "grad_norm": 0.0,
      "learning_rate": 9.352154433815209e-07,
      "loss": 0.9198,
      "step": 6938
    },
    {
      "epoch": 0.8655357365598104,
      "grad_norm": 0.0,
      "learning_rate": 9.335102367309057e-07,
      "loss": 0.8641,
      "step": 6939
    },
    {
      "epoch": 0.8656604714980666,
      "grad_norm": 0.0,
      "learning_rate": 9.318065099599815e-07,
      "loss": 0.9374,
      "step": 6940
    },
    {
      "epoch": 0.8657852064363228,
      "grad_norm": 0.0,
      "learning_rate": 9.301042633468438e-07,
      "loss": 0.9136,
      "step": 6941
    },
    {
      "epoch": 0.865909941374579,
      "grad_norm": 0.0,
      "learning_rate": 9.2840349716934e-07,
      "loss": 0.9501,
      "step": 6942
    },
    {
      "epoch": 0.8660346763128353,
      "grad_norm": 0.0,
      "learning_rate": 9.267042117050817e-07,
      "loss": 0.9623,
      "step": 6943
    },
    {
      "epoch": 0.8661594112510914,
      "grad_norm": 0.0,
      "learning_rate": 9.250064072314302e-07,
      "loss": 0.9139,
      "step": 6944
    },
    {
      "epoch": 0.8662841461893477,
      "grad_norm": 0.0,
      "learning_rate": 9.233100840255127e-07,
      "loss": 0.889,
      "step": 6945
    },
    {
      "epoch": 0.8664088811276038,
      "grad_norm": 0.0,
      "learning_rate": 9.216152423642122e-07,
      "loss": 0.925,
      "step": 6946
    },
    {
      "epoch": 0.8665336160658601,
      "grad_norm": 0.0,
      "learning_rate": 9.199218825241673e-07,
      "loss": 0.9498,
      "step": 6947
    },
    {
      "epoch": 0.8666583510041163,
      "grad_norm": 0.0,
      "learning_rate": 9.182300047817816e-07,
      "loss": 0.8929,
      "step": 6948
    },
    {
      "epoch": 0.8667830859423724,
      "grad_norm": 0.0,
      "learning_rate": 9.165396094132073e-07,
      "loss": 0.9379,
      "step": 6949
    },
    {
      "epoch": 0.8669078208806287,
      "grad_norm": 0.0,
      "learning_rate": 9.148506966943615e-07,
      "loss": 0.8752,
      "step": 6950
    },
    {
      "epoch": 0.8670325558188848,
      "grad_norm": 0.0,
      "learning_rate": 9.131632669009194e-07,
      "loss": 0.871,
      "step": 6951
    },
    {
      "epoch": 0.8671572907571411,
      "grad_norm": 0.0,
      "learning_rate": 9.114773203083061e-07,
      "loss": 0.8757,
      "step": 6952
    },
    {
      "epoch": 0.8672820256953973,
      "grad_norm": 0.0,
      "learning_rate": 9.097928571917181e-07,
      "loss": 0.9036,
      "step": 6953
    },
    {
      "epoch": 0.8674067606336535,
      "grad_norm": 0.0,
      "learning_rate": 9.081098778260966e-07,
      "loss": 0.9284,
      "step": 6954
    },
    {
      "epoch": 0.8675314955719097,
      "grad_norm": 0.0,
      "learning_rate": 9.064283824861486e-07,
      "loss": 0.8183,
      "step": 6955
    },
    {
      "epoch": 0.8676562305101659,
      "grad_norm": 0.0,
      "learning_rate": 9.047483714463379e-07,
      "loss": 0.9764,
      "step": 6956
    },
    {
      "epoch": 0.8677809654484221,
      "grad_norm": 0.0,
      "learning_rate": 9.030698449808805e-07,
      "loss": 0.8869,
      "step": 6957
    },
    {
      "epoch": 0.8679057003866784,
      "grad_norm": 0.0,
      "learning_rate": 9.013928033637576e-07,
      "loss": 0.9355,
      "step": 6958
    },
    {
      "epoch": 0.8680304353249345,
      "grad_norm": 0.0,
      "learning_rate": 8.997172468687033e-07,
      "loss": 0.8907,
      "step": 6959
    },
    {
      "epoch": 0.8681551702631907,
      "grad_norm": 0.0,
      "learning_rate": 8.980431757692132e-07,
      "loss": 0.9041,
      "step": 6960
    },
    {
      "epoch": 0.8682799052014469,
      "grad_norm": 0.0,
      "learning_rate": 8.963705903385344e-07,
      "loss": 0.8934,
      "step": 6961
    },
    {
      "epoch": 0.8684046401397031,
      "grad_norm": 0.0,
      "learning_rate": 8.946994908496753e-07,
      "loss": 0.8377,
      "step": 6962
    },
    {
      "epoch": 0.8685293750779594,
      "grad_norm": 0.0,
      "learning_rate": 8.930298775754054e-07,
      "loss": 0.8507,
      "step": 6963
    },
    {
      "epoch": 0.8686541100162155,
      "grad_norm": 0.0,
      "learning_rate": 8.91361750788241e-07,
      "loss": 0.8969,
      "step": 6964
    },
    {
      "epoch": 0.8687788449544718,
      "grad_norm": 0.0,
      "learning_rate": 8.896951107604701e-07,
      "loss": 0.9294,
      "step": 6965
    },
    {
      "epoch": 0.8689035798927279,
      "grad_norm": 0.0,
      "learning_rate": 8.880299577641249e-07,
      "loss": 0.9186,
      "step": 6966
    },
    {
      "epoch": 0.8690283148309842,
      "grad_norm": 0.0,
      "learning_rate": 8.863662920710015e-07,
      "loss": 0.918,
      "step": 6967
    },
    {
      "epoch": 0.8691530497692403,
      "grad_norm": 0.0,
      "learning_rate": 8.847041139526546e-07,
      "loss": 0.9162,
      "step": 6968
    },
    {
      "epoch": 0.8692777847074966,
      "grad_norm": 0.0,
      "learning_rate": 8.830434236803909e-07,
      "loss": 0.9074,
      "step": 6969
    },
    {
      "epoch": 0.8694025196457528,
      "grad_norm": 0.0,
      "learning_rate": 8.813842215252766e-07,
      "loss": 0.8985,
      "step": 6970
    },
    {
      "epoch": 0.869527254584009,
      "grad_norm": 0.0,
      "learning_rate": 8.797265077581362e-07,
      "loss": 0.9525,
      "step": 6971
    },
    {
      "epoch": 0.8696519895222652,
      "grad_norm": 0.0,
      "learning_rate": 8.780702826495502e-07,
      "loss": 0.8857,
      "step": 6972
    },
    {
      "epoch": 0.8697767244605213,
      "grad_norm": 0.0,
      "learning_rate": 8.764155464698598e-07,
      "loss": 0.9371,
      "step": 6973
    },
    {
      "epoch": 0.8699014593987776,
      "grad_norm": 0.0,
      "learning_rate": 8.747622994891536e-07,
      "loss": 0.9085,
      "step": 6974
    },
    {
      "epoch": 0.8700261943370338,
      "grad_norm": 0.0,
      "learning_rate": 8.731105419772868e-07,
      "loss": 0.9053,
      "step": 6975
    },
    {
      "epoch": 0.87015092927529,
      "grad_norm": 0.0,
      "learning_rate": 8.714602742038658e-07,
      "loss": 0.8931,
      "step": 6976
    },
    {
      "epoch": 0.8702756642135462,
      "grad_norm": 0.0,
      "learning_rate": 8.698114964382598e-07,
      "loss": 0.9657,
      "step": 6977
    },
    {
      "epoch": 0.8704003991518025,
      "grad_norm": 0.0,
      "learning_rate": 8.681642089495867e-07,
      "loss": 0.9253,
      "step": 6978
    },
    {
      "epoch": 0.8705251340900586,
      "grad_norm": 0.0,
      "learning_rate": 8.665184120067271e-07,
      "loss": 0.9063,
      "step": 6979
    },
    {
      "epoch": 0.8706498690283149,
      "grad_norm": 0.0,
      "learning_rate": 8.648741058783184e-07,
      "loss": 0.8842,
      "step": 6980
    },
    {
      "epoch": 0.870774603966571,
      "grad_norm": 0.0,
      "learning_rate": 8.632312908327467e-07,
      "loss": 0.8707,
      "step": 6981
    },
    {
      "epoch": 0.8708993389048273,
      "grad_norm": 0.0,
      "learning_rate": 8.615899671381689e-07,
      "loss": 0.8978,
      "step": 6982
    },
    {
      "epoch": 0.8710240738430834,
      "grad_norm": 0.0,
      "learning_rate": 8.599501350624861e-07,
      "loss": 0.901,
      "step": 6983
    },
    {
      "epoch": 0.8711488087813396,
      "grad_norm": 0.0,
      "learning_rate": 8.583117948733566e-07,
      "loss": 0.921,
      "step": 6984
    },
    {
      "epoch": 0.8712735437195959,
      "grad_norm": 0.0,
      "learning_rate": 8.566749468382074e-07,
      "loss": 0.8723,
      "step": 6985
    },
    {
      "epoch": 0.871398278657852,
      "grad_norm": 0.0,
      "learning_rate": 8.55039591224206e-07,
      "loss": 0.9148,
      "step": 6986
    },
    {
      "epoch": 0.8715230135961083,
      "grad_norm": 0.0,
      "learning_rate": 8.534057282982877e-07,
      "loss": 0.8554,
      "step": 6987
    },
    {
      "epoch": 0.8716477485343644,
      "grad_norm": 0.0,
      "learning_rate": 8.517733583271382e-07,
      "loss": 0.8634,
      "step": 6988
    },
    {
      "epoch": 0.8717724834726207,
      "grad_norm": 0.0,
      "learning_rate": 8.501424815772041e-07,
      "loss": 0.8788,
      "step": 6989
    },
    {
      "epoch": 0.8718972184108769,
      "grad_norm": 0.0,
      "learning_rate": 8.485130983146816e-07,
      "loss": 0.8947,
      "step": 6990
    },
    {
      "epoch": 0.8720219533491331,
      "grad_norm": 0.0,
      "learning_rate": 8.468852088055291e-07,
      "loss": 0.8321,
      "step": 6991
    },
    {
      "epoch": 0.8721466882873893,
      "grad_norm": 0.0,
      "learning_rate": 8.452588133154593e-07,
      "loss": 0.9384,
      "step": 6992
    },
    {
      "epoch": 0.8722714232256455,
      "grad_norm": 0.0,
      "learning_rate": 8.436339121099413e-07,
      "loss": 0.8862,
      "step": 6993
    },
    {
      "epoch": 0.8723961581639017,
      "grad_norm": 0.0,
      "learning_rate": 8.420105054542005e-07,
      "loss": 0.9008,
      "step": 6994
    },
    {
      "epoch": 0.872520893102158,
      "grad_norm": 0.0,
      "learning_rate": 8.40388593613215e-07,
      "loss": 0.911,
      "step": 6995
    },
    {
      "epoch": 0.8726456280404141,
      "grad_norm": 0.0,
      "learning_rate": 8.387681768517242e-07,
      "loss": 0.9297,
      "step": 6996
    },
    {
      "epoch": 0.8727703629786703,
      "grad_norm": 0.0,
      "learning_rate": 8.37149255434222e-07,
      "loss": 0.8583,
      "step": 6997
    },
    {
      "epoch": 0.8728950979169265,
      "grad_norm": 0.0,
      "learning_rate": 8.355318296249515e-07,
      "loss": 0.888,
      "step": 6998
    },
    {
      "epoch": 0.8730198328551827,
      "grad_norm": 0.0,
      "learning_rate": 8.339158996879248e-07,
      "loss": 0.9295,
      "step": 6999
    },
    {
      "epoch": 0.873144567793439,
      "grad_norm": 0.0,
      "learning_rate": 8.323014658869e-07,
      "loss": 0.9154,
      "step": 7000
    },
    {
      "epoch": 0.8732693027316951,
      "grad_norm": 0.0,
      "learning_rate": 8.306885284853872e-07,
      "loss": 0.8878,
      "step": 7001
    },
    {
      "epoch": 0.8733940376699514,
      "grad_norm": 0.0,
      "learning_rate": 8.29077087746668e-07,
      "loss": 0.8918,
      "step": 7002
    },
    {
      "epoch": 0.8735187726082075,
      "grad_norm": 0.0,
      "learning_rate": 8.274671439337645e-07,
      "loss": 0.8781,
      "step": 7003
    },
    {
      "epoch": 0.8736435075464638,
      "grad_norm": 0.0,
      "learning_rate": 8.258586973094606e-07,
      "loss": 0.8833,
      "step": 7004
    },
    {
      "epoch": 0.87376824248472,
      "grad_norm": 0.0,
      "learning_rate": 8.242517481362966e-07,
      "loss": 0.8957,
      "step": 7005
    },
    {
      "epoch": 0.8738929774229762,
      "grad_norm": 0.0,
      "learning_rate": 8.226462966765681e-07,
      "loss": 0.9262,
      "step": 7006
    },
    {
      "epoch": 0.8740177123612324,
      "grad_norm": 0.0,
      "learning_rate": 8.21042343192322e-07,
      "loss": 0.9017,
      "step": 7007
    },
    {
      "epoch": 0.8741424472994885,
      "grad_norm": 0.0,
      "learning_rate": 8.194398879453647e-07,
      "loss": 0.9593,
      "step": 7008
    },
    {
      "epoch": 0.8742671822377448,
      "grad_norm": 0.0,
      "learning_rate": 8.178389311972612e-07,
      "loss": 0.9225,
      "step": 7009
    },
    {
      "epoch": 0.874391917176001,
      "grad_norm": 0.0,
      "learning_rate": 8.162394732093215e-07,
      "loss": 0.9477,
      "step": 7010
    },
    {
      "epoch": 0.8745166521142572,
      "grad_norm": 0.0,
      "learning_rate": 8.146415142426245e-07,
      "loss": 0.9166,
      "step": 7011
    },
    {
      "epoch": 0.8746413870525134,
      "grad_norm": 0.0,
      "learning_rate": 8.130450545579926e-07,
      "loss": 0.8976,
      "step": 7012
    },
    {
      "epoch": 0.8747661219907696,
      "grad_norm": 0.0,
      "learning_rate": 8.114500944160086e-07,
      "loss": 0.859,
      "step": 7013
    },
    {
      "epoch": 0.8748908569290258,
      "grad_norm": 0.0,
      "learning_rate": 8.098566340770131e-07,
      "loss": 0.9434,
      "step": 7014
    },
    {
      "epoch": 0.8750155918672821,
      "grad_norm": 0.0,
      "learning_rate": 8.082646738010957e-07,
      "loss": 0.8747,
      "step": 7015
    },
    {
      "epoch": 0.8751403268055382,
      "grad_norm": 0.0,
      "learning_rate": 8.066742138481066e-07,
      "loss": 0.8966,
      "step": 7016
    },
    {
      "epoch": 0.8752650617437945,
      "grad_norm": 0.0,
      "learning_rate": 8.050852544776489e-07,
      "loss": 0.8715,
      "step": 7017
    },
    {
      "epoch": 0.8753897966820506,
      "grad_norm": 0.0,
      "learning_rate": 8.034977959490775e-07,
      "loss": 0.8961,
      "step": 7018
    },
    {
      "epoch": 0.8755145316203069,
      "grad_norm": 0.0,
      "learning_rate": 8.019118385215107e-07,
      "loss": 0.876,
      "step": 7019
    },
    {
      "epoch": 0.8756392665585631,
      "grad_norm": 0.0,
      "learning_rate": 8.003273824538138e-07,
      "loss": 0.9036,
      "step": 7020
    },
    {
      "epoch": 0.8757640014968192,
      "grad_norm": 0.0,
      "learning_rate": 7.987444280046098e-07,
      "loss": 0.9325,
      "step": 7021
    },
    {
      "epoch": 0.8758887364350755,
      "grad_norm": 0.0,
      "learning_rate": 7.971629754322774e-07,
      "loss": 0.8872,
      "step": 7022
    },
    {
      "epoch": 0.8760134713733316,
      "grad_norm": 0.0,
      "learning_rate": 7.955830249949525e-07,
      "loss": 0.9262,
      "step": 7023
    },
    {
      "epoch": 0.8761382063115879,
      "grad_norm": 0.0,
      "learning_rate": 7.940045769505167e-07,
      "loss": 0.9182,
      "step": 7024
    },
    {
      "epoch": 0.876262941249844,
      "grad_norm": 0.0,
      "learning_rate": 7.924276315566171e-07,
      "loss": 0.9059,
      "step": 7025
    },
    {
      "epoch": 0.8763876761881003,
      "grad_norm": 0.0,
      "learning_rate": 7.9085218907065e-07,
      "loss": 0.9152,
      "step": 7026
    },
    {
      "epoch": 0.8765124111263565,
      "grad_norm": 0.0,
      "learning_rate": 7.892782497497642e-07,
      "loss": 0.8744,
      "step": 7027
    },
    {
      "epoch": 0.8766371460646127,
      "grad_norm": 0.0,
      "learning_rate": 7.877058138508731e-07,
      "loss": 0.9161,
      "step": 7028
    },
    {
      "epoch": 0.8767618810028689,
      "grad_norm": 0.0,
      "learning_rate": 7.861348816306314e-07,
      "loss": 0.8621,
      "step": 7029
    },
    {
      "epoch": 0.8768866159411252,
      "grad_norm": 0.0,
      "learning_rate": 7.845654533454572e-07,
      "loss": 0.8915,
      "step": 7030
    },
    {
      "epoch": 0.8770113508793813,
      "grad_norm": 0.0,
      "learning_rate": 7.829975292515224e-07,
      "loss": 0.9104,
      "step": 7031
    },
    {
      "epoch": 0.8771360858176375,
      "grad_norm": 0.0,
      "learning_rate": 7.814311096047478e-07,
      "loss": 0.941,
      "step": 7032
    },
    {
      "epoch": 0.8772608207558937,
      "grad_norm": 0.0,
      "learning_rate": 7.798661946608166e-07,
      "loss": 0.9021,
      "step": 7033
    },
    {
      "epoch": 0.8773855556941499,
      "grad_norm": 0.0,
      "learning_rate": 7.783027846751611e-07,
      "loss": 0.9334,
      "step": 7034
    },
    {
      "epoch": 0.8775102906324062,
      "grad_norm": 0.0,
      "learning_rate": 7.767408799029675e-07,
      "loss": 0.8856,
      "step": 7035
    },
    {
      "epoch": 0.8776350255706623,
      "grad_norm": 0.0,
      "learning_rate": 7.751804805991792e-07,
      "loss": 0.9069,
      "step": 7036
    },
    {
      "epoch": 0.8777597605089186,
      "grad_norm": 0.0,
      "learning_rate": 7.736215870184927e-07,
      "loss": 0.869,
      "step": 7037
    },
    {
      "epoch": 0.8778844954471747,
      "grad_norm": 0.0,
      "learning_rate": 7.720641994153588e-07,
      "loss": 0.8713,
      "step": 7038
    },
    {
      "epoch": 0.878009230385431,
      "grad_norm": 0.0,
      "learning_rate": 7.705083180439832e-07,
      "loss": 0.9099,
      "step": 7039
    },
    {
      "epoch": 0.8781339653236871,
      "grad_norm": 0.0,
      "learning_rate": 7.68953943158326e-07,
      "loss": 0.8793,
      "step": 7040
    },
    {
      "epoch": 0.8782587002619434,
      "grad_norm": 0.0,
      "learning_rate": 7.674010750120964e-07,
      "loss": 0.9462,
      "step": 7041
    },
    {
      "epoch": 0.8783834352001996,
      "grad_norm": 0.0,
      "learning_rate": 7.65849713858764e-07,
      "loss": 0.9359,
      "step": 7042
    },
    {
      "epoch": 0.8785081701384558,
      "grad_norm": 0.0,
      "learning_rate": 7.642998599515517e-07,
      "loss": 0.8695,
      "step": 7043
    },
    {
      "epoch": 0.878632905076712,
      "grad_norm": 0.0,
      "learning_rate": 7.627515135434293e-07,
      "loss": 0.9376,
      "step": 7044
    },
    {
      "epoch": 0.8787576400149681,
      "grad_norm": 0.0,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.8557,
      "step": 7045
    },
    {
      "epoch": 0.8788823749532244,
      "grad_norm": 0.0,
      "learning_rate": 7.59659344235143e-07,
      "loss": 0.9316,
      "step": 7046
    },
    {
      "epoch": 0.8790071098914806,
      "grad_norm": 0.0,
      "learning_rate": 7.581155218396918e-07,
      "loss": 0.8795,
      "step": 7047
    },
    {
      "epoch": 0.8791318448297368,
      "grad_norm": 0.0,
      "learning_rate": 7.565732079527765e-07,
      "loss": 0.8967,
      "step": 7048
    },
    {
      "epoch": 0.879256579767993,
      "grad_norm": 0.0,
      "learning_rate": 7.550324028261391e-07,
      "loss": 0.9366,
      "step": 7049
    },
    {
      "epoch": 0.8793813147062493,
      "grad_norm": 0.0,
      "learning_rate": 7.534931067112772e-07,
      "loss": 0.908,
      "step": 7050
    },
    {
      "epoch": 0.8795060496445054,
      "grad_norm": 0.0,
      "learning_rate": 7.519553198594453e-07,
      "loss": 0.8828,
      "step": 7051
    },
    {
      "epoch": 0.8796307845827617,
      "grad_norm": 0.0,
      "learning_rate": 7.504190425216451e-07,
      "loss": 0.8864,
      "step": 7052
    },
    {
      "epoch": 0.8797555195210178,
      "grad_norm": 0.0,
      "learning_rate": 7.48884274948638e-07,
      "loss": 0.9045,
      "step": 7053
    },
    {
      "epoch": 0.8798802544592741,
      "grad_norm": 0.0,
      "learning_rate": 7.47351017390936e-07,
      "loss": 0.8915,
      "step": 7054
    },
    {
      "epoch": 0.8800049893975302,
      "grad_norm": 0.0,
      "learning_rate": 7.458192700988065e-07,
      "loss": 0.8924,
      "step": 7055
    },
    {
      "epoch": 0.8801297243357864,
      "grad_norm": 0.0,
      "learning_rate": 7.442890333222685e-07,
      "loss": 0.9038,
      "step": 7056
    },
    {
      "epoch": 0.8802544592740427,
      "grad_norm": 0.0,
      "learning_rate": 7.427603073110967e-07,
      "loss": 0.8641,
      "step": 7057
    },
    {
      "epoch": 0.8803791942122988,
      "grad_norm": 0.0,
      "learning_rate": 7.412330923148148e-07,
      "loss": 0.8745,
      "step": 7058
    },
    {
      "epoch": 0.8805039291505551,
      "grad_norm": 0.0,
      "learning_rate": 7.397073885827055e-07,
      "loss": 0.8628,
      "step": 7059
    },
    {
      "epoch": 0.8806286640888112,
      "grad_norm": 0.0,
      "learning_rate": 7.381831963638019e-07,
      "loss": 0.9094,
      "step": 7060
    },
    {
      "epoch": 0.8807533990270675,
      "grad_norm": 0.0,
      "learning_rate": 7.366605159068874e-07,
      "loss": 0.9138,
      "step": 7061
    },
    {
      "epoch": 0.8808781339653237,
      "grad_norm": 0.0,
      "learning_rate": 7.351393474605051e-07,
      "loss": 0.9378,
      "step": 7062
    },
    {
      "epoch": 0.8810028689035799,
      "grad_norm": 0.0,
      "learning_rate": 7.336196912729488e-07,
      "loss": 0.8596,
      "step": 7063
    },
    {
      "epoch": 0.8811276038418361,
      "grad_norm": 0.0,
      "learning_rate": 7.321015475922599e-07,
      "loss": 0.8884,
      "step": 7064
    },
    {
      "epoch": 0.8812523387800923,
      "grad_norm": 0.0,
      "learning_rate": 7.305849166662426e-07,
      "loss": 0.9192,
      "step": 7065
    },
    {
      "epoch": 0.8813770737183485,
      "grad_norm": 0.0,
      "learning_rate": 7.290697987424477e-07,
      "loss": 0.9202,
      "step": 7066
    },
    {
      "epoch": 0.8815018086566048,
      "grad_norm": 0.0,
      "learning_rate": 7.275561940681796e-07,
      "loss": 0.9148,
      "step": 7067
    },
    {
      "epoch": 0.8816265435948609,
      "grad_norm": 0.0,
      "learning_rate": 7.260441028904996e-07,
      "loss": 0.869,
      "step": 7068
    },
    {
      "epoch": 0.8817512785331171,
      "grad_norm": 0.0,
      "learning_rate": 7.245335254562147e-07,
      "loss": 0.9158,
      "step": 7069
    },
    {
      "epoch": 0.8818760134713733,
      "grad_norm": 0.0,
      "learning_rate": 7.23024462011892e-07,
      "loss": 0.9235,
      "step": 7070
    },
    {
      "epoch": 0.8820007484096295,
      "grad_norm": 0.0,
      "learning_rate": 7.21516912803849e-07,
      "loss": 0.9076,
      "step": 7071
    },
    {
      "epoch": 0.8821254833478858,
      "grad_norm": 0.0,
      "learning_rate": 7.200108780781556e-07,
      "loss": 0.9267,
      "step": 7072
    },
    {
      "epoch": 0.8822502182861419,
      "grad_norm": 0.0,
      "learning_rate": 7.185063580806318e-07,
      "loss": 0.96,
      "step": 7073
    },
    {
      "epoch": 0.8823749532243982,
      "grad_norm": 0.0,
      "learning_rate": 7.170033530568587e-07,
      "loss": 0.9198,
      "step": 7074
    },
    {
      "epoch": 0.8824996881626543,
      "grad_norm": 0.0,
      "learning_rate": 7.155018632521593e-07,
      "loss": 0.896,
      "step": 7075
    },
    {
      "epoch": 0.8826244231009106,
      "grad_norm": 0.0,
      "learning_rate": 7.140018889116163e-07,
      "loss": 0.905,
      "step": 7076
    },
    {
      "epoch": 0.8827491580391668,
      "grad_norm": 0.0,
      "learning_rate": 7.12503430280066e-07,
      "loss": 0.8942,
      "step": 7077
    },
    {
      "epoch": 0.882873892977423,
      "grad_norm": 0.0,
      "learning_rate": 7.110064876020906e-07,
      "loss": 0.9353,
      "step": 7078
    },
    {
      "epoch": 0.8829986279156792,
      "grad_norm": 0.0,
      "learning_rate": 7.095110611220312e-07,
      "loss": 0.8922,
      "step": 7079
    },
    {
      "epoch": 0.8831233628539353,
      "grad_norm": 0.0,
      "learning_rate": 7.080171510839795e-07,
      "loss": 0.8476,
      "step": 7080
    },
    {
      "epoch": 0.8832480977921916,
      "grad_norm": 0.0,
      "learning_rate": 7.065247577317747e-07,
      "loss": 0.8553,
      "step": 7081
    },
    {
      "epoch": 0.8833728327304478,
      "grad_norm": 0.0,
      "learning_rate": 7.050338813090207e-07,
      "loss": 0.922,
      "step": 7082
    },
    {
      "epoch": 0.883497567668704,
      "grad_norm": 0.0,
      "learning_rate": 7.035445220590597e-07,
      "loss": 0.9053,
      "step": 7083
    },
    {
      "epoch": 0.8836223026069602,
      "grad_norm": 0.0,
      "learning_rate": 7.02056680224995e-07,
      "loss": 0.8988,
      "step": 7084
    },
    {
      "epoch": 0.8837470375452164,
      "grad_norm": 0.0,
      "learning_rate": 7.005703560496823e-07,
      "loss": 0.8892,
      "step": 7085
    },
    {
      "epoch": 0.8838717724834726,
      "grad_norm": 0.0,
      "learning_rate": 6.990855497757221e-07,
      "loss": 0.9113,
      "step": 7086
    },
    {
      "epoch": 0.8839965074217289,
      "grad_norm": 0.0,
      "learning_rate": 6.976022616454759e-07,
      "loss": 0.9485,
      "step": 7087
    },
    {
      "epoch": 0.884121242359985,
      "grad_norm": 0.0,
      "learning_rate": 6.961204919010523e-07,
      "loss": 0.9455,
      "step": 7088
    },
    {
      "epoch": 0.8842459772982413,
      "grad_norm": 0.0,
      "learning_rate": 6.946402407843156e-07,
      "loss": 0.9347,
      "step": 7089
    },
    {
      "epoch": 0.8843707122364974,
      "grad_norm": 0.0,
      "learning_rate": 6.931615085368748e-07,
      "loss": 0.9015,
      "step": 7090
    },
    {
      "epoch": 0.8844954471747537,
      "grad_norm": 0.0,
      "learning_rate": 6.916842954001035e-07,
      "loss": 0.8859,
      "step": 7091
    },
    {
      "epoch": 0.8846201821130099,
      "grad_norm": 0.0,
      "learning_rate": 6.902086016151166e-07,
      "loss": 0.9246,
      "step": 7092
    },
    {
      "epoch": 0.884744917051266,
      "grad_norm": 0.0,
      "learning_rate": 6.887344274227803e-07,
      "loss": 0.9216,
      "step": 7093
    },
    {
      "epoch": 0.8848696519895223,
      "grad_norm": 0.0,
      "learning_rate": 6.872617730637255e-07,
      "loss": 0.9205,
      "step": 7094
    },
    {
      "epoch": 0.8849943869277784,
      "grad_norm": 0.0,
      "learning_rate": 6.857906387783197e-07,
      "loss": 0.9383,
      "step": 7095
    },
    {
      "epoch": 0.8851191218660347,
      "grad_norm": 0.0,
      "learning_rate": 6.843210248066922e-07,
      "loss": 0.8575,
      "step": 7096
    },
    {
      "epoch": 0.8852438568042909,
      "grad_norm": 0.0,
      "learning_rate": 6.828529313887211e-07,
      "loss": 0.8909,
      "step": 7097
    },
    {
      "epoch": 0.8853685917425471,
      "grad_norm": 0.0,
      "learning_rate": 6.813863587640345e-07,
      "loss": 0.857,
      "step": 7098
    },
    {
      "epoch": 0.8854933266808033,
      "grad_norm": 0.0,
      "learning_rate": 6.799213071720156e-07,
      "loss": 0.8616,
      "step": 7099
    },
    {
      "epoch": 0.8856180616190595,
      "grad_norm": 0.0,
      "learning_rate": 6.784577768517963e-07,
      "loss": 0.9264,
      "step": 7100
    },
    {
      "epoch": 0.8857427965573157,
      "grad_norm": 0.0,
      "learning_rate": 6.769957680422634e-07,
      "loss": 0.8783,
      "step": 7101
    },
    {
      "epoch": 0.885867531495572,
      "grad_norm": 0.0,
      "learning_rate": 6.755352809820537e-07,
      "loss": 0.8784,
      "step": 7102
    },
    {
      "epoch": 0.8859922664338281,
      "grad_norm": 0.0,
      "learning_rate": 6.740763159095532e-07,
      "loss": 0.8801,
      "step": 7103
    },
    {
      "epoch": 0.8861170013720843,
      "grad_norm": 0.0,
      "learning_rate": 6.726188730629035e-07,
      "loss": 0.8634,
      "step": 7104
    },
    {
      "epoch": 0.8862417363103405,
      "grad_norm": 0.0,
      "learning_rate": 6.711629526799946e-07,
      "loss": 0.9062,
      "step": 7105
    },
    {
      "epoch": 0.8863664712485967,
      "grad_norm": 0.0,
      "learning_rate": 6.697085549984727e-07,
      "loss": 0.8676,
      "step": 7106
    },
    {
      "epoch": 0.886491206186853,
      "grad_norm": 0.0,
      "learning_rate": 6.682556802557272e-07,
      "loss": 0.8642,
      "step": 7107
    },
    {
      "epoch": 0.8866159411251091,
      "grad_norm": 0.0,
      "learning_rate": 6.668043286889092e-07,
      "loss": 0.9215,
      "step": 7108
    },
    {
      "epoch": 0.8867406760633654,
      "grad_norm": 0.0,
      "learning_rate": 6.653545005349138e-07,
      "loss": 0.9014,
      "step": 7109
    },
    {
      "epoch": 0.8868654110016215,
      "grad_norm": 0.0,
      "learning_rate": 6.63906196030385e-07,
      "loss": 0.8935,
      "step": 7110
    },
    {
      "epoch": 0.8869901459398778,
      "grad_norm": 0.0,
      "learning_rate": 6.624594154117303e-07,
      "loss": 0.9067,
      "step": 7111
    },
    {
      "epoch": 0.887114880878134,
      "grad_norm": 0.0,
      "learning_rate": 6.610141589150953e-07,
      "loss": 0.8867,
      "step": 7112
    },
    {
      "epoch": 0.8872396158163902,
      "grad_norm": 0.0,
      "learning_rate": 6.595704267763848e-07,
      "loss": 0.9005,
      "step": 7113
    },
    {
      "epoch": 0.8873643507546464,
      "grad_norm": 0.0,
      "learning_rate": 6.581282192312521e-07,
      "loss": 0.9888,
      "step": 7114
    },
    {
      "epoch": 0.8874890856929026,
      "grad_norm": 0.0,
      "learning_rate": 6.566875365151004e-07,
      "loss": 0.8895,
      "step": 7115
    },
    {
      "epoch": 0.8876138206311588,
      "grad_norm": 0.0,
      "learning_rate": 6.552483788630858e-07,
      "loss": 0.8679,
      "step": 7116
    },
    {
      "epoch": 0.8877385555694149,
      "grad_norm": 0.0,
      "learning_rate": 6.538107465101162e-07,
      "loss": 0.8807,
      "step": 7117
    },
    {
      "epoch": 0.8878632905076712,
      "grad_norm": 0.0,
      "learning_rate": 6.523746396908493e-07,
      "loss": 0.9327,
      "step": 7118
    },
    {
      "epoch": 0.8879880254459274,
      "grad_norm": 0.0,
      "learning_rate": 6.50940058639693e-07,
      "loss": 0.9099,
      "step": 7119
    },
    {
      "epoch": 0.8881127603841836,
      "grad_norm": 0.0,
      "learning_rate": 6.495070035908068e-07,
      "loss": 0.8917,
      "step": 7120
    },
    {
      "epoch": 0.8882374953224398,
      "grad_norm": 0.0,
      "learning_rate": 6.480754747781037e-07,
      "loss": 0.8744,
      "step": 7121
    },
    {
      "epoch": 0.888362230260696,
      "grad_norm": 0.0,
      "learning_rate": 6.466454724352422e-07,
      "loss": 0.8649,
      "step": 7122
    },
    {
      "epoch": 0.8884869651989522,
      "grad_norm": 0.0,
      "learning_rate": 6.45216996795639e-07,
      "loss": 0.8668,
      "step": 7123
    },
    {
      "epoch": 0.8886117001372085,
      "grad_norm": 0.0,
      "learning_rate": 6.437900480924519e-07,
      "loss": 0.8683,
      "step": 7124
    },
    {
      "epoch": 0.8887364350754646,
      "grad_norm": 0.0,
      "learning_rate": 6.423646265585993e-07,
      "loss": 0.9056,
      "step": 7125
    },
    {
      "epoch": 0.8888611700137209,
      "grad_norm": 0.0,
      "learning_rate": 6.409407324267448e-07,
      "loss": 0.9306,
      "step": 7126
    },
    {
      "epoch": 0.888985904951977,
      "grad_norm": 0.0,
      "learning_rate": 6.395183659293003e-07,
      "loss": 0.8771,
      "step": 7127
    },
    {
      "epoch": 0.8891106398902332,
      "grad_norm": 0.0,
      "learning_rate": 6.38097527298438e-07,
      "loss": 0.8783,
      "step": 7128
    },
    {
      "epoch": 0.8892353748284895,
      "grad_norm": 0.0,
      "learning_rate": 6.36678216766069e-07,
      "loss": 0.928,
      "step": 7129
    },
    {
      "epoch": 0.8893601097667456,
      "grad_norm": 0.0,
      "learning_rate": 6.352604345638647e-07,
      "loss": 0.849,
      "step": 7130
    },
    {
      "epoch": 0.8894848447050019,
      "grad_norm": 0.0,
      "learning_rate": 6.33844180923242e-07,
      "loss": 0.8808,
      "step": 7131
    },
    {
      "epoch": 0.889609579643258,
      "grad_norm": 0.0,
      "learning_rate": 6.324294560753663e-07,
      "loss": 0.8724,
      "step": 7132
    },
    {
      "epoch": 0.8897343145815143,
      "grad_norm": 0.0,
      "learning_rate": 6.310162602511593e-07,
      "loss": 0.9039,
      "step": 7133
    },
    {
      "epoch": 0.8898590495197705,
      "grad_norm": 0.0,
      "learning_rate": 6.296045936812889e-07,
      "loss": 0.912,
      "step": 7134
    },
    {
      "epoch": 0.8899837844580267,
      "grad_norm": 0.0,
      "learning_rate": 6.281944565961772e-07,
      "loss": 0.9116,
      "step": 7135
    },
    {
      "epoch": 0.8901085193962829,
      "grad_norm": 0.0,
      "learning_rate": 6.267858492259904e-07,
      "loss": 0.9256,
      "step": 7136
    },
    {
      "epoch": 0.8902332543345391,
      "grad_norm": 0.0,
      "learning_rate": 6.253787718006499e-07,
      "loss": 0.9292,
      "step": 7137
    },
    {
      "epoch": 0.8903579892727953,
      "grad_norm": 0.0,
      "learning_rate": 6.239732245498287e-07,
      "loss": 0.9062,
      "step": 7138
    },
    {
      "epoch": 0.8904827242110516,
      "grad_norm": 0.0,
      "learning_rate": 6.225692077029421e-07,
      "loss": 0.9039,
      "step": 7139
    },
    {
      "epoch": 0.8906074591493077,
      "grad_norm": 0.0,
      "learning_rate": 6.211667214891682e-07,
      "loss": 0.8733,
      "step": 7140
    },
    {
      "epoch": 0.8907321940875639,
      "grad_norm": 0.0,
      "learning_rate": 6.197657661374223e-07,
      "loss": 0.895,
      "step": 7141
    },
    {
      "epoch": 0.8908569290258201,
      "grad_norm": 0.0,
      "learning_rate": 6.183663418763774e-07,
      "loss": 0.865,
      "step": 7142
    },
    {
      "epoch": 0.8909816639640763,
      "grad_norm": 0.0,
      "learning_rate": 6.169684489344574e-07,
      "loss": 0.8622,
      "step": 7143
    },
    {
      "epoch": 0.8911063989023326,
      "grad_norm": 0.0,
      "learning_rate": 6.155720875398297e-07,
      "loss": 0.8816,
      "step": 7144
    },
    {
      "epoch": 0.8912311338405887,
      "grad_norm": 0.0,
      "learning_rate": 6.141772579204175e-07,
      "loss": 0.9683,
      "step": 7145
    },
    {
      "epoch": 0.891355868778845,
      "grad_norm": 0.0,
      "learning_rate": 6.12783960303892e-07,
      "loss": 0.9106,
      "step": 7146
    },
    {
      "epoch": 0.8914806037171011,
      "grad_norm": 0.0,
      "learning_rate": 6.113921949176749e-07,
      "loss": 0.9217,
      "step": 7147
    },
    {
      "epoch": 0.8916053386553574,
      "grad_norm": 0.0,
      "learning_rate": 6.100019619889374e-07,
      "loss": 0.964,
      "step": 7148
    },
    {
      "epoch": 0.8917300735936136,
      "grad_norm": 0.0,
      "learning_rate": 6.086132617445995e-07,
      "loss": 0.8911,
      "step": 7149
    },
    {
      "epoch": 0.8918548085318698,
      "grad_norm": 0.0,
      "learning_rate": 6.072260944113328e-07,
      "loss": 0.8892,
      "step": 7150
    },
    {
      "epoch": 0.891979543470126,
      "grad_norm": 0.0,
      "learning_rate": 6.058404602155576e-07,
      "loss": 0.8817,
      "step": 7151
    },
    {
      "epoch": 0.8921042784083821,
      "grad_norm": 0.0,
      "learning_rate": 6.044563593834473e-07,
      "loss": 0.8742,
      "step": 7152
    },
    {
      "epoch": 0.8922290133466384,
      "grad_norm": 0.0,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.8943,
      "step": 7153
    },
    {
      "epoch": 0.8923537482848946,
      "grad_norm": 0.0,
      "learning_rate": 6.01692758713639e-07,
      "loss": 0.8634,
      "step": 7154
    },
    {
      "epoch": 0.8924784832231508,
      "grad_norm": 0.0,
      "learning_rate": 6.003132593270345e-07,
      "loss": 0.9119,
      "step": 7155
    },
    {
      "epoch": 0.892603218161407,
      "grad_norm": 0.0,
      "learning_rate": 5.989352942062676e-07,
      "loss": 0.9218,
      "step": 7156
    },
    {
      "epoch": 0.8927279530996632,
      "grad_norm": 0.0,
      "learning_rate": 5.975588635762631e-07,
      "loss": 0.8478,
      "step": 7157
    },
    {
      "epoch": 0.8928526880379194,
      "grad_norm": 0.0,
      "learning_rate": 5.961839676616854e-07,
      "loss": 0.9027,
      "step": 7158
    },
    {
      "epoch": 0.8929774229761757,
      "grad_norm": 0.0,
      "learning_rate": 5.948106066869519e-07,
      "loss": 0.9285,
      "step": 7159
    },
    {
      "epoch": 0.8931021579144318,
      "grad_norm": 0.0,
      "learning_rate": 5.934387808762332e-07,
      "loss": 0.9507,
      "step": 7160
    },
    {
      "epoch": 0.8932268928526881,
      "grad_norm": 0.0,
      "learning_rate": 5.920684904534424e-07,
      "loss": 0.9344,
      "step": 7161
    },
    {
      "epoch": 0.8933516277909442,
      "grad_norm": 0.0,
      "learning_rate": 5.906997356422461e-07,
      "loss": 0.9213,
      "step": 7162
    },
    {
      "epoch": 0.8934763627292005,
      "grad_norm": 0.0,
      "learning_rate": 5.893325166660601e-07,
      "loss": 0.9003,
      "step": 7163
    },
    {
      "epoch": 0.8936010976674567,
      "grad_norm": 0.0,
      "learning_rate": 5.879668337480505e-07,
      "loss": 0.8852,
      "step": 7164
    },
    {
      "epoch": 0.8937258326057128,
      "grad_norm": 0.0,
      "learning_rate": 5.866026871111275e-07,
      "loss": 0.9388,
      "step": 7165
    },
    {
      "epoch": 0.8938505675439691,
      "grad_norm": 0.0,
      "learning_rate": 5.852400769779576e-07,
      "loss": 0.9145,
      "step": 7166
    },
    {
      "epoch": 0.8939753024822252,
      "grad_norm": 0.0,
      "learning_rate": 5.838790035709507e-07,
      "loss": 0.9562,
      "step": 7167
    },
    {
      "epoch": 0.8941000374204815,
      "grad_norm": 0.0,
      "learning_rate": 5.825194671122702e-07,
      "loss": 0.8612,
      "step": 7168
    },
    {
      "epoch": 0.8942247723587377,
      "grad_norm": 0.0,
      "learning_rate": 5.811614678238276e-07,
      "loss": 0.8921,
      "step": 7169
    },
    {
      "epoch": 0.8943495072969939,
      "grad_norm": 0.0,
      "learning_rate": 5.798050059272786e-07,
      "loss": 0.9066,
      "step": 7170
    },
    {
      "epoch": 0.8944742422352501,
      "grad_norm": 0.0,
      "learning_rate": 5.784500816440353e-07,
      "loss": 0.8579,
      "step": 7171
    },
    {
      "epoch": 0.8945989771735063,
      "grad_norm": 0.0,
      "learning_rate": 5.77096695195255e-07,
      "loss": 0.8784,
      "step": 7172
    },
    {
      "epoch": 0.8947237121117625,
      "grad_norm": 0.0,
      "learning_rate": 5.757448468018423e-07,
      "loss": 0.9211,
      "step": 7173
    },
    {
      "epoch": 0.8948484470500188,
      "grad_norm": 0.0,
      "learning_rate": 5.743945366844572e-07,
      "loss": 0.9145,
      "step": 7174
    },
    {
      "epoch": 0.8949731819882749,
      "grad_norm": 0.0,
      "learning_rate": 5.730457650635024e-07,
      "loss": 0.8543,
      "step": 7175
    },
    {
      "epoch": 0.8950979169265311,
      "grad_norm": 0.0,
      "learning_rate": 5.716985321591272e-07,
      "loss": 0.8829,
      "step": 7176
    },
    {
      "epoch": 0.8952226518647873,
      "grad_norm": 0.0,
      "learning_rate": 5.703528381912415e-07,
      "loss": 0.8728,
      "step": 7177
    },
    {
      "epoch": 0.8953473868030435,
      "grad_norm": 0.0,
      "learning_rate": 5.690086833794906e-07,
      "loss": 0.9117,
      "step": 7178
    },
    {
      "epoch": 0.8954721217412998,
      "grad_norm": 0.0,
      "learning_rate": 5.676660679432766e-07,
      "loss": 0.92,
      "step": 7179
    },
    {
      "epoch": 0.8955968566795559,
      "grad_norm": 0.0,
      "learning_rate": 5.663249921017477e-07,
      "loss": 0.9202,
      "step": 7180
    },
    {
      "epoch": 0.8957215916178122,
      "grad_norm": 0.0,
      "learning_rate": 5.649854560738033e-07,
      "loss": 0.9218,
      "step": 7181
    },
    {
      "epoch": 0.8958463265560683,
      "grad_norm": 0.0,
      "learning_rate": 5.636474600780872e-07,
      "loss": 0.8951,
      "step": 7182
    },
    {
      "epoch": 0.8959710614943246,
      "grad_norm": 0.0,
      "learning_rate": 5.623110043329937e-07,
      "loss": 0.8825,
      "step": 7183
    },
    {
      "epoch": 0.8960957964325807,
      "grad_norm": 0.0,
      "learning_rate": 5.609760890566673e-07,
      "loss": 0.9077,
      "step": 7184
    },
    {
      "epoch": 0.896220531370837,
      "grad_norm": 0.0,
      "learning_rate": 5.596427144670002e-07,
      "loss": 0.9116,
      "step": 7185
    },
    {
      "epoch": 0.8963452663090932,
      "grad_norm": 0.0,
      "learning_rate": 5.583108807816329e-07,
      "loss": 0.8539,
      "step": 7186
    },
    {
      "epoch": 0.8964700012473494,
      "grad_norm": 0.0,
      "learning_rate": 5.569805882179524e-07,
      "loss": 0.9241,
      "step": 7187
    },
    {
      "epoch": 0.8965947361856056,
      "grad_norm": 0.0,
      "learning_rate": 5.556518369930963e-07,
      "loss": 0.9078,
      "step": 7188
    },
    {
      "epoch": 0.8967194711238617,
      "grad_norm": 0.0,
      "learning_rate": 5.543246273239533e-07,
      "loss": 0.8964,
      "step": 7189
    },
    {
      "epoch": 0.896844206062118,
      "grad_norm": 0.0,
      "learning_rate": 5.529989594271523e-07,
      "loss": 0.9167,
      "step": 7190
    },
    {
      "epoch": 0.8969689410003742,
      "grad_norm": 0.0,
      "learning_rate": 5.516748335190825e-07,
      "loss": 0.8754,
      "step": 7191
    },
    {
      "epoch": 0.8970936759386304,
      "grad_norm": 0.0,
      "learning_rate": 5.503522498158698e-07,
      "loss": 0.8906,
      "step": 7192
    },
    {
      "epoch": 0.8972184108768866,
      "grad_norm": 0.0,
      "learning_rate": 5.490312085333904e-07,
      "loss": 0.8806,
      "step": 7193
    },
    {
      "epoch": 0.8973431458151429,
      "grad_norm": 0.0,
      "learning_rate": 5.477117098872797e-07,
      "loss": 0.8683,
      "step": 7194
    },
    {
      "epoch": 0.897467880753399,
      "grad_norm": 0.0,
      "learning_rate": 5.463937540929065e-07,
      "loss": 0.8776,
      "step": 7195
    },
    {
      "epoch": 0.8975926156916553,
      "grad_norm": 0.0,
      "learning_rate": 5.450773413653976e-07,
      "loss": 0.8858,
      "step": 7196
    },
    {
      "epoch": 0.8977173506299114,
      "grad_norm": 0.0,
      "learning_rate": 5.437624719196222e-07,
      "loss": 0.9138,
      "step": 7197
    },
    {
      "epoch": 0.8978420855681677,
      "grad_norm": 0.0,
      "learning_rate": 5.424491459702053e-07,
      "loss": 0.8459,
      "step": 7198
    },
    {
      "epoch": 0.8979668205064238,
      "grad_norm": 0.0,
      "learning_rate": 5.411373637315087e-07,
      "loss": 0.8426,
      "step": 7199
    },
    {
      "epoch": 0.89809155544468,
      "grad_norm": 0.0,
      "learning_rate": 5.398271254176502e-07,
      "loss": 0.931,
      "step": 7200
    },
    {
      "epoch": 0.8982162903829363,
      "grad_norm": 0.0,
      "learning_rate": 5.385184312424973e-07,
      "loss": 0.8968,
      "step": 7201
    },
    {
      "epoch": 0.8983410253211924,
      "grad_norm": 0.0,
      "learning_rate": 5.37211281419655e-07,
      "loss": 0.9135,
      "step": 7202
    },
    {
      "epoch": 0.8984657602594487,
      "grad_norm": 0.0,
      "learning_rate": 5.359056761624903e-07,
      "loss": 0.9114,
      "step": 7203
    },
    {
      "epoch": 0.8985904951977048,
      "grad_norm": 0.0,
      "learning_rate": 5.346016156841072e-07,
      "loss": 0.9116,
      "step": 7204
    },
    {
      "epoch": 0.8987152301359611,
      "grad_norm": 0.0,
      "learning_rate": 5.332991001973609e-07,
      "loss": 0.9268,
      "step": 7205
    },
    {
      "epoch": 0.8988399650742173,
      "grad_norm": 0.0,
      "learning_rate": 5.319981299148569e-07,
      "loss": 0.8891,
      "step": 7206
    },
    {
      "epoch": 0.8989647000124735,
      "grad_norm": 0.0,
      "learning_rate": 5.306987050489442e-07,
      "loss": 0.9017,
      "step": 7207
    },
    {
      "epoch": 0.8990894349507297,
      "grad_norm": 0.0,
      "learning_rate": 5.294008258117223e-07,
      "loss": 0.8815,
      "step": 7208
    },
    {
      "epoch": 0.899214169888986,
      "grad_norm": 0.0,
      "learning_rate": 5.281044924150391e-07,
      "loss": 0.9195,
      "step": 7209
    },
    {
      "epoch": 0.8993389048272421,
      "grad_norm": 0.0,
      "learning_rate": 5.268097050704857e-07,
      "loss": 0.8855,
      "step": 7210
    },
    {
      "epoch": 0.8994636397654984,
      "grad_norm": 0.0,
      "learning_rate": 5.255164639894084e-07,
      "loss": 0.8736,
      "step": 7211
    },
    {
      "epoch": 0.8995883747037545,
      "grad_norm": 0.0,
      "learning_rate": 5.242247693828939e-07,
      "loss": 0.9449,
      "step": 7212
    },
    {
      "epoch": 0.8997131096420107,
      "grad_norm": 0.0,
      "learning_rate": 5.229346214617792e-07,
      "loss": 0.8741,
      "step": 7213
    },
    {
      "epoch": 0.8998378445802669,
      "grad_norm": 0.0,
      "learning_rate": 5.216460204366502e-07,
      "loss": 0.8851,
      "step": 7214
    },
    {
      "epoch": 0.8999625795185231,
      "grad_norm": 0.0,
      "learning_rate": 5.203589665178399e-07,
      "loss": 0.8927,
      "step": 7215
    },
    {
      "epoch": 0.9000873144567794,
      "grad_norm": 0.0,
      "learning_rate": 5.190734599154257e-07,
      "loss": 0.919,
      "step": 7216
    },
    {
      "epoch": 0.9002120493950355,
      "grad_norm": 0.0,
      "learning_rate": 5.177895008392353e-07,
      "loss": 0.9188,
      "step": 7217
    },
    {
      "epoch": 0.9003367843332918,
      "grad_norm": 0.0,
      "learning_rate": 5.165070894988444e-07,
      "loss": 0.9608,
      "step": 7218
    },
    {
      "epoch": 0.9004615192715479,
      "grad_norm": 0.0,
      "learning_rate": 5.152262261035723e-07,
      "loss": 0.9113,
      "step": 7219
    },
    {
      "epoch": 0.9005862542098042,
      "grad_norm": 0.0,
      "learning_rate": 5.139469108624917e-07,
      "loss": 0.8857,
      "step": 7220
    },
    {
      "epoch": 0.9007109891480604,
      "grad_norm": 0.0,
      "learning_rate": 5.126691439844178e-07,
      "loss": 0.9039,
      "step": 7221
    },
    {
      "epoch": 0.9008357240863166,
      "grad_norm": 0.0,
      "learning_rate": 5.113929256779105e-07,
      "loss": 0.8615,
      "step": 7222
    },
    {
      "epoch": 0.9009604590245728,
      "grad_norm": 0.0,
      "learning_rate": 5.101182561512874e-07,
      "loss": 0.8558,
      "step": 7223
    },
    {
      "epoch": 0.9010851939628289,
      "grad_norm": 0.0,
      "learning_rate": 5.088451356126023e-07,
      "loss": 0.8726,
      "step": 7224
    },
    {
      "epoch": 0.9012099289010852,
      "grad_norm": 0.0,
      "learning_rate": 5.075735642696611e-07,
      "loss": 0.8453,
      "step": 7225
    },
    {
      "epoch": 0.9013346638393414,
      "grad_norm": 0.0,
      "learning_rate": 5.063035423300189e-07,
      "loss": 0.9495,
      "step": 7226
    },
    {
      "epoch": 0.9014593987775976,
      "grad_norm": 0.0,
      "learning_rate": 5.05035070000971e-07,
      "loss": 0.9087,
      "step": 7227
    },
    {
      "epoch": 0.9015841337158538,
      "grad_norm": 0.0,
      "learning_rate": 5.037681474895673e-07,
      "loss": 0.8933,
      "step": 7228
    },
    {
      "epoch": 0.90170886865411,
      "grad_norm": 0.0,
      "learning_rate": 5.025027750026001e-07,
      "loss": 0.9071,
      "step": 7229
    },
    {
      "epoch": 0.9018336035923662,
      "grad_norm": 0.0,
      "learning_rate": 5.012389527466111e-07,
      "loss": 0.9128,
      "step": 7230
    },
    {
      "epoch": 0.9019583385306225,
      "grad_norm": 0.0,
      "learning_rate": 4.999766809278872e-07,
      "loss": 0.8867,
      "step": 7231
    },
    {
      "epoch": 0.9020830734688786,
      "grad_norm": 0.0,
      "learning_rate": 4.987159597524649e-07,
      "loss": 0.8539,
      "step": 7232
    },
    {
      "epoch": 0.9022078084071349,
      "grad_norm": 0.0,
      "learning_rate": 4.974567894261218e-07,
      "loss": 0.8345,
      "step": 7233
    },
    {
      "epoch": 0.902332543345391,
      "grad_norm": 0.0,
      "learning_rate": 4.961991701543889e-07,
      "loss": 0.88,
      "step": 7234
    },
    {
      "epoch": 0.9024572782836473,
      "grad_norm": 0.0,
      "learning_rate": 4.949431021425422e-07,
      "loss": 0.8648,
      "step": 7235
    },
    {
      "epoch": 0.9025820132219035,
      "grad_norm": 0.0,
      "learning_rate": 4.936885855955997e-07,
      "loss": 0.8706,
      "step": 7236
    },
    {
      "epoch": 0.9027067481601596,
      "grad_norm": 0.0,
      "learning_rate": 4.924356207183356e-07,
      "loss": 0.9001,
      "step": 7237
    },
    {
      "epoch": 0.9028314830984159,
      "grad_norm": 0.0,
      "learning_rate": 4.911842077152618e-07,
      "loss": 0.9489,
      "step": 7238
    },
    {
      "epoch": 0.902956218036672,
      "grad_norm": 0.0,
      "learning_rate": 4.899343467906392e-07,
      "loss": 0.8877,
      "step": 7239
    },
    {
      "epoch": 0.9030809529749283,
      "grad_norm": 0.0,
      "learning_rate": 4.886860381484814e-07,
      "loss": 0.8613,
      "step": 7240
    },
    {
      "epoch": 0.9032056879131845,
      "grad_norm": 0.0,
      "learning_rate": 4.874392819925389e-07,
      "loss": 0.9414,
      "step": 7241
    },
    {
      "epoch": 0.9033304228514407,
      "grad_norm": 0.0,
      "learning_rate": 4.861940785263164e-07,
      "loss": 0.902,
      "step": 7242
    },
    {
      "epoch": 0.9034551577896969,
      "grad_norm": 0.0,
      "learning_rate": 4.849504279530637e-07,
      "loss": 0.8701,
      "step": 7243
    },
    {
      "epoch": 0.9035798927279531,
      "grad_norm": 0.0,
      "learning_rate": 4.837083304757717e-07,
      "loss": 0.921,
      "step": 7244
    },
    {
      "epoch": 0.9037046276662093,
      "grad_norm": 0.0,
      "learning_rate": 4.82467786297186e-07,
      "loss": 0.8491,
      "step": 7245
    },
    {
      "epoch": 0.9038293626044656,
      "grad_norm": 0.0,
      "learning_rate": 4.812287956197925e-07,
      "loss": 0.8865,
      "step": 7246
    },
    {
      "epoch": 0.9039540975427217,
      "grad_norm": 0.0,
      "learning_rate": 4.799913586458271e-07,
      "loss": 0.921,
      "step": 7247
    },
    {
      "epoch": 0.9040788324809779,
      "grad_norm": 0.0,
      "learning_rate": 4.787554755772683e-07,
      "loss": 0.8605,
      "step": 7248
    },
    {
      "epoch": 0.9042035674192341,
      "grad_norm": 0.0,
      "learning_rate": 4.775211466158469e-07,
      "loss": 0.884,
      "step": 7249
    },
    {
      "epoch": 0.9043283023574903,
      "grad_norm": 0.0,
      "learning_rate": 4.76288371963034e-07,
      "loss": 0.9059,
      "step": 7250
    },
    {
      "epoch": 0.9044530372957466,
      "grad_norm": 0.0,
      "learning_rate": 4.750571518200509e-07,
      "loss": 0.8863,
      "step": 7251
    },
    {
      "epoch": 0.9045777722340027,
      "grad_norm": 0.0,
      "learning_rate": 4.7382748638786336e-07,
      "loss": 0.8797,
      "step": 7252
    },
    {
      "epoch": 0.904702507172259,
      "grad_norm": 0.0,
      "learning_rate": 4.7259937586718205e-07,
      "loss": 0.8966,
      "step": 7253
    },
    {
      "epoch": 0.9048272421105151,
      "grad_norm": 0.0,
      "learning_rate": 4.713728204584678e-07,
      "loss": 0.9068,
      "step": 7254
    },
    {
      "epoch": 0.9049519770487714,
      "grad_norm": 0.0,
      "learning_rate": 4.7014782036192473e-07,
      "loss": 0.8811,
      "step": 7255
    },
    {
      "epoch": 0.9050767119870275,
      "grad_norm": 0.0,
      "learning_rate": 4.6892437577750195e-07,
      "loss": 0.946,
      "step": 7256
    },
    {
      "epoch": 0.9052014469252838,
      "grad_norm": 0.0,
      "learning_rate": 4.677024869049007e-07,
      "loss": 0.9249,
      "step": 7257
    },
    {
      "epoch": 0.90532618186354,
      "grad_norm": 0.0,
      "learning_rate": 4.6648215394355935e-07,
      "loss": 0.8559,
      "step": 7258
    },
    {
      "epoch": 0.9054509168017962,
      "grad_norm": 0.0,
      "learning_rate": 4.652633770926696e-07,
      "loss": 0.9264,
      "step": 7259
    },
    {
      "epoch": 0.9055756517400524,
      "grad_norm": 0.0,
      "learning_rate": 4.640461565511678e-07,
      "loss": 0.9101,
      "step": 7260
    },
    {
      "epoch": 0.9057003866783085,
      "grad_norm": 0.0,
      "learning_rate": 4.628304925177318e-07,
      "loss": 0.8991,
      "step": 7261
    },
    {
      "epoch": 0.9058251216165648,
      "grad_norm": 0.0,
      "learning_rate": 4.6161638519078956e-07,
      "loss": 0.8572,
      "step": 7262
    },
    {
      "epoch": 0.905949856554821,
      "grad_norm": 0.0,
      "learning_rate": 4.604038347685147e-07,
      "loss": 0.8777,
      "step": 7263
    },
    {
      "epoch": 0.9060745914930772,
      "grad_norm": 0.0,
      "learning_rate": 4.591928414488278e-07,
      "loss": 0.8836,
      "step": 7264
    },
    {
      "epoch": 0.9061993264313334,
      "grad_norm": 0.0,
      "learning_rate": 4.579834054293886e-07,
      "loss": 0.9031,
      "step": 7265
    },
    {
      "epoch": 0.9063240613695897,
      "grad_norm": 0.0,
      "learning_rate": 4.567755269076124e-07,
      "loss": 0.8895,
      "step": 7266
    },
    {
      "epoch": 0.9064487963078458,
      "grad_norm": 0.0,
      "learning_rate": 4.555692060806527e-07,
      "loss": 0.8654,
      "step": 7267
    },
    {
      "epoch": 0.9065735312461021,
      "grad_norm": 0.0,
      "learning_rate": 4.543644431454131e-07,
      "loss": 0.8792,
      "step": 7268
    },
    {
      "epoch": 0.9066982661843582,
      "grad_norm": 0.0,
      "learning_rate": 4.5316123829854063e-07,
      "loss": 0.945,
      "step": 7269
    },
    {
      "epoch": 0.9068230011226145,
      "grad_norm": 0.0,
      "learning_rate": 4.519595917364272e-07,
      "loss": 0.911,
      "step": 7270
    },
    {
      "epoch": 0.9069477360608706,
      "grad_norm": 0.0,
      "learning_rate": 4.507595036552137e-07,
      "loss": 0.9051,
      "step": 7271
    },
    {
      "epoch": 0.9070724709991268,
      "grad_norm": 0.0,
      "learning_rate": 4.495609742507856e-07,
      "loss": 0.8828,
      "step": 7272
    },
    {
      "epoch": 0.9071972059373831,
      "grad_norm": 0.0,
      "learning_rate": 4.4836400371876975e-07,
      "loss": 0.8941,
      "step": 7273
    },
    {
      "epoch": 0.9073219408756392,
      "grad_norm": 0.0,
      "learning_rate": 4.4716859225454433e-07,
      "loss": 0.8965,
      "step": 7274
    },
    {
      "epoch": 0.9074466758138955,
      "grad_norm": 0.0,
      "learning_rate": 4.4597474005323105e-07,
      "loss": 0.8671,
      "step": 7275
    },
    {
      "epoch": 0.9075714107521516,
      "grad_norm": 0.0,
      "learning_rate": 4.44782447309694e-07,
      "loss": 0.9482,
      "step": 7276
    },
    {
      "epoch": 0.9076961456904079,
      "grad_norm": 0.0,
      "learning_rate": 4.4359171421854973e-07,
      "loss": 0.8912,
      "step": 7277
    },
    {
      "epoch": 0.9078208806286641,
      "grad_norm": 0.0,
      "learning_rate": 4.424025409741517e-07,
      "loss": 0.8995,
      "step": 7278
    },
    {
      "epoch": 0.9079456155669203,
      "grad_norm": 0.0,
      "learning_rate": 4.412149277706046e-07,
      "loss": 0.8982,
      "step": 7279
    },
    {
      "epoch": 0.9080703505051765,
      "grad_norm": 0.0,
      "learning_rate": 4.400288748017578e-07,
      "loss": 0.9225,
      "step": 7280
    },
    {
      "epoch": 0.9081950854434327,
      "grad_norm": 0.0,
      "learning_rate": 4.388443822612043e-07,
      "loss": 0.8503,
      "step": 7281
    },
    {
      "epoch": 0.9083198203816889,
      "grad_norm": 0.0,
      "learning_rate": 4.376614503422805e-07,
      "loss": 0.8687,
      "step": 7282
    },
    {
      "epoch": 0.9084445553199452,
      "grad_norm": 0.0,
      "learning_rate": 4.364800792380763e-07,
      "loss": 0.9048,
      "step": 7283
    },
    {
      "epoch": 0.9085692902582013,
      "grad_norm": 0.0,
      "learning_rate": 4.3530026914141767e-07,
      "loss": 0.8618,
      "step": 7284
    },
    {
      "epoch": 0.9086940251964575,
      "grad_norm": 0.0,
      "learning_rate": 4.3412202024487706e-07,
      "loss": 0.9358,
      "step": 7285
    },
    {
      "epoch": 0.9088187601347137,
      "grad_norm": 0.0,
      "learning_rate": 4.3294533274078063e-07,
      "loss": 0.8739,
      "step": 7286
    },
    {
      "epoch": 0.9089434950729699,
      "grad_norm": 0.0,
      "learning_rate": 4.317702068211882e-07,
      "loss": 0.8642,
      "step": 7287
    },
    {
      "epoch": 0.9090682300112262,
      "grad_norm": 0.0,
      "learning_rate": 4.305966426779118e-07,
      "loss": 0.8918,
      "step": 7288
    },
    {
      "epoch": 0.9091929649494823,
      "grad_norm": 0.0,
      "learning_rate": 4.294246405025071e-07,
      "loss": 0.8761,
      "step": 7289
    },
    {
      "epoch": 0.9093176998877386,
      "grad_norm": 0.0,
      "learning_rate": 4.2825420048627333e-07,
      "loss": 0.8589,
      "step": 7290
    },
    {
      "epoch": 0.9094424348259947,
      "grad_norm": 0.0,
      "learning_rate": 4.270853228202565e-07,
      "loss": 0.9139,
      "step": 7291
    },
    {
      "epoch": 0.909567169764251,
      "grad_norm": 0.0,
      "learning_rate": 4.2591800769524625e-07,
      "loss": 0.8803,
      "step": 7292
    },
    {
      "epoch": 0.9096919047025072,
      "grad_norm": 0.0,
      "learning_rate": 4.247522553017791e-07,
      "loss": 0.8846,
      "step": 7293
    },
    {
      "epoch": 0.9098166396407634,
      "grad_norm": 0.0,
      "learning_rate": 4.2358806583013503e-07,
      "loss": 0.8802,
      "step": 7294
    },
    {
      "epoch": 0.9099413745790196,
      "grad_norm": 0.0,
      "learning_rate": 4.2242543947033863e-07,
      "loss": 0.8655,
      "step": 7295
    },
    {
      "epoch": 0.9100661095172757,
      "grad_norm": 0.0,
      "learning_rate": 4.2126437641215823e-07,
      "loss": 0.926,
      "step": 7296
    },
    {
      "epoch": 0.910190844455532,
      "grad_norm": 0.0,
      "learning_rate": 4.2010487684511105e-07,
      "loss": 0.906,
      "step": 7297
    },
    {
      "epoch": 0.9103155793937882,
      "grad_norm": 0.0,
      "learning_rate": 4.189469409584568e-07,
      "loss": 0.9212,
      "step": 7298
    },
    {
      "epoch": 0.9104403143320444,
      "grad_norm": 0.0,
      "learning_rate": 4.1779056894119763e-07,
      "loss": 0.9373,
      "step": 7299
    },
    {
      "epoch": 0.9105650492703006,
      "grad_norm": 0.0,
      "learning_rate": 4.166357609820837e-07,
      "loss": 0.8797,
      "step": 7300
    },
    {
      "epoch": 0.9106897842085568,
      "grad_norm": 0.0,
      "learning_rate": 4.154825172696098e-07,
      "loss": 0.9436,
      "step": 7301
    },
    {
      "epoch": 0.910814519146813,
      "grad_norm": 0.0,
      "learning_rate": 4.143308379920108e-07,
      "loss": 0.9297,
      "step": 7302
    },
    {
      "epoch": 0.9109392540850693,
      "grad_norm": 0.0,
      "learning_rate": 4.131807233372753e-07,
      "loss": 0.9129,
      "step": 7303
    },
    {
      "epoch": 0.9110639890233254,
      "grad_norm": 0.0,
      "learning_rate": 4.120321734931265e-07,
      "loss": 0.9704,
      "step": 7304
    },
    {
      "epoch": 0.9111887239615817,
      "grad_norm": 0.0,
      "learning_rate": 4.1088518864703884e-07,
      "loss": 0.8871,
      "step": 7305
    },
    {
      "epoch": 0.9113134588998378,
      "grad_norm": 0.0,
      "learning_rate": 4.0973976898622923e-07,
      "loss": 0.8868,
      "step": 7306
    },
    {
      "epoch": 0.9114381938380941,
      "grad_norm": 0.0,
      "learning_rate": 4.0859591469765704e-07,
      "loss": 0.8702,
      "step": 7307
    },
    {
      "epoch": 0.9115629287763503,
      "grad_norm": 0.0,
      "learning_rate": 4.0745362596803063e-07,
      "loss": 0.9075,
      "step": 7308
    },
    {
      "epoch": 0.9116876637146064,
      "grad_norm": 0.0,
      "learning_rate": 4.063129029837987e-07,
      "loss": 0.9223,
      "step": 7309
    },
    {
      "epoch": 0.9118123986528627,
      "grad_norm": 0.0,
      "learning_rate": 4.0517374593115666e-07,
      "loss": 0.8843,
      "step": 7310
    },
    {
      "epoch": 0.9119371335911188,
      "grad_norm": 0.0,
      "learning_rate": 4.0403615499604365e-07,
      "loss": 0.8915,
      "step": 7311
    },
    {
      "epoch": 0.9120618685293751,
      "grad_norm": 0.0,
      "learning_rate": 4.029001303641422e-07,
      "loss": 0.8863,
      "step": 7312
    },
    {
      "epoch": 0.9121866034676313,
      "grad_norm": 0.0,
      "learning_rate": 4.017656722208807e-07,
      "loss": 0.8587,
      "step": 7313
    },
    {
      "epoch": 0.9123113384058875,
      "grad_norm": 0.0,
      "learning_rate": 4.006327807514321e-07,
      "loss": 0.93,
      "step": 7314
    },
    {
      "epoch": 0.9124360733441437,
      "grad_norm": 0.0,
      "learning_rate": 3.99501456140714e-07,
      "loss": 0.9696,
      "step": 7315
    },
    {
      "epoch": 0.9125608082823999,
      "grad_norm": 0.0,
      "learning_rate": 3.983716985733832e-07,
      "loss": 0.8744,
      "step": 7316
    },
    {
      "epoch": 0.9126855432206561,
      "grad_norm": 0.0,
      "learning_rate": 3.9724350823384774e-07,
      "loss": 0.9268,
      "step": 7317
    },
    {
      "epoch": 0.9128102781589124,
      "grad_norm": 0.0,
      "learning_rate": 3.961168853062569e-07,
      "loss": 0.8998,
      "step": 7318
    },
    {
      "epoch": 0.9129350130971685,
      "grad_norm": 0.0,
      "learning_rate": 3.949918299745004e-07,
      "loss": 0.9351,
      "step": 7319
    },
    {
      "epoch": 0.9130597480354247,
      "grad_norm": 0.0,
      "learning_rate": 3.9386834242222114e-07,
      "loss": 0.9033,
      "step": 7320
    },
    {
      "epoch": 0.9131844829736809,
      "grad_norm": 0.0,
      "learning_rate": 3.9274642283279486e-07,
      "loss": 0.9036,
      "step": 7321
    },
    {
      "epoch": 0.9133092179119371,
      "grad_norm": 0.0,
      "learning_rate": 3.916260713893516e-07,
      "loss": 0.8771,
      "step": 7322
    },
    {
      "epoch": 0.9134339528501934,
      "grad_norm": 0.0,
      "learning_rate": 3.9050728827475957e-07,
      "loss": 0.8732,
      "step": 7323
    },
    {
      "epoch": 0.9135586877884495,
      "grad_norm": 0.0,
      "learning_rate": 3.893900736716305e-07,
      "loss": 0.8968,
      "step": 7324
    },
    {
      "epoch": 0.9136834227267058,
      "grad_norm": 0.0,
      "learning_rate": 3.8827442776232403e-07,
      "loss": 0.9269,
      "step": 7325
    },
    {
      "epoch": 0.9138081576649619,
      "grad_norm": 0.0,
      "learning_rate": 3.871603507289412e-07,
      "loss": 0.9112,
      "step": 7326
    },
    {
      "epoch": 0.9139328926032182,
      "grad_norm": 0.0,
      "learning_rate": 3.8604784275332765e-07,
      "loss": 0.8995,
      "step": 7327
    },
    {
      "epoch": 0.9140576275414743,
      "grad_norm": 0.0,
      "learning_rate": 3.8493690401707253e-07,
      "loss": 0.9161,
      "step": 7328
    },
    {
      "epoch": 0.9141823624797306,
      "grad_norm": 0.0,
      "learning_rate": 3.838275347015086e-07,
      "loss": 0.8739,
      "step": 7329
    },
    {
      "epoch": 0.9143070974179868,
      "grad_norm": 0.0,
      "learning_rate": 3.827197349877143e-07,
      "loss": 0.8893,
      "step": 7330
    },
    {
      "epoch": 0.914431832356243,
      "grad_norm": 0.0,
      "learning_rate": 3.816135050565073e-07,
      "loss": 0.8423,
      "step": 7331
    },
    {
      "epoch": 0.9145565672944992,
      "grad_norm": 0.0,
      "learning_rate": 3.8050884508845756e-07,
      "loss": 0.8688,
      "step": 7332
    },
    {
      "epoch": 0.9146813022327553,
      "grad_norm": 0.0,
      "learning_rate": 3.794057552638686e-07,
      "loss": 0.8632,
      "step": 7333
    },
    {
      "epoch": 0.9148060371710116,
      "grad_norm": 0.0,
      "learning_rate": 3.783042357627942e-07,
      "loss": 0.8979,
      "step": 7334
    },
    {
      "epoch": 0.9149307721092678,
      "grad_norm": 0.0,
      "learning_rate": 3.772042867650305e-07,
      "loss": 0.9351,
      "step": 7335
    },
    {
      "epoch": 0.915055507047524,
      "grad_norm": 0.0,
      "learning_rate": 3.761059084501162e-07,
      "loss": 0.8753,
      "step": 7336
    },
    {
      "epoch": 0.9151802419857802,
      "grad_norm": 0.0,
      "learning_rate": 3.7500910099733443e-07,
      "loss": 0.8898,
      "step": 7337
    },
    {
      "epoch": 0.9153049769240365,
      "grad_norm": 0.0,
      "learning_rate": 3.739138645857121e-07,
      "loss": 0.8834,
      "step": 7338
    },
    {
      "epoch": 0.9154297118622926,
      "grad_norm": 0.0,
      "learning_rate": 3.7282019939401835e-07,
      "loss": 0.9774,
      "step": 7339
    },
    {
      "epoch": 0.9155544468005489,
      "grad_norm": 0.0,
      "learning_rate": 3.717281056007704e-07,
      "loss": 0.9196,
      "step": 7340
    },
    {
      "epoch": 0.915679181738805,
      "grad_norm": 0.0,
      "learning_rate": 3.706375833842213e-07,
      "loss": 0.902,
      "step": 7341
    },
    {
      "epoch": 0.9158039166770613,
      "grad_norm": 0.0,
      "learning_rate": 3.6954863292237297e-07,
      "loss": 0.9557,
      "step": 7342
    },
    {
      "epoch": 0.9159286516153174,
      "grad_norm": 0.0,
      "learning_rate": 3.684612543929711e-07,
      "loss": 0.8764,
      "step": 7343
    },
    {
      "epoch": 0.9160533865535736,
      "grad_norm": 0.0,
      "learning_rate": 3.6737544797350254e-07,
      "loss": 0.8967,
      "step": 7344
    },
    {
      "epoch": 0.9161781214918299,
      "grad_norm": 0.0,
      "learning_rate": 3.662912138411967e-07,
      "loss": 0.8926,
      "step": 7345
    },
    {
      "epoch": 0.916302856430086,
      "grad_norm": 0.0,
      "learning_rate": 3.6520855217302973e-07,
      "loss": 0.9201,
      "step": 7346
    },
    {
      "epoch": 0.9164275913683423,
      "grad_norm": 0.0,
      "learning_rate": 3.6412746314572035e-07,
      "loss": 0.8884,
      "step": 7347
    },
    {
      "epoch": 0.9165523263065984,
      "grad_norm": 0.0,
      "learning_rate": 3.630479469357251e-07,
      "loss": 0.8925,
      "step": 7348
    },
    {
      "epoch": 0.9166770612448547,
      "grad_norm": 0.0,
      "learning_rate": 3.6197000371925307e-07,
      "loss": 0.8872,
      "step": 7349
    },
    {
      "epoch": 0.9168017961831109,
      "grad_norm": 0.0,
      "learning_rate": 3.6089363367224907e-07,
      "loss": 0.8595,
      "step": 7350
    },
    {
      "epoch": 0.9169265311213671,
      "grad_norm": 0.0,
      "learning_rate": 3.5981883697040363e-07,
      "loss": 0.872,
      "step": 7351
    },
    {
      "epoch": 0.9170512660596233,
      "grad_norm": 0.0,
      "learning_rate": 3.587456137891532e-07,
      "loss": 0.9063,
      "step": 7352
    },
    {
      "epoch": 0.9171760009978795,
      "grad_norm": 0.0,
      "learning_rate": 3.576739643036731e-07,
      "loss": 0.8968,
      "step": 7353
    },
    {
      "epoch": 0.9173007359361357,
      "grad_norm": 0.0,
      "learning_rate": 3.566038886888823e-07,
      "loss": 0.8876,
      "step": 7354
    },
    {
      "epoch": 0.917425470874392,
      "grad_norm": 0.0,
      "learning_rate": 3.555353871194467e-07,
      "loss": 0.8938,
      "step": 7355
    },
    {
      "epoch": 0.9175502058126481,
      "grad_norm": 0.0,
      "learning_rate": 3.544684597697723e-07,
      "loss": 0.8969,
      "step": 7356
    },
    {
      "epoch": 0.9176749407509043,
      "grad_norm": 0.0,
      "learning_rate": 3.5340310681400646e-07,
      "loss": 0.923,
      "step": 7357
    },
    {
      "epoch": 0.9177996756891605,
      "grad_norm": 0.0,
      "learning_rate": 3.5233932842604237e-07,
      "loss": 0.8867,
      "step": 7358
    },
    {
      "epoch": 0.9179244106274167,
      "grad_norm": 0.0,
      "learning_rate": 3.512771247795166e-07,
      "loss": 0.9104,
      "step": 7359
    },
    {
      "epoch": 0.918049145565673,
      "grad_norm": 0.0,
      "learning_rate": 3.5021649604780714e-07,
      "loss": 0.9001,
      "step": 7360
    },
    {
      "epoch": 0.9181738805039291,
      "grad_norm": 0.0,
      "learning_rate": 3.491574424040356e-07,
      "loss": 0.8423,
      "step": 7361
    },
    {
      "epoch": 0.9182986154421854,
      "grad_norm": 0.0,
      "learning_rate": 3.4809996402106473e-07,
      "loss": 0.8892,
      "step": 7362
    },
    {
      "epoch": 0.9184233503804415,
      "grad_norm": 0.0,
      "learning_rate": 3.470440610715031e-07,
      "loss": 0.8957,
      "step": 7363
    },
    {
      "epoch": 0.9185480853186978,
      "grad_norm": 0.0,
      "learning_rate": 3.4598973372770186e-07,
      "loss": 0.9158,
      "step": 7364
    },
    {
      "epoch": 0.918672820256954,
      "grad_norm": 0.0,
      "learning_rate": 3.4493698216174986e-07,
      "loss": 0.9003,
      "step": 7365
    },
    {
      "epoch": 0.9187975551952102,
      "grad_norm": 0.0,
      "learning_rate": 3.4388580654548755e-07,
      "loss": 0.8386,
      "step": 7366
    },
    {
      "epoch": 0.9189222901334664,
      "grad_norm": 0.0,
      "learning_rate": 3.428362070504887e-07,
      "loss": 0.8474,
      "step": 7367
    },
    {
      "epoch": 0.9190470250717225,
      "grad_norm": 0.0,
      "learning_rate": 3.4178818384807746e-07,
      "loss": 0.8877,
      "step": 7368
    },
    {
      "epoch": 0.9191717600099788,
      "grad_norm": 0.0,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 0.8841,
      "step": 7369
    },
    {
      "epoch": 0.919296494948235,
      "grad_norm": 0.0,
      "learning_rate": 3.396968670050138e-07,
      "loss": 0.9,
      "step": 7370
    },
    {
      "epoch": 0.9194212298864912,
      "grad_norm": 0.0,
      "learning_rate": 3.3865357370571616e-07,
      "loss": 0.8675,
      "step": 7371
    },
    {
      "epoch": 0.9195459648247474,
      "grad_norm": 0.0,
      "learning_rate": 3.3761185738171665e-07,
      "loss": 0.8677,
      "step": 7372
    },
    {
      "epoch": 0.9196706997630036,
      "grad_norm": 0.0,
      "learning_rate": 3.3657171820305035e-07,
      "loss": 0.899,
      "step": 7373
    },
    {
      "epoch": 0.9197954347012598,
      "grad_norm": 0.0,
      "learning_rate": 3.3553315633949144e-07,
      "loss": 0.9082,
      "step": 7374
    },
    {
      "epoch": 0.9199201696395161,
      "grad_norm": 0.0,
      "learning_rate": 3.3449617196056217e-07,
      "loss": 0.842,
      "step": 7375
    },
    {
      "epoch": 0.9200449045777722,
      "grad_norm": 0.0,
      "learning_rate": 3.3346076523552486e-07,
      "loss": 0.8963,
      "step": 7376
    },
    {
      "epoch": 0.9201696395160285,
      "grad_norm": 0.0,
      "learning_rate": 3.3242693633337986e-07,
      "loss": 0.8884,
      "step": 7377
    },
    {
      "epoch": 0.9202943744542846,
      "grad_norm": 0.0,
      "learning_rate": 3.3139468542288e-07,
      "loss": 0.9053,
      "step": 7378
    },
    {
      "epoch": 0.9204191093925409,
      "grad_norm": 0.0,
      "learning_rate": 3.303640126725094e-07,
      "loss": 0.8664,
      "step": 7379
    },
    {
      "epoch": 0.9205438443307971,
      "grad_norm": 0.0,
      "learning_rate": 3.2933491825050237e-07,
      "loss": 0.8656,
      "step": 7380
    },
    {
      "epoch": 0.9206685792690532,
      "grad_norm": 0.0,
      "learning_rate": 3.283074023248345e-07,
      "loss": 0.9353,
      "step": 7381
    },
    {
      "epoch": 0.9207933142073095,
      "grad_norm": 0.0,
      "learning_rate": 3.2728146506321944e-07,
      "loss": 0.9221,
      "step": 7382
    },
    {
      "epoch": 0.9209180491455656,
      "grad_norm": 0.0,
      "learning_rate": 3.262571066331166e-07,
      "loss": 0.8847,
      "step": 7383
    },
    {
      "epoch": 0.9210427840838219,
      "grad_norm": 0.0,
      "learning_rate": 3.2523432720172776e-07,
      "loss": 0.8498,
      "step": 7384
    },
    {
      "epoch": 0.921167519022078,
      "grad_norm": 0.0,
      "learning_rate": 3.2421312693599605e-07,
      "loss": 0.9386,
      "step": 7385
    },
    {
      "epoch": 0.9212922539603343,
      "grad_norm": 0.0,
      "learning_rate": 3.231935060026092e-07,
      "loss": 0.8934,
      "step": 7386
    },
    {
      "epoch": 0.9214169888985905,
      "grad_norm": 0.0,
      "learning_rate": 3.2217546456799086e-07,
      "loss": 0.8891,
      "step": 7387
    },
    {
      "epoch": 0.9215417238368467,
      "grad_norm": 0.0,
      "learning_rate": 3.211590027983136e-07,
      "loss": 0.8851,
      "step": 7388
    },
    {
      "epoch": 0.9216664587751029,
      "grad_norm": 0.0,
      "learning_rate": 3.201441208594891e-07,
      "loss": 0.9241,
      "step": 7389
    },
    {
      "epoch": 0.9217911937133592,
      "grad_norm": 0.0,
      "learning_rate": 3.191308189171738e-07,
      "loss": 0.8694,
      "step": 7390
    },
    {
      "epoch": 0.9219159286516153,
      "grad_norm": 0.0,
      "learning_rate": 3.1811909713676093e-07,
      "loss": 0.9128,
      "step": 7391
    },
    {
      "epoch": 0.9220406635898715,
      "grad_norm": 0.0,
      "learning_rate": 3.171089556833895e-07,
      "loss": 0.8938,
      "step": 7392
    },
    {
      "epoch": 0.9221653985281277,
      "grad_norm": 0.0,
      "learning_rate": 3.161003947219421e-07,
      "loss": 0.9072,
      "step": 7393
    },
    {
      "epoch": 0.9222901334663839,
      "grad_norm": 0.0,
      "learning_rate": 3.150934144170381e-07,
      "loss": 0.8745,
      "step": 7394
    },
    {
      "epoch": 0.9224148684046402,
      "grad_norm": 0.0,
      "learning_rate": 3.140880149330461e-07,
      "loss": 0.8937,
      "step": 7395
    },
    {
      "epoch": 0.9225396033428963,
      "grad_norm": 0.0,
      "learning_rate": 3.1308419643406915e-07,
      "loss": 0.8846,
      "step": 7396
    },
    {
      "epoch": 0.9226643382811526,
      "grad_norm": 0.0,
      "learning_rate": 3.1208195908395747e-07,
      "loss": 0.9338,
      "step": 7397
    },
    {
      "epoch": 0.9227890732194087,
      "grad_norm": 0.0,
      "learning_rate": 3.1108130304630223e-07,
      "loss": 0.8792,
      "step": 7398
    },
    {
      "epoch": 0.922913808157665,
      "grad_norm": 0.0,
      "learning_rate": 3.1008222848443293e-07,
      "loss": 0.931,
      "step": 7399
    },
    {
      "epoch": 0.9230385430959211,
      "grad_norm": 0.0,
      "learning_rate": 3.090847355614257e-07,
      "loss": 0.9059,
      "step": 7400
    },
    {
      "epoch": 0.9231632780341774,
      "grad_norm": 0.0,
      "learning_rate": 3.080888244400959e-07,
      "loss": 0.9079,
      "step": 7401
    },
    {
      "epoch": 0.9232880129724336,
      "grad_norm": 0.0,
      "learning_rate": 3.070944952830035e-07,
      "loss": 0.9114,
      "step": 7402
    },
    {
      "epoch": 0.9234127479106898,
      "grad_norm": 0.0,
      "learning_rate": 3.061017482524442e-07,
      "loss": 0.9354,
      "step": 7403
    },
    {
      "epoch": 0.923537482848946,
      "grad_norm": 0.0,
      "learning_rate": 3.051105835104629e-07,
      "loss": 0.9345,
      "step": 7404
    },
    {
      "epoch": 0.9236622177872021,
      "grad_norm": 0.0,
      "learning_rate": 3.0412100121884005e-07,
      "loss": 0.8351,
      "step": 7405
    },
    {
      "epoch": 0.9237869527254584,
      "grad_norm": 0.0,
      "learning_rate": 3.0313300153910097e-07,
      "loss": 0.9563,
      "step": 7406
    },
    {
      "epoch": 0.9239116876637146,
      "grad_norm": 0.0,
      "learning_rate": 3.0214658463251443e-07,
      "loss": 0.871,
      "step": 7407
    },
    {
      "epoch": 0.9240364226019708,
      "grad_norm": 0.0,
      "learning_rate": 3.0116175066008613e-07,
      "loss": 0.8778,
      "step": 7408
    },
    {
      "epoch": 0.924161157540227,
      "grad_norm": 0.0,
      "learning_rate": 3.001784997825652e-07,
      "loss": 0.895,
      "step": 7409
    },
    {
      "epoch": 0.9242858924784833,
      "grad_norm": 0.0,
      "learning_rate": 2.991968321604466e-07,
      "loss": 0.8792,
      "step": 7410
    },
    {
      "epoch": 0.9244106274167394,
      "grad_norm": 0.0,
      "learning_rate": 2.982167479539577e-07,
      "loss": 0.9286,
      "step": 7411
    },
    {
      "epoch": 0.9245353623549957,
      "grad_norm": 0.0,
      "learning_rate": 2.9723824732307946e-07,
      "loss": 0.8801,
      "step": 7412
    },
    {
      "epoch": 0.9246600972932518,
      "grad_norm": 0.0,
      "learning_rate": 2.9626133042752303e-07,
      "loss": 0.8526,
      "step": 7413
    },
    {
      "epoch": 0.9247848322315081,
      "grad_norm": 0.0,
      "learning_rate": 2.9528599742674635e-07,
      "loss": 0.9459,
      "step": 7414
    },
    {
      "epoch": 0.9249095671697642,
      "grad_norm": 0.0,
      "learning_rate": 2.94312248479951e-07,
      "loss": 0.87,
      "step": 7415
    },
    {
      "epoch": 0.9250343021080205,
      "grad_norm": 0.0,
      "learning_rate": 2.9334008374607423e-07,
      "loss": 0.914,
      "step": 7416
    },
    {
      "epoch": 0.9251590370462767,
      "grad_norm": 0.0,
      "learning_rate": 2.9236950338380033e-07,
      "loss": 0.9526,
      "step": 7417
    },
    {
      "epoch": 0.9252837719845328,
      "grad_norm": 0.0,
      "learning_rate": 2.9140050755155134e-07,
      "loss": 0.8616,
      "step": 7418
    },
    {
      "epoch": 0.9254085069227891,
      "grad_norm": 0.0,
      "learning_rate": 2.9043309640749193e-07,
      "loss": 0.8633,
      "step": 7419
    },
    {
      "epoch": 0.9255332418610452,
      "grad_norm": 0.0,
      "learning_rate": 2.894672701095269e-07,
      "loss": 0.8667,
      "step": 7420
    },
    {
      "epoch": 0.9256579767993015,
      "grad_norm": 0.0,
      "learning_rate": 2.885030288153057e-07,
      "loss": 0.9011,
      "step": 7421
    },
    {
      "epoch": 0.9257827117375577,
      "grad_norm": 0.0,
      "learning_rate": 2.875403726822146e-07,
      "loss": 0.884,
      "step": 7422
    },
    {
      "epoch": 0.9259074466758139,
      "grad_norm": 0.0,
      "learning_rate": 2.865793018673857e-07,
      "loss": 0.8934,
      "step": 7423
    },
    {
      "epoch": 0.9260321816140701,
      "grad_norm": 0.0,
      "learning_rate": 2.85619816527688e-07,
      "loss": 0.8621,
      "step": 7424
    },
    {
      "epoch": 0.9261569165523263,
      "grad_norm": 0.0,
      "learning_rate": 2.846619168197351e-07,
      "loss": 0.8823,
      "step": 7425
    },
    {
      "epoch": 0.9262816514905825,
      "grad_norm": 0.0,
      "learning_rate": 2.8370560289987856e-07,
      "loss": 0.8418,
      "step": 7426
    },
    {
      "epoch": 0.9264063864288388,
      "grad_norm": 0.0,
      "learning_rate": 2.827508749242147e-07,
      "loss": 0.8895,
      "step": 7427
    },
    {
      "epoch": 0.9265311213670949,
      "grad_norm": 0.0,
      "learning_rate": 2.817977330485788e-07,
      "loss": 0.8996,
      "step": 7428
    },
    {
      "epoch": 0.9266558563053511,
      "grad_norm": 0.0,
      "learning_rate": 2.8084617742854645e-07,
      "loss": 0.9208,
      "step": 7429
    },
    {
      "epoch": 0.9267805912436073,
      "grad_norm": 0.0,
      "learning_rate": 2.798962082194379e-07,
      "loss": 0.9421,
      "step": 7430
    },
    {
      "epoch": 0.9269053261818635,
      "grad_norm": 0.0,
      "learning_rate": 2.789478255763078e-07,
      "loss": 0.8318,
      "step": 7431
    },
    {
      "epoch": 0.9270300611201198,
      "grad_norm": 0.0,
      "learning_rate": 2.780010296539615e-07,
      "loss": 0.9212,
      "step": 7432
    },
    {
      "epoch": 0.9271547960583759,
      "grad_norm": 0.0,
      "learning_rate": 2.7705582060693627e-07,
      "loss": 0.9186,
      "step": 7433
    },
    {
      "epoch": 0.9272795309966322,
      "grad_norm": 0.0,
      "learning_rate": 2.7611219858951543e-07,
      "loss": 0.8703,
      "step": 7434
    },
    {
      "epoch": 0.9274042659348883,
      "grad_norm": 0.0,
      "learning_rate": 2.7517016375572137e-07,
      "loss": 0.9037,
      "step": 7435
    },
    {
      "epoch": 0.9275290008731446,
      "grad_norm": 0.0,
      "learning_rate": 2.7422971625931884e-07,
      "loss": 0.8975,
      "step": 7436
    },
    {
      "epoch": 0.9276537358114008,
      "grad_norm": 0.0,
      "learning_rate": 2.732908562538106e-07,
      "loss": 0.8994,
      "step": 7437
    },
    {
      "epoch": 0.927778470749657,
      "grad_norm": 0.0,
      "learning_rate": 2.72353583892444e-07,
      "loss": 0.8952,
      "step": 7438
    },
    {
      "epoch": 0.9279032056879132,
      "grad_norm": 0.0,
      "learning_rate": 2.714178993282068e-07,
      "loss": 0.8638,
      "step": 7439
    },
    {
      "epoch": 0.9280279406261694,
      "grad_norm": 0.0,
      "learning_rate": 2.7048380271382124e-07,
      "loss": 0.8724,
      "step": 7440
    },
    {
      "epoch": 0.9281526755644256,
      "grad_norm": 0.0,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 0.9504,
      "step": 7441
    },
    {
      "epoch": 0.9282774105026818,
      "grad_norm": 0.0,
      "learning_rate": 2.686203739442328e-07,
      "loss": 0.8856,
      "step": 7442
    },
    {
      "epoch": 0.928402145440938,
      "grad_norm": 0.0,
      "learning_rate": 2.6769104209318665e-07,
      "loss": 0.8873,
      "step": 7443
    },
    {
      "epoch": 0.9285268803791942,
      "grad_norm": 0.0,
      "learning_rate": 2.6676329880031213e-07,
      "loss": 0.9414,
      "step": 7444
    },
    {
      "epoch": 0.9286516153174504,
      "grad_norm": 0.0,
      "learning_rate": 2.658371442170404e-07,
      "loss": 0.8791,
      "step": 7445
    },
    {
      "epoch": 0.9287763502557066,
      "grad_norm": 0.0,
      "learning_rate": 2.649125784945439e-07,
      "loss": 0.8711,
      "step": 7446
    },
    {
      "epoch": 0.9289010851939629,
      "grad_norm": 0.0,
      "learning_rate": 2.6398960178373513e-07,
      "loss": 0.8688,
      "step": 7447
    },
    {
      "epoch": 0.929025820132219,
      "grad_norm": 0.0,
      "learning_rate": 2.6306821423526474e-07,
      "loss": 0.8938,
      "step": 7448
    },
    {
      "epoch": 0.9291505550704753,
      "grad_norm": 0.0,
      "learning_rate": 2.6214841599953024e-07,
      "loss": 0.9234,
      "step": 7449
    },
    {
      "epoch": 0.9292752900087314,
      "grad_norm": 0.0,
      "learning_rate": 2.612302072266637e-07,
      "loss": 0.9034,
      "step": 7450
    },
    {
      "epoch": 0.9294000249469877,
      "grad_norm": 0.0,
      "learning_rate": 2.603135880665397e-07,
      "loss": 0.9246,
      "step": 7451
    },
    {
      "epoch": 0.9295247598852439,
      "grad_norm": 0.0,
      "learning_rate": 2.593985586687764e-07,
      "loss": 0.8776,
      "step": 7452
    },
    {
      "epoch": 0.9296494948235,
      "grad_norm": 0.0,
      "learning_rate": 2.5848511918272535e-07,
      "loss": 0.9365,
      "step": 7453
    },
    {
      "epoch": 0.9297742297617563,
      "grad_norm": 0.0,
      "learning_rate": 2.575732697574851e-07,
      "loss": 0.9067,
      "step": 7454
    },
    {
      "epoch": 0.9298989647000124,
      "grad_norm": 0.0,
      "learning_rate": 2.5666301054189323e-07,
      "loss": 0.8993,
      "step": 7455
    },
    {
      "epoch": 0.9300236996382687,
      "grad_norm": 0.0,
      "learning_rate": 2.557543416845265e-07,
      "loss": 0.855,
      "step": 7456
    },
    {
      "epoch": 0.9301484345765249,
      "grad_norm": 0.0,
      "learning_rate": 2.548472633337007e-07,
      "loss": 0.9184,
      "step": 7457
    },
    {
      "epoch": 0.9302731695147811,
      "grad_norm": 0.0,
      "learning_rate": 2.5394177563747736e-07,
      "loss": 0.903,
      "step": 7458
    },
    {
      "epoch": 0.9303979044530373,
      "grad_norm": 0.0,
      "learning_rate": 2.530378787436527e-07,
      "loss": 0.9007,
      "step": 7459
    },
    {
      "epoch": 0.9305226393912935,
      "grad_norm": 0.0,
      "learning_rate": 2.5213557279976317e-07,
      "loss": 0.8215,
      "step": 7460
    },
    {
      "epoch": 0.9306473743295497,
      "grad_norm": 0.0,
      "learning_rate": 2.512348579530932e-07,
      "loss": 0.9348,
      "step": 7461
    },
    {
      "epoch": 0.930772109267806,
      "grad_norm": 0.0,
      "learning_rate": 2.503357343506585e-07,
      "loss": 0.9138,
      "step": 7462
    },
    {
      "epoch": 0.9308968442060621,
      "grad_norm": 0.0,
      "learning_rate": 2.494382021392194e-07,
      "loss": 0.8914,
      "step": 7463
    },
    {
      "epoch": 0.9310215791443184,
      "grad_norm": 0.0,
      "learning_rate": 2.485422614652766e-07,
      "loss": 0.879,
      "step": 7464
    },
    {
      "epoch": 0.9311463140825745,
      "grad_norm": 0.0,
      "learning_rate": 2.4764791247506967e-07,
      "loss": 0.8431,
      "step": 7465
    },
    {
      "epoch": 0.9312710490208307,
      "grad_norm": 0.0,
      "learning_rate": 2.467551553145775e-07,
      "loss": 0.8611,
      "step": 7466
    },
    {
      "epoch": 0.931395783959087,
      "grad_norm": 0.0,
      "learning_rate": 2.458639901295223e-07,
      "loss": 0.8962,
      "step": 7467
    },
    {
      "epoch": 0.9315205188973431,
      "grad_norm": 0.0,
      "learning_rate": 2.449744170653645e-07,
      "loss": 0.9378,
      "step": 7468
    },
    {
      "epoch": 0.9316452538355994,
      "grad_norm": 0.0,
      "learning_rate": 2.4408643626730565e-07,
      "loss": 0.8594,
      "step": 7469
    },
    {
      "epoch": 0.9317699887738555,
      "grad_norm": 0.0,
      "learning_rate": 2.432000478802854e-07,
      "loss": 0.8856,
      "step": 7470
    },
    {
      "epoch": 0.9318947237121118,
      "grad_norm": 0.0,
      "learning_rate": 2.4231525204898354e-07,
      "loss": 0.8759,
      "step": 7471
    },
    {
      "epoch": 0.932019458650368,
      "grad_norm": 0.0,
      "learning_rate": 2.414320489178223e-07,
      "loss": 0.9563,
      "step": 7472
    },
    {
      "epoch": 0.9321441935886242,
      "grad_norm": 0.0,
      "learning_rate": 2.405504386309643e-07,
      "loss": 0.9137,
      "step": 7473
    },
    {
      "epoch": 0.9322689285268804,
      "grad_norm": 0.0,
      "learning_rate": 2.3967042133230756e-07,
      "loss": 0.9255,
      "step": 7474
    },
    {
      "epoch": 0.9323936634651366,
      "grad_norm": 0.0,
      "learning_rate": 2.387919971654962e-07,
      "loss": 0.8995,
      "step": 7475
    },
    {
      "epoch": 0.9325183984033928,
      "grad_norm": 0.0,
      "learning_rate": 2.3791516627390877e-07,
      "loss": 0.9135,
      "step": 7476
    },
    {
      "epoch": 0.9326431333416489,
      "grad_norm": 0.0,
      "learning_rate": 2.370399288006664e-07,
      "loss": 0.9318,
      "step": 7477
    },
    {
      "epoch": 0.9327678682799052,
      "grad_norm": 0.0,
      "learning_rate": 2.3616628488863147e-07,
      "loss": 0.9006,
      "step": 7478
    },
    {
      "epoch": 0.9328926032181614,
      "grad_norm": 0.0,
      "learning_rate": 2.3529423468040324e-07,
      "loss": 0.8881,
      "step": 7479
    },
    {
      "epoch": 0.9330173381564176,
      "grad_norm": 0.0,
      "learning_rate": 2.344237783183223e-07,
      "loss": 0.9048,
      "step": 7480
    },
    {
      "epoch": 0.9331420730946738,
      "grad_norm": 0.0,
      "learning_rate": 2.3355491594447055e-07,
      "loss": 0.9148,
      "step": 7481
    },
    {
      "epoch": 0.93326680803293,
      "grad_norm": 0.0,
      "learning_rate": 2.3268764770066564e-07,
      "loss": 0.9373,
      "step": 7482
    },
    {
      "epoch": 0.9333915429711862,
      "grad_norm": 0.0,
      "learning_rate": 2.3182197372846994e-07,
      "loss": 0.8942,
      "step": 7483
    },
    {
      "epoch": 0.9335162779094425,
      "grad_norm": 0.0,
      "learning_rate": 2.3095789416918258e-07,
      "loss": 0.8659,
      "step": 7484
    },
    {
      "epoch": 0.9336410128476986,
      "grad_norm": 0.0,
      "learning_rate": 2.300954091638441e-07,
      "loss": 0.8843,
      "step": 7485
    },
    {
      "epoch": 0.9337657477859549,
      "grad_norm": 0.0,
      "learning_rate": 2.292345188532308e-07,
      "loss": 0.9311,
      "step": 7486
    },
    {
      "epoch": 0.933890482724211,
      "grad_norm": 0.0,
      "learning_rate": 2.2837522337786467e-07,
      "loss": 0.9732,
      "step": 7487
    },
    {
      "epoch": 0.9340152176624673,
      "grad_norm": 0.0,
      "learning_rate": 2.2751752287800354e-07,
      "loss": 0.9255,
      "step": 7488
    },
    {
      "epoch": 0.9341399526007235,
      "grad_norm": 0.0,
      "learning_rate": 2.2666141749364434e-07,
      "loss": 0.8712,
      "step": 7489
    },
    {
      "epoch": 0.9342646875389796,
      "grad_norm": 0.0,
      "learning_rate": 2.2580690736452858e-07,
      "loss": 0.9262,
      "step": 7490
    },
    {
      "epoch": 0.9343894224772359,
      "grad_norm": 0.0,
      "learning_rate": 2.2495399263012919e-07,
      "loss": 0.8903,
      "step": 7491
    },
    {
      "epoch": 0.934514157415492,
      "grad_norm": 0.0,
      "learning_rate": 2.2410267342966695e-07,
      "loss": 0.8995,
      "step": 7492
    },
    {
      "epoch": 0.9346388923537483,
      "grad_norm": 0.0,
      "learning_rate": 2.2325294990209744e-07,
      "loss": 0.8804,
      "step": 7493
    },
    {
      "epoch": 0.9347636272920045,
      "grad_norm": 0.0,
      "learning_rate": 2.2240482218611524e-07,
      "loss": 0.9049,
      "step": 7494
    },
    {
      "epoch": 0.9348883622302607,
      "grad_norm": 0.0,
      "learning_rate": 2.2155829042015963e-07,
      "loss": 0.8846,
      "step": 7495
    },
    {
      "epoch": 0.9350130971685169,
      "grad_norm": 0.0,
      "learning_rate": 2.207133547424045e-07,
      "loss": 0.9314,
      "step": 7496
    },
    {
      "epoch": 0.9351378321067731,
      "grad_norm": 0.0,
      "learning_rate": 2.198700152907629e-07,
      "loss": 0.9429,
      "step": 7497
    },
    {
      "epoch": 0.9352625670450293,
      "grad_norm": 0.0,
      "learning_rate": 2.1902827220289137e-07,
      "loss": 0.9296,
      "step": 7498
    },
    {
      "epoch": 0.9353873019832856,
      "grad_norm": 0.0,
      "learning_rate": 2.181881256161822e-07,
      "loss": 0.8922,
      "step": 7499
    },
    {
      "epoch": 0.9355120369215417,
      "grad_norm": 0.0,
      "learning_rate": 2.1734957566776903e-07,
      "loss": 0.9488,
      "step": 7500
    },
    {
      "epoch": 0.9356367718597979,
      "grad_norm": 0.0,
      "learning_rate": 2.1651262249452465e-07,
      "loss": 0.8941,
      "step": 7501
    },
    {
      "epoch": 0.9357615067980541,
      "grad_norm": 0.0,
      "learning_rate": 2.1567726623306084e-07,
      "loss": 0.829,
      "step": 7502
    },
    {
      "epoch": 0.9358862417363103,
      "grad_norm": 0.0,
      "learning_rate": 2.1484350701972745e-07,
      "loss": 0.9013,
      "step": 7503
    },
    {
      "epoch": 0.9360109766745666,
      "grad_norm": 0.0,
      "learning_rate": 2.140113449906167e-07,
      "loss": 0.8815,
      "step": 7504
    },
    {
      "epoch": 0.9361357116128227,
      "grad_norm": 0.0,
      "learning_rate": 2.1318078028155886e-07,
      "loss": 0.9027,
      "step": 7505
    },
    {
      "epoch": 0.936260446551079,
      "grad_norm": 0.0,
      "learning_rate": 2.123518130281199e-07,
      "loss": 0.9177,
      "step": 7506
    },
    {
      "epoch": 0.9363851814893351,
      "grad_norm": 0.0,
      "learning_rate": 2.1152444336561272e-07,
      "loss": 0.9073,
      "step": 7507
    },
    {
      "epoch": 0.9365099164275914,
      "grad_norm": 0.0,
      "learning_rate": 2.106986714290804e-07,
      "loss": 0.9127,
      "step": 7508
    },
    {
      "epoch": 0.9366346513658476,
      "grad_norm": 0.0,
      "learning_rate": 2.0987449735331178e-07,
      "loss": 0.8471,
      "step": 7509
    },
    {
      "epoch": 0.9367593863041038,
      "grad_norm": 0.0,
      "learning_rate": 2.090519212728348e-07,
      "loss": 0.8948,
      "step": 7510
    },
    {
      "epoch": 0.93688412124236,
      "grad_norm": 0.0,
      "learning_rate": 2.0823094332191097e-07,
      "loss": 0.9388,
      "step": 7511
    },
    {
      "epoch": 0.9370088561806162,
      "grad_norm": 0.0,
      "learning_rate": 2.074115636345464e-07,
      "loss": 0.8949,
      "step": 7512
    },
    {
      "epoch": 0.9371335911188724,
      "grad_norm": 0.0,
      "learning_rate": 2.0659378234448524e-07,
      "loss": 0.9098,
      "step": 7513
    },
    {
      "epoch": 0.9372583260571286,
      "grad_norm": 0.0,
      "learning_rate": 2.0577759958520848e-07,
      "loss": 0.8785,
      "step": 7514
    },
    {
      "epoch": 0.9373830609953848,
      "grad_norm": 0.0,
      "learning_rate": 2.049630154899396e-07,
      "loss": 0.8663,
      "step": 7515
    },
    {
      "epoch": 0.937507795933641,
      "grad_norm": 0.0,
      "learning_rate": 2.0415003019163659e-07,
      "loss": 0.9339,
      "step": 7516
    },
    {
      "epoch": 0.9376325308718972,
      "grad_norm": 0.0,
      "learning_rate": 2.0333864382300116e-07,
      "loss": 0.9135,
      "step": 7517
    },
    {
      "epoch": 0.9377572658101534,
      "grad_norm": 0.0,
      "learning_rate": 2.025288565164707e-07,
      "loss": 0.8269,
      "step": 7518
    },
    {
      "epoch": 0.9378820007484097,
      "grad_norm": 0.0,
      "learning_rate": 2.01720668404225e-07,
      "loss": 0.8628,
      "step": 7519
    },
    {
      "epoch": 0.9380067356866658,
      "grad_norm": 0.0,
      "learning_rate": 2.0091407961817967e-07,
      "loss": 0.8911,
      "step": 7520
    },
    {
      "epoch": 0.9381314706249221,
      "grad_norm": 0.0,
      "learning_rate": 2.0010909028998826e-07,
      "loss": 0.8801,
      "step": 7521
    },
    {
      "epoch": 0.9382562055631782,
      "grad_norm": 0.0,
      "learning_rate": 1.9930570055104903e-07,
      "loss": 0.9371,
      "step": 7522
    },
    {
      "epoch": 0.9383809405014345,
      "grad_norm": 0.0,
      "learning_rate": 1.9850391053249153e-07,
      "loss": 0.9138,
      "step": 7523
    },
    {
      "epoch": 0.9385056754396907,
      "grad_norm": 0.0,
      "learning_rate": 1.9770372036519214e-07,
      "loss": 0.8872,
      "step": 7524
    },
    {
      "epoch": 0.9386304103779468,
      "grad_norm": 0.0,
      "learning_rate": 1.9690513017975976e-07,
      "loss": 0.9266,
      "step": 7525
    },
    {
      "epoch": 0.9387551453162031,
      "grad_norm": 0.0,
      "learning_rate": 1.9610814010654343e-07,
      "loss": 0.93,
      "step": 7526
    },
    {
      "epoch": 0.9388798802544592,
      "grad_norm": 0.0,
      "learning_rate": 1.953127502756358e-07,
      "loss": 0.8915,
      "step": 7527
    },
    {
      "epoch": 0.9390046151927155,
      "grad_norm": 0.0,
      "learning_rate": 1.945189608168596e-07,
      "loss": 0.8825,
      "step": 7528
    },
    {
      "epoch": 0.9391293501309717,
      "grad_norm": 0.0,
      "learning_rate": 1.9372677185978462e-07,
      "loss": 0.902,
      "step": 7529
    },
    {
      "epoch": 0.9392540850692279,
      "grad_norm": 0.0,
      "learning_rate": 1.9293618353371512e-07,
      "loss": 0.8839,
      "step": 7530
    },
    {
      "epoch": 0.9393788200074841,
      "grad_norm": 0.0,
      "learning_rate": 1.921471959676957e-07,
      "loss": 0.943,
      "step": 7531
    },
    {
      "epoch": 0.9395035549457403,
      "grad_norm": 0.0,
      "learning_rate": 1.913598092905089e-07,
      "loss": 0.9413,
      "step": 7532
    },
    {
      "epoch": 0.9396282898839965,
      "grad_norm": 0.0,
      "learning_rate": 1.9057402363067413e-07,
      "loss": 0.9131,
      "step": 7533
    },
    {
      "epoch": 0.9397530248222528,
      "grad_norm": 0.0,
      "learning_rate": 1.8978983911645322e-07,
      "loss": 0.8776,
      "step": 7534
    },
    {
      "epoch": 0.9398777597605089,
      "grad_norm": 0.0,
      "learning_rate": 1.890072558758449e-07,
      "loss": 0.8929,
      "step": 7535
    },
    {
      "epoch": 0.9400024946987652,
      "grad_norm": 0.0,
      "learning_rate": 1.88226274036587e-07,
      "loss": 0.9101,
      "step": 7536
    },
    {
      "epoch": 0.9401272296370213,
      "grad_norm": 0.0,
      "learning_rate": 1.874468937261531e-07,
      "loss": 0.9418,
      "step": 7537
    },
    {
      "epoch": 0.9402519645752775,
      "grad_norm": 0.0,
      "learning_rate": 1.8666911507176032e-07,
      "loss": 0.8994,
      "step": 7538
    },
    {
      "epoch": 0.9403766995135338,
      "grad_norm": 0.0,
      "learning_rate": 1.8589293820036048e-07,
      "loss": 0.9639,
      "step": 7539
    },
    {
      "epoch": 0.9405014344517899,
      "grad_norm": 0.0,
      "learning_rate": 1.851183632386444e-07,
      "loss": 0.9067,
      "step": 7540
    },
    {
      "epoch": 0.9406261693900462,
      "grad_norm": 0.0,
      "learning_rate": 1.8434539031304433e-07,
      "loss": 0.8969,
      "step": 7541
    },
    {
      "epoch": 0.9407509043283023,
      "grad_norm": 0.0,
      "learning_rate": 1.835740195497271e-07,
      "loss": 0.9227,
      "step": 7542
    },
    {
      "epoch": 0.9408756392665586,
      "grad_norm": 0.0,
      "learning_rate": 1.8280425107460088e-07,
      "loss": 0.9154,
      "step": 7543
    },
    {
      "epoch": 0.9410003742048147,
      "grad_norm": 0.0,
      "learning_rate": 1.8203608501331183e-07,
      "loss": 0.9572,
      "step": 7544
    },
    {
      "epoch": 0.941125109143071,
      "grad_norm": 0.0,
      "learning_rate": 1.8126952149124188e-07,
      "loss": 0.9075,
      "step": 7545
    },
    {
      "epoch": 0.9412498440813272,
      "grad_norm": 0.0,
      "learning_rate": 1.805045606335143e-07,
      "loss": 0.8909,
      "step": 7546
    },
    {
      "epoch": 0.9413745790195834,
      "grad_norm": 0.0,
      "learning_rate": 1.797412025649914e-07,
      "loss": 0.9335,
      "step": 7547
    },
    {
      "epoch": 0.9414993139578396,
      "grad_norm": 0.0,
      "learning_rate": 1.789794474102724e-07,
      "loss": 0.9451,
      "step": 7548
    },
    {
      "epoch": 0.9416240488960957,
      "grad_norm": 0.0,
      "learning_rate": 1.7821929529369343e-07,
      "loss": 0.8611,
      "step": 7549
    },
    {
      "epoch": 0.941748783834352,
      "grad_norm": 0.0,
      "learning_rate": 1.7746074633933075e-07,
      "loss": 0.9088,
      "step": 7550
    },
    {
      "epoch": 0.9418735187726082,
      "grad_norm": 0.0,
      "learning_rate": 1.7670380067099868e-07,
      "loss": 0.9104,
      "step": 7551
    },
    {
      "epoch": 0.9419982537108644,
      "grad_norm": 0.0,
      "learning_rate": 1.7594845841225062e-07,
      "loss": 0.8983,
      "step": 7552
    },
    {
      "epoch": 0.9421229886491206,
      "grad_norm": 0.0,
      "learning_rate": 1.7519471968637792e-07,
      "loss": 0.8798,
      "step": 7553
    },
    {
      "epoch": 0.9422477235873769,
      "grad_norm": 0.0,
      "learning_rate": 1.7444258461640885e-07,
      "loss": 0.9278,
      "step": 7554
    },
    {
      "epoch": 0.942372458525633,
      "grad_norm": 0.0,
      "learning_rate": 1.7369205332510964e-07,
      "loss": 0.8571,
      "step": 7555
    },
    {
      "epoch": 0.9424971934638893,
      "grad_norm": 0.0,
      "learning_rate": 1.7294312593498897e-07,
      "loss": 0.8594,
      "step": 7556
    },
    {
      "epoch": 0.9426219284021454,
      "grad_norm": 0.0,
      "learning_rate": 1.7219580256828683e-07,
      "loss": 0.952,
      "step": 7557
    },
    {
      "epoch": 0.9427466633404017,
      "grad_norm": 0.0,
      "learning_rate": 1.7145008334698898e-07,
      "loss": 0.8899,
      "step": 7558
    },
    {
      "epoch": 0.9428713982786578,
      "grad_norm": 0.0,
      "learning_rate": 1.707059683928125e-07,
      "loss": 0.848,
      "step": 7559
    },
    {
      "epoch": 0.9429961332169141,
      "grad_norm": 0.0,
      "learning_rate": 1.6996345782721692e-07,
      "loss": 0.8932,
      "step": 7560
    },
    {
      "epoch": 0.9431208681551703,
      "grad_norm": 0.0,
      "learning_rate": 1.6922255177140078e-07,
      "loss": 0.9069,
      "step": 7561
    },
    {
      "epoch": 0.9432456030934264,
      "grad_norm": 0.0,
      "learning_rate": 1.684832503462952e-07,
      "loss": 0.9193,
      "step": 7562
    },
    {
      "epoch": 0.9433703380316827,
      "grad_norm": 0.0,
      "learning_rate": 1.677455536725736e-07,
      "loss": 0.9228,
      "step": 7563
    },
    {
      "epoch": 0.9434950729699388,
      "grad_norm": 0.0,
      "learning_rate": 1.6700946187064858e-07,
      "loss": 0.8607,
      "step": 7564
    },
    {
      "epoch": 0.9436198079081951,
      "grad_norm": 0.0,
      "learning_rate": 1.6627497506066737e-07,
      "loss": 0.9416,
      "step": 7565
    },
    {
      "epoch": 0.9437445428464513,
      "grad_norm": 0.0,
      "learning_rate": 1.6554209336251738e-07,
      "loss": 0.9042,
      "step": 7566
    },
    {
      "epoch": 0.9438692777847075,
      "grad_norm": 0.0,
      "learning_rate": 1.648108168958229e-07,
      "loss": 0.9139,
      "step": 7567
    },
    {
      "epoch": 0.9439940127229637,
      "grad_norm": 0.0,
      "learning_rate": 1.6408114577994627e-07,
      "loss": 0.9037,
      "step": 7568
    },
    {
      "epoch": 0.94411874766122,
      "grad_norm": 0.0,
      "learning_rate": 1.6335308013398888e-07,
      "loss": 0.8511,
      "step": 7569
    },
    {
      "epoch": 0.9442434825994761,
      "grad_norm": 0.0,
      "learning_rate": 1.626266200767901e-07,
      "loss": 0.8994,
      "step": 7570
    },
    {
      "epoch": 0.9443682175377324,
      "grad_norm": 0.0,
      "learning_rate": 1.6190176572692506e-07,
      "loss": 0.9138,
      "step": 7571
    },
    {
      "epoch": 0.9444929524759885,
      "grad_norm": 0.0,
      "learning_rate": 1.6117851720270805e-07,
      "loss": 0.8811,
      "step": 7572
    },
    {
      "epoch": 0.9446176874142447,
      "grad_norm": 0.0,
      "learning_rate": 1.6045687462219352e-07,
      "loss": 0.8582,
      "step": 7573
    },
    {
      "epoch": 0.9447424223525009,
      "grad_norm": 0.0,
      "learning_rate": 1.5973683810317054e-07,
      "loss": 0.8444,
      "step": 7574
    },
    {
      "epoch": 0.9448671572907571,
      "grad_norm": 0.0,
      "learning_rate": 1.5901840776316623e-07,
      "loss": 0.8555,
      "step": 7575
    },
    {
      "epoch": 0.9449918922290134,
      "grad_norm": 0.0,
      "learning_rate": 1.5830158371944793e-07,
      "loss": 0.9203,
      "step": 7576
    },
    {
      "epoch": 0.9451166271672695,
      "grad_norm": 0.0,
      "learning_rate": 1.5758636608901977e-07,
      "loss": 0.9209,
      "step": 7577
    },
    {
      "epoch": 0.9452413621055258,
      "grad_norm": 0.0,
      "learning_rate": 1.5687275498862174e-07,
      "loss": 0.9287,
      "step": 7578
    },
    {
      "epoch": 0.9453660970437819,
      "grad_norm": 0.0,
      "learning_rate": 1.56160750534734e-07,
      "loss": 0.9617,
      "step": 7579
    },
    {
      "epoch": 0.9454908319820382,
      "grad_norm": 0.0,
      "learning_rate": 1.5545035284357357e-07,
      "loss": 0.9329,
      "step": 7580
    },
    {
      "epoch": 0.9456155669202944,
      "grad_norm": 0.0,
      "learning_rate": 1.5474156203109548e-07,
      "loss": 0.8667,
      "step": 7581
    },
    {
      "epoch": 0.9457403018585506,
      "grad_norm": 0.0,
      "learning_rate": 1.5403437821299272e-07,
      "loss": 0.9338,
      "step": 7582
    },
    {
      "epoch": 0.9458650367968068,
      "grad_norm": 0.0,
      "learning_rate": 1.53328801504693e-07,
      "loss": 0.9413,
      "step": 7583
    },
    {
      "epoch": 0.945989771735063,
      "grad_norm": 0.0,
      "learning_rate": 1.5262483202136747e-07,
      "loss": 0.929,
      "step": 7584
    },
    {
      "epoch": 0.9461145066733192,
      "grad_norm": 0.0,
      "learning_rate": 1.519224698779198e-07,
      "loss": 0.9473,
      "step": 7585
    },
    {
      "epoch": 0.9462392416115754,
      "grad_norm": 0.0,
      "learning_rate": 1.5122171518899274e-07,
      "loss": 0.9266,
      "step": 7586
    },
    {
      "epoch": 0.9463639765498316,
      "grad_norm": 0.0,
      "learning_rate": 1.5052256806897037e-07,
      "loss": 0.9455,
      "step": 7587
    },
    {
      "epoch": 0.9464887114880878,
      "grad_norm": 0.0,
      "learning_rate": 1.4982502863196692e-07,
      "loss": 0.8761,
      "step": 7588
    },
    {
      "epoch": 0.946613446426344,
      "grad_norm": 0.0,
      "learning_rate": 1.491290969918402e-07,
      "loss": 0.9291,
      "step": 7589
    },
    {
      "epoch": 0.9467381813646002,
      "grad_norm": 0.0,
      "learning_rate": 1.4843477326218493e-07,
      "loss": 0.8982,
      "step": 7590
    },
    {
      "epoch": 0.9468629163028565,
      "grad_norm": 0.0,
      "learning_rate": 1.4774205755633041e-07,
      "loss": 0.8605,
      "step": 7591
    },
    {
      "epoch": 0.9469876512411126,
      "grad_norm": 0.0,
      "learning_rate": 1.4705094998734738e-07,
      "loss": 0.9081,
      "step": 7592
    },
    {
      "epoch": 0.9471123861793689,
      "grad_norm": 0.0,
      "learning_rate": 1.4636145066803886e-07,
      "loss": 0.8844,
      "step": 7593
    },
    {
      "epoch": 0.947237121117625,
      "grad_norm": 0.0,
      "learning_rate": 1.4567355971095266e-07,
      "loss": 0.9255,
      "step": 7594
    },
    {
      "epoch": 0.9473618560558813,
      "grad_norm": 0.0,
      "learning_rate": 1.449872772283656e-07,
      "loss": 0.8577,
      "step": 7595
    },
    {
      "epoch": 0.9474865909941375,
      "grad_norm": 0.0,
      "learning_rate": 1.4430260333229919e-07,
      "loss": 0.9313,
      "step": 7596
    },
    {
      "epoch": 0.9476113259323936,
      "grad_norm": 0.0,
      "learning_rate": 1.4361953813450846e-07,
      "loss": 0.9021,
      "step": 7597
    },
    {
      "epoch": 0.9477360608706499,
      "grad_norm": 0.0,
      "learning_rate": 1.429380817464865e-07,
      "loss": 0.9226,
      "step": 7598
    },
    {
      "epoch": 0.947860795808906,
      "grad_norm": 0.0,
      "learning_rate": 1.422582342794665e-07,
      "loss": 0.8746,
      "step": 7599
    },
    {
      "epoch": 0.9479855307471623,
      "grad_norm": 0.0,
      "learning_rate": 1.4157999584441417e-07,
      "loss": 0.9176,
      "step": 7600
    },
    {
      "epoch": 0.9481102656854185,
      "grad_norm": 0.0,
      "learning_rate": 1.409033665520354e-07,
      "loss": 0.8661,
      "step": 7601
    },
    {
      "epoch": 0.9482350006236747,
      "grad_norm": 0.0,
      "learning_rate": 1.4022834651277407e-07,
      "loss": 0.8692,
      "step": 7602
    },
    {
      "epoch": 0.9483597355619309,
      "grad_norm": 0.0,
      "learning_rate": 1.395549358368087e-07,
      "loss": 0.8369,
      "step": 7603
    },
    {
      "epoch": 0.9484844705001871,
      "grad_norm": 0.0,
      "learning_rate": 1.3888313463406022e-07,
      "loss": 0.8579,
      "step": 7604
    },
    {
      "epoch": 0.9486092054384433,
      "grad_norm": 0.0,
      "learning_rate": 1.3821294301418097e-07,
      "loss": 0.9185,
      "step": 7605
    },
    {
      "epoch": 0.9487339403766996,
      "grad_norm": 0.0,
      "learning_rate": 1.3754436108656234e-07,
      "loss": 0.887,
      "step": 7606
    },
    {
      "epoch": 0.9488586753149557,
      "grad_norm": 0.0,
      "learning_rate": 1.3687738896033588e-07,
      "loss": 0.9236,
      "step": 7607
    },
    {
      "epoch": 0.948983410253212,
      "grad_norm": 0.0,
      "learning_rate": 1.3621202674436783e-07,
      "loss": 0.8483,
      "step": 7608
    },
    {
      "epoch": 0.9491081451914681,
      "grad_norm": 0.0,
      "learning_rate": 1.3554827454726137e-07,
      "loss": 0.8624,
      "step": 7609
    },
    {
      "epoch": 0.9492328801297243,
      "grad_norm": 0.0,
      "learning_rate": 1.348861324773576e-07,
      "loss": 0.9151,
      "step": 7610
    },
    {
      "epoch": 0.9493576150679806,
      "grad_norm": 0.0,
      "learning_rate": 1.342256006427356e-07,
      "loss": 0.9069,
      "step": 7611
    },
    {
      "epoch": 0.9494823500062367,
      "grad_norm": 0.0,
      "learning_rate": 1.3356667915121025e-07,
      "loss": 0.9221,
      "step": 7612
    },
    {
      "epoch": 0.949607084944493,
      "grad_norm": 0.0,
      "learning_rate": 1.3290936811033438e-07,
      "loss": 0.8767,
      "step": 7613
    },
    {
      "epoch": 0.9497318198827491,
      "grad_norm": 0.0,
      "learning_rate": 1.3225366762739778e-07,
      "loss": 0.9172,
      "step": 7614
    },
    {
      "epoch": 0.9498565548210054,
      "grad_norm": 0.0,
      "learning_rate": 1.315995778094259e-07,
      "loss": 0.8875,
      "step": 7615
    },
    {
      "epoch": 0.9499812897592615,
      "grad_norm": 0.0,
      "learning_rate": 1.3094709876318447e-07,
      "loss": 0.9052,
      "step": 7616
    },
    {
      "epoch": 0.9501060246975178,
      "grad_norm": 0.0,
      "learning_rate": 1.3029623059517493e-07,
      "loss": 0.9255,
      "step": 7617
    },
    {
      "epoch": 0.950230759635774,
      "grad_norm": 0.0,
      "learning_rate": 1.2964697341163234e-07,
      "loss": 0.9131,
      "step": 7618
    },
    {
      "epoch": 0.9503554945740302,
      "grad_norm": 0.0,
      "learning_rate": 1.2899932731853637e-07,
      "loss": 0.8915,
      "step": 7619
    },
    {
      "epoch": 0.9504802295122864,
      "grad_norm": 0.0,
      "learning_rate": 1.2835329242159467e-07,
      "loss": 0.908,
      "step": 7620
    },
    {
      "epoch": 0.9506049644505425,
      "grad_norm": 0.0,
      "learning_rate": 1.2770886882625954e-07,
      "loss": 0.8914,
      "step": 7621
    },
    {
      "epoch": 0.9507296993887988,
      "grad_norm": 0.0,
      "learning_rate": 1.2706605663771575e-07,
      "loss": 0.938,
      "step": 7622
    },
    {
      "epoch": 0.950854434327055,
      "grad_norm": 0.0,
      "learning_rate": 1.2642485596088606e-07,
      "loss": 0.905,
      "step": 7623
    },
    {
      "epoch": 0.9509791692653112,
      "grad_norm": 0.0,
      "learning_rate": 1.2578526690043337e-07,
      "loss": 0.9486,
      "step": 7624
    },
    {
      "epoch": 0.9511039042035674,
      "grad_norm": 0.0,
      "learning_rate": 1.2514728956075084e-07,
      "loss": 0.9075,
      "step": 7625
    },
    {
      "epoch": 0.9512286391418237,
      "grad_norm": 0.0,
      "learning_rate": 1.245109240459741e-07,
      "loss": 0.8558,
      "step": 7626
    },
    {
      "epoch": 0.9513533740800798,
      "grad_norm": 0.0,
      "learning_rate": 1.2387617045997446e-07,
      "loss": 0.83,
      "step": 7627
    },
    {
      "epoch": 0.9514781090183361,
      "grad_norm": 0.0,
      "learning_rate": 1.2324302890636131e-07,
      "loss": 0.9527,
      "step": 7628
    },
    {
      "epoch": 0.9516028439565922,
      "grad_norm": 0.0,
      "learning_rate": 1.226114994884753e-07,
      "loss": 0.8275,
      "step": 7629
    },
    {
      "epoch": 0.9517275788948485,
      "grad_norm": 0.0,
      "learning_rate": 1.219815823094006e-07,
      "loss": 0.9063,
      "step": 7630
    },
    {
      "epoch": 0.9518523138331046,
      "grad_norm": 0.0,
      "learning_rate": 1.2135327747195613e-07,
      "loss": 0.9119,
      "step": 7631
    },
    {
      "epoch": 0.9519770487713609,
      "grad_norm": 0.0,
      "learning_rate": 1.2072658507869428e-07,
      "loss": 0.8869,
      "step": 7632
    },
    {
      "epoch": 0.9521017837096171,
      "grad_norm": 0.0,
      "learning_rate": 1.201015052319099e-07,
      "loss": 0.9219,
      "step": 7633
    },
    {
      "epoch": 0.9522265186478732,
      "grad_norm": 0.0,
      "learning_rate": 1.1947803803363135e-07,
      "loss": 0.8855,
      "step": 7634
    },
    {
      "epoch": 0.9523512535861295,
      "grad_norm": 0.0,
      "learning_rate": 1.1885618358562278e-07,
      "loss": 0.892,
      "step": 7635
    },
    {
      "epoch": 0.9524759885243856,
      "grad_norm": 0.0,
      "learning_rate": 1.1823594198938749e-07,
      "loss": 0.9188,
      "step": 7636
    },
    {
      "epoch": 0.9526007234626419,
      "grad_norm": 0.0,
      "learning_rate": 1.1761731334616333e-07,
      "loss": 0.9172,
      "step": 7637
    },
    {
      "epoch": 0.9527254584008981,
      "grad_norm": 0.0,
      "learning_rate": 1.1700029775692845e-07,
      "loss": 0.9467,
      "step": 7638
    },
    {
      "epoch": 0.9528501933391543,
      "grad_norm": 0.0,
      "learning_rate": 1.1638489532239339e-07,
      "loss": 0.8491,
      "step": 7639
    },
    {
      "epoch": 0.9529749282774105,
      "grad_norm": 0.0,
      "learning_rate": 1.157711061430078e-07,
      "loss": 0.9475,
      "step": 7640
    },
    {
      "epoch": 0.9530996632156667,
      "grad_norm": 0.0,
      "learning_rate": 1.151589303189582e-07,
      "loss": 0.8778,
      "step": 7641
    },
    {
      "epoch": 0.9532243981539229,
      "grad_norm": 0.0,
      "learning_rate": 1.1454836795016577e-07,
      "loss": 0.9326,
      "step": 7642
    },
    {
      "epoch": 0.9533491330921792,
      "grad_norm": 0.0,
      "learning_rate": 1.1393941913629081e-07,
      "loss": 0.9006,
      "step": 7643
    },
    {
      "epoch": 0.9534738680304353,
      "grad_norm": 0.0,
      "learning_rate": 1.1333208397672823e-07,
      "loss": 0.9291,
      "step": 7644
    },
    {
      "epoch": 0.9535986029686915,
      "grad_norm": 0.0,
      "learning_rate": 1.127263625706132e-07,
      "loss": 0.8597,
      "step": 7645
    },
    {
      "epoch": 0.9537233379069477,
      "grad_norm": 0.0,
      "learning_rate": 1.1212225501680995e-07,
      "loss": 0.8884,
      "step": 7646
    },
    {
      "epoch": 0.9538480728452039,
      "grad_norm": 0.0,
      "learning_rate": 1.1151976141392851e-07,
      "loss": 0.8921,
      "step": 7647
    },
    {
      "epoch": 0.9539728077834602,
      "grad_norm": 0.0,
      "learning_rate": 1.1091888186030908e-07,
      "loss": 0.942,
      "step": 7648
    },
    {
      "epoch": 0.9540975427217163,
      "grad_norm": 0.0,
      "learning_rate": 1.1031961645402878e-07,
      "loss": 0.8983,
      "step": 7649
    },
    {
      "epoch": 0.9542222776599726,
      "grad_norm": 0.0,
      "learning_rate": 1.0972196529290602e-07,
      "loss": 0.9207,
      "step": 7650
    },
    {
      "epoch": 0.9543470125982287,
      "grad_norm": 0.0,
      "learning_rate": 1.0912592847449056e-07,
      "loss": 0.8908,
      "step": 7651
    },
    {
      "epoch": 0.954471747536485,
      "grad_norm": 0.0,
      "learning_rate": 1.0853150609607122e-07,
      "loss": 0.8591,
      "step": 7652
    },
    {
      "epoch": 0.9545964824747412,
      "grad_norm": 0.0,
      "learning_rate": 1.0793869825467262e-07,
      "loss": 0.9064,
      "step": 7653
    },
    {
      "epoch": 0.9547212174129974,
      "grad_norm": 0.0,
      "learning_rate": 1.0734750504705516e-07,
      "loss": 0.926,
      "step": 7654
    },
    {
      "epoch": 0.9548459523512536,
      "grad_norm": 0.0,
      "learning_rate": 1.0675792656971828e-07,
      "loss": 0.8967,
      "step": 7655
    },
    {
      "epoch": 0.9549706872895098,
      "grad_norm": 0.0,
      "learning_rate": 1.0616996291889504e-07,
      "loss": 0.8772,
      "step": 7656
    },
    {
      "epoch": 0.955095422227766,
      "grad_norm": 0.0,
      "learning_rate": 1.055836141905553e-07,
      "loss": 0.8822,
      "step": 7657
    },
    {
      "epoch": 0.9552201571660222,
      "grad_norm": 0.0,
      "learning_rate": 1.0499888048040585e-07,
      "loss": 0.9095,
      "step": 7658
    },
    {
      "epoch": 0.9553448921042784,
      "grad_norm": 0.0,
      "learning_rate": 1.0441576188389146e-07,
      "loss": 0.9051,
      "step": 7659
    },
    {
      "epoch": 0.9554696270425346,
      "grad_norm": 0.0,
      "learning_rate": 1.0383425849619155e-07,
      "loss": 0.8806,
      "step": 7660
    },
    {
      "epoch": 0.9555943619807908,
      "grad_norm": 0.0,
      "learning_rate": 1.032543704122213e-07,
      "loss": 0.9049,
      "step": 7661
    },
    {
      "epoch": 0.955719096919047,
      "grad_norm": 0.0,
      "learning_rate": 1.0267609772663278e-07,
      "loss": 0.902,
      "step": 7662
    },
    {
      "epoch": 0.9558438318573033,
      "grad_norm": 0.0,
      "learning_rate": 1.0209944053381605e-07,
      "loss": 0.9329,
      "step": 7663
    },
    {
      "epoch": 0.9559685667955594,
      "grad_norm": 0.0,
      "learning_rate": 1.0152439892789356e-07,
      "loss": 0.9231,
      "step": 7664
    },
    {
      "epoch": 0.9560933017338157,
      "grad_norm": 0.0,
      "learning_rate": 1.0095097300273026e-07,
      "loss": 0.904,
      "step": 7665
    },
    {
      "epoch": 0.9562180366720718,
      "grad_norm": 0.0,
      "learning_rate": 1.0037916285192129e-07,
      "loss": 0.9254,
      "step": 7666
    },
    {
      "epoch": 0.9563427716103281,
      "grad_norm": 0.0,
      "learning_rate": 9.980896856880084e-08,
      "loss": 0.8829,
      "step": 7667
    },
    {
      "epoch": 0.9564675065485843,
      "grad_norm": 0.0,
      "learning_rate": 9.924039024643895e-08,
      "loss": 0.8809,
      "step": 7668
    },
    {
      "epoch": 0.9565922414868404,
      "grad_norm": 0.0,
      "learning_rate": 9.867342797764135e-08,
      "loss": 0.9055,
      "step": 7669
    },
    {
      "epoch": 0.9567169764250967,
      "grad_norm": 0.0,
      "learning_rate": 9.810808185495291e-08,
      "loss": 0.8878,
      "step": 7670
    },
    {
      "epoch": 0.9568417113633528,
      "grad_norm": 0.0,
      "learning_rate": 9.754435197064981e-08,
      "loss": 0.8701,
      "step": 7671
    },
    {
      "epoch": 0.9569664463016091,
      "grad_norm": 0.0,
      "learning_rate": 9.698223841674847e-08,
      "loss": 0.905,
      "step": 7672
    },
    {
      "epoch": 0.9570911812398653,
      "grad_norm": 0.0,
      "learning_rate": 9.642174128499882e-08,
      "loss": 0.9274,
      "step": 7673
    },
    {
      "epoch": 0.9572159161781215,
      "grad_norm": 0.0,
      "learning_rate": 9.586286066688877e-08,
      "loss": 0.9317,
      "step": 7674
    },
    {
      "epoch": 0.9573406511163777,
      "grad_norm": 0.0,
      "learning_rate": 9.530559665364203e-08,
      "loss": 0.9305,
      "step": 7675
    },
    {
      "epoch": 0.9574653860546339,
      "grad_norm": 0.0,
      "learning_rate": 9.474994933621807e-08,
      "loss": 0.8765,
      "step": 7676
    },
    {
      "epoch": 0.9575901209928901,
      "grad_norm": 0.0,
      "learning_rate": 9.41959188053132e-08,
      "loss": 0.8806,
      "step": 7677
    },
    {
      "epoch": 0.9577148559311464,
      "grad_norm": 0.0,
      "learning_rate": 9.364350515135623e-08,
      "loss": 0.889,
      "step": 7678
    },
    {
      "epoch": 0.9578395908694025,
      "grad_norm": 0.0,
      "learning_rate": 9.309270846451835e-08,
      "loss": 0.9496,
      "step": 7679
    },
    {
      "epoch": 0.9579643258076588,
      "grad_norm": 0.0,
      "learning_rate": 9.254352883470207e-08,
      "loss": 0.9457,
      "step": 7680
    },
    {
      "epoch": 0.9580890607459149,
      "grad_norm": 0.0,
      "learning_rate": 9.199596635154684e-08,
      "loss": 0.8631,
      "step": 7681
    },
    {
      "epoch": 0.9582137956841711,
      "grad_norm": 0.0,
      "learning_rate": 9.145002110443002e-08,
      "loss": 0.8919,
      "step": 7682
    },
    {
      "epoch": 0.9583385306224274,
      "grad_norm": 0.0,
      "learning_rate": 9.090569318246256e-08,
      "loss": 0.9048,
      "step": 7683
    },
    {
      "epoch": 0.9584632655606835,
      "grad_norm": 0.0,
      "learning_rate": 9.036298267449228e-08,
      "loss": 0.8901,
      "step": 7684
    },
    {
      "epoch": 0.9585880004989398,
      "grad_norm": 0.0,
      "learning_rate": 8.982188966910498e-08,
      "loss": 0.9972,
      "step": 7685
    },
    {
      "epoch": 0.9587127354371959,
      "grad_norm": 0.0,
      "learning_rate": 8.928241425461781e-08,
      "loss": 0.9174,
      "step": 7686
    },
    {
      "epoch": 0.9588374703754522,
      "grad_norm": 0.0,
      "learning_rate": 8.874455651909031e-08,
      "loss": 0.8634,
      "step": 7687
    },
    {
      "epoch": 0.9589622053137083,
      "grad_norm": 0.0,
      "learning_rate": 8.820831655031115e-08,
      "loss": 0.8793,
      "step": 7688
    },
    {
      "epoch": 0.9590869402519646,
      "grad_norm": 0.0,
      "learning_rate": 8.76736944358103e-08,
      "loss": 0.8837,
      "step": 7689
    },
    {
      "epoch": 0.9592116751902208,
      "grad_norm": 0.0,
      "learning_rate": 8.714069026285022e-08,
      "loss": 0.9178,
      "step": 7690
    },
    {
      "epoch": 0.959336410128477,
      "grad_norm": 0.0,
      "learning_rate": 8.660930411843127e-08,
      "loss": 0.8979,
      "step": 7691
    },
    {
      "epoch": 0.9594611450667332,
      "grad_norm": 0.0,
      "learning_rate": 8.607953608928965e-08,
      "loss": 0.8992,
      "step": 7692
    },
    {
      "epoch": 0.9595858800049893,
      "grad_norm": 0.0,
      "learning_rate": 8.555138626189619e-08,
      "loss": 0.8898,
      "step": 7693
    },
    {
      "epoch": 0.9597106149432456,
      "grad_norm": 0.0,
      "learning_rate": 8.502485472245859e-08,
      "loss": 0.9016,
      "step": 7694
    },
    {
      "epoch": 0.9598353498815018,
      "grad_norm": 0.0,
      "learning_rate": 8.44999415569192e-08,
      "loss": 0.9199,
      "step": 7695
    },
    {
      "epoch": 0.959960084819758,
      "grad_norm": 0.0,
      "learning_rate": 8.39766468509573e-08,
      "loss": 0.8971,
      "step": 7696
    },
    {
      "epoch": 0.9600848197580142,
      "grad_norm": 0.0,
      "learning_rate": 8.345497068998897e-08,
      "loss": 0.9349,
      "step": 7697
    },
    {
      "epoch": 0.9602095546962705,
      "grad_norm": 0.0,
      "learning_rate": 8.29349131591628e-08,
      "loss": 0.9186,
      "step": 7698
    },
    {
      "epoch": 0.9603342896345266,
      "grad_norm": 0.0,
      "learning_rate": 8.241647434336864e-08,
      "loss": 0.8457,
      "step": 7699
    },
    {
      "epoch": 0.9604590245727829,
      "grad_norm": 0.0,
      "learning_rate": 8.189965432722546e-08,
      "loss": 0.8777,
      "step": 7700
    },
    {
      "epoch": 0.960583759511039,
      "grad_norm": 0.0,
      "learning_rate": 8.138445319509248e-08,
      "loss": 0.8905,
      "step": 7701
    },
    {
      "epoch": 0.9607084944492953,
      "grad_norm": 0.0,
      "learning_rate": 8.087087103106461e-08,
      "loss": 0.9152,
      "step": 7702
    },
    {
      "epoch": 0.9608332293875514,
      "grad_norm": 0.0,
      "learning_rate": 8.035890791896928e-08,
      "loss": 0.8667,
      "step": 7703
    },
    {
      "epoch": 0.9609579643258077,
      "grad_norm": 0.0,
      "learning_rate": 7.984856394237294e-08,
      "loss": 0.8869,
      "step": 7704
    },
    {
      "epoch": 0.9610826992640639,
      "grad_norm": 0.0,
      "learning_rate": 7.933983918457677e-08,
      "loss": 0.975,
      "step": 7705
    },
    {
      "epoch": 0.96120743420232,
      "grad_norm": 0.0,
      "learning_rate": 7.883273372861766e-08,
      "loss": 0.9293,
      "step": 7706
    },
    {
      "epoch": 0.9613321691405763,
      "grad_norm": 0.0,
      "learning_rate": 7.832724765726828e-08,
      "loss": 0.8765,
      "step": 7707
    },
    {
      "epoch": 0.9614569040788324,
      "grad_norm": 0.0,
      "learning_rate": 7.782338105303489e-08,
      "loss": 0.9171,
      "step": 7708
    },
    {
      "epoch": 0.9615816390170887,
      "grad_norm": 0.0,
      "learning_rate": 7.732113399816276e-08,
      "loss": 0.9198,
      "step": 7709
    },
    {
      "epoch": 0.9617063739553449,
      "grad_norm": 0.0,
      "learning_rate": 7.68205065746297e-08,
      "loss": 0.8547,
      "step": 7710
    },
    {
      "epoch": 0.9618311088936011,
      "grad_norm": 0.0,
      "learning_rate": 7.632149886415363e-08,
      "loss": 0.8904,
      "step": 7711
    },
    {
      "epoch": 0.9619558438318573,
      "grad_norm": 0.0,
      "learning_rate": 7.582411094818276e-08,
      "loss": 0.8791,
      "step": 7712
    },
    {
      "epoch": 0.9620805787701135,
      "grad_norm": 0.0,
      "learning_rate": 7.532834290790436e-08,
      "loss": 0.8689,
      "step": 7713
    },
    {
      "epoch": 0.9622053137083697,
      "grad_norm": 0.0,
      "learning_rate": 7.483419482423926e-08,
      "loss": 0.9058,
      "step": 7714
    },
    {
      "epoch": 0.962330048646626,
      "grad_norm": 0.0,
      "learning_rate": 7.434166677784626e-08,
      "loss": 0.8769,
      "step": 7715
    },
    {
      "epoch": 0.9624547835848821,
      "grad_norm": 0.0,
      "learning_rate": 7.385075884911775e-08,
      "loss": 0.881,
      "step": 7716
    },
    {
      "epoch": 0.9625795185231383,
      "grad_norm": 0.0,
      "learning_rate": 7.336147111818181e-08,
      "loss": 0.9376,
      "step": 7717
    },
    {
      "epoch": 0.9627042534613945,
      "grad_norm": 0.0,
      "learning_rate": 7.287380366490238e-08,
      "loss": 0.9145,
      "step": 7718
    },
    {
      "epoch": 0.9628289883996507,
      "grad_norm": 0.0,
      "learning_rate": 7.238775656888019e-08,
      "loss": 0.8818,
      "step": 7719
    },
    {
      "epoch": 0.962953723337907,
      "grad_norm": 0.0,
      "learning_rate": 7.19033299094496e-08,
      "loss": 0.9109,
      "step": 7720
    },
    {
      "epoch": 0.9630784582761631,
      "grad_norm": 0.0,
      "learning_rate": 7.142052376568176e-08,
      "loss": 0.9415,
      "step": 7721
    },
    {
      "epoch": 0.9632031932144194,
      "grad_norm": 0.0,
      "learning_rate": 7.093933821638255e-08,
      "loss": 0.9449,
      "step": 7722
    },
    {
      "epoch": 0.9633279281526755,
      "grad_norm": 0.0,
      "learning_rate": 7.045977334009357e-08,
      "loss": 0.9573,
      "step": 7723
    },
    {
      "epoch": 0.9634526630909318,
      "grad_norm": 0.0,
      "learning_rate": 6.998182921509222e-08,
      "loss": 0.9048,
      "step": 7724
    },
    {
      "epoch": 0.963577398029188,
      "grad_norm": 0.0,
      "learning_rate": 6.950550591938943e-08,
      "loss": 0.9145,
      "step": 7725
    },
    {
      "epoch": 0.9637021329674442,
      "grad_norm": 0.0,
      "learning_rate": 6.903080353073522e-08,
      "loss": 0.8535,
      "step": 7726
    },
    {
      "epoch": 0.9638268679057004,
      "grad_norm": 0.0,
      "learning_rate": 6.855772212661204e-08,
      "loss": 0.8912,
      "step": 7727
    },
    {
      "epoch": 0.9639516028439566,
      "grad_norm": 0.0,
      "learning_rate": 6.808626178423927e-08,
      "loss": 0.9118,
      "step": 7728
    },
    {
      "epoch": 0.9640763377822128,
      "grad_norm": 0.0,
      "learning_rate": 6.761642258056977e-08,
      "loss": 0.8686,
      "step": 7729
    },
    {
      "epoch": 0.964201072720469,
      "grad_norm": 0.0,
      "learning_rate": 6.714820459229554e-08,
      "loss": 0.907,
      "step": 7730
    },
    {
      "epoch": 0.9643258076587252,
      "grad_norm": 0.0,
      "learning_rate": 6.668160789583878e-08,
      "loss": 0.9098,
      "step": 7731
    },
    {
      "epoch": 0.9644505425969814,
      "grad_norm": 0.0,
      "learning_rate": 6.621663256736078e-08,
      "loss": 0.9226,
      "step": 7732
    },
    {
      "epoch": 0.9645752775352376,
      "grad_norm": 0.0,
      "learning_rate": 6.575327868275972e-08,
      "loss": 0.8909,
      "step": 7733
    },
    {
      "epoch": 0.9647000124734938,
      "grad_norm": 0.0,
      "learning_rate": 6.529154631766288e-08,
      "loss": 0.9185,
      "step": 7734
    },
    {
      "epoch": 0.9648247474117501,
      "grad_norm": 0.0,
      "learning_rate": 6.483143554743776e-08,
      "loss": 0.8981,
      "step": 7735
    },
    {
      "epoch": 0.9649494823500062,
      "grad_norm": 0.0,
      "learning_rate": 6.437294644718872e-08,
      "loss": 0.8773,
      "step": 7736
    },
    {
      "epoch": 0.9650742172882625,
      "grad_norm": 0.0,
      "learning_rate": 6.391607909174813e-08,
      "loss": 0.8735,
      "step": 7737
    },
    {
      "epoch": 0.9651989522265186,
      "grad_norm": 0.0,
      "learning_rate": 6.346083355569188e-08,
      "loss": 0.9003,
      "step": 7738
    },
    {
      "epoch": 0.9653236871647749,
      "grad_norm": 0.0,
      "learning_rate": 6.30072099133261e-08,
      "loss": 0.8931,
      "step": 7739
    },
    {
      "epoch": 0.9654484221030311,
      "grad_norm": 0.0,
      "learning_rate": 6.255520823869487e-08,
      "loss": 0.9295,
      "step": 7740
    },
    {
      "epoch": 0.9655731570412872,
      "grad_norm": 0.0,
      "learning_rate": 6.210482860557366e-08,
      "loss": 0.8875,
      "step": 7741
    },
    {
      "epoch": 0.9656978919795435,
      "grad_norm": 0.0,
      "learning_rate": 6.165607108747807e-08,
      "loss": 0.9029,
      "step": 7742
    },
    {
      "epoch": 0.9658226269177996,
      "grad_norm": 0.0,
      "learning_rate": 6.120893575765618e-08,
      "loss": 0.8948,
      "step": 7743
    },
    {
      "epoch": 0.9659473618560559,
      "grad_norm": 0.0,
      "learning_rate": 6.076342268909075e-08,
      "loss": 0.8757,
      "step": 7744
    },
    {
      "epoch": 0.966072096794312,
      "grad_norm": 0.0,
      "learning_rate": 6.031953195450136e-08,
      "loss": 0.8849,
      "step": 7745
    },
    {
      "epoch": 0.9661968317325683,
      "grad_norm": 0.0,
      "learning_rate": 5.987726362634339e-08,
      "loss": 0.8985,
      "step": 7746
    },
    {
      "epoch": 0.9663215666708245,
      "grad_norm": 0.0,
      "learning_rate": 5.943661777680354e-08,
      "loss": 0.8624,
      "step": 7747
    },
    {
      "epoch": 0.9664463016090807,
      "grad_norm": 0.0,
      "learning_rate": 5.8997594477809836e-08,
      "loss": 0.8854,
      "step": 7748
    },
    {
      "epoch": 0.9665710365473369,
      "grad_norm": 0.0,
      "learning_rate": 5.8560193801018294e-08,
      "loss": 0.9344,
      "step": 7749
    },
    {
      "epoch": 0.9666957714855932,
      "grad_norm": 0.0,
      "learning_rate": 5.812441581782624e-08,
      "loss": 0.8512,
      "step": 7750
    },
    {
      "epoch": 0.9668205064238493,
      "grad_norm": 0.0,
      "learning_rate": 5.769026059936233e-08,
      "loss": 0.953,
      "step": 7751
    },
    {
      "epoch": 0.9669452413621056,
      "grad_norm": 0.0,
      "learning_rate": 5.7257728216492114e-08,
      "loss": 0.8709,
      "step": 7752
    },
    {
      "epoch": 0.9670699763003617,
      "grad_norm": 0.0,
      "learning_rate": 5.682681873981577e-08,
      "loss": 0.9364,
      "step": 7753
    },
    {
      "epoch": 0.9671947112386179,
      "grad_norm": 0.0,
      "learning_rate": 5.6397532239668154e-08,
      "loss": 0.9063,
      "step": 7754
    },
    {
      "epoch": 0.9673194461768742,
      "grad_norm": 0.0,
      "learning_rate": 5.5969868786120983e-08,
      "loss": 0.8966,
      "step": 7755
    },
    {
      "epoch": 0.9674441811151303,
      "grad_norm": 0.0,
      "learning_rate": 5.554382844897843e-08,
      "loss": 0.8591,
      "step": 7756
    },
    {
      "epoch": 0.9675689160533866,
      "grad_norm": 0.0,
      "learning_rate": 5.5119411297781534e-08,
      "loss": 0.9179,
      "step": 7757
    },
    {
      "epoch": 0.9676936509916427,
      "grad_norm": 0.0,
      "learning_rate": 5.4696617401805984e-08,
      "loss": 0.8999,
      "step": 7758
    },
    {
      "epoch": 0.967818385929899,
      "grad_norm": 0.0,
      "learning_rate": 5.427544683006103e-08,
      "loss": 0.9069,
      "step": 7759
    },
    {
      "epoch": 0.9679431208681551,
      "grad_norm": 0.0,
      "learning_rate": 5.385589965129501e-08,
      "loss": 0.9313,
      "step": 7760
    },
    {
      "epoch": 0.9680678558064114,
      "grad_norm": 0.0,
      "learning_rate": 5.3437975933985366e-08,
      "loss": 0.8845,
      "step": 7761
    },
    {
      "epoch": 0.9681925907446676,
      "grad_norm": 0.0,
      "learning_rate": 5.302167574635087e-08,
      "loss": 0.8739,
      "step": 7762
    },
    {
      "epoch": 0.9683173256829238,
      "grad_norm": 0.0,
      "learning_rate": 5.260699915633938e-08,
      "loss": 0.8518,
      "step": 7763
    },
    {
      "epoch": 0.96844206062118,
      "grad_norm": 0.0,
      "learning_rate": 5.219394623163898e-08,
      "loss": 0.8796,
      "step": 7764
    },
    {
      "epoch": 0.9685667955594361,
      "grad_norm": 0.0,
      "learning_rate": 5.178251703967019e-08,
      "loss": 0.9339,
      "step": 7765
    },
    {
      "epoch": 0.9686915304976924,
      "grad_norm": 0.0,
      "learning_rate": 5.137271164758595e-08,
      "loss": 0.9018,
      "step": 7766
    },
    {
      "epoch": 0.9688162654359486,
      "grad_norm": 0.0,
      "learning_rate": 5.096453012227942e-08,
      "loss": 0.8565,
      "step": 7767
    },
    {
      "epoch": 0.9689410003742048,
      "grad_norm": 0.0,
      "learning_rate": 5.0557972530375086e-08,
      "loss": 0.9718,
      "step": 7768
    },
    {
      "epoch": 0.969065735312461,
      "grad_norm": 0.0,
      "learning_rate": 5.015303893823431e-08,
      "loss": 0.8612,
      "step": 7769
    },
    {
      "epoch": 0.9691904702507173,
      "grad_norm": 0.0,
      "learning_rate": 4.974972941195089e-08,
      "loss": 0.8851,
      "step": 7770
    },
    {
      "epoch": 0.9693152051889734,
      "grad_norm": 0.0,
      "learning_rate": 4.93480440173566e-08,
      "loss": 0.9046,
      "step": 7771
    },
    {
      "epoch": 0.9694399401272297,
      "grad_norm": 0.0,
      "learning_rate": 4.894798282001567e-08,
      "loss": 0.9093,
      "step": 7772
    },
    {
      "epoch": 0.9695646750654858,
      "grad_norm": 0.0,
      "learning_rate": 4.85495458852292e-08,
      "loss": 0.8918,
      "step": 7773
    },
    {
      "epoch": 0.9696894100037421,
      "grad_norm": 0.0,
      "learning_rate": 4.815273327803183e-08,
      "loss": 0.9367,
      "step": 7774
    },
    {
      "epoch": 0.9698141449419982,
      "grad_norm": 0.0,
      "learning_rate": 4.775754506319175e-08,
      "loss": 0.9284,
      "step": 7775
    },
    {
      "epoch": 0.9699388798802545,
      "grad_norm": 0.0,
      "learning_rate": 4.736398130521513e-08,
      "loss": 0.8901,
      "step": 7776
    },
    {
      "epoch": 0.9700636148185107,
      "grad_norm": 0.0,
      "learning_rate": 4.6972042068341714e-08,
      "loss": 0.86,
      "step": 7777
    },
    {
      "epoch": 0.9701883497567668,
      "grad_norm": 0.0,
      "learning_rate": 4.6581727416544766e-08,
      "loss": 0.9244,
      "step": 7778
    },
    {
      "epoch": 0.9703130846950231,
      "grad_norm": 0.0,
      "learning_rate": 4.619303741353554e-08,
      "loss": 0.9051,
      "step": 7779
    },
    {
      "epoch": 0.9704378196332792,
      "grad_norm": 0.0,
      "learning_rate": 4.580597212275661e-08,
      "loss": 0.9047,
      "step": 7780
    },
    {
      "epoch": 0.9705625545715355,
      "grad_norm": 0.0,
      "learning_rate": 4.5420531607385244e-08,
      "loss": 0.896,
      "step": 7781
    },
    {
      "epoch": 0.9706872895097917,
      "grad_norm": 0.0,
      "learning_rate": 4.503671593033776e-08,
      "loss": 0.9003,
      "step": 7782
    },
    {
      "epoch": 0.9708120244480479,
      "grad_norm": 0.0,
      "learning_rate": 4.4654525154260717e-08,
      "loss": 0.9246,
      "step": 7783
    },
    {
      "epoch": 0.9709367593863041,
      "grad_norm": 0.0,
      "learning_rate": 4.4273959341538666e-08,
      "loss": 0.8897,
      "step": 7784
    },
    {
      "epoch": 0.9710614943245603,
      "grad_norm": 0.0,
      "learning_rate": 4.389501855428857e-08,
      "loss": 0.9427,
      "step": 7785
    },
    {
      "epoch": 0.9711862292628165,
      "grad_norm": 0.0,
      "learning_rate": 4.35177028543643e-08,
      "loss": 0.8948,
      "step": 7786
    },
    {
      "epoch": 0.9713109642010728,
      "grad_norm": 0.0,
      "learning_rate": 4.314201230335102e-08,
      "loss": 0.8841,
      "step": 7787
    },
    {
      "epoch": 0.9714356991393289,
      "grad_norm": 0.0,
      "learning_rate": 4.2767946962573025e-08,
      "loss": 0.9072,
      "step": 7788
    },
    {
      "epoch": 0.9715604340775851,
      "grad_norm": 0.0,
      "learning_rate": 4.2395506893085916e-08,
      "loss": 0.8639,
      "step": 7789
    },
    {
      "epoch": 0.9716851690158413,
      "grad_norm": 0.0,
      "learning_rate": 4.202469215568217e-08,
      "loss": 0.9051,
      "step": 7790
    },
    {
      "epoch": 0.9718099039540975,
      "grad_norm": 0.0,
      "learning_rate": 4.1655502810888926e-08,
      "loss": 0.9275,
      "step": 7791
    },
    {
      "epoch": 0.9719346388923538,
      "grad_norm": 0.0,
      "learning_rate": 4.128793891896465e-08,
      "loss": 0.9081,
      "step": 7792
    },
    {
      "epoch": 0.9720593738306099,
      "grad_norm": 0.0,
      "learning_rate": 4.0922000539906914e-08,
      "loss": 0.8862,
      "step": 7793
    },
    {
      "epoch": 0.9721841087688662,
      "grad_norm": 0.0,
      "learning_rate": 4.0557687733445704e-08,
      "loss": 0.9036,
      "step": 7794
    },
    {
      "epoch": 0.9723088437071223,
      "grad_norm": 0.0,
      "learning_rate": 4.019500055904568e-08,
      "loss": 0.9608,
      "step": 7795
    },
    {
      "epoch": 0.9724335786453786,
      "grad_norm": 0.0,
      "learning_rate": 3.983393907590727e-08,
      "loss": 0.8862,
      "step": 7796
    },
    {
      "epoch": 0.9725583135836348,
      "grad_norm": 0.0,
      "learning_rate": 3.947450334296443e-08,
      "loss": 0.8921,
      "step": 7797
    },
    {
      "epoch": 0.972683048521891,
      "grad_norm": 0.0,
      "learning_rate": 3.911669341888469e-08,
      "loss": 0.9224,
      "step": 7798
    },
    {
      "epoch": 0.9728077834601472,
      "grad_norm": 0.0,
      "learning_rate": 3.876050936207465e-08,
      "loss": 0.9133,
      "step": 7799
    },
    {
      "epoch": 0.9729325183984034,
      "grad_norm": 0.0,
      "learning_rate": 3.8405951230670035e-08,
      "loss": 0.9125,
      "step": 7800
    },
    {
      "epoch": 0.9730572533366596,
      "grad_norm": 0.0,
      "learning_rate": 3.805301908254455e-08,
      "loss": 0.9039,
      "step": 7801
    },
    {
      "epoch": 0.9731819882749158,
      "grad_norm": 0.0,
      "learning_rate": 3.770171297530545e-08,
      "loss": 0.8742,
      "step": 7802
    },
    {
      "epoch": 0.973306723213172,
      "grad_norm": 0.0,
      "learning_rate": 3.735203296629575e-08,
      "loss": 0.8548,
      "step": 7803
    },
    {
      "epoch": 0.9734314581514282,
      "grad_norm": 0.0,
      "learning_rate": 3.70039791125909e-08,
      "loss": 0.8835,
      "step": 7804
    },
    {
      "epoch": 0.9735561930896844,
      "grad_norm": 0.0,
      "learning_rate": 3.665755147100214e-08,
      "loss": 0.8854,
      "step": 7805
    },
    {
      "epoch": 0.9736809280279406,
      "grad_norm": 0.0,
      "learning_rate": 3.6312750098076445e-08,
      "loss": 0.9352,
      "step": 7806
    },
    {
      "epoch": 0.9738056629661969,
      "grad_norm": 0.0,
      "learning_rate": 3.5969575050091024e-08,
      "loss": 0.9126,
      "step": 7807
    },
    {
      "epoch": 0.973930397904453,
      "grad_norm": 0.0,
      "learning_rate": 3.56280263830644e-08,
      "loss": 0.877,
      "step": 7808
    },
    {
      "epoch": 0.9740551328427093,
      "grad_norm": 0.0,
      "learning_rate": 3.528810415274309e-08,
      "loss": 0.9235,
      "step": 7809
    },
    {
      "epoch": 0.9741798677809654,
      "grad_norm": 0.0,
      "learning_rate": 3.4949808414612705e-08,
      "loss": 0.9081,
      "step": 7810
    },
    {
      "epoch": 0.9743046027192217,
      "grad_norm": 0.0,
      "learning_rate": 3.461313922389131e-08,
      "loss": 0.8242,
      "step": 7811
    },
    {
      "epoch": 0.9744293376574779,
      "grad_norm": 0.0,
      "learning_rate": 3.427809663553161e-08,
      "loss": 0.8448,
      "step": 7812
    },
    {
      "epoch": 0.974554072595734,
      "grad_norm": 0.0,
      "learning_rate": 3.394468070422097e-08,
      "loss": 0.9392,
      "step": 7813
    },
    {
      "epoch": 0.9746788075339903,
      "grad_norm": 0.0,
      "learning_rate": 3.36128914843814e-08,
      "loss": 0.9167,
      "step": 7814
    },
    {
      "epoch": 0.9748035424722464,
      "grad_norm": 0.0,
      "learning_rate": 3.328272903016849e-08,
      "loss": 0.8646,
      "step": 7815
    },
    {
      "epoch": 0.9749282774105027,
      "grad_norm": 0.0,
      "learning_rate": 3.2954193395473566e-08,
      "loss": 0.8904,
      "step": 7816
    },
    {
      "epoch": 0.9750530123487589,
      "grad_norm": 0.0,
      "learning_rate": 3.262728463392262e-08,
      "loss": 0.8839,
      "step": 7817
    },
    {
      "epoch": 0.9751777472870151,
      "grad_norm": 0.0,
      "learning_rate": 3.230200279887519e-08,
      "loss": 0.9266,
      "step": 7818
    },
    {
      "epoch": 0.9753024822252713,
      "grad_norm": 0.0,
      "learning_rate": 3.197834794342436e-08,
      "loss": 0.9223,
      "step": 7819
    },
    {
      "epoch": 0.9754272171635275,
      "grad_norm": 0.0,
      "learning_rate": 3.1656320120400094e-08,
      "loss": 0.8816,
      "step": 7820
    },
    {
      "epoch": 0.9755519521017837,
      "grad_norm": 0.0,
      "learning_rate": 3.13359193823648e-08,
      "loss": 0.9249,
      "step": 7821
    },
    {
      "epoch": 0.97567668704004,
      "grad_norm": 0.0,
      "learning_rate": 3.101714578161552e-08,
      "loss": 0.8956,
      "step": 7822
    },
    {
      "epoch": 0.9758014219782961,
      "grad_norm": 0.0,
      "learning_rate": 3.069999937018508e-08,
      "loss": 0.8872,
      "step": 7823
    },
    {
      "epoch": 0.9759261569165524,
      "grad_norm": 0.0,
      "learning_rate": 3.038448019983986e-08,
      "loss": 0.9167,
      "step": 7824
    },
    {
      "epoch": 0.9760508918548085,
      "grad_norm": 0.0,
      "learning_rate": 3.0070588322079765e-08,
      "loss": 0.8777,
      "step": 7825
    },
    {
      "epoch": 0.9761756267930647,
      "grad_norm": 0.0,
      "learning_rate": 2.9758323788140475e-08,
      "loss": 0.8657,
      "step": 7826
    },
    {
      "epoch": 0.976300361731321,
      "grad_norm": 0.0,
      "learning_rate": 2.9447686648990115e-08,
      "loss": 0.8782,
      "step": 7827
    },
    {
      "epoch": 0.9764250966695771,
      "grad_norm": 0.0,
      "learning_rate": 2.9138676955333676e-08,
      "loss": 0.9222,
      "step": 7828
    },
    {
      "epoch": 0.9765498316078334,
      "grad_norm": 0.0,
      "learning_rate": 2.8831294757609707e-08,
      "loss": 0.8781,
      "step": 7829
    },
    {
      "epoch": 0.9766745665460895,
      "grad_norm": 0.0,
      "learning_rate": 2.852554010598918e-08,
      "loss": 0.9293,
      "step": 7830
    },
    {
      "epoch": 0.9767993014843458,
      "grad_norm": 0.0,
      "learning_rate": 2.8221413050381064e-08,
      "loss": 0.9366,
      "step": 7831
    },
    {
      "epoch": 0.976924036422602,
      "grad_norm": 0.0,
      "learning_rate": 2.791891364042565e-08,
      "loss": 0.971,
      "step": 7832
    },
    {
      "epoch": 0.9770487713608582,
      "grad_norm": 0.0,
      "learning_rate": 2.7618041925496775e-08,
      "loss": 0.8669,
      "step": 7833
    },
    {
      "epoch": 0.9771735062991144,
      "grad_norm": 0.0,
      "learning_rate": 2.7318797954707378e-08,
      "loss": 0.8849,
      "step": 7834
    },
    {
      "epoch": 0.9772982412373706,
      "grad_norm": 0.0,
      "learning_rate": 2.7021181776899495e-08,
      "loss": 0.9066,
      "step": 7835
    },
    {
      "epoch": 0.9774229761756268,
      "grad_norm": 0.0,
      "learning_rate": 2.6725193440650942e-08,
      "loss": 0.9093,
      "step": 7836
    },
    {
      "epoch": 0.9775477111138829,
      "grad_norm": 0.0,
      "learning_rate": 2.643083299427751e-08,
      "loss": 0.9029,
      "step": 7837
    },
    {
      "epoch": 0.9776724460521392,
      "grad_norm": 0.0,
      "learning_rate": 2.6138100485822994e-08,
      "loss": 0.94,
      "step": 7838
    },
    {
      "epoch": 0.9777971809903954,
      "grad_norm": 0.0,
      "learning_rate": 2.5846995963071387e-08,
      "loss": 0.894,
      "step": 7839
    },
    {
      "epoch": 0.9779219159286516,
      "grad_norm": 0.0,
      "learning_rate": 2.5557519473535796e-08,
      "loss": 0.8974,
      "step": 7840
    },
    {
      "epoch": 0.9780466508669078,
      "grad_norm": 0.0,
      "learning_rate": 2.5269671064467315e-08,
      "loss": 0.8398,
      "step": 7841
    },
    {
      "epoch": 0.978171385805164,
      "grad_norm": 0.0,
      "learning_rate": 2.498345078285058e-08,
      "loss": 0.9112,
      "step": 7842
    },
    {
      "epoch": 0.9782961207434202,
      "grad_norm": 0.0,
      "learning_rate": 2.469885867540378e-08,
      "loss": 0.9315,
      "step": 7843
    },
    {
      "epoch": 0.9784208556816765,
      "grad_norm": 0.0,
      "learning_rate": 2.4415894788578642e-08,
      "loss": 0.9008,
      "step": 7844
    },
    {
      "epoch": 0.9785455906199326,
      "grad_norm": 0.0,
      "learning_rate": 2.4134559168562666e-08,
      "loss": 0.8574,
      "step": 7845
    },
    {
      "epoch": 0.9786703255581889,
      "grad_norm": 0.0,
      "learning_rate": 2.3854851861276895e-08,
      "loss": 0.9293,
      "step": 7846
    },
    {
      "epoch": 0.978795060496445,
      "grad_norm": 0.0,
      "learning_rate": 2.357677291237592e-08,
      "loss": 0.8705,
      "step": 7847
    },
    {
      "epoch": 0.9789197954347013,
      "grad_norm": 0.0,
      "learning_rate": 2.3300322367251215e-08,
      "loss": 0.8992,
      "step": 7848
    },
    {
      "epoch": 0.9790445303729575,
      "grad_norm": 0.0,
      "learning_rate": 2.302550027102335e-08,
      "loss": 0.9168,
      "step": 7849
    },
    {
      "epoch": 0.9791692653112136,
      "grad_norm": 0.0,
      "learning_rate": 2.2752306668553102e-08,
      "loss": 0.8829,
      "step": 7850
    },
    {
      "epoch": 0.9792940002494699,
      "grad_norm": 0.0,
      "learning_rate": 2.2480741604430368e-08,
      "loss": 0.9285,
      "step": 7851
    },
    {
      "epoch": 0.979418735187726,
      "grad_norm": 0.0,
      "learning_rate": 2.2210805122983013e-08,
      "loss": 0.8863,
      "step": 7852
    },
    {
      "epoch": 0.9795434701259823,
      "grad_norm": 0.0,
      "learning_rate": 2.1942497268270245e-08,
      "loss": 0.8889,
      "step": 7853
    },
    {
      "epoch": 0.9796682050642385,
      "grad_norm": 0.0,
      "learning_rate": 2.1675818084089252e-08,
      "loss": 0.8897,
      "step": 7854
    },
    {
      "epoch": 0.9797929400024947,
      "grad_norm": 0.0,
      "learning_rate": 2.1410767613965212e-08,
      "loss": 0.9029,
      "step": 7855
    },
    {
      "epoch": 0.9799176749407509,
      "grad_norm": 0.0,
      "learning_rate": 2.1147345901162407e-08,
      "loss": 0.9085,
      "step": 7856
    },
    {
      "epoch": 0.9800424098790071,
      "grad_norm": 0.0,
      "learning_rate": 2.088555298867978e-08,
      "loss": 0.9842,
      "step": 7857
    },
    {
      "epoch": 0.9801671448172633,
      "grad_norm": 0.0,
      "learning_rate": 2.062538891924537e-08,
      "loss": 0.859,
      "step": 7858
    },
    {
      "epoch": 0.9802918797555196,
      "grad_norm": 0.0,
      "learning_rate": 2.036685373532743e-08,
      "loss": 0.8708,
      "step": 7859
    },
    {
      "epoch": 0.9804166146937757,
      "grad_norm": 0.0,
      "learning_rate": 2.0109947479123315e-08,
      "loss": 0.9026,
      "step": 7860
    },
    {
      "epoch": 0.9805413496320319,
      "grad_norm": 0.0,
      "learning_rate": 1.985467019256726e-08,
      "loss": 0.8809,
      "step": 7861
    },
    {
      "epoch": 0.9806660845702881,
      "grad_norm": 0.0,
      "learning_rate": 1.9601021917327045e-08,
      "loss": 0.9482,
      "step": 7862
    },
    {
      "epoch": 0.9807908195085443,
      "grad_norm": 0.0,
      "learning_rate": 1.9349002694805107e-08,
      "loss": 0.8909,
      "step": 7863
    },
    {
      "epoch": 0.9809155544468006,
      "grad_norm": 0.0,
      "learning_rate": 1.909861256613632e-08,
      "loss": 0.8499,
      "step": 7864
    },
    {
      "epoch": 0.9810402893850567,
      "grad_norm": 0.0,
      "learning_rate": 1.8849851572191325e-08,
      "loss": 0.9245,
      "step": 7865
    },
    {
      "epoch": 0.981165024323313,
      "grad_norm": 0.0,
      "learning_rate": 1.8602719753573196e-08,
      "loss": 0.9164,
      "step": 7866
    },
    {
      "epoch": 0.9812897592615691,
      "grad_norm": 0.0,
      "learning_rate": 1.8357217150620775e-08,
      "loss": 0.8717,
      "step": 7867
    },
    {
      "epoch": 0.9814144941998254,
      "grad_norm": 0.0,
      "learning_rate": 1.811334380340757e-08,
      "loss": 0.8924,
      "step": 7868
    },
    {
      "epoch": 0.9815392291380816,
      "grad_norm": 0.0,
      "learning_rate": 1.7871099751737287e-08,
      "loss": 0.9328,
      "step": 7869
    },
    {
      "epoch": 0.9816639640763378,
      "grad_norm": 0.0,
      "learning_rate": 1.7630485035151635e-08,
      "loss": 0.9034,
      "step": 7870
    },
    {
      "epoch": 0.981788699014594,
      "grad_norm": 0.0,
      "learning_rate": 1.739149969292586e-08,
      "loss": 0.8827,
      "step": 7871
    },
    {
      "epoch": 0.9819134339528502,
      "grad_norm": 0.0,
      "learning_rate": 1.7154143764067653e-08,
      "loss": 0.8815,
      "step": 7872
    },
    {
      "epoch": 0.9820381688911064,
      "grad_norm": 0.0,
      "learning_rate": 1.6918417287318245e-08,
      "loss": 0.8992,
      "step": 7873
    },
    {
      "epoch": 0.9821629038293626,
      "grad_norm": 0.0,
      "learning_rate": 1.6684320301155744e-08,
      "loss": 0.9141,
      "step": 7874
    },
    {
      "epoch": 0.9822876387676188,
      "grad_norm": 0.0,
      "learning_rate": 1.6451852843789585e-08,
      "loss": 0.8656,
      "step": 7875
    },
    {
      "epoch": 0.982412373705875,
      "grad_norm": 0.0,
      "learning_rate": 1.622101495316497e-08,
      "loss": 0.9087,
      "step": 7876
    },
    {
      "epoch": 0.9825371086441312,
      "grad_norm": 0.0,
      "learning_rate": 1.599180666696065e-08,
      "loss": 0.8736,
      "step": 7877
    },
    {
      "epoch": 0.9826618435823874,
      "grad_norm": 0.0,
      "learning_rate": 1.5764228022588924e-08,
      "loss": 0.9154,
      "step": 7878
    },
    {
      "epoch": 0.9827865785206437,
      "grad_norm": 0.0,
      "learning_rate": 1.5538279057196736e-08,
      "loss": 0.8951,
      "step": 7879
    },
    {
      "epoch": 0.9829113134588998,
      "grad_norm": 0.0,
      "learning_rate": 1.5313959807663482e-08,
      "loss": 0.8482,
      "step": 7880
    },
    {
      "epoch": 0.9830360483971561,
      "grad_norm": 0.0,
      "learning_rate": 1.509127031060431e-08,
      "loss": 0.9096,
      "step": 7881
    },
    {
      "epoch": 0.9831607833354122,
      "grad_norm": 0.0,
      "learning_rate": 1.487021060236904e-08,
      "loss": 0.8942,
      "step": 7882
    },
    {
      "epoch": 0.9832855182736685,
      "grad_norm": 0.0,
      "learning_rate": 1.4650780719038804e-08,
      "loss": 0.8728,
      "step": 7883
    },
    {
      "epoch": 0.9834102532119247,
      "grad_norm": 0.0,
      "learning_rate": 1.44329806964294e-08,
      "loss": 0.9246,
      "step": 7884
    },
    {
      "epoch": 0.9835349881501808,
      "grad_norm": 0.0,
      "learning_rate": 1.4216810570093498e-08,
      "loss": 0.9511,
      "step": 7885
    },
    {
      "epoch": 0.9836597230884371,
      "grad_norm": 0.0,
      "learning_rate": 1.4002270375313986e-08,
      "loss": 0.9524,
      "step": 7886
    },
    {
      "epoch": 0.9837844580266932,
      "grad_norm": 0.0,
      "learning_rate": 1.378936014710841e-08,
      "loss": 0.9212,
      "step": 7887
    },
    {
      "epoch": 0.9839091929649495,
      "grad_norm": 0.0,
      "learning_rate": 1.3578079920231192e-08,
      "loss": 0.8966,
      "step": 7888
    },
    {
      "epoch": 0.9840339279032057,
      "grad_norm": 0.0,
      "learning_rate": 1.3368429729168075e-08,
      "loss": 0.8629,
      "step": 7889
    },
    {
      "epoch": 0.9841586628414619,
      "grad_norm": 0.0,
      "learning_rate": 1.3160409608137249e-08,
      "loss": 0.8892,
      "step": 7890
    },
    {
      "epoch": 0.9842833977797181,
      "grad_norm": 0.0,
      "learning_rate": 1.2954019591095989e-08,
      "loss": 0.9285,
      "step": 7891
    },
    {
      "epoch": 0.9844081327179743,
      "grad_norm": 0.0,
      "learning_rate": 1.2749259711729577e-08,
      "loss": 0.8867,
      "step": 7892
    },
    {
      "epoch": 0.9845328676562305,
      "grad_norm": 0.0,
      "learning_rate": 1.2546130003461276e-08,
      "loss": 0.9011,
      "step": 7893
    },
    {
      "epoch": 0.9846576025944868,
      "grad_norm": 0.0,
      "learning_rate": 1.2344630499447895e-08,
      "loss": 0.9644,
      "step": 7894
    },
    {
      "epoch": 0.9847823375327429,
      "grad_norm": 0.0,
      "learning_rate": 1.2144761232577573e-08,
      "loss": 0.9497,
      "step": 7895
    },
    {
      "epoch": 0.9849070724709992,
      "grad_norm": 0.0,
      "learning_rate": 1.1946522235474212e-08,
      "loss": 0.9315,
      "step": 7896
    },
    {
      "epoch": 0.9850318074092553,
      "grad_norm": 0.0,
      "learning_rate": 1.1749913540496372e-08,
      "loss": 0.886,
      "step": 7897
    },
    {
      "epoch": 0.9851565423475115,
      "grad_norm": 0.0,
      "learning_rate": 1.155493517973394e-08,
      "loss": 0.9215,
      "step": 7898
    },
    {
      "epoch": 0.9852812772857678,
      "grad_norm": 0.0,
      "learning_rate": 1.1361587185014788e-08,
      "loss": 0.8804,
      "step": 7899
    },
    {
      "epoch": 0.9854060122240239,
      "grad_norm": 0.0,
      "learning_rate": 1.1169869587895899e-08,
      "loss": 0.9167,
      "step": 7900
    },
    {
      "epoch": 0.9855307471622802,
      "grad_norm": 0.0,
      "learning_rate": 1.0979782419671126e-08,
      "loss": 0.835,
      "step": 7901
    },
    {
      "epoch": 0.9856554821005363,
      "grad_norm": 0.0,
      "learning_rate": 1.0791325711367872e-08,
      "loss": 0.8562,
      "step": 7902
    },
    {
      "epoch": 0.9857802170387926,
      "grad_norm": 0.0,
      "learning_rate": 1.060449949374709e-08,
      "loss": 0.8799,
      "step": 7903
    },
    {
      "epoch": 0.9859049519770487,
      "grad_norm": 0.0,
      "learning_rate": 1.041930379730327e-08,
      "loss": 0.8679,
      "step": 7904
    },
    {
      "epoch": 0.986029686915305,
      "grad_norm": 0.0,
      "learning_rate": 1.023573865226446e-08,
      "loss": 0.9221,
      "step": 7905
    },
    {
      "epoch": 0.9861544218535612,
      "grad_norm": 0.0,
      "learning_rate": 1.0053804088594465e-08,
      "loss": 0.9329,
      "step": 7906
    },
    {
      "epoch": 0.9862791567918174,
      "grad_norm": 0.0,
      "learning_rate": 9.873500135987312e-09,
      "loss": 0.9115,
      "step": 7907
    },
    {
      "epoch": 0.9864038917300736,
      "grad_norm": 0.0,
      "learning_rate": 9.694826823876125e-09,
      "loss": 0.8862,
      "step": 7908
    },
    {
      "epoch": 0.9865286266683297,
      "grad_norm": 0.0,
      "learning_rate": 9.517784181422018e-09,
      "loss": 0.9125,
      "step": 7909
    },
    {
      "epoch": 0.986653361606586,
      "grad_norm": 0.0,
      "learning_rate": 9.34237223752521e-09,
      "loss": 0.9544,
      "step": 7910
    },
    {
      "epoch": 0.9867780965448422,
      "grad_norm": 0.0,
      "learning_rate": 9.16859102081613e-09,
      "loss": 0.8789,
      "step": 7911
    },
    {
      "epoch": 0.9869028314830984,
      "grad_norm": 0.0,
      "learning_rate": 8.996440559659869e-09,
      "loss": 0.9045,
      "step": 7912
    },
    {
      "epoch": 0.9870275664213546,
      "grad_norm": 0.0,
      "learning_rate": 8.82592088215728e-09,
      "loss": 0.8678,
      "step": 7913
    },
    {
      "epoch": 0.9871523013596109,
      "grad_norm": 0.0,
      "learning_rate": 8.657032016139433e-09,
      "loss": 0.8873,
      "step": 7914
    },
    {
      "epoch": 0.987277036297867,
      "grad_norm": 0.0,
      "learning_rate": 8.489773989174276e-09,
      "loss": 0.9097,
      "step": 7915
    },
    {
      "epoch": 0.9874017712361233,
      "grad_norm": 0.0,
      "learning_rate": 8.324146828563307e-09,
      "loss": 0.9168,
      "step": 7916
    },
    {
      "epoch": 0.9875265061743794,
      "grad_norm": 0.0,
      "learning_rate": 8.160150561340452e-09,
      "loss": 0.8912,
      "step": 7917
    },
    {
      "epoch": 0.9876512411126357,
      "grad_norm": 0.0,
      "learning_rate": 7.997785214273191e-09,
      "loss": 0.9099,
      "step": 7918
    },
    {
      "epoch": 0.9877759760508918,
      "grad_norm": 0.0,
      "learning_rate": 7.837050813863655e-09,
      "loss": 0.8735,
      "step": 7919
    },
    {
      "epoch": 0.9879007109891481,
      "grad_norm": 0.0,
      "learning_rate": 7.677947386349749e-09,
      "loss": 0.9124,
      "step": 7920
    },
    {
      "epoch": 0.9880254459274043,
      "grad_norm": 0.0,
      "learning_rate": 7.520474957699586e-09,
      "loss": 0.9486,
      "step": 7921
    },
    {
      "epoch": 0.9881501808656604,
      "grad_norm": 0.0,
      "learning_rate": 7.3646335536159406e-09,
      "loss": 0.8811,
      "step": 7922
    },
    {
      "epoch": 0.9882749158039167,
      "grad_norm": 0.0,
      "learning_rate": 7.210423199538464e-09,
      "loss": 0.8515,
      "step": 7923
    },
    {
      "epoch": 0.9883996507421728,
      "grad_norm": 0.0,
      "learning_rate": 7.05784392063591e-09,
      "loss": 0.8952,
      "step": 7924
    },
    {
      "epoch": 0.9885243856804291,
      "grad_norm": 0.0,
      "learning_rate": 6.906895741813913e-09,
      "loss": 0.8921,
      "step": 7925
    },
    {
      "epoch": 0.9886491206186853,
      "grad_norm": 0.0,
      "learning_rate": 6.757578687710542e-09,
      "loss": 0.8559,
      "step": 7926
    },
    {
      "epoch": 0.9887738555569415,
      "grad_norm": 0.0,
      "learning_rate": 6.609892782699634e-09,
      "loss": 0.9066,
      "step": 7927
    },
    {
      "epoch": 0.9888985904951977,
      "grad_norm": 0.0,
      "learning_rate": 6.463838050885241e-09,
      "loss": 0.8654,
      "step": 7928
    },
    {
      "epoch": 0.989023325433454,
      "grad_norm": 0.0,
      "learning_rate": 6.319414516108291e-09,
      "loss": 0.8514,
      "step": 7929
    },
    {
      "epoch": 0.9891480603717101,
      "grad_norm": 0.0,
      "learning_rate": 6.176622201942151e-09,
      "loss": 0.9158,
      "step": 7930
    },
    {
      "epoch": 0.9892727953099664,
      "grad_norm": 0.0,
      "learning_rate": 6.035461131693732e-09,
      "loss": 0.8874,
      "step": 7931
    },
    {
      "epoch": 0.9893975302482225,
      "grad_norm": 0.0,
      "learning_rate": 5.895931328405713e-09,
      "loss": 0.918,
      "step": 7932
    },
    {
      "epoch": 0.9895222651864787,
      "grad_norm": 0.0,
      "learning_rate": 5.758032814850989e-09,
      "loss": 0.8788,
      "step": 7933
    },
    {
      "epoch": 0.9896470001247349,
      "grad_norm": 0.0,
      "learning_rate": 5.6217656135382214e-09,
      "loss": 0.9283,
      "step": 7934
    },
    {
      "epoch": 0.9897717350629911,
      "grad_norm": 0.0,
      "learning_rate": 5.4871297467107285e-09,
      "loss": 0.9546,
      "step": 7935
    },
    {
      "epoch": 0.9898964700012474,
      "grad_norm": 0.0,
      "learning_rate": 5.354125236343155e-09,
      "loss": 0.8538,
      "step": 7936
    },
    {
      "epoch": 0.9900212049395035,
      "grad_norm": 0.0,
      "learning_rate": 5.222752104147022e-09,
      "loss": 0.9094,
      "step": 7937
    },
    {
      "epoch": 0.9901459398777598,
      "grad_norm": 0.0,
      "learning_rate": 5.093010371564066e-09,
      "loss": 0.9857,
      "step": 7938
    },
    {
      "epoch": 0.9902706748160159,
      "grad_norm": 0.0,
      "learning_rate": 4.964900059771794e-09,
      "loss": 0.8774,
      "step": 7939
    },
    {
      "epoch": 0.9903954097542722,
      "grad_norm": 0.0,
      "learning_rate": 4.838421189681253e-09,
      "loss": 0.8955,
      "step": 7940
    },
    {
      "epoch": 0.9905201446925284,
      "grad_norm": 0.0,
      "learning_rate": 4.713573781938152e-09,
      "loss": 0.9069,
      "step": 7941
    },
    {
      "epoch": 0.9906448796307846,
      "grad_norm": 0.0,
      "learning_rate": 4.5903578569184146e-09,
      "loss": 0.9288,
      "step": 7942
    },
    {
      "epoch": 0.9907696145690408,
      "grad_norm": 0.0,
      "learning_rate": 4.468773434734841e-09,
      "loss": 0.9408,
      "step": 7943
    },
    {
      "epoch": 0.990894349507297,
      "grad_norm": 0.0,
      "learning_rate": 4.348820535233778e-09,
      "loss": 0.8892,
      "step": 7944
    },
    {
      "epoch": 0.9910190844455532,
      "grad_norm": 0.0,
      "learning_rate": 4.230499177994007e-09,
      "loss": 0.8844,
      "step": 7945
    },
    {
      "epoch": 0.9911438193838094,
      "grad_norm": 0.0,
      "learning_rate": 4.1138093823289705e-09,
      "loss": 0.9492,
      "step": 7946
    },
    {
      "epoch": 0.9912685543220656,
      "grad_norm": 0.0,
      "learning_rate": 3.998751167284543e-09,
      "loss": 0.885,
      "step": 7947
    },
    {
      "epoch": 0.9913932892603218,
      "grad_norm": 0.0,
      "learning_rate": 3.885324551642367e-09,
      "loss": 0.9219,
      "step": 7948
    },
    {
      "epoch": 0.991518024198578,
      "grad_norm": 0.0,
      "learning_rate": 3.773529553916522e-09,
      "loss": 0.8798,
      "step": 7949
    },
    {
      "epoch": 0.9916427591368342,
      "grad_norm": 0.0,
      "learning_rate": 3.6633661923524133e-09,
      "loss": 0.8823,
      "step": 7950
    },
    {
      "epoch": 0.9917674940750905,
      "grad_norm": 0.0,
      "learning_rate": 3.5548344849345438e-09,
      "loss": 0.8941,
      "step": 7951
    },
    {
      "epoch": 0.9918922290133466,
      "grad_norm": 0.0,
      "learning_rate": 3.447934449375412e-09,
      "loss": 0.8966,
      "step": 7952
    },
    {
      "epoch": 0.9920169639516029,
      "grad_norm": 0.0,
      "learning_rate": 3.3426661031255024e-09,
      "loss": 0.902,
      "step": 7953
    },
    {
      "epoch": 0.992141698889859,
      "grad_norm": 0.0,
      "learning_rate": 3.239029463367738e-09,
      "loss": 0.9254,
      "step": 7954
    },
    {
      "epoch": 0.9922664338281153,
      "grad_norm": 0.0,
      "learning_rate": 3.137024547016365e-09,
      "loss": 0.8662,
      "step": 7955
    },
    {
      "epoch": 0.9923911687663715,
      "grad_norm": 0.0,
      "learning_rate": 3.0366513707213995e-09,
      "loss": 0.865,
      "step": 7956
    },
    {
      "epoch": 0.9925159037046276,
      "grad_norm": 0.0,
      "learning_rate": 2.9379099508675125e-09,
      "loss": 0.8738,
      "step": 7957
    },
    {
      "epoch": 0.9926406386428839,
      "grad_norm": 0.0,
      "learning_rate": 2.8408003035718114e-09,
      "loss": 0.9308,
      "step": 7958
    },
    {
      "epoch": 0.99276537358114,
      "grad_norm": 0.0,
      "learning_rate": 2.7453224446838402e-09,
      "loss": 0.8843,
      "step": 7959
    },
    {
      "epoch": 0.9928901085193963,
      "grad_norm": 0.0,
      "learning_rate": 2.6514763897889096e-09,
      "loss": 0.9275,
      "step": 7960
    },
    {
      "epoch": 0.9930148434576525,
      "grad_norm": 0.0,
      "learning_rate": 2.559262154204767e-09,
      "loss": 0.8948,
      "step": 7961
    },
    {
      "epoch": 0.9931395783959087,
      "grad_norm": 0.0,
      "learning_rate": 2.4686797529827055e-09,
      "loss": 0.9405,
      "step": 7962
    },
    {
      "epoch": 0.9932643133341649,
      "grad_norm": 0.0,
      "learning_rate": 2.379729200908676e-09,
      "loss": 0.9327,
      "step": 7963
    },
    {
      "epoch": 0.9933890482724211,
      "grad_norm": 0.0,
      "learning_rate": 2.2924105125021746e-09,
      "loss": 0.8432,
      "step": 7964
    },
    {
      "epoch": 0.9935137832106773,
      "grad_norm": 0.0,
      "learning_rate": 2.2067237020140243e-09,
      "loss": 0.8749,
      "step": 7965
    },
    {
      "epoch": 0.9936385181489336,
      "grad_norm": 0.0,
      "learning_rate": 2.122668783431925e-09,
      "loss": 0.8908,
      "step": 7966
    },
    {
      "epoch": 0.9937632530871897,
      "grad_norm": 0.0,
      "learning_rate": 2.0402457704749024e-09,
      "loss": 0.8775,
      "step": 7967
    },
    {
      "epoch": 0.993887988025446,
      "grad_norm": 0.0,
      "learning_rate": 1.9594546765977495e-09,
      "loss": 0.8641,
      "step": 7968
    },
    {
      "epoch": 0.9940127229637021,
      "grad_norm": 0.0,
      "learning_rate": 1.8802955149865854e-09,
      "loss": 0.9346,
      "step": 7969
    },
    {
      "epoch": 0.9941374579019583,
      "grad_norm": 0.0,
      "learning_rate": 1.8027682985610751e-09,
      "loss": 0.928,
      "step": 7970
    },
    {
      "epoch": 0.9942621928402146,
      "grad_norm": 0.0,
      "learning_rate": 1.7268730399777612e-09,
      "loss": 0.894,
      "step": 7971
    },
    {
      "epoch": 0.9943869277784707,
      "grad_norm": 0.0,
      "learning_rate": 1.652609751624512e-09,
      "loss": 0.8259,
      "step": 7972
    },
    {
      "epoch": 0.994511662716727,
      "grad_norm": 0.0,
      "learning_rate": 1.5799784456216328e-09,
      "loss": 0.8796,
      "step": 7973
    },
    {
      "epoch": 0.9946363976549831,
      "grad_norm": 0.0,
      "learning_rate": 1.5089791338251947e-09,
      "loss": 0.9803,
      "step": 7974
    },
    {
      "epoch": 0.9947611325932394,
      "grad_norm": 0.0,
      "learning_rate": 1.4396118278237058e-09,
      "loss": 0.8724,
      "step": 7975
    },
    {
      "epoch": 0.9948858675314955,
      "grad_norm": 0.0,
      "learning_rate": 1.3718765389392208e-09,
      "loss": 0.9159,
      "step": 7976
    },
    {
      "epoch": 0.9950106024697518,
      "grad_norm": 0.0,
      "learning_rate": 1.3057732782295606e-09,
      "loss": 0.906,
      "step": 7977
    },
    {
      "epoch": 0.995135337408008,
      "grad_norm": 0.0,
      "learning_rate": 1.2413020564827626e-09,
      "loss": 0.8765,
      "step": 7978
    },
    {
      "epoch": 0.9952600723462642,
      "grad_norm": 0.0,
      "learning_rate": 1.178462884222631e-09,
      "loss": 0.9213,
      "step": 7979
    },
    {
      "epoch": 0.9953848072845204,
      "grad_norm": 0.0,
      "learning_rate": 1.1172557717065157e-09,
      "loss": 0.8926,
      "step": 7980
    },
    {
      "epoch": 0.9955095422227765,
      "grad_norm": 0.0,
      "learning_rate": 1.0576807289253143e-09,
      "loss": 0.9095,
      "step": 7981
    },
    {
      "epoch": 0.9956342771610328,
      "grad_norm": 0.0,
      "learning_rate": 9.997377656012496e-10,
      "loss": 0.8696,
      "step": 7982
    },
    {
      "epoch": 0.995759012099289,
      "grad_norm": 0.0,
      "learning_rate": 9.43426891193422e-10,
      "loss": 0.9217,
      "step": 7983
    },
    {
      "epoch": 0.9958837470375452,
      "grad_norm": 0.0,
      "learning_rate": 8.887481148944776e-10,
      "loss": 0.9402,
      "step": 7984
    },
    {
      "epoch": 0.9960084819758014,
      "grad_norm": 0.0,
      "learning_rate": 8.357014456272794e-10,
      "loss": 0.8834,
      "step": 7985
    },
    {
      "epoch": 0.9961332169140577,
      "grad_norm": 0.0,
      "learning_rate": 7.842868920504564e-10,
      "loss": 0.8635,
      "step": 7986
    },
    {
      "epoch": 0.9962579518523138,
      "grad_norm": 0.0,
      "learning_rate": 7.34504462557295e-10,
      "loss": 0.8803,
      "step": 7987
    },
    {
      "epoch": 0.9963826867905701,
      "grad_norm": 0.0,
      "learning_rate": 6.863541652735173e-10,
      "loss": 0.8678,
      "step": 7988
    },
    {
      "epoch": 0.9965074217288262,
      "grad_norm": 0.0,
      "learning_rate": 6.398360080572819e-10,
      "loss": 0.9099,
      "step": 7989
    },
    {
      "epoch": 0.9966321566670825,
      "grad_norm": 0.0,
      "learning_rate": 5.949499985025142e-10,
      "loss": 0.9496,
      "step": 7990
    },
    {
      "epoch": 0.9967568916053386,
      "grad_norm": 0.0,
      "learning_rate": 5.516961439355761e-10,
      "loss": 0.9352,
      "step": 7991
    },
    {
      "epoch": 0.9968816265435949,
      "grad_norm": 0.0,
      "learning_rate": 5.100744514163758e-10,
      "loss": 0.9465,
      "step": 7992
    },
    {
      "epoch": 0.9970063614818511,
      "grad_norm": 0.0,
      "learning_rate": 4.700849277383679e-10,
      "loss": 0.9046,
      "step": 7993
    },
    {
      "epoch": 0.9971310964201072,
      "grad_norm": 0.0,
      "learning_rate": 4.317275794296638e-10,
      "loss": 0.94,
      "step": 7994
    },
    {
      "epoch": 0.9972558313583635,
      "grad_norm": 0.0,
      "learning_rate": 3.950024127508112e-10,
      "loss": 0.943,
      "step": 7995
    },
    {
      "epoch": 0.9973805662966196,
      "grad_norm": 0.0,
      "learning_rate": 3.5990943369590413e-10,
      "loss": 0.9167,
      "step": 7996
    },
    {
      "epoch": 0.9975053012348759,
      "grad_norm": 0.0,
      "learning_rate": 3.2644864799369346e-10,
      "loss": 0.8945,
      "step": 7997
    },
    {
      "epoch": 0.9976300361731321,
      "grad_norm": 0.0,
      "learning_rate": 2.94620061104256e-10,
      "loss": 0.9371,
      "step": 7998
    },
    {
      "epoch": 0.9977547711113883,
      "grad_norm": 0.0,
      "learning_rate": 2.64423678225656e-10,
      "loss": 0.8776,
      "step": 7999
    },
    {
      "epoch": 0.9978795060496445,
      "grad_norm": 0.0,
      "learning_rate": 2.358595042839529e-10,
      "loss": 0.8685,
      "step": 8000
    },
    {
      "epoch": 0.9980042409879007,
      "grad_norm": 0.0,
      "learning_rate": 2.0892754394208348e-10,
      "loss": 0.9104,
      "step": 8001
    },
    {
      "epoch": 0.9981289759261569,
      "grad_norm": 0.0,
      "learning_rate": 1.8362780159764115e-10,
      "loss": 0.8852,
      "step": 8002
    },
    {
      "epoch": 0.9982537108644132,
      "grad_norm": 0.0,
      "learning_rate": 1.5996028137843511e-10,
      "loss": 0.8884,
      "step": 8003
    },
    {
      "epoch": 0.9983784458026693,
      "grad_norm": 0.0,
      "learning_rate": 1.379249871491517e-10,
      "loss": 0.8657,
      "step": 8004
    },
    {
      "epoch": 0.9985031807409255,
      "grad_norm": 0.0,
      "learning_rate": 1.1752192250469307e-10,
      "loss": 0.8889,
      "step": 8005
    },
    {
      "epoch": 0.9986279156791817,
      "grad_norm": 0.0,
      "learning_rate": 9.875109077683853e-11,
      "loss": 0.898,
      "step": 8006
    },
    {
      "epoch": 0.9987526506174379,
      "grad_norm": 0.0,
      "learning_rate": 8.16124950286934e-11,
      "loss": 0.8613,
      "step": 8007
    },
    {
      "epoch": 0.9988773855556942,
      "grad_norm": 0.0,
      "learning_rate": 6.610613805690947e-11,
      "loss": 0.9277,
      "step": 8008
    },
    {
      "epoch": 0.9990021204939503,
      "grad_norm": 0.0,
      "learning_rate": 5.223202239501568e-11,
      "loss": 0.9622,
      "step": 8009
    },
    {
      "epoch": 0.9991268554322066,
      "grad_norm": 0.0,
      "learning_rate": 3.999015030564657e-11,
      "loss": 0.9087,
      "step": 8010
    },
    {
      "epoch": 0.9992515903704627,
      "grad_norm": 0.0,
      "learning_rate": 2.938052378720357e-11,
      "loss": 0.8766,
      "step": 8011
    },
    {
      "epoch": 0.999376325308719,
      "grad_norm": 0.0,
      "learning_rate": 2.0403144571634615e-11,
      "loss": 0.8852,
      "step": 8012
    },
    {
      "epoch": 0.9995010602469752,
      "grad_norm": 0.0,
      "learning_rate": 1.3058014124434082e-11,
      "loss": 0.906,
      "step": 8013
    },
    {
      "epoch": 0.9996257951852314,
      "grad_norm": 0.0,
      "learning_rate": 7.345133644642843e-12,
      "loss": 0.8907,
      "step": 8014
    },
    {
      "epoch": 0.9997505301234876,
      "grad_norm": 0.0,
      "learning_rate": 3.2645040637380163e-12,
      "loss": 0.941,
      "step": 8015
    },
    {
      "epoch": 0.9998752650617438,
      "grad_norm": 0.0,
      "learning_rate": 8.161260489636391e-13,
      "loss": 0.9176,
      "step": 8016
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 0.8931,
      "step": 8017
    },
    {
      "epoch": 1.0,
      "step": 8017,
      "total_flos": 7.133618647935201e+20,
      "train_loss": 1.0184825783405158,
      "train_runtime": 70667.3144,
      "train_samples_per_second": 116.156,
      "train_steps_per_second": 0.113
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 8017,
  "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": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.133618647935201e+20,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}