{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5198,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 8.51450312862667,
      "learning_rate": 1.282051282051282e-07,
      "loss": 1.3357,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.135789479491661,
      "learning_rate": 2.564102564102564e-07,
      "loss": 1.461,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.282849336738456,
      "learning_rate": 3.846153846153847e-07,
      "loss": 1.398,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.687822040542483,
      "learning_rate": 5.128205128205128e-07,
      "loss": 1.3237,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.661914361071114,
      "learning_rate": 6.41025641025641e-07,
      "loss": 1.3816,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.380877563911993,
      "learning_rate": 7.692307692307694e-07,
      "loss": 1.3695,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.175991991797468,
      "learning_rate": 8.974358974358975e-07,
      "loss": 1.352,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.351552881137372,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 1.3267,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.0585709462222335,
      "learning_rate": 1.153846153846154e-06,
      "loss": 1.3748,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.630206850623948,
      "learning_rate": 1.282051282051282e-06,
      "loss": 1.2804,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.182035756781344,
      "learning_rate": 1.4102564102564104e-06,
      "loss": 1.2426,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1480737783434,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 1.3279,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5746232830481954,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.291,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7681846788527262,
      "learning_rate": 1.794871794871795e-06,
      "loss": 1.1735,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.686360476622005,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 1.2223,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6572300312646266,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 1.2537,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6743287804232662,
      "learning_rate": 2.1794871794871797e-06,
      "loss": 0.2836,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.106563001553971,
      "learning_rate": 2.307692307692308e-06,
      "loss": 1.1418,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2464256977611963,
      "learning_rate": 2.435897435897436e-06,
      "loss": 1.159,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.147622263221434,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.1831,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.931431056314639,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 1.1379,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.737803323589056,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.1919,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.313978230167394,
      "learning_rate": 2.948717948717949e-06,
      "loss": 1.0642,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9153030433380873,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.1272,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7341586296976115,
      "learning_rate": 3.205128205128206e-06,
      "loss": 1.129,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.876535951317794,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.0854,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2766072814741807,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 1.1111,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8574186238802948,
      "learning_rate": 3.58974358974359e-06,
      "loss": 1.0837,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8310944565586837,
      "learning_rate": 3.7179487179487184e-06,
      "loss": 1.0433,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8299522346424648,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.0852,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7890917635589956,
      "learning_rate": 3.974358974358974e-06,
      "loss": 1.0324,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9350250273957001,
      "learning_rate": 4.102564102564103e-06,
      "loss": 1.0884,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6435127516443646,
      "learning_rate": 4.230769230769231e-06,
      "loss": 1.038,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5566218014850959,
      "learning_rate": 4.358974358974359e-06,
      "loss": 1.0003,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4317415796739794,
      "learning_rate": 4.487179487179488e-06,
      "loss": 1.0106,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4143910302999692,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.9953,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4582234928618028,
      "learning_rate": 4.743589743589744e-06,
      "loss": 1.0622,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3891378421030436,
      "learning_rate": 4.871794871794872e-06,
      "loss": 1.0152,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.334712980776743,
      "learning_rate": 5e-06,
      "loss": 1.0149,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.382328966901326,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.9817,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3519398990624543,
      "learning_rate": 5.256410256410257e-06,
      "loss": 0.9885,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2302698546074466,
      "learning_rate": 5.384615384615385e-06,
      "loss": 0.9711,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.217176272032287,
      "learning_rate": 5.512820512820514e-06,
      "loss": 1.007,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2257574567421499,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.9658,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3086340108058478,
      "learning_rate": 5.769230769230769e-06,
      "loss": 0.9674,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3253457661119452,
      "learning_rate": 5.897435897435898e-06,
      "loss": 0.9906,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2904156745893463,
      "learning_rate": 6.025641025641026e-06,
      "loss": 1.0504,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3082323527196393,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.949,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3007740481996355,
      "learning_rate": 6.282051282051282e-06,
      "loss": 1.0321,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2442438245248184,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.0037,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3322219374791289,
      "learning_rate": 6.538461538461539e-06,
      "loss": 1.0033,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3443550377164626,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.9771,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2554218806094641,
      "learning_rate": 6.794871794871796e-06,
      "loss": 0.9755,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2446550326862067,
      "learning_rate": 6.923076923076923e-06,
      "loss": 0.9717,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4534623109624805,
      "learning_rate": 7.051282051282053e-06,
      "loss": 1.0598,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2136377492170274,
      "learning_rate": 7.17948717948718e-06,
      "loss": 1.0373,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2145596544604598,
      "learning_rate": 7.307692307692308e-06,
      "loss": 0.9913,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1899223936594614,
      "learning_rate": 7.435897435897437e-06,
      "loss": 1.0131,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2831662539623923,
      "learning_rate": 7.564102564102564e-06,
      "loss": 0.9908,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.26548884389373,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.9676,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.280062690501541,
      "learning_rate": 7.820512820512822e-06,
      "loss": 0.9925,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2121786947525532,
      "learning_rate": 7.948717948717949e-06,
      "loss": 1.0349,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3160031521107878,
      "learning_rate": 8.076923076923077e-06,
      "loss": 1.0228,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.33048141435525,
      "learning_rate": 8.205128205128205e-06,
      "loss": 0.96,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.197961669682093,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.9497,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3680892603273216,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.9702,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2740425876568267,
      "learning_rate": 8.58974358974359e-06,
      "loss": 0.9624,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2107892019656168,
      "learning_rate": 8.717948717948719e-06,
      "loss": 0.9714,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3125357738299954,
      "learning_rate": 8.846153846153847e-06,
      "loss": 1.013,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1945948265595083,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.9509,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2466255304417562,
      "learning_rate": 9.102564102564104e-06,
      "loss": 0.9285,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2482854670299255,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.9872,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2187136215197034,
      "learning_rate": 9.358974358974359e-06,
      "loss": 0.9582,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2792544885704191,
      "learning_rate": 9.487179487179487e-06,
      "loss": 0.9714,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3368664000522912,
      "learning_rate": 9.615384615384616e-06,
      "loss": 0.9595,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2832694071005875,
      "learning_rate": 9.743589743589744e-06,
      "loss": 0.9611,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1929058609766512,
      "learning_rate": 9.871794871794872e-06,
      "loss": 0.9396,
      "step": 77
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.183462485031287,
      "learning_rate": 1e-05,
      "loss": 1.0078,
      "step": 78
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2653531870612071,
      "learning_rate": 1.012820512820513e-05,
      "loss": 0.9685,
      "step": 79
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3370385998428251,
      "learning_rate": 1.0256410256410256e-05,
      "loss": 0.9314,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.231081987641431,
      "learning_rate": 1.0384615384615386e-05,
      "loss": 0.9679,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2229555309488849,
      "learning_rate": 1.0512820512820514e-05,
      "loss": 0.9344,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.230399971318872,
      "learning_rate": 1.0641025641025643e-05,
      "loss": 0.9197,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.322911215841961,
      "learning_rate": 1.076923076923077e-05,
      "loss": 0.9561,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3946143053469953,
      "learning_rate": 1.0897435897435898e-05,
      "loss": 0.2919,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4772334721976073,
      "learning_rate": 1.1025641025641028e-05,
      "loss": 0.9533,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2456818659632007,
      "learning_rate": 1.1153846153846154e-05,
      "loss": 0.9594,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.267023993124228,
      "learning_rate": 1.1282051282051283e-05,
      "loss": 0.9586,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2272380076169542,
      "learning_rate": 1.1410256410256411e-05,
      "loss": 0.9729,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2546124176274378,
      "learning_rate": 1.1538461538461538e-05,
      "loss": 0.9734,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2537751400368864,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.9771,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3052062440635954,
      "learning_rate": 1.1794871794871796e-05,
      "loss": 0.9825,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.342557781934961,
      "learning_rate": 1.1923076923076925e-05,
      "loss": 0.9515,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.328400880370958,
      "learning_rate": 1.2051282051282051e-05,
      "loss": 0.9133,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3575623461883737,
      "learning_rate": 1.217948717948718e-05,
      "loss": 0.9629,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3134183890283484,
      "learning_rate": 1.230769230769231e-05,
      "loss": 0.9583,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1836548486919165,
      "learning_rate": 1.2435897435897436e-05,
      "loss": 0.9332,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2858393118086517,
      "learning_rate": 1.2564102564102565e-05,
      "loss": 0.9466,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1833789896362548,
      "learning_rate": 1.2692307692307693e-05,
      "loss": 0.9328,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1973467989126954,
      "learning_rate": 1.2820512820512823e-05,
      "loss": 0.9808,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3565292483478921,
      "learning_rate": 1.294871794871795e-05,
      "loss": 1.042,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2979377880472796,
      "learning_rate": 1.3076923076923078e-05,
      "loss": 0.9561,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2242505564662811,
      "learning_rate": 1.3205128205128207e-05,
      "loss": 0.9489,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1445716714068175,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.8959,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2474001475905299,
      "learning_rate": 1.3461538461538463e-05,
      "loss": 0.9925,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.258650071777052,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 0.9609,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1950567844987592,
      "learning_rate": 1.3717948717948718e-05,
      "loss": 0.9923,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2155824828879507,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 0.9003,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2675983512936937,
      "learning_rate": 1.3974358974358975e-05,
      "loss": 0.9778,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.235970927264469,
      "learning_rate": 1.4102564102564105e-05,
      "loss": 0.94,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2820727268029593,
      "learning_rate": 1.4230769230769232e-05,
      "loss": 0.9848,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1249878679058445,
      "learning_rate": 1.435897435897436e-05,
      "loss": 0.9572,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1871911204206116,
      "learning_rate": 1.4487179487179489e-05,
      "loss": 0.9029,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2732089324247469,
      "learning_rate": 1.4615384615384615e-05,
      "loss": 0.9792,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.262559897499703,
      "learning_rate": 1.4743589743589745e-05,
      "loss": 0.9403,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.171718496958663,
      "learning_rate": 1.4871794871794874e-05,
      "loss": 0.9425,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2645833363217485,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.9072,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.230636830430442,
      "learning_rate": 1.5128205128205129e-05,
      "loss": 0.9311,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2026005880641237,
      "learning_rate": 1.5256410256410257e-05,
      "loss": 0.9631,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2082320367956438,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 0.9555,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.237262475400901,
      "learning_rate": 1.5512820512820516e-05,
      "loss": 0.9738,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2552477137228524,
      "learning_rate": 1.5641025641025644e-05,
      "loss": 0.9203,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3787122978733872,
      "learning_rate": 1.576923076923077e-05,
      "loss": 0.9587,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2340312389309394,
      "learning_rate": 1.5897435897435897e-05,
      "loss": 0.9818,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2354919017454544,
      "learning_rate": 1.602564102564103e-05,
      "loss": 0.9413,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2115158159455037,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 0.9868,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2394357258406212,
      "learning_rate": 1.6282051282051282e-05,
      "loss": 0.9436,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1422321101314494,
      "learning_rate": 1.641025641025641e-05,
      "loss": 0.9163,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2335892183928456,
      "learning_rate": 1.653846153846154e-05,
      "loss": 0.9445,
      "step": 129
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2308909473361989,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.966,
      "step": 130
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1671506442943451,
      "learning_rate": 1.6794871794871796e-05,
      "loss": 0.886,
      "step": 131
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.241851431292281,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 0.9907,
      "step": 132
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2397869121417024,
      "learning_rate": 1.7051282051282053e-05,
      "loss": 0.9646,
      "step": 133
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2229779946715076,
      "learning_rate": 1.717948717948718e-05,
      "loss": 0.9661,
      "step": 134
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3089218601913348,
      "learning_rate": 1.730769230769231e-05,
      "loss": 1.0012,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1603581260112832,
      "learning_rate": 1.7435897435897438e-05,
      "loss": 0.3314,
      "step": 136
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.96145940531011,
      "learning_rate": 1.7564102564102566e-05,
      "loss": 0.3221,
      "step": 137
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0965542544266973,
      "learning_rate": 1.7692307692307694e-05,
      "loss": 0.9823,
      "step": 138
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.283729269899266,
      "learning_rate": 1.7820512820512823e-05,
      "loss": 0.9839,
      "step": 139
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1719129274111726,
      "learning_rate": 1.794871794871795e-05,
      "loss": 0.9779,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.093799424544862,
      "learning_rate": 1.807692307692308e-05,
      "loss": 0.9206,
      "step": 141
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2206457700606652,
      "learning_rate": 1.8205128205128208e-05,
      "loss": 0.9582,
      "step": 142
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2563084337892594,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.9071,
      "step": 143
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2006614783082044,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 0.9212,
      "step": 144
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2142392026348914,
      "learning_rate": 1.8589743589743593e-05,
      "loss": 0.8658,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1398936190164322,
      "learning_rate": 1.8717948717948718e-05,
      "loss": 0.8885,
      "step": 146
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1149617749669338,
      "learning_rate": 1.8846153846153846e-05,
      "loss": 0.9747,
      "step": 147
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.219780699305036,
      "learning_rate": 1.8974358974358975e-05,
      "loss": 0.8837,
      "step": 148
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.194216548741588,
      "learning_rate": 1.9102564102564106e-05,
      "loss": 1.0043,
      "step": 149
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3607612182684818,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.9704,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2437580381620708,
      "learning_rate": 1.935897435897436e-05,
      "loss": 0.9136,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1419014743783718,
      "learning_rate": 1.9487179487179488e-05,
      "loss": 0.9224,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1487350046084015,
      "learning_rate": 1.9615384615384617e-05,
      "loss": 0.9445,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1692361609257214,
      "learning_rate": 1.9743589743589745e-05,
      "loss": 0.9611,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1501202844507379,
      "learning_rate": 1.9871794871794873e-05,
      "loss": 0.9482,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2434592797338346,
      "learning_rate": 2e-05,
      "loss": 0.9643,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1668087974803811,
      "learning_rate": 1.9999998058827844e-05,
      "loss": 0.9427,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2172324358161632,
      "learning_rate": 1.9999992235312136e-05,
      "loss": 0.9236,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2264783463644897,
      "learning_rate": 1.9999982529455127e-05,
      "loss": 0.8932,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1769404431839507,
      "learning_rate": 1.9999968941260596e-05,
      "loss": 0.9755,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0583067168109546,
      "learning_rate": 1.9999951470733808e-05,
      "loss": 0.8972,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1313481232086917,
      "learning_rate": 1.9999930117881548e-05,
      "loss": 0.8687,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2413947226707565,
      "learning_rate": 1.9999904882712115e-05,
      "loss": 0.9275,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2040140366899663,
      "learning_rate": 1.99998757652353e-05,
      "loss": 0.9549,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1258349456753027,
      "learning_rate": 1.9999842765462403e-05,
      "loss": 0.9688,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1044989955837305,
      "learning_rate": 1.999980588340624e-05,
      "loss": 0.9044,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1771017197281342,
      "learning_rate": 1.9999765119081132e-05,
      "loss": 0.9617,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.167892837928055,
      "learning_rate": 1.9999720472502902e-05,
      "loss": 0.9805,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.403918537893246,
      "learning_rate": 1.9999671943688885e-05,
      "loss": 0.3634,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.185787273375159,
      "learning_rate": 1.9999619532657915e-05,
      "loss": 0.9396,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2161402796440486,
      "learning_rate": 1.9999563239430352e-05,
      "loss": 0.9359,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2355264685001197,
      "learning_rate": 1.9999503064028043e-05,
      "loss": 0.942,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1644412127118722,
      "learning_rate": 1.999943900647435e-05,
      "loss": 0.917,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.23701859955091,
      "learning_rate": 1.9999371066794146e-05,
      "loss": 0.9917,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1965905230074592,
      "learning_rate": 1.9999299245013805e-05,
      "loss": 0.9256,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1038834984068802,
      "learning_rate": 1.999922354116121e-05,
      "loss": 0.9554,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1274549802290965,
      "learning_rate": 1.999914395526575e-05,
      "loss": 0.9167,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1967405280867125,
      "learning_rate": 1.9999060487358333e-05,
      "loss": 0.9833,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1330703335053616,
      "learning_rate": 1.9998973137471352e-05,
      "loss": 0.9388,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1728601947692268,
      "learning_rate": 1.9998881905638727e-05,
      "loss": 0.953,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0983032919180202,
      "learning_rate": 1.9998786791895874e-05,
      "loss": 0.9258,
      "step": 181
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.148853477451541,
      "learning_rate": 1.999868779627972e-05,
      "loss": 0.9135,
      "step": 182
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1878992167638494,
      "learning_rate": 1.9998584918828695e-05,
      "loss": 0.9585,
      "step": 183
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.164622504746696,
      "learning_rate": 1.9998478159582747e-05,
      "loss": 1.0198,
      "step": 184
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1291285961212867,
      "learning_rate": 1.999836751858332e-05,
      "loss": 0.9259,
      "step": 185
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.188258292874918,
      "learning_rate": 1.9998252995873367e-05,
      "loss": 0.9311,
      "step": 186
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1485619941772611,
      "learning_rate": 1.999813459149735e-05,
      "loss": 0.9728,
      "step": 187
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.175369693643261,
      "learning_rate": 1.9998012305501243e-05,
      "loss": 0.9187,
      "step": 188
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.27079974899039,
      "learning_rate": 1.999788613793251e-05,
      "loss": 0.9651,
      "step": 189
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1302366343584613,
      "learning_rate": 1.999775608884015e-05,
      "loss": 0.9358,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0486101037348394,
      "learning_rate": 1.9997622158274635e-05,
      "loss": 0.923,
      "step": 191
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1373881349988662,
      "learning_rate": 1.9997484346287973e-05,
      "loss": 0.9483,
      "step": 192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1383681062850532,
      "learning_rate": 1.9997342652933668e-05,
      "loss": 0.9296,
      "step": 193
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.220834703029841,
      "learning_rate": 1.9997197078266723e-05,
      "loss": 0.9298,
      "step": 194
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0486578152635608,
      "learning_rate": 1.999704762234366e-05,
      "loss": 0.9602,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.171108032219634,
      "learning_rate": 1.99968942852225e-05,
      "loss": 0.9256,
      "step": 196
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1423888679408472,
      "learning_rate": 1.9996737066962778e-05,
      "loss": 0.8827,
      "step": 197
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.168600418922881,
      "learning_rate": 1.9996575967625525e-05,
      "loss": 0.9528,
      "step": 198
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.144641282513704,
      "learning_rate": 1.999641098727329e-05,
      "loss": 0.9586,
      "step": 199
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1896754534275924,
      "learning_rate": 1.999624212597013e-05,
      "loss": 0.8821,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.125211309250738,
      "learning_rate": 1.9996069383781587e-05,
      "loss": 0.9257,
      "step": 201
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1536009975477848,
      "learning_rate": 1.9995892760774738e-05,
      "loss": 0.9472,
      "step": 202
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1768089091418108,
      "learning_rate": 1.9995712257018153e-05,
      "loss": 0.9824,
      "step": 203
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1977294438457617,
      "learning_rate": 1.9995527872581903e-05,
      "loss": 0.9165,
      "step": 204
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.220795637058411,
      "learning_rate": 1.9995339607537578e-05,
      "loss": 0.9327,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1400585733871065,
      "learning_rate": 1.9995147461958267e-05,
      "loss": 0.9503,
      "step": 206
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6396273349769477,
      "learning_rate": 1.999495143591857e-05,
      "loss": 0.3727,
      "step": 207
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1171714778022996,
      "learning_rate": 1.999475152949459e-05,
      "loss": 0.9058,
      "step": 208
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2147288155003513,
      "learning_rate": 1.9994547742763935e-05,
      "loss": 0.9294,
      "step": 209
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1495006179468015,
      "learning_rate": 1.9994340075805724e-05,
      "loss": 0.9748,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2256800738881273,
      "learning_rate": 1.9994128528700583e-05,
      "loss": 1.0014,
      "step": 211
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1317747909454503,
      "learning_rate": 1.9993913101530635e-05,
      "loss": 0.917,
      "step": 212
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0916164504805597,
      "learning_rate": 1.9993693794379525e-05,
      "loss": 0.8644,
      "step": 213
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1011261767035505,
      "learning_rate": 1.9993470607332387e-05,
      "loss": 0.9566,
      "step": 214
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1493960788733046,
      "learning_rate": 1.999324354047588e-05,
      "loss": 0.9773,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0628386671695187,
      "learning_rate": 1.9993012593898146e-05,
      "loss": 0.9227,
      "step": 216
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1170186326974756,
      "learning_rate": 1.9992777767688857e-05,
      "loss": 0.9052,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0822695317401287,
      "learning_rate": 1.9992539061939175e-05,
      "loss": 0.9783,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2287385765536187,
      "learning_rate": 1.999229647674178e-05,
      "loss": 0.9289,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.115382615365123,
      "learning_rate": 1.9992050012190845e-05,
      "loss": 0.9284,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9019672647733321,
      "learning_rate": 1.9991799668382058e-05,
      "loss": 0.3409,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1234349221003008,
      "learning_rate": 1.9991545445412614e-05,
      "loss": 0.9268,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1635886956601753,
      "learning_rate": 1.9991287343381208e-05,
      "loss": 1.027,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2225291845759534,
      "learning_rate": 1.9991025362388044e-05,
      "loss": 0.9243,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0753852683369352,
      "learning_rate": 1.9990759502534835e-05,
      "loss": 0.9294,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.088477123962133,
      "learning_rate": 1.9990489763924796e-05,
      "loss": 0.9213,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1195902173061787,
      "learning_rate": 1.9990216146662648e-05,
      "loss": 0.9309,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1410348436365578,
      "learning_rate": 1.9989938650854618e-05,
      "loss": 0.9771,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.139398298539973,
      "learning_rate": 1.998965727660844e-05,
      "loss": 0.9238,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2340251200021093,
      "learning_rate": 1.9989372024033352e-05,
      "loss": 0.9364,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1502394157879723,
      "learning_rate": 1.99890828932401e-05,
      "loss": 0.9528,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1578112238928098,
      "learning_rate": 1.9988789884340938e-05,
      "loss": 0.8623,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1285596104147748,
      "learning_rate": 1.9988492997449615e-05,
      "loss": 0.9045,
      "step": 233
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1186453054365697,
      "learning_rate": 1.9988192232681398e-05,
      "loss": 0.8694,
      "step": 234
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1524805959573174,
      "learning_rate": 1.9987887590153055e-05,
      "loss": 0.9578,
      "step": 235
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2521593203776042,
      "learning_rate": 1.9987579069982856e-05,
      "loss": 0.9286,
      "step": 236
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0944420868561155,
      "learning_rate": 1.9987266672290577e-05,
      "loss": 0.9238,
      "step": 237
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0696539967586831,
      "learning_rate": 1.9986950397197503e-05,
      "loss": 0.9655,
      "step": 238
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2115223322181758,
      "learning_rate": 1.9986630244826425e-05,
      "loss": 0.919,
      "step": 239
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.184294018990164,
      "learning_rate": 1.998630621530164e-05,
      "loss": 0.9159,
      "step": 240
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.098643954326505,
      "learning_rate": 1.998597830874894e-05,
      "loss": 0.886,
      "step": 241
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.136917710097713,
      "learning_rate": 1.9985646525295634e-05,
      "loss": 0.9774,
      "step": 242
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0826403309996027,
      "learning_rate": 1.998531086507053e-05,
      "loss": 0.935,
      "step": 243
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0664784717906421,
      "learning_rate": 1.9984971328203945e-05,
      "loss": 0.972,
      "step": 244
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0712310100238078,
      "learning_rate": 1.9984627914827698e-05,
      "loss": 0.9551,
      "step": 245
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.152975196907916,
      "learning_rate": 1.9984280625075115e-05,
      "loss": 0.9432,
      "step": 246
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0370750974973089,
      "learning_rate": 1.9983929459081022e-05,
      "loss": 0.8786,
      "step": 247
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7234297990775279,
      "learning_rate": 1.998357441698176e-05,
      "loss": 0.3342,
      "step": 248
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0678452430634673,
      "learning_rate": 1.998321549891516e-05,
      "loss": 0.9333,
      "step": 249
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.054806116286314,
      "learning_rate": 1.9982852705020572e-05,
      "loss": 0.96,
      "step": 250
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1146527392028827,
      "learning_rate": 1.9982486035438848e-05,
      "loss": 0.9381,
      "step": 251
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0324924990549156,
      "learning_rate": 1.9982115490312334e-05,
      "loss": 0.8905,
      "step": 252
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7192477946225247,
      "learning_rate": 1.9981741069784894e-05,
      "loss": 0.329,
      "step": 253
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.12973086102753,
      "learning_rate": 1.9981362774001886e-05,
      "loss": 0.9616,
      "step": 254
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.073955569539675,
      "learning_rate": 1.9980980603110185e-05,
      "loss": 0.9589,
      "step": 255
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6557953782223579,
      "learning_rate": 1.9980594557258158e-05,
      "loss": 0.3367,
      "step": 256
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6430648611436847,
      "learning_rate": 1.9980204636595682e-05,
      "loss": 0.3356,
      "step": 257
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6195986705043429,
      "learning_rate": 1.9979810841274135e-05,
      "loss": 0.3053,
      "step": 258
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1841950901366154,
      "learning_rate": 1.9979413171446403e-05,
      "loss": 0.8887,
      "step": 259
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1235841380255114,
      "learning_rate": 1.9979011627266884e-05,
      "loss": 0.9403,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0777013070758232,
      "learning_rate": 1.997860620889146e-05,
      "loss": 0.9354,
      "step": 261
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1619347212681885,
      "learning_rate": 1.997819691647753e-05,
      "loss": 0.9151,
      "step": 262
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2090233667861765,
      "learning_rate": 1.9977783750184e-05,
      "loss": 0.9224,
      "step": 263
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0068931595704151,
      "learning_rate": 1.9977366710171274e-05,
      "loss": 0.8727,
      "step": 264
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0887730442284358,
      "learning_rate": 1.9976945796601258e-05,
      "loss": 0.9557,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1037919729158103,
      "learning_rate": 1.9976521009637366e-05,
      "loss": 0.8774,
      "step": 266
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0924901902003337,
      "learning_rate": 1.997609234944452e-05,
      "loss": 0.9451,
      "step": 267
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.101329008933356,
      "learning_rate": 1.9975659816189137e-05,
      "loss": 0.8507,
      "step": 268
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2850327015684637,
      "learning_rate": 1.997522341003914e-05,
      "loss": 0.9388,
      "step": 269
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1047350869777726,
      "learning_rate": 1.9974783131163957e-05,
      "loss": 0.9488,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9932698719287122,
      "learning_rate": 1.9974338979734523e-05,
      "loss": 0.873,
      "step": 271
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1725233859198567,
      "learning_rate": 1.997389095592327e-05,
      "loss": 0.8942,
      "step": 272
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.074936752083998,
      "learning_rate": 1.9973439059904133e-05,
      "loss": 0.9712,
      "step": 273
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0382908159587438,
      "learning_rate": 1.9972983291852565e-05,
      "loss": 0.9407,
      "step": 274
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1229736105199453,
      "learning_rate": 1.9972523651945496e-05,
      "loss": 0.9007,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0939738982161797,
      "learning_rate": 1.9972060140361384e-05,
      "loss": 0.9603,
      "step": 276
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0281719564225267,
      "learning_rate": 1.997159275728018e-05,
      "loss": 0.9114,
      "step": 277
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1490616221602226,
      "learning_rate": 1.9971121502883332e-05,
      "loss": 0.9149,
      "step": 278
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1286788739496818,
      "learning_rate": 1.9970646377353802e-05,
      "loss": 0.9116,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1320926060085195,
      "learning_rate": 1.997016738087605e-05,
      "loss": 0.9507,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9695125955551456,
      "learning_rate": 1.9969684513636035e-05,
      "loss": 0.8669,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1283730071241458,
      "learning_rate": 1.9969197775821227e-05,
      "loss": 0.9288,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.11455183387599,
      "learning_rate": 1.9968707167620593e-05,
      "loss": 0.922,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0987848079231104,
      "learning_rate": 1.9968212689224603e-05,
      "loss": 0.8885,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.020775966913813,
      "learning_rate": 1.996771434082523e-05,
      "loss": 0.8829,
      "step": 285
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0723191749076573,
      "learning_rate": 1.9967212122615958e-05,
      "loss": 0.9281,
      "step": 286
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8477566440961237,
      "learning_rate": 1.9966706034791752e-05,
      "loss": 0.312,
      "step": 287
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.091601393610562,
      "learning_rate": 1.9966196077549106e-05,
      "loss": 0.9501,
      "step": 288
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6811255367705601,
      "learning_rate": 1.996568225108599e-05,
      "loss": 0.3113,
      "step": 289
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0536287645147577,
      "learning_rate": 1.99651645556019e-05,
      "loss": 0.9056,
      "step": 290
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1067552950880963,
      "learning_rate": 1.9964642991297817e-05,
      "loss": 0.9697,
      "step": 291
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.10500521891306,
      "learning_rate": 1.996411755837623e-05,
      "loss": 0.926,
      "step": 292
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0553316114242433,
      "learning_rate": 1.9963588257041137e-05,
      "loss": 0.9881,
      "step": 293
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1344137499761449,
      "learning_rate": 1.996305508749802e-05,
      "loss": 0.9693,
      "step": 294
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.015859869474842,
      "learning_rate": 1.9962518049953887e-05,
      "loss": 0.9155,
      "step": 295
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0859651979037555,
      "learning_rate": 1.9961977144617225e-05,
      "loss": 0.8795,
      "step": 296
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.105448767309579,
      "learning_rate": 1.996143237169803e-05,
      "loss": 0.9088,
      "step": 297
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0407238415142812,
      "learning_rate": 1.996088373140781e-05,
      "loss": 0.8896,
      "step": 298
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.045023202885212,
      "learning_rate": 1.9960331223959564e-05,
      "loss": 0.9447,
      "step": 299
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.945399250333677,
      "learning_rate": 1.995977484956779e-05,
      "loss": 0.873,
      "step": 300
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0499436701217568,
      "learning_rate": 1.9959214608448495e-05,
      "loss": 0.8922,
      "step": 301
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0765492130175547,
      "learning_rate": 1.9958650500819183e-05,
      "loss": 0.9211,
      "step": 302
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.15902083561024,
      "learning_rate": 1.995808252689886e-05,
      "loss": 0.9558,
      "step": 303
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.058221561227104,
      "learning_rate": 1.9957510686908034e-05,
      "loss": 0.9074,
      "step": 304
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0813224664577539,
      "learning_rate": 1.9956934981068713e-05,
      "loss": 0.9305,
      "step": 305
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0801793345496462,
      "learning_rate": 1.9956355409604402e-05,
      "loss": 0.9282,
      "step": 306
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0942611681303511,
      "learning_rate": 1.9955771972740118e-05,
      "loss": 0.9289,
      "step": 307
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9912675373942823,
      "learning_rate": 1.9955184670702363e-05,
      "loss": 0.9148,
      "step": 308
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0847145986781577,
      "learning_rate": 1.995459350371915e-05,
      "loss": 0.9294,
      "step": 309
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0860738538952093,
      "learning_rate": 1.9953998472019996e-05,
      "loss": 1.0075,
      "step": 310
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0927183507665676,
      "learning_rate": 1.995339957583591e-05,
      "loss": 0.9745,
      "step": 311
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0941941738814143,
      "learning_rate": 1.9952796815399403e-05,
      "loss": 0.3362,
      "step": 312
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0182140016940744,
      "learning_rate": 1.9952190190944484e-05,
      "loss": 0.865,
      "step": 313
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.51676982970469,
      "learning_rate": 1.9951579702706668e-05,
      "loss": 0.9247,
      "step": 314
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0675052450887197,
      "learning_rate": 1.9950965350922975e-05,
      "loss": 0.9304,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.106780566822522,
      "learning_rate": 1.9950347135831907e-05,
      "loss": 0.9134,
      "step": 316
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0691137339903958,
      "learning_rate": 1.994972505767348e-05,
      "loss": 0.8082,
      "step": 317
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.021543444230282,
      "learning_rate": 1.994909911668921e-05,
      "loss": 0.9069,
      "step": 318
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.13610317478637,
      "learning_rate": 1.99484693131221e-05,
      "loss": 0.9757,
      "step": 319
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1763189015651083,
      "learning_rate": 1.994783564721667e-05,
      "loss": 0.9601,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1698018776310886,
      "learning_rate": 1.9947198119218924e-05,
      "loss": 0.9464,
      "step": 321
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0601892098723118,
      "learning_rate": 1.994655672937638e-05,
      "loss": 0.8932,
      "step": 322
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0714251178438507,
      "learning_rate": 1.9945911477938044e-05,
      "loss": 0.9383,
      "step": 323
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6902933711193322,
      "learning_rate": 1.994526236515442e-05,
      "loss": 0.3302,
      "step": 324
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6597391038302063,
      "learning_rate": 1.994460939127753e-05,
      "loss": 0.342,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.080775540622738,
      "learning_rate": 1.9943952556560863e-05,
      "loss": 0.9,
      "step": 326
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9853180147335264,
      "learning_rate": 1.9943291861259433e-05,
      "loss": 0.8382,
      "step": 327
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0227041615591088,
      "learning_rate": 1.9942627305629747e-05,
      "loss": 0.9535,
      "step": 328
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0924742267091372,
      "learning_rate": 1.9941958889929808e-05,
      "loss": 0.9436,
      "step": 329
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0899696170596782,
      "learning_rate": 1.9941286614419113e-05,
      "loss": 0.9297,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0968414800431812,
      "learning_rate": 1.994061047935867e-05,
      "loss": 0.8832,
      "step": 331
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.162785698328206,
      "learning_rate": 1.9939930485010968e-05,
      "loss": 0.9551,
      "step": 332
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1619270031343534,
      "learning_rate": 1.9939246631640014e-05,
      "loss": 0.9164,
      "step": 333
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.103187867879266,
      "learning_rate": 1.99385589195113e-05,
      "loss": 0.917,
      "step": 334
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0416620968678691,
      "learning_rate": 1.9937867348891815e-05,
      "loss": 0.8733,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.074889493390474,
      "learning_rate": 1.9937171920050057e-05,
      "loss": 0.9351,
      "step": 336
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7718677087489636,
      "learning_rate": 1.9936472633256012e-05,
      "loss": 0.3085,
      "step": 337
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0857213535478414,
      "learning_rate": 1.9935769488781167e-05,
      "loss": 0.9222,
      "step": 338
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.115567477534618,
      "learning_rate": 1.993506248689851e-05,
      "loss": 0.8847,
      "step": 339
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0826310801255548,
      "learning_rate": 1.993435162788252e-05,
      "loss": 0.9746,
      "step": 340
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.085727206672529,
      "learning_rate": 1.993363691200918e-05,
      "loss": 0.9724,
      "step": 341
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0193187878876966,
      "learning_rate": 1.9932918339555965e-05,
      "loss": 0.9061,
      "step": 342
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.110163681990614,
      "learning_rate": 1.9932195910801848e-05,
      "loss": 1.0174,
      "step": 343
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.040860545311735,
      "learning_rate": 1.9931469626027305e-05,
      "loss": 0.9356,
      "step": 344
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0238158143541243,
      "learning_rate": 1.9930739485514304e-05,
      "loss": 0.946,
      "step": 345
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0628990811414993,
      "learning_rate": 1.9930005489546308e-05,
      "loss": 0.9461,
      "step": 346
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.082836670118977,
      "learning_rate": 1.9929267638408277e-05,
      "loss": 0.9123,
      "step": 347
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6428384456488929,
      "learning_rate": 1.9928525932386678e-05,
      "loss": 0.3412,
      "step": 348
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5620606106604066,
      "learning_rate": 1.9927780371769463e-05,
      "loss": 0.3181,
      "step": 349
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0959545701638165,
      "learning_rate": 1.9927030956846083e-05,
      "loss": 0.9027,
      "step": 350
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.167904179241011,
      "learning_rate": 1.992627768790749e-05,
      "loss": 0.8743,
      "step": 351
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0912316997158107,
      "learning_rate": 1.9925520565246125e-05,
      "loss": 0.953,
      "step": 352
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1461010674767642,
      "learning_rate": 1.9924759589155932e-05,
      "loss": 0.9438,
      "step": 353
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9658119461171629,
      "learning_rate": 1.9923994759932344e-05,
      "loss": 0.9072,
      "step": 354
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1595096232073885,
      "learning_rate": 1.9923226077872296e-05,
      "loss": 0.9433,
      "step": 355
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0456262519499235,
      "learning_rate": 1.9922453543274223e-05,
      "loss": 0.9272,
      "step": 356
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0539235042281192,
      "learning_rate": 1.9921677156438044e-05,
      "loss": 0.9497,
      "step": 357
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0510332293334428,
      "learning_rate": 1.9920896917665178e-05,
      "loss": 0.9375,
      "step": 358
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0266201124371113,
      "learning_rate": 1.992011282725854e-05,
      "loss": 0.9173,
      "step": 359
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1703875785785696,
      "learning_rate": 1.9919324885522548e-05,
      "loss": 0.9832,
      "step": 360
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0352205523051436,
      "learning_rate": 1.99185330927631e-05,
      "loss": 0.8881,
      "step": 361
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.096274952582758,
      "learning_rate": 1.99177374492876e-05,
      "loss": 0.8909,
      "step": 362
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1410706458657647,
      "learning_rate": 1.991693795540494e-05,
      "loss": 0.9682,
      "step": 363
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6099073311662304,
      "learning_rate": 1.9916134611425522e-05,
      "loss": 0.3071,
      "step": 364
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1282936576153242,
      "learning_rate": 1.9915327417661226e-05,
      "loss": 0.8863,
      "step": 365
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2742718249999099,
      "learning_rate": 1.991451637442543e-05,
      "loss": 0.8958,
      "step": 366
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6516176261325937,
      "learning_rate": 1.9913701482033008e-05,
      "loss": 0.3086,
      "step": 367
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9740689010309594,
      "learning_rate": 1.9912882740800336e-05,
      "loss": 0.8527,
      "step": 368
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1030909991461444,
      "learning_rate": 1.9912060151045273e-05,
      "loss": 0.9118,
      "step": 369
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0453121951431348,
      "learning_rate": 1.9911233713087172e-05,
      "loss": 0.9291,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9664614890423704,
      "learning_rate": 1.9910403427246895e-05,
      "loss": 0.8467,
      "step": 371
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6365119599744067,
      "learning_rate": 1.990956929384678e-05,
      "loss": 0.3252,
      "step": 372
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0541073161426353,
      "learning_rate": 1.990873131321067e-05,
      "loss": 0.9849,
      "step": 373
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9836223838818082,
      "learning_rate": 1.9907889485663897e-05,
      "loss": 0.9149,
      "step": 374
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1193961499632623,
      "learning_rate": 1.9907043811533283e-05,
      "loss": 0.9477,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0555481143463739,
      "learning_rate": 1.9906194291147155e-05,
      "loss": 0.9597,
      "step": 376
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0718975366876373,
      "learning_rate": 1.9905340924835322e-05,
      "loss": 0.9149,
      "step": 377
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0047554087343782,
      "learning_rate": 1.9904483712929094e-05,
      "loss": 0.9008,
      "step": 378
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.10640602366125,
      "learning_rate": 1.9903622655761267e-05,
      "loss": 0.9682,
      "step": 379
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0696622509575167,
      "learning_rate": 1.990275775366613e-05,
      "loss": 0.9222,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1860369361920624,
      "learning_rate": 1.9901889006979473e-05,
      "loss": 0.9479,
      "step": 381
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1420770907821463,
      "learning_rate": 1.990101641603857e-05,
      "loss": 0.9245,
      "step": 382
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0987625594348243,
      "learning_rate": 1.9900139981182193e-05,
      "loss": 0.9883,
      "step": 383
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6545814292167129,
      "learning_rate": 1.9899259702750604e-05,
      "loss": 0.2777,
      "step": 384
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0598892089105028,
      "learning_rate": 1.9898375581085555e-05,
      "loss": 0.8772,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0325400071462147,
      "learning_rate": 1.9897487616530296e-05,
      "loss": 0.8666,
      "step": 386
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0224285564651,
      "learning_rate": 1.9896595809429565e-05,
      "loss": 0.9539,
      "step": 387
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.034282228070939,
      "learning_rate": 1.9895700160129593e-05,
      "loss": 0.9072,
      "step": 388
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0372829740963514,
      "learning_rate": 1.9894800668978095e-05,
      "loss": 0.9398,
      "step": 389
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0482750376448904,
      "learning_rate": 1.9893897336324292e-05,
      "loss": 0.8566,
      "step": 390
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9431650741441973,
      "learning_rate": 1.9892990162518884e-05,
      "loss": 0.8703,
      "step": 391
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1486010962046234,
      "learning_rate": 1.9892079147914072e-05,
      "loss": 0.9095,
      "step": 392
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0154576018046595,
      "learning_rate": 1.9891164292863537e-05,
      "loss": 0.8969,
      "step": 393
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0462029997159128,
      "learning_rate": 1.9890245597722465e-05,
      "loss": 0.9176,
      "step": 394
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0202704308681423,
      "learning_rate": 1.9889323062847516e-05,
      "loss": 0.9393,
      "step": 395
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.058435857287852,
      "learning_rate": 1.988839668859686e-05,
      "loss": 0.9168,
      "step": 396
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1868266151982283,
      "learning_rate": 1.988746647533014e-05,
      "loss": 0.9127,
      "step": 397
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9623823283565845,
      "learning_rate": 1.9886532423408495e-05,
      "loss": 0.9521,
      "step": 398
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0355647093205946,
      "learning_rate": 1.9885594533194564e-05,
      "loss": 0.9603,
      "step": 399
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0603512012446292,
      "learning_rate": 1.9884652805052465e-05,
      "loss": 0.8956,
      "step": 400
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0247659247133944,
      "learning_rate": 1.9883707239347804e-05,
      "loss": 0.9115,
      "step": 401
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0890478302916504,
      "learning_rate": 1.988275783644769e-05,
      "loss": 0.9288,
      "step": 402
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.019957943440732,
      "learning_rate": 1.988180459672071e-05,
      "loss": 0.8811,
      "step": 403
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0614086470044446,
      "learning_rate": 1.988084752053695e-05,
      "loss": 0.9587,
      "step": 404
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9954483089191754,
      "learning_rate": 1.9879886608267967e-05,
      "loss": 0.8682,
      "step": 405
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.083969440658036,
      "learning_rate": 1.9878921860286832e-05,
      "loss": 0.9714,
      "step": 406
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.070418989712176,
      "learning_rate": 1.9877953276968088e-05,
      "loss": 0.9369,
      "step": 407
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.086963577033811,
      "learning_rate": 1.9876980858687777e-05,
      "loss": 0.8777,
      "step": 408
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9915674961216239,
      "learning_rate": 1.9876004605823417e-05,
      "loss": 0.9011,
      "step": 409
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.134733378121224,
      "learning_rate": 1.987502451875403e-05,
      "loss": 0.9015,
      "step": 410
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0720135552311367,
      "learning_rate": 1.987404059786012e-05,
      "loss": 0.9437,
      "step": 411
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0229180953769659,
      "learning_rate": 1.9873052843523676e-05,
      "loss": 0.889,
      "step": 412
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1040790827310152,
      "learning_rate": 1.987206125612818e-05,
      "loss": 0.9596,
      "step": 413
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1379390408413494,
      "learning_rate": 1.98710658360586e-05,
      "loss": 0.9336,
      "step": 414
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0827688628828125,
      "learning_rate": 1.987006658370139e-05,
      "loss": 0.9083,
      "step": 415
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.059370864607608,
      "learning_rate": 1.9869063499444495e-05,
      "loss": 0.9284,
      "step": 416
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.046903954112636,
      "learning_rate": 1.9868056583677346e-05,
      "loss": 0.9023,
      "step": 417
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9867294226942861,
      "learning_rate": 1.9867045836790867e-05,
      "loss": 0.8821,
      "step": 418
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.659035344391336,
      "learning_rate": 1.9866031259177463e-05,
      "loss": 0.2994,
      "step": 419
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2277195546660273,
      "learning_rate": 1.9865012851231022e-05,
      "loss": 0.9772,
      "step": 420
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1581040089896628,
      "learning_rate": 1.9863990613346936e-05,
      "loss": 0.9138,
      "step": 421
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1089023629298191,
      "learning_rate": 1.986296454592206e-05,
      "loss": 0.9378,
      "step": 422
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0249983825231066,
      "learning_rate": 1.9861934649354763e-05,
      "loss": 0.8659,
      "step": 423
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0553460303760007,
      "learning_rate": 1.9860900924044873e-05,
      "loss": 0.8738,
      "step": 424
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6407031038548503,
      "learning_rate": 1.9859863370393726e-05,
      "loss": 0.3237,
      "step": 425
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.187792381843799,
      "learning_rate": 1.9858821988804132e-05,
      "loss": 0.8677,
      "step": 426
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1213165711399207,
      "learning_rate": 1.9857776779680393e-05,
      "loss": 0.9495,
      "step": 427
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1344210894472007,
      "learning_rate": 1.98567277434283e-05,
      "loss": 0.9507,
      "step": 428
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9819374805223658,
      "learning_rate": 1.9855674880455115e-05,
      "loss": 0.9083,
      "step": 429
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0090267923965066,
      "learning_rate": 1.98546181911696e-05,
      "loss": 0.9068,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1788922255796073,
      "learning_rate": 1.9853557675982e-05,
      "loss": 0.9281,
      "step": 431
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0864529304112507,
      "learning_rate": 1.985249333530404e-05,
      "loss": 0.9533,
      "step": 432
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0765914341453022,
      "learning_rate": 1.9851425169548938e-05,
      "loss": 0.9431,
      "step": 433
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0362685775930396,
      "learning_rate": 1.9850353179131392e-05,
      "loss": 0.8992,
      "step": 434
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.05651398563554,
      "learning_rate": 1.9849277364467585e-05,
      "loss": 0.9515,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6505528196221511,
      "learning_rate": 1.984819772597518e-05,
      "loss": 0.3295,
      "step": 436
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0917954386230806,
      "learning_rate": 1.9847114264073336e-05,
      "loss": 0.8716,
      "step": 437
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0420758775513914,
      "learning_rate": 1.984602697918269e-05,
      "loss": 0.9254,
      "step": 438
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0556918205860562,
      "learning_rate": 1.9844935871725363e-05,
      "loss": 0.9038,
      "step": 439
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0238214336392735,
      "learning_rate": 1.9843840942124956e-05,
      "loss": 0.943,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0121691303821772,
      "learning_rate": 1.9842742190806566e-05,
      "loss": 0.8835,
      "step": 441
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0249235279134765,
      "learning_rate": 1.984163961819676e-05,
      "loss": 0.9526,
      "step": 442
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9762645991854163,
      "learning_rate": 1.9840533224723595e-05,
      "loss": 0.8924,
      "step": 443
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0417945320409139,
      "learning_rate": 1.9839423010816616e-05,
      "loss": 0.9028,
      "step": 444
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1250201285982142,
      "learning_rate": 1.983830897690684e-05,
      "loss": 0.8821,
      "step": 445
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1053361322414739,
      "learning_rate": 1.9837191123426777e-05,
      "loss": 0.9697,
      "step": 446
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1092814107785585,
      "learning_rate": 1.983606945081042e-05,
      "loss": 0.9625,
      "step": 447
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9977480651947346,
      "learning_rate": 1.983494395949323e-05,
      "loss": 0.942,
      "step": 448
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0207349197117936,
      "learning_rate": 1.983381464991217e-05,
      "loss": 0.9516,
      "step": 449
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9791070246434724,
      "learning_rate": 1.9832681522505676e-05,
      "loss": 0.8922,
      "step": 450
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9757480751514966,
      "learning_rate": 1.9831544577713663e-05,
      "loss": 0.8642,
      "step": 451
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9383184894539532,
      "learning_rate": 1.983040381597754e-05,
      "loss": 0.9019,
      "step": 452
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1233970083843534,
      "learning_rate": 1.982925923774018e-05,
      "loss": 0.9536,
      "step": 453
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7011005543322711,
      "learning_rate": 1.9828110843445954e-05,
      "loss": 0.3104,
      "step": 454
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6518983126787029,
      "learning_rate": 1.982695863354071e-05,
      "loss": 0.3519,
      "step": 455
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0787401745299774,
      "learning_rate": 1.9825802608471767e-05,
      "loss": 0.9098,
      "step": 456
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0252003125283013,
      "learning_rate": 1.982464276868794e-05,
      "loss": 0.8554,
      "step": 457
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1558332092759849,
      "learning_rate": 1.982347911463952e-05,
      "loss": 0.9292,
      "step": 458
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0838887572000833,
      "learning_rate": 1.9822311646778277e-05,
      "loss": 1.008,
      "step": 459
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.023485868253778,
      "learning_rate": 1.982114036555746e-05,
      "loss": 0.9498,
      "step": 460
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1699150547758583,
      "learning_rate": 1.9819965271431797e-05,
      "loss": 0.9226,
      "step": 461
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0826001540553698,
      "learning_rate": 1.9818786364857506e-05,
      "loss": 0.9364,
      "step": 462
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0885637674191846,
      "learning_rate": 1.9817603646292278e-05,
      "loss": 0.9003,
      "step": 463
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9401856464275979,
      "learning_rate": 1.9816417116195287e-05,
      "loss": 0.8963,
      "step": 464
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0146517603244287,
      "learning_rate": 1.9815226775027182e-05,
      "loss": 0.8972,
      "step": 465
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7006391428096176,
      "learning_rate": 1.9814032623250093e-05,
      "loss": 0.3164,
      "step": 466
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0848628860575587,
      "learning_rate": 1.9812834661327632e-05,
      "loss": 0.9448,
      "step": 467
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0412986909090827,
      "learning_rate": 1.9811632889724888e-05,
      "loss": 0.8916,
      "step": 468
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0108495542380274,
      "learning_rate": 1.9810427308908437e-05,
      "loss": 0.8829,
      "step": 469
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0681490845902217,
      "learning_rate": 1.9809217919346318e-05,
      "loss": 0.927,
      "step": 470
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1409826044713254,
      "learning_rate": 1.980800472150806e-05,
      "loss": 0.9079,
      "step": 471
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.093956014909104,
      "learning_rate": 1.9806787715864674e-05,
      "loss": 0.9302,
      "step": 472
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.047110649353734,
      "learning_rate": 1.9805566902888637e-05,
      "loss": 0.8881,
      "step": 473
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.050562616473448,
      "learning_rate": 1.9804342283053916e-05,
      "loss": 0.9463,
      "step": 474
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9507778169523744,
      "learning_rate": 1.980311385683594e-05,
      "loss": 0.8333,
      "step": 475
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.00203007884491,
      "learning_rate": 1.980188162471164e-05,
      "loss": 0.9326,
      "step": 476
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0394276373113438,
      "learning_rate": 1.98006455871594e-05,
      "loss": 0.8821,
      "step": 477
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1097597394023153,
      "learning_rate": 1.97994057446591e-05,
      "loss": 0.9271,
      "step": 478
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.12060733869808,
      "learning_rate": 1.979816209769209e-05,
      "loss": 0.896,
      "step": 479
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2067987322609348,
      "learning_rate": 1.9796914646741187e-05,
      "loss": 0.9561,
      "step": 480
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0730068998438107,
      "learning_rate": 1.9795663392290702e-05,
      "loss": 0.9295,
      "step": 481
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8602860712136935,
      "learning_rate": 1.9794408334826415e-05,
      "loss": 0.3296,
      "step": 482
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1086451584610169,
      "learning_rate": 1.979314947483558e-05,
      "loss": 0.9311,
      "step": 483
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9962343787516814,
      "learning_rate": 1.9791886812806932e-05,
      "loss": 0.9053,
      "step": 484
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0677753365435056,
      "learning_rate": 1.9790620349230676e-05,
      "loss": 0.8539,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6737126301525597,
      "learning_rate": 1.9789350084598504e-05,
      "loss": 0.3154,
      "step": 486
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0699606321627981,
      "learning_rate": 1.9788076019403565e-05,
      "loss": 0.9311,
      "step": 487
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0368480364327806,
      "learning_rate": 1.9786798154140507e-05,
      "loss": 0.886,
      "step": 488
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9490360489641462,
      "learning_rate": 1.9785516489305437e-05,
      "loss": 0.8801,
      "step": 489
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1954084958536368,
      "learning_rate": 1.9784231025395936e-05,
      "loss": 0.9359,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9656720647927435,
      "learning_rate": 1.9782941762911075e-05,
      "loss": 0.958,
      "step": 491
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0027987209861144,
      "learning_rate": 1.9781648702351383e-05,
      "loss": 0.9038,
      "step": 492
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9752529269707755,
      "learning_rate": 1.9780351844218874e-05,
      "loss": 0.9622,
      "step": 493
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0157854284757784,
      "learning_rate": 1.977905118901703e-05,
      "loss": 0.9373,
      "step": 494
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.926111866468234,
      "learning_rate": 1.977774673725081e-05,
      "loss": 0.8996,
      "step": 495
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.072471605599371,
      "learning_rate": 1.977643848942665e-05,
      "loss": 0.9404,
      "step": 496
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.066222416486999,
      "learning_rate": 1.977512644605246e-05,
      "loss": 0.9086,
      "step": 497
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7419541098646472,
      "learning_rate": 1.9773810607637612e-05,
      "loss": 0.3349,
      "step": 498
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0126453575175096,
      "learning_rate": 1.9772490974692962e-05,
      "loss": 0.9339,
      "step": 499
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0561151464898475,
      "learning_rate": 1.9771167547730844e-05,
      "loss": 0.9027,
      "step": 500
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0269300941146178,
      "learning_rate": 1.976984032726505e-05,
      "loss": 0.9068,
      "step": 501
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0954055505298483,
      "learning_rate": 1.976850931381086e-05,
      "loss": 0.9194,
      "step": 502
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.034558600626248,
      "learning_rate": 1.976717450788501e-05,
      "loss": 0.9143,
      "step": 503
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0350502169032825,
      "learning_rate": 1.9765835910005726e-05,
      "loss": 0.9588,
      "step": 504
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0314925679076354,
      "learning_rate": 1.9764493520692685e-05,
      "loss": 0.8901,
      "step": 505
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0901843840982026,
      "learning_rate": 1.9763147340467067e-05,
      "loss": 0.912,
      "step": 506
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9780987589935486,
      "learning_rate": 1.9761797369851498e-05,
      "loss": 0.8939,
      "step": 507
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9650460773850774,
      "learning_rate": 1.9760443609370074e-05,
      "loss": 0.8686,
      "step": 508
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9141095513069339,
      "learning_rate": 1.975908605954838e-05,
      "loss": 0.3398,
      "step": 509
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.073117355639936,
      "learning_rate": 1.9757724720913466e-05,
      "loss": 0.9429,
      "step": 510
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0756311713374274,
      "learning_rate": 1.9756359593993845e-05,
      "loss": 0.8867,
      "step": 511
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9403096749576504,
      "learning_rate": 1.975499067931951e-05,
      "loss": 0.9198,
      "step": 512
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0076977295729554,
      "learning_rate": 1.975361797742192e-05,
      "loss": 0.8636,
      "step": 513
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0930328365945423,
      "learning_rate": 1.9752241488834002e-05,
      "loss": 0.8471,
      "step": 514
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0737329228647663,
      "learning_rate": 1.975086121409016e-05,
      "loss": 0.9502,
      "step": 515
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0107408267855122,
      "learning_rate": 1.974947715372626e-05,
      "loss": 0.9221,
      "step": 516
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0740691622705583,
      "learning_rate": 1.974808930827965e-05,
      "loss": 0.9185,
      "step": 517
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0389721022627627,
      "learning_rate": 1.9746697678289128e-05,
      "loss": 0.9137,
      "step": 518
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0877191061228284,
      "learning_rate": 1.9745302264294982e-05,
      "loss": 0.9728,
      "step": 519
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0922048960867272,
      "learning_rate": 1.9743903066838954e-05,
      "loss": 0.9373,
      "step": 520
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0080218081601238,
      "learning_rate": 1.9742500086464266e-05,
      "loss": 0.892,
      "step": 521
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0274831658895955,
      "learning_rate": 1.9741093323715597e-05,
      "loss": 0.8544,
      "step": 522
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0741899245902944,
      "learning_rate": 1.9739682779139107e-05,
      "loss": 0.9322,
      "step": 523
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0186044673056691,
      "learning_rate": 1.9738268453282414e-05,
      "loss": 0.8625,
      "step": 524
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9741556770954737,
      "learning_rate": 1.9736850346694608e-05,
      "loss": 0.8652,
      "step": 525
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0645144689889867,
      "learning_rate": 1.973542845992625e-05,
      "loss": 0.9755,
      "step": 526
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1031249709406246,
      "learning_rate": 1.9734002793529362e-05,
      "loss": 0.9263,
      "step": 527
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9356476320351311,
      "learning_rate": 1.9732573348057437e-05,
      "loss": 0.9016,
      "step": 528
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.998276805799876,
      "learning_rate": 1.973114012406544e-05,
      "loss": 0.8689,
      "step": 529
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0059198249087018,
      "learning_rate": 1.9729703122109788e-05,
      "loss": 0.9257,
      "step": 530
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.00101772340023,
      "learning_rate": 1.9728262342748384e-05,
      "loss": 0.9166,
      "step": 531
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9426540603287022,
      "learning_rate": 1.9726817786540584e-05,
      "loss": 0.899,
      "step": 532
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.101397633685596,
      "learning_rate": 1.9725369454047215e-05,
      "loss": 0.8587,
      "step": 533
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1211013385593964,
      "learning_rate": 1.9723917345830568e-05,
      "loss": 0.8894,
      "step": 534
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1235025561550558,
      "learning_rate": 1.9722461462454405e-05,
      "loss": 0.9118,
      "step": 535
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.046277013084084,
      "learning_rate": 1.9721001804483947e-05,
      "loss": 0.8968,
      "step": 536
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9919709418440723,
      "learning_rate": 1.9719538372485887e-05,
      "loss": 0.9436,
      "step": 537
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0206416156424283,
      "learning_rate": 1.9718071167028376e-05,
      "loss": 0.8688,
      "step": 538
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1630443562362291,
      "learning_rate": 1.9716600188681038e-05,
      "loss": 0.8833,
      "step": 539
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9763970065556874,
      "learning_rate": 1.971512543801495e-05,
      "loss": 0.8654,
      "step": 540
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9943744095530666,
      "learning_rate": 1.9713646915602663e-05,
      "loss": 0.8865,
      "step": 541
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0377006475159611,
      "learning_rate": 1.9712164622018197e-05,
      "loss": 0.9498,
      "step": 542
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.977695697745445,
      "learning_rate": 1.9710678557837024e-05,
      "loss": 0.97,
      "step": 543
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0727984986046473,
      "learning_rate": 1.9709188723636088e-05,
      "loss": 0.9442,
      "step": 544
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0600801851822703,
      "learning_rate": 1.970769511999379e-05,
      "loss": 0.9294,
      "step": 545
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0373886856128742,
      "learning_rate": 1.9706197747490004e-05,
      "loss": 0.9299,
      "step": 546
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9928032082122716,
      "learning_rate": 1.9704696606706055e-05,
      "loss": 0.9053,
      "step": 547
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9847366665808235,
      "learning_rate": 1.9703191698224742e-05,
      "loss": 0.8271,
      "step": 548
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.052355692437453,
      "learning_rate": 1.9701683022630323e-05,
      "loss": 0.9005,
      "step": 549
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9865624994419027,
      "learning_rate": 1.9700170580508514e-05,
      "loss": 0.9264,
      "step": 550
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9575108959843508,
      "learning_rate": 1.9698654372446495e-05,
      "loss": 0.8997,
      "step": 551
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0184174202401635,
      "learning_rate": 1.969713439903292e-05,
      "loss": 0.9634,
      "step": 552
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0592845592954634,
      "learning_rate": 1.9695610660857886e-05,
      "loss": 0.9273,
      "step": 553
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1163332263910863,
      "learning_rate": 1.9694083158512965e-05,
      "loss": 0.9407,
      "step": 554
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0303853335791702,
      "learning_rate": 1.9692551892591185e-05,
      "loss": 0.9594,
      "step": 555
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9523928272390737,
      "learning_rate": 1.9691016863687037e-05,
      "loss": 0.8885,
      "step": 556
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.111725044098634,
      "learning_rate": 1.968947807239647e-05,
      "loss": 0.8945,
      "step": 557
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0725906694543348,
      "learning_rate": 1.9687935519316897e-05,
      "loss": 0.9645,
      "step": 558
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0996575442379461,
      "learning_rate": 1.9686389205047186e-05,
      "loss": 0.3233,
      "step": 559
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9852247073799738,
      "learning_rate": 1.9684839130187678e-05,
      "loss": 0.9728,
      "step": 560
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0888390099082714,
      "learning_rate": 1.968328529534016e-05,
      "loss": 0.9052,
      "step": 561
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1140388351748562,
      "learning_rate": 1.9681727701107885e-05,
      "loss": 0.8627,
      "step": 562
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1099164975603883,
      "learning_rate": 1.9680166348095568e-05,
      "loss": 0.9213,
      "step": 563
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0293415809057698,
      "learning_rate": 1.967860123690937e-05,
      "loss": 0.8934,
      "step": 564
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9867220026054747,
      "learning_rate": 1.9677032368156934e-05,
      "loss": 0.9315,
      "step": 565
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9848899487812687,
      "learning_rate": 1.967545974244734e-05,
      "loss": 0.8749,
      "step": 566
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0113523790616061,
      "learning_rate": 1.9673883360391138e-05,
      "loss": 0.9005,
      "step": 567
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0388745929232106,
      "learning_rate": 1.9672303222600333e-05,
      "loss": 0.8308,
      "step": 568
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9867149417183403,
      "learning_rate": 1.967071932968839e-05,
      "loss": 0.9552,
      "step": 569
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8576158855847958,
      "learning_rate": 1.9669131682270232e-05,
      "loss": 0.3215,
      "step": 570
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0408422762342415,
      "learning_rate": 1.9667540280962235e-05,
      "loss": 0.9323,
      "step": 571
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.924350776930073,
      "learning_rate": 1.966594512638224e-05,
      "loss": 0.8338,
      "step": 572
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0447424719004375,
      "learning_rate": 1.9664346219149538e-05,
      "loss": 0.8937,
      "step": 573
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0302644069330094,
      "learning_rate": 1.966274355988488e-05,
      "loss": 0.9192,
      "step": 574
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9326157881940959,
      "learning_rate": 1.9661137149210473e-05,
      "loss": 0.8886,
      "step": 575
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0245061690704076,
      "learning_rate": 1.9659526987749987e-05,
      "loss": 0.8825,
      "step": 576
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9418529165862767,
      "learning_rate": 1.9657913076128532e-05,
      "loss": 0.9306,
      "step": 577
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0859006930792712,
      "learning_rate": 1.965629541497269e-05,
      "loss": 0.9177,
      "step": 578
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.070762797550408,
      "learning_rate": 1.9654674004910493e-05,
      "loss": 0.8666,
      "step": 579
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.976309952266866,
      "learning_rate": 1.9653048846571427e-05,
      "loss": 0.9106,
      "step": 580
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0467189918999822,
      "learning_rate": 1.9651419940586437e-05,
      "loss": 0.9007,
      "step": 581
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0052924637247473,
      "learning_rate": 1.964978728758791e-05,
      "loss": 0.9711,
      "step": 582
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.060102685664137,
      "learning_rate": 1.9648150888209715e-05,
      "loss": 0.8908,
      "step": 583
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0326645475904999,
      "learning_rate": 1.9646510743087144e-05,
      "loss": 0.9414,
      "step": 584
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0656783039396664,
      "learning_rate": 1.964486685285697e-05,
      "loss": 0.9426,
      "step": 585
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0942576375608535,
      "learning_rate": 1.9643219218157395e-05,
      "loss": 0.8611,
      "step": 586
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9473630203731953,
      "learning_rate": 1.9641567839628092e-05,
      "loss": 0.8998,
      "step": 587
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9962602307938256,
      "learning_rate": 1.963991271791019e-05,
      "loss": 0.8208,
      "step": 588
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0483546917809532,
      "learning_rate": 1.9638253853646255e-05,
      "loss": 0.8316,
      "step": 589
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0528547949294274,
      "learning_rate": 1.9636591247480323e-05,
      "loss": 0.8749,
      "step": 590
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1422346764526765,
      "learning_rate": 1.9634924900057867e-05,
      "loss": 0.9876,
      "step": 591
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0348841462493712,
      "learning_rate": 1.963325481202583e-05,
      "loss": 0.8766,
      "step": 592
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0221700534651696,
      "learning_rate": 1.963158098403259e-05,
      "loss": 0.8574,
      "step": 593
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0023043460303467,
      "learning_rate": 1.9629903416727987e-05,
      "loss": 0.8726,
      "step": 594
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.700267110053785,
      "learning_rate": 1.962822211076331e-05,
      "loss": 0.328,
      "step": 595
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0559942433136433,
      "learning_rate": 1.96265370667913e-05,
      "loss": 0.8895,
      "step": 596
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.068822484223205,
      "learning_rate": 1.9624848285466146e-05,
      "loss": 0.8663,
      "step": 597
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9751919316155465,
      "learning_rate": 1.9623155767443498e-05,
      "loss": 0.8959,
      "step": 598
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9870814270030241,
      "learning_rate": 1.9621459513380445e-05,
      "loss": 0.9107,
      "step": 599
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.969400910395,
      "learning_rate": 1.9619759523935532e-05,
      "loss": 0.9355,
      "step": 600
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.039953361257672,
      "learning_rate": 1.9618055799768757e-05,
      "loss": 0.845,
      "step": 601
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9524451062605711,
      "learning_rate": 1.961634834154156e-05,
      "loss": 0.9049,
      "step": 602
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6398051998003157,
      "learning_rate": 1.9614637149916834e-05,
      "loss": 0.3602,
      "step": 603
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0069835298670196,
      "learning_rate": 1.9612922225558924e-05,
      "loss": 0.9498,
      "step": 604
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1280363280181527,
      "learning_rate": 1.961120356913363e-05,
      "loss": 0.8635,
      "step": 605
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0579986139833863,
      "learning_rate": 1.960948118130818e-05,
      "loss": 0.9659,
      "step": 606
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0390630648995063,
      "learning_rate": 1.9607755062751273e-05,
      "loss": 0.9035,
      "step": 607
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9952440274593469,
      "learning_rate": 1.9606025214133046e-05,
      "loss": 0.9255,
      "step": 608
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9726695231935732,
      "learning_rate": 1.9604291636125084e-05,
      "loss": 0.9108,
      "step": 609
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0564757769831479,
      "learning_rate": 1.960255432940043e-05,
      "loss": 0.9302,
      "step": 610
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0397521762790012,
      "learning_rate": 1.9600813294633552e-05,
      "loss": 0.9036,
      "step": 611
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6532296629598495,
      "learning_rate": 1.9599068532500394e-05,
      "loss": 0.3059,
      "step": 612
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0271925085122335,
      "learning_rate": 1.9597320043678322e-05,
      "loss": 0.8837,
      "step": 613
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1240313939410338,
      "learning_rate": 1.9595567828846166e-05,
      "loss": 0.9012,
      "step": 614
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0140589267273943,
      "learning_rate": 1.9593811888684192e-05,
      "loss": 0.8393,
      "step": 615
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9740147330602928,
      "learning_rate": 1.9592052223874115e-05,
      "loss": 0.9075,
      "step": 616
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9822987503950591,
      "learning_rate": 1.959028883509911e-05,
      "loss": 0.8981,
      "step": 617
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9565288666556554,
      "learning_rate": 1.9588521723043764e-05,
      "loss": 0.921,
      "step": 618
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.014654559607114,
      "learning_rate": 1.958675088839415e-05,
      "loss": 0.8913,
      "step": 619
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0315667425935544,
      "learning_rate": 1.9584976331837758e-05,
      "loss": 0.8552,
      "step": 620
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.017538758402361,
      "learning_rate": 1.9583198054063535e-05,
      "loss": 0.8756,
      "step": 621
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0085205258937346,
      "learning_rate": 1.9581416055761865e-05,
      "loss": 0.872,
      "step": 622
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7177303332082328,
      "learning_rate": 1.9579630337624585e-05,
      "loss": 0.3191,
      "step": 623
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0202586246227798,
      "learning_rate": 1.9577840900344974e-05,
      "loss": 0.949,
      "step": 624
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0273845514020463,
      "learning_rate": 1.9576047744617752e-05,
      "loss": 0.908,
      "step": 625
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5945886300676403,
      "learning_rate": 1.957425087113908e-05,
      "loss": 0.3274,
      "step": 626
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.142814973385369,
      "learning_rate": 1.9572450280606568e-05,
      "loss": 0.957,
      "step": 627
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9795496644207347,
      "learning_rate": 1.9570645973719273e-05,
      "loss": 0.9166,
      "step": 628
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0968135733331537,
      "learning_rate": 1.9568837951177677e-05,
      "loss": 0.9464,
      "step": 629
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0361264756248911,
      "learning_rate": 1.9567026213683728e-05,
      "loss": 0.9509,
      "step": 630
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7927393302455275,
      "learning_rate": 1.9565210761940798e-05,
      "loss": 0.3494,
      "step": 631
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0618706451638107,
      "learning_rate": 1.956339159665371e-05,
      "loss": 0.8816,
      "step": 632
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9681287434692115,
      "learning_rate": 1.956156871852873e-05,
      "loss": 0.8839,
      "step": 633
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6249896242353356,
      "learning_rate": 1.9559742128273558e-05,
      "loss": 0.3159,
      "step": 634
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.990949063167081,
      "learning_rate": 1.9557911826597337e-05,
      "loss": 0.9002,
      "step": 635
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0629238701415464,
      "learning_rate": 1.9556077814210662e-05,
      "loss": 0.8723,
      "step": 636
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0146532281187925,
      "learning_rate": 1.955424009182555e-05,
      "loss": 0.9641,
      "step": 637
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0193263137724529,
      "learning_rate": 1.955239866015547e-05,
      "loss": 0.9105,
      "step": 638
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.084243135239892,
      "learning_rate": 1.9550553519915335e-05,
      "loss": 0.9412,
      "step": 639
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1101230211674127,
      "learning_rate": 1.954870467182149e-05,
      "loss": 0.8869,
      "step": 640
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9456380056363715,
      "learning_rate": 1.954685211659172e-05,
      "loss": 0.945,
      "step": 641
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9831918877860092,
      "learning_rate": 1.9544995854945248e-05,
      "loss": 0.8604,
      "step": 642
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9998988685050143,
      "learning_rate": 1.954313588760274e-05,
      "loss": 0.8597,
      "step": 643
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0616374863690747,
      "learning_rate": 1.9541272215286304e-05,
      "loss": 0.9092,
      "step": 644
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0144445211650113,
      "learning_rate": 1.9539404838719477e-05,
      "loss": 0.8758,
      "step": 645
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9445177872607831,
      "learning_rate": 1.9537533758627242e-05,
      "loss": 0.8999,
      "step": 646
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.041491997548341,
      "learning_rate": 1.953565897573601e-05,
      "loss": 0.8467,
      "step": 647
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0086013170559758,
      "learning_rate": 1.9533780490773645e-05,
      "loss": 0.883,
      "step": 648
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.086510032644491,
      "learning_rate": 1.9531898304469435e-05,
      "loss": 0.8858,
      "step": 649
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8114459546485628,
      "learning_rate": 1.953001241755411e-05,
      "loss": 0.3096,
      "step": 650
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0016909490063928,
      "learning_rate": 1.952812283075984e-05,
      "loss": 0.8366,
      "step": 651
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9991056050958834,
      "learning_rate": 1.952622954482022e-05,
      "loss": 0.9087,
      "step": 652
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9172441506578952,
      "learning_rate": 1.9524332560470293e-05,
      "loss": 0.8762,
      "step": 653
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.021182761055864,
      "learning_rate": 1.9522431878446536e-05,
      "loss": 0.8867,
      "step": 654
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0558404528100394,
      "learning_rate": 1.9520527499486856e-05,
      "loss": 0.8569,
      "step": 655
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.053141661891204,
      "learning_rate": 1.95186194243306e-05,
      "loss": 0.8939,
      "step": 656
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0247369237913335,
      "learning_rate": 1.9516707653718546e-05,
      "loss": 0.9464,
      "step": 657
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0242473709412496,
      "learning_rate": 1.9514792188392914e-05,
      "loss": 0.8862,
      "step": 658
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9213526061939562,
      "learning_rate": 1.9512873029097347e-05,
      "loss": 0.851,
      "step": 659
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9023800604964348,
      "learning_rate": 1.9510950176576933e-05,
      "loss": 0.8957,
      "step": 660
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9840843503878705,
      "learning_rate": 1.950902363157819e-05,
      "loss": 0.8998,
      "step": 661
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1062045973937527,
      "learning_rate": 1.950709339484907e-05,
      "loss": 0.8801,
      "step": 662
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9743020836044194,
      "learning_rate": 1.9505159467138954e-05,
      "loss": 0.906,
      "step": 663
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9959868200726724,
      "learning_rate": 1.9503221849198655e-05,
      "loss": 0.325,
      "step": 664
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0707015264635562,
      "learning_rate": 1.9501280541780435e-05,
      "loss": 0.9033,
      "step": 665
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1067829446036443,
      "learning_rate": 1.9499335545637968e-05,
      "loss": 0.9604,
      "step": 666
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6391890155390987,
      "learning_rate": 1.949738686152637e-05,
      "loss": 0.3355,
      "step": 667
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1462482171801385,
      "learning_rate": 1.9495434490202188e-05,
      "loss": 0.9481,
      "step": 668
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9635986376256932,
      "learning_rate": 1.94934784324234e-05,
      "loss": 0.8352,
      "step": 669
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0069021749862115,
      "learning_rate": 1.9491518688949417e-05,
      "loss": 0.9216,
      "step": 670
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0192626903372848,
      "learning_rate": 1.9489555260541074e-05,
      "loss": 0.8996,
      "step": 671
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0556614344887922,
      "learning_rate": 1.948758814796064e-05,
      "loss": 0.943,
      "step": 672
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.921277452503479,
      "learning_rate": 1.9485617351971827e-05,
      "loss": 0.8389,
      "step": 673
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9938894141350295,
      "learning_rate": 1.9483642873339753e-05,
      "loss": 0.893,
      "step": 674
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.066203979812592,
      "learning_rate": 1.9481664712830987e-05,
      "loss": 0.9191,
      "step": 675
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.075255373248265,
      "learning_rate": 1.9479682871213515e-05,
      "loss": 0.8978,
      "step": 676
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.995800637146686,
      "learning_rate": 1.9477697349256756e-05,
      "loss": 0.9189,
      "step": 677
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9785425023996115,
      "learning_rate": 1.947570814773156e-05,
      "loss": 0.9005,
      "step": 678
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9484412204665139,
      "learning_rate": 1.9473715267410206e-05,
      "loss": 0.891,
      "step": 679
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.937973424465193,
      "learning_rate": 1.9471718709066392e-05,
      "loss": 0.941,
      "step": 680
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1802257253837194,
      "learning_rate": 1.9469718473475256e-05,
      "loss": 0.935,
      "step": 681
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.008437740824162,
      "learning_rate": 1.9467714561413358e-05,
      "loss": 0.9307,
      "step": 682
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9237544154345879,
      "learning_rate": 1.9465706973658683e-05,
      "loss": 0.8248,
      "step": 683
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0078539918335259,
      "learning_rate": 1.9463695710990648e-05,
      "loss": 0.3513,
      "step": 684
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9570704241685571,
      "learning_rate": 1.946168077419009e-05,
      "loss": 0.8546,
      "step": 685
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.105140576034043,
      "learning_rate": 1.9459662164039283e-05,
      "loss": 0.931,
      "step": 686
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9141565445414078,
      "learning_rate": 1.9457639881321917e-05,
      "loss": 0.877,
      "step": 687
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0097781632261076,
      "learning_rate": 1.9455613926823115e-05,
      "loss": 0.8482,
      "step": 688
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9479844617693002,
      "learning_rate": 1.945358430132942e-05,
      "loss": 0.8926,
      "step": 689
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.202081249228092,
      "learning_rate": 1.9451551005628803e-05,
      "loss": 1.0096,
      "step": 690
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9553351078491228,
      "learning_rate": 1.9449514040510654e-05,
      "loss": 0.8628,
      "step": 691
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.040393470769721,
      "learning_rate": 1.9447473406765803e-05,
      "loss": 0.8698,
      "step": 692
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8588652403424635,
      "learning_rate": 1.9445429105186487e-05,
      "loss": 0.8272,
      "step": 693
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0298422885292464,
      "learning_rate": 1.9443381136566382e-05,
      "loss": 0.944,
      "step": 694
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9703615335203106,
      "learning_rate": 1.9441329501700568e-05,
      "loss": 0.8825,
      "step": 695
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1116791911887947,
      "learning_rate": 1.943927420138557e-05,
      "loss": 0.9409,
      "step": 696
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9537126193479966,
      "learning_rate": 1.9437215236419322e-05,
      "loss": 0.8212,
      "step": 697
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9200129916893326,
      "learning_rate": 1.9435152607601187e-05,
      "loss": 0.8869,
      "step": 698
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9859983048566212,
      "learning_rate": 1.943308631573195e-05,
      "loss": 0.8524,
      "step": 699
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9448920825617617,
      "learning_rate": 1.9431016361613816e-05,
      "loss": 0.8489,
      "step": 700
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9683125602072604,
      "learning_rate": 1.9428942746050406e-05,
      "loss": 0.9088,
      "step": 701
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0568688680696083,
      "learning_rate": 1.9426865469846773e-05,
      "loss": 0.8989,
      "step": 702
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.037932138273204,
      "learning_rate": 1.9424784533809393e-05,
      "loss": 0.8786,
      "step": 703
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9291792828340132,
      "learning_rate": 1.942269993874615e-05,
      "loss": 0.8476,
      "step": 704
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9982258123988965,
      "learning_rate": 1.9420611685466358e-05,
      "loss": 0.8185,
      "step": 705
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9926539927426091,
      "learning_rate": 1.9418519774780748e-05,
      "loss": 0.8959,
      "step": 706
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.047837582367839,
      "learning_rate": 1.9416424207501474e-05,
      "loss": 0.9062,
      "step": 707
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9402102769517983,
      "learning_rate": 1.9414324984442102e-05,
      "loss": 0.8642,
      "step": 708
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1235993713867702,
      "learning_rate": 1.9412222106417632e-05,
      "loss": 0.9037,
      "step": 709
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9734247709000003,
      "learning_rate": 1.9410115574244462e-05,
      "loss": 0.8621,
      "step": 710
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.985382054086815,
      "learning_rate": 1.9408005388740433e-05,
      "loss": 0.907,
      "step": 711
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0089427958029917,
      "learning_rate": 1.9405891550724778e-05,
      "loss": 0.8744,
      "step": 712
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0237323048829965,
      "learning_rate": 1.940377406101817e-05,
      "loss": 0.8691,
      "step": 713
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.105245615816628,
      "learning_rate": 1.9401652920442694e-05,
      "loss": 0.9363,
      "step": 714
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0511871833636357,
      "learning_rate": 1.9399528129821842e-05,
      "loss": 0.9281,
      "step": 715
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0046065841037237,
      "learning_rate": 1.939739968998054e-05,
      "loss": 0.8631,
      "step": 716
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9069262670740837,
      "learning_rate": 1.939526760174511e-05,
      "loss": 0.8926,
      "step": 717
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0966120821784187,
      "learning_rate": 1.939313186594331e-05,
      "loss": 0.9145,
      "step": 718
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.990262948725273,
      "learning_rate": 1.9390992483404308e-05,
      "loss": 0.9214,
      "step": 719
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0295589043825053,
      "learning_rate": 1.938884945495868e-05,
      "loss": 0.8654,
      "step": 720
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0123169862727623,
      "learning_rate": 1.9386702781438425e-05,
      "loss": 0.9007,
      "step": 721
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9596730515576505,
      "learning_rate": 1.938455246367696e-05,
      "loss": 0.9164,
      "step": 722
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9819070739190467,
      "learning_rate": 1.9382398502509107e-05,
      "loss": 0.9138,
      "step": 723
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9648950390369071,
      "learning_rate": 1.938024089877111e-05,
      "loss": 0.8621,
      "step": 724
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0425410367000871,
      "learning_rate": 1.9378079653300624e-05,
      "loss": 0.9311,
      "step": 725
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9591927583612092,
      "learning_rate": 1.9375914766936723e-05,
      "loss": 0.348,
      "step": 726
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1193923738818012,
      "learning_rate": 1.9373746240519884e-05,
      "loss": 0.896,
      "step": 727
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9265840759969621,
      "learning_rate": 1.937157407489201e-05,
      "loss": 0.8852,
      "step": 728
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.024109608212135,
      "learning_rate": 1.9369398270896403e-05,
      "loss": 0.8651,
      "step": 729
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9724060228881831,
      "learning_rate": 1.936721882937779e-05,
      "loss": 0.9485,
      "step": 730
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9575219154328227,
      "learning_rate": 1.9365035751182307e-05,
      "loss": 0.882,
      "step": 731
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9768261218684733,
      "learning_rate": 1.93628490371575e-05,
      "loss": 0.8681,
      "step": 732
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9818696624121271,
      "learning_rate": 1.9360658688152322e-05,
      "loss": 0.9191,
      "step": 733
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9860102347779445,
      "learning_rate": 1.9358464705017143e-05,
      "loss": 0.8224,
      "step": 734
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9643198791407219,
      "learning_rate": 1.9356267088603745e-05,
      "loss": 0.7989,
      "step": 735
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9444401831668692,
      "learning_rate": 1.9354065839765316e-05,
      "loss": 0.9566,
      "step": 736
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1127621703763082,
      "learning_rate": 1.9351860959356462e-05,
      "loss": 0.9131,
      "step": 737
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9912640747616286,
      "learning_rate": 1.9349652448233187e-05,
      "loss": 0.8803,
      "step": 738
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.953423968117197,
      "learning_rate": 1.934744030725291e-05,
      "loss": 0.8755,
      "step": 739
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0145729351681743,
      "learning_rate": 1.934522453727447e-05,
      "loss": 0.8724,
      "step": 740
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.159667992240504,
      "learning_rate": 1.93430051391581e-05,
      "loss": 0.8609,
      "step": 741
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4035763337764988,
      "learning_rate": 1.934078211376544e-05,
      "loss": 0.9248,
      "step": 742
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9588028316884541,
      "learning_rate": 1.9338555461959554e-05,
      "loss": 0.8873,
      "step": 743
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.004098781211947,
      "learning_rate": 1.93363251846049e-05,
      "loss": 0.8615,
      "step": 744
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0076171041944468,
      "learning_rate": 1.9334091282567352e-05,
      "loss": 0.8793,
      "step": 745
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1019190114554638,
      "learning_rate": 1.9331853756714185e-05,
      "loss": 0.9519,
      "step": 746
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9997542950684323,
      "learning_rate": 1.9329612607914088e-05,
      "loss": 0.9242,
      "step": 747
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1014557907866487,
      "learning_rate": 1.9327367837037142e-05,
      "loss": 0.8882,
      "step": 748
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0342007015540966,
      "learning_rate": 1.9325119444954855e-05,
      "loss": 0.8745,
      "step": 749
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8605200061919471,
      "learning_rate": 1.9322867432540126e-05,
      "loss": 0.352,
      "step": 750
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0256166718731885,
      "learning_rate": 1.9320611800667268e-05,
      "loss": 0.9057,
      "step": 751
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0148165167083076,
      "learning_rate": 1.9318352550211986e-05,
      "loss": 0.8432,
      "step": 752
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0292582018298924,
      "learning_rate": 1.9316089682051403e-05,
      "loss": 0.8666,
      "step": 753
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.012086527350091,
      "learning_rate": 1.9313823197064042e-05,
      "loss": 0.8153,
      "step": 754
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0095567980751223,
      "learning_rate": 1.9311553096129835e-05,
      "loss": 0.9326,
      "step": 755
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.009817360160476,
      "learning_rate": 1.9309279380130112e-05,
      "loss": 0.8934,
      "step": 756
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0233525543959727,
      "learning_rate": 1.93070020499476e-05,
      "loss": 0.8912,
      "step": 757
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.026526183817067,
      "learning_rate": 1.930472110646645e-05,
      "loss": 0.8281,
      "step": 758
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6194829422237531,
      "learning_rate": 1.9302436550572187e-05,
      "loss": 0.3008,
      "step": 759
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9546538290057851,
      "learning_rate": 1.930014838315177e-05,
      "loss": 0.843,
      "step": 760
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0229195672428766,
      "learning_rate": 1.9297856605093534e-05,
      "loss": 0.8855,
      "step": 761
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0164667165133634,
      "learning_rate": 1.9295561217287226e-05,
      "loss": 0.8853,
      "step": 762
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0883343958320837,
      "learning_rate": 1.9293262220624002e-05,
      "loss": 0.8947,
      "step": 763
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9494309155128452,
      "learning_rate": 1.9290959615996407e-05,
      "loss": 0.9085,
      "step": 764
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9902548284694317,
      "learning_rate": 1.9288653404298392e-05,
      "loss": 0.9342,
      "step": 765
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0142642600469607,
      "learning_rate": 1.9286343586425307e-05,
      "loss": 0.8678,
      "step": 766
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0530662695742425,
      "learning_rate": 1.9284030163273907e-05,
      "loss": 0.9063,
      "step": 767
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.004719952114633,
      "learning_rate": 1.9281713135742333e-05,
      "loss": 0.864,
      "step": 768
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9813915196427986,
      "learning_rate": 1.9279392504730147e-05,
      "loss": 0.9056,
      "step": 769
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9767295937637226,
      "learning_rate": 1.9277068271138287e-05,
      "loss": 0.8496,
      "step": 770
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.999935284491193,
      "learning_rate": 1.9274740435869107e-05,
      "loss": 0.9119,
      "step": 771
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9423968461083797,
      "learning_rate": 1.927240899982635e-05,
      "loss": 0.8476,
      "step": 772
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9239096866385611,
      "learning_rate": 1.9270073963915162e-05,
      "loss": 0.9107,
      "step": 773
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9853646855950166,
      "learning_rate": 1.9267735329042086e-05,
      "loss": 0.929,
      "step": 774
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9890292391732491,
      "learning_rate": 1.9265393096115056e-05,
      "loss": 0.8522,
      "step": 775
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0095850929092314,
      "learning_rate": 1.926304726604341e-05,
      "loss": 0.8566,
      "step": 776
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9805512823365669,
      "learning_rate": 1.9260697839737875e-05,
      "loss": 0.9079,
      "step": 777
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0715348875354558,
      "learning_rate": 1.925834481811059e-05,
      "loss": 0.8224,
      "step": 778
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.945850839457018,
      "learning_rate": 1.9255988202075065e-05,
      "loss": 0.8939,
      "step": 779
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0126678373174234,
      "learning_rate": 1.925362799254623e-05,
      "loss": 0.9102,
      "step": 780
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9659566032175706,
      "learning_rate": 1.9251264190440398e-05,
      "loss": 0.9111,
      "step": 781
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0254775780185623,
      "learning_rate": 1.9248896796675277e-05,
      "loss": 0.8857,
      "step": 782
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.966190476087147,
      "learning_rate": 1.924652581216997e-05,
      "loss": 0.8943,
      "step": 783
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.946334883153325,
      "learning_rate": 1.9244151237844975e-05,
      "loss": 0.9176,
      "step": 784
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0178051091791205,
      "learning_rate": 1.9241773074622182e-05,
      "loss": 0.8905,
      "step": 785
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0123459505118566,
      "learning_rate": 1.923939132342488e-05,
      "loss": 0.8552,
      "step": 786
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0555821827233511,
      "learning_rate": 1.923700598517775e-05,
      "loss": 0.9328,
      "step": 787
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9982417062498129,
      "learning_rate": 1.923461706080685e-05,
      "loss": 0.8895,
      "step": 788
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9591122833788951,
      "learning_rate": 1.923222455123965e-05,
      "loss": 0.8959,
      "step": 789
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8646305926303756,
      "learning_rate": 1.9229828457405005e-05,
      "loss": 0.8479,
      "step": 790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9353634637331191,
      "learning_rate": 1.9227428780233162e-05,
      "loss": 0.9177,
      "step": 791
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0257511476964827,
      "learning_rate": 1.922502552065576e-05,
      "loss": 0.8732,
      "step": 792
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.047035475208104,
      "learning_rate": 1.922261867960582e-05,
      "loss": 0.857,
      "step": 793
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.972355986308515,
      "learning_rate": 1.9220208258017763e-05,
      "loss": 0.8743,
      "step": 794
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0399716966598544,
      "learning_rate": 1.92177942568274e-05,
      "loss": 0.9053,
      "step": 795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9167434176697662,
      "learning_rate": 1.921537667697193e-05,
      "loss": 0.8935,
      "step": 796
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9220908147750175,
      "learning_rate": 1.9212955519389938e-05,
      "loss": 0.9226,
      "step": 797
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0164624692962692,
      "learning_rate": 1.9210530785021405e-05,
      "loss": 0.8926,
      "step": 798
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9390305826133173,
      "learning_rate": 1.9208102474807692e-05,
      "loss": 0.9083,
      "step": 799
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9904308731932963,
      "learning_rate": 1.920567058969155e-05,
      "loss": 0.869,
      "step": 800
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0589571299226501,
      "learning_rate": 1.920323513061713e-05,
      "loss": 0.9272,
      "step": 801
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9812060128458221,
      "learning_rate": 1.9200796098529956e-05,
      "loss": 0.8737,
      "step": 802
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0347557272560106,
      "learning_rate": 1.919835349437694e-05,
      "loss": 0.9165,
      "step": 803
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0544527388603995,
      "learning_rate": 1.9195907319106394e-05,
      "loss": 0.8366,
      "step": 804
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1155554894522726,
      "learning_rate": 1.9193457573667996e-05,
      "loss": 0.9157,
      "step": 805
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0416630669372666,
      "learning_rate": 1.919100425901283e-05,
      "loss": 0.9144,
      "step": 806
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.013275643613882,
      "learning_rate": 1.9188547376093355e-05,
      "loss": 0.8536,
      "step": 807
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9849876757641738,
      "learning_rate": 1.918608692586342e-05,
      "loss": 0.8575,
      "step": 808
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0230270546181703,
      "learning_rate": 1.918362290927825e-05,
      "loss": 0.8767,
      "step": 809
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0300768355045773,
      "learning_rate": 1.9181155327294468e-05,
      "loss": 0.8989,
      "step": 810
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9201965734337391,
      "learning_rate": 1.9178684180870072e-05,
      "loss": 0.8711,
      "step": 811
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9689782235762201,
      "learning_rate": 1.9176209470964446e-05,
      "loss": 0.8757,
      "step": 812
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0598294007276035,
      "learning_rate": 1.9173731198538354e-05,
      "loss": 0.8758,
      "step": 813
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9813993981757178,
      "learning_rate": 1.9171249364553956e-05,
      "loss": 0.886,
      "step": 814
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0290571130091501,
      "learning_rate": 1.9168763969974773e-05,
      "loss": 0.8732,
      "step": 815
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9288181566195929,
      "learning_rate": 1.916627501576573e-05,
      "loss": 0.9195,
      "step": 816
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9902895251600329,
      "learning_rate": 1.916378250289312e-05,
      "loss": 0.8228,
      "step": 817
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.018378178566844,
      "learning_rate": 1.9161286432324628e-05,
      "loss": 0.9145,
      "step": 818
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0271428427586473,
      "learning_rate": 1.9158786805029307e-05,
      "loss": 0.8222,
      "step": 819
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0426762305495705,
      "learning_rate": 1.9156283621977603e-05,
      "loss": 0.9189,
      "step": 820
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0489699451634227,
      "learning_rate": 1.9153776884141336e-05,
      "loss": 0.9251,
      "step": 821
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1023309822140657,
      "learning_rate": 1.915126659249371e-05,
      "loss": 0.9444,
      "step": 822
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9917218823428156,
      "learning_rate": 1.9148752748009304e-05,
      "loss": 0.8525,
      "step": 823
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9736513863736533,
      "learning_rate": 1.914623535166408e-05,
      "loss": 0.8065,
      "step": 824
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7035095977630922,
      "learning_rate": 1.9143714404435382e-05,
      "loss": 0.3557,
      "step": 825
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0501685690662765,
      "learning_rate": 1.9141189907301922e-05,
      "loss": 0.8994,
      "step": 826
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0338529861211843,
      "learning_rate": 1.9138661861243802e-05,
      "loss": 0.8274,
      "step": 827
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8947890769827717,
      "learning_rate": 1.913613026724249e-05,
      "loss": 0.8839,
      "step": 828
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.032333923998269,
      "learning_rate": 1.9133595126280848e-05,
      "loss": 0.8658,
      "step": 829
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9352001257373778,
      "learning_rate": 1.9131056439343095e-05,
      "loss": 0.892,
      "step": 830
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9116018914660895,
      "learning_rate": 1.9128514207414838e-05,
      "loss": 0.8464,
      "step": 831
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.03458760079023,
      "learning_rate": 1.9125968431483068e-05,
      "loss": 0.8977,
      "step": 832
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0237835797629515,
      "learning_rate": 1.9123419112536132e-05,
      "loss": 0.858,
      "step": 833
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9723167924618649,
      "learning_rate": 1.912086625156377e-05,
      "loss": 0.9539,
      "step": 834
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.076965610161451,
      "learning_rate": 1.911830984955709e-05,
      "loss": 0.9416,
      "step": 835
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0138424862305402,
      "learning_rate": 1.911574990750857e-05,
      "loss": 0.856,
      "step": 836
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9858536783072386,
      "learning_rate": 1.9113186426412073e-05,
      "loss": 0.8403,
      "step": 837
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0225718211754873,
      "learning_rate": 1.9110619407262828e-05,
      "loss": 0.8865,
      "step": 838
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.023521869627053,
      "learning_rate": 1.9108048851057447e-05,
      "loss": 0.9002,
      "step": 839
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9769953426511312,
      "learning_rate": 1.9105474758793897e-05,
      "loss": 0.9055,
      "step": 840
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9753747779321191,
      "learning_rate": 1.9102897131471536e-05,
      "loss": 0.8973,
      "step": 841
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.033691936131496,
      "learning_rate": 1.9100315970091088e-05,
      "loss": 0.8673,
      "step": 842
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9561806897477386,
      "learning_rate": 1.9097731275654645e-05,
      "loss": 0.8961,
      "step": 843
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0468614747735896,
      "learning_rate": 1.909514304916568e-05,
      "loss": 0.8966,
      "step": 844
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9821747538715515,
      "learning_rate": 1.9092551291629026e-05,
      "loss": 0.9024,
      "step": 845
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.048918061334102,
      "learning_rate": 1.9089956004050893e-05,
      "loss": 0.8922,
      "step": 846
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9455142741779373,
      "learning_rate": 1.908735718743887e-05,
      "loss": 0.8906,
      "step": 847
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8827859434895523,
      "learning_rate": 1.908475484280189e-05,
      "loss": 0.8921,
      "step": 848
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9932513837442609,
      "learning_rate": 1.908214897115029e-05,
      "loss": 0.8583,
      "step": 849
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9873873241424637,
      "learning_rate": 1.907953957349575e-05,
      "loss": 0.9383,
      "step": 850
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9568113076726463,
      "learning_rate": 1.907692665085133e-05,
      "loss": 0.8235,
      "step": 851
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0019621049281515,
      "learning_rate": 1.9074310204231457e-05,
      "loss": 0.8696,
      "step": 852
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9495532176043411,
      "learning_rate": 1.9071690234651923e-05,
      "loss": 0.8477,
      "step": 853
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.898832589203252,
      "learning_rate": 1.9069066743129893e-05,
      "loss": 0.9642,
      "step": 854
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0670787918406623,
      "learning_rate": 1.90664397306839e-05,
      "loss": 0.9002,
      "step": 855
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9966316373170804,
      "learning_rate": 1.9063809198333832e-05,
      "loss": 0.8769,
      "step": 856
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9921297899963173,
      "learning_rate": 1.9061175147100957e-05,
      "loss": 0.8519,
      "step": 857
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7193360595640876,
      "learning_rate": 1.905853757800791e-05,
      "loss": 0.3627,
      "step": 858
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0797688138728536,
      "learning_rate": 1.9055896492078675e-05,
      "loss": 0.9193,
      "step": 859
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9921579424611826,
      "learning_rate": 1.905325189033862e-05,
      "loss": 0.8809,
      "step": 860
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9372583605620389,
      "learning_rate": 1.905060377381447e-05,
      "loss": 0.9029,
      "step": 861
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.984768014885644,
      "learning_rate": 1.904795214353431e-05,
      "loss": 0.9135,
      "step": 862
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0404221438139252,
      "learning_rate": 1.90452970005276e-05,
      "loss": 0.914,
      "step": 863
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9981908802805378,
      "learning_rate": 1.9042638345825155e-05,
      "loss": 0.9093,
      "step": 864
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.024601535079956,
      "learning_rate": 1.9039976180459158e-05,
      "loss": 0.9626,
      "step": 865
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.047652710091501,
      "learning_rate": 1.9037310505463153e-05,
      "loss": 0.8882,
      "step": 866
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9211333873683571,
      "learning_rate": 1.9034641321872043e-05,
      "loss": 0.9203,
      "step": 867
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8859036586536466,
      "learning_rate": 1.9031968630722104e-05,
      "loss": 0.8455,
      "step": 868
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0150197334643887,
      "learning_rate": 1.902929243305096e-05,
      "loss": 0.8724,
      "step": 869
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.925380319432884,
      "learning_rate": 1.902661272989761e-05,
      "loss": 0.8028,
      "step": 870
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0377507321256039,
      "learning_rate": 1.9023929522302394e-05,
      "loss": 0.8973,
      "step": 871
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9932203029595837,
      "learning_rate": 1.9021242811307044e-05,
      "loss": 0.9157,
      "step": 872
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0313632542335749,
      "learning_rate": 1.901855259795462e-05,
      "loss": 0.8644,
      "step": 873
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.97947378393248,
      "learning_rate": 1.9015858883289556e-05,
      "loss": 0.8481,
      "step": 874
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.395748202566758,
      "learning_rate": 1.9013161668357655e-05,
      "loss": 0.8702,
      "step": 875
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9986298286541507,
      "learning_rate": 1.901046095420606e-05,
      "loss": 0.9285,
      "step": 876
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0431229621573457,
      "learning_rate": 1.9007756741883284e-05,
      "loss": 0.8191,
      "step": 877
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9756496559466518,
      "learning_rate": 1.9005049032439193e-05,
      "loss": 0.8848,
      "step": 878
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0395633648370124,
      "learning_rate": 1.9002337826925012e-05,
      "loss": 0.9088,
      "step": 879
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.891110250912047,
      "learning_rate": 1.899962312639333e-05,
      "loss": 0.342,
      "step": 880
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9357576580755308,
      "learning_rate": 1.8996904931898085e-05,
      "loss": 0.857,
      "step": 881
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9296175727321284,
      "learning_rate": 1.899418324449457e-05,
      "loss": 0.9144,
      "step": 882
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.041693034148782,
      "learning_rate": 1.8991458065239444e-05,
      "loss": 0.9522,
      "step": 883
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9474384889968834,
      "learning_rate": 1.8988729395190712e-05,
      "loss": 0.8459,
      "step": 884
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0135662247978106,
      "learning_rate": 1.8985997235407735e-05,
      "loss": 0.8908,
      "step": 885
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0916523340850075,
      "learning_rate": 1.898326158695124e-05,
      "loss": 0.8863,
      "step": 886
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0196514988191883,
      "learning_rate": 1.8980522450883287e-05,
      "loss": 0.8576,
      "step": 887
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.965209550858663,
      "learning_rate": 1.8977779828267314e-05,
      "loss": 0.9198,
      "step": 888
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9086801352972873,
      "learning_rate": 1.8975033720168094e-05,
      "loss": 0.8931,
      "step": 889
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.011780092711424,
      "learning_rate": 1.897228412765177e-05,
      "loss": 0.8658,
      "step": 890
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7612686988744196,
      "learning_rate": 1.896953105178582e-05,
      "loss": 0.3643,
      "step": 891
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.972750635718914,
      "learning_rate": 1.8966774493639084e-05,
      "loss": 0.8982,
      "step": 892
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8157833523080248,
      "learning_rate": 1.896401445428176e-05,
      "loss": 0.3419,
      "step": 893
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9914304814297066,
      "learning_rate": 1.896125093478538e-05,
      "loss": 0.9163,
      "step": 894
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6528737185755795,
      "learning_rate": 1.895848393622284e-05,
      "loss": 0.3372,
      "step": 895
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.914089048898719,
      "learning_rate": 1.895571345966839e-05,
      "loss": 0.8719,
      "step": 896
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0765758935561653,
      "learning_rate": 1.8952939506197622e-05,
      "loss": 0.9338,
      "step": 897
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9136568119539024,
      "learning_rate": 1.8950162076887477e-05,
      "loss": 0.8717,
      "step": 898
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9923626867553543,
      "learning_rate": 1.894738117281625e-05,
      "loss": 0.8645,
      "step": 899
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0245771269208253,
      "learning_rate": 1.8944596795063584e-05,
      "loss": 0.8907,
      "step": 900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.963750953004356,
      "learning_rate": 1.894180894471047e-05,
      "loss": 0.8826,
      "step": 901
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9674730048695991,
      "learning_rate": 1.8939017622839253e-05,
      "loss": 0.8069,
      "step": 902
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2055700641204234,
      "learning_rate": 1.8936222830533613e-05,
      "loss": 0.3373,
      "step": 903
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9699817393280212,
      "learning_rate": 1.8933424568878586e-05,
      "loss": 0.8333,
      "step": 904
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.049212390240577,
      "learning_rate": 1.8930622838960555e-05,
      "loss": 0.9212,
      "step": 905
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9927497631909207,
      "learning_rate": 1.8927817641867244e-05,
      "loss": 0.9144,
      "step": 906
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9611317546786723,
      "learning_rate": 1.8925008978687737e-05,
      "loss": 0.8544,
      "step": 907
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.038014364152612,
      "learning_rate": 1.8922196850512446e-05,
      "loss": 0.8952,
      "step": 908
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9631100025305463,
      "learning_rate": 1.8919381258433135e-05,
      "loss": 0.8699,
      "step": 909
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9752113309931695,
      "learning_rate": 1.8916562203542916e-05,
      "loss": 0.8877,
      "step": 910
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9389505568759904,
      "learning_rate": 1.8913739686936244e-05,
      "loss": 0.8325,
      "step": 911
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0609974376067517,
      "learning_rate": 1.8910913709708918e-05,
      "loss": 0.781,
      "step": 912
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9703410121299941,
      "learning_rate": 1.8908084272958077e-05,
      "loss": 0.8619,
      "step": 913
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9727082545001341,
      "learning_rate": 1.8905251377782206e-05,
      "loss": 0.8268,
      "step": 914
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7963912700321882,
      "learning_rate": 1.8902415025281136e-05,
      "loss": 0.3666,
      "step": 915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0348992923987697,
      "learning_rate": 1.889957521655603e-05,
      "loss": 0.9322,
      "step": 916
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0829512876554188,
      "learning_rate": 1.8896731952709408e-05,
      "loss": 0.9581,
      "step": 917
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0743489760878264,
      "learning_rate": 1.8893885234845117e-05,
      "loss": 0.9002,
      "step": 918
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0276447243266738,
      "learning_rate": 1.8891035064068354e-05,
      "loss": 0.8688,
      "step": 919
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9432843495136447,
      "learning_rate": 1.888818144148565e-05,
      "loss": 0.8777,
      "step": 920
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0159335553751017,
      "learning_rate": 1.888532436820488e-05,
      "loss": 0.8659,
      "step": 921
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9418331029061882,
      "learning_rate": 1.8882463845335263e-05,
      "loss": 0.8777,
      "step": 922
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9361412887867991,
      "learning_rate": 1.8879599873987343e-05,
      "loss": 0.8556,
      "step": 923
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0592380626826587,
      "learning_rate": 1.8876732455273022e-05,
      "loss": 0.8719,
      "step": 924
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9985481790658722,
      "learning_rate": 1.8873861590305527e-05,
      "loss": 0.89,
      "step": 925
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9176244362647558,
      "learning_rate": 1.8870987280199428e-05,
      "loss": 0.8649,
      "step": 926
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0577156150099387,
      "learning_rate": 1.886810952607063e-05,
      "loss": 0.842,
      "step": 927
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8533166311323892,
      "learning_rate": 1.8865228329036372e-05,
      "loss": 0.8584,
      "step": 928
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0006984980413158,
      "learning_rate": 1.886234369021524e-05,
      "loss": 0.8668,
      "step": 929
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6883521611863723,
      "learning_rate": 1.885945561072715e-05,
      "loss": 0.3476,
      "step": 930
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9306800596737632,
      "learning_rate": 1.885656409169335e-05,
      "loss": 0.9009,
      "step": 931
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9069552735826024,
      "learning_rate": 1.885366913423643e-05,
      "loss": 0.9076,
      "step": 932
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9415360275718712,
      "learning_rate": 1.8850770739480312e-05,
      "loss": 0.8456,
      "step": 933
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9035525581434465,
      "learning_rate": 1.8847868908550252e-05,
      "loss": 0.869,
      "step": 934
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9639181345371466,
      "learning_rate": 1.8844963642572837e-05,
      "loss": 0.9271,
      "step": 935
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0322382977250752,
      "learning_rate": 1.8842054942676e-05,
      "loss": 0.8953,
      "step": 936
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9378010568177635,
      "learning_rate": 1.8839142809988987e-05,
      "loss": 0.8159,
      "step": 937
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9809469176929816,
      "learning_rate": 1.88362272456424e-05,
      "loss": 0.922,
      "step": 938
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0011544241707129,
      "learning_rate": 1.8833308250768153e-05,
      "loss": 0.8882,
      "step": 939
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6867872644539501,
      "learning_rate": 1.8830385826499507e-05,
      "loss": 0.3261,
      "step": 940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9381440353302899,
      "learning_rate": 1.882745997397104e-05,
      "loss": 0.8449,
      "step": 941
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0281234024674604,
      "learning_rate": 1.8824530694318675e-05,
      "loss": 0.8174,
      "step": 942
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.007138371716712,
      "learning_rate": 1.882159798867966e-05,
      "loss": 0.8655,
      "step": 943
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9423952675030928,
      "learning_rate": 1.8818661858192562e-05,
      "loss": 0.8915,
      "step": 944
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9566356158651095,
      "learning_rate": 1.88157223039973e-05,
      "loss": 0.8459,
      "step": 945
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0130125358680158,
      "learning_rate": 1.8812779327235106e-05,
      "loss": 0.8649,
      "step": 946
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.986596697179356,
      "learning_rate": 1.880983292904854e-05,
      "loss": 0.8532,
      "step": 947
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.957536864768185,
      "learning_rate": 1.88068831105815e-05,
      "loss": 0.9365,
      "step": 948
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0432625667681412,
      "learning_rate": 1.8803929872979214e-05,
      "loss": 0.872,
      "step": 949
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9279290572236365,
      "learning_rate": 1.8800973217388215e-05,
      "loss": 0.8435,
      "step": 950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9467229911150754,
      "learning_rate": 1.879801314495639e-05,
      "loss": 0.8942,
      "step": 951
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9120011154362695,
      "learning_rate": 1.879504965683294e-05,
      "loss": 0.8975,
      "step": 952
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0141190077203315,
      "learning_rate": 1.8792082754168385e-05,
      "loss": 0.9561,
      "step": 953
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9975242559822155,
      "learning_rate": 1.878911243811459e-05,
      "loss": 0.8917,
      "step": 954
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9718712807109631,
      "learning_rate": 1.8786138709824726e-05,
      "loss": 0.8368,
      "step": 955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9860159522764359,
      "learning_rate": 1.8783161570453295e-05,
      "loss": 0.8109,
      "step": 956
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6426693929770062,
      "learning_rate": 1.878018102115614e-05,
      "loss": 0.3574,
      "step": 957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9895510507220153,
      "learning_rate": 1.8777197063090394e-05,
      "loss": 0.8799,
      "step": 958
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9772405158972344,
      "learning_rate": 1.877420969741454e-05,
      "loss": 0.862,
      "step": 959
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0122933204911715,
      "learning_rate": 1.877121892528838e-05,
      "loss": 0.8822,
      "step": 960
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.051036313517453,
      "learning_rate": 1.876822474787303e-05,
      "loss": 0.8664,
      "step": 961
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9553971684077619,
      "learning_rate": 1.8765227166330933e-05,
      "loss": 0.9328,
      "step": 962
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0293712887656543,
      "learning_rate": 1.8762226181825857e-05,
      "loss": 0.9339,
      "step": 963
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9758930962585538,
      "learning_rate": 1.875922179552288e-05,
      "loss": 0.9093,
      "step": 964
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9480858875567534,
      "learning_rate": 1.875621400858842e-05,
      "loss": 0.8326,
      "step": 965
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.950840366081411,
      "learning_rate": 1.875320282219019e-05,
      "loss": 0.9242,
      "step": 966
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9200853920744584,
      "learning_rate": 1.8750188237497247e-05,
      "loss": 0.8779,
      "step": 967
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9902728299474665,
      "learning_rate": 1.874717025567995e-05,
      "loss": 0.9065,
      "step": 968
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.018771536859431,
      "learning_rate": 1.874414887790999e-05,
      "loss": 0.9183,
      "step": 969
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9882425936293185,
      "learning_rate": 1.8741124105360363e-05,
      "loss": 0.9145,
      "step": 970
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9629821273230678,
      "learning_rate": 1.873809593920539e-05,
      "loss": 0.9128,
      "step": 971
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.025696601402018,
      "learning_rate": 1.8735064380620717e-05,
      "loss": 0.8595,
      "step": 972
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9955153024544002,
      "learning_rate": 1.873202943078329e-05,
      "loss": 0.8977,
      "step": 973
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8819433893490389,
      "learning_rate": 1.8728991090871387e-05,
      "loss": 0.8376,
      "step": 974
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.970559658370758,
      "learning_rate": 1.8725949362064596e-05,
      "loss": 0.9021,
      "step": 975
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9787827503572325,
      "learning_rate": 1.8722904245543817e-05,
      "loss": 0.9021,
      "step": 976
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.040810717277909,
      "learning_rate": 1.871985574249127e-05,
      "loss": 0.8225,
      "step": 977
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0311008168455402,
      "learning_rate": 1.8716803854090495e-05,
      "loss": 0.8583,
      "step": 978
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9823018158514842,
      "learning_rate": 1.8713748581526334e-05,
      "loss": 0.9182,
      "step": 979
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9810514410575801,
      "learning_rate": 1.871068992598495e-05,
      "loss": 0.8954,
      "step": 980
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9481384211237358,
      "learning_rate": 1.8707627888653816e-05,
      "loss": 0.8883,
      "step": 981
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9412155833805487,
      "learning_rate": 1.8704562470721728e-05,
      "loss": 0.8322,
      "step": 982
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.892596455652922,
      "learning_rate": 1.870149367337878e-05,
      "loss": 0.8548,
      "step": 983
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6965814958792538,
      "learning_rate": 1.8698421497816386e-05,
      "loss": 0.3665,
      "step": 984
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0844806158159288,
      "learning_rate": 1.869534594522727e-05,
      "loss": 0.85,
      "step": 985
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.507521383812343,
      "learning_rate": 1.8692267016805473e-05,
      "loss": 0.8687,
      "step": 986
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9400865333779943,
      "learning_rate": 1.8689184713746333e-05,
      "loss": 0.8439,
      "step": 987
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9101707082324502,
      "learning_rate": 1.868609903724651e-05,
      "loss": 0.7855,
      "step": 988
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9447850579024419,
      "learning_rate": 1.8683009988503972e-05,
      "loss": 0.8591,
      "step": 989
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6186311743771725,
      "learning_rate": 1.867991756871799e-05,
      "loss": 0.3669,
      "step": 990
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.925890949747509,
      "learning_rate": 1.867682177908915e-05,
      "loss": 0.8426,
      "step": 991
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0241286568404682,
      "learning_rate": 1.867372262081934e-05,
      "loss": 0.9099,
      "step": 992
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9443052396844083,
      "learning_rate": 1.8670620095111766e-05,
      "loss": 0.9002,
      "step": 993
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.018397775186854,
      "learning_rate": 1.8667514203170934e-05,
      "loss": 0.9124,
      "step": 994
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9506710998975957,
      "learning_rate": 1.8664404946202658e-05,
      "loss": 0.9449,
      "step": 995
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.009444761762686,
      "learning_rate": 1.8661292325414058e-05,
      "loss": 0.8455,
      "step": 996
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9785603620815688,
      "learning_rate": 1.865817634201356e-05,
      "loss": 0.8708,
      "step": 997
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9395763401670297,
      "learning_rate": 1.8655056997210893e-05,
      "loss": 0.8581,
      "step": 998
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9841306239728521,
      "learning_rate": 1.8651934292217097e-05,
      "loss": 0.8926,
      "step": 999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0248004797737689,
      "learning_rate": 1.864880822824452e-05,
      "loss": 0.8841,
      "step": 1000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.921197441705695,
      "learning_rate": 1.8645678806506795e-05,
      "loss": 0.8648,
      "step": 1001
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.987982297620746,
      "learning_rate": 1.864254602821888e-05,
      "loss": 0.8602,
      "step": 1002
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8803659410902438,
      "learning_rate": 1.8639409894597026e-05,
      "loss": 0.8552,
      "step": 1003
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9389140774457249,
      "learning_rate": 1.8636270406858786e-05,
      "loss": 0.8935,
      "step": 1004
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9078382682097731,
      "learning_rate": 1.8633127566223023e-05,
      "loss": 0.8833,
      "step": 1005
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9574426099011674,
      "learning_rate": 1.862998137390989e-05,
      "loss": 0.8458,
      "step": 1006
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8651518567090668,
      "learning_rate": 1.8626831831140845e-05,
      "loss": 0.8288,
      "step": 1007
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0263000549353067,
      "learning_rate": 1.8623678939138652e-05,
      "loss": 0.8878,
      "step": 1008
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9934558310895679,
      "learning_rate": 1.8620522699127374e-05,
      "loss": 0.8801,
      "step": 1009
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8879790773169591,
      "learning_rate": 1.8617363112332376e-05,
      "loss": 0.8688,
      "step": 1010
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0481791849564477,
      "learning_rate": 1.8614200179980307e-05,
      "loss": 0.875,
      "step": 1011
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9567507494553726,
      "learning_rate": 1.8611033903299136e-05,
      "loss": 0.8505,
      "step": 1012
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9166888120082278,
      "learning_rate": 1.8607864283518116e-05,
      "loss": 0.8853,
      "step": 1013
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9350919093827829,
      "learning_rate": 1.8604691321867804e-05,
      "loss": 0.8286,
      "step": 1014
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9094125849393647,
      "learning_rate": 1.8601515019580053e-05,
      "loss": 0.8743,
      "step": 1015
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0303784377933929,
      "learning_rate": 1.8598335377888012e-05,
      "loss": 0.9139,
      "step": 1016
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0374719200244558,
      "learning_rate": 1.8595152398026128e-05,
      "loss": 0.8633,
      "step": 1017
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6891452761213006,
      "learning_rate": 1.8591966081230142e-05,
      "loss": 0.3157,
      "step": 1018
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6706000796699109,
      "learning_rate": 1.8588776428737095e-05,
      "loss": 0.3348,
      "step": 1019
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9466775297577518,
      "learning_rate": 1.858558344178532e-05,
      "loss": 0.8718,
      "step": 1020
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0045118450006285,
      "learning_rate": 1.8582387121614437e-05,
      "loss": 0.8932,
      "step": 1021
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0134668814786145,
      "learning_rate": 1.857918746946538e-05,
      "loss": 0.8914,
      "step": 1022
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0637067541512826,
      "learning_rate": 1.8575984486580353e-05,
      "loss": 0.9725,
      "step": 1023
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0535398977296055,
      "learning_rate": 1.857277817420287e-05,
      "loss": 0.9108,
      "step": 1024
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9633838596158648,
      "learning_rate": 1.8569568533577727e-05,
      "loss": 0.8866,
      "step": 1025
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9979310180750609,
      "learning_rate": 1.8566355565951023e-05,
      "loss": 0.9519,
      "step": 1026
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0378266142508819,
      "learning_rate": 1.8563139272570142e-05,
      "loss": 0.8863,
      "step": 1027
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9764086926454589,
      "learning_rate": 1.8559919654683756e-05,
      "loss": 0.8972,
      "step": 1028
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.027804309761088,
      "learning_rate": 1.8556696713541833e-05,
      "loss": 0.9023,
      "step": 1029
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9456437644740294,
      "learning_rate": 1.855347045039563e-05,
      "loss": 0.8824,
      "step": 1030
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8991485593683348,
      "learning_rate": 1.8550240866497697e-05,
      "loss": 0.8666,
      "step": 1031
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.929226344387187,
      "learning_rate": 1.854700796310186e-05,
      "loss": 0.868,
      "step": 1032
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9833824311254056,
      "learning_rate": 1.8543771741463254e-05,
      "loss": 0.866,
      "step": 1033
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0432081950015273,
      "learning_rate": 1.8540532202838286e-05,
      "loss": 0.8951,
      "step": 1034
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9098216031770685,
      "learning_rate": 1.8537289348484658e-05,
      "loss": 0.8652,
      "step": 1035
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0109717684233752,
      "learning_rate": 1.8534043179661357e-05,
      "loss": 0.8133,
      "step": 1036
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.256135105295619,
      "learning_rate": 1.8530793697628658e-05,
      "loss": 0.3483,
      "step": 1037
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.025185892579088,
      "learning_rate": 1.8527540903648122e-05,
      "loss": 0.9518,
      "step": 1038
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.044930749703983,
      "learning_rate": 1.8524284798982595e-05,
      "loss": 0.9211,
      "step": 1039
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0628710315437742,
      "learning_rate": 1.852102538489621e-05,
      "loss": 0.8805,
      "step": 1040
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9549803710615234,
      "learning_rate": 1.8517762662654383e-05,
      "loss": 0.874,
      "step": 1041
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8245717415360597,
      "learning_rate": 1.851449663352381e-05,
      "loss": 0.3561,
      "step": 1042
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9846657529845916,
      "learning_rate": 1.851122729877249e-05,
      "loss": 0.8219,
      "step": 1043
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.944128758491443,
      "learning_rate": 1.8507954659669677e-05,
      "loss": 0.8172,
      "step": 1044
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0200692598849197,
      "learning_rate": 1.850467871748593e-05,
      "loss": 0.8727,
      "step": 1045
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9746374516083738,
      "learning_rate": 1.850139947349308e-05,
      "loss": 0.8502,
      "step": 1046
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9318365449849446,
      "learning_rate": 1.8498116928964244e-05,
      "loss": 0.9081,
      "step": 1047
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9776210224174555,
      "learning_rate": 1.849483108517381e-05,
      "loss": 0.8656,
      "step": 1048
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8750147550852347,
      "learning_rate": 1.849154194339747e-05,
      "loss": 0.8681,
      "step": 1049
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0160271203862166,
      "learning_rate": 1.8488249504912173e-05,
      "loss": 0.8887,
      "step": 1050
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8964603698493916,
      "learning_rate": 1.8484953770996163e-05,
      "loss": 0.857,
      "step": 1051
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9130756219495387,
      "learning_rate": 1.848165474292895e-05,
      "loss": 0.8005,
      "step": 1052
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0378161612049885,
      "learning_rate": 1.8478352421991334e-05,
      "loss": 0.8772,
      "step": 1053
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9947950378211264,
      "learning_rate": 1.847504680946539e-05,
      "loss": 0.9601,
      "step": 1054
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9952073447123031,
      "learning_rate": 1.847173790663447e-05,
      "loss": 0.8579,
      "step": 1055
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.932748465011439,
      "learning_rate": 1.8468425714783206e-05,
      "loss": 0.8723,
      "step": 1056
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8855449083284954,
      "learning_rate": 1.84651102351975e-05,
      "loss": 0.9086,
      "step": 1057
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9549077292523339,
      "learning_rate": 1.846179146916454e-05,
      "loss": 0.877,
      "step": 1058
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9455600160301628,
      "learning_rate": 1.8458469417972783e-05,
      "loss": 0.8636,
      "step": 1059
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.773897949855512,
      "learning_rate": 1.8455144082911965e-05,
      "loss": 0.3399,
      "step": 1060
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9011357489212315,
      "learning_rate": 1.8451815465273097e-05,
      "loss": 0.8541,
      "step": 1061
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9611347980145428,
      "learning_rate": 1.8448483566348456e-05,
      "loss": 0.8792,
      "step": 1062
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.931657231930603,
      "learning_rate": 1.8445148387431605e-05,
      "loss": 0.8607,
      "step": 1063
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9865603716463012,
      "learning_rate": 1.8441809929817382e-05,
      "loss": 0.8815,
      "step": 1064
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9092895633997763,
      "learning_rate": 1.8438468194801876e-05,
      "loss": 0.9119,
      "step": 1065
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9443420192498997,
      "learning_rate": 1.8435123183682475e-05,
      "loss": 0.8476,
      "step": 1066
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0307132265908217,
      "learning_rate": 1.8431774897757824e-05,
      "loss": 0.8116,
      "step": 1067
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9324142784557121,
      "learning_rate": 1.8428423338327847e-05,
      "loss": 0.8613,
      "step": 1068
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9812874453531177,
      "learning_rate": 1.8425068506693727e-05,
      "loss": 0.9482,
      "step": 1069
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9441153188221174,
      "learning_rate": 1.842171040415793e-05,
      "loss": 0.8316,
      "step": 1070
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9538151746561712,
      "learning_rate": 1.8418349032024185e-05,
      "loss": 0.8643,
      "step": 1071
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9547495099903158,
      "learning_rate": 1.8414984391597492e-05,
      "loss": 0.93,
      "step": 1072
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8964845710849756,
      "learning_rate": 1.8411616484184126e-05,
      "loss": 0.8623,
      "step": 1073
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9463050627364531,
      "learning_rate": 1.8408245311091618e-05,
      "loss": 0.8495,
      "step": 1074
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9172786222377669,
      "learning_rate": 1.8404870873628774e-05,
      "loss": 0.8248,
      "step": 1075
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9657387093899251,
      "learning_rate": 1.8401493173105675e-05,
      "loss": 0.9075,
      "step": 1076
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8941443712365168,
      "learning_rate": 1.8398112210833648e-05,
      "loss": 0.8427,
      "step": 1077
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9263543673029749,
      "learning_rate": 1.8394727988125308e-05,
      "loss": 0.861,
      "step": 1078
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9069673653760911,
      "learning_rate": 1.8391340506294524e-05,
      "loss": 0.863,
      "step": 1079
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9179448205791418,
      "learning_rate": 1.8387949766656434e-05,
      "loss": 0.8184,
      "step": 1080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9648171368830184,
      "learning_rate": 1.8384555770527438e-05,
      "loss": 0.8743,
      "step": 1081
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.970527495359082,
      "learning_rate": 1.8381158519225204e-05,
      "loss": 0.8742,
      "step": 1082
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9583960319132804,
      "learning_rate": 1.8377758014068662e-05,
      "loss": 0.874,
      "step": 1083
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9511770938802572,
      "learning_rate": 1.8374354256378e-05,
      "loss": 0.9028,
      "step": 1084
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7715438614618417,
      "learning_rate": 1.837094724747468e-05,
      "loss": 0.3687,
      "step": 1085
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9817302508487269,
      "learning_rate": 1.8367536988681422e-05,
      "loss": 0.8617,
      "step": 1086
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0388004857283948,
      "learning_rate": 1.83641234813222e-05,
      "loss": 0.9042,
      "step": 1087
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8975051101350897,
      "learning_rate": 1.8360706726722253e-05,
      "loss": 0.8682,
      "step": 1088
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0037973127743443,
      "learning_rate": 1.835728672620809e-05,
      "loss": 0.827,
      "step": 1089
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0529663439904464,
      "learning_rate": 1.8353863481107473e-05,
      "loss": 0.9057,
      "step": 1090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.972384884142141,
      "learning_rate": 1.835043699274942e-05,
      "loss": 0.8429,
      "step": 1091
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9630820728919557,
      "learning_rate": 1.8347007262464206e-05,
      "loss": 0.8679,
      "step": 1092
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0961171093169335,
      "learning_rate": 1.8343574291583385e-05,
      "loss": 0.8402,
      "step": 1093
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.999622675067265,
      "learning_rate": 1.8340138081439743e-05,
      "loss": 0.8182,
      "step": 1094
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9140985464717769,
      "learning_rate": 1.833669863336734e-05,
      "loss": 0.8702,
      "step": 1095
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9220282594680045,
      "learning_rate": 1.833325594870148e-05,
      "loss": 0.8487,
      "step": 1096
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9726417389135765,
      "learning_rate": 1.8329810028778747e-05,
      "loss": 0.9004,
      "step": 1097
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.988079989855183,
      "learning_rate": 1.8326360874936952e-05,
      "loss": 0.885,
      "step": 1098
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9878902789253315,
      "learning_rate": 1.8322908488515182e-05,
      "loss": 0.8859,
      "step": 1099
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0238190983291031,
      "learning_rate": 1.8319452870853772e-05,
      "loss": 0.9077,
      "step": 1100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.995478602338753,
      "learning_rate": 1.8315994023294306e-05,
      "loss": 0.9108,
      "step": 1101
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9260730068647788,
      "learning_rate": 1.8312531947179634e-05,
      "loss": 0.8789,
      "step": 1102
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9413032649770245,
      "learning_rate": 1.8309066643853854e-05,
      "loss": 0.9195,
      "step": 1103
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9351264449668469,
      "learning_rate": 1.8305598114662312e-05,
      "loss": 0.8859,
      "step": 1104
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9445978429459276,
      "learning_rate": 1.830212636095161e-05,
      "loss": 0.91,
      "step": 1105
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9815773490795457,
      "learning_rate": 1.8298651384069605e-05,
      "loss": 0.8493,
      "step": 1106
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7120926466863937,
      "learning_rate": 1.8295173185365405e-05,
      "loss": 0.3378,
      "step": 1107
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9958262344940756,
      "learning_rate": 1.829169176618936e-05,
      "loss": 0.8827,
      "step": 1108
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9633307580279853,
      "learning_rate": 1.828820712789308e-05,
      "loss": 0.8634,
      "step": 1109
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9310877346994232,
      "learning_rate": 1.828471927182942e-05,
      "loss": 0.8504,
      "step": 1110
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.032040914611013,
      "learning_rate": 1.828122819935249e-05,
      "loss": 0.8817,
      "step": 1111
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9607197856055588,
      "learning_rate": 1.8277733911817642e-05,
      "loss": 0.839,
      "step": 1112
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9600259854891728,
      "learning_rate": 1.8274236410581478e-05,
      "loss": 0.8146,
      "step": 1113
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.008002683480128,
      "learning_rate": 1.827073569700185e-05,
      "loss": 0.8984,
      "step": 1114
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1173645992891736,
      "learning_rate": 1.8267231772437854e-05,
      "loss": 0.8567,
      "step": 1115
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0791712528052337,
      "learning_rate": 1.8263724638249834e-05,
      "loss": 0.8942,
      "step": 1116
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9727036240063542,
      "learning_rate": 1.8260214295799382e-05,
      "loss": 0.836,
      "step": 1117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0021913393521464,
      "learning_rate": 1.825670074644933e-05,
      "loss": 0.8814,
      "step": 1118
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7059337370901708,
      "learning_rate": 1.8253183991563768e-05,
      "loss": 0.3443,
      "step": 1119
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9739875288391548,
      "learning_rate": 1.824966403250801e-05,
      "loss": 0.8738,
      "step": 1120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0390259111790447,
      "learning_rate": 1.8246140870648633e-05,
      "loss": 0.9055,
      "step": 1121
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6246854465528036,
      "learning_rate": 1.8242614507353446e-05,
      "loss": 0.362,
      "step": 1122
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9046935714744837,
      "learning_rate": 1.8239084943991507e-05,
      "loss": 0.8717,
      "step": 1123
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9268386392800794,
      "learning_rate": 1.823555218193311e-05,
      "loss": 0.8348,
      "step": 1124
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9952174678762654,
      "learning_rate": 1.8232016222549797e-05,
      "loss": 0.8856,
      "step": 1125
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6434939624257731,
      "learning_rate": 1.8228477067214352e-05,
      "loss": 0.3861,
      "step": 1126
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9604555843958786,
      "learning_rate": 1.8224934717300794e-05,
      "loss": 0.8605,
      "step": 1127
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9594104039281369,
      "learning_rate": 1.8221389174184385e-05,
      "loss": 0.8881,
      "step": 1128
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9486888382102782,
      "learning_rate": 1.8217840439241633e-05,
      "loss": 0.8025,
      "step": 1129
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8861995421023469,
      "learning_rate": 1.8214288513850267e-05,
      "loss": 0.9001,
      "step": 1130
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9812367745560965,
      "learning_rate": 1.8210733399389277e-05,
      "loss": 0.8302,
      "step": 1131
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.828008366302184,
      "learning_rate": 1.820717509723888e-05,
      "loss": 0.8184,
      "step": 1132
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9398867929371636,
      "learning_rate": 1.8203613608780525e-05,
      "loss": 0.8736,
      "step": 1133
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0088232063246516,
      "learning_rate": 1.8200048935396908e-05,
      "loss": 0.9112,
      "step": 1134
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0008540031952076,
      "learning_rate": 1.819648107847196e-05,
      "loss": 0.8183,
      "step": 1135
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9513806206058879,
      "learning_rate": 1.8192910039390844e-05,
      "loss": 0.8854,
      "step": 1136
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9813008012233102,
      "learning_rate": 1.8189335819539963e-05,
      "loss": 0.8973,
      "step": 1137
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8660513620408233,
      "learning_rate": 1.8185758420306947e-05,
      "loss": 0.8228,
      "step": 1138
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9886770167456332,
      "learning_rate": 1.818217784308067e-05,
      "loss": 0.9108,
      "step": 1139
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6335035746097271,
      "learning_rate": 1.817859408925123e-05,
      "loss": 0.3488,
      "step": 1140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0394448965070209,
      "learning_rate": 1.817500716020997e-05,
      "loss": 0.8541,
      "step": 1141
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9095591438034518,
      "learning_rate": 1.8171417057349457e-05,
      "loss": 0.8089,
      "step": 1142
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0208921125278818,
      "learning_rate": 1.816782378206349e-05,
      "loss": 0.8653,
      "step": 1143
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9451097394594975,
      "learning_rate": 1.8164227335747108e-05,
      "loss": 0.942,
      "step": 1144
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0529921626093994,
      "learning_rate": 1.8160627719796568e-05,
      "loss": 0.9094,
      "step": 1145
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9902060138339026,
      "learning_rate": 1.815702493560937e-05,
      "loss": 0.9179,
      "step": 1146
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8963491514701459,
      "learning_rate": 1.8153418984584238e-05,
      "loss": 0.89,
      "step": 1147
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9840247611143353,
      "learning_rate": 1.8149809868121125e-05,
      "loss": 0.8851,
      "step": 1148
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9493588993626518,
      "learning_rate": 1.8146197587621217e-05,
      "loss": 0.8475,
      "step": 1149
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9849343372342739,
      "learning_rate": 1.814258214448692e-05,
      "loss": 0.8963,
      "step": 1150
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8986079664584536,
      "learning_rate": 1.8138963540121878e-05,
      "loss": 0.8911,
      "step": 1151
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9921488263917064,
      "learning_rate": 1.813534177593096e-05,
      "loss": 0.8613,
      "step": 1152
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9551419562723557,
      "learning_rate": 1.8131716853320254e-05,
      "loss": 0.8872,
      "step": 1153
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9798801461763851,
      "learning_rate": 1.8128088773697086e-05,
      "loss": 0.8325,
      "step": 1154
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0175115085082844,
      "learning_rate": 1.8124457538469996e-05,
      "loss": 0.8803,
      "step": 1155
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9662103613866581,
      "learning_rate": 1.8120823149048753e-05,
      "loss": 0.8106,
      "step": 1156
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0696709219225051,
      "learning_rate": 1.811718560684436e-05,
      "loss": 0.8851,
      "step": 1157
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6667436605174416,
      "learning_rate": 1.8113544913269025e-05,
      "loss": 0.3576,
      "step": 1158
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8862416918230475,
      "learning_rate": 1.8109901069736202e-05,
      "loss": 0.8993,
      "step": 1159
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.001267709491352,
      "learning_rate": 1.8106254077660552e-05,
      "loss": 0.8642,
      "step": 1160
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0320405379681639,
      "learning_rate": 1.810260393845796e-05,
      "loss": 0.862,
      "step": 1161
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9543826305147441,
      "learning_rate": 1.809895065354554e-05,
      "loss": 0.8854,
      "step": 1162
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9420489935541324,
      "learning_rate": 1.8095294224341622e-05,
      "loss": 0.8569,
      "step": 1163
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0218185518443381,
      "learning_rate": 1.8091634652265755e-05,
      "loss": 0.8952,
      "step": 1164
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.947244204691924,
      "learning_rate": 1.8087971938738715e-05,
      "loss": 0.8393,
      "step": 1165
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8997578807668782,
      "learning_rate": 1.808430608518249e-05,
      "loss": 0.8476,
      "step": 1166
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9610682838012956,
      "learning_rate": 1.808063709302029e-05,
      "loss": 0.875,
      "step": 1167
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.935044898836728,
      "learning_rate": 1.807696496367655e-05,
      "loss": 0.872,
      "step": 1168
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8694885618299711,
      "learning_rate": 1.8073289698576913e-05,
      "loss": 0.8908,
      "step": 1169
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.988260587233124,
      "learning_rate": 1.8069611299148236e-05,
      "loss": 0.8542,
      "step": 1170
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9078681347223087,
      "learning_rate": 1.8065929766818617e-05,
      "loss": 0.8483,
      "step": 1171
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.738788385301926,
      "learning_rate": 1.806224510301734e-05,
      "loss": 0.3208,
      "step": 1172
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9738920367935925,
      "learning_rate": 1.8058557309174926e-05,
      "loss": 0.8777,
      "step": 1173
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9844476589283174,
      "learning_rate": 1.8054866386723096e-05,
      "loss": 0.8349,
      "step": 1174
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9877056714005591,
      "learning_rate": 1.80511723370948e-05,
      "loss": 0.8886,
      "step": 1175
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0258684357513237,
      "learning_rate": 1.804747516172419e-05,
      "loss": 0.9198,
      "step": 1176
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8848066737534694,
      "learning_rate": 1.8043774862046644e-05,
      "loss": 0.8378,
      "step": 1177
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9174470590880138,
      "learning_rate": 1.804007143949874e-05,
      "loss": 0.8845,
      "step": 1178
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9771917671312343,
      "learning_rate": 1.8036364895518272e-05,
      "loss": 0.7839,
      "step": 1179
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9758596647242445,
      "learning_rate": 1.8032655231544253e-05,
      "loss": 0.8232,
      "step": 1180
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0801985558837837,
      "learning_rate": 1.8028942449016903e-05,
      "loss": 0.8756,
      "step": 1181
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.013137207003489,
      "learning_rate": 1.8025226549377647e-05,
      "loss": 0.8358,
      "step": 1182
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9878720491332617,
      "learning_rate": 1.8021507534069133e-05,
      "loss": 0.8792,
      "step": 1183
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9331157024776532,
      "learning_rate": 1.8017785404535198e-05,
      "loss": 0.8093,
      "step": 1184
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0460408903574094,
      "learning_rate": 1.8014060162220916e-05,
      "loss": 0.8935,
      "step": 1185
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0259379767524426,
      "learning_rate": 1.801033180857254e-05,
      "loss": 0.8599,
      "step": 1186
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9055153919223407,
      "learning_rate": 1.8006600345037558e-05,
      "loss": 0.8933,
      "step": 1187
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9651622844869906,
      "learning_rate": 1.8002865773064644e-05,
      "loss": 0.8524,
      "step": 1188
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0140484623018196,
      "learning_rate": 1.799912809410369e-05,
      "loss": 0.8716,
      "step": 1189
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9557699219217266,
      "learning_rate": 1.799538730960579e-05,
      "loss": 0.8758,
      "step": 1190
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.985529949847647,
      "learning_rate": 1.799164342102325e-05,
      "loss": 0.8777,
      "step": 1191
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9064948908758821,
      "learning_rate": 1.7987896429809573e-05,
      "loss": 0.8644,
      "step": 1192
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7372307610871006,
      "learning_rate": 1.798414633741947e-05,
      "loss": 0.343,
      "step": 1193
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9567265117832868,
      "learning_rate": 1.7980393145308857e-05,
      "loss": 0.796,
      "step": 1194
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9030565773807138,
      "learning_rate": 1.797663685493485e-05,
      "loss": 0.9026,
      "step": 1195
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0446505657816685,
      "learning_rate": 1.7972877467755777e-05,
      "loss": 0.8909,
      "step": 1196
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8993710872362982,
      "learning_rate": 1.7969114985231152e-05,
      "loss": 0.8547,
      "step": 1197
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9515873134669062,
      "learning_rate": 1.796534940882171e-05,
      "loss": 0.9109,
      "step": 1198
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.871291997480676,
      "learning_rate": 1.7961580739989365e-05,
      "loss": 0.7782,
      "step": 1199
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8976170363370676,
      "learning_rate": 1.795780898019726e-05,
      "loss": 0.8871,
      "step": 1200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9072195891788813,
      "learning_rate": 1.795403413090971e-05,
      "loss": 0.9052,
      "step": 1201
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9054891878060444,
      "learning_rate": 1.7950256193592243e-05,
      "loss": 0.8109,
      "step": 1202
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9514963274216849,
      "learning_rate": 1.794647516971159e-05,
      "loss": 0.8557,
      "step": 1203
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9668962336067615,
      "learning_rate": 1.7942691060735666e-05,
      "loss": 0.8236,
      "step": 1204
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9885387544583619,
      "learning_rate": 1.79389038681336e-05,
      "loss": 0.8634,
      "step": 1205
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.059226832772945,
      "learning_rate": 1.7935113593375707e-05,
      "loss": 0.8071,
      "step": 1206
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9682109914939278,
      "learning_rate": 1.7931320237933503e-05,
      "loss": 0.8389,
      "step": 1207
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9188451479280468,
      "learning_rate": 1.79275238032797e-05,
      "loss": 0.9139,
      "step": 1208
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9447340210041916,
      "learning_rate": 1.7923724290888205e-05,
      "loss": 0.8465,
      "step": 1209
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9420378374763761,
      "learning_rate": 1.791992170223412e-05,
      "loss": 0.8665,
      "step": 1210
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9574100683730187,
      "learning_rate": 1.791611603879374e-05,
      "loss": 0.8516,
      "step": 1211
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9519437333895201,
      "learning_rate": 1.791230730204455e-05,
      "loss": 0.9004,
      "step": 1212
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9597082744191842,
      "learning_rate": 1.7908495493465236e-05,
      "loss": 0.8663,
      "step": 1213
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9144360709158345,
      "learning_rate": 1.7904680614535675e-05,
      "loss": 0.9388,
      "step": 1214
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9051835968185558,
      "learning_rate": 1.7900862666736935e-05,
      "loss": 0.8239,
      "step": 1215
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9554741408715728,
      "learning_rate": 1.789704165155127e-05,
      "loss": 0.8645,
      "step": 1216
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9544070518831257,
      "learning_rate": 1.7893217570462134e-05,
      "loss": 0.848,
      "step": 1217
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9598390346838902,
      "learning_rate": 1.7889390424954168e-05,
      "loss": 0.8451,
      "step": 1218
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8966705051386944,
      "learning_rate": 1.78855602165132e-05,
      "loss": 0.8591,
      "step": 1219
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0969621281036697,
      "learning_rate": 1.7881726946626244e-05,
      "loss": 0.8725,
      "step": 1220
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0519698824877544,
      "learning_rate": 1.787789061678151e-05,
      "loss": 0.8479,
      "step": 1221
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9167767965936483,
      "learning_rate": 1.78740512284684e-05,
      "loss": 0.8841,
      "step": 1222
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9556247046975325,
      "learning_rate": 1.787020878317749e-05,
      "loss": 0.8754,
      "step": 1223
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9050725303214799,
      "learning_rate": 1.7866363282400555e-05,
      "loss": 0.8605,
      "step": 1224
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9549019680093457,
      "learning_rate": 1.7862514727630543e-05,
      "loss": 0.8846,
      "step": 1225
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9613952022792592,
      "learning_rate": 1.7858663120361597e-05,
      "loss": 0.9212,
      "step": 1226
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8670916939342002,
      "learning_rate": 1.785480846208905e-05,
      "loss": 0.9184,
      "step": 1227
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.882004080628267,
      "learning_rate": 1.7850950754309405e-05,
      "loss": 0.8353,
      "step": 1228
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9108394168817177,
      "learning_rate": 1.7847089998520365e-05,
      "loss": 0.8456,
      "step": 1229
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9630143882068324,
      "learning_rate": 1.7843226196220803e-05,
      "loss": 0.9309,
      "step": 1230
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8540992506381067,
      "learning_rate": 1.783935934891078e-05,
      "loss": 0.8327,
      "step": 1231
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8991551497709254,
      "learning_rate": 1.7835489458091544e-05,
      "loss": 0.9029,
      "step": 1232
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9318875354488825,
      "learning_rate": 1.7831616525265515e-05,
      "loss": 0.8538,
      "step": 1233
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9298745026614752,
      "learning_rate": 1.7827740551936296e-05,
      "loss": 0.9261,
      "step": 1234
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8721371260131286,
      "learning_rate": 1.7823861539608686e-05,
      "loss": 0.8911,
      "step": 1235
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.903936388150881,
      "learning_rate": 1.7819979489788638e-05,
      "loss": 0.9328,
      "step": 1236
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9012213788709718,
      "learning_rate": 1.7816094403983298e-05,
      "loss": 0.8175,
      "step": 1237
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9279403124168664,
      "learning_rate": 1.7812206283701002e-05,
      "loss": 0.8875,
      "step": 1238
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9421012844556739,
      "learning_rate": 1.7808315130451244e-05,
      "loss": 0.867,
      "step": 1239
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.934552926342281,
      "learning_rate": 1.78044209457447e-05,
      "loss": 0.8193,
      "step": 1240
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9233630997963262,
      "learning_rate": 1.7800523731093232e-05,
      "loss": 0.9299,
      "step": 1241
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0518631278044193,
      "learning_rate": 1.7796623488009875e-05,
      "loss": 0.8626,
      "step": 1242
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9103317340988335,
      "learning_rate": 1.7792720218008826e-05,
      "loss": 0.8881,
      "step": 1243
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8890483345709617,
      "learning_rate": 1.7788813922605488e-05,
      "loss": 0.85,
      "step": 1244
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8987805522715748,
      "learning_rate": 1.7784904603316402e-05,
      "loss": 0.8458,
      "step": 1245
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9799230476296118,
      "learning_rate": 1.7780992261659305e-05,
      "loss": 0.8538,
      "step": 1246
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9594805226962042,
      "learning_rate": 1.777707689915311e-05,
      "loss": 0.8726,
      "step": 1247
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9596809764078332,
      "learning_rate": 1.777315851731789e-05,
      "loss": 0.8324,
      "step": 1248
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.866717826885279,
      "learning_rate": 1.7769237117674893e-05,
      "loss": 0.8549,
      "step": 1249
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9486683787408926,
      "learning_rate": 1.7765312701746543e-05,
      "loss": 0.8683,
      "step": 1250
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9822110469838201,
      "learning_rate": 1.7761385271056436e-05,
      "loss": 0.8896,
      "step": 1251
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9606701125854995,
      "learning_rate": 1.7757454827129338e-05,
      "loss": 0.8945,
      "step": 1252
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8225131746710052,
      "learning_rate": 1.7753521371491174e-05,
      "loss": 0.3449,
      "step": 1253
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9464143330837163,
      "learning_rate": 1.7749584905669057e-05,
      "loss": 0.7767,
      "step": 1254
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0465680684821153,
      "learning_rate": 1.774564543119125e-05,
      "loss": 0.8857,
      "step": 1255
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0360440973283846,
      "learning_rate": 1.7741702949587196e-05,
      "loss": 0.8908,
      "step": 1256
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9638703530026742,
      "learning_rate": 1.7737757462387507e-05,
      "loss": 0.913,
      "step": 1257
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9273253553119967,
      "learning_rate": 1.7733808971123946e-05,
      "loss": 0.9277,
      "step": 1258
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9789479616642742,
      "learning_rate": 1.7729857477329463e-05,
      "loss": 0.8516,
      "step": 1259
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9414576634142587,
      "learning_rate": 1.7725902982538162e-05,
      "loss": 0.8374,
      "step": 1260
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8837743523464174,
      "learning_rate": 1.772194548828531e-05,
      "loss": 0.8542,
      "step": 1261
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9532700117107941,
      "learning_rate": 1.7717984996107346e-05,
      "loss": 0.8362,
      "step": 1262
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7759506103627126,
      "learning_rate": 1.771402150754187e-05,
      "loss": 0.3611,
      "step": 1263
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9228472247854348,
      "learning_rate": 1.7710055024127637e-05,
      "loss": 0.8259,
      "step": 1264
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9719134456920342,
      "learning_rate": 1.7706085547404582e-05,
      "loss": 0.8598,
      "step": 1265
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8491138284645515,
      "learning_rate": 1.770211307891379e-05,
      "loss": 0.8838,
      "step": 1266
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.859571977278734,
      "learning_rate": 1.769813762019751e-05,
      "loss": 0.8228,
      "step": 1267
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9290652722281416,
      "learning_rate": 1.769415917279915e-05,
      "loss": 0.8144,
      "step": 1268
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9745814815382586,
      "learning_rate": 1.7690177738263284e-05,
      "loss": 0.8718,
      "step": 1269
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.987085793799981,
      "learning_rate": 1.7686193318135635e-05,
      "loss": 0.916,
      "step": 1270
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8786177495478376,
      "learning_rate": 1.76822059139631e-05,
      "loss": 0.8839,
      "step": 1271
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9122274070230761,
      "learning_rate": 1.7678215527293724e-05,
      "loss": 0.8596,
      "step": 1272
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9477912962032063,
      "learning_rate": 1.767422215967671e-05,
      "loss": 0.8977,
      "step": 1273
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9244032326315187,
      "learning_rate": 1.767022581266242e-05,
      "loss": 0.9161,
      "step": 1274
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9329725185260518,
      "learning_rate": 1.766622648780238e-05,
      "loss": 0.941,
      "step": 1275
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9958168360130557,
      "learning_rate": 1.766222418664926e-05,
      "loss": 0.8108,
      "step": 1276
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9393926420588763,
      "learning_rate": 1.765821891075689e-05,
      "loss": 0.8987,
      "step": 1277
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8939863323218828,
      "learning_rate": 1.7654210661680263e-05,
      "loss": 0.8615,
      "step": 1278
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8977270393393056,
      "learning_rate": 1.765019944097551e-05,
      "loss": 0.9122,
      "step": 1279
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6981225331360205,
      "learning_rate": 1.7646185250199936e-05,
      "loss": 0.3444,
      "step": 1280
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6905456085497725,
      "learning_rate": 1.7642168090911976e-05,
      "loss": 0.3214,
      "step": 1281
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.056397109211119,
      "learning_rate": 1.763814796467124e-05,
      "loss": 0.9113,
      "step": 1282
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9881979462887269,
      "learning_rate": 1.763412487303847e-05,
      "loss": 0.8594,
      "step": 1283
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0290882294530141,
      "learning_rate": 1.7630098817575578e-05,
      "loss": 0.9227,
      "step": 1284
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0533239955190492,
      "learning_rate": 1.762606979984561e-05,
      "loss": 0.8752,
      "step": 1285
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.027953727499225,
      "learning_rate": 1.7622037821412775e-05,
      "loss": 0.9032,
      "step": 1286
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.062873751476769,
      "learning_rate": 1.7618002883842426e-05,
      "loss": 0.841,
      "step": 1287
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9756847461383983,
      "learning_rate": 1.7613964988701057e-05,
      "loss": 0.8759,
      "step": 1288
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9044883759042869,
      "learning_rate": 1.7609924137556326e-05,
      "loss": 0.9208,
      "step": 1289
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0007517438885893,
      "learning_rate": 1.7605880331977022e-05,
      "loss": 0.8163,
      "step": 1290
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.904782237579859,
      "learning_rate": 1.76018335735331e-05,
      "loss": 0.9003,
      "step": 1291
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8975818892904778,
      "learning_rate": 1.7597783863795644e-05,
      "loss": 0.9142,
      "step": 1292
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0050149171352156,
      "learning_rate": 1.7593731204336895e-05,
      "loss": 0.8507,
      "step": 1293
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9088035349817715,
      "learning_rate": 1.7589675596730233e-05,
      "loss": 0.8978,
      "step": 1294
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8809764114517267,
      "learning_rate": 1.758561704255018e-05,
      "loss": 0.9105,
      "step": 1295
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8978223960019384,
      "learning_rate": 1.7581555543372413e-05,
      "loss": 0.8912,
      "step": 1296
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.936043912899347,
      "learning_rate": 1.7577491100773744e-05,
      "loss": 0.8609,
      "step": 1297
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8975710080113904,
      "learning_rate": 1.7573423716332128e-05,
      "loss": 0.8153,
      "step": 1298
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9587123928491116,
      "learning_rate": 1.7569353391626665e-05,
      "loss": 0.8437,
      "step": 1299
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9046520648903176,
      "learning_rate": 1.7565280128237595e-05,
      "loss": 0.8631,
      "step": 1300
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2080814509509372,
      "learning_rate": 1.75612039277463e-05,
      "loss": 0.3589,
      "step": 1301
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0397457188189705,
      "learning_rate": 1.75571247917353e-05,
      "loss": 0.8953,
      "step": 1302
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9275062980877427,
      "learning_rate": 1.7553042721788255e-05,
      "loss": 0.8508,
      "step": 1303
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6331590927420735,
      "learning_rate": 1.754895771948997e-05,
      "loss": 0.3749,
      "step": 1304
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9730020785826088,
      "learning_rate": 1.754486978642637e-05,
      "loss": 0.9076,
      "step": 1305
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9561547313711414,
      "learning_rate": 1.7540778924184553e-05,
      "loss": 0.8606,
      "step": 1306
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9710249453695874,
      "learning_rate": 1.7536685134352717e-05,
      "loss": 0.8797,
      "step": 1307
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9586924875433408,
      "learning_rate": 1.7532588418520215e-05,
      "loss": 0.9255,
      "step": 1308
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9900545593229638,
      "learning_rate": 1.7528488778277535e-05,
      "loss": 0.9215,
      "step": 1309
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9462828024538048,
      "learning_rate": 1.75243862152163e-05,
      "loss": 0.8238,
      "step": 1310
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9132222452521136,
      "learning_rate": 1.752028073092926e-05,
      "loss": 0.8062,
      "step": 1311
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0189215847582964,
      "learning_rate": 1.7516172327010314e-05,
      "loss": 0.8948,
      "step": 1312
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9410653950269818,
      "learning_rate": 1.751206100505448e-05,
      "loss": 0.9015,
      "step": 1313
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9885015541325021,
      "learning_rate": 1.7507946766657914e-05,
      "loss": 0.9272,
      "step": 1314
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9780583920695052,
      "learning_rate": 1.7503829613417905e-05,
      "loss": 0.8345,
      "step": 1315
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9418812028279031,
      "learning_rate": 1.749970954693288e-05,
      "loss": 0.833,
      "step": 1316
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9412177028638465,
      "learning_rate": 1.7495586568802384e-05,
      "loss": 0.8533,
      "step": 1317
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.031422458731205,
      "learning_rate": 1.7491460680627105e-05,
      "loss": 0.8788,
      "step": 1318
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0265392701151252,
      "learning_rate": 1.7487331884008845e-05,
      "loss": 0.8496,
      "step": 1319
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0080925754212984,
      "learning_rate": 1.7483200180550554e-05,
      "loss": 0.9374,
      "step": 1320
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9501206844315905,
      "learning_rate": 1.74790655718563e-05,
      "loss": 0.8704,
      "step": 1321
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0137607897458862,
      "learning_rate": 1.747492805953128e-05,
      "loss": 0.352,
      "step": 1322
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9495240205540482,
      "learning_rate": 1.7470787645181818e-05,
      "loss": 0.8356,
      "step": 1323
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.001600942798213,
      "learning_rate": 1.7466644330415362e-05,
      "loss": 0.8756,
      "step": 1324
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9419457957820736,
      "learning_rate": 1.7462498116840496e-05,
      "loss": 0.8469,
      "step": 1325
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0261991573826656,
      "learning_rate": 1.745834900606692e-05,
      "loss": 0.9254,
      "step": 1326
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0441371633937526,
      "learning_rate": 1.7454196999705458e-05,
      "loss": 0.8346,
      "step": 1327
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.969622340368838,
      "learning_rate": 1.7450042099368066e-05,
      "loss": 0.8996,
      "step": 1328
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9707939576292045,
      "learning_rate": 1.7445884306667823e-05,
      "loss": 0.8658,
      "step": 1329
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9587129004214763,
      "learning_rate": 1.7441723623218917e-05,
      "loss": 0.8354,
      "step": 1330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9157347662878492,
      "learning_rate": 1.7437560050636678e-05,
      "loss": 0.8426,
      "step": 1331
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9404707785271739,
      "learning_rate": 1.7433393590537543e-05,
      "loss": 0.8234,
      "step": 1332
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9522179353243918,
      "learning_rate": 1.7429224244539077e-05,
      "loss": 0.8706,
      "step": 1333
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9594263485418149,
      "learning_rate": 1.7425052014259965e-05,
      "loss": 0.8963,
      "step": 1334
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9780547411485991,
      "learning_rate": 1.7420876901320006e-05,
      "loss": 0.9099,
      "step": 1335
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0519264906338495,
      "learning_rate": 1.7416698907340128e-05,
      "loss": 0.8907,
      "step": 1336
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9645688350859514,
      "learning_rate": 1.741251803394237e-05,
      "loss": 0.9259,
      "step": 1337
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9921157860980591,
      "learning_rate": 1.740833428274989e-05,
      "loss": 0.8244,
      "step": 1338
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8682740920041911,
      "learning_rate": 1.7404147655386966e-05,
      "loss": 0.8689,
      "step": 1339
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.923085823929372,
      "learning_rate": 1.739995815347899e-05,
      "loss": 0.9048,
      "step": 1340
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0251361056477213,
      "learning_rate": 1.739576577865247e-05,
      "loss": 0.8931,
      "step": 1341
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9404453727368778,
      "learning_rate": 1.739157053253503e-05,
      "loss": 0.8605,
      "step": 1342
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0240155840310095,
      "learning_rate": 1.738737241675541e-05,
      "loss": 0.8611,
      "step": 1343
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9066952294332022,
      "learning_rate": 1.7383171432943466e-05,
      "loss": 0.8918,
      "step": 1344
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9313707398122598,
      "learning_rate": 1.737896758273016e-05,
      "loss": 0.8759,
      "step": 1345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9461924658131678,
      "learning_rate": 1.7374760867747574e-05,
      "loss": 0.9264,
      "step": 1346
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0391186672326336,
      "learning_rate": 1.7370551289628895e-05,
      "loss": 0.8368,
      "step": 1347
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9668995342755141,
      "learning_rate": 1.7366338850008432e-05,
      "loss": 0.8389,
      "step": 1348
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.994852173943171,
      "learning_rate": 1.73621235505216e-05,
      "loss": 0.8301,
      "step": 1349
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0420824582039552,
      "learning_rate": 1.7357905392804918e-05,
      "loss": 0.9432,
      "step": 1350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8737329970118888,
      "learning_rate": 1.735368437849602e-05,
      "loss": 0.8704,
      "step": 1351
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7070932789195868,
      "learning_rate": 1.7349460509233654e-05,
      "loss": 0.3515,
      "step": 1352
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.865837116361287,
      "learning_rate": 1.734523378665767e-05,
      "loss": 0.8763,
      "step": 1353
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9555972398413806,
      "learning_rate": 1.7341004212409026e-05,
      "loss": 0.9203,
      "step": 1354
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9514240583178237,
      "learning_rate": 1.7336771788129785e-05,
      "loss": 0.8233,
      "step": 1355
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9052324872183646,
      "learning_rate": 1.7332536515463126e-05,
      "loss": 0.9023,
      "step": 1356
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9602867523828692,
      "learning_rate": 1.7328298396053324e-05,
      "loss": 0.8595,
      "step": 1357
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0111257173279324,
      "learning_rate": 1.7324057431545768e-05,
      "loss": 0.9067,
      "step": 1358
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9867047681331311,
      "learning_rate": 1.7319813623586935e-05,
      "loss": 0.8804,
      "step": 1359
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9689090034911425,
      "learning_rate": 1.7315566973824433e-05,
      "loss": 0.9277,
      "step": 1360
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0068007396193648,
      "learning_rate": 1.7311317483906946e-05,
      "loss": 0.9246,
      "step": 1361
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8512082466293288,
      "learning_rate": 1.730706515548427e-05,
      "loss": 0.8938,
      "step": 1362
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9611260469747361,
      "learning_rate": 1.730280999020732e-05,
      "loss": 0.8754,
      "step": 1363
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9157457787921268,
      "learning_rate": 1.729855198972808e-05,
      "loss": 0.8702,
      "step": 1364
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9392385507882408,
      "learning_rate": 1.729429115569967e-05,
      "loss": 0.8592,
      "step": 1365
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8376330051043601,
      "learning_rate": 1.729002748977628e-05,
      "loss": 0.8496,
      "step": 1366
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9010413330988,
      "learning_rate": 1.7285760993613215e-05,
      "loss": 0.8798,
      "step": 1367
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9041735982700156,
      "learning_rate": 1.7281491668866874e-05,
      "loss": 0.8557,
      "step": 1368
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9438308927958647,
      "learning_rate": 1.727721951719476e-05,
      "loss": 0.8708,
      "step": 1369
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9690728690203271,
      "learning_rate": 1.7272944540255468e-05,
      "loss": 0.8814,
      "step": 1370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9758716977129109,
      "learning_rate": 1.726866673970869e-05,
      "loss": 0.9028,
      "step": 1371
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9031947173621965,
      "learning_rate": 1.7264386117215216e-05,
      "loss": 0.8883,
      "step": 1372
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9990898538608038,
      "learning_rate": 1.7260102674436933e-05,
      "loss": 0.8188,
      "step": 1373
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9751514223955624,
      "learning_rate": 1.7255816413036818e-05,
      "loss": 0.8347,
      "step": 1374
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6734448711319339,
      "learning_rate": 1.7251527334678946e-05,
      "loss": 0.3471,
      "step": 1375
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9559779875530791,
      "learning_rate": 1.7247235441028486e-05,
      "loss": 0.9194,
      "step": 1376
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.953365142511051,
      "learning_rate": 1.7242940733751696e-05,
      "loss": 0.8604,
      "step": 1377
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9539980743170202,
      "learning_rate": 1.7238643214515934e-05,
      "loss": 0.8897,
      "step": 1378
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9202678325006665,
      "learning_rate": 1.7234342884989642e-05,
      "loss": 0.8246,
      "step": 1379
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9570265042420878,
      "learning_rate": 1.7230039746842352e-05,
      "loss": 0.9008,
      "step": 1380
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0057917763161188,
      "learning_rate": 1.7225733801744698e-05,
      "loss": 0.8606,
      "step": 1381
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9796603438567899,
      "learning_rate": 1.7221425051368394e-05,
      "loss": 0.9031,
      "step": 1382
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0419395452696212,
      "learning_rate": 1.7217113497386245e-05,
      "loss": 0.856,
      "step": 1383
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9262802133964761,
      "learning_rate": 1.721279914147214e-05,
      "loss": 0.883,
      "step": 1384
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.94667110638933,
      "learning_rate": 1.7208481985301065e-05,
      "loss": 0.8875,
      "step": 1385
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8901513258551078,
      "learning_rate": 1.7204162030549093e-05,
      "loss": 0.8995,
      "step": 1386
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.820428451293344,
      "learning_rate": 1.7199839278893368e-05,
      "loss": 0.8401,
      "step": 1387
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0182094317992902,
      "learning_rate": 1.719551373201214e-05,
      "loss": 0.7928,
      "step": 1388
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9715124054545393,
      "learning_rate": 1.7191185391584736e-05,
      "loss": 0.8761,
      "step": 1389
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.954698221586228,
      "learning_rate": 1.7186854259291558e-05,
      "loss": 0.9345,
      "step": 1390
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9288012513813138,
      "learning_rate": 1.7182520336814105e-05,
      "loss": 0.8583,
      "step": 1391
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8692184858057738,
      "learning_rate": 1.717818362583496e-05,
      "loss": 0.838,
      "step": 1392
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0015162941109457,
      "learning_rate": 1.7173844128037777e-05,
      "loss": 0.8605,
      "step": 1393
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9505109210156145,
      "learning_rate": 1.71695018451073e-05,
      "loss": 0.8912,
      "step": 1394
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.85642472615397,
      "learning_rate": 1.7165156778729355e-05,
      "loss": 0.8407,
      "step": 1395
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8526127845229322,
      "learning_rate": 1.7160808930590845e-05,
      "loss": 0.9086,
      "step": 1396
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9671523719376158,
      "learning_rate": 1.7156458302379753e-05,
      "loss": 0.8962,
      "step": 1397
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9493290141508168,
      "learning_rate": 1.7152104895785147e-05,
      "loss": 0.8224,
      "step": 1398
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9753593789379112,
      "learning_rate": 1.7147748712497162e-05,
      "loss": 0.8742,
      "step": 1399
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.877771253207926,
      "learning_rate": 1.7143389754207026e-05,
      "loss": 0.907,
      "step": 1400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9247703071284201,
      "learning_rate": 1.713902802260703e-05,
      "loss": 0.8925,
      "step": 1401
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9958887373094057,
      "learning_rate": 1.7134663519390557e-05,
      "loss": 0.8937,
      "step": 1402
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8826537862564542,
      "learning_rate": 1.7130296246252048e-05,
      "loss": 0.865,
      "step": 1403
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8289887252650691,
      "learning_rate": 1.7125926204887034e-05,
      "loss": 0.8704,
      "step": 1404
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.035708442093661,
      "learning_rate": 1.712155339699211e-05,
      "loss": 0.8701,
      "step": 1405
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9579994641678127,
      "learning_rate": 1.7117177824264962e-05,
      "loss": 0.8855,
      "step": 1406
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6823087853078077,
      "learning_rate": 1.7112799488404327e-05,
      "loss": 0.3586,
      "step": 1407
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8801912181105376,
      "learning_rate": 1.7108418391110033e-05,
      "loss": 0.8411,
      "step": 1408
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9223968629494282,
      "learning_rate": 1.7104034534082968e-05,
      "loss": 0.8847,
      "step": 1409
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8990227424497484,
      "learning_rate": 1.7099647919025096e-05,
      "loss": 0.8378,
      "step": 1410
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8545743771943869,
      "learning_rate": 1.7095258547639456e-05,
      "loss": 0.8789,
      "step": 1411
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0448905116793399,
      "learning_rate": 1.709086642163015e-05,
      "loss": 0.8422,
      "step": 1412
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9197117974110159,
      "learning_rate": 1.7086471542702355e-05,
      "loss": 0.8804,
      "step": 1413
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8309579943159204,
      "learning_rate": 1.708207391256231e-05,
      "loss": 0.8506,
      "step": 1414
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9367873119167642,
      "learning_rate": 1.707767353291733e-05,
      "loss": 0.8187,
      "step": 1415
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0405466043469933,
      "learning_rate": 1.7073270405475796e-05,
      "loss": 0.8584,
      "step": 1416
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0094918626085747,
      "learning_rate": 1.7068864531947147e-05,
      "loss": 0.8317,
      "step": 1417
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9283301578660179,
      "learning_rate": 1.70644559140419e-05,
      "loss": 0.8877,
      "step": 1418
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9185178172303632,
      "learning_rate": 1.706004455347163e-05,
      "loss": 0.8442,
      "step": 1419
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9237497585801865,
      "learning_rate": 1.705563045194898e-05,
      "loss": 0.8699,
      "step": 1420
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0311970085288673,
      "learning_rate": 1.7051213611187657e-05,
      "loss": 0.8605,
      "step": 1421
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8810031267658764,
      "learning_rate": 1.704679403290243e-05,
      "loss": 0.8179,
      "step": 1422
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8713687741350276,
      "learning_rate": 1.7042371718809132e-05,
      "loss": 0.8821,
      "step": 1423
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8802045287970798,
      "learning_rate": 1.7037946670624652e-05,
      "loss": 0.9071,
      "step": 1424
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0006478232692055,
      "learning_rate": 1.7033518890066956e-05,
      "loss": 0.8441,
      "step": 1425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9404768316736005,
      "learning_rate": 1.7029088378855055e-05,
      "loss": 0.8603,
      "step": 1426
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9346473983684817,
      "learning_rate": 1.7024655138709025e-05,
      "loss": 0.8676,
      "step": 1427
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9758873096706631,
      "learning_rate": 1.7020219171350004e-05,
      "loss": 0.8356,
      "step": 1428
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7391366306554745,
      "learning_rate": 1.7015780478500187e-05,
      "loss": 0.3542,
      "step": 1429
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0299073429883019,
      "learning_rate": 1.701133906188283e-05,
      "loss": 0.8223,
      "step": 1430
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9282540995034538,
      "learning_rate": 1.700689492322224e-05,
      "loss": 0.8373,
      "step": 1431
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9705228032414721,
      "learning_rate": 1.700244806424379e-05,
      "loss": 0.8866,
      "step": 1432
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9333619032860123,
      "learning_rate": 1.6997998486673893e-05,
      "loss": 0.8286,
      "step": 1433
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9128753291722924,
      "learning_rate": 1.699354619224004e-05,
      "loss": 0.8795,
      "step": 1434
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9765253317435493,
      "learning_rate": 1.698909118267076e-05,
      "loss": 0.8482,
      "step": 1435
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.84580115113318,
      "learning_rate": 1.6984633459695646e-05,
      "loss": 0.8802,
      "step": 1436
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8679393391911872,
      "learning_rate": 1.6980173025045328e-05,
      "loss": 0.801,
      "step": 1437
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9375304016011884,
      "learning_rate": 1.697570988045151e-05,
      "loss": 0.9005,
      "step": 1438
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9967864246091177,
      "learning_rate": 1.6971244027646937e-05,
      "loss": 0.8997,
      "step": 1439
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9771690859415288,
      "learning_rate": 1.69667754683654e-05,
      "loss": 0.8263,
      "step": 1440
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9702304688603192,
      "learning_rate": 1.6962304204341758e-05,
      "loss": 0.8402,
      "step": 1441
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9871422947025689,
      "learning_rate": 1.6957830237311904e-05,
      "loss": 0.8238,
      "step": 1442
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8180517372225792,
      "learning_rate": 1.6953353569012784e-05,
      "loss": 0.8586,
      "step": 1443
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9491103856363544,
      "learning_rate": 1.6948874201182402e-05,
      "loss": 0.8432,
      "step": 1444
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9312937483616247,
      "learning_rate": 1.6944392135559798e-05,
      "loss": 0.8653,
      "step": 1445
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.935385929361679,
      "learning_rate": 1.6939907373885062e-05,
      "loss": 0.8341,
      "step": 1446
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9154728639287718,
      "learning_rate": 1.6935419917899335e-05,
      "loss": 0.8294,
      "step": 1447
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9704916603057421,
      "learning_rate": 1.6930929769344807e-05,
      "loss": 0.8603,
      "step": 1448
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8741532662202783,
      "learning_rate": 1.69264369299647e-05,
      "loss": 0.8604,
      "step": 1449
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9361223861482604,
      "learning_rate": 1.692194140150329e-05,
      "loss": 0.8765,
      "step": 1450
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8876549328117052,
      "learning_rate": 1.69174431857059e-05,
      "loss": 0.8788,
      "step": 1451
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7434286355904817,
      "learning_rate": 1.6912942284318898e-05,
      "loss": 0.3691,
      "step": 1452
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.958399953650501,
      "learning_rate": 1.6908438699089674e-05,
      "loss": 0.882,
      "step": 1453
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9407539892903738,
      "learning_rate": 1.690393243176668e-05,
      "loss": 0.8386,
      "step": 1454
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9124806359958285,
      "learning_rate": 1.6899423484099413e-05,
      "loss": 0.7746,
      "step": 1455
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9603142319854733,
      "learning_rate": 1.6894911857838394e-05,
      "loss": 0.8398,
      "step": 1456
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9196906249639387,
      "learning_rate": 1.689039755473519e-05,
      "loss": 0.8958,
      "step": 1457
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9218606049870551,
      "learning_rate": 1.6885880576542417e-05,
      "loss": 0.8021,
      "step": 1458
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9116366835355816,
      "learning_rate": 1.6881360925013712e-05,
      "loss": 0.8746,
      "step": 1459
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8722332163355812,
      "learning_rate": 1.6876838601903765e-05,
      "loss": 0.8722,
      "step": 1460
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.948634816725846,
      "learning_rate": 1.6872313608968296e-05,
      "loss": 0.8738,
      "step": 1461
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.975774148307411,
      "learning_rate": 1.6867785947964065e-05,
      "loss": 0.8802,
      "step": 1462
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9989313188481971,
      "learning_rate": 1.6863255620648866e-05,
      "loss": 0.8593,
      "step": 1463
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8679742164368862,
      "learning_rate": 1.685872262878152e-05,
      "loss": 0.8111,
      "step": 1464
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8748959338141457,
      "learning_rate": 1.6854186974121903e-05,
      "loss": 0.8882,
      "step": 1465
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8092134636725681,
      "learning_rate": 1.68496486584309e-05,
      "loss": 0.8636,
      "step": 1466
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9329117833948616,
      "learning_rate": 1.6845107683470453e-05,
      "loss": 0.8774,
      "step": 1467
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9152565593814938,
      "learning_rate": 1.6840564051003517e-05,
      "loss": 0.8373,
      "step": 1468
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9693844997221372,
      "learning_rate": 1.6836017762794087e-05,
      "loss": 0.8645,
      "step": 1469
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9590159824216442,
      "learning_rate": 1.6831468820607192e-05,
      "loss": 0.8404,
      "step": 1470
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9119179880184769,
      "learning_rate": 1.6826917226208886e-05,
      "loss": 0.9196,
      "step": 1471
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9110648122426552,
      "learning_rate": 1.6822362981366257e-05,
      "loss": 0.8368,
      "step": 1472
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9047719169096858,
      "learning_rate": 1.6817806087847417e-05,
      "loss": 0.8484,
      "step": 1473
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.050159108194291,
      "learning_rate": 1.681324654742151e-05,
      "loss": 0.8042,
      "step": 1474
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9614749953246081,
      "learning_rate": 1.6808684361858706e-05,
      "loss": 0.8449,
      "step": 1475
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9554579070884294,
      "learning_rate": 1.6804119532930202e-05,
      "loss": 0.8341,
      "step": 1476
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9611677784558422,
      "learning_rate": 1.6799552062408225e-05,
      "loss": 0.8336,
      "step": 1477
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.009482107005681,
      "learning_rate": 1.6794981952066018e-05,
      "loss": 0.8325,
      "step": 1478
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7798689398306743,
      "learning_rate": 1.6790409203677863e-05,
      "loss": 0.3757,
      "step": 1479
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7906827522058792,
      "learning_rate": 1.6785833819019052e-05,
      "loss": 0.8535,
      "step": 1480
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8983792648930081,
      "learning_rate": 1.678125579986591e-05,
      "loss": 0.8478,
      "step": 1481
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9199328753447427,
      "learning_rate": 1.677667514799578e-05,
      "loss": 0.8838,
      "step": 1482
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8823122315049025,
      "learning_rate": 1.6772091865187032e-05,
      "loss": 0.791,
      "step": 1483
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9353306487848541,
      "learning_rate": 1.676750595321905e-05,
      "loss": 0.8823,
      "step": 1484
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9177332453005533,
      "learning_rate": 1.6762917413872246e-05,
      "loss": 0.8172,
      "step": 1485
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9867419566799213,
      "learning_rate": 1.675832624892805e-05,
      "loss": 0.8649,
      "step": 1486
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9848527799017143,
      "learning_rate": 1.6753732460168907e-05,
      "loss": 0.7796,
      "step": 1487
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0108675321542762,
      "learning_rate": 1.674913604937828e-05,
      "loss": 0.8504,
      "step": 1488
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9260998753009939,
      "learning_rate": 1.6744537018340662e-05,
      "loss": 0.8531,
      "step": 1489
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0105825664420098,
      "learning_rate": 1.6739935368841555e-05,
      "loss": 0.8665,
      "step": 1490
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8861263677706575,
      "learning_rate": 1.6735331102667475e-05,
      "loss": 0.8456,
      "step": 1491
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9874986553296158,
      "learning_rate": 1.6730724221605955e-05,
      "loss": 0.8582,
      "step": 1492
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9653637137493576,
      "learning_rate": 1.6726114727445547e-05,
      "loss": 0.838,
      "step": 1493
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0312443271574854,
      "learning_rate": 1.6721502621975813e-05,
      "loss": 0.8738,
      "step": 1494
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.914397711731916,
      "learning_rate": 1.6716887906987332e-05,
      "loss": 0.8828,
      "step": 1495
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9584646069198112,
      "learning_rate": 1.6712270584271703e-05,
      "loss": 0.8885,
      "step": 1496
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9552756341266667,
      "learning_rate": 1.670765065562152e-05,
      "loss": 0.8968,
      "step": 1497
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9196908091392165,
      "learning_rate": 1.67030281228304e-05,
      "loss": 0.8328,
      "step": 1498
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9208967202778376,
      "learning_rate": 1.6698402987692968e-05,
      "loss": 0.8714,
      "step": 1499
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0019926213033055,
      "learning_rate": 1.6693775252004866e-05,
      "loss": 0.8886,
      "step": 1500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8787293399148045,
      "learning_rate": 1.668914491756274e-05,
      "loss": 0.8624,
      "step": 1501
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.874315587957002,
      "learning_rate": 1.668451198616424e-05,
      "loss": 0.8203,
      "step": 1502
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9227311133945645,
      "learning_rate": 1.6679876459608033e-05,
      "loss": 0.8441,
      "step": 1503
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7083666401882114,
      "learning_rate": 1.667523833969379e-05,
      "loss": 0.3279,
      "step": 1504
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9790468475266028,
      "learning_rate": 1.667059762822219e-05,
      "loss": 0.844,
      "step": 1505
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9291092600552696,
      "learning_rate": 1.666595432699491e-05,
      "loss": 0.8874,
      "step": 1506
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8854978813558275,
      "learning_rate": 1.6661308437814652e-05,
      "loss": 0.8383,
      "step": 1507
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.859477731947404,
      "learning_rate": 1.6656659962485097e-05,
      "loss": 0.8329,
      "step": 1508
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9423234693589986,
      "learning_rate": 1.6652008902810952e-05,
      "loss": 0.8868,
      "step": 1509
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.940957074646862,
      "learning_rate": 1.6647355260597915e-05,
      "loss": 0.8168,
      "step": 1510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9574970787894114,
      "learning_rate": 1.664269903765269e-05,
      "loss": 0.8618,
      "step": 1511
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0612094323922023,
      "learning_rate": 1.6638040235782983e-05,
      "loss": 0.9442,
      "step": 1512
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9215059611339891,
      "learning_rate": 1.6633378856797505e-05,
      "loss": 0.893,
      "step": 1513
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8537420315164669,
      "learning_rate": 1.662871490250596e-05,
      "loss": 0.8927,
      "step": 1514
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9949544738726046,
      "learning_rate": 1.662404837471905e-05,
      "loss": 0.879,
      "step": 1515
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8693390429567069,
      "learning_rate": 1.66193792752485e-05,
      "loss": 0.8741,
      "step": 1516
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.893704725916103,
      "learning_rate": 1.6614707605906995e-05,
      "loss": 0.852,
      "step": 1517
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9638455277621011,
      "learning_rate": 1.661003336850825e-05,
      "loss": 0.8905,
      "step": 1518
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9388476708789977,
      "learning_rate": 1.660535656486696e-05,
      "loss": 0.8843,
      "step": 1519
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9870002855799506,
      "learning_rate": 1.660067719679882e-05,
      "loss": 0.8582,
      "step": 1520
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9973083895121924,
      "learning_rate": 1.6595995266120528e-05,
      "loss": 0.941,
      "step": 1521
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9414884953760438,
      "learning_rate": 1.6591310774649766e-05,
      "loss": 0.8656,
      "step": 1522
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9673334653153137,
      "learning_rate": 1.6586623724205216e-05,
      "loss": 0.8302,
      "step": 1523
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9546960688657534,
      "learning_rate": 1.6581934116606554e-05,
      "loss": 0.8949,
      "step": 1524
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9629431699572075,
      "learning_rate": 1.657724195367444e-05,
      "loss": 0.8751,
      "step": 1525
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9016528418285694,
      "learning_rate": 1.657254723723054e-05,
      "loss": 0.8164,
      "step": 1526
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9177377265526541,
      "learning_rate": 1.6567849969097505e-05,
      "loss": 0.8468,
      "step": 1527
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8752347641074195,
      "learning_rate": 1.6563150151098973e-05,
      "loss": 0.839,
      "step": 1528
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7594418272405481,
      "learning_rate": 1.6558447785059577e-05,
      "loss": 0.3207,
      "step": 1529
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9910668706490896,
      "learning_rate": 1.655374287280494e-05,
      "loss": 0.9034,
      "step": 1530
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9828492961580876,
      "learning_rate": 1.6549035416161662e-05,
      "loss": 0.8964,
      "step": 1531
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9018478353215523,
      "learning_rate": 1.654432541695735e-05,
      "loss": 0.8341,
      "step": 1532
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8821433603153473,
      "learning_rate": 1.653961287702058e-05,
      "loss": 0.8742,
      "step": 1533
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9241207436376797,
      "learning_rate": 1.653489779818093e-05,
      "loss": 0.8398,
      "step": 1534
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8965517741518974,
      "learning_rate": 1.6530180182268946e-05,
      "loss": 0.793,
      "step": 1535
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8224172244291974,
      "learning_rate": 1.652546003111618e-05,
      "loss": 0.8166,
      "step": 1536
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8973979233846617,
      "learning_rate": 1.652073734655515e-05,
      "loss": 0.8367,
      "step": 1537
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8544825366701363,
      "learning_rate": 1.6516012130419366e-05,
      "loss": 0.8341,
      "step": 1538
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8880311762812442,
      "learning_rate": 1.6511284384543317e-05,
      "loss": 0.8086,
      "step": 1539
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0119798700267943,
      "learning_rate": 1.6506554110762483e-05,
      "loss": 0.863,
      "step": 1540
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9884341099567423,
      "learning_rate": 1.650182131091332e-05,
      "loss": 0.8731,
      "step": 1541
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9222804896171843,
      "learning_rate": 1.6497085986833252e-05,
      "loss": 0.8405,
      "step": 1542
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9650718453374061,
      "learning_rate": 1.6492348140360704e-05,
      "loss": 0.8521,
      "step": 1543
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9334005377612737,
      "learning_rate": 1.6487607773335074e-05,
      "loss": 0.843,
      "step": 1544
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8943425492512767,
      "learning_rate": 1.648286488759673e-05,
      "loss": 0.7814,
      "step": 1545
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8761518494653457,
      "learning_rate": 1.6478119484987026e-05,
      "loss": 0.8692,
      "step": 1546
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9230015158556538,
      "learning_rate": 1.6473371567348287e-05,
      "loss": 0.8759,
      "step": 1547
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6703177025066237,
      "learning_rate": 1.6468621136523823e-05,
      "loss": 0.3278,
      "step": 1548
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9020116410834526,
      "learning_rate": 1.646386819435791e-05,
      "loss": 0.8241,
      "step": 1549
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9333943288438883,
      "learning_rate": 1.6459112742695807e-05,
      "loss": 0.8608,
      "step": 1550
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9773800117449802,
      "learning_rate": 1.6454354783383748e-05,
      "loss": 0.8007,
      "step": 1551
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9601768687857349,
      "learning_rate": 1.644959431826893e-05,
      "loss": 0.9015,
      "step": 1552
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8709057322491991,
      "learning_rate": 1.6444831349199528e-05,
      "loss": 0.8348,
      "step": 1553
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9177734389155945,
      "learning_rate": 1.6440065878024697e-05,
      "loss": 0.8178,
      "step": 1554
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.882499788978848,
      "learning_rate": 1.6435297906594553e-05,
      "loss": 0.8911,
      "step": 1555
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9477839567711032,
      "learning_rate": 1.643052743676019e-05,
      "loss": 0.7694,
      "step": 1556
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9993930137244931,
      "learning_rate": 1.6425754470373667e-05,
      "loss": 0.8574,
      "step": 1557
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7966548087202375,
      "learning_rate": 1.642097900928801e-05,
      "loss": 0.3702,
      "step": 1558
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0149564638051334,
      "learning_rate": 1.6416201055357225e-05,
      "loss": 0.8183,
      "step": 1559
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9496450965416646,
      "learning_rate": 1.641142061043627e-05,
      "loss": 0.7875,
      "step": 1560
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8805121593364662,
      "learning_rate": 1.640663767638108e-05,
      "loss": 0.9264,
      "step": 1561
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9858745207032383,
      "learning_rate": 1.6401852255048564e-05,
      "loss": 0.8621,
      "step": 1562
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9683060698419903,
      "learning_rate": 1.6397064348296578e-05,
      "loss": 0.8442,
      "step": 1563
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6344032368224581,
      "learning_rate": 1.6392273957983955e-05,
      "loss": 0.3711,
      "step": 1564
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0706624245223124,
      "learning_rate": 1.638748108597049e-05,
      "loss": 0.8884,
      "step": 1565
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9265454407473649,
      "learning_rate": 1.6382685734116934e-05,
      "loss": 0.8642,
      "step": 1566
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9270774533276412,
      "learning_rate": 1.6377887904285018e-05,
      "loss": 0.837,
      "step": 1567
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9116679433932229,
      "learning_rate": 1.637308759833742e-05,
      "loss": 0.8258,
      "step": 1568
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9216176845543587,
      "learning_rate": 1.6368284818137787e-05,
      "loss": 0.8933,
      "step": 1569
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8959430291679986,
      "learning_rate": 1.636347956555072e-05,
      "loss": 0.8813,
      "step": 1570
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9599575453045383,
      "learning_rate": 1.635867184244178e-05,
      "loss": 0.8494,
      "step": 1571
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8782181700773556,
      "learning_rate": 1.63538616506775e-05,
      "loss": 0.8201,
      "step": 1572
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0197490170908485,
      "learning_rate": 1.6349048992125358e-05,
      "loss": 0.8927,
      "step": 1573
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9318504896848653,
      "learning_rate": 1.634423386865379e-05,
      "loss": 0.8553,
      "step": 1574
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8782920458233964,
      "learning_rate": 1.6339416282132196e-05,
      "loss": 0.8735,
      "step": 1575
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9680482166824512,
      "learning_rate": 1.633459623443093e-05,
      "loss": 0.9252,
      "step": 1576
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9260720077324531,
      "learning_rate": 1.6329773727421297e-05,
      "loss": 0.8859,
      "step": 1577
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8371085425872065,
      "learning_rate": 1.6324948762975567e-05,
      "loss": 0.8532,
      "step": 1578
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9434729315499106,
      "learning_rate": 1.632012134296695e-05,
      "loss": 0.8387,
      "step": 1579
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9245677657609859,
      "learning_rate": 1.6315291469269617e-05,
      "loss": 0.8257,
      "step": 1580
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7390497332426954,
      "learning_rate": 1.63104591437587e-05,
      "loss": 0.3775,
      "step": 1581
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6298860309577747,
      "learning_rate": 1.6305624368310265e-05,
      "loss": 0.3419,
      "step": 1582
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9072366817595753,
      "learning_rate": 1.630078714480134e-05,
      "loss": 0.841,
      "step": 1583
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9137206462296844,
      "learning_rate": 1.6295947475109904e-05,
      "loss": 0.8444,
      "step": 1584
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0020277237681396,
      "learning_rate": 1.629110536111488e-05,
      "loss": 0.8669,
      "step": 1585
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.940996512948944,
      "learning_rate": 1.628626080469615e-05,
      "loss": 0.8715,
      "step": 1586
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9080274520962657,
      "learning_rate": 1.628141380773453e-05,
      "loss": 0.8079,
      "step": 1587
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0567644387383153,
      "learning_rate": 1.6276564372111797e-05,
      "loss": 0.9092,
      "step": 1588
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9033069790036525,
      "learning_rate": 1.6271712499710663e-05,
      "loss": 0.8205,
      "step": 1589
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9771032425333117,
      "learning_rate": 1.62668581924148e-05,
      "loss": 0.8972,
      "step": 1590
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8411606747435716,
      "learning_rate": 1.6262001452108807e-05,
      "loss": 0.349,
      "step": 1591
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8968403587054132,
      "learning_rate": 1.6257142280678247e-05,
      "loss": 0.8809,
      "step": 1592
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0026748929271616,
      "learning_rate": 1.6252280680009613e-05,
      "loss": 0.9014,
      "step": 1593
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9174594373002899,
      "learning_rate": 1.6247416651990343e-05,
      "loss": 0.8409,
      "step": 1594
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8733972073490497,
      "learning_rate": 1.624255019850883e-05,
      "loss": 0.8051,
      "step": 1595
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8617765760538291,
      "learning_rate": 1.6237681321454387e-05,
      "loss": 0.8917,
      "step": 1596
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9123586564433416,
      "learning_rate": 1.623281002271729e-05,
      "loss": 0.8903,
      "step": 1597
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8617312035304858,
      "learning_rate": 1.6227936304188738e-05,
      "loss": 0.7963,
      "step": 1598
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8881495341050858,
      "learning_rate": 1.622306016776088e-05,
      "loss": 0.8284,
      "step": 1599
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9644808790291267,
      "learning_rate": 1.6218181615326795e-05,
      "loss": 0.876,
      "step": 1600
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9845343732299995,
      "learning_rate": 1.6213300648780515e-05,
      "loss": 0.9011,
      "step": 1601
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9276174305943505,
      "learning_rate": 1.620841727001699e-05,
      "loss": 0.8137,
      "step": 1602
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9327487696699454,
      "learning_rate": 1.6203531480932114e-05,
      "loss": 0.8222,
      "step": 1603
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9874057686195213,
      "learning_rate": 1.619864328342273e-05,
      "loss": 0.8506,
      "step": 1604
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.858726308654083,
      "learning_rate": 1.6193752679386593e-05,
      "loss": 0.8459,
      "step": 1605
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9403864406434129,
      "learning_rate": 1.6188859670722414e-05,
      "loss": 0.8269,
      "step": 1606
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9169609584507873,
      "learning_rate": 1.6183964259329817e-05,
      "loss": 0.8616,
      "step": 1607
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9141714230362586,
      "learning_rate": 1.6179066447109376e-05,
      "loss": 0.8301,
      "step": 1608
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9860616961475185,
      "learning_rate": 1.6174166235962588e-05,
      "loss": 0.8433,
      "step": 1609
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9616369633766064,
      "learning_rate": 1.6169263627791886e-05,
      "loss": 0.8815,
      "step": 1610
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.930501686783607,
      "learning_rate": 1.616435862450063e-05,
      "loss": 0.8455,
      "step": 1611
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8441606373380772,
      "learning_rate": 1.615945122799311e-05,
      "loss": 0.3731,
      "step": 1612
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7002547048042236,
      "learning_rate": 1.6154541440174547e-05,
      "loss": 0.3542,
      "step": 1613
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9283503304637784,
      "learning_rate": 1.614962926295109e-05,
      "loss": 0.8253,
      "step": 1614
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0685417465963103,
      "learning_rate": 1.6144714698229814e-05,
      "loss": 0.8056,
      "step": 1615
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9643883470169773,
      "learning_rate": 1.6139797747918725e-05,
      "loss": 0.8783,
      "step": 1616
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8841005773452021,
      "learning_rate": 1.613487841392675e-05,
      "loss": 0.8648,
      "step": 1617
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9609081124012467,
      "learning_rate": 1.612995669816375e-05,
      "loss": 0.857,
      "step": 1618
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9036318073096852,
      "learning_rate": 1.6125032602540492e-05,
      "loss": 0.8837,
      "step": 1619
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8711528357109345,
      "learning_rate": 1.6120106128968686e-05,
      "loss": 0.8931,
      "step": 1620
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9881681369990489,
      "learning_rate": 1.6115177279360965e-05,
      "loss": 0.8841,
      "step": 1621
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9684247032667483,
      "learning_rate": 1.611024605563087e-05,
      "loss": 0.807,
      "step": 1622
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9590336860751467,
      "learning_rate": 1.610531245969287e-05,
      "loss": 0.8566,
      "step": 1623
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9131587255759439,
      "learning_rate": 1.6100376493462368e-05,
      "loss": 0.9037,
      "step": 1624
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8564403255408799,
      "learning_rate": 1.6095438158855668e-05,
      "loss": 0.8098,
      "step": 1625
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0133196911692832,
      "learning_rate": 1.609049745779e-05,
      "loss": 0.3222,
      "step": 1626
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9347298766436942,
      "learning_rate": 1.6085554392183517e-05,
      "loss": 0.8767,
      "step": 1627
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8518763427829182,
      "learning_rate": 1.608060896395529e-05,
      "loss": 0.814,
      "step": 1628
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9329890000563096,
      "learning_rate": 1.60756611750253e-05,
      "loss": 0.8767,
      "step": 1629
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9387151727897931,
      "learning_rate": 1.6070711027314446e-05,
      "loss": 0.7957,
      "step": 1630
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9919693946003626,
      "learning_rate": 1.606575852274456e-05,
      "loss": 0.9462,
      "step": 1631
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6813985666689464,
      "learning_rate": 1.6060803663238357e-05,
      "loss": 0.3556,
      "step": 1632
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8859617169941388,
      "learning_rate": 1.6055846450719498e-05,
      "loss": 0.8602,
      "step": 1633
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.917942459156067,
      "learning_rate": 1.6050886887112535e-05,
      "loss": 0.863,
      "step": 1634
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8283186536537116,
      "learning_rate": 1.6045924974342945e-05,
      "loss": 0.8806,
      "step": 1635
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9495378326380399,
      "learning_rate": 1.604096071433711e-05,
      "loss": 0.7998,
      "step": 1636
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.852177629619996,
      "learning_rate": 1.6035994109022333e-05,
      "loss": 0.8768,
      "step": 1637
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9055444154698985,
      "learning_rate": 1.6031025160326814e-05,
      "loss": 0.8605,
      "step": 1638
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9290485654600965,
      "learning_rate": 1.6026053870179678e-05,
      "loss": 0.8096,
      "step": 1639
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9086366531381428,
      "learning_rate": 1.6021080240510943e-05,
      "loss": 0.8408,
      "step": 1640
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9742969239523744,
      "learning_rate": 1.601610427325155e-05,
      "loss": 0.8616,
      "step": 1641
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8658545804239492,
      "learning_rate": 1.6011125970333333e-05,
      "loss": 0.8676,
      "step": 1642
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8481479434975266,
      "learning_rate": 1.600614533368905e-05,
      "loss": 0.854,
      "step": 1643
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9531977644782048,
      "learning_rate": 1.6001162365252348e-05,
      "loss": 0.8176,
      "step": 1644
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9186003016563306,
      "learning_rate": 1.5996177066957787e-05,
      "loss": 0.8587,
      "step": 1645
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0640968472392511,
      "learning_rate": 1.5991189440740838e-05,
      "loss": 0.8333,
      "step": 1646
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9269257897963394,
      "learning_rate": 1.5986199488537867e-05,
      "loss": 0.893,
      "step": 1647
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9278685826552993,
      "learning_rate": 1.598120721228614e-05,
      "loss": 0.8605,
      "step": 1648
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8990973969770869,
      "learning_rate": 1.5976212613923836e-05,
      "loss": 0.8524,
      "step": 1649
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.964232448746923,
      "learning_rate": 1.5971215695390026e-05,
      "loss": 0.8413,
      "step": 1650
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8692671711820789,
      "learning_rate": 1.5966216458624692e-05,
      "loss": 0.8277,
      "step": 1651
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9247971893689358,
      "learning_rate": 1.5961214905568705e-05,
      "loss": 0.8542,
      "step": 1652
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9476696180334772,
      "learning_rate": 1.595621103816384e-05,
      "loss": 0.9068,
      "step": 1653
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0072588531876303,
      "learning_rate": 1.5951204858352772e-05,
      "loss": 0.9037,
      "step": 1654
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8614055380882291,
      "learning_rate": 1.594619636807907e-05,
      "loss": 0.8236,
      "step": 1655
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9414808007109455,
      "learning_rate": 1.5941185569287206e-05,
      "loss": 0.8924,
      "step": 1656
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9526805369018754,
      "learning_rate": 1.5936172463922542e-05,
      "loss": 0.8426,
      "step": 1657
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9663168961986072,
      "learning_rate": 1.593115705393134e-05,
      "loss": 0.8186,
      "step": 1658
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8915915924739047,
      "learning_rate": 1.5926139341260755e-05,
      "loss": 0.877,
      "step": 1659
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8764768201174455,
      "learning_rate": 1.5921119327858835e-05,
      "loss": 0.8726,
      "step": 1660
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8840660020916732,
      "learning_rate": 1.5916097015674518e-05,
      "loss": 0.8838,
      "step": 1661
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.024442060958909,
      "learning_rate": 1.5911072406657646e-05,
      "loss": 0.8785,
      "step": 1662
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8363038919003488,
      "learning_rate": 1.5906045502758943e-05,
      "loss": 0.8379,
      "step": 1663
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9426622953460951,
      "learning_rate": 1.590101630593002e-05,
      "loss": 0.8245,
      "step": 1664
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8816958522897141,
      "learning_rate": 1.5895984818123392e-05,
      "loss": 0.9343,
      "step": 1665
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8950524324091882,
      "learning_rate": 1.5890951041292453e-05,
      "loss": 0.8792,
      "step": 1666
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9368043671851101,
      "learning_rate": 1.588591497739149e-05,
      "loss": 0.8435,
      "step": 1667
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8814788037889155,
      "learning_rate": 1.5880876628375668e-05,
      "loss": 0.8208,
      "step": 1668
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.91096873942418,
      "learning_rate": 1.587583599620106e-05,
      "loss": 0.8205,
      "step": 1669
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8721096457392798,
      "learning_rate": 1.5870793082824604e-05,
      "loss": 0.7806,
      "step": 1670
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.968686215162025,
      "learning_rate": 1.5865747890204138e-05,
      "loss": 0.8559,
      "step": 1671
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0448955657356387,
      "learning_rate": 1.5860700420298377e-05,
      "loss": 0.3646,
      "step": 1672
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9149326554701476,
      "learning_rate": 1.5855650675066924e-05,
      "loss": 0.8759,
      "step": 1673
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9627977145279535,
      "learning_rate": 1.5850598656470265e-05,
      "loss": 0.8477,
      "step": 1674
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.938371513016805,
      "learning_rate": 1.584554436646976e-05,
      "loss": 0.8352,
      "step": 1675
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9859449137886535,
      "learning_rate": 1.5840487807027665e-05,
      "loss": 0.8598,
      "step": 1676
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9910941719940294,
      "learning_rate": 1.5835428980107113e-05,
      "loss": 0.8374,
      "step": 1677
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8529921348633568,
      "learning_rate": 1.583036788767211e-05,
      "loss": 0.8675,
      "step": 1678
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9112101994243816,
      "learning_rate": 1.5825304531687548e-05,
      "loss": 0.8365,
      "step": 1679
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8666219803148173,
      "learning_rate": 1.5820238914119195e-05,
      "loss": 0.8131,
      "step": 1680
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9562492399138658,
      "learning_rate": 1.5815171036933697e-05,
      "loss": 0.8024,
      "step": 1681
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9185041023129298,
      "learning_rate": 1.5810100902098582e-05,
      "loss": 0.8998,
      "step": 1682
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9528549264354593,
      "learning_rate": 1.580502851158225e-05,
      "loss": 0.8791,
      "step": 1683
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8904227507631272,
      "learning_rate": 1.5799953867353975e-05,
      "loss": 0.8635,
      "step": 1684
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9489312122757222,
      "learning_rate": 1.579487697138391e-05,
      "loss": 0.8673,
      "step": 1685
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1531571984318576,
      "learning_rate": 1.5789797825643086e-05,
      "loss": 0.8215,
      "step": 1686
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.903056910490669,
      "learning_rate": 1.5784716432103394e-05,
      "loss": 0.8769,
      "step": 1687
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9219313151981534,
      "learning_rate": 1.5779632792737608e-05,
      "loss": 0.9277,
      "step": 1688
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9472606730085892,
      "learning_rate": 1.5774546909519376e-05,
      "loss": 0.8522,
      "step": 1689
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8826684380610071,
      "learning_rate": 1.5769458784423206e-05,
      "loss": 0.8314,
      "step": 1690
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0333728761256764,
      "learning_rate": 1.5764368419424488e-05,
      "loss": 0.8566,
      "step": 1691
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8715195968716078,
      "learning_rate": 1.575927581649948e-05,
      "loss": 0.8735,
      "step": 1692
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9407481221874833,
      "learning_rate": 1.5754180977625303e-05,
      "loss": 0.9006,
      "step": 1693
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9442228001207122,
      "learning_rate": 1.574908390477995e-05,
      "loss": 0.8251,
      "step": 1694
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9056481673223961,
      "learning_rate": 1.5743984599942273e-05,
      "loss": 0.8584,
      "step": 1695
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9031621701765301,
      "learning_rate": 1.5738883065092005e-05,
      "loss": 0.8453,
      "step": 1696
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9188120736013115,
      "learning_rate": 1.5733779302209735e-05,
      "loss": 0.8129,
      "step": 1697
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.963904186116481,
      "learning_rate": 1.572867331327692e-05,
      "loss": 0.8274,
      "step": 1698
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9585526575265366,
      "learning_rate": 1.5723565100275884e-05,
      "loss": 0.867,
      "step": 1699
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9660309071629689,
      "learning_rate": 1.5718454665189806e-05,
      "loss": 0.8759,
      "step": 1700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9624420867380036,
      "learning_rate": 1.5713342010002733e-05,
      "loss": 0.8332,
      "step": 1701
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9145080147925806,
      "learning_rate": 1.5708227136699578e-05,
      "loss": 0.8196,
      "step": 1702
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8910518419954789,
      "learning_rate": 1.5703110047266105e-05,
      "loss": 0.7902,
      "step": 1703
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9387732208353655,
      "learning_rate": 1.569799074368895e-05,
      "loss": 0.8225,
      "step": 1704
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8960006426609993,
      "learning_rate": 1.5692869227955603e-05,
      "loss": 0.8051,
      "step": 1705
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0388878116633116,
      "learning_rate": 1.5687745502054407e-05,
      "loss": 0.786,
      "step": 1706
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0018011606710515,
      "learning_rate": 1.5682619567974575e-05,
      "loss": 0.8895,
      "step": 1707
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9715118103837384,
      "learning_rate": 1.567749142770617e-05,
      "loss": 0.9035,
      "step": 1708
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9422371667768359,
      "learning_rate": 1.5672361083240106e-05,
      "loss": 0.8427,
      "step": 1709
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9772126673986357,
      "learning_rate": 1.5667228536568167e-05,
      "loss": 0.8765,
      "step": 1710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8879570270946235,
      "learning_rate": 1.566209378968298e-05,
      "loss": 0.3643,
      "step": 1711
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9013171072290466,
      "learning_rate": 1.565695684457803e-05,
      "loss": 0.7759,
      "step": 1712
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9234724216909708,
      "learning_rate": 1.5651817703247666e-05,
      "loss": 0.8976,
      "step": 1713
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9784711922381659,
      "learning_rate": 1.5646676367687067e-05,
      "loss": 0.8603,
      "step": 1714
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9312921933121947,
      "learning_rate": 1.564153283989228e-05,
      "loss": 0.8145,
      "step": 1715
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7340206206824262,
      "learning_rate": 1.5636387121860207e-05,
      "loss": 0.3564,
      "step": 1716
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9248388178619025,
      "learning_rate": 1.5631239215588578e-05,
      "loss": 0.7578,
      "step": 1717
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0538303108317801,
      "learning_rate": 1.5626089123076004e-05,
      "loss": 0.8478,
      "step": 1718
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9991647565848257,
      "learning_rate": 1.5620936846321917e-05,
      "loss": 0.8876,
      "step": 1719
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8662263888751773,
      "learning_rate": 1.561578238732661e-05,
      "loss": 0.8251,
      "step": 1720
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9211123681694305,
      "learning_rate": 1.561062574809123e-05,
      "loss": 0.8165,
      "step": 1721
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9533348167678284,
      "learning_rate": 1.5605466930617747e-05,
      "loss": 0.8669,
      "step": 1722
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8714475586747789,
      "learning_rate": 1.5600305936909005e-05,
      "loss": 0.8416,
      "step": 1723
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8683200316325933,
      "learning_rate": 1.559514276896867e-05,
      "loss": 0.9106,
      "step": 1724
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0301746687167541,
      "learning_rate": 1.558997742880127e-05,
      "loss": 0.9133,
      "step": 1725
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9207385567583036,
      "learning_rate": 1.5584809918412158e-05,
      "loss": 0.8445,
      "step": 1726
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8725945703318834,
      "learning_rate": 1.557964023980755e-05,
      "loss": 0.8178,
      "step": 1727
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8959531391771841,
      "learning_rate": 1.5574468394994486e-05,
      "loss": 0.3562,
      "step": 1728
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8809089140537366,
      "learning_rate": 1.5569294385980856e-05,
      "loss": 0.8377,
      "step": 1729
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9070740463523246,
      "learning_rate": 1.556411821477539e-05,
      "loss": 0.8067,
      "step": 1730
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9695836030971217,
      "learning_rate": 1.5558939883387657e-05,
      "loss": 0.8653,
      "step": 1731
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9643847544363983,
      "learning_rate": 1.5553759393828058e-05,
      "loss": 0.7739,
      "step": 1732
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9203340657045889,
      "learning_rate": 1.554857674810784e-05,
      "loss": 0.8999,
      "step": 1733
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9822371767344402,
      "learning_rate": 1.554339194823909e-05,
      "loss": 0.8635,
      "step": 1734
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9350876497380692,
      "learning_rate": 1.553820499623472e-05,
      "loss": 0.7636,
      "step": 1735
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8898428192669671,
      "learning_rate": 1.553301589410848e-05,
      "loss": 0.8056,
      "step": 1736
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8828444885212114,
      "learning_rate": 1.5527824643874968e-05,
      "loss": 0.8527,
      "step": 1737
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9467614201076696,
      "learning_rate": 1.5522631247549598e-05,
      "loss": 0.8261,
      "step": 1738
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9070562718952324,
      "learning_rate": 1.5517435707148628e-05,
      "loss": 0.8632,
      "step": 1739
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9939451290080973,
      "learning_rate": 1.5512238024689144e-05,
      "loss": 0.8998,
      "step": 1740
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9215284891937242,
      "learning_rate": 1.550703820218907e-05,
      "loss": 0.8483,
      "step": 1741
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9420461674739535,
      "learning_rate": 1.550183624166715e-05,
      "loss": 0.8574,
      "step": 1742
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9918015678205718,
      "learning_rate": 1.549663214514297e-05,
      "loss": 0.8775,
      "step": 1743
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8750762616179327,
      "learning_rate": 1.5491425914636934e-05,
      "loss": 0.7885,
      "step": 1744
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.81383493144181,
      "learning_rate": 1.5486217552170283e-05,
      "loss": 0.8644,
      "step": 1745
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9241407322432063,
      "learning_rate": 1.548100705976508e-05,
      "loss": 0.8661,
      "step": 1746
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9349902923628352,
      "learning_rate": 1.5475794439444226e-05,
      "loss": 0.9156,
      "step": 1747
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8889720654699763,
      "learning_rate": 1.5470579693231432e-05,
      "loss": 0.8121,
      "step": 1748
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9417640502712347,
      "learning_rate": 1.5465362823151245e-05,
      "loss": 0.8362,
      "step": 1749
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9245206196204288,
      "learning_rate": 1.5460143831229026e-05,
      "loss": 0.8847,
      "step": 1750
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0612620463525837,
      "learning_rate": 1.545492271949098e-05,
      "loss": 0.8906,
      "step": 1751
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9013102433111873,
      "learning_rate": 1.544969948996411e-05,
      "loss": 0.862,
      "step": 1752
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.840734510524284,
      "learning_rate": 1.544447414467626e-05,
      "loss": 0.8058,
      "step": 1753
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9759824305566552,
      "learning_rate": 1.5439246685656093e-05,
      "loss": 0.8537,
      "step": 1754
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9109292560626803,
      "learning_rate": 1.5434017114933082e-05,
      "loss": 0.8268,
      "step": 1755
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.946731715802161,
      "learning_rate": 1.5428785434537527e-05,
      "loss": 0.8832,
      "step": 1756
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9161769497649447,
      "learning_rate": 1.542355164650055e-05,
      "loss": 0.853,
      "step": 1757
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8924453740191084,
      "learning_rate": 1.541831575285408e-05,
      "loss": 0.829,
      "step": 1758
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9221581455088456,
      "learning_rate": 1.541307775563088e-05,
      "loss": 0.8756,
      "step": 1759
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9029272345444195,
      "learning_rate": 1.540783765686452e-05,
      "loss": 0.8359,
      "step": 1760
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.948910939848515,
      "learning_rate": 1.540259545858938e-05,
      "loss": 0.8868,
      "step": 1761
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0151757330963906,
      "learning_rate": 1.539735116284067e-05,
      "loss": 0.9,
      "step": 1762
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9879029796575579,
      "learning_rate": 1.53921047716544e-05,
      "loss": 0.7987,
      "step": 1763
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9158500626487664,
      "learning_rate": 1.53868562870674e-05,
      "loss": 0.7881,
      "step": 1764
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9454761680677035,
      "learning_rate": 1.5381605711117318e-05,
      "loss": 0.8526,
      "step": 1765
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6845205461694421,
      "learning_rate": 1.5376353045842604e-05,
      "loss": 0.3553,
      "step": 1766
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9857118057212051,
      "learning_rate": 1.5371098293282526e-05,
      "loss": 0.904,
      "step": 1767
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9109045030889309,
      "learning_rate": 1.5365841455477158e-05,
      "loss": 0.8703,
      "step": 1768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9749596405683608,
      "learning_rate": 1.5360582534467382e-05,
      "loss": 0.8324,
      "step": 1769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0662890424874079,
      "learning_rate": 1.5355321532294897e-05,
      "loss": 0.8299,
      "step": 1770
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5917609957521507,
      "learning_rate": 1.5350058451002204e-05,
      "loss": 0.3561,
      "step": 1771
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0380436192460951,
      "learning_rate": 1.5344793292632614e-05,
      "loss": 0.7959,
      "step": 1772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.957947617087189,
      "learning_rate": 1.533952605923024e-05,
      "loss": 0.8089,
      "step": 1773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8904400422217335,
      "learning_rate": 1.5334256752840007e-05,
      "loss": 0.8114,
      "step": 1774
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9331656620726193,
      "learning_rate": 1.532898537550764e-05,
      "loss": 0.8587,
      "step": 1775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9197415446693271,
      "learning_rate": 1.532371192927966e-05,
      "loss": 0.8352,
      "step": 1776
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.873866987477969,
      "learning_rate": 1.5318436416203412e-05,
      "loss": 0.7682,
      "step": 1777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8530070136496641,
      "learning_rate": 1.531315883832703e-05,
      "loss": 0.856,
      "step": 1778
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9193812267768797,
      "learning_rate": 1.530787919769945e-05,
      "loss": 0.7795,
      "step": 1779
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6746619514542871,
      "learning_rate": 1.5302597496370408e-05,
      "loss": 0.3771,
      "step": 1780
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9245727855304479,
      "learning_rate": 1.5297313736390447e-05,
      "loss": 0.8062,
      "step": 1781
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9073442717916587,
      "learning_rate": 1.5292027919810898e-05,
      "loss": 0.8525,
      "step": 1782
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9074286417004753,
      "learning_rate": 1.52867400486839e-05,
      "loss": 0.8097,
      "step": 1783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8989728024838403,
      "learning_rate": 1.528145012506239e-05,
      "loss": 0.8357,
      "step": 1784
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9977297428353041,
      "learning_rate": 1.5276158151000096e-05,
      "loss": 0.8523,
      "step": 1785
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6348609850853256,
      "learning_rate": 1.5270864128551542e-05,
      "loss": 0.3455,
      "step": 1786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8934830987373341,
      "learning_rate": 1.5265568059772053e-05,
      "loss": 0.8315,
      "step": 1787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8593120753757034,
      "learning_rate": 1.5260269946717746e-05,
      "loss": 0.8624,
      "step": 1788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9579923911170706,
      "learning_rate": 1.5254969791445526e-05,
      "loss": 0.8728,
      "step": 1789
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8498194455366958,
      "learning_rate": 1.5249667596013102e-05,
      "loss": 0.7932,
      "step": 1790
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9830882140468343,
      "learning_rate": 1.5244363362478967e-05,
      "loss": 0.8397,
      "step": 1791
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8747521145384063,
      "learning_rate": 1.5239057092902404e-05,
      "loss": 0.8539,
      "step": 1792
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9006508970276589,
      "learning_rate": 1.523374878934349e-05,
      "loss": 0.846,
      "step": 1793
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9251315924799437,
      "learning_rate": 1.5228438453863095e-05,
      "loss": 0.8499,
      "step": 1794
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.870936798434442,
      "learning_rate": 1.522312608852287e-05,
      "loss": 0.8319,
      "step": 1795
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9233172106024027,
      "learning_rate": 1.5217811695385263e-05,
      "loss": 0.8779,
      "step": 1796
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8598287725482912,
      "learning_rate": 1.52124952765135e-05,
      "loss": 0.8771,
      "step": 1797
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6335721153420865,
      "learning_rate": 1.5207176833971598e-05,
      "loss": 0.3755,
      "step": 1798
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6368214628243776,
      "learning_rate": 1.520185636982436e-05,
      "loss": 0.3503,
      "step": 1799
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8967100713272995,
      "learning_rate": 1.5196533886137376e-05,
      "loss": 0.8583,
      "step": 1800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9383755343219106,
      "learning_rate": 1.5191209384977014e-05,
      "loss": 0.831,
      "step": 1801
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9107845400751471,
      "learning_rate": 1.5185882868410431e-05,
      "loss": 0.8448,
      "step": 1802
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6515005590321009,
      "learning_rate": 1.5180554338505564e-05,
      "loss": 0.3509,
      "step": 1803
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9666347443577025,
      "learning_rate": 1.517522379733113e-05,
      "loss": 0.807,
      "step": 1804
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8658835855323628,
      "learning_rate": 1.5169891246956629e-05,
      "loss": 0.8292,
      "step": 1805
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8828545386033965,
      "learning_rate": 1.5164556689452346e-05,
      "loss": 0.8091,
      "step": 1806
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.871015685800264,
      "learning_rate": 1.5159220126889329e-05,
      "loss": 0.7853,
      "step": 1807
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9595036218950984,
      "learning_rate": 1.5153881561339426e-05,
      "loss": 0.8744,
      "step": 1808
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9159233207141578,
      "learning_rate": 1.5148540994875242e-05,
      "loss": 0.8285,
      "step": 1809
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9024958211462425,
      "learning_rate": 1.5143198429570181e-05,
      "loss": 0.8807,
      "step": 1810
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6502736828821828,
      "learning_rate": 1.5137853867498403e-05,
      "loss": 0.351,
      "step": 1811
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9296779748534861,
      "learning_rate": 1.5132507310734847e-05,
      "loss": 0.8709,
      "step": 1812
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9175904105591312,
      "learning_rate": 1.5127158761355241e-05,
      "loss": 0.9037,
      "step": 1813
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9240681421397033,
      "learning_rate": 1.512180822143607e-05,
      "loss": 0.8328,
      "step": 1814
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9028494110105718,
      "learning_rate": 1.5116455693054594e-05,
      "loss": 0.8504,
      "step": 1815
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9750954167603443,
      "learning_rate": 1.5111101178288858e-05,
      "loss": 0.8259,
      "step": 1816
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8833039303729264,
      "learning_rate": 1.510574467921766e-05,
      "loss": 0.8284,
      "step": 1817
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9117401358957243,
      "learning_rate": 1.5100386197920585e-05,
      "loss": 0.7711,
      "step": 1818
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8879802986569884,
      "learning_rate": 1.5095025736477977e-05,
      "loss": 0.8712,
      "step": 1819
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8600879351481723,
      "learning_rate": 1.5089663296970952e-05,
      "loss": 0.8165,
      "step": 1820
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.958570241732143,
      "learning_rate": 1.5084298881481388e-05,
      "loss": 0.8048,
      "step": 1821
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9655996294463675,
      "learning_rate": 1.5078932492091942e-05,
      "loss": 0.838,
      "step": 1822
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9345204793216182,
      "learning_rate": 1.5073564130886032e-05,
      "loss": 0.8781,
      "step": 1823
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9811374527818862,
      "learning_rate": 1.506819379994784e-05,
      "loss": 0.7485,
      "step": 1824
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8848499876027218,
      "learning_rate": 1.5062821501362308e-05,
      "loss": 0.8174,
      "step": 1825
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9305920686875305,
      "learning_rate": 1.5057447237215152e-05,
      "loss": 0.8829,
      "step": 1826
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9322810963001693,
      "learning_rate": 1.5052071009592846e-05,
      "loss": 0.8239,
      "step": 1827
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9361000933979026,
      "learning_rate": 1.5046692820582625e-05,
      "loss": 0.8553,
      "step": 1828
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8682291032802651,
      "learning_rate": 1.504131267227249e-05,
      "loss": 0.851,
      "step": 1829
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8893024531298159,
      "learning_rate": 1.5035930566751198e-05,
      "loss": 0.8908,
      "step": 1830
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8948667761907209,
      "learning_rate": 1.5030546506108268e-05,
      "loss": 0.89,
      "step": 1831
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8757957269978255,
      "learning_rate": 1.5025160492433976e-05,
      "loss": 0.817,
      "step": 1832
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9466176596768207,
      "learning_rate": 1.501977252781936e-05,
      "loss": 0.8485,
      "step": 1833
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9312382224877516,
      "learning_rate": 1.5014382614356213e-05,
      "loss": 0.8665,
      "step": 1834
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.975962314707832,
      "learning_rate": 1.5008990754137088e-05,
      "loss": 0.729,
      "step": 1835
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9397541486313047,
      "learning_rate": 1.5003596949255284e-05,
      "loss": 0.8054,
      "step": 1836
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8858544313774939,
      "learning_rate": 1.4998201201804867e-05,
      "loss": 0.8489,
      "step": 1837
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9371338798708746,
      "learning_rate": 1.499280351388065e-05,
      "loss": 0.8353,
      "step": 1838
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8808646886521502,
      "learning_rate": 1.49874038875782e-05,
      "loss": 0.8103,
      "step": 1839
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9295419748895162,
      "learning_rate": 1.498200232499384e-05,
      "loss": 0.8245,
      "step": 1840
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0371415773967052,
      "learning_rate": 1.4976598828224643e-05,
      "loss": 0.8658,
      "step": 1841
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8710559125239278,
      "learning_rate": 1.497119339936843e-05,
      "loss": 0.7736,
      "step": 1842
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9273890220171412,
      "learning_rate": 1.4965786040523779e-05,
      "loss": 0.8605,
      "step": 1843
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9344202997974553,
      "learning_rate": 1.496037675379001e-05,
      "loss": 0.8561,
      "step": 1844
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7536029370752119,
      "learning_rate": 1.4954965541267192e-05,
      "loss": 0.3303,
      "step": 1845
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0077521965463638,
      "learning_rate": 1.494955240505615e-05,
      "loss": 0.7944,
      "step": 1846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9763180597270974,
      "learning_rate": 1.494413734725844e-05,
      "loss": 0.814,
      "step": 1847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9316668681681081,
      "learning_rate": 1.4938720369976385e-05,
      "loss": 0.8298,
      "step": 1848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9569150408917994,
      "learning_rate": 1.4933301475313036e-05,
      "loss": 0.8676,
      "step": 1849
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9528370159812843,
      "learning_rate": 1.4927880665372197e-05,
      "loss": 0.9052,
      "step": 1850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9108912008593442,
      "learning_rate": 1.4922457942258411e-05,
      "loss": 0.8403,
      "step": 1851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.946056800029388,
      "learning_rate": 1.4917033308076967e-05,
      "loss": 0.8642,
      "step": 1852
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0145538193485275,
      "learning_rate": 1.4911606764933892e-05,
      "loss": 0.8804,
      "step": 1853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8499611353820287,
      "learning_rate": 1.490617831493596e-05,
      "loss": 0.8552,
      "step": 1854
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8441235956405478,
      "learning_rate": 1.4900747960190682e-05,
      "loss": 0.8202,
      "step": 1855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9005926598809061,
      "learning_rate": 1.489531570280631e-05,
      "loss": 0.8767,
      "step": 1856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9242007729332997,
      "learning_rate": 1.488988154489183e-05,
      "loss": 0.7922,
      "step": 1857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9228296910575406,
      "learning_rate": 1.4884445488556972e-05,
      "loss": 0.8149,
      "step": 1858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8964125420322703,
      "learning_rate": 1.4879007535912198e-05,
      "loss": 0.8633,
      "step": 1859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7180197280234195,
      "learning_rate": 1.4873567689068708e-05,
      "loss": 0.3472,
      "step": 1860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8395972538183191,
      "learning_rate": 1.4868125950138442e-05,
      "loss": 0.8631,
      "step": 1861
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8876020396417427,
      "learning_rate": 1.4862682321234064e-05,
      "loss": 0.8394,
      "step": 1862
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8359938487219355,
      "learning_rate": 1.4857236804468983e-05,
      "loss": 0.8759,
      "step": 1863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8760612263745683,
      "learning_rate": 1.4851789401957338e-05,
      "loss": 0.81,
      "step": 1864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9115704232218258,
      "learning_rate": 1.4846340115813993e-05,
      "loss": 0.8049,
      "step": 1865
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9262757106317488,
      "learning_rate": 1.484088894815455e-05,
      "loss": 0.8585,
      "step": 1866
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9658383700675881,
      "learning_rate": 1.4835435901095341e-05,
      "loss": 0.9151,
      "step": 1867
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6784647326937336,
      "learning_rate": 1.4829980976753426e-05,
      "loss": 0.3584,
      "step": 1868
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8087183934296753,
      "learning_rate": 1.4824524177246597e-05,
      "loss": 0.8469,
      "step": 1869
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9301601268422226,
      "learning_rate": 1.4819065504693365e-05,
      "loss": 0.7787,
      "step": 1870
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9225403669730645,
      "learning_rate": 1.4813604961212984e-05,
      "loss": 0.8959,
      "step": 1871
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9033379733446967,
      "learning_rate": 1.4808142548925417e-05,
      "loss": 0.8918,
      "step": 1872
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9469403590862796,
      "learning_rate": 1.4802678269951365e-05,
      "loss": 0.8457,
      "step": 1873
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9696760253280345,
      "learning_rate": 1.4797212126412243e-05,
      "loss": 0.8437,
      "step": 1874
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9335528797716516,
      "learning_rate": 1.4791744120430202e-05,
      "loss": 0.7956,
      "step": 1875
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9098264535167683,
      "learning_rate": 1.4786274254128112e-05,
      "loss": 0.8817,
      "step": 1876
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9162566177066078,
      "learning_rate": 1.4780802529629559e-05,
      "loss": 0.8389,
      "step": 1877
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9583471157420339,
      "learning_rate": 1.4775328949058856e-05,
      "loss": 0.8422,
      "step": 1878
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9527335954765002,
      "learning_rate": 1.4769853514541037e-05,
      "loss": 0.9189,
      "step": 1879
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.921785717010861,
      "learning_rate": 1.4764376228201848e-05,
      "loss": 0.7965,
      "step": 1880
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9214426056803401,
      "learning_rate": 1.475889709216777e-05,
      "loss": 0.8264,
      "step": 1881
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8791886140326173,
      "learning_rate": 1.4753416108565985e-05,
      "loss": 0.8858,
      "step": 1882
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9260768745325552,
      "learning_rate": 1.47479332795244e-05,
      "loss": 0.8475,
      "step": 1883
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9810659996962456,
      "learning_rate": 1.4742448607171644e-05,
      "loss": 0.882,
      "step": 1884
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6662044696150002,
      "learning_rate": 1.473696209363705e-05,
      "loss": 0.3858,
      "step": 1885
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9387119069574016,
      "learning_rate": 1.4731473741050673e-05,
      "loss": 0.8298,
      "step": 1886
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9764975265107826,
      "learning_rate": 1.4725983551543279e-05,
      "loss": 0.8262,
      "step": 1887
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.932424906436865,
      "learning_rate": 1.472049152724635e-05,
      "loss": 0.8288,
      "step": 1888
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9005514087192231,
      "learning_rate": 1.471499767029208e-05,
      "loss": 0.8941,
      "step": 1889
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9622131687436255,
      "learning_rate": 1.470950198281337e-05,
      "loss": 0.8752,
      "step": 1890
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9958825217005961,
      "learning_rate": 1.470400446694384e-05,
      "loss": 0.8018,
      "step": 1891
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.010010125702113,
      "learning_rate": 1.4698505124817811e-05,
      "loss": 0.8238,
      "step": 1892
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9235846975506805,
      "learning_rate": 1.4693003958570318e-05,
      "loss": 0.8183,
      "step": 1893
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6698502124821001,
      "learning_rate": 1.4687500970337103e-05,
      "loss": 0.3411,
      "step": 1894
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9134571924596682,
      "learning_rate": 1.4681996162254618e-05,
      "loss": 0.8722,
      "step": 1895
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9263451414161206,
      "learning_rate": 1.4676489536460015e-05,
      "loss": 0.9047,
      "step": 1896
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9144015318976983,
      "learning_rate": 1.467098109509116e-05,
      "loss": 0.8937,
      "step": 1897
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8507542684487138,
      "learning_rate": 1.4665470840286614e-05,
      "loss": 0.8264,
      "step": 1898
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9538816146770269,
      "learning_rate": 1.4659958774185654e-05,
      "loss": 0.8626,
      "step": 1899
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7979151475437121,
      "learning_rate": 1.4654444898928249e-05,
      "loss": 0.8418,
      "step": 1900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8610524046568713,
      "learning_rate": 1.4648929216655077e-05,
      "loss": 0.8809,
      "step": 1901
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8823657106556122,
      "learning_rate": 1.4643411729507517e-05,
      "loss": 0.8696,
      "step": 1902
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.858443181588155,
      "learning_rate": 1.4637892439627644e-05,
      "loss": 0.8516,
      "step": 1903
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9764252680676682,
      "learning_rate": 1.4632371349158241e-05,
      "loss": 0.9016,
      "step": 1904
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9253674923771901,
      "learning_rate": 1.4626848460242782e-05,
      "loss": 0.868,
      "step": 1905
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8644581479035671,
      "learning_rate": 1.4621323775025444e-05,
      "loss": 0.8236,
      "step": 1906
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9478669143410589,
      "learning_rate": 1.4615797295651099e-05,
      "loss": 0.8351,
      "step": 1907
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9361414436107189,
      "learning_rate": 1.4610269024265317e-05,
      "loss": 0.8836,
      "step": 1908
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9254491811305535,
      "learning_rate": 1.4604738963014365e-05,
      "loss": 0.818,
      "step": 1909
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6941995332491423,
      "learning_rate": 1.4599207114045202e-05,
      "loss": 0.3855,
      "step": 1910
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8651615657375262,
      "learning_rate": 1.4593673479505482e-05,
      "loss": 0.8522,
      "step": 1911
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8227519128701712,
      "learning_rate": 1.4588138061543551e-05,
      "loss": 0.8388,
      "step": 1912
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9200072417214321,
      "learning_rate": 1.458260086230845e-05,
      "loss": 0.9133,
      "step": 1913
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5704729902772816,
      "learning_rate": 1.4577061883949912e-05,
      "loss": 0.3338,
      "step": 1914
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8697042648085789,
      "learning_rate": 1.4571521128618358e-05,
      "loss": 0.8039,
      "step": 1915
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9315820739049158,
      "learning_rate": 1.4565978598464895e-05,
      "loss": 0.8604,
      "step": 1916
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9389323966896426,
      "learning_rate": 1.4560434295641338e-05,
      "loss": 0.8311,
      "step": 1917
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9315739873730015,
      "learning_rate": 1.455488822230016e-05,
      "loss": 0.7925,
      "step": 1918
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.95464838877711,
      "learning_rate": 1.4549340380594545e-05,
      "loss": 0.8185,
      "step": 1919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8672806767208509,
      "learning_rate": 1.454379077267836e-05,
      "loss": 0.8368,
      "step": 1920
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9241013762954439,
      "learning_rate": 1.4538239400706147e-05,
      "loss": 0.838,
      "step": 1921
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9551547519245643,
      "learning_rate": 1.4532686266833143e-05,
      "loss": 0.8598,
      "step": 1922
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7125674550563859,
      "learning_rate": 1.4527131373215265e-05,
      "loss": 0.3532,
      "step": 1923
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9739595197944078,
      "learning_rate": 1.4521574722009115e-05,
      "loss": 0.8386,
      "step": 1924
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5967574252326919,
      "learning_rate": 1.4516016315371974e-05,
      "loss": 0.3265,
      "step": 1925
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9243422913692647,
      "learning_rate": 1.4510456155461807e-05,
      "loss": 0.8816,
      "step": 1926
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8553447658037779,
      "learning_rate": 1.4504894244437264e-05,
      "loss": 0.8131,
      "step": 1927
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0183887392221476,
      "learning_rate": 1.4499330584457667e-05,
      "loss": 0.8646,
      "step": 1928
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9039948344715546,
      "learning_rate": 1.4493765177683017e-05,
      "loss": 0.8393,
      "step": 1929
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8529814239418307,
      "learning_rate": 1.4488198026274007e-05,
      "loss": 0.8738,
      "step": 1930
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8978504783405167,
      "learning_rate": 1.4482629132391985e-05,
      "loss": 0.8818,
      "step": 1931
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9301720714244804,
      "learning_rate": 1.4477058498198993e-05,
      "loss": 0.8611,
      "step": 1932
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9788440478702308,
      "learning_rate": 1.4471486125857743e-05,
      "loss": 0.3554,
      "step": 1933
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8690600578866797,
      "learning_rate": 1.446591201753162e-05,
      "loss": 0.7956,
      "step": 1934
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8867848224379985,
      "learning_rate": 1.4460336175384688e-05,
      "loss": 0.8692,
      "step": 1935
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.952698310384124,
      "learning_rate": 1.4454758601581675e-05,
      "loss": 0.8409,
      "step": 1936
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8799079892475987,
      "learning_rate": 1.4449179298287999e-05,
      "loss": 0.8608,
      "step": 1937
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9268570615435013,
      "learning_rate": 1.4443598267669723e-05,
      "loss": 0.8092,
      "step": 1938
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9558579424608453,
      "learning_rate": 1.4438015511893602e-05,
      "loss": 0.7995,
      "step": 1939
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9531919927777873,
      "learning_rate": 1.4432431033127056e-05,
      "loss": 0.9062,
      "step": 1940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9229030743608305,
      "learning_rate": 1.442684483353817e-05,
      "loss": 0.8418,
      "step": 1941
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8367214803390673,
      "learning_rate": 1.4421256915295697e-05,
      "loss": 0.748,
      "step": 1942
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9223254542119035,
      "learning_rate": 1.4415667280569064e-05,
      "loss": 0.7876,
      "step": 1943
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8821116830330472,
      "learning_rate": 1.4410075931528356e-05,
      "loss": 0.8714,
      "step": 1944
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9170443987937573,
      "learning_rate": 1.4404482870344322e-05,
      "loss": 0.8475,
      "step": 1945
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.943908765899527,
      "learning_rate": 1.4398888099188396e-05,
      "loss": 0.8972,
      "step": 1946
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9127056862563817,
      "learning_rate": 1.4393291620232646e-05,
      "loss": 0.8252,
      "step": 1947
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9171273295185493,
      "learning_rate": 1.4387693435649826e-05,
      "loss": 0.8437,
      "step": 1948
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9334718000347808,
      "learning_rate": 1.4382093547613338e-05,
      "loss": 0.8761,
      "step": 1949
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0072172198695404,
      "learning_rate": 1.4376491958297263e-05,
      "loss": 0.8041,
      "step": 1950
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9389920680790379,
      "learning_rate": 1.4370888669876317e-05,
      "loss": 0.8883,
      "step": 1951
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.918805051749973,
      "learning_rate": 1.4365283684525895e-05,
      "loss": 0.8496,
      "step": 1952
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8822777769662488,
      "learning_rate": 1.4359677004422045e-05,
      "loss": 0.7904,
      "step": 1953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8914287763740717,
      "learning_rate": 1.4354068631741476e-05,
      "loss": 0.8516,
      "step": 1954
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9391197085806189,
      "learning_rate": 1.4348458568661548e-05,
      "loss": 0.9238,
      "step": 1955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9212150661479673,
      "learning_rate": 1.434284681736028e-05,
      "loss": 0.8754,
      "step": 1956
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9429941721790444,
      "learning_rate": 1.4337233380016354e-05,
      "loss": 0.8879,
      "step": 1957
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8807826696015281,
      "learning_rate": 1.433161825880909e-05,
      "loss": 0.8383,
      "step": 1958
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9000818153287816,
      "learning_rate": 1.432600145591848e-05,
      "loss": 0.8495,
      "step": 1959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7542343824572151,
      "learning_rate": 1.4320382973525151e-05,
      "loss": 0.3628,
      "step": 1960
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8718276018511801,
      "learning_rate": 1.43147628138104e-05,
      "loss": 0.846,
      "step": 1961
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9146490193215882,
      "learning_rate": 1.4309140978956161e-05,
      "loss": 0.8854,
      "step": 1962
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9102450958517628,
      "learning_rate": 1.430351747114503e-05,
      "loss": 0.8633,
      "step": 1963
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9395831429778632,
      "learning_rate": 1.429789229256024e-05,
      "loss": 0.7907,
      "step": 1964
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6233636944074703,
      "learning_rate": 1.429226544538568e-05,
      "loss": 0.2781,
      "step": 1965
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.61996243659808,
      "learning_rate": 1.4286636931805887e-05,
      "loss": 0.3004,
      "step": 1966
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.87042626049094,
      "learning_rate": 1.4281006754006045e-05,
      "loss": 0.8428,
      "step": 1967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8567422586648958,
      "learning_rate": 1.427537491417198e-05,
      "loss": 0.8447,
      "step": 1968
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8386373599190023,
      "learning_rate": 1.426974141449017e-05,
      "loss": 0.7783,
      "step": 1969
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.915459994318423,
      "learning_rate": 1.4264106257147732e-05,
      "loss": 0.853,
      "step": 1970
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9407377963836193,
      "learning_rate": 1.4258469444332423e-05,
      "loss": 0.8339,
      "step": 1971
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.899816004139802,
      "learning_rate": 1.4252830978232658e-05,
      "loss": 0.9051,
      "step": 1972
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8644635215658549,
      "learning_rate": 1.4247190861037474e-05,
      "loss": 0.8065,
      "step": 1973
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9656993218363943,
      "learning_rate": 1.4241549094936567e-05,
      "loss": 0.868,
      "step": 1974
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8612824028340191,
      "learning_rate": 1.4235905682120255e-05,
      "loss": 0.8172,
      "step": 1975
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9466239868690243,
      "learning_rate": 1.4230260624779512e-05,
      "loss": 0.8422,
      "step": 1976
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9060858700423675,
      "learning_rate": 1.4224613925105947e-05,
      "loss": 0.7904,
      "step": 1977
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.893128790613575,
      "learning_rate": 1.4218965585291792e-05,
      "loss": 0.8529,
      "step": 1978
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9072678491654574,
      "learning_rate": 1.4213315607529939e-05,
      "loss": 0.838,
      "step": 1979
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9499851879783491,
      "learning_rate": 1.4207663994013896e-05,
      "loss": 0.8134,
      "step": 1980
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.88902786402998,
      "learning_rate": 1.4202010746937815e-05,
      "loss": 0.8567,
      "step": 1981
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9413954587769024,
      "learning_rate": 1.4196355868496485e-05,
      "loss": 0.8793,
      "step": 1982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8978284341788665,
      "learning_rate": 1.4190699360885323e-05,
      "loss": 0.8524,
      "step": 1983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8917206857516385,
      "learning_rate": 1.4185041226300376e-05,
      "loss": 0.7756,
      "step": 1984
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8786612518891446,
      "learning_rate": 1.4179381466938332e-05,
      "loss": 0.8058,
      "step": 1985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9674952948883856,
      "learning_rate": 1.4173720084996501e-05,
      "loss": 0.8172,
      "step": 1986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9444449305383978,
      "learning_rate": 1.4168057082672828e-05,
      "loss": 0.918,
      "step": 1987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.874601267697836,
      "learning_rate": 1.4162392462165884e-05,
      "loss": 0.8582,
      "step": 1988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8792918076823568,
      "learning_rate": 1.4156726225674874e-05,
      "loss": 0.8057,
      "step": 1989
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.92149274207962,
      "learning_rate": 1.415105837539962e-05,
      "loss": 0.8504,
      "step": 1990
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8889781746299023,
      "learning_rate": 1.414538891354058e-05,
      "loss": 0.8371,
      "step": 1991
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8991292034876013,
      "learning_rate": 1.4139717842298835e-05,
      "loss": 0.8546,
      "step": 1992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8782259528200983,
      "learning_rate": 1.4134045163876086e-05,
      "loss": 0.8731,
      "step": 1993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9490398304095193,
      "learning_rate": 1.4128370880474667e-05,
      "loss": 0.843,
      "step": 1994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8820937084619873,
      "learning_rate": 1.412269499429753e-05,
      "loss": 0.8419,
      "step": 1995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9493849058486733,
      "learning_rate": 1.4117017507548244e-05,
      "loss": 0.8542,
      "step": 1996
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9825252945170443,
      "learning_rate": 1.4111338422431013e-05,
      "loss": 0.857,
      "step": 1997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.981505514784676,
      "learning_rate": 1.4105657741150648e-05,
      "loss": 0.7696,
      "step": 1998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.915162154796651,
      "learning_rate": 1.4099975465912584e-05,
      "loss": 0.8602,
      "step": 1999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8667836697495711,
      "learning_rate": 1.4094291598922877e-05,
      "loss": 0.8527,
      "step": 2000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9178462763512861,
      "learning_rate": 1.40886061423882e-05,
      "loss": 0.8362,
      "step": 2001
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9106445348294747,
      "learning_rate": 1.4082919098515846e-05,
      "loss": 0.8319,
      "step": 2002
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9052161871966199,
      "learning_rate": 1.407723046951372e-05,
      "loss": 0.773,
      "step": 2003
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9097225395006232,
      "learning_rate": 1.4071540257590341e-05,
      "loss": 0.8413,
      "step": 2004
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9015904096701783,
      "learning_rate": 1.4065848464954848e-05,
      "loss": 0.8776,
      "step": 2005
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8699943608847669,
      "learning_rate": 1.4060155093816988e-05,
      "loss": 0.8016,
      "step": 2006
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0832549770657776,
      "learning_rate": 1.4054460146387124e-05,
      "loss": 0.3479,
      "step": 2007
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9055339328875904,
      "learning_rate": 1.4048763624876233e-05,
      "loss": 0.7758,
      "step": 2008
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0122417199263003,
      "learning_rate": 1.4043065531495904e-05,
      "loss": 0.8268,
      "step": 2009
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.973823789650218,
      "learning_rate": 1.4037365868458325e-05,
      "loss": 0.8092,
      "step": 2010
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8286505043399948,
      "learning_rate": 1.4031664637976305e-05,
      "loss": 0.8161,
      "step": 2011
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6487522364999441,
      "learning_rate": 1.402596184226326e-05,
      "loss": 0.3537,
      "step": 2012
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9051456503967307,
      "learning_rate": 1.4020257483533208e-05,
      "loss": 0.8503,
      "step": 2013
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0167848003233255,
      "learning_rate": 1.401455156400078e-05,
      "loss": 0.8411,
      "step": 2014
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8859794573860409,
      "learning_rate": 1.400884408588121e-05,
      "loss": 0.8325,
      "step": 2015
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.907210876590499,
      "learning_rate": 1.400313505139034e-05,
      "loss": 0.8682,
      "step": 2016
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8010320087660571,
      "learning_rate": 1.3997424462744607e-05,
      "loss": 0.8397,
      "step": 2017
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9479719313610894,
      "learning_rate": 1.3991712322161065e-05,
      "loss": 0.8727,
      "step": 2018
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9530254905916028,
      "learning_rate": 1.3985998631857359e-05,
      "loss": 0.8362,
      "step": 2019
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9050902025080043,
      "learning_rate": 1.398028339405174e-05,
      "loss": 0.8669,
      "step": 2020
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7567838361711046,
      "learning_rate": 1.3974566610963068e-05,
      "loss": 0.3658,
      "step": 2021
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.002927106911553,
      "learning_rate": 1.3968848284810785e-05,
      "loss": 0.8129,
      "step": 2022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9420462592917253,
      "learning_rate": 1.3963128417814951e-05,
      "loss": 0.8167,
      "step": 2023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9974594888551499,
      "learning_rate": 1.3957407012196204e-05,
      "loss": 0.9196,
      "step": 2024
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0076955973066963,
      "learning_rate": 1.3951684070175802e-05,
      "loss": 0.8923,
      "step": 2025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9676084267087733,
      "learning_rate": 1.3945959593975582e-05,
      "loss": 0.8783,
      "step": 2026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8478418183629071,
      "learning_rate": 1.3940233585817984e-05,
      "loss": 0.8247,
      "step": 2027
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8652095060509584,
      "learning_rate": 1.3934506047926042e-05,
      "loss": 0.7978,
      "step": 2028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9171601890345387,
      "learning_rate": 1.3928776982523384e-05,
      "loss": 0.867,
      "step": 2029
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9318791132698803,
      "learning_rate": 1.3923046391834229e-05,
      "loss": 0.8647,
      "step": 2030
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8606640838619062,
      "learning_rate": 1.3917314278083391e-05,
      "loss": 0.7663,
      "step": 2031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6441653312912322,
      "learning_rate": 1.3911580643496272e-05,
      "loss": 0.3645,
      "step": 2032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9112368487996431,
      "learning_rate": 1.3905845490298867e-05,
      "loss": 0.7829,
      "step": 2033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9412589673967356,
      "learning_rate": 1.390010882071776e-05,
      "loss": 0.8799,
      "step": 2034
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.640191916727278,
      "learning_rate": 1.3894370636980128e-05,
      "loss": 0.3624,
      "step": 2035
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8755253692468701,
      "learning_rate": 1.3888630941313728e-05,
      "loss": 0.8227,
      "step": 2036
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9895200054762981,
      "learning_rate": 1.3882889735946901e-05,
      "loss": 0.908,
      "step": 2037
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6693950423903552,
      "learning_rate": 1.3877147023108592e-05,
      "loss": 0.3223,
      "step": 2038
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8863945362505439,
      "learning_rate": 1.3871402805028314e-05,
      "loss": 0.8897,
      "step": 2039
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6308681364528729,
      "learning_rate": 1.3865657083936167e-05,
      "loss": 0.3526,
      "step": 2040
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9586522815092596,
      "learning_rate": 1.3859909862062844e-05,
      "loss": 0.8691,
      "step": 2041
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8796339506053823,
      "learning_rate": 1.385416114163961e-05,
      "loss": 0.8762,
      "step": 2042
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9814711525822407,
      "learning_rate": 1.3848410924898321e-05,
      "loss": 0.8055,
      "step": 2043
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9276132223428695,
      "learning_rate": 1.3842659214071406e-05,
      "loss": 0.8152,
      "step": 2044
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.900305236586342,
      "learning_rate": 1.3836906011391878e-05,
      "loss": 0.8832,
      "step": 2045
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8576558122264133,
      "learning_rate": 1.3831151319093323e-05,
      "loss": 0.8544,
      "step": 2046
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9005762351611063,
      "learning_rate": 1.382539513940992e-05,
      "loss": 0.7916,
      "step": 2047
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0082953284349392,
      "learning_rate": 1.3819637474576411e-05,
      "loss": 0.88,
      "step": 2048
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8734128914705562,
      "learning_rate": 1.381387832682812e-05,
      "loss": 0.7992,
      "step": 2049
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8795024823042235,
      "learning_rate": 1.380811769840095e-05,
      "loss": 0.821,
      "step": 2050
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9376278336860242,
      "learning_rate": 1.3802355591531366e-05,
      "loss": 0.8467,
      "step": 2051
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9436857287628149,
      "learning_rate": 1.3796592008456427e-05,
      "loss": 0.8115,
      "step": 2052
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6760160565033244,
      "learning_rate": 1.3790826951413747e-05,
      "loss": 0.3548,
      "step": 2053
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9311483786418009,
      "learning_rate": 1.3785060422641526e-05,
      "loss": 0.8043,
      "step": 2054
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9256512514419022,
      "learning_rate": 1.3779292424378521e-05,
      "loss": 0.8169,
      "step": 2055
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9177827321954057,
      "learning_rate": 1.3773522958864076e-05,
      "loss": 0.8008,
      "step": 2056
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9352855957121892,
      "learning_rate": 1.3767752028338091e-05,
      "loss": 0.8384,
      "step": 2057
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9676237785734642,
      "learning_rate": 1.376197963504104e-05,
      "loss": 0.9023,
      "step": 2058
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9021416058140242,
      "learning_rate": 1.3756205781213965e-05,
      "loss": 0.8147,
      "step": 2059
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9975322157639436,
      "learning_rate": 1.375043046909848e-05,
      "loss": 0.8099,
      "step": 2060
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9638406615583308,
      "learning_rate": 1.3744653700936752e-05,
      "loss": 0.8438,
      "step": 2061
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0101458356642041,
      "learning_rate": 1.3738875478971526e-05,
      "loss": 0.789,
      "step": 2062
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8992264985134729,
      "learning_rate": 1.3733095805446107e-05,
      "loss": 0.9198,
      "step": 2063
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9379110016529751,
      "learning_rate": 1.372731468260436e-05,
      "loss": 0.8021,
      "step": 2064
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9321017182672903,
      "learning_rate": 1.372153211269072e-05,
      "loss": 0.8453,
      "step": 2065
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8897345316220132,
      "learning_rate": 1.3715748097950176e-05,
      "loss": 0.8257,
      "step": 2066
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9309637796943091,
      "learning_rate": 1.3709962640628284e-05,
      "loss": 0.8658,
      "step": 2067
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9644736949060442,
      "learning_rate": 1.3704175742971158e-05,
      "loss": 0.8578,
      "step": 2068
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9366767336938698,
      "learning_rate": 1.369838740722547e-05,
      "loss": 0.8047,
      "step": 2069
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8626594313284064,
      "learning_rate": 1.3692597635638452e-05,
      "loss": 0.7759,
      "step": 2070
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9534494896120499,
      "learning_rate": 1.368680643045789e-05,
      "loss": 0.8031,
      "step": 2071
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.901035059858045,
      "learning_rate": 1.3681013793932132e-05,
      "loss": 0.8203,
      "step": 2072
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.967083828418263,
      "learning_rate": 1.3675219728310076e-05,
      "loss": 0.8227,
      "step": 2073
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9184528205730447,
      "learning_rate": 1.3669424235841185e-05,
      "loss": 0.8298,
      "step": 2074
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9446609185914175,
      "learning_rate": 1.3663627318775459e-05,
      "loss": 0.8391,
      "step": 2075
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9228408238881118,
      "learning_rate": 1.3657828979363468e-05,
      "loss": 0.8556,
      "step": 2076
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8825079889345769,
      "learning_rate": 1.3652029219856324e-05,
      "loss": 0.8571,
      "step": 2077
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6836854951464316,
      "learning_rate": 1.3646228042505694e-05,
      "loss": 0.3817,
      "step": 2078
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8796275212525831,
      "learning_rate": 1.3640425449563793e-05,
      "loss": 0.8019,
      "step": 2079
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8915882759838223,
      "learning_rate": 1.3634621443283389e-05,
      "loss": 0.8445,
      "step": 2080
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9387134395462204,
      "learning_rate": 1.36288160259178e-05,
      "loss": 0.8347,
      "step": 2081
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9784215474541588,
      "learning_rate": 1.3623009199720882e-05,
      "loss": 0.8373,
      "step": 2082
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9044560983311913,
      "learning_rate": 1.3617200966947053e-05,
      "loss": 0.8705,
      "step": 2083
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8512930194856594,
      "learning_rate": 1.3611391329851262e-05,
      "loss": 0.8702,
      "step": 2084
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9033223327407928,
      "learning_rate": 1.3605580290689013e-05,
      "loss": 0.8352,
      "step": 2085
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9134922314466967,
      "learning_rate": 1.3599767851716353e-05,
      "loss": 0.8805,
      "step": 2086
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9316009289603682,
      "learning_rate": 1.3593954015189867e-05,
      "loss": 0.8499,
      "step": 2087
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8867876632889755,
      "learning_rate": 1.3588138783366692e-05,
      "loss": 0.8545,
      "step": 2088
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0362419542604158,
      "learning_rate": 1.3582322158504495e-05,
      "loss": 0.8197,
      "step": 2089
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8760110121111988,
      "learning_rate": 1.3576504142861496e-05,
      "loss": 0.8334,
      "step": 2090
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9195046325250839,
      "learning_rate": 1.3570684738696444e-05,
      "loss": 0.8891,
      "step": 2091
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8808883074916823,
      "learning_rate": 1.3564863948268631e-05,
      "loss": 0.7974,
      "step": 2092
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.932711830504739,
      "learning_rate": 1.3559041773837898e-05,
      "loss": 0.9133,
      "step": 2093
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8559617324216907,
      "learning_rate": 1.3553218217664603e-05,
      "loss": 0.8167,
      "step": 2094
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.91348841971254,
      "learning_rate": 1.3547393282009656e-05,
      "loss": 0.8456,
      "step": 2095
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8589329674561517,
      "learning_rate": 1.3541566969134496e-05,
      "loss": 0.7502,
      "step": 2096
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8903198219958853,
      "learning_rate": 1.3535739281301102e-05,
      "loss": 0.8602,
      "step": 2097
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8568085121366,
      "learning_rate": 1.3529910220771975e-05,
      "loss": 0.7843,
      "step": 2098
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8586703877071293,
      "learning_rate": 1.3524079789810163e-05,
      "loss": 0.7875,
      "step": 2099
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9292435981247513,
      "learning_rate": 1.3518247990679241e-05,
      "loss": 0.7943,
      "step": 2100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8638828734511771,
      "learning_rate": 1.3512414825643312e-05,
      "loss": 0.8634,
      "step": 2101
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9673952353423241,
      "learning_rate": 1.3506580296967011e-05,
      "loss": 0.8152,
      "step": 2102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8696079006635374,
      "learning_rate": 1.3500744406915505e-05,
      "loss": 0.7838,
      "step": 2103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8960367048035516,
      "learning_rate": 1.3494907157754485e-05,
      "loss": 0.8633,
      "step": 2104
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9707624031873656,
      "learning_rate": 1.348906855175017e-05,
      "loss": 0.8155,
      "step": 2105
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8575011043775633,
      "learning_rate": 1.3483228591169315e-05,
      "loss": 0.8365,
      "step": 2106
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8783090920481267,
      "learning_rate": 1.347738727827919e-05,
      "loss": 0.7768,
      "step": 2107
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9778995430416199,
      "learning_rate": 1.3471544615347591e-05,
      "loss": 0.7901,
      "step": 2108
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9173891242561194,
      "learning_rate": 1.3465700604642847e-05,
      "loss": 0.8292,
      "step": 2109
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9665450247341244,
      "learning_rate": 1.34598552484338e-05,
      "loss": 0.8498,
      "step": 2110
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6578151347465485,
      "learning_rate": 1.3454008548989816e-05,
      "loss": 0.3328,
      "step": 2111
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9439427976122899,
      "learning_rate": 1.3448160508580789e-05,
      "loss": 0.7919,
      "step": 2112
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9633581945461809,
      "learning_rate": 1.3442311129477133e-05,
      "loss": 0.8999,
      "step": 2113
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6537624972134877,
      "learning_rate": 1.343646041394977e-05,
      "loss": 0.3325,
      "step": 2114
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9698733658514711,
      "learning_rate": 1.3430608364270156e-05,
      "loss": 0.8507,
      "step": 2115
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8965900798781624,
      "learning_rate": 1.3424754982710256e-05,
      "loss": 0.887,
      "step": 2116
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8482232270052977,
      "learning_rate": 1.3418900271542552e-05,
      "loss": 0.8201,
      "step": 2117
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8821826701470357,
      "learning_rate": 1.3413044233040045e-05,
      "loss": 0.8623,
      "step": 2118
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9584074134826128,
      "learning_rate": 1.3407186869476253e-05,
      "loss": 0.84,
      "step": 2119
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8231912303858337,
      "learning_rate": 1.3401328183125208e-05,
      "loss": 0.7827,
      "step": 2120
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8479658549701137,
      "learning_rate": 1.339546817626145e-05,
      "loss": 0.8356,
      "step": 2121
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.946978343458939,
      "learning_rate": 1.3389606851160037e-05,
      "loss": 0.843,
      "step": 2122
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9775916994273204,
      "learning_rate": 1.3383744210096537e-05,
      "loss": 0.8443,
      "step": 2123
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8877097432639058,
      "learning_rate": 1.3377880255347026e-05,
      "loss": 0.8487,
      "step": 2124
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8704148020642273,
      "learning_rate": 1.3372014989188098e-05,
      "loss": 0.8606,
      "step": 2125
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9716709438486486,
      "learning_rate": 1.3366148413896851e-05,
      "loss": 0.7935,
      "step": 2126
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8808588080288063,
      "learning_rate": 1.3360280531750886e-05,
      "loss": 0.8139,
      "step": 2127
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9508482606019398,
      "learning_rate": 1.3354411345028324e-05,
      "loss": 0.951,
      "step": 2128
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8276898886050423,
      "learning_rate": 1.3348540856007782e-05,
      "loss": 0.8865,
      "step": 2129
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8539946786241285,
      "learning_rate": 1.3342669066968385e-05,
      "loss": 0.8267,
      "step": 2130
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8775057046021246,
      "learning_rate": 1.3336795980189763e-05,
      "loss": 0.8475,
      "step": 2131
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.833951573932897,
      "learning_rate": 1.3330921597952056e-05,
      "loss": 0.8305,
      "step": 2132
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8662937184154199,
      "learning_rate": 1.3325045922535896e-05,
      "loss": 0.82,
      "step": 2133
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8303251510386033,
      "learning_rate": 1.3319168956222423e-05,
      "loss": 0.8659,
      "step": 2134
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9979167876081959,
      "learning_rate": 1.331329070129328e-05,
      "loss": 0.8238,
      "step": 2135
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9312871106325198,
      "learning_rate": 1.3307411160030608e-05,
      "loss": 0.8391,
      "step": 2136
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9026161787363646,
      "learning_rate": 1.3301530334717046e-05,
      "loss": 0.8527,
      "step": 2137
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9053079281926635,
      "learning_rate": 1.3295648227635729e-05,
      "loss": 0.8836,
      "step": 2138
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8771991267875974,
      "learning_rate": 1.32897648410703e-05,
      "loss": 0.8198,
      "step": 2139
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9246855755245613,
      "learning_rate": 1.328388017730489e-05,
      "loss": 0.8378,
      "step": 2140
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8941377754091024,
      "learning_rate": 1.327799423862413e-05,
      "loss": 0.8146,
      "step": 2141
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.64500144439335,
      "learning_rate": 1.3272107027313142e-05,
      "loss": 0.3328,
      "step": 2142
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9130899155117017,
      "learning_rate": 1.3266218545657541e-05,
      "loss": 0.8534,
      "step": 2143
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9497898470213203,
      "learning_rate": 1.326032879594344e-05,
      "loss": 0.8704,
      "step": 2144
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9028703445219217,
      "learning_rate": 1.3254437780457448e-05,
      "loss": 0.8092,
      "step": 2145
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9273774140421257,
      "learning_rate": 1.3248545501486654e-05,
      "loss": 0.8451,
      "step": 2146
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9255614971800743,
      "learning_rate": 1.3242651961318646e-05,
      "loss": 0.856,
      "step": 2147
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.908409815387708,
      "learning_rate": 1.32367571622415e-05,
      "loss": 0.7987,
      "step": 2148
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8943328416145526,
      "learning_rate": 1.3230861106543777e-05,
      "loss": 0.8953,
      "step": 2149
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9147903537967061,
      "learning_rate": 1.3224963796514532e-05,
      "loss": 0.8673,
      "step": 2150
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9276415103311112,
      "learning_rate": 1.32190652344433e-05,
      "loss": 0.8385,
      "step": 2151
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8722580359797597,
      "learning_rate": 1.3213165422620111e-05,
      "loss": 0.805,
      "step": 2152
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8944028407960756,
      "learning_rate": 1.3207264363335472e-05,
      "loss": 0.823,
      "step": 2153
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6362287183929279,
      "learning_rate": 1.3201362058880375e-05,
      "loss": 0.332,
      "step": 2154
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9485319589319248,
      "learning_rate": 1.3195458511546307e-05,
      "loss": 0.8258,
      "step": 2155
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8305269396087958,
      "learning_rate": 1.3189553723625217e-05,
      "loss": 0.8179,
      "step": 2156
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9186815451847754,
      "learning_rate": 1.318364769740955e-05,
      "loss": 0.883,
      "step": 2157
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8130072663069422,
      "learning_rate": 1.3177740435192235e-05,
      "loss": 0.8159,
      "step": 2158
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9720995906270598,
      "learning_rate": 1.3171831939266668e-05,
      "loss": 0.8451,
      "step": 2159
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.020579839747264,
      "learning_rate": 1.3165922211926734e-05,
      "loss": 0.8835,
      "step": 2160
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8343260433575109,
      "learning_rate": 1.3160011255466791e-05,
      "loss": 0.7821,
      "step": 2161
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9231564535345474,
      "learning_rate": 1.3154099072181677e-05,
      "loss": 0.8306,
      "step": 2162
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9668529084854499,
      "learning_rate": 1.3148185664366704e-05,
      "loss": 0.8422,
      "step": 2163
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8913924208524517,
      "learning_rate": 1.314227103431766e-05,
      "loss": 0.8194,
      "step": 2164
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6786011104436539,
      "learning_rate": 1.3136355184330809e-05,
      "loss": 0.322,
      "step": 2165
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9545860978270105,
      "learning_rate": 1.3130438116702888e-05,
      "loss": 0.8299,
      "step": 2166
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9066285784437548,
      "learning_rate": 1.3124519833731106e-05,
      "loss": 0.8278,
      "step": 2167
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9031653125625679,
      "learning_rate": 1.3118600337713146e-05,
      "loss": 0.7915,
      "step": 2168
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9166004742783158,
      "learning_rate": 1.3112679630947156e-05,
      "loss": 0.8564,
      "step": 2169
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9179013679001377,
      "learning_rate": 1.310675771573176e-05,
      "loss": 0.879,
      "step": 2170
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9659073277455126,
      "learning_rate": 1.310083459436605e-05,
      "loss": 0.8372,
      "step": 2171
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9068067368953961,
      "learning_rate": 1.3094910269149587e-05,
      "loss": 0.8026,
      "step": 2172
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8899072665505431,
      "learning_rate": 1.3088984742382395e-05,
      "loss": 0.8381,
      "step": 2173
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8810144813086953,
      "learning_rate": 1.3083058016364972e-05,
      "loss": 0.8324,
      "step": 2174
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.951261192589121,
      "learning_rate": 1.3077130093398274e-05,
      "loss": 0.8699,
      "step": 2175
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9315421930382464,
      "learning_rate": 1.3071200975783725e-05,
      "loss": 0.8089,
      "step": 2176
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9063992639437459,
      "learning_rate": 1.3065270665823206e-05,
      "loss": 0.8738,
      "step": 2177
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8681085197877371,
      "learning_rate": 1.3059339165819082e-05,
      "loss": 0.8101,
      "step": 2178
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.923497961995955,
      "learning_rate": 1.3053406478074155e-05,
      "loss": 0.8429,
      "step": 2179
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8960729107422843,
      "learning_rate": 1.3047472604891701e-05,
      "loss": 0.8719,
      "step": 2180
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9174083619968908,
      "learning_rate": 1.3041537548575455e-05,
      "loss": 0.772,
      "step": 2181
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.85496325015295,
      "learning_rate": 1.303560131142961e-05,
      "loss": 0.8525,
      "step": 2182
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8863540579001733,
      "learning_rate": 1.3029663895758814e-05,
      "loss": 0.8531,
      "step": 2183
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9316277916692143,
      "learning_rate": 1.3023725303868183e-05,
      "loss": 0.7872,
      "step": 2184
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9103693777793167,
      "learning_rate": 1.3017785538063277e-05,
      "loss": 0.8218,
      "step": 2185
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9218996655116388,
      "learning_rate": 1.3011844600650121e-05,
      "loss": 0.8592,
      "step": 2186
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9519352385664571,
      "learning_rate": 1.300590249393519e-05,
      "loss": 0.8406,
      "step": 2187
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8730123474933638,
      "learning_rate": 1.2999959220225416e-05,
      "loss": 0.8737,
      "step": 2188
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8719007977690524,
      "learning_rate": 1.299401478182818e-05,
      "loss": 0.854,
      "step": 2189
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6415756668891844,
      "learning_rate": 1.2988069181051314e-05,
      "loss": 0.3422,
      "step": 2190
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8453095919879439,
      "learning_rate": 1.2982122420203114e-05,
      "loss": 0.8581,
      "step": 2191
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8885338728511852,
      "learning_rate": 1.2976174501592313e-05,
      "loss": 0.837,
      "step": 2192
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8453704594590639,
      "learning_rate": 1.2970225427528098e-05,
      "loss": 0.8136,
      "step": 2193
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8134869984693511,
      "learning_rate": 1.2964275200320104e-05,
      "loss": 0.8421,
      "step": 2194
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8366949309631647,
      "learning_rate": 1.2958323822278413e-05,
      "loss": 0.8444,
      "step": 2195
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8833972916141275,
      "learning_rate": 1.2952371295713558e-05,
      "loss": 0.8581,
      "step": 2196
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9498489928359702,
      "learning_rate": 1.2946417622936512e-05,
      "loss": 0.8192,
      "step": 2197
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9044295334295956,
      "learning_rate": 1.2940462806258696e-05,
      "loss": 0.8388,
      "step": 2198
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.866335739963497,
      "learning_rate": 1.2934506847991976e-05,
      "loss": 0.8208,
      "step": 2199
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8849439242633889,
      "learning_rate": 1.2928549750448661e-05,
      "loss": 0.7718,
      "step": 2200
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8959221049274818,
      "learning_rate": 1.2922591515941498e-05,
      "loss": 0.8457,
      "step": 2201
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9099592230441483,
      "learning_rate": 1.2916632146783683e-05,
      "loss": 0.8034,
      "step": 2202
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9263804491985094,
      "learning_rate": 1.2910671645288841e-05,
      "loss": 0.8884,
      "step": 2203
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9055949729834711,
      "learning_rate": 1.2904710013771054e-05,
      "loss": 0.772,
      "step": 2204
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9755269409599453,
      "learning_rate": 1.2898747254544826e-05,
      "loss": 0.8673,
      "step": 2205
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9304092644847091,
      "learning_rate": 1.2892783369925105e-05,
      "loss": 0.8124,
      "step": 2206
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8738500255237043,
      "learning_rate": 1.2886818362227283e-05,
      "loss": 0.8218,
      "step": 2207
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9350577122855949,
      "learning_rate": 1.2880852233767174e-05,
      "loss": 0.8459,
      "step": 2208
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9006245278510514,
      "learning_rate": 1.2874884986861038e-05,
      "loss": 0.8534,
      "step": 2209
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9056213556764512,
      "learning_rate": 1.2868916623825561e-05,
      "loss": 0.8542,
      "step": 2210
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9273056506565264,
      "learning_rate": 1.2862947146977876e-05,
      "loss": 0.8103,
      "step": 2211
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9115237564352413,
      "learning_rate": 1.2856976558635532e-05,
      "loss": 0.8367,
      "step": 2212
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8775218088008436,
      "learning_rate": 1.2851004861116519e-05,
      "loss": 0.8688,
      "step": 2213
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9312994445205127,
      "learning_rate": 1.2845032056739257e-05,
      "loss": 0.8462,
      "step": 2214
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8999122806917376,
      "learning_rate": 1.2839058147822595e-05,
      "loss": 0.8384,
      "step": 2215
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8669459026818231,
      "learning_rate": 1.2833083136685803e-05,
      "loss": 0.8393,
      "step": 2216
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8327103707907066,
      "learning_rate": 1.2827107025648595e-05,
      "loss": 0.8351,
      "step": 2217
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9096548713748628,
      "learning_rate": 1.2821129817031099e-05,
      "loss": 0.839,
      "step": 2218
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0154928775963457,
      "learning_rate": 1.2815151513153874e-05,
      "loss": 0.8568,
      "step": 2219
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9445014939657639,
      "learning_rate": 1.2809172116337903e-05,
      "loss": 0.8684,
      "step": 2220
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9061180743997501,
      "learning_rate": 1.2803191628904594e-05,
      "loss": 0.8522,
      "step": 2221
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8805631011916942,
      "learning_rate": 1.2797210053175779e-05,
      "loss": 0.8018,
      "step": 2222
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8771426203487475,
      "learning_rate": 1.2791227391473706e-05,
      "loss": 0.8307,
      "step": 2223
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9199223307033391,
      "learning_rate": 1.2785243646121059e-05,
      "loss": 0.8063,
      "step": 2224
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.955676725673217,
      "learning_rate": 1.277925881944093e-05,
      "loss": 0.8396,
      "step": 2225
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9287970941172812,
      "learning_rate": 1.2773272913756833e-05,
      "loss": 0.8705,
      "step": 2226
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9590004269670487,
      "learning_rate": 1.2767285931392705e-05,
      "loss": 0.8989,
      "step": 2227
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9439883504678238,
      "learning_rate": 1.27612978746729e-05,
      "loss": 0.8359,
      "step": 2228
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9387079876706972,
      "learning_rate": 1.2755308745922182e-05,
      "loss": 0.8721,
      "step": 2229
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8886689572495989,
      "learning_rate": 1.2749318547465742e-05,
      "loss": 0.8324,
      "step": 2230
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9561251350375432,
      "learning_rate": 1.2743327281629181e-05,
      "loss": 0.7893,
      "step": 2231
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9981662677431808,
      "learning_rate": 1.2737334950738512e-05,
      "loss": 0.8538,
      "step": 2232
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6326961774673527,
      "learning_rate": 1.273134155712017e-05,
      "loss": 0.3464,
      "step": 2233
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6475048122240991,
      "learning_rate": 1.272534710310099e-05,
      "loss": 0.372,
      "step": 2234
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8493526055662766,
      "learning_rate": 1.2719351591008228e-05,
      "loss": 0.8657,
      "step": 2235
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8810452441709623,
      "learning_rate": 1.2713355023169547e-05,
      "loss": 0.7396,
      "step": 2236
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9537886943129791,
      "learning_rate": 1.2707357401913022e-05,
      "loss": 0.8632,
      "step": 2237
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8540196146046531,
      "learning_rate": 1.270135872956714e-05,
      "loss": 0.824,
      "step": 2238
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8545864779356266,
      "learning_rate": 1.2695359008460785e-05,
      "loss": 0.8357,
      "step": 2239
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9464892211625475,
      "learning_rate": 1.2689358240923264e-05,
      "loss": 0.7599,
      "step": 2240
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9164971733455652,
      "learning_rate": 1.2683356429284273e-05,
      "loss": 0.7926,
      "step": 2241
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.893355183139102,
      "learning_rate": 1.2677353575873926e-05,
      "loss": 0.8204,
      "step": 2242
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8930424751696118,
      "learning_rate": 1.2671349683022736e-05,
      "loss": 0.8313,
      "step": 2243
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6579519278549545,
      "learning_rate": 1.2665344753061622e-05,
      "loss": 0.3661,
      "step": 2244
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6423559610992513,
      "learning_rate": 1.2659338788321904e-05,
      "loss": 0.369,
      "step": 2245
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9900709346283105,
      "learning_rate": 1.2653331791135308e-05,
      "loss": 0.8335,
      "step": 2246
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0027779560264818,
      "learning_rate": 1.2647323763833952e-05,
      "loss": 0.8612,
      "step": 2247
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9705071960325581,
      "learning_rate": 1.264131470875036e-05,
      "loss": 0.8096,
      "step": 2248
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9737412948014224,
      "learning_rate": 1.2635304628217452e-05,
      "loss": 0.872,
      "step": 2249
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.979317583931037,
      "learning_rate": 1.2629293524568555e-05,
      "loss": 0.9045,
      "step": 2250
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6893161355267412,
      "learning_rate": 1.2623281400137383e-05,
      "loss": 0.3462,
      "step": 2251
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.901785511805362,
      "learning_rate": 1.2617268257258051e-05,
      "loss": 0.7784,
      "step": 2252
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0131132991948384,
      "learning_rate": 1.2611254098265063e-05,
      "loss": 0.8783,
      "step": 2253
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9129056359837359,
      "learning_rate": 1.2605238925493326e-05,
      "loss": 0.8625,
      "step": 2254
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9046808823212116,
      "learning_rate": 1.2599222741278136e-05,
      "loss": 0.8887,
      "step": 2255
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9616673879403604,
      "learning_rate": 1.2593205547955185e-05,
      "loss": 0.7906,
      "step": 2256
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8998216564653858,
      "learning_rate": 1.2587187347860554e-05,
      "loss": 0.8769,
      "step": 2257
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.910372537231332,
      "learning_rate": 1.2581168143330716e-05,
      "loss": 0.7691,
      "step": 2258
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8993893196956334,
      "learning_rate": 1.2575147936702531e-05,
      "loss": 0.8248,
      "step": 2259
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9366492299065973,
      "learning_rate": 1.2569126730313255e-05,
      "loss": 0.8187,
      "step": 2260
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9753635000647769,
      "learning_rate": 1.2563104526500523e-05,
      "loss": 0.8586,
      "step": 2261
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6423213818675634,
      "learning_rate": 1.2557081327602361e-05,
      "loss": 0.3634,
      "step": 2262
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9730790349957007,
      "learning_rate": 1.2551057135957187e-05,
      "loss": 0.805,
      "step": 2263
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9835889248477745,
      "learning_rate": 1.2545031953903796e-05,
      "loss": 0.8292,
      "step": 2264
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8509058391822822,
      "learning_rate": 1.2539005783781374e-05,
      "loss": 0.8032,
      "step": 2265
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9201302661869103,
      "learning_rate": 1.2532978627929486e-05,
      "loss": 0.878,
      "step": 2266
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8715376950856221,
      "learning_rate": 1.2526950488688083e-05,
      "loss": 0.8628,
      "step": 2267
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0594520047935554,
      "learning_rate": 1.2520921368397492e-05,
      "loss": 0.8472,
      "step": 2268
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9144952990672605,
      "learning_rate": 1.2514891269398429e-05,
      "loss": 0.8154,
      "step": 2269
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9323786785058538,
      "learning_rate": 1.2508860194031986e-05,
      "loss": 0.8331,
      "step": 2270
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6531853342179946,
      "learning_rate": 1.2502828144639629e-05,
      "loss": 0.3535,
      "step": 2271
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8172144673745517,
      "learning_rate": 1.2496795123563218e-05,
      "loss": 0.8294,
      "step": 2272
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9585194322520708,
      "learning_rate": 1.249076113314497e-05,
      "loss": 0.8365,
      "step": 2273
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.89519365989807,
      "learning_rate": 1.248472617572749e-05,
      "loss": 0.746,
      "step": 2274
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9387569658594638,
      "learning_rate": 1.2478690253653756e-05,
      "loss": 0.8735,
      "step": 2275
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9611843934866328,
      "learning_rate": 1.2472653369267122e-05,
      "loss": 0.8313,
      "step": 2276
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9632024458396176,
      "learning_rate": 1.2466615524911316e-05,
      "loss": 0.7389,
      "step": 2277
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8962283916932735,
      "learning_rate": 1.2460576722930432e-05,
      "loss": 0.8268,
      "step": 2278
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9502085783307822,
      "learning_rate": 1.2454536965668949e-05,
      "loss": 0.8415,
      "step": 2279
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.963630794028161,
      "learning_rate": 1.24484962554717e-05,
      "loss": 0.8963,
      "step": 2280
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0111113288834166,
      "learning_rate": 1.24424545946839e-05,
      "loss": 0.772,
      "step": 2281
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8908786505288011,
      "learning_rate": 1.2436411985651131e-05,
      "loss": 0.8771,
      "step": 2282
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8591463028521654,
      "learning_rate": 1.2430368430719342e-05,
      "loss": 0.7957,
      "step": 2283
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9878761532963412,
      "learning_rate": 1.242432393223485e-05,
      "loss": 0.8614,
      "step": 2284
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9725571750925845,
      "learning_rate": 1.2418278492544328e-05,
      "loss": 0.8227,
      "step": 2285
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9818059237419009,
      "learning_rate": 1.2412232113994841e-05,
      "loss": 0.8052,
      "step": 2286
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8830315748764603,
      "learning_rate": 1.2406184798933786e-05,
      "loss": 0.8361,
      "step": 2287
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8112724611279631,
      "learning_rate": 1.2400136549708945e-05,
      "loss": 0.8353,
      "step": 2288
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8260284072677478,
      "learning_rate": 1.239408736866846e-05,
      "loss": 0.845,
      "step": 2289
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9537160078777562,
      "learning_rate": 1.2388037258160823e-05,
      "loss": 0.7968,
      "step": 2290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8915895209747542,
      "learning_rate": 1.23819862205349e-05,
      "loss": 0.801,
      "step": 2291
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8862832895674161,
      "learning_rate": 1.2375934258139917e-05,
      "loss": 0.8198,
      "step": 2292
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8606859305949497,
      "learning_rate": 1.2369881373325448e-05,
      "loss": 0.8629,
      "step": 2293
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9497982243685315,
      "learning_rate": 1.236382756844143e-05,
      "loss": 0.8937,
      "step": 2294
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6762961781757095,
      "learning_rate": 1.2357772845838159e-05,
      "loss": 0.3539,
      "step": 2295
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8568416705101688,
      "learning_rate": 1.2351717207866292e-05,
      "loss": 0.7969,
      "step": 2296
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8734465176137307,
      "learning_rate": 1.2345660656876832e-05,
      "loss": 0.8018,
      "step": 2297
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9032712075057601,
      "learning_rate": 1.233960319522114e-05,
      "loss": 0.858,
      "step": 2298
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9836359317334388,
      "learning_rate": 1.2333544825250938e-05,
      "loss": 0.8721,
      "step": 2299
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9340220172594565,
      "learning_rate": 1.2327485549318285e-05,
      "loss": 0.8131,
      "step": 2300
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8960014170298839,
      "learning_rate": 1.2321425369775601e-05,
      "loss": 0.8454,
      "step": 2301
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9008152342214373,
      "learning_rate": 1.2315364288975665e-05,
      "loss": 0.813,
      "step": 2302
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9131202701498977,
      "learning_rate": 1.2309302309271587e-05,
      "loss": 0.8109,
      "step": 2303
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0206983564074195,
      "learning_rate": 1.2303239433016842e-05,
      "loss": 0.8461,
      "step": 2304
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9220550589076495,
      "learning_rate": 1.2297175662565248e-05,
      "loss": 0.8481,
      "step": 2305
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6235653676515943,
      "learning_rate": 1.229111100027097e-05,
      "loss": 0.3407,
      "step": 2306
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0115158126526735,
      "learning_rate": 1.228504544848851e-05,
      "loss": 0.8674,
      "step": 2307
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.616148954053492,
      "learning_rate": 1.2278979009572736e-05,
      "loss": 0.3497,
      "step": 2308
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8879986302236202,
      "learning_rate": 1.2272911685878841e-05,
      "loss": 0.7589,
      "step": 2309
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.896482646593493,
      "learning_rate": 1.2266843479762372e-05,
      "loss": 0.8293,
      "step": 2310
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8858071673225315,
      "learning_rate": 1.2260774393579209e-05,
      "loss": 0.8088,
      "step": 2311
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8442499526481856,
      "learning_rate": 1.2254704429685593e-05,
      "loss": 0.8517,
      "step": 2312
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9546282737342803,
      "learning_rate": 1.2248633590438084e-05,
      "loss": 0.8545,
      "step": 2313
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8282326883291615,
      "learning_rate": 1.2242561878193589e-05,
      "loss": 0.8047,
      "step": 2314
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9811556215384327,
      "learning_rate": 1.2236489295309362e-05,
      "loss": 0.8615,
      "step": 2315
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6445124034582596,
      "learning_rate": 1.2230415844142984e-05,
      "loss": 0.3161,
      "step": 2316
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9878716666098286,
      "learning_rate": 1.2224341527052378e-05,
      "loss": 0.8575,
      "step": 2317
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8603976727977397,
      "learning_rate": 1.2218266346395811e-05,
      "loss": 0.8318,
      "step": 2318
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0334077371751302,
      "learning_rate": 1.221219030453187e-05,
      "loss": 0.8675,
      "step": 2319
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8908892603384936,
      "learning_rate": 1.220611340381948e-05,
      "loss": 0.8512,
      "step": 2320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7903778781340928,
      "learning_rate": 1.2200035646617912e-05,
      "loss": 0.8281,
      "step": 2321
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.902788796635266,
      "learning_rate": 1.2193957035286757e-05,
      "loss": 0.7896,
      "step": 2322
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9418615155690113,
      "learning_rate": 1.2187877572185937e-05,
      "loss": 0.832,
      "step": 2323
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.988159608516871,
      "learning_rate": 1.2181797259675713e-05,
      "loss": 0.8861,
      "step": 2324
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8900192743541402,
      "learning_rate": 1.2175716100116677e-05,
      "loss": 0.8666,
      "step": 2325
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9062537193735849,
      "learning_rate": 1.2169634095869736e-05,
      "loss": 0.8186,
      "step": 2326
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9787782202027756,
      "learning_rate": 1.2163551249296132e-05,
      "loss": 0.8819,
      "step": 2327
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9481001967636833,
      "learning_rate": 1.2157467562757443e-05,
      "loss": 0.8236,
      "step": 2328
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8813955037768413,
      "learning_rate": 1.2151383038615563e-05,
      "loss": 0.8096,
      "step": 2329
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9314959746653316,
      "learning_rate": 1.214529767923271e-05,
      "loss": 0.8077,
      "step": 2330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8890738420473207,
      "learning_rate": 1.2139211486971436e-05,
      "loss": 0.828,
      "step": 2331
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9048551254732008,
      "learning_rate": 1.213312446419461e-05,
      "loss": 0.8388,
      "step": 2332
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8800127547905899,
      "learning_rate": 1.2127036613265418e-05,
      "loss": 0.8614,
      "step": 2333
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9205800091986429,
      "learning_rate": 1.2120947936547375e-05,
      "loss": 0.8052,
      "step": 2334
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6826430090398615,
      "learning_rate": 1.2114858436404322e-05,
      "loss": 0.3161,
      "step": 2335
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.66812808453715,
      "learning_rate": 1.2108768115200405e-05,
      "loss": 0.3621,
      "step": 2336
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.002241309658275,
      "learning_rate": 1.2102676975300095e-05,
      "loss": 0.8584,
      "step": 2337
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9785358210105112,
      "learning_rate": 1.209658501906819e-05,
      "loss": 0.8372,
      "step": 2338
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8695932423964282,
      "learning_rate": 1.2090492248869795e-05,
      "loss": 0.7667,
      "step": 2339
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8900326536160054,
      "learning_rate": 1.2084398667070325e-05,
      "loss": 0.8732,
      "step": 2340
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8542069689350129,
      "learning_rate": 1.2078304276035527e-05,
      "loss": 0.8289,
      "step": 2341
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8839606461014088,
      "learning_rate": 1.2072209078131451e-05,
      "loss": 0.8248,
      "step": 2342
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.94506845989143,
      "learning_rate": 1.2066113075724461e-05,
      "loss": 0.8427,
      "step": 2343
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9385461741738601,
      "learning_rate": 1.206001627118124e-05,
      "loss": 0.8514,
      "step": 2344
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8627032987351473,
      "learning_rate": 1.2053918666868776e-05,
      "loss": 0.7995,
      "step": 2345
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8691377270039453,
      "learning_rate": 1.2047820265154362e-05,
      "loss": 0.7825,
      "step": 2346
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9558191671750059,
      "learning_rate": 1.2041721068405614e-05,
      "loss": 0.8465,
      "step": 2347
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8872032400322101,
      "learning_rate": 1.203562107899045e-05,
      "loss": 0.8025,
      "step": 2348
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8742343542874927,
      "learning_rate": 1.2029520299277095e-05,
      "loss": 0.8092,
      "step": 2349
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9475790176403909,
      "learning_rate": 1.2023418731634078e-05,
      "loss": 0.8884,
      "step": 2350
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9648074165445263,
      "learning_rate": 1.2017316378430244e-05,
      "loss": 0.8248,
      "step": 2351
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0096723464552133,
      "learning_rate": 1.2011213242034733e-05,
      "loss": 0.8166,
      "step": 2352
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8750261536749051,
      "learning_rate": 1.2005109324816992e-05,
      "loss": 0.7909,
      "step": 2353
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.008537211304414,
      "learning_rate": 1.1999004629146775e-05,
      "loss": 0.9152,
      "step": 2354
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.878774318131039,
      "learning_rate": 1.1992899157394133e-05,
      "loss": 0.8494,
      "step": 2355
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9029011531001719,
      "learning_rate": 1.1986792911929418e-05,
      "loss": 0.8014,
      "step": 2356
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9073348592797953,
      "learning_rate": 1.198068589512329e-05,
      "loss": 0.7974,
      "step": 2357
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9976168405208554,
      "learning_rate": 1.1974578109346702e-05,
      "loss": 0.804,
      "step": 2358
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7823764881181438,
      "learning_rate": 1.1968469556970905e-05,
      "loss": 0.3157,
      "step": 2359
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9466774062933332,
      "learning_rate": 1.1962360240367445e-05,
      "loss": 0.8946,
      "step": 2360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8668886670731502,
      "learning_rate": 1.1956250161908179e-05,
      "loss": 0.8526,
      "step": 2361
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9779884622989756,
      "learning_rate": 1.195013932396524e-05,
      "loss": 0.7624,
      "step": 2362
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.022704763496662,
      "learning_rate": 1.1944027728911072e-05,
      "loss": 0.8551,
      "step": 2363
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9048481719588787,
      "learning_rate": 1.1937915379118406e-05,
      "loss": 0.844,
      "step": 2364
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.975985182584186,
      "learning_rate": 1.1931802276960265e-05,
      "loss": 0.8906,
      "step": 2365
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8993541922042338,
      "learning_rate": 1.1925688424809965e-05,
      "loss": 0.8579,
      "step": 2366
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.92822602821753,
      "learning_rate": 1.1919573825041115e-05,
      "loss": 0.8731,
      "step": 2367
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8764646062611547,
      "learning_rate": 1.1913458480027614e-05,
      "loss": 0.8364,
      "step": 2368
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8366904022464043,
      "learning_rate": 1.1907342392143646e-05,
      "loss": 0.8976,
      "step": 2369
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9206202861620859,
      "learning_rate": 1.1901225563763694e-05,
      "loss": 0.7958,
      "step": 2370
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9327183938161066,
      "learning_rate": 1.1895107997262516e-05,
      "loss": 0.753,
      "step": 2371
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7952482036389207,
      "learning_rate": 1.1888989695015166e-05,
      "loss": 0.7825,
      "step": 2372
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9394984611072418,
      "learning_rate": 1.1882870659396968e-05,
      "loss": 0.8304,
      "step": 2373
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9345270711637232,
      "learning_rate": 1.1876750892783558e-05,
      "loss": 0.8281,
      "step": 2374
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9359103525315124,
      "learning_rate": 1.1870630397550831e-05,
      "loss": 0.8726,
      "step": 2375
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9729774159820782,
      "learning_rate": 1.1864509176074974e-05,
      "loss": 0.8647,
      "step": 2376
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.927079009795979,
      "learning_rate": 1.185838723073246e-05,
      "loss": 0.8631,
      "step": 2377
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8748926697994657,
      "learning_rate": 1.1852264563900038e-05,
      "loss": 0.8817,
      "step": 2378
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8691588418637364,
      "learning_rate": 1.1846141177954733e-05,
      "loss": 0.8352,
      "step": 2379
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8773189763357292,
      "learning_rate": 1.1840017075273861e-05,
      "loss": 0.8379,
      "step": 2380
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8842576581342518,
      "learning_rate": 1.1833892258235008e-05,
      "loss": 0.8574,
      "step": 2381
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9184604947137909,
      "learning_rate": 1.1827766729216035e-05,
      "loss": 0.8148,
      "step": 2382
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9458176366902252,
      "learning_rate": 1.1821640490595086e-05,
      "loss": 0.8802,
      "step": 2383
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9070667965638365,
      "learning_rate": 1.181551354475058e-05,
      "loss": 0.7924,
      "step": 2384
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8577290425487507,
      "learning_rate": 1.1809385894061206e-05,
      "loss": 0.7939,
      "step": 2385
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9779564161144946,
      "learning_rate": 1.1803257540905926e-05,
      "loss": 0.8739,
      "step": 2386
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9814218518314121,
      "learning_rate": 1.1797128487663982e-05,
      "loss": 0.8614,
      "step": 2387
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9132063684243544,
      "learning_rate": 1.1790998736714882e-05,
      "loss": 0.8281,
      "step": 2388
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8783636346693541,
      "learning_rate": 1.1784868290438404e-05,
      "loss": 0.7927,
      "step": 2389
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9578141404098198,
      "learning_rate": 1.1778737151214606e-05,
      "loss": 0.8213,
      "step": 2390
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.898408968747149,
      "learning_rate": 1.17726053214238e-05,
      "loss": 0.7761,
      "step": 2391
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8890658739202771,
      "learning_rate": 1.1766472803446577e-05,
      "loss": 0.7894,
      "step": 2392
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8602489796998779,
      "learning_rate": 1.1760339599663788e-05,
      "loss": 0.8295,
      "step": 2393
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9526931440720994,
      "learning_rate": 1.1754205712456556e-05,
      "loss": 0.8235,
      "step": 2394
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9613225466381516,
      "learning_rate": 1.1748071144206266e-05,
      "loss": 0.8443,
      "step": 2395
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0594767409420043,
      "learning_rate": 1.1741935897294572e-05,
      "loss": 0.8114,
      "step": 2396
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7276494899254101,
      "learning_rate": 1.1735799974103388e-05,
      "loss": 0.3412,
      "step": 2397
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9076228940387036,
      "learning_rate": 1.1729663377014888e-05,
      "loss": 0.8518,
      "step": 2398
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8769485711402156,
      "learning_rate": 1.172352610841151e-05,
      "loss": 0.7485,
      "step": 2399
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9012739267850697,
      "learning_rate": 1.1717388170675954e-05,
      "loss": 0.8297,
      "step": 2400
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9182604313648371,
      "learning_rate": 1.1711249566191179e-05,
      "loss": 0.8375,
      "step": 2401
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8698521451502552,
      "learning_rate": 1.17051102973404e-05,
      "loss": 0.8208,
      "step": 2402
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.932319868463604,
      "learning_rate": 1.1698970366507096e-05,
      "loss": 0.7975,
      "step": 2403
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8850233465007697,
      "learning_rate": 1.1692829776074999e-05,
      "loss": 0.8588,
      "step": 2404
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8650017382809384,
      "learning_rate": 1.1686688528428099e-05,
      "loss": 0.8137,
      "step": 2405
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8603786346255067,
      "learning_rate": 1.1680546625950635e-05,
      "loss": 0.8448,
      "step": 2406
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9136110552706779,
      "learning_rate": 1.167440407102711e-05,
      "loss": 0.8943,
      "step": 2407
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8309379967366447,
      "learning_rate": 1.1668260866042271e-05,
      "loss": 0.8028,
      "step": 2408
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8240233881605197,
      "learning_rate": 1.1662117013381126e-05,
      "loss": 0.7329,
      "step": 2409
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8992265398117297,
      "learning_rate": 1.1655972515428928e-05,
      "loss": 0.8455,
      "step": 2410
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9635478455006036,
      "learning_rate": 1.1649827374571182e-05,
      "loss": 0.866,
      "step": 2411
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9229661985122403,
      "learning_rate": 1.1643681593193642e-05,
      "loss": 0.8067,
      "step": 2412
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9112130103739976,
      "learning_rate": 1.1637535173682318e-05,
      "loss": 0.8093,
      "step": 2413
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8754646597402577,
      "learning_rate": 1.1631388118423457e-05,
      "loss": 0.8564,
      "step": 2414
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.672122357232946,
      "learning_rate": 1.1625240429803553e-05,
      "loss": 0.3369,
      "step": 2415
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6333771779400473,
      "learning_rate": 1.1619092110209361e-05,
      "loss": 0.3678,
      "step": 2416
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9366243944954127,
      "learning_rate": 1.1612943162027863e-05,
      "loss": 0.8743,
      "step": 2417
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0305816430071892,
      "learning_rate": 1.1606793587646295e-05,
      "loss": 0.7951,
      "step": 2418
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0395552438897293,
      "learning_rate": 1.160064338945213e-05,
      "loss": 0.8556,
      "step": 2419
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9730724703047997,
      "learning_rate": 1.1594492569833093e-05,
      "loss": 0.863,
      "step": 2420
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6002783115416471,
      "learning_rate": 1.1588341131177137e-05,
      "loss": 0.3584,
      "step": 2421
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9856690500458851,
      "learning_rate": 1.1582189075872467e-05,
      "loss": 0.8565,
      "step": 2422
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9641957860638242,
      "learning_rate": 1.1576036406307523e-05,
      "loss": 0.8246,
      "step": 2423
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8811058714008839,
      "learning_rate": 1.156988312487098e-05,
      "loss": 0.8337,
      "step": 2424
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9049308993234049,
      "learning_rate": 1.1563729233951757e-05,
      "loss": 0.8756,
      "step": 2425
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9152834201119553,
      "learning_rate": 1.1557574735939003e-05,
      "loss": 0.8254,
      "step": 2426
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8797411516172416,
      "learning_rate": 1.1551419633222107e-05,
      "loss": 0.8083,
      "step": 2427
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8832245398957476,
      "learning_rate": 1.1545263928190692e-05,
      "loss": 0.8359,
      "step": 2428
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8797888615008629,
      "learning_rate": 1.1539107623234618e-05,
      "loss": 0.8482,
      "step": 2429
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9205040543568732,
      "learning_rate": 1.153295072074397e-05,
      "loss": 0.8524,
      "step": 2430
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9498591666469491,
      "learning_rate": 1.1526793223109072e-05,
      "loss": 0.8332,
      "step": 2431
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9622396030805644,
      "learning_rate": 1.1520635132720475e-05,
      "loss": 0.8081,
      "step": 2432
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9267423630231612,
      "learning_rate": 1.1514476451968961e-05,
      "loss": 0.8536,
      "step": 2433
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.861735009274987,
      "learning_rate": 1.1508317183245545e-05,
      "loss": 0.7988,
      "step": 2434
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9267849908145733,
      "learning_rate": 1.1502157328941466e-05,
      "loss": 0.8066,
      "step": 2435
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9099020013178674,
      "learning_rate": 1.149599689144819e-05,
      "loss": 0.8523,
      "step": 2436
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8973292415249488,
      "learning_rate": 1.1489835873157414e-05,
      "loss": 0.8826,
      "step": 2437
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8750962737670304,
      "learning_rate": 1.1483674276461053e-05,
      "loss": 0.8317,
      "step": 2438
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9374366496705961,
      "learning_rate": 1.1477512103751254e-05,
      "loss": 0.8636,
      "step": 2439
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9528082283090977,
      "learning_rate": 1.1471349357420384e-05,
      "loss": 0.8358,
      "step": 2440
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.04542207622013,
      "learning_rate": 1.1465186039861033e-05,
      "loss": 0.8193,
      "step": 2441
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9358239903301908,
      "learning_rate": 1.1459022153466016e-05,
      "loss": 0.862,
      "step": 2442
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0608384781328,
      "learning_rate": 1.1452857700628362e-05,
      "loss": 0.8394,
      "step": 2443
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9173346301290893,
      "learning_rate": 1.1446692683741326e-05,
      "loss": 0.8059,
      "step": 2444
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9166246892473237,
      "learning_rate": 1.1440527105198377e-05,
      "loss": 0.8382,
      "step": 2445
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8818513474432851,
      "learning_rate": 1.143436096739321e-05,
      "loss": 0.7959,
      "step": 2446
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9262648624067222,
      "learning_rate": 1.1428194272719729e-05,
      "loss": 0.8244,
      "step": 2447
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9358893451537988,
      "learning_rate": 1.1422027023572052e-05,
      "loss": 0.8699,
      "step": 2448
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8760760324195949,
      "learning_rate": 1.1415859222344525e-05,
      "loss": 0.8314,
      "step": 2449
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8941169465447399,
      "learning_rate": 1.14096908714317e-05,
      "loss": 0.8304,
      "step": 2450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8971277893447626,
      "learning_rate": 1.1403521973228342e-05,
      "loss": 0.805,
      "step": 2451
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8938968469764148,
      "learning_rate": 1.1397352530129428e-05,
      "loss": 0.8663,
      "step": 2452
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9840371642797014,
      "learning_rate": 1.139118254453015e-05,
      "loss": 0.8658,
      "step": 2453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.865795139590691,
      "learning_rate": 1.1385012018825907e-05,
      "loss": 0.813,
      "step": 2454
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7899338672234327,
      "learning_rate": 1.1378840955412313e-05,
      "loss": 0.349,
      "step": 2455
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8888789508146394,
      "learning_rate": 1.1372669356685185e-05,
      "loss": 0.8775,
      "step": 2456
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9337067696512601,
      "learning_rate": 1.1366497225040549e-05,
      "loss": 0.8115,
      "step": 2457
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0049400269279054,
      "learning_rate": 1.1360324562874643e-05,
      "loss": 0.8315,
      "step": 2458
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9274517415587367,
      "learning_rate": 1.1354151372583901e-05,
      "loss": 0.8645,
      "step": 2459
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8834436404164846,
      "learning_rate": 1.1347977656564974e-05,
      "loss": 0.8277,
      "step": 2460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.877795833508024,
      "learning_rate": 1.1341803417214705e-05,
      "loss": 0.8128,
      "step": 2461
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9546222910525111,
      "learning_rate": 1.1335628656930153e-05,
      "loss": 0.8283,
      "step": 2462
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0208948626564218,
      "learning_rate": 1.132945337810857e-05,
      "loss": 0.7877,
      "step": 2463
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8941971268608675,
      "learning_rate": 1.132327758314741e-05,
      "loss": 0.7987,
      "step": 2464
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9004575258192079,
      "learning_rate": 1.131710127444433e-05,
      "loss": 0.8716,
      "step": 2465
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8727286774221223,
      "learning_rate": 1.1310924454397187e-05,
      "loss": 0.8285,
      "step": 2466
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8972928729858601,
      "learning_rate": 1.1304747125404031e-05,
      "loss": 0.8227,
      "step": 2467
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9658239425305535,
      "learning_rate": 1.129856928986312e-05,
      "loss": 0.8807,
      "step": 2468
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9332341810779711,
      "learning_rate": 1.12923909501729e-05,
      "loss": 0.7974,
      "step": 2469
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0152280867524228,
      "learning_rate": 1.1286212108732015e-05,
      "loss": 0.8179,
      "step": 2470
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8688135672930029,
      "learning_rate": 1.1280032767939302e-05,
      "loss": 0.7869,
      "step": 2471
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9294654412949372,
      "learning_rate": 1.1273852930193798e-05,
      "loss": 0.8154,
      "step": 2472
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9820573722582673,
      "learning_rate": 1.1267672597894725e-05,
      "loss": 0.8746,
      "step": 2473
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8934510045792191,
      "learning_rate": 1.12614917734415e-05,
      "loss": 0.7025,
      "step": 2474
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8551044766324492,
      "learning_rate": 1.1255310459233737e-05,
      "loss": 0.8128,
      "step": 2475
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9324637888532037,
      "learning_rate": 1.1249128657671233e-05,
      "loss": 0.8956,
      "step": 2476
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8580456302695965,
      "learning_rate": 1.1242946371153974e-05,
      "loss": 0.8396,
      "step": 2477
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9751795777066251,
      "learning_rate": 1.1236763602082136e-05,
      "loss": 0.8571,
      "step": 2478
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9372182317088138,
      "learning_rate": 1.1230580352856088e-05,
      "loss": 0.8245,
      "step": 2479
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8230379777295255,
      "learning_rate": 1.1224396625876375e-05,
      "loss": 0.8378,
      "step": 2480
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8682754193624708,
      "learning_rate": 1.1218212423543734e-05,
      "loss": 0.7911,
      "step": 2481
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9412688205946572,
      "learning_rate": 1.1212027748259086e-05,
      "loss": 0.8996,
      "step": 2482
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8825757749201112,
      "learning_rate": 1.1205842602423537e-05,
      "loss": 0.8022,
      "step": 2483
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9019297758675981,
      "learning_rate": 1.1199656988438373e-05,
      "loss": 0.7976,
      "step": 2484
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.926127688733699,
      "learning_rate": 1.1193470908705055e-05,
      "loss": 0.8804,
      "step": 2485
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9612874867114664,
      "learning_rate": 1.1187284365625241e-05,
      "loss": 0.8398,
      "step": 2486
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9741433415943679,
      "learning_rate": 1.1181097361600754e-05,
      "loss": 0.8194,
      "step": 2487
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8857750932528216,
      "learning_rate": 1.1174909899033608e-05,
      "loss": 0.744,
      "step": 2488
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0229978470180545,
      "learning_rate": 1.1168721980325987e-05,
      "loss": 0.8704,
      "step": 2489
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8314835955041695,
      "learning_rate": 1.1162533607880251e-05,
      "loss": 0.8394,
      "step": 2490
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8787786334822267,
      "learning_rate": 1.1156344784098942e-05,
      "loss": 0.8525,
      "step": 2491
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9734266721963658,
      "learning_rate": 1.1150155511384772e-05,
      "loss": 0.8393,
      "step": 2492
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9002929985477961,
      "learning_rate": 1.1143965792140631e-05,
      "loss": 0.8678,
      "step": 2493
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9549717101537958,
      "learning_rate": 1.1137775628769584e-05,
      "loss": 0.8833,
      "step": 2494
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9145772734601845,
      "learning_rate": 1.1131585023674863e-05,
      "loss": 0.8682,
      "step": 2495
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9335751455590071,
      "learning_rate": 1.1125393979259874e-05,
      "loss": 0.7836,
      "step": 2496
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8768430208751595,
      "learning_rate": 1.1119202497928192e-05,
      "loss": 0.8151,
      "step": 2497
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9371895471024949,
      "learning_rate": 1.1113010582083568e-05,
      "loss": 0.8515,
      "step": 2498
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8751961265280099,
      "learning_rate": 1.1106818234129913e-05,
      "loss": 0.8339,
      "step": 2499
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.883972482194603,
      "learning_rate": 1.1100625456471307e-05,
      "loss": 0.8303,
      "step": 2500
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9859250864563864,
      "learning_rate": 1.1094432251512006e-05,
      "loss": 0.8133,
      "step": 2501
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8994035979085674,
      "learning_rate": 1.1088238621656422e-05,
      "loss": 0.8133,
      "step": 2502
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9482013981391613,
      "learning_rate": 1.1082044569309138e-05,
      "loss": 0.8652,
      "step": 2503
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9258963244531584,
      "learning_rate": 1.1075850096874894e-05,
      "loss": 0.8534,
      "step": 2504
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7496378304814073,
      "learning_rate": 1.1069655206758603e-05,
      "loss": 0.3426,
      "step": 2505
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8728140964968927,
      "learning_rate": 1.1063459901365325e-05,
      "loss": 0.8768,
      "step": 2506
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0123649366810248,
      "learning_rate": 1.1057264183100303e-05,
      "loss": 0.8361,
      "step": 2507
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8779146402149405,
      "learning_rate": 1.1051068054368921e-05,
      "loss": 0.8127,
      "step": 2508
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8268494766642677,
      "learning_rate": 1.104487151757673e-05,
      "loss": 0.8227,
      "step": 2509
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9624081009789859,
      "learning_rate": 1.1038674575129442e-05,
      "loss": 0.7779,
      "step": 2510
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9997951225051258,
      "learning_rate": 1.1032477229432921e-05,
      "loss": 0.849,
      "step": 2511
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6278664535212514,
      "learning_rate": 1.1026279482893187e-05,
      "loss": 0.3572,
      "step": 2512
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8747940769784894,
      "learning_rate": 1.1020081337916425e-05,
      "loss": 0.8454,
      "step": 2513
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6327995720988032,
      "learning_rate": 1.1013882796908963e-05,
      "loss": 0.3666,
      "step": 2514
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.936919357126992,
      "learning_rate": 1.1007683862277292e-05,
      "loss": 0.818,
      "step": 2515
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9016411363460419,
      "learning_rate": 1.1001484536428052e-05,
      "loss": 0.8038,
      "step": 2516
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8847212138856001,
      "learning_rate": 1.0995284821768029e-05,
      "loss": 0.8386,
      "step": 2517
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8934161929640436,
      "learning_rate": 1.098908472070417e-05,
      "loss": 0.8384,
      "step": 2518
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9295348409728265,
      "learning_rate": 1.0982884235643567e-05,
      "loss": 0.8577,
      "step": 2519
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9253348726028913,
      "learning_rate": 1.0976683368993464e-05,
      "loss": 0.8543,
      "step": 2520
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9239660358217657,
      "learning_rate": 1.0970482123161249e-05,
      "loss": 0.9181,
      "step": 2521
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8705739594864179,
      "learning_rate": 1.0964280500554459e-05,
      "loss": 0.8447,
      "step": 2522
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9448062205017324,
      "learning_rate": 1.0958078503580776e-05,
      "loss": 0.8684,
      "step": 2523
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9484473165199675,
      "learning_rate": 1.0951876134648032e-05,
      "loss": 0.8463,
      "step": 2524
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9702838464963998,
      "learning_rate": 1.0945673396164198e-05,
      "loss": 0.8332,
      "step": 2525
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9423477910012408,
      "learning_rate": 1.0939470290537389e-05,
      "loss": 0.8471,
      "step": 2526
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8897501899606944,
      "learning_rate": 1.0933266820175868e-05,
      "loss": 0.7911,
      "step": 2527
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6340111143651316,
      "learning_rate": 1.0927062987488035e-05,
      "loss": 0.343,
      "step": 2528
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8546727224097606,
      "learning_rate": 1.0920858794882429e-05,
      "loss": 0.85,
      "step": 2529
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8727870096777536,
      "learning_rate": 1.0914654244767736e-05,
      "loss": 0.86,
      "step": 2530
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6271992496282396,
      "learning_rate": 1.0908449339552769e-05,
      "loss": 0.3172,
      "step": 2531
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9277200473130776,
      "learning_rate": 1.0902244081646489e-05,
      "loss": 0.7743,
      "step": 2532
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.905019439271064,
      "learning_rate": 1.0896038473457993e-05,
      "loss": 0.7735,
      "step": 2533
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8557152210699065,
      "learning_rate": 1.0889832517396511e-05,
      "loss": 0.8214,
      "step": 2534
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9137477729017378,
      "learning_rate": 1.0883626215871408e-05,
      "loss": 0.85,
      "step": 2535
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9107016110876013,
      "learning_rate": 1.0877419571292183e-05,
      "loss": 0.8475,
      "step": 2536
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9336601179117388,
      "learning_rate": 1.0871212586068469e-05,
      "loss": 0.8701,
      "step": 2537
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9404937148209042,
      "learning_rate": 1.0865005262610033e-05,
      "loss": 0.8277,
      "step": 2538
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.798926221518447,
      "learning_rate": 1.085879760332677e-05,
      "loss": 0.814,
      "step": 2539
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.931988071535048,
      "learning_rate": 1.085258961062871e-05,
      "loss": 0.8082,
      "step": 2540
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.865847312222079,
      "learning_rate": 1.0846381286926007e-05,
      "loss": 0.7872,
      "step": 2541
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9369259273699448,
      "learning_rate": 1.0840172634628948e-05,
      "loss": 0.8246,
      "step": 2542
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.932716603123185,
      "learning_rate": 1.0833963656147944e-05,
      "loss": 0.801,
      "step": 2543
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.682153899210074,
      "learning_rate": 1.082775435389353e-05,
      "loss": 0.3388,
      "step": 2544
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9591920652282206,
      "learning_rate": 1.0821544730276379e-05,
      "loss": 0.821,
      "step": 2545
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8330646226635915,
      "learning_rate": 1.0815334787707277e-05,
      "loss": 0.8113,
      "step": 2546
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8583634319751613,
      "learning_rate": 1.0809124528597138e-05,
      "loss": 0.8236,
      "step": 2547
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8708659268917964,
      "learning_rate": 1.0802913955356998e-05,
      "loss": 0.8295,
      "step": 2548
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9439356890374719,
      "learning_rate": 1.0796703070398016e-05,
      "loss": 0.7875,
      "step": 2549
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.892524294643804,
      "learning_rate": 1.079049187613147e-05,
      "loss": 0.836,
      "step": 2550
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9803012804851093,
      "learning_rate": 1.0784280374968761e-05,
      "loss": 0.8257,
      "step": 2551
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8958024542019513,
      "learning_rate": 1.0778068569321403e-05,
      "loss": 0.8828,
      "step": 2552
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0041454842797932,
      "learning_rate": 1.077185646160104e-05,
      "loss": 0.8345,
      "step": 2553
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9620565168556444,
      "learning_rate": 1.0765644054219422e-05,
      "loss": 0.831,
      "step": 2554
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9407734424618007,
      "learning_rate": 1.0759431349588421e-05,
      "loss": 0.7998,
      "step": 2555
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.010146191082973,
      "learning_rate": 1.0753218350120023e-05,
      "loss": 0.8788,
      "step": 2556
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.852171333004536,
      "learning_rate": 1.0747005058226325e-05,
      "loss": 0.8095,
      "step": 2557
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.867098903103863,
      "learning_rate": 1.0740791476319543e-05,
      "loss": 0.7871,
      "step": 2558
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9678420339403134,
      "learning_rate": 1.0734577606812007e-05,
      "loss": 0.808,
      "step": 2559
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9464479473938867,
      "learning_rate": 1.0728363452116149e-05,
      "loss": 0.7983,
      "step": 2560
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9219137773096744,
      "learning_rate": 1.0722149014644523e-05,
      "loss": 0.8018,
      "step": 2561
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9475999685764789,
      "learning_rate": 1.0715934296809782e-05,
      "loss": 0.8147,
      "step": 2562
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8892806678239912,
      "learning_rate": 1.0709719301024698e-05,
      "loss": 0.8281,
      "step": 2563
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.023509755568712,
      "learning_rate": 1.0703504029702148e-05,
      "loss": 0.8654,
      "step": 2564
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9580827493163907,
      "learning_rate": 1.0697288485255107e-05,
      "loss": 0.8408,
      "step": 2565
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6583476203354245,
      "learning_rate": 1.0691072670096669e-05,
      "loss": 0.3403,
      "step": 2566
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8953797260230527,
      "learning_rate": 1.0684856586640026e-05,
      "loss": 0.8076,
      "step": 2567
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9055271483723223,
      "learning_rate": 1.0678640237298476e-05,
      "loss": 0.8241,
      "step": 2568
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8737349000626337,
      "learning_rate": 1.0672423624485423e-05,
      "loss": 0.8534,
      "step": 2569
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9139511207891733,
      "learning_rate": 1.0666206750614363e-05,
      "loss": 0.8174,
      "step": 2570
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8844025698040143,
      "learning_rate": 1.0659989618098904e-05,
      "loss": 0.8455,
      "step": 2571
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.892037311099729,
      "learning_rate": 1.065377222935275e-05,
      "loss": 0.8625,
      "step": 2572
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9168811640494859,
      "learning_rate": 1.0647554586789708e-05,
      "loss": 0.8755,
      "step": 2573
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9336426639312455,
      "learning_rate": 1.064133669282368e-05,
      "loss": 0.8263,
      "step": 2574
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8812702232117681,
      "learning_rate": 1.0635118549868668e-05,
      "loss": 0.8432,
      "step": 2575
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9033271780268637,
      "learning_rate": 1.0628900160338764e-05,
      "loss": 0.7788,
      "step": 2576
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9143495300073218,
      "learning_rate": 1.0622681526648167e-05,
      "loss": 0.7917,
      "step": 2577
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9171851069973169,
      "learning_rate": 1.0616462651211156e-05,
      "loss": 0.8258,
      "step": 2578
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8464260571159359,
      "learning_rate": 1.0610243536442125e-05,
      "loss": 0.7698,
      "step": 2579
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6547095765765105,
      "learning_rate": 1.0604024184755539e-05,
      "loss": 0.3497,
      "step": 2580
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.930451132852199,
      "learning_rate": 1.0597804598565969e-05,
      "loss": 0.8978,
      "step": 2581
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9061429989782361,
      "learning_rate": 1.0591584780288069e-05,
      "loss": 0.8254,
      "step": 2582
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8855459153252606,
      "learning_rate": 1.0585364732336587e-05,
      "loss": 0.8134,
      "step": 2583
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8572347760404551,
      "learning_rate": 1.0579144457126365e-05,
      "loss": 0.7911,
      "step": 2584
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8703362235237988,
      "learning_rate": 1.057292395707232e-05,
      "loss": 0.8144,
      "step": 2585
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9164333919729767,
      "learning_rate": 1.0566703234589471e-05,
      "loss": 0.8321,
      "step": 2586
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6108288404714763,
      "learning_rate": 1.0560482292092912e-05,
      "loss": 0.3712,
      "step": 2587
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8926748983939895,
      "learning_rate": 1.0554261131997833e-05,
      "loss": 0.856,
      "step": 2588
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8870294538333673,
      "learning_rate": 1.0548039756719497e-05,
      "loss": 0.8301,
      "step": 2589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.941408842922138,
      "learning_rate": 1.054181816867326e-05,
      "loss": 0.8537,
      "step": 2590
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9173617182079558,
      "learning_rate": 1.053559637027455e-05,
      "loss": 0.8716,
      "step": 2591
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.991493581420146,
      "learning_rate": 1.0529374363938888e-05,
      "loss": 0.8946,
      "step": 2592
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.930432704676571,
      "learning_rate": 1.0523152152081875e-05,
      "loss": 0.7486,
      "step": 2593
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9548262371723343,
      "learning_rate": 1.051692973711918e-05,
      "loss": 0.8387,
      "step": 2594
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8979290372137383,
      "learning_rate": 1.0510707121466568e-05,
      "loss": 0.8673,
      "step": 2595
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6386814477623091,
      "learning_rate": 1.0504484307539864e-05,
      "loss": 0.3183,
      "step": 2596
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8713817845909924,
      "learning_rate": 1.0498261297754984e-05,
      "loss": 0.8233,
      "step": 2597
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9797663160060371,
      "learning_rate": 1.0492038094527907e-05,
      "loss": 0.8467,
      "step": 2598
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9173954347708996,
      "learning_rate": 1.0485814700274706e-05,
      "loss": 0.7973,
      "step": 2599
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.884211052079598,
      "learning_rate": 1.047959111741151e-05,
      "loss": 0.7669,
      "step": 2600
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8949968077068898,
      "learning_rate": 1.0473367348354529e-05,
      "loss": 0.8765,
      "step": 2601
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9597388514621974,
      "learning_rate": 1.0467143395520044e-05,
      "loss": 0.822,
      "step": 2602
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9036057690056982,
      "learning_rate": 1.046091926132441e-05,
      "loss": 0.8079,
      "step": 2603
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9898325244152997,
      "learning_rate": 1.0454694948184045e-05,
      "loss": 0.8043,
      "step": 2604
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8998428982004754,
      "learning_rate": 1.044847045851545e-05,
      "loss": 0.8365,
      "step": 2605
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8881013296323274,
      "learning_rate": 1.044224579473518e-05,
      "loss": 0.8671,
      "step": 2606
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.913114779317727,
      "learning_rate": 1.0436020959259862e-05,
      "loss": 0.7972,
      "step": 2607
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8895648712655045,
      "learning_rate": 1.0429795954506203e-05,
      "loss": 0.7991,
      "step": 2608
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8720348085008901,
      "learning_rate": 1.0423570782890951e-05,
      "loss": 0.8357,
      "step": 2609
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9413141357895105,
      "learning_rate": 1.0417345446830938e-05,
      "loss": 0.7807,
      "step": 2610
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8892204053724623,
      "learning_rate": 1.0411119948743052e-05,
      "loss": 0.7819,
      "step": 2611
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9051892497083076,
      "learning_rate": 1.0404894291044247e-05,
      "loss": 0.8158,
      "step": 2612
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8821486234757627,
      "learning_rate": 1.0398668476151538e-05,
      "loss": 0.7902,
      "step": 2613
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9078302440883009,
      "learning_rate": 1.0392442506482e-05,
      "loss": 0.8985,
      "step": 2614
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.894513904880208,
      "learning_rate": 1.038621638445277e-05,
      "loss": 0.8206,
      "step": 2615
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9408220546368526,
      "learning_rate": 1.037999011248104e-05,
      "loss": 0.8858,
      "step": 2616
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8751649051909806,
      "learning_rate": 1.0373763692984062e-05,
      "loss": 0.8376,
      "step": 2617
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8818219403965178,
      "learning_rate": 1.0367537128379154e-05,
      "loss": 0.8285,
      "step": 2618
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9349232818516506,
      "learning_rate": 1.0361310421083677e-05,
      "loss": 0.8093,
      "step": 2619
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9787255928485379,
      "learning_rate": 1.0355083573515052e-05,
      "loss": 0.8168,
      "step": 2620
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.86892362990586,
      "learning_rate": 1.0348856588090764e-05,
      "loss": 0.8042,
      "step": 2621
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8726120469723883,
      "learning_rate": 1.0342629467228331e-05,
      "loss": 0.8562,
      "step": 2622
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9443148834500905,
      "learning_rate": 1.0336402213345345e-05,
      "loss": 0.7695,
      "step": 2623
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9332011416826533,
      "learning_rate": 1.0330174828859434e-05,
      "loss": 0.8308,
      "step": 2624
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.902158975519002,
      "learning_rate": 1.0323947316188288e-05,
      "loss": 0.8236,
      "step": 2625
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.865353627468692,
      "learning_rate": 1.031771967774964e-05,
      "loss": 0.7999,
      "step": 2626
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9624402253095137,
      "learning_rate": 1.0311491915961271e-05,
      "loss": 0.8372,
      "step": 2627
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.909793653497614,
      "learning_rate": 1.030526403324102e-05,
      "loss": 0.8595,
      "step": 2628
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2118785954402884,
      "learning_rate": 1.0299036032006759e-05,
      "loss": 0.8245,
      "step": 2629
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9559848694640516,
      "learning_rate": 1.0292807914676412e-05,
      "loss": 0.8034,
      "step": 2630
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9351525140012878,
      "learning_rate": 1.0286579683667952e-05,
      "loss": 0.8291,
      "step": 2631
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9822193611515928,
      "learning_rate": 1.0280351341399392e-05,
      "loss": 0.8131,
      "step": 2632
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8542103389758862,
      "learning_rate": 1.027412289028879e-05,
      "loss": 0.8343,
      "step": 2633
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9192279694885683,
      "learning_rate": 1.0267894332754243e-05,
      "loss": 0.854,
      "step": 2634
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6451265418136436,
      "learning_rate": 1.0261665671213891e-05,
      "loss": 0.3592,
      "step": 2635
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8933481897626346,
      "learning_rate": 1.0255436908085919e-05,
      "loss": 0.7855,
      "step": 2636
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8846412466268327,
      "learning_rate": 1.024920804578854e-05,
      "loss": 0.839,
      "step": 2637
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6341483839980708,
      "learning_rate": 1.0242979086740019e-05,
      "loss": 0.3601,
      "step": 2638
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8379826961268309,
      "learning_rate": 1.023675003335865e-05,
      "loss": 0.8131,
      "step": 2639
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9296513891519859,
      "learning_rate": 1.0230520888062765e-05,
      "loss": 0.7811,
      "step": 2640
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6329424965195707,
      "learning_rate": 1.0224291653270739e-05,
      "loss": 0.3449,
      "step": 2641
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0433433102128264,
      "learning_rate": 1.0218062331400969e-05,
      "loss": 0.8743,
      "step": 2642
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9739050889798161,
      "learning_rate": 1.0211832924871889e-05,
      "loss": 0.8266,
      "step": 2643
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8939654003079988,
      "learning_rate": 1.0205603436101978e-05,
      "loss": 0.8213,
      "step": 2644
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9673194546999769,
      "learning_rate": 1.0199373867509734e-05,
      "loss": 0.8799,
      "step": 2645
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9415450211493752,
      "learning_rate": 1.019314422151369e-05,
      "loss": 0.8267,
      "step": 2646
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8779915679354771,
      "learning_rate": 1.0186914500532408e-05,
      "loss": 0.8349,
      "step": 2647
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9639534267921943,
      "learning_rate": 1.0180684706984483e-05,
      "loss": 0.8175,
      "step": 2648
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9203627529889155,
      "learning_rate": 1.0174454843288533e-05,
      "loss": 0.8115,
      "step": 2649
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9788846002690266,
      "learning_rate": 1.0168224911863205e-05,
      "loss": 0.8586,
      "step": 2650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9971828171879129,
      "learning_rate": 1.0161994915127173e-05,
      "loss": 0.848,
      "step": 2651
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9463616266812321,
      "learning_rate": 1.015576485549914e-05,
      "loss": 0.7672,
      "step": 2652
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9993906440464315,
      "learning_rate": 1.0149534735397823e-05,
      "loss": 0.8397,
      "step": 2653
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8582899575153747,
      "learning_rate": 1.0143304557241979e-05,
      "loss": 0.8217,
      "step": 2654
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9384601863996282,
      "learning_rate": 1.0137074323450372e-05,
      "loss": 0.8189,
      "step": 2655
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0439098841023378,
      "learning_rate": 1.0130844036441787e-05,
      "loss": 0.8399,
      "step": 2656
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9484411519754461,
      "learning_rate": 1.0124613698635043e-05,
      "loss": 0.7982,
      "step": 2657
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9246822177143305,
      "learning_rate": 1.0118383312448973e-05,
      "loss": 0.8363,
      "step": 2658
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8652313740422064,
      "learning_rate": 1.0112152880302426e-05,
      "loss": 0.8779,
      "step": 2659
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6842057545392694,
      "learning_rate": 1.0105922404614265e-05,
      "loss": 0.3438,
      "step": 2660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8714344208203978,
      "learning_rate": 1.0099691887803385e-05,
      "loss": 0.8578,
      "step": 2661
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9280064610481337,
      "learning_rate": 1.0093461332288678e-05,
      "loss": 0.8776,
      "step": 2662
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8809863226883596,
      "learning_rate": 1.0087230740489065e-05,
      "loss": 0.8773,
      "step": 2663
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6529362640968233,
      "learning_rate": 1.0081000114823473e-05,
      "loss": 0.3449,
      "step": 2664
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8323990848674785,
      "learning_rate": 1.007476945771085e-05,
      "loss": 0.8492,
      "step": 2665
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9397195583958897,
      "learning_rate": 1.006853877157015e-05,
      "loss": 0.8451,
      "step": 2666
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.943546349815915,
      "learning_rate": 1.0062308058820337e-05,
      "loss": 0.8285,
      "step": 2667
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9512911058234472,
      "learning_rate": 1.0056077321880393e-05,
      "loss": 0.8502,
      "step": 2668
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8749452667222266,
      "learning_rate": 1.0049846563169297e-05,
      "loss": 0.8371,
      "step": 2669
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8512840864416977,
      "learning_rate": 1.0043615785106051e-05,
      "loss": 0.8174,
      "step": 2670
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8796370390073013,
      "learning_rate": 1.0037384990109658e-05,
      "loss": 0.8378,
      "step": 2671
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8752908809902543,
      "learning_rate": 1.0031154180599123e-05,
      "loss": 0.793,
      "step": 2672
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9001034412589243,
      "learning_rate": 1.0024923358993458e-05,
      "loss": 0.8416,
      "step": 2673
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9070631509846198,
      "learning_rate": 1.0018692527711695e-05,
      "loss": 0.8589,
      "step": 2674
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8729936324276313,
      "learning_rate": 1.0012461689172846e-05,
      "loss": 0.8282,
      "step": 2675
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.874229214595415,
      "learning_rate": 1.0006230845795937e-05,
      "loss": 0.7913,
      "step": 2676
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8711394897510107,
      "learning_rate": 1e-05,
      "loss": 0.8475,
      "step": 2677
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8942815553699416,
      "learning_rate": 9.993769154204063e-06,
      "loss": 0.7736,
      "step": 2678
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.930126589127633,
      "learning_rate": 9.987538310827159e-06,
      "loss": 0.842,
      "step": 2679
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8913746433532589,
      "learning_rate": 9.981307472288308e-06,
      "loss": 0.8075,
      "step": 2680
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.90201527494254,
      "learning_rate": 9.975076641006542e-06,
      "loss": 0.8337,
      "step": 2681
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8928784259399958,
      "learning_rate": 9.968845819400883e-06,
      "loss": 0.81,
      "step": 2682
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9488528978098529,
      "learning_rate": 9.962615009890346e-06,
      "loss": 0.8322,
      "step": 2683
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9578564898138324,
      "learning_rate": 9.956384214893949e-06,
      "loss": 0.8849,
      "step": 2684
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9369435253246341,
      "learning_rate": 9.950153436830707e-06,
      "loss": 0.8449,
      "step": 2685
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0069379595267043,
      "learning_rate": 9.94392267811961e-06,
      "loss": 0.9104,
      "step": 2686
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9299758472235661,
      "learning_rate": 9.937691941179665e-06,
      "loss": 0.8312,
      "step": 2687
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0045507812062406,
      "learning_rate": 9.931461228429856e-06,
      "loss": 0.835,
      "step": 2688
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.897433111882256,
      "learning_rate": 9.925230542289151e-06,
      "loss": 0.8488,
      "step": 2689
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9585672649441522,
      "learning_rate": 9.91899988517653e-06,
      "loss": 0.8203,
      "step": 2690
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6486090358162822,
      "learning_rate": 9.912769259510938e-06,
      "loss": 0.3531,
      "step": 2691
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9408644738700991,
      "learning_rate": 9.906538667711324e-06,
      "loss": 0.8527,
      "step": 2692
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9298418819947663,
      "learning_rate": 9.90030811219662e-06,
      "loss": 0.819,
      "step": 2693
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.92515152967437,
      "learning_rate": 9.894077595385736e-06,
      "loss": 0.8332,
      "step": 2694
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9080868982474979,
      "learning_rate": 9.887847119697577e-06,
      "loss": 0.7833,
      "step": 2695
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8830221163861659,
      "learning_rate": 9.881616687551032e-06,
      "loss": 0.894,
      "step": 2696
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9077065753714084,
      "learning_rate": 9.875386301364958e-06,
      "loss": 0.8002,
      "step": 2697
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8913075023136413,
      "learning_rate": 9.869155963558215e-06,
      "loss": 0.7523,
      "step": 2698
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8452708456063243,
      "learning_rate": 9.862925676549635e-06,
      "loss": 0.8139,
      "step": 2699
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9852799779703769,
      "learning_rate": 9.856695442758023e-06,
      "loss": 0.8312,
      "step": 2700
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9002648632167846,
      "learning_rate": 9.850465264602175e-06,
      "loss": 0.8144,
      "step": 2701
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8739219284249765,
      "learning_rate": 9.844235144500865e-06,
      "loss": 0.7848,
      "step": 2702
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.85575428684359,
      "learning_rate": 9.83800508487283e-06,
      "loss": 0.7859,
      "step": 2703
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8876988575594946,
      "learning_rate": 9.831775088136797e-06,
      "loss": 0.7934,
      "step": 2704
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9083444346966596,
      "learning_rate": 9.82554515671147e-06,
      "loss": 0.7814,
      "step": 2705
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8644174542858927,
      "learning_rate": 9.819315293015519e-06,
      "loss": 0.8098,
      "step": 2706
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8818210425497185,
      "learning_rate": 9.813085499467594e-06,
      "loss": 0.8305,
      "step": 2707
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8634053620872911,
      "learning_rate": 9.806855778486314e-06,
      "loss": 0.8465,
      "step": 2708
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9106215788579637,
      "learning_rate": 9.800626132490268e-06,
      "loss": 0.8414,
      "step": 2709
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8830706214536302,
      "learning_rate": 9.794396563898022e-06,
      "loss": 0.7939,
      "step": 2710
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8875585383515582,
      "learning_rate": 9.788167075128113e-06,
      "loss": 0.758,
      "step": 2711
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8339731028060823,
      "learning_rate": 9.781937668599035e-06,
      "loss": 0.7845,
      "step": 2712
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8886174882238808,
      "learning_rate": 9.775708346729263e-06,
      "loss": 0.7868,
      "step": 2713
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8907875576178206,
      "learning_rate": 9.769479111937238e-06,
      "loss": 0.8095,
      "step": 2714
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6853908667990091,
      "learning_rate": 9.763249966641352e-06,
      "loss": 0.3395,
      "step": 2715
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8492719021138081,
      "learning_rate": 9.757020913259986e-06,
      "loss": 0.8585,
      "step": 2716
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9296776201019715,
      "learning_rate": 9.750791954211464e-06,
      "loss": 0.8544,
      "step": 2717
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9092702413323978,
      "learning_rate": 9.744563091914085e-06,
      "loss": 0.8144,
      "step": 2718
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8757607196492998,
      "learning_rate": 9.738334328786114e-06,
      "loss": 0.8435,
      "step": 2719
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8393931390110503,
      "learning_rate": 9.732105667245759e-06,
      "loss": 0.8678,
      "step": 2720
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.868529922796247,
      "learning_rate": 9.725877109711212e-06,
      "loss": 0.8789,
      "step": 2721
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9045944754057672,
      "learning_rate": 9.719648658600611e-06,
      "loss": 0.8279,
      "step": 2722
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9248766251795416,
      "learning_rate": 9.71342031633205e-06,
      "loss": 0.812,
      "step": 2723
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9365693770727488,
      "learning_rate": 9.70719208532359e-06,
      "loss": 0.8615,
      "step": 2724
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8535902089750859,
      "learning_rate": 9.700963967993246e-06,
      "loss": 0.8049,
      "step": 2725
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9347720809054822,
      "learning_rate": 9.694735966758982e-06,
      "loss": 0.8641,
      "step": 2726
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9108322504892093,
      "learning_rate": 9.688508084038729e-06,
      "loss": 0.7885,
      "step": 2727
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9369042039417748,
      "learning_rate": 9.682280322250365e-06,
      "loss": 0.7824,
      "step": 2728
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8671251193476686,
      "learning_rate": 9.676052683811715e-06,
      "loss": 0.8308,
      "step": 2729
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9139869575431803,
      "learning_rate": 9.669825171140568e-06,
      "loss": 0.8361,
      "step": 2730
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8050914910990501,
      "learning_rate": 9.66359778665466e-06,
      "loss": 0.7872,
      "step": 2731
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8385024519865856,
      "learning_rate": 9.657370532771672e-06,
      "loss": 0.8111,
      "step": 2732
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9233988834308862,
      "learning_rate": 9.651143411909241e-06,
      "loss": 0.8214,
      "step": 2733
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9282802081589839,
      "learning_rate": 9.64491642648495e-06,
      "loss": 0.87,
      "step": 2734
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8510276519884985,
      "learning_rate": 9.638689578916326e-06,
      "loss": 0.8576,
      "step": 2735
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9010726363748109,
      "learning_rate": 9.632462871620847e-06,
      "loss": 0.8631,
      "step": 2736
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9654170137835699,
      "learning_rate": 9.62623630701594e-06,
      "loss": 0.8267,
      "step": 2737
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6560297391298098,
      "learning_rate": 9.620009887518963e-06,
      "loss": 0.3306,
      "step": 2738
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8735912394551512,
      "learning_rate": 9.613783615547233e-06,
      "loss": 0.8201,
      "step": 2739
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9270013203140282,
      "learning_rate": 9.607557493518006e-06,
      "loss": 0.8005,
      "step": 2740
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9227608581906446,
      "learning_rate": 9.601331523848464e-06,
      "loss": 0.8037,
      "step": 2741
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8712663031906092,
      "learning_rate": 9.595105708955758e-06,
      "loss": 0.7571,
      "step": 2742
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9353997589207167,
      "learning_rate": 9.588880051256951e-06,
      "loss": 0.8719,
      "step": 2743
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9199468883628774,
      "learning_rate": 9.582654553169064e-06,
      "loss": 0.7961,
      "step": 2744
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9372591065571392,
      "learning_rate": 9.576429217109054e-06,
      "loss": 0.806,
      "step": 2745
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6231509345251269,
      "learning_rate": 9.5702040454938e-06,
      "loss": 0.3281,
      "step": 2746
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9584334477373007,
      "learning_rate": 9.563979040740138e-06,
      "loss": 0.787,
      "step": 2747
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6911126989586137,
      "learning_rate": 9.557754205264826e-06,
      "loss": 0.3441,
      "step": 2748
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9384467110763755,
      "learning_rate": 9.551529541484554e-06,
      "loss": 0.8581,
      "step": 2749
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.014290461482116,
      "learning_rate": 9.545305051815957e-06,
      "loss": 0.8563,
      "step": 2750
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9468069280947159,
      "learning_rate": 9.539080738675597e-06,
      "loss": 0.875,
      "step": 2751
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9137722819597663,
      "learning_rate": 9.53285660447996e-06,
      "loss": 0.8502,
      "step": 2752
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9266704308604238,
      "learning_rate": 9.526632651645476e-06,
      "loss": 0.8157,
      "step": 2753
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.898396332461035,
      "learning_rate": 9.520408882588497e-06,
      "loss": 0.8393,
      "step": 2754
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9652916455298866,
      "learning_rate": 9.514185299725299e-06,
      "loss": 0.8436,
      "step": 2755
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8231561412591036,
      "learning_rate": 9.507961905472093e-06,
      "loss": 0.7993,
      "step": 2756
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8699390794228699,
      "learning_rate": 9.501738702245023e-06,
      "loss": 0.8267,
      "step": 2757
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8734571809654034,
      "learning_rate": 9.495515692460138e-06,
      "loss": 0.813,
      "step": 2758
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9233026011595122,
      "learning_rate": 9.489292878533436e-06,
      "loss": 0.7881,
      "step": 2759
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8551414031645883,
      "learning_rate": 9.483070262880823e-06,
      "loss": 0.9083,
      "step": 2760
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9750705243874158,
      "learning_rate": 9.476847847918126e-06,
      "loss": 0.8473,
      "step": 2761
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9155902619676595,
      "learning_rate": 9.47062563606111e-06,
      "loss": 0.8758,
      "step": 2762
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8551366392166557,
      "learning_rate": 9.464403629725454e-06,
      "loss": 0.7954,
      "step": 2763
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9045002964188006,
      "learning_rate": 9.458181831326744e-06,
      "loss": 0.8127,
      "step": 2764
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9532490536208348,
      "learning_rate": 9.451960243280506e-06,
      "loss": 0.8299,
      "step": 2765
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0144537995122842,
      "learning_rate": 9.44573886800217e-06,
      "loss": 0.8517,
      "step": 2766
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9367602956735837,
      "learning_rate": 9.43951770790709e-06,
      "loss": 0.8491,
      "step": 2767
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0260509774927078,
      "learning_rate": 9.433296765410534e-06,
      "loss": 0.8085,
      "step": 2768
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6184530283347967,
      "learning_rate": 9.427076042927683e-06,
      "loss": 0.3343,
      "step": 2769
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9424672699176908,
      "learning_rate": 9.420855542873638e-06,
      "loss": 0.8224,
      "step": 2770
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9161161469753342,
      "learning_rate": 9.414635267663416e-06,
      "loss": 0.7875,
      "step": 2771
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8539558872073146,
      "learning_rate": 9.408415219711934e-06,
      "loss": 0.7535,
      "step": 2772
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8827482086429652,
      "learning_rate": 9.402195401434036e-06,
      "loss": 0.738,
      "step": 2773
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8805397922529531,
      "learning_rate": 9.395975815244468e-06,
      "loss": 0.798,
      "step": 2774
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9395482898382149,
      "learning_rate": 9.389756463557878e-06,
      "loss": 0.8446,
      "step": 2775
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8891341486054283,
      "learning_rate": 9.383537348788844e-06,
      "loss": 0.845,
      "step": 2776
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6823567605164959,
      "learning_rate": 9.377318473351838e-06,
      "loss": 0.3594,
      "step": 2777
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.629994222579997,
      "learning_rate": 9.371099839661238e-06,
      "loss": 0.3787,
      "step": 2778
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9088554312435095,
      "learning_rate": 9.364881450131335e-06,
      "loss": 0.7951,
      "step": 2779
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9342518682965034,
      "learning_rate": 9.358663307176323e-06,
      "loss": 0.8103,
      "step": 2780
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9190139096284582,
      "learning_rate": 9.352445413210294e-06,
      "loss": 0.8437,
      "step": 2781
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.906549552469079,
      "learning_rate": 9.346227770647251e-06,
      "loss": 0.8398,
      "step": 2782
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8824036298058661,
      "learning_rate": 9.3400103819011e-06,
      "loss": 0.7847,
      "step": 2783
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9311930755544104,
      "learning_rate": 9.33379324938564e-06,
      "loss": 0.8137,
      "step": 2784
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9348846839053557,
      "learning_rate": 9.327576375514582e-06,
      "loss": 0.8437,
      "step": 2785
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9223522869200028,
      "learning_rate": 9.321359762701527e-06,
      "loss": 0.8182,
      "step": 2786
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.646815829612988,
      "learning_rate": 9.315143413359975e-06,
      "loss": 0.3675,
      "step": 2787
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9025439761691967,
      "learning_rate": 9.308927329903333e-06,
      "loss": 0.7732,
      "step": 2788
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8695422238581125,
      "learning_rate": 9.302711514744897e-06,
      "loss": 0.806,
      "step": 2789
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9064134498423921,
      "learning_rate": 9.296495970297855e-06,
      "loss": 0.8496,
      "step": 2790
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9252325586814739,
      "learning_rate": 9.290280698975307e-06,
      "loss": 0.7653,
      "step": 2791
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8561462772796694,
      "learning_rate": 9.284065703190221e-06,
      "loss": 0.8261,
      "step": 2792
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8651562276288275,
      "learning_rate": 9.27785098535548e-06,
      "loss": 0.8304,
      "step": 2793
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9148134018960902,
      "learning_rate": 9.271636547883856e-06,
      "loss": 0.7717,
      "step": 2794
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9096855340020398,
      "learning_rate": 9.265422393187998e-06,
      "loss": 0.796,
      "step": 2795
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9611971850030903,
      "learning_rate": 9.259208523680457e-06,
      "loss": 0.8437,
      "step": 2796
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8455806448291012,
      "learning_rate": 9.252994941773679e-06,
      "loss": 0.7711,
      "step": 2797
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8903317847193496,
      "learning_rate": 9.24678164987998e-06,
      "loss": 0.8088,
      "step": 2798
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8664463243770525,
      "learning_rate": 9.24056865041158e-06,
      "loss": 0.7652,
      "step": 2799
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9443367849087784,
      "learning_rate": 9.234355945780581e-06,
      "loss": 0.807,
      "step": 2800
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0797540571269888,
      "learning_rate": 9.228143538398963e-06,
      "loss": 0.8487,
      "step": 2801
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9111085503394909,
      "learning_rate": 9.221931430678598e-06,
      "loss": 0.8365,
      "step": 2802
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9839497172156995,
      "learning_rate": 9.215719625031245e-06,
      "loss": 0.8351,
      "step": 2803
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.987099781667182,
      "learning_rate": 9.209508123868534e-06,
      "loss": 0.8554,
      "step": 2804
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.842246838955819,
      "learning_rate": 9.203296929601986e-06,
      "loss": 0.7901,
      "step": 2805
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9063140593694151,
      "learning_rate": 9.197086044643004e-06,
      "loss": 0.7917,
      "step": 2806
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9020195733210299,
      "learning_rate": 9.190875471402865e-06,
      "loss": 0.7838,
      "step": 2807
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9181433146456777,
      "learning_rate": 9.184665212292723e-06,
      "loss": 0.8064,
      "step": 2808
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8390331494103472,
      "learning_rate": 9.178455269723623e-06,
      "loss": 0.7931,
      "step": 2809
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6213421488880165,
      "learning_rate": 9.172245646106471e-06,
      "loss": 0.346,
      "step": 2810
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9001027284907431,
      "learning_rate": 9.166036343852061e-06,
      "loss": 0.8079,
      "step": 2811
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6222445538325835,
      "learning_rate": 9.159827365371055e-06,
      "loss": 0.3662,
      "step": 2812
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9685210067395333,
      "learning_rate": 9.153618713073995e-06,
      "loss": 0.8654,
      "step": 2813
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8920263399343821,
      "learning_rate": 9.14741038937129e-06,
      "loss": 0.8034,
      "step": 2814
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.930963396296525,
      "learning_rate": 9.141202396673232e-06,
      "loss": 0.8352,
      "step": 2815
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0205379412050215,
      "learning_rate": 9.13499473738997e-06,
      "loss": 0.8234,
      "step": 2816
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.890498204574965,
      "learning_rate": 9.128787413931536e-06,
      "loss": 0.8417,
      "step": 2817
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9021567553691938,
      "learning_rate": 9.122580428707822e-06,
      "loss": 0.8107,
      "step": 2818
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9262342637325413,
      "learning_rate": 9.116373784128597e-06,
      "loss": 0.79,
      "step": 2819
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.891461232591358,
      "learning_rate": 9.110167482603494e-06,
      "loss": 0.7757,
      "step": 2820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8487291403479028,
      "learning_rate": 9.10396152654201e-06,
      "loss": 0.8358,
      "step": 2821
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8726465556322136,
      "learning_rate": 9.097755918353513e-06,
      "loss": 0.8337,
      "step": 2822
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9516588395032781,
      "learning_rate": 9.091550660447236e-06,
      "loss": 0.8381,
      "step": 2823
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8583467959470671,
      "learning_rate": 9.08534575523227e-06,
      "loss": 0.8397,
      "step": 2824
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9293923789868531,
      "learning_rate": 9.079141205117573e-06,
      "loss": 0.8059,
      "step": 2825
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9881335857769045,
      "learning_rate": 9.072937012511968e-06,
      "loss": 0.8072,
      "step": 2826
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9032635342306711,
      "learning_rate": 9.066733179824134e-06,
      "loss": 0.8939,
      "step": 2827
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8725047321698488,
      "learning_rate": 9.060529709462613e-06,
      "loss": 0.7357,
      "step": 2828
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9011113671956883,
      "learning_rate": 9.054326603835807e-06,
      "loss": 0.8503,
      "step": 2829
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9148046787231339,
      "learning_rate": 9.048123865351971e-06,
      "loss": 0.8547,
      "step": 2830
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0008008157150496,
      "learning_rate": 9.041921496419225e-06,
      "loss": 0.8554,
      "step": 2831
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9080813799289716,
      "learning_rate": 9.035719499445545e-06,
      "loss": 0.8142,
      "step": 2832
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9307720382123568,
      "learning_rate": 9.029517876838755e-06,
      "loss": 0.7749,
      "step": 2833
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6849004888633169,
      "learning_rate": 9.023316631006536e-06,
      "loss": 0.3354,
      "step": 2834
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8977711857846856,
      "learning_rate": 9.017115764356436e-06,
      "loss": 0.8241,
      "step": 2835
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6204754068842,
      "learning_rate": 9.010915279295833e-06,
      "loss": 0.351,
      "step": 2836
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8670610792675758,
      "learning_rate": 9.004715178231975e-06,
      "loss": 0.7973,
      "step": 2837
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6556276042601157,
      "learning_rate": 8.998515463571953e-06,
      "loss": 0.3316,
      "step": 2838
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8717404789179308,
      "learning_rate": 8.992316137722711e-06,
      "loss": 0.7839,
      "step": 2839
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9488404106475103,
      "learning_rate": 8.986117203091042e-06,
      "loss": 0.7986,
      "step": 2840
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8433926836263373,
      "learning_rate": 8.97991866208358e-06,
      "loss": 0.8811,
      "step": 2841
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9003445984357026,
      "learning_rate": 8.973720517106814e-06,
      "loss": 0.7712,
      "step": 2842
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9322842025979723,
      "learning_rate": 8.967522770567086e-06,
      "loss": 0.861,
      "step": 2843
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9010553757024541,
      "learning_rate": 8.961325424870561e-06,
      "loss": 0.7988,
      "step": 2844
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9053159557674239,
      "learning_rate": 8.955128482423271e-06,
      "loss": 0.867,
      "step": 2845
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9151701485967116,
      "learning_rate": 8.948931945631082e-06,
      "loss": 0.8554,
      "step": 2846
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8947955653582053,
      "learning_rate": 8.9427358168997e-06,
      "loss": 0.8627,
      "step": 2847
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9014171235726715,
      "learning_rate": 8.936540098634675e-06,
      "loss": 0.7905,
      "step": 2848
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9757720339712482,
      "learning_rate": 8.930344793241404e-06,
      "loss": 0.7953,
      "step": 2849
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8610061340463989,
      "learning_rate": 8.924149903125108e-06,
      "loss": 0.7729,
      "step": 2850
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.011601105914278,
      "learning_rate": 8.917955430690865e-06,
      "loss": 0.8367,
      "step": 2851
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8679393869408992,
      "learning_rate": 8.91176137834358e-06,
      "loss": 0.815,
      "step": 2852
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8337782494380428,
      "learning_rate": 8.905567748487997e-06,
      "loss": 0.7942,
      "step": 2853
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.758028321262565,
      "learning_rate": 8.899374543528695e-06,
      "loss": 0.8301,
      "step": 2854
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6412696907250184,
      "learning_rate": 8.893181765870094e-06,
      "loss": 0.3656,
      "step": 2855
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9348839233141436,
      "learning_rate": 8.886989417916435e-06,
      "loss": 0.8014,
      "step": 2856
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8943693756872015,
      "learning_rate": 8.88079750207181e-06,
      "loss": 0.8049,
      "step": 2857
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9390916312240684,
      "learning_rate": 8.87460602074013e-06,
      "loss": 0.8037,
      "step": 2858
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9107412550236499,
      "learning_rate": 8.86841497632514e-06,
      "loss": 0.7752,
      "step": 2859
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9102775009221158,
      "learning_rate": 8.862224371230418e-06,
      "loss": 0.7744,
      "step": 2860
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9068398703761597,
      "learning_rate": 8.85603420785937e-06,
      "loss": 0.7794,
      "step": 2861
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8969534981975771,
      "learning_rate": 8.84984448861523e-06,
      "loss": 0.8762,
      "step": 2862
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9226074882460866,
      "learning_rate": 8.84365521590106e-06,
      "loss": 0.7954,
      "step": 2863
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8578588367781853,
      "learning_rate": 8.837466392119752e-06,
      "loss": 0.7327,
      "step": 2864
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.862870301551924,
      "learning_rate": 8.831278019674017e-06,
      "loss": 0.7763,
      "step": 2865
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8743630688387457,
      "learning_rate": 8.825090100966396e-06,
      "loss": 0.7955,
      "step": 2866
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9470412603104771,
      "learning_rate": 8.818902638399247e-06,
      "loss": 0.8077,
      "step": 2867
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8654504557581757,
      "learning_rate": 8.81271563437476e-06,
      "loss": 0.817,
      "step": 2868
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.862852295966076,
      "learning_rate": 8.806529091294948e-06,
      "loss": 0.7991,
      "step": 2869
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.854765825666986,
      "learning_rate": 8.800343011561633e-06,
      "loss": 0.8417,
      "step": 2870
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8708057502137654,
      "learning_rate": 8.794157397576464e-06,
      "loss": 0.8081,
      "step": 2871
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6897948332785355,
      "learning_rate": 8.787972251740916e-06,
      "loss": 0.3497,
      "step": 2872
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8630507621872269,
      "learning_rate": 8.781787576456269e-06,
      "loss": 0.7724,
      "step": 2873
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9281850111871295,
      "learning_rate": 8.775603374123627e-06,
      "loss": 0.823,
      "step": 2874
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8393108991037416,
      "learning_rate": 8.769419647143917e-06,
      "loss": 0.7862,
      "step": 2875
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9260935359795776,
      "learning_rate": 8.763236397917865e-06,
      "loss": 0.8079,
      "step": 2876
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9054280345042903,
      "learning_rate": 8.757053628846028e-06,
      "loss": 0.8206,
      "step": 2877
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.923532903782874,
      "learning_rate": 8.75087134232877e-06,
      "loss": 0.8063,
      "step": 2878
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9004547424460166,
      "learning_rate": 8.744689540766265e-06,
      "loss": 0.7992,
      "step": 2879
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9327844573353299,
      "learning_rate": 8.738508226558499e-06,
      "loss": 0.8201,
      "step": 2880
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8882990271747728,
      "learning_rate": 8.73232740210528e-06,
      "loss": 0.7966,
      "step": 2881
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8937549324999928,
      "learning_rate": 8.726147069806206e-06,
      "loss": 0.8127,
      "step": 2882
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8720299624925851,
      "learning_rate": 8.719967232060698e-06,
      "loss": 0.8186,
      "step": 2883
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8770240074569635,
      "learning_rate": 8.713787891267988e-06,
      "loss": 0.7486,
      "step": 2884
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.946889574703758,
      "learning_rate": 8.707609049827102e-06,
      "loss": 0.8099,
      "step": 2885
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.901620807360295,
      "learning_rate": 8.70143071013688e-06,
      "loss": 0.7921,
      "step": 2886
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.988777271523026,
      "learning_rate": 8.695252874595972e-06,
      "loss": 0.7756,
      "step": 2887
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9578918752404674,
      "learning_rate": 8.689075545602816e-06,
      "loss": 0.7813,
      "step": 2888
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.703672399955851,
      "learning_rate": 8.68289872555567e-06,
      "loss": 0.3641,
      "step": 2889
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8550627920428079,
      "learning_rate": 8.676722416852594e-06,
      "loss": 0.8006,
      "step": 2890
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8683166770705795,
      "learning_rate": 8.670546621891434e-06,
      "loss": 0.8037,
      "step": 2891
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8841314507455754,
      "learning_rate": 8.66437134306985e-06,
      "loss": 0.8228,
      "step": 2892
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7889081352797939,
      "learning_rate": 8.658196582785297e-06,
      "loss": 0.7727,
      "step": 2893
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9040444802770972,
      "learning_rate": 8.652022343435027e-06,
      "loss": 0.8173,
      "step": 2894
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9615561654551227,
      "learning_rate": 8.645848627416102e-06,
      "loss": 0.8151,
      "step": 2895
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8738343643024133,
      "learning_rate": 8.63967543712536e-06,
      "loss": 0.8122,
      "step": 2896
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9278610917185689,
      "learning_rate": 8.633502774959453e-06,
      "loss": 0.8212,
      "step": 2897
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9725950678576002,
      "learning_rate": 8.627330643314818e-06,
      "loss": 0.8183,
      "step": 2898
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8605018981239586,
      "learning_rate": 8.62115904458769e-06,
      "loss": 0.7759,
      "step": 2899
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9648566263272085,
      "learning_rate": 8.614987981174093e-06,
      "loss": 0.771,
      "step": 2900
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9594701252484124,
      "learning_rate": 8.608817455469854e-06,
      "loss": 0.8059,
      "step": 2901
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9489446661788397,
      "learning_rate": 8.602647469870573e-06,
      "loss": 0.8859,
      "step": 2902
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8886649533672951,
      "learning_rate": 8.596478026771658e-06,
      "loss": 0.8578,
      "step": 2903
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8894961008948492,
      "learning_rate": 8.590309128568303e-06,
      "loss": 0.838,
      "step": 2904
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8904751206592825,
      "learning_rate": 8.584140777655476e-06,
      "loss": 0.8502,
      "step": 2905
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8440764488845747,
      "learning_rate": 8.57797297642795e-06,
      "loss": 0.7928,
      "step": 2906
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8590292947806343,
      "learning_rate": 8.571805727280278e-06,
      "loss": 0.805,
      "step": 2907
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8369782466602471,
      "learning_rate": 8.565639032606794e-06,
      "loss": 0.8432,
      "step": 2908
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8981633279823792,
      "learning_rate": 8.559472894801623e-06,
      "loss": 0.7703,
      "step": 2909
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9168632726981396,
      "learning_rate": 8.553307316258678e-06,
      "loss": 0.7735,
      "step": 2910
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8973659879207941,
      "learning_rate": 8.547142299371642e-06,
      "loss": 0.7833,
      "step": 2911
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8586376233157236,
      "learning_rate": 8.540977846533986e-06,
      "loss": 0.8145,
      "step": 2912
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8959160550719872,
      "learning_rate": 8.534813960138968e-06,
      "loss": 0.8172,
      "step": 2913
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8720511699211637,
      "learning_rate": 8.528650642579618e-06,
      "loss": 0.7846,
      "step": 2914
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9102793101790946,
      "learning_rate": 8.52248789624875e-06,
      "loss": 0.8447,
      "step": 2915
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9955505213979047,
      "learning_rate": 8.516325723538949e-06,
      "loss": 0.8977,
      "step": 2916
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9089087479242515,
      "learning_rate": 8.510164126842591e-06,
      "loss": 0.7876,
      "step": 2917
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9191484680258619,
      "learning_rate": 8.504003108551814e-06,
      "loss": 0.8433,
      "step": 2918
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9010386260781592,
      "learning_rate": 8.497842671058539e-06,
      "loss": 0.8672,
      "step": 2919
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9588614765506035,
      "learning_rate": 8.491682816754456e-06,
      "loss": 0.7938,
      "step": 2920
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9026915863674069,
      "learning_rate": 8.485523548031044e-06,
      "loss": 0.8494,
      "step": 2921
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9365272119145712,
      "learning_rate": 8.479364867279529e-06,
      "loss": 0.8331,
      "step": 2922
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9536533485243786,
      "learning_rate": 8.47320677689093e-06,
      "loss": 0.8363,
      "step": 2923
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9366467446177726,
      "learning_rate": 8.467049279256034e-06,
      "loss": 0.8078,
      "step": 2924
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9248706859283505,
      "learning_rate": 8.460892376765387e-06,
      "loss": 0.8761,
      "step": 2925
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9101503567955059,
      "learning_rate": 8.45473607180931e-06,
      "loss": 0.7839,
      "step": 2926
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9076987068112641,
      "learning_rate": 8.448580366777898e-06,
      "loss": 0.8306,
      "step": 2927
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9006473202242683,
      "learning_rate": 8.442425264061e-06,
      "loss": 0.8419,
      "step": 2928
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9446425995257687,
      "learning_rate": 8.436270766048245e-06,
      "loss": 0.7888,
      "step": 2929
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9239549946669774,
      "learning_rate": 8.430116875129023e-06,
      "loss": 0.827,
      "step": 2930
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8831047393570554,
      "learning_rate": 8.42396359369248e-06,
      "loss": 0.8087,
      "step": 2931
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.93678267113599,
      "learning_rate": 8.417810924127533e-06,
      "loss": 0.8551,
      "step": 2932
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8754294321681435,
      "learning_rate": 8.411658868822866e-06,
      "loss": 0.8243,
      "step": 2933
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8437420379132752,
      "learning_rate": 8.40550743016691e-06,
      "loss": 0.7534,
      "step": 2934
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9281668169642902,
      "learning_rate": 8.39935661054787e-06,
      "loss": 0.7813,
      "step": 2935
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.85969212665586,
      "learning_rate": 8.393206412353709e-06,
      "loss": 0.8201,
      "step": 2936
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8465334552597953,
      "learning_rate": 8.38705683797214e-06,
      "loss": 0.8173,
      "step": 2937
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9489805187270017,
      "learning_rate": 8.38090788979064e-06,
      "loss": 0.7707,
      "step": 2938
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9206987292873049,
      "learning_rate": 8.374759570196448e-06,
      "loss": 0.7592,
      "step": 2939
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9561085012475512,
      "learning_rate": 8.368611881576547e-06,
      "loss": 0.8258,
      "step": 2940
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9090317326355709,
      "learning_rate": 8.362464826317687e-06,
      "loss": 0.811,
      "step": 2941
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9563429689336302,
      "learning_rate": 8.35631840680636e-06,
      "loss": 0.804,
      "step": 2942
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8886227884988657,
      "learning_rate": 8.35017262542882e-06,
      "loss": 0.8252,
      "step": 2943
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9191240861150869,
      "learning_rate": 8.344027484571075e-06,
      "loss": 0.8274,
      "step": 2944
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8897647481976063,
      "learning_rate": 8.337882986618877e-06,
      "loss": 0.7695,
      "step": 2945
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9627236599694662,
      "learning_rate": 8.331739133957729e-06,
      "loss": 0.7894,
      "step": 2946
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.981644522675891,
      "learning_rate": 8.325595928972894e-06,
      "loss": 0.7916,
      "step": 2947
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8152322581555713,
      "learning_rate": 8.319453374049367e-06,
      "loss": 0.798,
      "step": 2948
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8682597991766394,
      "learning_rate": 8.313311471571903e-06,
      "loss": 0.7727,
      "step": 2949
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8312895789731203,
      "learning_rate": 8.307170223925003e-06,
      "loss": 0.7809,
      "step": 2950
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9249148766433075,
      "learning_rate": 8.301029633492907e-06,
      "loss": 0.8251,
      "step": 2951
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8785919887262441,
      "learning_rate": 8.294889702659602e-06,
      "loss": 0.8377,
      "step": 2952
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8700526325391463,
      "learning_rate": 8.288750433808828e-06,
      "loss": 0.8436,
      "step": 2953
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8877766075458277,
      "learning_rate": 8.282611829324049e-06,
      "loss": 0.8476,
      "step": 2954
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8961776841723762,
      "learning_rate": 8.276473891588492e-06,
      "loss": 0.7876,
      "step": 2955
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0209973422826852,
      "learning_rate": 8.270336622985116e-06,
      "loss": 0.8352,
      "step": 2956
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8787845075585228,
      "learning_rate": 8.264200025896616e-06,
      "loss": 0.8379,
      "step": 2957
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9116814611999065,
      "learning_rate": 8.258064102705428e-06,
      "loss": 0.8127,
      "step": 2958
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8938340871994415,
      "learning_rate": 8.251928855793736e-06,
      "loss": 0.8379,
      "step": 2959
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8988003467132636,
      "learning_rate": 8.245794287543447e-06,
      "loss": 0.7782,
      "step": 2960
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8507703712289355,
      "learning_rate": 8.239660400336213e-06,
      "loss": 0.7891,
      "step": 2961
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9836128286699736,
      "learning_rate": 8.233527196553428e-06,
      "loss": 0.8726,
      "step": 2962
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9056047639042526,
      "learning_rate": 8.227394678576204e-06,
      "loss": 0.7982,
      "step": 2963
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8685148815471998,
      "learning_rate": 8.221262848785395e-06,
      "loss": 0.797,
      "step": 2964
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9620440851479932,
      "learning_rate": 8.215131709561597e-06,
      "loss": 0.868,
      "step": 2965
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8928781177167705,
      "learning_rate": 8.20900126328512e-06,
      "loss": 0.7911,
      "step": 2966
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9648295359973789,
      "learning_rate": 8.202871512336023e-06,
      "loss": 0.8096,
      "step": 2967
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9084126501246927,
      "learning_rate": 8.196742459094079e-06,
      "loss": 0.8575,
      "step": 2968
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6731531030852652,
      "learning_rate": 8.190614105938796e-06,
      "loss": 0.349,
      "step": 2969
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9675381970724168,
      "learning_rate": 8.184486455249424e-06,
      "loss": 0.7989,
      "step": 2970
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8951061371936494,
      "learning_rate": 8.178359509404916e-06,
      "loss": 0.8258,
      "step": 2971
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8520442984260591,
      "learning_rate": 8.172233270783966e-06,
      "loss": 0.7778,
      "step": 2972
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8811060958797043,
      "learning_rate": 8.166107741764997e-06,
      "loss": 0.7929,
      "step": 2973
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.939342301656453,
      "learning_rate": 8.15998292472614e-06,
      "loss": 0.8054,
      "step": 2974
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9713198010913632,
      "learning_rate": 8.153858822045267e-06,
      "loss": 0.8279,
      "step": 2975
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9597478949846393,
      "learning_rate": 8.147735436099967e-06,
      "loss": 0.7929,
      "step": 2976
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9015423101157358,
      "learning_rate": 8.141612769267543e-06,
      "loss": 0.8089,
      "step": 2977
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9000042198178669,
      "learning_rate": 8.135490823925027e-06,
      "loss": 0.7885,
      "step": 2978
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9327499874293148,
      "learning_rate": 8.129369602449176e-06,
      "loss": 0.8125,
      "step": 2979
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8451094481781865,
      "learning_rate": 8.123249107216446e-06,
      "loss": 0.7825,
      "step": 2980
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8964908862968295,
      "learning_rate": 8.117129340603032e-06,
      "loss": 0.8106,
      "step": 2981
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9341085273665833,
      "learning_rate": 8.111010304984841e-06,
      "loss": 0.7743,
      "step": 2982
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9299159947514029,
      "learning_rate": 8.104892002737488e-06,
      "loss": 0.7273,
      "step": 2983
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8125984292461604,
      "learning_rate": 8.098774436236308e-06,
      "loss": 0.7923,
      "step": 2984
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9391407380323356,
      "learning_rate": 8.092657607856356e-06,
      "loss": 0.839,
      "step": 2985
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6410370212647633,
      "learning_rate": 8.086541519972388e-06,
      "loss": 0.3223,
      "step": 2986
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9029695782347806,
      "learning_rate": 8.080426174958886e-06,
      "loss": 0.8161,
      "step": 2987
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9028981789577178,
      "learning_rate": 8.074311575190039e-06,
      "loss": 0.816,
      "step": 2988
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8634051436582038,
      "learning_rate": 8.068197723039738e-06,
      "loss": 0.7869,
      "step": 2989
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6559059376103494,
      "learning_rate": 8.062084620881598e-06,
      "loss": 0.3652,
      "step": 2990
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.647831475268762,
      "learning_rate": 8.055972271088933e-06,
      "loss": 0.3347,
      "step": 2991
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8616118857066699,
      "learning_rate": 8.049860676034762e-06,
      "loss": 0.772,
      "step": 2992
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9503469515238651,
      "learning_rate": 8.043749838091828e-06,
      "loss": 0.7712,
      "step": 2993
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8949228037808951,
      "learning_rate": 8.037639759632558e-06,
      "loss": 0.7977,
      "step": 2994
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9111786492992983,
      "learning_rate": 8.031530443029099e-06,
      "loss": 0.8381,
      "step": 2995
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8135468666109006,
      "learning_rate": 8.025421890653303e-06,
      "loss": 0.8057,
      "step": 2996
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9295398617219444,
      "learning_rate": 8.019314104876712e-06,
      "loss": 0.7949,
      "step": 2997
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8745895620798525,
      "learning_rate": 8.013207088070582e-06,
      "loss": 0.8086,
      "step": 2998
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6077688289387373,
      "learning_rate": 8.007100842605872e-06,
      "loss": 0.3443,
      "step": 2999
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9629106489298914,
      "learning_rate": 8.000995370853227e-06,
      "loss": 0.7546,
      "step": 3000
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9009075068505787,
      "learning_rate": 7.994890675183008e-06,
      "loss": 0.8064,
      "step": 3001
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8455223620985411,
      "learning_rate": 7.98878675796527e-06,
      "loss": 0.7289,
      "step": 3002
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6264435037158163,
      "learning_rate": 7.98268362156976e-06,
      "loss": 0.3462,
      "step": 3003
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.021109175200724,
      "learning_rate": 7.976581268365924e-06,
      "loss": 0.8681,
      "step": 3004
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8823914190919896,
      "learning_rate": 7.97047970072291e-06,
      "loss": 0.7972,
      "step": 3005
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9868149674795346,
      "learning_rate": 7.964378921009552e-06,
      "loss": 0.8199,
      "step": 3006
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9801402790853327,
      "learning_rate": 7.958278931594385e-06,
      "loss": 0.79,
      "step": 3007
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8859188618362488,
      "learning_rate": 7.952179734845642e-06,
      "loss": 0.8215,
      "step": 3008
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8468078985894798,
      "learning_rate": 7.946081333131227e-06,
      "loss": 0.7957,
      "step": 3009
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8849797413435665,
      "learning_rate": 7.93998372881876e-06,
      "loss": 0.8124,
      "step": 3010
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9616712907683098,
      "learning_rate": 7.93388692427554e-06,
      "loss": 0.8478,
      "step": 3011
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8561668775274952,
      "learning_rate": 7.92779092186855e-06,
      "loss": 0.8195,
      "step": 3012
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0054449417426998,
      "learning_rate": 7.921695723964473e-06,
      "loss": 0.801,
      "step": 3013
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8326698086486156,
      "learning_rate": 7.915601332929678e-06,
      "loss": 0.8083,
      "step": 3014
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8766406292082866,
      "learning_rate": 7.90950775113021e-06,
      "loss": 0.8007,
      "step": 3015
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8620963459203537,
      "learning_rate": 7.903414980931813e-06,
      "loss": 0.8101,
      "step": 3016
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9046346075465268,
      "learning_rate": 7.897323024699907e-06,
      "loss": 0.84,
      "step": 3017
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9236377814595449,
      "learning_rate": 7.8912318847996e-06,
      "loss": 0.838,
      "step": 3018
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8723767109158214,
      "learning_rate": 7.885141563595685e-06,
      "loss": 0.869,
      "step": 3019
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8896412990463735,
      "learning_rate": 7.879052063452626e-06,
      "loss": 0.8121,
      "step": 3020
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9222767865951825,
      "learning_rate": 7.872963386734584e-06,
      "loss": 0.8379,
      "step": 3021
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9828941756795614,
      "learning_rate": 7.866875535805394e-06,
      "loss": 0.87,
      "step": 3022
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9080825542330396,
      "learning_rate": 7.860788513028566e-06,
      "loss": 0.7674,
      "step": 3023
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8592950542609101,
      "learning_rate": 7.85470232076729e-06,
      "loss": 0.7611,
      "step": 3024
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8595075668963955,
      "learning_rate": 7.848616961384442e-06,
      "loss": 0.7508,
      "step": 3025
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8963136228567984,
      "learning_rate": 7.842532437242559e-06,
      "loss": 0.8105,
      "step": 3026
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8661150025772038,
      "learning_rate": 7.83644875070387e-06,
      "loss": 0.7722,
      "step": 3027
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9161172440118561,
      "learning_rate": 7.83036590413027e-06,
      "loss": 0.7817,
      "step": 3028
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8966847396590685,
      "learning_rate": 7.824283899883327e-06,
      "loss": 0.8402,
      "step": 3029
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9727710053911692,
      "learning_rate": 7.818202740324287e-06,
      "loss": 0.8077,
      "step": 3030
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5996390737834433,
      "learning_rate": 7.812122427814068e-06,
      "loss": 0.3297,
      "step": 3031
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8721022972476102,
      "learning_rate": 7.806042964713248e-06,
      "loss": 0.8039,
      "step": 3032
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8881535851388281,
      "learning_rate": 7.79996435338209e-06,
      "loss": 0.8221,
      "step": 3033
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9521242084843881,
      "learning_rate": 7.793886596180521e-06,
      "loss": 0.8443,
      "step": 3034
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6134161535030874,
      "learning_rate": 7.787809695468134e-06,
      "loss": 0.3544,
      "step": 3035
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8996293969116891,
      "learning_rate": 7.78173365360419e-06,
      "loss": 0.7968,
      "step": 3036
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8563083930165744,
      "learning_rate": 7.775658472947623e-06,
      "loss": 0.8261,
      "step": 3037
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.970350692033099,
      "learning_rate": 7.769584155857019e-06,
      "loss": 0.851,
      "step": 3038
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9016466504400048,
      "learning_rate": 7.763510704690645e-06,
      "loss": 0.733,
      "step": 3039
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8720163455990934,
      "learning_rate": 7.757438121806414e-06,
      "loss": 0.8651,
      "step": 3040
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9871417654251717,
      "learning_rate": 7.75136640956192e-06,
      "loss": 0.8186,
      "step": 3041
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8993060501975796,
      "learning_rate": 7.745295570314412e-06,
      "loss": 0.8294,
      "step": 3042
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8999307524988095,
      "learning_rate": 7.739225606420793e-06,
      "loss": 0.7737,
      "step": 3043
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8879898543130486,
      "learning_rate": 7.733156520237633e-06,
      "loss": 0.8081,
      "step": 3044
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9618960892019607,
      "learning_rate": 7.727088314121165e-06,
      "loss": 0.8131,
      "step": 3045
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8614396806410094,
      "learning_rate": 7.721020990427268e-06,
      "loss": 0.8097,
      "step": 3046
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9124758458216881,
      "learning_rate": 7.714954551511489e-06,
      "loss": 0.7444,
      "step": 3047
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8389922359407591,
      "learning_rate": 7.708888999729036e-06,
      "loss": 0.8249,
      "step": 3048
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0057237737416553,
      "learning_rate": 7.702824337434756e-06,
      "loss": 0.8096,
      "step": 3049
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.888304140567456,
      "learning_rate": 7.69676056698316e-06,
      "loss": 0.787,
      "step": 3050
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9100893881442869,
      "learning_rate": 7.690697690728417e-06,
      "loss": 0.767,
      "step": 3051
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9771973278642437,
      "learning_rate": 7.68463571102434e-06,
      "loss": 0.8585,
      "step": 3052
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9554191216596909,
      "learning_rate": 7.678574630224399e-06,
      "loss": 0.8463,
      "step": 3053
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9691849856052909,
      "learning_rate": 7.672514450681721e-06,
      "loss": 0.8012,
      "step": 3054
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9132273473987697,
      "learning_rate": 7.666455174749066e-06,
      "loss": 0.8173,
      "step": 3055
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8819276887638757,
      "learning_rate": 7.66039680477886e-06,
      "loss": 0.8137,
      "step": 3056
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6092286029591505,
      "learning_rate": 7.654339343123173e-06,
      "loss": 0.3719,
      "step": 3057
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9954604771288952,
      "learning_rate": 7.648282792133711e-06,
      "loss": 0.8016,
      "step": 3058
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8983288192580546,
      "learning_rate": 7.642227154161841e-06,
      "loss": 0.8085,
      "step": 3059
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9205778157704319,
      "learning_rate": 7.636172431558575e-06,
      "loss": 0.8028,
      "step": 3060
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8552695859968034,
      "learning_rate": 7.630118626674557e-06,
      "loss": 0.7892,
      "step": 3061
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8362468334178449,
      "learning_rate": 7.6240657418600846e-06,
      "loss": 0.7732,
      "step": 3062
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8400551564535934,
      "learning_rate": 7.618013779465101e-06,
      "loss": 0.8086,
      "step": 3063
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9663378561455378,
      "learning_rate": 7.611962741839178e-06,
      "loss": 0.8543,
      "step": 3064
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9266102139933176,
      "learning_rate": 7.6059126313315466e-06,
      "loss": 0.8299,
      "step": 3065
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9801960406528545,
      "learning_rate": 7.599863450291056e-06,
      "loss": 0.781,
      "step": 3066
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9304320462972632,
      "learning_rate": 7.593815201066215e-06,
      "loss": 0.8342,
      "step": 3067
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9968896696601784,
      "learning_rate": 7.587767886005164e-06,
      "loss": 0.8713,
      "step": 3068
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9048692425626969,
      "learning_rate": 7.581721507455672e-06,
      "loss": 0.8301,
      "step": 3069
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9303969165437285,
      "learning_rate": 7.575676067765154e-06,
      "loss": 0.8184,
      "step": 3070
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.894010328377246,
      "learning_rate": 7.569631569280662e-06,
      "loss": 0.7935,
      "step": 3071
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8941977901775198,
      "learning_rate": 7.563588014348871e-06,
      "loss": 0.8051,
      "step": 3072
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6009386017702366,
      "learning_rate": 7.5575454053161e-06,
      "loss": 0.3213,
      "step": 3073
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.838048436291068,
      "learning_rate": 7.551503744528304e-06,
      "loss": 0.7895,
      "step": 3074
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8417327667262106,
      "learning_rate": 7.545463034331054e-06,
      "loss": 0.851,
      "step": 3075
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9525946070613072,
      "learning_rate": 7.539423277069568e-06,
      "loss": 0.8216,
      "step": 3076
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0850034910555955,
      "learning_rate": 7.53338447508869e-06,
      "loss": 0.761,
      "step": 3077
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8980749796980405,
      "learning_rate": 7.52734663073288e-06,
      "loss": 0.781,
      "step": 3078
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9252791745911499,
      "learning_rate": 7.521309746346246e-06,
      "loss": 0.7745,
      "step": 3079
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9261431257127103,
      "learning_rate": 7.515273824272516e-06,
      "loss": 0.8159,
      "step": 3080
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8969905033319198,
      "learning_rate": 7.509238866855033e-06,
      "loss": 0.8051,
      "step": 3081
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9471579411836506,
      "learning_rate": 7.503204876436785e-06,
      "loss": 0.7489,
      "step": 3082
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.928504808406632,
      "learning_rate": 7.497171855360372e-06,
      "loss": 0.8079,
      "step": 3083
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8512228837557256,
      "learning_rate": 7.491139805968018e-06,
      "loss": 0.8024,
      "step": 3084
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8429127440737798,
      "learning_rate": 7.485108730601571e-06,
      "loss": 0.8209,
      "step": 3085
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9005446130381719,
      "learning_rate": 7.4790786316025125e-06,
      "loss": 0.7674,
      "step": 3086
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9364235248413175,
      "learning_rate": 7.473049511311921e-06,
      "loss": 0.8908,
      "step": 3087
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0061366337983793,
      "learning_rate": 7.467021372070515e-06,
      "loss": 0.8031,
      "step": 3088
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9587802737748812,
      "learning_rate": 7.46099421621863e-06,
      "loss": 0.8115,
      "step": 3089
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9127066881585993,
      "learning_rate": 7.4549680460962044e-06,
      "loss": 0.7969,
      "step": 3090
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8431038068082084,
      "learning_rate": 7.448942864042819e-06,
      "loss": 0.7478,
      "step": 3091
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8804454650397777,
      "learning_rate": 7.4429186723976425e-06,
      "loss": 0.8141,
      "step": 3092
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9449527997984056,
      "learning_rate": 7.43689547349948e-06,
      "loss": 0.8018,
      "step": 3093
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8869598993953937,
      "learning_rate": 7.43087326968675e-06,
      "loss": 0.7819,
      "step": 3094
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9360286043800446,
      "learning_rate": 7.42485206329747e-06,
      "loss": 0.817,
      "step": 3095
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.943890183714738,
      "learning_rate": 7.418831856669286e-06,
      "loss": 0.8759,
      "step": 3096
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9125696358325882,
      "learning_rate": 7.41281265213945e-06,
      "loss": 0.7716,
      "step": 3097
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.868597122965294,
      "learning_rate": 7.406794452044816e-06,
      "loss": 0.7701,
      "step": 3098
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.682941018699608,
      "learning_rate": 7.400777258721865e-06,
      "loss": 0.3651,
      "step": 3099
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9459271406873224,
      "learning_rate": 7.394761074506679e-06,
      "loss": 0.8073,
      "step": 3100
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9540814835154555,
      "learning_rate": 7.3887459017349405e-06,
      "loss": 0.8041,
      "step": 3101
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8656930028179741,
      "learning_rate": 7.382731742741953e-06,
      "loss": 0.7445,
      "step": 3102
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9036409151924134,
      "learning_rate": 7.376718599862621e-06,
      "loss": 0.8306,
      "step": 3103
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8870537605273829,
      "learning_rate": 7.370706475431446e-06,
      "loss": 0.7577,
      "step": 3104
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8923558599084171,
      "learning_rate": 7.364695371782547e-06,
      "loss": 0.8088,
      "step": 3105
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8495913025424678,
      "learning_rate": 7.358685291249644e-06,
      "loss": 0.787,
      "step": 3106
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8923500610134905,
      "learning_rate": 7.352676236166051e-06,
      "loss": 0.7993,
      "step": 3107
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9413066398039539,
      "learning_rate": 7.346668208864695e-06,
      "loss": 0.7733,
      "step": 3108
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.927817174328612,
      "learning_rate": 7.3406612116781e-06,
      "loss": 0.7979,
      "step": 3109
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8787436635580761,
      "learning_rate": 7.33465524693838e-06,
      "loss": 0.742,
      "step": 3110
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9230468076717209,
      "learning_rate": 7.328650316977265e-06,
      "loss": 0.7802,
      "step": 3111
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8578285798361892,
      "learning_rate": 7.322646424126079e-06,
      "loss": 0.7961,
      "step": 3112
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1023973736301973,
      "learning_rate": 7.316643570715729e-06,
      "loss": 0.795,
      "step": 3113
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9236773963738355,
      "learning_rate": 7.310641759076742e-06,
      "loss": 0.7713,
      "step": 3114
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0199567862012349,
      "learning_rate": 7.304640991539216e-06,
      "loss": 0.8775,
      "step": 3115
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8846238131325247,
      "learning_rate": 7.2986412704328625e-06,
      "loss": 0.7913,
      "step": 3116
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8465501010728887,
      "learning_rate": 7.292642598086982e-06,
      "loss": 0.8248,
      "step": 3117
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8768389687014375,
      "learning_rate": 7.286644976830457e-06,
      "loss": 0.7548,
      "step": 3118
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9772374465960456,
      "learning_rate": 7.280648408991775e-06,
      "loss": 0.7606,
      "step": 3119
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9803039917684566,
      "learning_rate": 7.274652896899015e-06,
      "loss": 0.9065,
      "step": 3120
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.893818483602827,
      "learning_rate": 7.268658442879834e-06,
      "loss": 0.8689,
      "step": 3121
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9594962750505325,
      "learning_rate": 7.262665049261489e-06,
      "loss": 0.7489,
      "step": 3122
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.900629384290254,
      "learning_rate": 7.256672718370824e-06,
      "loss": 0.8634,
      "step": 3123
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8787171247768543,
      "learning_rate": 7.250681452534261e-06,
      "loss": 0.7326,
      "step": 3124
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8918120380212007,
      "learning_rate": 7.2446912540778196e-06,
      "loss": 0.7874,
      "step": 3125
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9057997519484395,
      "learning_rate": 7.238702125327106e-06,
      "loss": 0.7928,
      "step": 3126
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8780780938371464,
      "learning_rate": 7.232714068607296e-06,
      "loss": 0.8079,
      "step": 3127
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8242257134032676,
      "learning_rate": 7.226727086243168e-06,
      "loss": 0.7747,
      "step": 3128
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9155535016455626,
      "learning_rate": 7.220741180559074e-06,
      "loss": 0.8242,
      "step": 3129
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.972960322124405,
      "learning_rate": 7.214756353878942e-06,
      "loss": 0.7713,
      "step": 3130
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9583482996377302,
      "learning_rate": 7.208772608526293e-06,
      "loss": 0.8007,
      "step": 3131
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.923203465185541,
      "learning_rate": 7.202789946824227e-06,
      "loss": 0.7288,
      "step": 3132
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9452359938499277,
      "learning_rate": 7.1968083710954075e-06,
      "loss": 0.8058,
      "step": 3133
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5850309402104871,
      "learning_rate": 7.1908278836621e-06,
      "loss": 0.3685,
      "step": 3134
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9055282335536384,
      "learning_rate": 7.184848486846128e-06,
      "loss": 0.795,
      "step": 3135
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9172905897509362,
      "learning_rate": 7.178870182968904e-06,
      "loss": 0.7572,
      "step": 3136
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9552900395172296,
      "learning_rate": 7.1728929743514065e-06,
      "loss": 0.8179,
      "step": 3137
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8660663206621552,
      "learning_rate": 7.166916863314199e-06,
      "loss": 0.8319,
      "step": 3138
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0484727481864542,
      "learning_rate": 7.1609418521774095e-06,
      "loss": 0.8173,
      "step": 3139
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8799006242515753,
      "learning_rate": 7.154967943260748e-06,
      "loss": 0.8797,
      "step": 3140
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9428769864115999,
      "learning_rate": 7.148995138883483e-06,
      "loss": 0.8409,
      "step": 3141
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9860901750926063,
      "learning_rate": 7.143023441364471e-06,
      "loss": 0.8076,
      "step": 3142
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9224678596694819,
      "learning_rate": 7.13705285302213e-06,
      "loss": 0.7884,
      "step": 3143
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9420159704342941,
      "learning_rate": 7.131083376174441e-06,
      "loss": 0.8141,
      "step": 3144
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9533759226783683,
      "learning_rate": 7.125115013138966e-06,
      "loss": 0.8329,
      "step": 3145
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8288882442243611,
      "learning_rate": 7.119147766232832e-06,
      "loss": 0.7628,
      "step": 3146
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8581619046662188,
      "learning_rate": 7.113181637772721e-06,
      "loss": 0.8283,
      "step": 3147
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7558969545633211,
      "learning_rate": 7.107216630074895e-06,
      "loss": 0.7538,
      "step": 3148
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6051577577874107,
      "learning_rate": 7.1012527454551795e-06,
      "loss": 0.3254,
      "step": 3149
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8981387713321135,
      "learning_rate": 7.09528998622895e-06,
      "loss": 0.8487,
      "step": 3150
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8848781581232625,
      "learning_rate": 7.089328354711159e-06,
      "loss": 0.7647,
      "step": 3151
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8975894652266261,
      "learning_rate": 7.083367853216323e-06,
      "loss": 0.8497,
      "step": 3152
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9349171627394431,
      "learning_rate": 7.077408484058505e-06,
      "loss": 0.8236,
      "step": 3153
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8572295765911783,
      "learning_rate": 7.071450249551342e-06,
      "loss": 0.7455,
      "step": 3154
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8987878468792212,
      "learning_rate": 7.065493152008026e-06,
      "loss": 0.8483,
      "step": 3155
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.984013596700327,
      "learning_rate": 7.059537193741306e-06,
      "loss": 0.8211,
      "step": 3156
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9497561449428296,
      "learning_rate": 7.053582377063489e-06,
      "loss": 0.8096,
      "step": 3157
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8765278594334195,
      "learning_rate": 7.047628704286446e-06,
      "loss": 0.7914,
      "step": 3158
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9828199292123492,
      "learning_rate": 7.041676177721588e-06,
      "loss": 0.8824,
      "step": 3159
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6416491933710876,
      "learning_rate": 7.035724799679898e-06,
      "loss": 0.3578,
      "step": 3160
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.636282492374853,
      "learning_rate": 7.029774572471904e-06,
      "loss": 0.3603,
      "step": 3161
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9066438778566619,
      "learning_rate": 7.023825498407689e-06,
      "loss": 0.8178,
      "step": 3162
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8742298206217719,
      "learning_rate": 7.0178775797968855e-06,
      "loss": 0.8108,
      "step": 3163
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9946903039861719,
      "learning_rate": 7.011930818948688e-06,
      "loss": 0.8037,
      "step": 3164
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9612617734208126,
      "learning_rate": 7.005985218171825e-06,
      "loss": 0.851,
      "step": 3165
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.966188657665486,
      "learning_rate": 7.000040779774591e-06,
      "loss": 0.8173,
      "step": 3166
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9216213172772308,
      "learning_rate": 6.994097506064812e-06,
      "loss": 0.8282,
      "step": 3167
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8930368794002056,
      "learning_rate": 6.9881553993498805e-06,
      "loss": 0.7939,
      "step": 3168
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9305460509554879,
      "learning_rate": 6.9822144619367275e-06,
      "loss": 0.8492,
      "step": 3169
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8524376114065846,
      "learning_rate": 6.97627469613182e-06,
      "loss": 0.7974,
      "step": 3170
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0406046249729757,
      "learning_rate": 6.970336104241186e-06,
      "loss": 0.7696,
      "step": 3171
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9906862667257431,
      "learning_rate": 6.9643986885703955e-06,
      "loss": 0.7897,
      "step": 3172
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0108746149320018,
      "learning_rate": 6.958462451424547e-06,
      "loss": 0.8113,
      "step": 3173
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.93105851287248,
      "learning_rate": 6.952527395108302e-06,
      "loss": 0.7887,
      "step": 3174
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9210365263681604,
      "learning_rate": 6.9465935219258504e-06,
      "loss": 0.7986,
      "step": 3175
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6690018076603848,
      "learning_rate": 6.9406608341809215e-06,
      "loss": 0.3337,
      "step": 3176
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9415492708686202,
      "learning_rate": 6.934729334176793e-06,
      "loss": 0.7811,
      "step": 3177
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6000038770911127,
      "learning_rate": 6.928799024216282e-06,
      "loss": 0.3402,
      "step": 3178
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8697872649536379,
      "learning_rate": 6.92286990660173e-06,
      "loss": 0.7971,
      "step": 3179
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.813815806336729,
      "learning_rate": 6.91694198363503e-06,
      "loss": 0.7882,
      "step": 3180
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9605991428809861,
      "learning_rate": 6.911015257617606e-06,
      "loss": 0.8374,
      "step": 3181
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9690918425045214,
      "learning_rate": 6.905089730850416e-06,
      "loss": 0.8284,
      "step": 3182
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9565023127109424,
      "learning_rate": 6.8991654056339505e-06,
      "loss": 0.8017,
      "step": 3183
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9289513849794059,
      "learning_rate": 6.893242284268244e-06,
      "loss": 0.8116,
      "step": 3184
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0865727402378287,
      "learning_rate": 6.887320369052848e-06,
      "loss": 0.8217,
      "step": 3185
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9092106230007583,
      "learning_rate": 6.8813996622868584e-06,
      "loss": 0.8489,
      "step": 3186
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9144055966149908,
      "learning_rate": 6.8754801662688964e-06,
      "loss": 0.8113,
      "step": 3187
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.927032336459728,
      "learning_rate": 6.869561883297116e-06,
      "loss": 0.7806,
      "step": 3188
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8636642726704217,
      "learning_rate": 6.863644815669197e-06,
      "loss": 0.7926,
      "step": 3189
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9591955111052248,
      "learning_rate": 6.857728965682344e-06,
      "loss": 0.8398,
      "step": 3190
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9714302664194066,
      "learning_rate": 6.851814335633298e-06,
      "loss": 0.7581,
      "step": 3191
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9030154199525647,
      "learning_rate": 6.8459009278183275e-06,
      "loss": 0.8152,
      "step": 3192
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9168324910370076,
      "learning_rate": 6.839988744533211e-06,
      "loss": 0.7953,
      "step": 3193
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8574196135506319,
      "learning_rate": 6.834077788073268e-06,
      "loss": 0.7959,
      "step": 3194
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9110904564421373,
      "learning_rate": 6.828168060733336e-06,
      "loss": 0.7885,
      "step": 3195
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9220713806812779,
      "learning_rate": 6.822259564807768e-06,
      "loss": 0.8212,
      "step": 3196
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.931976936800054,
      "learning_rate": 6.81635230259045e-06,
      "loss": 0.8363,
      "step": 3197
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9456884464409329,
      "learning_rate": 6.810446276374789e-06,
      "loss": 0.7543,
      "step": 3198
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.954508640416024,
      "learning_rate": 6.8045414884536975e-06,
      "loss": 0.8331,
      "step": 3199
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8578020424207564,
      "learning_rate": 6.7986379411196255e-06,
      "loss": 0.8031,
      "step": 3200
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9000091811432456,
      "learning_rate": 6.7927356366645315e-06,
      "loss": 0.8191,
      "step": 3201
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8099969175886812,
      "learning_rate": 6.786834577379893e-06,
      "loss": 0.7936,
      "step": 3202
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8718813004392114,
      "learning_rate": 6.780934765556702e-06,
      "loss": 0.7352,
      "step": 3203
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8888419029260286,
      "learning_rate": 6.775036203485472e-06,
      "loss": 0.7915,
      "step": 3204
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9123380981002144,
      "learning_rate": 6.769138893456225e-06,
      "loss": 0.7965,
      "step": 3205
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8844771367494798,
      "learning_rate": 6.763242837758504e-06,
      "loss": 0.8211,
      "step": 3206
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8925071636973982,
      "learning_rate": 6.757348038681357e-06,
      "loss": 0.823,
      "step": 3207
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9167367220812779,
      "learning_rate": 6.751454498513349e-06,
      "loss": 0.7948,
      "step": 3208
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8732841079909026,
      "learning_rate": 6.745562219542554e-06,
      "loss": 0.8501,
      "step": 3209
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8659092307550642,
      "learning_rate": 6.7396712040565625e-06,
      "loss": 0.7383,
      "step": 3210
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8904974212371529,
      "learning_rate": 6.733781454342463e-06,
      "loss": 0.7702,
      "step": 3211
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9187076478659426,
      "learning_rate": 6.727892972686861e-06,
      "loss": 0.7586,
      "step": 3212
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7558986364993409,
      "learning_rate": 6.722005761375873e-06,
      "loss": 0.3506,
      "step": 3213
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9067390365628254,
      "learning_rate": 6.716119822695111e-06,
      "loss": 0.7928,
      "step": 3214
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8955731651458422,
      "learning_rate": 6.710235158929703e-06,
      "loss": 0.8309,
      "step": 3215
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8403216201485296,
      "learning_rate": 6.704351772364274e-06,
      "loss": 0.7998,
      "step": 3216
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9325988609200968,
      "learning_rate": 6.698469665282958e-06,
      "loss": 0.8581,
      "step": 3217
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9750997503197982,
      "learning_rate": 6.692588839969397e-06,
      "loss": 0.798,
      "step": 3218
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9307631040131953,
      "learning_rate": 6.6867092987067214e-06,
      "loss": 0.8449,
      "step": 3219
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9755306872626124,
      "learning_rate": 6.680831043777579e-06,
      "loss": 0.8166,
      "step": 3220
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8751647234842586,
      "learning_rate": 6.674954077464108e-06,
      "loss": 0.8558,
      "step": 3221
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8770431864451785,
      "learning_rate": 6.6690784020479484e-06,
      "loss": 0.8272,
      "step": 3222
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9300400249407006,
      "learning_rate": 6.6632040198102364e-06,
      "loss": 0.7942,
      "step": 3223
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8670749434193954,
      "learning_rate": 6.657330933031619e-06,
      "loss": 0.8103,
      "step": 3224
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9666032022779086,
      "learning_rate": 6.651459143992221e-06,
      "loss": 0.8713,
      "step": 3225
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8901458190216331,
      "learning_rate": 6.645588654971677e-06,
      "loss": 0.8175,
      "step": 3226
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8660964133462812,
      "learning_rate": 6.639719468249115e-06,
      "loss": 0.8416,
      "step": 3227
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8942404985322258,
      "learning_rate": 6.633851586103153e-06,
      "loss": 0.803,
      "step": 3228
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9007725186319743,
      "learning_rate": 6.627985010811903e-06,
      "loss": 0.7904,
      "step": 3229
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8712374695328513,
      "learning_rate": 6.622119744652977e-06,
      "loss": 0.7815,
      "step": 3230
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8400867609594589,
      "learning_rate": 6.616255789903467e-06,
      "loss": 0.7705,
      "step": 3231
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9558437515814884,
      "learning_rate": 6.610393148839964e-06,
      "loss": 0.8364,
      "step": 3232
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8895117201108962,
      "learning_rate": 6.6045318237385526e-06,
      "loss": 0.8811,
      "step": 3233
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9587795561810505,
      "learning_rate": 6.598671816874794e-06,
      "loss": 0.8171,
      "step": 3234
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.053771605216768,
      "learning_rate": 6.5928131305237465e-06,
      "loss": 0.7817,
      "step": 3235
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.882412777374006,
      "learning_rate": 6.586955766959958e-06,
      "loss": 0.8338,
      "step": 3236
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9341872635074979,
      "learning_rate": 6.581099728457451e-06,
      "loss": 0.7511,
      "step": 3237
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8826853094267656,
      "learning_rate": 6.5752450172897466e-06,
      "loss": 0.8438,
      "step": 3238
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9645201457937224,
      "learning_rate": 6.569391635729847e-06,
      "loss": 0.8376,
      "step": 3239
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9220114294828193,
      "learning_rate": 6.563539586050233e-06,
      "loss": 0.8043,
      "step": 3240
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0600513602308248,
      "learning_rate": 6.557688870522871e-06,
      "loss": 0.8777,
      "step": 3241
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9474321743708568,
      "learning_rate": 6.551839491419213e-06,
      "loss": 0.7813,
      "step": 3242
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8703379502172973,
      "learning_rate": 6.545991451010185e-06,
      "loss": 0.8006,
      "step": 3243
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9033540009325384,
      "learning_rate": 6.5401447515662065e-06,
      "loss": 0.7729,
      "step": 3244
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.887032353353209,
      "learning_rate": 6.5342993953571556e-06,
      "loss": 0.7709,
      "step": 3245
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8934866451976539,
      "learning_rate": 6.52845538465241e-06,
      "loss": 0.7632,
      "step": 3246
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9908279259143117,
      "learning_rate": 6.522612721720813e-06,
      "loss": 0.8479,
      "step": 3247
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9104114922061196,
      "learning_rate": 6.5167714088306865e-06,
      "loss": 0.818,
      "step": 3248
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8184528153807836,
      "learning_rate": 6.51093144824983e-06,
      "loss": 0.7847,
      "step": 3249
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8905535574385885,
      "learning_rate": 6.505092842245519e-06,
      "loss": 0.802,
      "step": 3250
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9223237179368532,
      "learning_rate": 6.499255593084498e-06,
      "loss": 0.8368,
      "step": 3251
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9054931442942447,
      "learning_rate": 6.493419703032991e-06,
      "loss": 0.8291,
      "step": 3252
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9559104623615382,
      "learning_rate": 6.487585174356691e-06,
      "loss": 0.8315,
      "step": 3253
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.006849691221201,
      "learning_rate": 6.481752009320761e-06,
      "loss": 0.8255,
      "step": 3254
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.908780429129069,
      "learning_rate": 6.4759202101898366e-06,
      "loss": 0.7806,
      "step": 3255
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.93427518541516,
      "learning_rate": 6.4700897792280285e-06,
      "loss": 0.8548,
      "step": 3256
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8806632771512825,
      "learning_rate": 6.464260718698902e-06,
      "loss": 0.8243,
      "step": 3257
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9257915670789789,
      "learning_rate": 6.458433030865503e-06,
      "loss": 0.8027,
      "step": 3258
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.931724522194878,
      "learning_rate": 6.452606717990346e-06,
      "loss": 0.7804,
      "step": 3259
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8616758757447319,
      "learning_rate": 6.4467817823354005e-06,
      "loss": 0.8624,
      "step": 3260
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.902345815646916,
      "learning_rate": 6.440958226162104e-06,
      "loss": 0.787,
      "step": 3261
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8944581957985375,
      "learning_rate": 6.43513605173137e-06,
      "loss": 0.7794,
      "step": 3262
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8694662753821185,
      "learning_rate": 6.4293152613035594e-06,
      "loss": 0.7761,
      "step": 3263
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8932002439423659,
      "learning_rate": 6.4234958571385095e-06,
      "loss": 0.7959,
      "step": 3264
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8318048457927059,
      "learning_rate": 6.4176778414955075e-06,
      "loss": 0.799,
      "step": 3265
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.889111674912736,
      "learning_rate": 6.4118612166333124e-06,
      "loss": 0.7423,
      "step": 3266
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.882760810050881,
      "learning_rate": 6.4060459848101354e-06,
      "loss": 0.8106,
      "step": 3267
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8863104426867531,
      "learning_rate": 6.400232148283651e-06,
      "loss": 0.7767,
      "step": 3268
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8575286869307077,
      "learning_rate": 6.3944197093109885e-06,
      "loss": 0.7907,
      "step": 3269
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9358100162670082,
      "learning_rate": 6.388608670148741e-06,
      "loss": 0.7503,
      "step": 3270
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9918062643233598,
      "learning_rate": 6.38279903305295e-06,
      "loss": 0.8656,
      "step": 3271
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9115631991045925,
      "learning_rate": 6.376990800279119e-06,
      "loss": 0.7917,
      "step": 3272
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9870131186602463,
      "learning_rate": 6.3711839740822035e-06,
      "loss": 0.8292,
      "step": 3273
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8973095757658632,
      "learning_rate": 6.3653785567166125e-06,
      "loss": 0.8419,
      "step": 3274
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9070851845368323,
      "learning_rate": 6.359574550436209e-06,
      "loss": 0.7619,
      "step": 3275
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8728332563545449,
      "learning_rate": 6.3537719574943105e-06,
      "loss": 0.8181,
      "step": 3276
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9003170147800488,
      "learning_rate": 6.347970780143678e-06,
      "loss": 0.7955,
      "step": 3277
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6334394473733783,
      "learning_rate": 6.342171020636533e-06,
      "loss": 0.3317,
      "step": 3278
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9048411131486348,
      "learning_rate": 6.336372681224543e-06,
      "loss": 0.7789,
      "step": 3279
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8867567375633325,
      "learning_rate": 6.330575764158819e-06,
      "loss": 0.8495,
      "step": 3280
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9818444083470453,
      "learning_rate": 6.324780271689923e-06,
      "loss": 0.8235,
      "step": 3281
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9675379600396952,
      "learning_rate": 6.318986206067872e-06,
      "loss": 0.8688,
      "step": 3282
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6286700888006959,
      "learning_rate": 6.313193569542113e-06,
      "loss": 0.3558,
      "step": 3283
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8707271621651945,
      "learning_rate": 6.30740236436155e-06,
      "loss": 0.8045,
      "step": 3284
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9218601991134071,
      "learning_rate": 6.301612592774533e-06,
      "loss": 0.7884,
      "step": 3285
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9997841188258512,
      "learning_rate": 6.295824257028844e-06,
      "loss": 0.7928,
      "step": 3286
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8809800808052786,
      "learning_rate": 6.290037359371717e-06,
      "loss": 0.849,
      "step": 3287
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8452200440412654,
      "learning_rate": 6.284251902049827e-06,
      "loss": 0.8515,
      "step": 3288
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8763766940425712,
      "learning_rate": 6.278467887309283e-06,
      "loss": 0.7847,
      "step": 3289
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0508913694613438,
      "learning_rate": 6.272685317395644e-06,
      "loss": 0.8182,
      "step": 3290
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8767991622452508,
      "learning_rate": 6.266904194553896e-06,
      "loss": 0.8032,
      "step": 3291
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9268928544759419,
      "learning_rate": 6.261124521028477e-06,
      "loss": 0.8226,
      "step": 3292
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8428892444761845,
      "learning_rate": 6.255346299063252e-06,
      "loss": 0.7986,
      "step": 3293
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8496009116463769,
      "learning_rate": 6.249569530901525e-06,
      "loss": 0.8048,
      "step": 3294
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9749669664125381,
      "learning_rate": 6.243794218786034e-06,
      "loss": 0.7832,
      "step": 3295
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0169311473800233,
      "learning_rate": 6.238020364958964e-06,
      "loss": 0.8586,
      "step": 3296
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9189426040001417,
      "learning_rate": 6.232247971661912e-06,
      "loss": 0.7978,
      "step": 3297
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9340136710386168,
      "learning_rate": 6.2264770411359256e-06,
      "loss": 0.7979,
      "step": 3298
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9138720335091015,
      "learning_rate": 6.22070757562148e-06,
      "loss": 0.795,
      "step": 3299
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9716345889202367,
      "learning_rate": 6.214939577358479e-06,
      "loss": 0.7802,
      "step": 3300
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9349474807288428,
      "learning_rate": 6.209173048586253e-06,
      "loss": 0.8324,
      "step": 3301
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9340918005030123,
      "learning_rate": 6.203407991543577e-06,
      "loss": 0.8245,
      "step": 3302
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9286487521673094,
      "learning_rate": 6.197644408468635e-06,
      "loss": 0.8649,
      "step": 3303
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8698365014841728,
      "learning_rate": 6.191882301599052e-06,
      "loss": 0.8437,
      "step": 3304
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8959531125405886,
      "learning_rate": 6.186121673171882e-06,
      "loss": 0.7832,
      "step": 3305
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.882357215422682,
      "learning_rate": 6.180362525423591e-06,
      "loss": 0.8508,
      "step": 3306
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9762991137562015,
      "learning_rate": 6.174604860590081e-06,
      "loss": 0.8251,
      "step": 3307
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9286888652178208,
      "learning_rate": 6.168848680906678e-06,
      "loss": 0.7752,
      "step": 3308
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8778693928450553,
      "learning_rate": 6.163093988608127e-06,
      "loss": 0.7823,
      "step": 3309
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9121598152165593,
      "learning_rate": 6.157340785928595e-06,
      "loss": 0.781,
      "step": 3310
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.895875094062653,
      "learning_rate": 6.151589075101681e-06,
      "loss": 0.804,
      "step": 3311
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9635748366712207,
      "learning_rate": 6.145838858360391e-06,
      "loss": 0.7874,
      "step": 3312
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9529711291818556,
      "learning_rate": 6.140090137937158e-06,
      "loss": 0.7643,
      "step": 3313
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9043731780847785,
      "learning_rate": 6.134342916063838e-06,
      "loss": 0.8321,
      "step": 3314
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9436861852035275,
      "learning_rate": 6.128597194971691e-06,
      "loss": 0.8097,
      "step": 3315
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9322680236628427,
      "learning_rate": 6.122852976891413e-06,
      "loss": 0.8076,
      "step": 3316
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9362462101328525,
      "learning_rate": 6.117110264053101e-06,
      "loss": 0.764,
      "step": 3317
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9751863669680978,
      "learning_rate": 6.111369058686276e-06,
      "loss": 0.7558,
      "step": 3318
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9407808394938816,
      "learning_rate": 6.105629363019875e-06,
      "loss": 0.7704,
      "step": 3319
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8735349583452319,
      "learning_rate": 6.099891179282242e-06,
      "loss": 0.6935,
      "step": 3320
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.660669833796213,
      "learning_rate": 6.094154509701133e-06,
      "loss": 0.3248,
      "step": 3321
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8661990347671281,
      "learning_rate": 6.088419356503732e-06,
      "loss": 0.7408,
      "step": 3322
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8438811092701023,
      "learning_rate": 6.082685721916612e-06,
      "loss": 0.8052,
      "step": 3323
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.911181012536647,
      "learning_rate": 6.076953608165772e-06,
      "loss": 0.8888,
      "step": 3324
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9097003076142177,
      "learning_rate": 6.07122301747662e-06,
      "loss": 0.8132,
      "step": 3325
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9656243365315978,
      "learning_rate": 6.065493952073961e-06,
      "loss": 0.8228,
      "step": 3326
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9225317639331806,
      "learning_rate": 6.0597664141820176e-06,
      "loss": 0.8431,
      "step": 3327
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8626745154290263,
      "learning_rate": 6.054040406024422e-06,
      "loss": 0.7484,
      "step": 3328
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9900188543037469,
      "learning_rate": 6.0483159298242e-06,
      "loss": 0.8479,
      "step": 3329
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.89004318319486,
      "learning_rate": 6.042592987803796e-06,
      "loss": 0.7661,
      "step": 3330
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8537589987581478,
      "learning_rate": 6.036871582185054e-06,
      "loss": 0.8186,
      "step": 3331
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9018405226081864,
      "learning_rate": 6.031151715189217e-06,
      "loss": 0.7854,
      "step": 3332
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9256650116334434,
      "learning_rate": 6.025433389036935e-06,
      "loss": 0.8084,
      "step": 3333
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9574120863449502,
      "learning_rate": 6.019716605948261e-06,
      "loss": 0.8269,
      "step": 3334
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0950142213855458,
      "learning_rate": 6.014001368142643e-06,
      "loss": 0.8681,
      "step": 3335
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.891170121039297,
      "learning_rate": 6.008287677838937e-06,
      "loss": 0.7244,
      "step": 3336
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9261214524269106,
      "learning_rate": 6.002575537255395e-06,
      "loss": 0.8642,
      "step": 3337
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9220124370557526,
      "learning_rate": 5.996864948609662e-06,
      "loss": 0.8452,
      "step": 3338
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6355360203336539,
      "learning_rate": 5.9911559141187924e-06,
      "loss": 0.3594,
      "step": 3339
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8527327026225408,
      "learning_rate": 5.9854484359992235e-06,
      "loss": 0.8592,
      "step": 3340
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9029944761387311,
      "learning_rate": 5.979742516466793e-06,
      "loss": 0.8335,
      "step": 3341
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0124535176535376,
      "learning_rate": 5.974038157736746e-06,
      "loss": 0.8083,
      "step": 3342
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8737212829590905,
      "learning_rate": 5.968335362023697e-06,
      "loss": 0.7959,
      "step": 3343
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9007047449960218,
      "learning_rate": 5.962634131541676e-06,
      "loss": 0.8511,
      "step": 3344
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.029732412956407,
      "learning_rate": 5.956934468504101e-06,
      "loss": 0.7916,
      "step": 3345
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0404809424993084,
      "learning_rate": 5.951236375123768e-06,
      "loss": 0.822,
      "step": 3346
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.944838049390634,
      "learning_rate": 5.945539853612876e-06,
      "loss": 0.8177,
      "step": 3347
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.980857432855144,
      "learning_rate": 5.939844906183016e-06,
      "loss": 0.8709,
      "step": 3348
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8987710477434153,
      "learning_rate": 5.934151535045156e-06,
      "loss": 0.8308,
      "step": 3349
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8701597603637885,
      "learning_rate": 5.92845974240966e-06,
      "loss": 0.7514,
      "step": 3350
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9088603332358378,
      "learning_rate": 5.922769530486283e-06,
      "loss": 0.8209,
      "step": 3351
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8580434589245047,
      "learning_rate": 5.917080901484156e-06,
      "loss": 0.7679,
      "step": 3352
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8845692274763326,
      "learning_rate": 5.9113938576118e-06,
      "loss": 0.7616,
      "step": 3353
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9098846165531439,
      "learning_rate": 5.905708401077128e-06,
      "loss": 0.8539,
      "step": 3354
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8760011393364391,
      "learning_rate": 5.900024534087421e-06,
      "loss": 0.7993,
      "step": 3355
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9090750549045444,
      "learning_rate": 5.894342258849355e-06,
      "loss": 0.8093,
      "step": 3356
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8804309817195937,
      "learning_rate": 5.88866157756899e-06,
      "loss": 0.7973,
      "step": 3357
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.915108115278716,
      "learning_rate": 5.882982492451757e-06,
      "loss": 0.8256,
      "step": 3358
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9162019065398838,
      "learning_rate": 5.877305005702471e-06,
      "loss": 0.7801,
      "step": 3359
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9627143850186335,
      "learning_rate": 5.871629119525335e-06,
      "loss": 0.8568,
      "step": 3360
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8470621108831629,
      "learning_rate": 5.865954836123915e-06,
      "loss": 0.7818,
      "step": 3361
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9685759093528372,
      "learning_rate": 5.860282157701167e-06,
      "loss": 0.7736,
      "step": 3362
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9507860074427731,
      "learning_rate": 5.854611086459423e-06,
      "loss": 0.7915,
      "step": 3363
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9548908272563116,
      "learning_rate": 5.8489416246003814e-06,
      "loss": 0.7887,
      "step": 3364
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.981132048103368,
      "learning_rate": 5.8432737743251315e-06,
      "loss": 0.8166,
      "step": 3365
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0219987616688238,
      "learning_rate": 5.8376075378341194e-06,
      "loss": 0.8037,
      "step": 3366
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9443061862389283,
      "learning_rate": 5.831942917327172e-06,
      "loss": 0.804,
      "step": 3367
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.862555961352647,
      "learning_rate": 5.826279915003503e-06,
      "loss": 0.805,
      "step": 3368
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9199634454531093,
      "learning_rate": 5.8206185330616725e-06,
      "loss": 0.8365,
      "step": 3369
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9833664628468438,
      "learning_rate": 5.814958773699625e-06,
      "loss": 0.8313,
      "step": 3370
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8827701510709564,
      "learning_rate": 5.809300639114683e-06,
      "loss": 0.7664,
      "step": 3371
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9751116966990292,
      "learning_rate": 5.803644131503516e-06,
      "loss": 0.7912,
      "step": 3372
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8571840058947209,
      "learning_rate": 5.797989253062186e-06,
      "loss": 0.8341,
      "step": 3373
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9327436448787725,
      "learning_rate": 5.792336005986105e-06,
      "loss": 0.8566,
      "step": 3374
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9352392450430022,
      "learning_rate": 5.786684392470064e-06,
      "loss": 0.7547,
      "step": 3375
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9230669688565332,
      "learning_rate": 5.781034414708208e-06,
      "loss": 0.8031,
      "step": 3376
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8897319703712987,
      "learning_rate": 5.775386074894058e-06,
      "loss": 0.7935,
      "step": 3377
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8970292170357786,
      "learning_rate": 5.769739375220489e-06,
      "loss": 0.807,
      "step": 3378
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9517916145267793,
      "learning_rate": 5.7640943178797445e-06,
      "loss": 0.8021,
      "step": 3379
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9937458254449809,
      "learning_rate": 5.7584509050634395e-06,
      "loss": 0.7632,
      "step": 3380
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8828336575492021,
      "learning_rate": 5.752809138962525e-06,
      "loss": 0.7964,
      "step": 3381
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9983055112993984,
      "learning_rate": 5.747169021767342e-06,
      "loss": 0.7495,
      "step": 3382
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8372494656725854,
      "learning_rate": 5.7415305556675805e-06,
      "loss": 0.826,
      "step": 3383
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9056266636424698,
      "learning_rate": 5.73589374285227e-06,
      "loss": 0.7794,
      "step": 3384
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9911819949411086,
      "learning_rate": 5.730258585509832e-06,
      "loss": 0.7591,
      "step": 3385
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.854883013311933,
      "learning_rate": 5.724625085828022e-06,
      "loss": 0.7829,
      "step": 3386
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9484227423053941,
      "learning_rate": 5.718993245993958e-06,
      "loss": 0.8408,
      "step": 3387
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9754039598977263,
      "learning_rate": 5.713363068194115e-06,
      "loss": 0.8005,
      "step": 3388
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8882355056175681,
      "learning_rate": 5.7077345546143235e-06,
      "loss": 0.7885,
      "step": 3389
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8820668707058634,
      "learning_rate": 5.702107707439766e-06,
      "loss": 0.7995,
      "step": 3390
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.898091044694131,
      "learning_rate": 5.6964825288549745e-06,
      "loss": 0.8401,
      "step": 3391
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9701391211181191,
      "learning_rate": 5.690859021043842e-06,
      "loss": 0.7336,
      "step": 3392
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8261018837822723,
      "learning_rate": 5.685237186189601e-06,
      "loss": 0.7749,
      "step": 3393
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9068747158712686,
      "learning_rate": 5.679617026474853e-06,
      "loss": 0.7499,
      "step": 3394
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8534281874635383,
      "learning_rate": 5.673998544081527e-06,
      "loss": 0.8061,
      "step": 3395
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.032854339408431,
      "learning_rate": 5.6683817411909114e-06,
      "loss": 0.8119,
      "step": 3396
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.618742281327215,
      "learning_rate": 5.662766619983653e-06,
      "loss": 0.3483,
      "step": 3397
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9101785517127509,
      "learning_rate": 5.65715318263972e-06,
      "loss": 0.7756,
      "step": 3398
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.964408151418888,
      "learning_rate": 5.651541431338454e-06,
      "loss": 0.8005,
      "step": 3399
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9455116351384658,
      "learning_rate": 5.645931368258527e-06,
      "loss": 0.8036,
      "step": 3400
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9496015067178616,
      "learning_rate": 5.640322995577958e-06,
      "loss": 0.7447,
      "step": 3401
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8710235732617739,
      "learning_rate": 5.634716315474109e-06,
      "loss": 0.8556,
      "step": 3402
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9114607545843209,
      "learning_rate": 5.629111330123689e-06,
      "loss": 0.777,
      "step": 3403
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9309342543515639,
      "learning_rate": 5.623508041702743e-06,
      "loss": 0.7663,
      "step": 3404
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9322504279841963,
      "learning_rate": 5.617906452386659e-06,
      "loss": 0.8256,
      "step": 3405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9759482154027382,
      "learning_rate": 5.612306564350179e-06,
      "loss": 0.8301,
      "step": 3406
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9587270467680283,
      "learning_rate": 5.6067083797673535e-06,
      "loss": 0.8053,
      "step": 3407
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8811855867214288,
      "learning_rate": 5.601111900811607e-06,
      "loss": 0.8267,
      "step": 3408
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9031059367742273,
      "learning_rate": 5.595517129655681e-06,
      "loss": 0.7479,
      "step": 3409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.974692289298073,
      "learning_rate": 5.589924068471648e-06,
      "loss": 0.8036,
      "step": 3410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.871538327427673,
      "learning_rate": 5.58433271943094e-06,
      "loss": 0.7169,
      "step": 3411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9673204535909509,
      "learning_rate": 5.578743084704306e-06,
      "loss": 0.7877,
      "step": 3412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9521404768537644,
      "learning_rate": 5.573155166461833e-06,
      "loss": 0.8066,
      "step": 3413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9095975341290748,
      "learning_rate": 5.567568966872947e-06,
      "loss": 0.8312,
      "step": 3414
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9646981307762109,
      "learning_rate": 5.5619844881064e-06,
      "loss": 0.8482,
      "step": 3415
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8430321860905127,
      "learning_rate": 5.556401732330281e-06,
      "loss": 0.7794,
      "step": 3416
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.941712217697136,
      "learning_rate": 5.550820701712007e-06,
      "loss": 0.8516,
      "step": 3417
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9829543215592875,
      "learning_rate": 5.545241398418326e-06,
      "loss": 0.8044,
      "step": 3418
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.899144007868431,
      "learning_rate": 5.539663824615312e-06,
      "loss": 0.7781,
      "step": 3419
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9774180434083056,
      "learning_rate": 5.534087982468384e-06,
      "loss": 0.7739,
      "step": 3420
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9069206088374879,
      "learning_rate": 5.5285138741422615e-06,
      "loss": 0.8164,
      "step": 3421
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9378248176044642,
      "learning_rate": 5.522941501801008e-06,
      "loss": 0.8363,
      "step": 3422
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8806183256148444,
      "learning_rate": 5.517370867608021e-06,
      "loss": 0.7577,
      "step": 3423
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9606268663431288,
      "learning_rate": 5.511801973725997e-06,
      "loss": 0.7607,
      "step": 3424
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9380407915653187,
      "learning_rate": 5.506234822316983e-06,
      "loss": 0.7846,
      "step": 3425
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9137854337047425,
      "learning_rate": 5.500669415542336e-06,
      "loss": 0.8711,
      "step": 3426
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9396315203775052,
      "learning_rate": 5.495105755562738e-06,
      "loss": 0.7964,
      "step": 3427
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9146193337600917,
      "learning_rate": 5.4895438445381945e-06,
      "loss": 0.7961,
      "step": 3428
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.908475190092035,
      "learning_rate": 5.48398368462803e-06,
      "loss": 0.815,
      "step": 3429
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.020731768149628,
      "learning_rate": 5.4784252779908905e-06,
      "loss": 0.8438,
      "step": 3430
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9175505848606937,
      "learning_rate": 5.4728686267847354e-06,
      "loss": 0.7992,
      "step": 3431
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9169769633698054,
      "learning_rate": 5.467313733166863e-06,
      "loss": 0.8052,
      "step": 3432
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0038247970644942,
      "learning_rate": 5.461760599293855e-06,
      "loss": 0.8051,
      "step": 3433
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8653677152507062,
      "learning_rate": 5.456209227321643e-06,
      "loss": 0.8265,
      "step": 3434
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9486996890624889,
      "learning_rate": 5.450659619405458e-06,
      "loss": 0.804,
      "step": 3435
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9388693094130365,
      "learning_rate": 5.445111777699842e-06,
      "loss": 0.7435,
      "step": 3436
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9087913395201153,
      "learning_rate": 5.439565704358667e-06,
      "loss": 0.7711,
      "step": 3437
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9105822513750034,
      "learning_rate": 5.434021401535105e-06,
      "loss": 0.827,
      "step": 3438
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9308333031630616,
      "learning_rate": 5.428478871381646e-06,
      "loss": 0.7438,
      "step": 3439
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8626148064755859,
      "learning_rate": 5.422938116050092e-06,
      "loss": 0.8242,
      "step": 3440
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9151045905306782,
      "learning_rate": 5.417399137691552e-06,
      "loss": 0.8389,
      "step": 3441
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6543084801883043,
      "learning_rate": 5.411861938456453e-06,
      "loss": 0.3724,
      "step": 3442
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8461259166119001,
      "learning_rate": 5.406326520494522e-06,
      "loss": 0.8004,
      "step": 3443
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9615025523577454,
      "learning_rate": 5.400792885954802e-06,
      "loss": 0.797,
      "step": 3444
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9502323684690408,
      "learning_rate": 5.395261036985635e-06,
      "loss": 0.8392,
      "step": 3445
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9570783452080148,
      "learning_rate": 5.389730975734686e-06,
      "loss": 0.8216,
      "step": 3446
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8878592206322244,
      "learning_rate": 5.384202704348902e-06,
      "loss": 0.8199,
      "step": 3447
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9766705143320804,
      "learning_rate": 5.378676224974557e-06,
      "loss": 0.8234,
      "step": 3448
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9054686846045996,
      "learning_rate": 5.373151539757224e-06,
      "loss": 0.7893,
      "step": 3449
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.947640347224495,
      "learning_rate": 5.367628650841761e-06,
      "loss": 0.8008,
      "step": 3450
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9461511998931234,
      "learning_rate": 5.362107560372358e-06,
      "loss": 0.7889,
      "step": 3451
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8817796818583945,
      "learning_rate": 5.356588270492487e-06,
      "loss": 0.7989,
      "step": 3452
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.962217513499361,
      "learning_rate": 5.351070783344926e-06,
      "loss": 0.8409,
      "step": 3453
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9112400635001678,
      "learning_rate": 5.3455551010717545e-06,
      "loss": 0.8517,
      "step": 3454
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9355391223821187,
      "learning_rate": 5.34004122581435e-06,
      "loss": 0.8404,
      "step": 3455
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9835868903951838,
      "learning_rate": 5.334529159713389e-06,
      "loss": 0.8645,
      "step": 3456
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9277106800315987,
      "learning_rate": 5.329018904908841e-06,
      "loss": 0.7794,
      "step": 3457
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7929018672445299,
      "learning_rate": 5.323510463539989e-06,
      "loss": 0.7265,
      "step": 3458
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.868961645969808,
      "learning_rate": 5.318003837745382e-06,
      "loss": 0.8328,
      "step": 3459
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8896624034587266,
      "learning_rate": 5.3124990296628974e-06,
      "loss": 0.7882,
      "step": 3460
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9008102066444923,
      "learning_rate": 5.306996041429688e-06,
      "loss": 0.8664,
      "step": 3461
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8608152388173771,
      "learning_rate": 5.301494875182192e-06,
      "loss": 0.7014,
      "step": 3462
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6592645101367504,
      "learning_rate": 5.295995533056162e-06,
      "loss": 0.3644,
      "step": 3463
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8778017480457997,
      "learning_rate": 5.290498017186631e-06,
      "loss": 0.8397,
      "step": 3464
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9561072168054056,
      "learning_rate": 5.2850023297079235e-06,
      "loss": 0.8558,
      "step": 3465
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9250278369840537,
      "learning_rate": 5.279508472753654e-06,
      "loss": 0.8164,
      "step": 3466
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9467760864877051,
      "learning_rate": 5.274016448456725e-06,
      "loss": 0.8158,
      "step": 3467
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8744642928004293,
      "learning_rate": 5.2685262589493314e-06,
      "loss": 0.8238,
      "step": 3468
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8853458112497895,
      "learning_rate": 5.263037906362953e-06,
      "loss": 0.7874,
      "step": 3469
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5833666811567016,
      "learning_rate": 5.257551392828359e-06,
      "loss": 0.3382,
      "step": 3470
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9112318079454849,
      "learning_rate": 5.252066720475597e-06,
      "loss": 0.7916,
      "step": 3471
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9463649051101968,
      "learning_rate": 5.246583891434018e-06,
      "loss": 0.8298,
      "step": 3472
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8401958654768457,
      "learning_rate": 5.241102907832232e-06,
      "loss": 0.8152,
      "step": 3473
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.896770151513034,
      "learning_rate": 5.235623771798151e-06,
      "loss": 0.8184,
      "step": 3474
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8672246686062405,
      "learning_rate": 5.23014648545897e-06,
      "loss": 0.8218,
      "step": 3475
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9404036273445235,
      "learning_rate": 5.224671050941146e-06,
      "loss": 0.8363,
      "step": 3476
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9487029378148641,
      "learning_rate": 5.2191974703704425e-06,
      "loss": 0.749,
      "step": 3477
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9045769401191277,
      "learning_rate": 5.213725745871889e-06,
      "loss": 0.803,
      "step": 3478
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8264469669186376,
      "learning_rate": 5.208255879569799e-06,
      "loss": 0.7457,
      "step": 3479
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9185659980472989,
      "learning_rate": 5.20278787358776e-06,
      "loss": 0.8071,
      "step": 3480
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8245316783284444,
      "learning_rate": 5.197321730048641e-06,
      "loss": 0.7831,
      "step": 3481
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8521134928658195,
      "learning_rate": 5.1918574510745865e-06,
      "loss": 0.7761,
      "step": 3482
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9242139693777036,
      "learning_rate": 5.186395038787017e-06,
      "loss": 0.8527,
      "step": 3483
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.864079650535489,
      "learning_rate": 5.180934495306638e-06,
      "loss": 0.7826,
      "step": 3484
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8520240742615859,
      "learning_rate": 5.175475822753404e-06,
      "loss": 0.8064,
      "step": 3485
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9232624447415759,
      "learning_rate": 5.170019023246574e-06,
      "loss": 0.8193,
      "step": 3486
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8912452380576795,
      "learning_rate": 5.16456409890466e-06,
      "loss": 0.7682,
      "step": 3487
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.904557263206528,
      "learning_rate": 5.159111051845451e-06,
      "loss": 0.8258,
      "step": 3488
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8525320100055634,
      "learning_rate": 5.153659884186013e-06,
      "loss": 0.8177,
      "step": 3489
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9092928205766065,
      "learning_rate": 5.148210598042665e-06,
      "loss": 0.8203,
      "step": 3490
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9443910236525055,
      "learning_rate": 5.142763195531017e-06,
      "loss": 0.8232,
      "step": 3491
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9477203016548692,
      "learning_rate": 5.137317678765939e-06,
      "loss": 0.8163,
      "step": 3492
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0620438325345511,
      "learning_rate": 5.131874049861563e-06,
      "loss": 0.8141,
      "step": 3493
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8662926436179214,
      "learning_rate": 5.126432310931295e-06,
      "loss": 0.8211,
      "step": 3494
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8990758272571899,
      "learning_rate": 5.120992464087807e-06,
      "loss": 0.7766,
      "step": 3495
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9287082167414092,
      "learning_rate": 5.115554511443033e-06,
      "loss": 0.8148,
      "step": 3496
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9362115773326014,
      "learning_rate": 5.1101184551081705e-06,
      "loss": 0.7793,
      "step": 3497
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8888374360410692,
      "learning_rate": 5.104684297193694e-06,
      "loss": 0.7605,
      "step": 3498
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9571754788447719,
      "learning_rate": 5.099252039809317e-06,
      "loss": 0.827,
      "step": 3499
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8730204980127758,
      "learning_rate": 5.09382168506404e-06,
      "loss": 0.8287,
      "step": 3500
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9731097861162947,
      "learning_rate": 5.088393235066114e-06,
      "loss": 0.7562,
      "step": 3501
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8752357040913773,
      "learning_rate": 5.082966691923037e-06,
      "loss": 0.7671,
      "step": 3502
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9174707512698735,
      "learning_rate": 5.077542057741592e-06,
      "loss": 0.7711,
      "step": 3503
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9277825862504434,
      "learning_rate": 5.0721193346278066e-06,
      "loss": 0.7992,
      "step": 3504
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.912713674913226,
      "learning_rate": 5.066698524686966e-06,
      "loss": 0.8607,
      "step": 3505
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0073120292808082,
      "learning_rate": 5.061279630023618e-06,
      "loss": 0.8015,
      "step": 3506
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9488095639762408,
      "learning_rate": 5.055862652741562e-06,
      "loss": 0.8194,
      "step": 3507
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9726314995146373,
      "learning_rate": 5.050447594943856e-06,
      "loss": 0.8144,
      "step": 3508
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9042849421486261,
      "learning_rate": 5.045034458732808e-06,
      "loss": 0.7698,
      "step": 3509
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6824861420826083,
      "learning_rate": 5.0396232462099945e-06,
      "loss": 0.3582,
      "step": 3510
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0003169157579999,
      "learning_rate": 5.034213959476222e-06,
      "loss": 0.8342,
      "step": 3511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0065627976338345,
      "learning_rate": 5.028806600631569e-06,
      "loss": 0.8045,
      "step": 3512
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9255491536171078,
      "learning_rate": 5.023401171775357e-06,
      "loss": 0.7626,
      "step": 3513
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9707978689750395,
      "learning_rate": 5.017997675006161e-06,
      "loss": 0.8206,
      "step": 3514
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8802185378656407,
      "learning_rate": 5.012596112421806e-06,
      "loss": 0.7845,
      "step": 3515
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9250148561571797,
      "learning_rate": 5.007196486119355e-06,
      "loss": 0.8088,
      "step": 3516
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9023258226283749,
      "learning_rate": 5.001798798195136e-06,
      "loss": 0.7686,
      "step": 3517
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9768138526263169,
      "learning_rate": 4.996403050744719e-06,
      "loss": 0.8235,
      "step": 3518
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9681630080659644,
      "learning_rate": 4.991009245862917e-06,
      "loss": 0.8209,
      "step": 3519
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.039020045948542,
      "learning_rate": 4.985617385643789e-06,
      "loss": 0.864,
      "step": 3520
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8922324691671892,
      "learning_rate": 4.980227472180643e-06,
      "loss": 0.7373,
      "step": 3521
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8965371248572804,
      "learning_rate": 4.974839507566027e-06,
      "loss": 0.7774,
      "step": 3522
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9487068230504054,
      "learning_rate": 4.969453493891733e-06,
      "loss": 0.7939,
      "step": 3523
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9241599433151817,
      "learning_rate": 4.9640694332488075e-06,
      "loss": 0.7515,
      "step": 3524
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.90512482790068,
      "learning_rate": 4.958687327727511e-06,
      "loss": 0.8281,
      "step": 3525
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.859048884189087,
      "learning_rate": 4.953307179417376e-06,
      "loss": 0.7448,
      "step": 3526
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8634631451569753,
      "learning_rate": 4.947928990407156e-06,
      "loss": 0.7586,
      "step": 3527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.018390626646393,
      "learning_rate": 4.94255276278485e-06,
      "loss": 0.842,
      "step": 3528
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5949339489391032,
      "learning_rate": 4.937178498637696e-06,
      "loss": 0.3188,
      "step": 3529
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.914784110066213,
      "learning_rate": 4.931806200052165e-06,
      "loss": 0.8198,
      "step": 3530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0169501688304254,
      "learning_rate": 4.926435869113971e-06,
      "loss": 0.7822,
      "step": 3531
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6088923352310568,
      "learning_rate": 4.92106750790806e-06,
      "loss": 0.3255,
      "step": 3532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0262465223445916,
      "learning_rate": 4.915701118518616e-06,
      "loss": 0.8166,
      "step": 3533
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8970856396833505,
      "learning_rate": 4.910336703029055e-06,
      "loss": 0.8038,
      "step": 3534
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9355097487823956,
      "learning_rate": 4.904974263522025e-06,
      "loss": 0.84,
      "step": 3535
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9209814605626327,
      "learning_rate": 4.899613802079419e-06,
      "loss": 0.7655,
      "step": 3536
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.908146752574269,
      "learning_rate": 4.8942553207823395e-06,
      "loss": 0.839,
      "step": 3537
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9284450807822859,
      "learning_rate": 4.888898821711144e-06,
      "loss": 0.7736,
      "step": 3538
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8120428166539175,
      "learning_rate": 4.883544306945407e-06,
      "loss": 0.8226,
      "step": 3539
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9922820509892123,
      "learning_rate": 4.878191778563934e-06,
      "loss": 0.7867,
      "step": 3540
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8905850684370319,
      "learning_rate": 4.872841238644766e-06,
      "loss": 0.7854,
      "step": 3541
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9207545730123466,
      "learning_rate": 4.867492689265154e-06,
      "loss": 0.8279,
      "step": 3542
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9042018420509628,
      "learning_rate": 4.8621461325016015e-06,
      "loss": 0.7767,
      "step": 3543
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8777259162139578,
      "learning_rate": 4.856801570429822e-06,
      "loss": 0.8524,
      "step": 3544
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9112646244190127,
      "learning_rate": 4.851459005124759e-06,
      "loss": 0.8068,
      "step": 3545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8537041150604617,
      "learning_rate": 4.846118438660578e-06,
      "loss": 0.7746,
      "step": 3546
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9354925972368501,
      "learning_rate": 4.840779873110675e-06,
      "loss": 0.8103,
      "step": 3547
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6547545650627044,
      "learning_rate": 4.83544331054766e-06,
      "loss": 0.3032,
      "step": 3548
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.878240699836159,
      "learning_rate": 4.83010875304337e-06,
      "loss": 0.8151,
      "step": 3549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.924346518038654,
      "learning_rate": 4.824776202668875e-06,
      "loss": 0.8125,
      "step": 3550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8873331957009043,
      "learning_rate": 4.819445661494437e-06,
      "loss": 0.7699,
      "step": 3551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8882854777049307,
      "learning_rate": 4.8141171315895694e-06,
      "loss": 0.796,
      "step": 3552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.873052710768591,
      "learning_rate": 4.808790615022987e-06,
      "loss": 0.8454,
      "step": 3553
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9021887207411133,
      "learning_rate": 4.803466113862626e-06,
      "loss": 0.8474,
      "step": 3554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.933100199073192,
      "learning_rate": 4.798143630175642e-06,
      "loss": 0.7431,
      "step": 3555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.928738261973101,
      "learning_rate": 4.792823166028405e-06,
      "loss": 0.8095,
      "step": 3556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9906330212625959,
      "learning_rate": 4.787504723486505e-06,
      "loss": 0.8347,
      "step": 3557
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9197362561412737,
      "learning_rate": 4.7821883046147414e-06,
      "loss": 0.8395,
      "step": 3558
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.89888288249482,
      "learning_rate": 4.776873911477133e-06,
      "loss": 0.8332,
      "step": 3559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0136943907203282,
      "learning_rate": 4.771561546136908e-06,
      "loss": 0.7976,
      "step": 3560
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9409951403307463,
      "learning_rate": 4.766251210656509e-06,
      "loss": 0.7861,
      "step": 3561
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8952214223516566,
      "learning_rate": 4.760942907097601e-06,
      "loss": 0.7775,
      "step": 3562
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6490693416126403,
      "learning_rate": 4.755636637521035e-06,
      "loss": 0.3421,
      "step": 3563
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9937185272463941,
      "learning_rate": 4.750332403986902e-06,
      "loss": 0.8013,
      "step": 3564
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8897242991359666,
      "learning_rate": 4.7450302085544735e-06,
      "loss": 0.7703,
      "step": 3565
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8241155348462704,
      "learning_rate": 4.739730053282255e-06,
      "loss": 0.7934,
      "step": 3566
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8750376855555766,
      "learning_rate": 4.734431940227951e-06,
      "loss": 0.8056,
      "step": 3567
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9988109323223237,
      "learning_rate": 4.7291358714484594e-06,
      "loss": 0.8353,
      "step": 3568
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9419329518243553,
      "learning_rate": 4.723841848999907e-06,
      "loss": 0.8019,
      "step": 3569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0321395851138286,
      "learning_rate": 4.718549874937612e-06,
      "loss": 0.8252,
      "step": 3570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8787259657387192,
      "learning_rate": 4.713259951316103e-06,
      "loss": 0.7555,
      "step": 3571
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9039355617238541,
      "learning_rate": 4.707972080189106e-06,
      "loss": 0.7569,
      "step": 3572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9936258589005653,
      "learning_rate": 4.702686263609559e-06,
      "loss": 0.7826,
      "step": 3573
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9200535717167393,
      "learning_rate": 4.697402503629596e-06,
      "loss": 0.7632,
      "step": 3574
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9837403837004961,
      "learning_rate": 4.69212080230055e-06,
      "loss": 0.8365,
      "step": 3575
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9497066757672191,
      "learning_rate": 4.686841161672974e-06,
      "loss": 0.845,
      "step": 3576
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9858311661729944,
      "learning_rate": 4.681563583796587e-06,
      "loss": 0.8313,
      "step": 3577
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8787251696543732,
      "learning_rate": 4.67628807072034e-06,
      "loss": 0.7626,
      "step": 3578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9918358275301637,
      "learning_rate": 4.6710146244923645e-06,
      "loss": 0.8367,
      "step": 3579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9079490784424992,
      "learning_rate": 4.665743247159995e-06,
      "loss": 0.8048,
      "step": 3580
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8533579747047433,
      "learning_rate": 4.660473940769761e-06,
      "loss": 0.8047,
      "step": 3581
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9477698812656608,
      "learning_rate": 4.655206707367388e-06,
      "loss": 0.7824,
      "step": 3582
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8833350034719417,
      "learning_rate": 4.649941548997797e-06,
      "loss": 0.8221,
      "step": 3583
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8892541482978952,
      "learning_rate": 4.644678467705101e-06,
      "loss": 0.7873,
      "step": 3584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8513485534210088,
      "learning_rate": 4.639417465532622e-06,
      "loss": 0.7624,
      "step": 3585
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.895645065953611,
      "learning_rate": 4.634158544522849e-06,
      "loss": 0.7924,
      "step": 3586
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8493424571799859,
      "learning_rate": 4.628901706717476e-06,
      "loss": 0.8041,
      "step": 3587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9275924126025425,
      "learning_rate": 4.623646954157399e-06,
      "loss": 0.7778,
      "step": 3588
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8991458083251213,
      "learning_rate": 4.618394288882681e-06,
      "loss": 0.8278,
      "step": 3589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8973050498376289,
      "learning_rate": 4.613143712932603e-06,
      "loss": 0.7944,
      "step": 3590
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6149458178222174,
      "learning_rate": 4.607895228345603e-06,
      "loss": 0.3731,
      "step": 3591
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8597382754354694,
      "learning_rate": 4.602648837159333e-06,
      "loss": 0.7728,
      "step": 3592
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6275930301242421,
      "learning_rate": 4.597404541410622e-06,
      "loss": 0.3561,
      "step": 3593
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9315085799830811,
      "learning_rate": 4.592162343135483e-06,
      "loss": 0.785,
      "step": 3594
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9053612888119205,
      "learning_rate": 4.586922244369122e-06,
      "loss": 0.792,
      "step": 3595
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9101265901943261,
      "learning_rate": 4.5816842471459224e-06,
      "loss": 0.754,
      "step": 3596
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9185673315618529,
      "learning_rate": 4.576448353499457e-06,
      "loss": 0.8327,
      "step": 3597
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9206735882828124,
      "learning_rate": 4.571214565462477e-06,
      "loss": 0.7472,
      "step": 3598
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8604700861216604,
      "learning_rate": 4.565982885066923e-06,
      "loss": 0.7257,
      "step": 3599
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.959922521750005,
      "learning_rate": 4.560753314343912e-06,
      "loss": 0.7899,
      "step": 3600
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8380662945403673,
      "learning_rate": 4.555525855323738e-06,
      "loss": 0.7505,
      "step": 3601
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8593728332329658,
      "learning_rate": 4.5503005100358945e-06,
      "loss": 0.7436,
      "step": 3602
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6004219631874937,
      "learning_rate": 4.545077280509022e-06,
      "loss": 0.3438,
      "step": 3603
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.585872915801442,
      "learning_rate": 4.539856168770974e-06,
      "loss": 0.3319,
      "step": 3604
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9333732594749871,
      "learning_rate": 4.534637176848758e-06,
      "loss": 0.8068,
      "step": 3605
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9884775038972663,
      "learning_rate": 4.52942030676857e-06,
      "loss": 0.8615,
      "step": 3606
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9259382715336266,
      "learning_rate": 4.524205560555774e-06,
      "loss": 0.7299,
      "step": 3607
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8878568014811203,
      "learning_rate": 4.5189929402349175e-06,
      "loss": 0.7691,
      "step": 3608
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9066665773332193,
      "learning_rate": 4.513782447829717e-06,
      "loss": 0.8155,
      "step": 3609
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9413819175114126,
      "learning_rate": 4.508574085363065e-06,
      "loss": 0.7958,
      "step": 3610
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9702884777398391,
      "learning_rate": 4.503367854857035e-06,
      "loss": 0.8292,
      "step": 3611
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8182835525299927,
      "learning_rate": 4.498163758332853e-06,
      "loss": 0.7148,
      "step": 3612
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9176688077096794,
      "learning_rate": 4.492961797810932e-06,
      "loss": 0.8434,
      "step": 3613
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9168581859598011,
      "learning_rate": 4.4877619753108605e-06,
      "loss": 0.7397,
      "step": 3614
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9034682112837701,
      "learning_rate": 4.4825642928513746e-06,
      "loss": 0.7614,
      "step": 3615
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9744747496270048,
      "learning_rate": 4.477368752450409e-06,
      "loss": 0.8085,
      "step": 3616
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9314101928112091,
      "learning_rate": 4.472175356125036e-06,
      "loss": 0.7694,
      "step": 3617
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8680379081826342,
      "learning_rate": 4.466984105891521e-06,
      "loss": 0.7638,
      "step": 3618
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.868378303432429,
      "learning_rate": 4.461795003765285e-06,
      "loss": 0.8564,
      "step": 3619
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9261369258371421,
      "learning_rate": 4.456608051760914e-06,
      "loss": 0.8165,
      "step": 3620
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.866036217791606,
      "learning_rate": 4.45142325189216e-06,
      "loss": 0.7998,
      "step": 3621
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9039901904404615,
      "learning_rate": 4.446240606171945e-06,
      "loss": 0.8024,
      "step": 3622
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9656314131153821,
      "learning_rate": 4.4410601166123475e-06,
      "loss": 0.8083,
      "step": 3623
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6380904234753876,
      "learning_rate": 4.4358817852246124e-06,
      "loss": 0.3389,
      "step": 3624
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8913880571237748,
      "learning_rate": 4.430705614019147e-06,
      "loss": 0.7471,
      "step": 3625
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0130656541408485,
      "learning_rate": 4.425531605005519e-06,
      "loss": 0.8495,
      "step": 3626
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9103761749289868,
      "learning_rate": 4.420359760192452e-06,
      "loss": 0.819,
      "step": 3627
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9281462968313929,
      "learning_rate": 4.4151900815878455e-06,
      "loss": 0.75,
      "step": 3628
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9556308182735098,
      "learning_rate": 4.410022571198734e-06,
      "loss": 0.8045,
      "step": 3629
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8975453080273252,
      "learning_rate": 4.404857231031332e-06,
      "loss": 0.7881,
      "step": 3630
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9233142211511964,
      "learning_rate": 4.399694063090999e-06,
      "loss": 0.8415,
      "step": 3631
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9603615661656331,
      "learning_rate": 4.394533069382255e-06,
      "loss": 0.8504,
      "step": 3632
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.922872252333285,
      "learning_rate": 4.3893742519087754e-06,
      "loss": 0.7669,
      "step": 3633
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8777963385336816,
      "learning_rate": 4.3842176126733914e-06,
      "loss": 0.803,
      "step": 3634
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9334348649130485,
      "learning_rate": 4.379063153678087e-06,
      "loss": 0.7847,
      "step": 3635
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9088300334531552,
      "learning_rate": 4.373910876923997e-06,
      "loss": 0.8533,
      "step": 3636
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9101690545480922,
      "learning_rate": 4.368760784411423e-06,
      "loss": 0.7702,
      "step": 3637
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9387196394488951,
      "learning_rate": 4.363612878139799e-06,
      "loss": 0.7881,
      "step": 3638
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9286249547911235,
      "learning_rate": 4.3584671601077224e-06,
      "loss": 0.8414,
      "step": 3639
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9583430645385345,
      "learning_rate": 4.353323632312938e-06,
      "loss": 0.8379,
      "step": 3640
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9500515528642419,
      "learning_rate": 4.348182296752336e-06,
      "loss": 0.8342,
      "step": 3641
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8657687131105768,
      "learning_rate": 4.343043155421971e-06,
      "loss": 0.8879,
      "step": 3642
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9660031621210414,
      "learning_rate": 4.3379062103170214e-06,
      "loss": 0.819,
      "step": 3643
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9256824111689461,
      "learning_rate": 4.332771463431837e-06,
      "loss": 0.7824,
      "step": 3644
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8636521860269754,
      "learning_rate": 4.327638916759898e-06,
      "loss": 0.785,
      "step": 3645
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9726247267708757,
      "learning_rate": 4.322508572293836e-06,
      "loss": 0.7961,
      "step": 3646
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9232528983231846,
      "learning_rate": 4.317380432025428e-06,
      "loss": 0.8001,
      "step": 3647
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9241353745371422,
      "learning_rate": 4.312254497945595e-06,
      "loss": 0.6901,
      "step": 3648
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8842170536535531,
      "learning_rate": 4.3071307720444015e-06,
      "loss": 0.7722,
      "step": 3649
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9118254116662003,
      "learning_rate": 4.3020092563110485e-06,
      "loss": 0.7897,
      "step": 3650
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0050365828036085,
      "learning_rate": 4.2968899527338984e-06,
      "loss": 0.7827,
      "step": 3651
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8358463304205329,
      "learning_rate": 4.291772863300428e-06,
      "loss": 0.7935,
      "step": 3652
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9873234211033919,
      "learning_rate": 4.2866579899972686e-06,
      "loss": 0.8469,
      "step": 3653
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.952483116042539,
      "learning_rate": 4.281545334810201e-06,
      "loss": 0.7686,
      "step": 3654
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8987826416507552,
      "learning_rate": 4.276434899724119e-06,
      "loss": 0.8545,
      "step": 3655
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8781773983647108,
      "learning_rate": 4.27132668672308e-06,
      "loss": 0.7618,
      "step": 3656
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8159148205057767,
      "learning_rate": 4.266220697790266e-06,
      "loss": 0.8108,
      "step": 3657
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9357114078021441,
      "learning_rate": 4.2611169349079985e-06,
      "loss": 0.8327,
      "step": 3658
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8917744958198338,
      "learning_rate": 4.25601540005773e-06,
      "loss": 0.8198,
      "step": 3659
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9870978009177905,
      "learning_rate": 4.250916095220056e-06,
      "loss": 0.819,
      "step": 3660
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0435188420856156,
      "learning_rate": 4.2458190223747e-06,
      "loss": 0.7842,
      "step": 3661
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.862209377848389,
      "learning_rate": 4.240724183500518e-06,
      "loss": 0.8388,
      "step": 3662
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0620375284807644,
      "learning_rate": 4.2356315805755135e-06,
      "loss": 0.8509,
      "step": 3663
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9131549350148387,
      "learning_rate": 4.230541215576798e-06,
      "loss": 0.8257,
      "step": 3664
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9534533205856246,
      "learning_rate": 4.225453090480631e-06,
      "loss": 0.8449,
      "step": 3665
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6653481285542087,
      "learning_rate": 4.220367207262398e-06,
      "loss": 0.3663,
      "step": 3666
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9148842560788213,
      "learning_rate": 4.21528356789661e-06,
      "loss": 0.7982,
      "step": 3667
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8977869512349389,
      "learning_rate": 4.210202174356922e-06,
      "loss": 0.8341,
      "step": 3668
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8699440262020789,
      "learning_rate": 4.20512302861609e-06,
      "loss": 0.7618,
      "step": 3669
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9504060663434063,
      "learning_rate": 4.2000461326460274e-06,
      "loss": 0.7921,
      "step": 3670
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9650084873975309,
      "learning_rate": 4.194971488417753e-06,
      "loss": 0.8109,
      "step": 3671
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9612276758066001,
      "learning_rate": 4.189899097901421e-06,
      "loss": 0.853,
      "step": 3672
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9753365410291932,
      "learning_rate": 4.184828963066305e-06,
      "loss": 0.8317,
      "step": 3673
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9256262997886414,
      "learning_rate": 4.179761085880809e-06,
      "loss": 0.7914,
      "step": 3674
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.945854121502618,
      "learning_rate": 4.174695468312456e-06,
      "loss": 0.7659,
      "step": 3675
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.926474963197543,
      "learning_rate": 4.16963211232789e-06,
      "loss": 0.7758,
      "step": 3676
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.932440120679715,
      "learning_rate": 4.16457101989289e-06,
      "loss": 0.8317,
      "step": 3677
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.92045401106965,
      "learning_rate": 4.159512192972337e-06,
      "loss": 0.8137,
      "step": 3678
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8860107161244485,
      "learning_rate": 4.15445563353024e-06,
      "loss": 0.7733,
      "step": 3679
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5734237215102803,
      "learning_rate": 4.149401343529742e-06,
      "loss": 0.3179,
      "step": 3680
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8490827909893586,
      "learning_rate": 4.144349324933077e-06,
      "loss": 0.7535,
      "step": 3681
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8837897132112416,
      "learning_rate": 4.139299579701623e-06,
      "loss": 0.7655,
      "step": 3682
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0005654743448777,
      "learning_rate": 4.134252109795863e-06,
      "loss": 0.7727,
      "step": 3683
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9208829784663308,
      "learning_rate": 4.129206917175397e-06,
      "loss": 0.8256,
      "step": 3684
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8631999475936971,
      "learning_rate": 4.124164003798944e-06,
      "loss": 0.8164,
      "step": 3685
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9823982781501832,
      "learning_rate": 4.119123371624335e-06,
      "loss": 0.8101,
      "step": 3686
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9125191229700969,
      "learning_rate": 4.114085022608517e-06,
      "loss": 0.7927,
      "step": 3687
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0763216016205124,
      "learning_rate": 4.109048958707552e-06,
      "loss": 0.8585,
      "step": 3688
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0665564784930084,
      "learning_rate": 4.104015181876613e-06,
      "loss": 0.8011,
      "step": 3689
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9682817555622907,
      "learning_rate": 4.09898369406998e-06,
      "loss": 0.7966,
      "step": 3690
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9920813874981601,
      "learning_rate": 4.0939544972410636e-06,
      "loss": 0.8302,
      "step": 3691
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8417303281399883,
      "learning_rate": 4.0889275933423576e-06,
      "loss": 0.7786,
      "step": 3692
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8693865934318991,
      "learning_rate": 4.0839029843254815e-06,
      "loss": 0.7577,
      "step": 3693
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8817119981264545,
      "learning_rate": 4.078880672141171e-06,
      "loss": 0.7547,
      "step": 3694
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9172642733774574,
      "learning_rate": 4.073860658739246e-06,
      "loss": 0.8466,
      "step": 3695
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9043850345975923,
      "learning_rate": 4.068842946068661e-06,
      "loss": 0.7537,
      "step": 3696
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8289374581123911,
      "learning_rate": 4.063827536077459e-06,
      "loss": 0.7688,
      "step": 3697
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9308200109601307,
      "learning_rate": 4.058814430712796e-06,
      "loss": 0.7828,
      "step": 3698
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9710713967349989,
      "learning_rate": 4.0538036319209325e-06,
      "loss": 0.8072,
      "step": 3699
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9222919465303355,
      "learning_rate": 4.0487951416472324e-06,
      "loss": 0.7951,
      "step": 3700
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9497942609859896,
      "learning_rate": 4.043788961836164e-06,
      "loss": 0.8294,
      "step": 3701
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.916153966719797,
      "learning_rate": 4.038785094431295e-06,
      "loss": 0.8081,
      "step": 3702
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.606837091672731,
      "learning_rate": 4.0337835413753116e-06,
      "loss": 0.3064,
      "step": 3703
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8971169321444822,
      "learning_rate": 4.0287843046099765e-06,
      "loss": 0.8056,
      "step": 3704
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.002399994653951,
      "learning_rate": 4.0237873860761645e-06,
      "loss": 0.8017,
      "step": 3705
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9784829915671664,
      "learning_rate": 4.018792787713865e-06,
      "loss": 0.7872,
      "step": 3706
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.878310973122219,
      "learning_rate": 4.013800511462135e-06,
      "loss": 0.7961,
      "step": 3707
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9125104313196607,
      "learning_rate": 4.008810559259162e-06,
      "loss": 0.7437,
      "step": 3708
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8455716479948326,
      "learning_rate": 4.003822933042213e-06,
      "loss": 0.7811,
      "step": 3709
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9038120543549172,
      "learning_rate": 3.998837634747655e-06,
      "loss": 0.768,
      "step": 3710
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9423285283379873,
      "learning_rate": 3.993854666310955e-06,
      "loss": 0.7798,
      "step": 3711
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9095119907221378,
      "learning_rate": 3.98887402966667e-06,
      "loss": 0.8394,
      "step": 3712
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9369767380288095,
      "learning_rate": 3.983895726748455e-06,
      "loss": 0.7798,
      "step": 3713
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9523016703621346,
      "learning_rate": 3.97891975948906e-06,
      "loss": 0.8194,
      "step": 3714
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9184163326390579,
      "learning_rate": 3.973946129820326e-06,
      "loss": 0.8045,
      "step": 3715
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9251730297631321,
      "learning_rate": 3.968974839673186e-06,
      "loss": 0.697,
      "step": 3716
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9292779732253466,
      "learning_rate": 3.964005890977672e-06,
      "loss": 0.8164,
      "step": 3717
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5710089424707564,
      "learning_rate": 3.9590392856628946e-06,
      "loss": 0.3278,
      "step": 3718
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8898314661724199,
      "learning_rate": 3.954075025657058e-06,
      "loss": 0.7706,
      "step": 3719
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.87234253615286,
      "learning_rate": 3.949113112887471e-06,
      "loss": 0.8263,
      "step": 3720
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8600752934408742,
      "learning_rate": 3.944153549280506e-06,
      "loss": 0.866,
      "step": 3721
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8569403047652074,
      "learning_rate": 3.939196336761645e-06,
      "loss": 0.755,
      "step": 3722
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8924574632180897,
      "learning_rate": 3.934241477255445e-06,
      "loss": 0.8319,
      "step": 3723
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9198942856083479,
      "learning_rate": 3.929288972685555e-06,
      "loss": 0.756,
      "step": 3724
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0177261230956407,
      "learning_rate": 3.924338824974705e-06,
      "loss": 0.8427,
      "step": 3725
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6122327927577137,
      "learning_rate": 3.919391036044715e-06,
      "loss": 0.3235,
      "step": 3726
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8280525306547116,
      "learning_rate": 3.914445607816486e-06,
      "loss": 0.7408,
      "step": 3727
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8672588844978508,
      "learning_rate": 3.909502542210001e-06,
      "loss": 0.8039,
      "step": 3728
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.864255984756994,
      "learning_rate": 3.904561841144338e-06,
      "loss": 0.7731,
      "step": 3729
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9400151514873779,
      "learning_rate": 3.899623506537635e-06,
      "loss": 0.8065,
      "step": 3730
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9852315240753249,
      "learning_rate": 3.894687540307127e-06,
      "loss": 0.7902,
      "step": 3731
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8990640436800402,
      "learning_rate": 3.8897539443691355e-06,
      "loss": 0.8721,
      "step": 3732
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9098844576452563,
      "learning_rate": 3.884822720639036e-06,
      "loss": 0.8353,
      "step": 3733
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9242506269564833,
      "learning_rate": 3.879893871031314e-06,
      "loss": 0.7731,
      "step": 3734
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9493022683767174,
      "learning_rate": 3.874967397459511e-06,
      "loss": 0.7777,
      "step": 3735
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9217183892577459,
      "learning_rate": 3.870043301836256e-06,
      "loss": 0.8072,
      "step": 3736
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0019758482869725,
      "learning_rate": 3.86512158607325e-06,
      "loss": 0.8425,
      "step": 3737
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.026430682460471,
      "learning_rate": 3.860202252081276e-06,
      "loss": 0.8243,
      "step": 3738
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9361247581704105,
      "learning_rate": 3.855285301770188e-06,
      "loss": 0.7831,
      "step": 3739
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8502776262459499,
      "learning_rate": 3.850370737048913e-06,
      "loss": 0.7659,
      "step": 3740
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9427799883870333,
      "learning_rate": 3.8454585598254565e-06,
      "loss": 0.7772,
      "step": 3741
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9083995556675819,
      "learning_rate": 3.840548772006891e-06,
      "loss": 0.7718,
      "step": 3742
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8536211079279141,
      "learning_rate": 3.835641375499375e-06,
      "loss": 0.8023,
      "step": 3743
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8118793229940751,
      "learning_rate": 3.830736372208118e-06,
      "loss": 0.8133,
      "step": 3744
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9553093599821824,
      "learning_rate": 3.8258337640374125e-06,
      "loss": 0.7878,
      "step": 3745
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9652633758350265,
      "learning_rate": 3.820933552890629e-06,
      "loss": 0.7594,
      "step": 3746
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8781064926104922,
      "learning_rate": 3.816035740670185e-06,
      "loss": 0.7862,
      "step": 3747
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9258789174957323,
      "learning_rate": 3.811140329277591e-06,
      "loss": 0.7963,
      "step": 3748
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6634780861171986,
      "learning_rate": 3.8062473206134088e-06,
      "loss": 0.3289,
      "step": 3749
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9397692972377151,
      "learning_rate": 3.8013567165772735e-06,
      "loss": 0.8277,
      "step": 3750
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8932426187805154,
      "learning_rate": 3.7964685190678874e-06,
      "loss": 0.8066,
      "step": 3751
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9280807863537107,
      "learning_rate": 3.7915827299830154e-06,
      "loss": 0.7868,
      "step": 3752
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9000539417822102,
      "learning_rate": 3.7866993512194895e-06,
      "loss": 0.765,
      "step": 3753
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0030551398281662,
      "learning_rate": 3.7818183846732024e-06,
      "loss": 0.7996,
      "step": 3754
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8552922009234679,
      "learning_rate": 3.776939832239125e-06,
      "loss": 0.666,
      "step": 3755
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9110787069972419,
      "learning_rate": 3.7720636958112623e-06,
      "loss": 0.745,
      "step": 3756
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9451176004286698,
      "learning_rate": 3.7671899772827113e-06,
      "loss": 0.8167,
      "step": 3757
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9289355565449846,
      "learning_rate": 3.7623186785456156e-06,
      "loss": 0.7577,
      "step": 3758
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8872738153308526,
      "learning_rate": 3.757449801491172e-06,
      "loss": 0.7649,
      "step": 3759
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8626710077450064,
      "learning_rate": 3.7525833480096575e-06,
      "loss": 0.688,
      "step": 3760
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9428667399771639,
      "learning_rate": 3.7477193199903903e-06,
      "loss": 0.8354,
      "step": 3761
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9602769003543302,
      "learning_rate": 3.7428577193217563e-06,
      "loss": 0.824,
      "step": 3762
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8411597840107657,
      "learning_rate": 3.737998547891195e-06,
      "loss": 0.7993,
      "step": 3763
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8988075050974379,
      "learning_rate": 3.7331418075852053e-06,
      "loss": 0.8333,
      "step": 3764
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0145411721531914,
      "learning_rate": 3.728287500289339e-06,
      "loss": 0.8284,
      "step": 3765
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9853916014758581,
      "learning_rate": 3.7234356278882076e-06,
      "loss": 0.8017,
      "step": 3766
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9014490595115382,
      "learning_rate": 3.718586192265473e-06,
      "loss": 0.7658,
      "step": 3767
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9003778719722925,
      "learning_rate": 3.7137391953038516e-06,
      "loss": 0.8618,
      "step": 3768
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.025560105713503,
      "learning_rate": 3.7088946388851223e-06,
      "loss": 0.7602,
      "step": 3769
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9307608879292972,
      "learning_rate": 3.7040525248901003e-06,
      "loss": 0.747,
      "step": 3770
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9563123377258909,
      "learning_rate": 3.6992128551986617e-06,
      "loss": 0.8459,
      "step": 3771
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9190706387767555,
      "learning_rate": 3.6943756316897406e-06,
      "loss": 0.7961,
      "step": 3772
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8424256681638376,
      "learning_rate": 3.6895408562413027e-06,
      "loss": 0.8106,
      "step": 3773
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9145276082744154,
      "learning_rate": 3.684708530730382e-06,
      "loss": 0.7943,
      "step": 3774
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.929616081250431,
      "learning_rate": 3.6798786570330526e-06,
      "loss": 0.7716,
      "step": 3775
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.916584371592113,
      "learning_rate": 3.6750512370244363e-06,
      "loss": 0.7816,
      "step": 3776
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9824248590196525,
      "learning_rate": 3.670226272578704e-06,
      "loss": 0.805,
      "step": 3777
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9700092612379301,
      "learning_rate": 3.6654037655690732e-06,
      "loss": 0.7754,
      "step": 3778
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9490639247332491,
      "learning_rate": 3.660583717867807e-06,
      "loss": 0.8308,
      "step": 3779
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9059340342641323,
      "learning_rate": 3.655766131346211e-06,
      "loss": 0.8629,
      "step": 3780
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9178368265643119,
      "learning_rate": 3.650951007874648e-06,
      "loss": 0.8466,
      "step": 3781
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.900150019714018,
      "learning_rate": 3.6461383493225012e-06,
      "loss": 0.8009,
      "step": 3782
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0095279918262572,
      "learning_rate": 3.6413281575582194e-06,
      "loss": 0.7727,
      "step": 3783
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0109422497236087,
      "learning_rate": 3.6365204344492867e-06,
      "loss": 0.8323,
      "step": 3784
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9615823108290004,
      "learning_rate": 3.6317151818622154e-06,
      "loss": 0.7457,
      "step": 3785
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9161166566724152,
      "learning_rate": 3.62691240166258e-06,
      "loss": 0.7797,
      "step": 3786
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.945462156155603,
      "learning_rate": 3.6221120957149826e-06,
      "loss": 0.8367,
      "step": 3787
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9543563333994236,
      "learning_rate": 3.617314265883066e-06,
      "loss": 0.7829,
      "step": 3788
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0266996561124555,
      "learning_rate": 3.612518914029515e-06,
      "loss": 0.8137,
      "step": 3789
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8472312766174432,
      "learning_rate": 3.6077260420160487e-06,
      "loss": 0.8122,
      "step": 3790
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9239697545637713,
      "learning_rate": 3.602935651703424e-06,
      "loss": 0.82,
      "step": 3791
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9393017828744792,
      "learning_rate": 3.598147744951438e-06,
      "loss": 0.7602,
      "step": 3792
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9183384626383398,
      "learning_rate": 3.5933623236189198e-06,
      "loss": 0.8076,
      "step": 3793
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9528738816201527,
      "learning_rate": 3.58857938956373e-06,
      "loss": 0.8105,
      "step": 3794
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6487987256150152,
      "learning_rate": 3.58379894464278e-06,
      "loss": 0.3208,
      "step": 3795
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9459386230554001,
      "learning_rate": 3.57902099071199e-06,
      "loss": 0.8165,
      "step": 3796
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0544731789956656,
      "learning_rate": 3.5742455296263346e-06,
      "loss": 0.8551,
      "step": 3797
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9074926371126291,
      "learning_rate": 3.569472563239814e-06,
      "loss": 0.8152,
      "step": 3798
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8814891432158416,
      "learning_rate": 3.5647020934054465e-06,
      "loss": 0.7172,
      "step": 3799
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9633827814806312,
      "learning_rate": 3.559934121975304e-06,
      "loss": 0.745,
      "step": 3800
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9187623265177074,
      "learning_rate": 3.5551686508004735e-06,
      "loss": 0.8608,
      "step": 3801
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9533466260745015,
      "learning_rate": 3.550405681731074e-06,
      "loss": 0.8167,
      "step": 3802
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9197254286263087,
      "learning_rate": 3.5456452166162547e-06,
      "loss": 0.8096,
      "step": 3803
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9384201136531252,
      "learning_rate": 3.540887257304193e-06,
      "loss": 0.7923,
      "step": 3804
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9788762606314427,
      "learning_rate": 3.5361318056420925e-06,
      "loss": 0.8089,
      "step": 3805
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0502880924081504,
      "learning_rate": 3.531378863476178e-06,
      "loss": 0.815,
      "step": 3806
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6044625249396228,
      "learning_rate": 3.5266284326517165e-06,
      "loss": 0.3214,
      "step": 3807
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.021363294721096,
      "learning_rate": 3.5218805150129755e-06,
      "loss": 0.8115,
      "step": 3808
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8652538096632651,
      "learning_rate": 3.5171351124032703e-06,
      "loss": 0.7773,
      "step": 3809
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6027687330969653,
      "learning_rate": 3.51239222666493e-06,
      "loss": 0.3396,
      "step": 3810
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9907357583747116,
      "learning_rate": 3.507651859639295e-06,
      "loss": 0.8129,
      "step": 3811
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9574845859379618,
      "learning_rate": 3.5029140131667493e-06,
      "loss": 0.8152,
      "step": 3812
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9074554461319645,
      "learning_rate": 3.4981786890866853e-06,
      "loss": 0.7702,
      "step": 3813
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9016039581007935,
      "learning_rate": 3.493445889237518e-06,
      "loss": 0.7884,
      "step": 3814
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8779094389255431,
      "learning_rate": 3.4887156154566847e-06,
      "loss": 0.8013,
      "step": 3815
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8477948882679076,
      "learning_rate": 3.4839878695806385e-06,
      "loss": 0.8188,
      "step": 3816
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.960423005196755,
      "learning_rate": 3.4792626534448547e-06,
      "loss": 0.8102,
      "step": 3817
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9369817058553301,
      "learning_rate": 3.4745399688838243e-06,
      "loss": 0.8577,
      "step": 3818
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9090765597256338,
      "learning_rate": 3.469819817731056e-06,
      "loss": 0.8098,
      "step": 3819
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.858659912614471,
      "learning_rate": 3.4651022018190715e-06,
      "loss": 0.7817,
      "step": 3820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9409851275183723,
      "learning_rate": 3.460387122979423e-06,
      "loss": 0.7983,
      "step": 3821
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.848147218636278,
      "learning_rate": 3.455674583042652e-06,
      "loss": 0.7956,
      "step": 3822
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.59638751281959,
      "learning_rate": 3.4509645838383386e-06,
      "loss": 0.353,
      "step": 3823
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8633221561570475,
      "learning_rate": 3.446257127195066e-06,
      "loss": 0.7814,
      "step": 3824
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9431908724288878,
      "learning_rate": 3.4415522149404233e-06,
      "loss": 0.802,
      "step": 3825
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8850187886534494,
      "learning_rate": 3.436849848901028e-06,
      "loss": 0.8212,
      "step": 3826
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8431163901720273,
      "learning_rate": 3.432150030902497e-06,
      "loss": 0.839,
      "step": 3827
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8773837687550976,
      "learning_rate": 3.427452762769462e-06,
      "loss": 0.8059,
      "step": 3828
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9859314852778939,
      "learning_rate": 3.4227580463255628e-06,
      "loss": 0.7959,
      "step": 3829
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6294670192712657,
      "learning_rate": 3.4180658833934523e-06,
      "loss": 0.343,
      "step": 3830
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9754084756472022,
      "learning_rate": 3.4133762757947873e-06,
      "loss": 0.8084,
      "step": 3831
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9111167689349098,
      "learning_rate": 3.4086892253502344e-06,
      "loss": 0.8064,
      "step": 3832
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9034039846537317,
      "learning_rate": 3.4040047338794756e-06,
      "loss": 0.7398,
      "step": 3833
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9424517808887491,
      "learning_rate": 3.3993228032011784e-06,
      "loss": 0.7823,
      "step": 3834
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8792269132362961,
      "learning_rate": 3.3946434351330415e-06,
      "loss": 0.7658,
      "step": 3835
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6259160805769047,
      "learning_rate": 3.3899666314917512e-06,
      "loss": 0.3337,
      "step": 3836
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8673002128445269,
      "learning_rate": 3.385292394093006e-06,
      "loss": 0.8087,
      "step": 3837
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8532834147709366,
      "learning_rate": 3.3806207247515068e-06,
      "loss": 0.8119,
      "step": 3838
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9975484398443187,
      "learning_rate": 3.375951625280948e-06,
      "loss": 0.7744,
      "step": 3839
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0220800764573161,
      "learning_rate": 3.3712850974940437e-06,
      "loss": 0.7763,
      "step": 3840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9251300904942581,
      "learning_rate": 3.3666211432024974e-06,
      "loss": 0.769,
      "step": 3841
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9591894359060232,
      "learning_rate": 3.361959764217018e-06,
      "loss": 0.8218,
      "step": 3842
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9055577232212361,
      "learning_rate": 3.357300962347313e-06,
      "loss": 0.8511,
      "step": 3843
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8685401193485591,
      "learning_rate": 3.3526447394020887e-06,
      "loss": 0.8133,
      "step": 3844
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8774463048956342,
      "learning_rate": 3.3479910971890516e-06,
      "loss": 0.8241,
      "step": 3845
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8917764565803398,
      "learning_rate": 3.343340037514903e-06,
      "loss": 0.8169,
      "step": 3846
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9356995522208004,
      "learning_rate": 3.3386915621853533e-06,
      "loss": 0.8119,
      "step": 3847
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9970589642014346,
      "learning_rate": 3.3340456730050887e-06,
      "loss": 0.8419,
      "step": 3848
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9831293115692574,
      "learning_rate": 3.3294023717778122e-06,
      "loss": 0.8074,
      "step": 3849
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9432455449801467,
      "learning_rate": 3.324761660306215e-06,
      "loss": 0.8334,
      "step": 3850
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9846403293235677,
      "learning_rate": 3.3201235403919683e-06,
      "loss": 0.7709,
      "step": 3851
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6264150938704788,
      "learning_rate": 3.3154880138357626e-06,
      "loss": 0.3222,
      "step": 3852
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9196370728235349,
      "learning_rate": 3.3108550824372632e-06,
      "loss": 0.7899,
      "step": 3853
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9632459017957622,
      "learning_rate": 3.306224747995136e-06,
      "loss": 0.7844,
      "step": 3854
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9425093630084058,
      "learning_rate": 3.301597012307034e-06,
      "loss": 0.7176,
      "step": 3855
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9941625682839172,
      "learning_rate": 3.2969718771696047e-06,
      "loss": 0.817,
      "step": 3856
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9168382333618911,
      "learning_rate": 3.292349344378486e-06,
      "loss": 0.7317,
      "step": 3857
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9719011409040538,
      "learning_rate": 3.287729415728298e-06,
      "loss": 0.7684,
      "step": 3858
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.887170840224854,
      "learning_rate": 3.283112093012669e-06,
      "loss": 0.7842,
      "step": 3859
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8592578753938587,
      "learning_rate": 3.278497378024187e-06,
      "loss": 0.7468,
      "step": 3860
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.93335350875479,
      "learning_rate": 3.2738852725544547e-06,
      "loss": 0.8421,
      "step": 3861
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8903541124226438,
      "learning_rate": 3.2692757783940467e-06,
      "loss": 0.7791,
      "step": 3862
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9502443845393894,
      "learning_rate": 3.264668897332527e-06,
      "loss": 0.7775,
      "step": 3863
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9453899057771106,
      "learning_rate": 3.2600646311584494e-06,
      "loss": 0.7736,
      "step": 3864
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8659950007891597,
      "learning_rate": 3.2554629816593375e-06,
      "loss": 0.8425,
      "step": 3865
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9167574184868257,
      "learning_rate": 3.250863950621721e-06,
      "loss": 0.7228,
      "step": 3866
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9024902665769222,
      "learning_rate": 3.2462675398310984e-06,
      "loss": 0.7311,
      "step": 3867
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9420372240340888,
      "learning_rate": 3.241673751071954e-06,
      "loss": 0.763,
      "step": 3868
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8773769823091826,
      "learning_rate": 3.2370825861277567e-06,
      "loss": 0.7718,
      "step": 3869
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8983971542649613,
      "learning_rate": 3.2324940467809527e-06,
      "loss": 0.7919,
      "step": 3870
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9598017581935206,
      "learning_rate": 3.2279081348129713e-06,
      "loss": 0.7934,
      "step": 3871
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.862285490135528,
      "learning_rate": 3.223324852004219e-06,
      "loss": 0.7436,
      "step": 3872
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9272355584514826,
      "learning_rate": 3.2187442001340942e-06,
      "loss": 0.8163,
      "step": 3873
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0249923961363903,
      "learning_rate": 3.21416618098095e-06,
      "loss": 0.8229,
      "step": 3874
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9801039148754099,
      "learning_rate": 3.2095907963221396e-06,
      "loss": 0.7746,
      "step": 3875
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6375123465443372,
      "learning_rate": 3.2050180479339865e-06,
      "loss": 0.3409,
      "step": 3876
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8904973748277651,
      "learning_rate": 3.2004479375917783e-06,
      "loss": 0.7739,
      "step": 3877
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6027645501796205,
      "learning_rate": 3.1958804670698008e-06,
      "loss": 0.3491,
      "step": 3878
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0029647526419572,
      "learning_rate": 3.191315638141297e-06,
      "loss": 0.7935,
      "step": 3879
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.89424378175305,
      "learning_rate": 3.1867534525784937e-06,
      "loss": 0.7409,
      "step": 3880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9599840863794487,
      "learning_rate": 3.182193912152586e-06,
      "loss": 0.7868,
      "step": 3881
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9692213285198317,
      "learning_rate": 3.177637018633746e-06,
      "loss": 0.7974,
      "step": 3882
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8911449263453768,
      "learning_rate": 3.1730827737911163e-06,
      "loss": 0.7792,
      "step": 3883
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8797982849520214,
      "learning_rate": 3.1685311793928077e-06,
      "loss": 0.8048,
      "step": 3884
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8494206824629105,
      "learning_rate": 3.163982237205917e-06,
      "loss": 0.7159,
      "step": 3885
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9436072041043417,
      "learning_rate": 3.1594359489964853e-06,
      "loss": 0.8364,
      "step": 3886
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8404604372050297,
      "learning_rate": 3.15489231652955e-06,
      "loss": 0.7625,
      "step": 3887
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9836279786807196,
      "learning_rate": 3.150351341569101e-06,
      "loss": 0.7861,
      "step": 3888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9219603834879888,
      "learning_rate": 3.1458130258781006e-06,
      "loss": 0.8608,
      "step": 3889
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9526479506307071,
      "learning_rate": 3.141277371218484e-06,
      "loss": 0.7814,
      "step": 3890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.962405774261848,
      "learning_rate": 3.136744379351139e-06,
      "loss": 0.7763,
      "step": 3891
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6072745526876893,
      "learning_rate": 3.1322140520359366e-06,
      "loss": 0.3438,
      "step": 3892
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9500348523140263,
      "learning_rate": 3.1276863910317057e-06,
      "loss": 0.8102,
      "step": 3893
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8570281507756382,
      "learning_rate": 3.1231613980962373e-06,
      "loss": 0.7761,
      "step": 3894
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9510017944015908,
      "learning_rate": 3.1186390749862904e-06,
      "loss": 0.8299,
      "step": 3895
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9801391576192314,
      "learning_rate": 3.1141194234575878e-06,
      "loss": 0.7688,
      "step": 3896
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9056793514752146,
      "learning_rate": 3.1096024452648123e-06,
      "loss": 0.7509,
      "step": 3897
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9678406103853336,
      "learning_rate": 3.1050881421616076e-06,
      "loss": 0.8231,
      "step": 3898
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8625465986081032,
      "learning_rate": 3.100576515900591e-06,
      "loss": 0.7541,
      "step": 3899
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9790319931015794,
      "learning_rate": 3.0960675682333186e-06,
      "loss": 0.8383,
      "step": 3900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9530215626181748,
      "learning_rate": 3.0915613009103296e-06,
      "loss": 0.8024,
      "step": 3901
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9036228748040248,
      "learning_rate": 3.0870577156811077e-06,
      "loss": 0.7468,
      "step": 3902
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8811540620849014,
      "learning_rate": 3.0825568142940998e-06,
      "loss": 0.7758,
      "step": 3903
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9981420435667785,
      "learning_rate": 3.0780585984967113e-06,
      "loss": 0.7924,
      "step": 3904
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9068658082474116,
      "learning_rate": 3.073563070035305e-06,
      "loss": 0.8202,
      "step": 3905
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9092761458744537,
      "learning_rate": 3.069070230655198e-06,
      "loss": 0.7986,
      "step": 3906
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9202356266624818,
      "learning_rate": 3.0645800821006667e-06,
      "loss": 0.7744,
      "step": 3907
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8922955740869184,
      "learning_rate": 3.060092626114941e-06,
      "loss": 0.7509,
      "step": 3908
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8894908278113423,
      "learning_rate": 3.0556078644402066e-06,
      "loss": 0.7886,
      "step": 3909
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9285946293332782,
      "learning_rate": 3.051125798817598e-06,
      "loss": 0.7422,
      "step": 3910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9602494056202034,
      "learning_rate": 3.0466464309872167e-06,
      "loss": 0.7593,
      "step": 3911
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9035479610569661,
      "learning_rate": 3.042169762688096e-06,
      "loss": 0.741,
      "step": 3912
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8942562715512703,
      "learning_rate": 3.0376957956582452e-06,
      "loss": 0.8194,
      "step": 3913
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9092476516688031,
      "learning_rate": 3.0332245316346e-06,
      "loss": 0.8502,
      "step": 3914
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9076907718753942,
      "learning_rate": 3.0287559723530667e-06,
      "loss": 0.8454,
      "step": 3915
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9227492470007473,
      "learning_rate": 3.024290119548495e-06,
      "loss": 0.8063,
      "step": 3916
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8682048048386585,
      "learning_rate": 3.019826974954674e-06,
      "loss": 0.7439,
      "step": 3917
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8701232003904373,
      "learning_rate": 3.0153665403043586e-06,
      "loss": 0.7753,
      "step": 3918
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.955783257271466,
      "learning_rate": 3.01090881732924e-06,
      "loss": 0.7939,
      "step": 3919
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9641569249471746,
      "learning_rate": 3.0064538077599603e-06,
      "loss": 0.7681,
      "step": 3920
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8923339595872951,
      "learning_rate": 3.002001513326107e-06,
      "loss": 0.7289,
      "step": 3921
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9657530438146585,
      "learning_rate": 2.9975519357562155e-06,
      "loss": 0.7779,
      "step": 3922
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9118284466994308,
      "learning_rate": 2.9931050767777626e-06,
      "loss": 0.7998,
      "step": 3923
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8263074748661284,
      "learning_rate": 2.9886609381171703e-06,
      "loss": 0.6963,
      "step": 3924
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9066686932222329,
      "learning_rate": 2.984219521499816e-06,
      "loss": 0.7474,
      "step": 3925
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8754971548174341,
      "learning_rate": 2.9797808286499976e-06,
      "loss": 0.7591,
      "step": 3926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.621494268662893,
      "learning_rate": 2.9753448612909775e-06,
      "loss": 0.3058,
      "step": 3927
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8773082305440989,
      "learning_rate": 2.9709116211449484e-06,
      "loss": 0.7702,
      "step": 3928
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.837148387369686,
      "learning_rate": 2.966481109933047e-06,
      "loss": 0.7952,
      "step": 3929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9615487554763124,
      "learning_rate": 2.9620533293753495e-06,
      "loss": 0.7727,
      "step": 3930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8948705453484388,
      "learning_rate": 2.957628281190873e-06,
      "loss": 0.7474,
      "step": 3931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9255391451441685,
      "learning_rate": 2.9532059670975732e-06,
      "loss": 0.8142,
      "step": 3932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9026440598301692,
      "learning_rate": 2.948786388812346e-06,
      "loss": 0.7481,
      "step": 3933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.907061981906316,
      "learning_rate": 2.9443695480510225e-06,
      "loss": 0.7633,
      "step": 3934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6031856834780076,
      "learning_rate": 2.9399554465283742e-06,
      "loss": 0.298,
      "step": 3935
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0109728916144016,
      "learning_rate": 2.935544085958102e-06,
      "loss": 0.7595,
      "step": 3936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9078782252885981,
      "learning_rate": 2.931135468052858e-06,
      "loss": 0.8209,
      "step": 3937
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0110948762745922,
      "learning_rate": 2.926729594524207e-06,
      "loss": 0.8021,
      "step": 3938
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8905995068200706,
      "learning_rate": 2.9223264670826746e-06,
      "loss": 0.8143,
      "step": 3939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8649155720199928,
      "learning_rate": 2.9179260874376915e-06,
      "loss": 0.8032,
      "step": 3940
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0176291501016157,
      "learning_rate": 2.9135284572976486e-06,
      "loss": 0.7713,
      "step": 3941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9975299558015833,
      "learning_rate": 2.9091335783698517e-06,
      "loss": 0.7674,
      "step": 3942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8893359192274206,
      "learning_rate": 2.9047414523605467e-06,
      "loss": 0.7897,
      "step": 3943
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8896456939809895,
      "learning_rate": 2.9003520809749053e-06,
      "loss": 0.7749,
      "step": 3944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9067411570331365,
      "learning_rate": 2.8959654659170354e-06,
      "loss": 0.8187,
      "step": 3945
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.937709049173102,
      "learning_rate": 2.8915816088899696e-06,
      "loss": 0.8206,
      "step": 3946
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8433054292545751,
      "learning_rate": 2.8872005115956746e-06,
      "loss": 0.8053,
      "step": 3947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5722784711824801,
      "learning_rate": 2.8828221757350406e-06,
      "loss": 0.3125,
      "step": 3948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9042236028609925,
      "learning_rate": 2.8784466030078905e-06,
      "loss": 0.7792,
      "step": 3949
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9057248700555998,
      "learning_rate": 2.874073795112967e-06,
      "loss": 0.7369,
      "step": 3950
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9608285009100604,
      "learning_rate": 2.8697037537479565e-06,
      "loss": 0.8041,
      "step": 3951
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6016523265886421,
      "learning_rate": 2.8653364806094454e-06,
      "loss": 0.3499,
      "step": 3952
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9140597543614061,
      "learning_rate": 2.86097197739297e-06,
      "loss": 0.8382,
      "step": 3953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9099518413123862,
      "learning_rate": 2.856610245792976e-06,
      "loss": 0.8132,
      "step": 3954
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9192264358203488,
      "learning_rate": 2.8522512875028396e-06,
      "loss": 0.8118,
      "step": 3955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9170122619215043,
      "learning_rate": 2.847895104214856e-06,
      "loss": 0.7824,
      "step": 3956
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.916154356950468,
      "learning_rate": 2.843541697620249e-06,
      "loss": 0.7517,
      "step": 3957
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.006348131120469,
      "learning_rate": 2.8391910694091584e-06,
      "loss": 0.7485,
      "step": 3958
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9150267590516977,
      "learning_rate": 2.8348432212706443e-06,
      "loss": 0.7985,
      "step": 3959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9130853450232639,
      "learning_rate": 2.8304981548927025e-06,
      "loss": 0.7452,
      "step": 3960
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0262130149438609,
      "learning_rate": 2.826155871962227e-06,
      "loss": 0.839,
      "step": 3961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9727795317617526,
      "learning_rate": 2.8218163741650415e-06,
      "loss": 0.837,
      "step": 3962
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9145358067094624,
      "learning_rate": 2.817479663185898e-06,
      "loss": 0.7924,
      "step": 3963
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9571403283121809,
      "learning_rate": 2.813145740708445e-06,
      "loss": 0.7475,
      "step": 3964
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9513426135201535,
      "learning_rate": 2.808814608415271e-06,
      "loss": 0.8503,
      "step": 3965
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8822110541033928,
      "learning_rate": 2.8044862679878605e-06,
      "loss": 0.8601,
      "step": 3966
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9544554023023886,
      "learning_rate": 2.800160721106633e-06,
      "loss": 0.7505,
      "step": 3967
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0801794028784035,
      "learning_rate": 2.7958379694509108e-06,
      "loss": 0.8006,
      "step": 3968
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9550531620291347,
      "learning_rate": 2.791518014698935e-06,
      "loss": 0.7837,
      "step": 3969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0304268243697041,
      "learning_rate": 2.787200858527862e-06,
      "loss": 0.7563,
      "step": 3970
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6311030404478309,
      "learning_rate": 2.7828865026137584e-06,
      "loss": 0.3037,
      "step": 3971
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8726277525433543,
      "learning_rate": 2.7785749486316085e-06,
      "loss": 0.7893,
      "step": 3972
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8005138831857073,
      "learning_rate": 2.774266198255303e-06,
      "loss": 0.7709,
      "step": 3973
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9965322921083731,
      "learning_rate": 2.7699602531576496e-06,
      "loss": 0.7824,
      "step": 3974
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8699510633304274,
      "learning_rate": 2.765657115010364e-06,
      "loss": 0.7795,
      "step": 3975
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9581458710915906,
      "learning_rate": 2.7613567854840685e-06,
      "loss": 0.8236,
      "step": 3976
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8577138949559125,
      "learning_rate": 2.7570592662483086e-06,
      "loss": 0.8181,
      "step": 3977
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9156146585153879,
      "learning_rate": 2.752764558971517e-06,
      "loss": 0.8442,
      "step": 3978
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.884173002543246,
      "learning_rate": 2.748472665321056e-06,
      "loss": 0.854,
      "step": 3979
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8538143340554506,
      "learning_rate": 2.744183586963185e-06,
      "loss": 0.7612,
      "step": 3980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9818915773561689,
      "learning_rate": 2.739897325563069e-06,
      "loss": 0.7902,
      "step": 3981
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.874370154147911,
      "learning_rate": 2.7356138827847856e-06,
      "loss": 0.8011,
      "step": 3982
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9226430652372434,
      "learning_rate": 2.731333260291311e-06,
      "loss": 0.8007,
      "step": 3983
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9201810081031793,
      "learning_rate": 2.7270554597445343e-06,
      "loss": 0.8125,
      "step": 3984
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9058547954085784,
      "learning_rate": 2.7227804828052384e-06,
      "loss": 0.7483,
      "step": 3985
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9465244256892463,
      "learning_rate": 2.7185083311331283e-06,
      "loss": 0.8086,
      "step": 3986
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9733545258127441,
      "learning_rate": 2.7142390063867896e-06,
      "loss": 0.819,
      "step": 3987
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6164587481549991,
      "learning_rate": 2.709972510223725e-06,
      "loss": 0.3223,
      "step": 3988
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9637411587363879,
      "learning_rate": 2.7057088443003343e-06,
      "loss": 0.8243,
      "step": 3989
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9214575130817333,
      "learning_rate": 2.7014480102719174e-06,
      "loss": 0.7992,
      "step": 3990
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9273252973130645,
      "learning_rate": 2.697190009792685e-06,
      "loss": 0.836,
      "step": 3991
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9022215867845718,
      "learning_rate": 2.692934844515729e-06,
      "loss": 0.8041,
      "step": 3992
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9443674874928005,
      "learning_rate": 2.6886825160930587e-06,
      "loss": 0.8151,
      "step": 3993
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8667943213226655,
      "learning_rate": 2.6844330261755715e-06,
      "loss": 0.8151,
      "step": 3994
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.967018348439748,
      "learning_rate": 2.6801863764130653e-06,
      "loss": 0.7729,
      "step": 3995
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.905205529895287,
      "learning_rate": 2.675942568454236e-06,
      "loss": 0.784,
      "step": 3996
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5996665616676364,
      "learning_rate": 2.671701603946678e-06,
      "loss": 0.3063,
      "step": 3997
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8904035095710309,
      "learning_rate": 2.667463484536876e-06,
      "loss": 0.7676,
      "step": 3998
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.91588637115171,
      "learning_rate": 2.6632282118702147e-06,
      "loss": 0.821,
      "step": 3999
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.931213630185678,
      "learning_rate": 2.65899578759098e-06,
      "loss": 0.8206,
      "step": 4000
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9957414401394535,
      "learning_rate": 2.654766213342335e-06,
      "loss": 0.8228,
      "step": 4001
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8704978926685915,
      "learning_rate": 2.650539490766346e-06,
      "loss": 0.7681,
      "step": 4002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8842941803385187,
      "learning_rate": 2.646315621503983e-06,
      "loss": 0.788,
      "step": 4003
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.927771337927262,
      "learning_rate": 2.642094607195085e-06,
      "loss": 0.7946,
      "step": 4004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0038039097545384,
      "learning_rate": 2.6378764494784027e-06,
      "loss": 0.8301,
      "step": 4005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8722971532266178,
      "learning_rate": 2.633661149991569e-06,
      "loss": 0.8011,
      "step": 4006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.947358231343761,
      "learning_rate": 2.6294487103711064e-06,
      "loss": 0.7919,
      "step": 4007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9512216366004074,
      "learning_rate": 2.6252391322524297e-06,
      "loss": 0.8003,
      "step": 4008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8876109194585653,
      "learning_rate": 2.6210324172698432e-06,
      "loss": 0.8373,
      "step": 4009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9251040421813855,
      "learning_rate": 2.6168285670565374e-06,
      "loss": 0.826,
      "step": 4010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6373352091770039,
      "learning_rate": 2.6126275832445892e-06,
      "loss": 0.3596,
      "step": 4011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9050643529243408,
      "learning_rate": 2.6084294674649734e-06,
      "loss": 0.8488,
      "step": 4012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.970562762103769,
      "learning_rate": 2.6042342213475346e-06,
      "loss": 0.7905,
      "step": 4013
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.854163923425117,
      "learning_rate": 2.6000418465210143e-06,
      "loss": 0.7821,
      "step": 4014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9326458059034255,
      "learning_rate": 2.595852344613038e-06,
      "loss": 0.7759,
      "step": 4015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8443320201903799,
      "learning_rate": 2.5916657172501103e-06,
      "loss": 0.7521,
      "step": 4016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8287038898269482,
      "learning_rate": 2.587481966057633e-06,
      "loss": 0.6743,
      "step": 4017
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0425706786403959,
      "learning_rate": 2.583301092659872e-06,
      "loss": 0.818,
      "step": 4018
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9305432287230739,
      "learning_rate": 2.5791230986799944e-06,
      "loss": 0.7487,
      "step": 4019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9907044051337833,
      "learning_rate": 2.5749479857400383e-06,
      "loss": 0.7885,
      "step": 4020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9918953199487682,
      "learning_rate": 2.5707757554609247e-06,
      "loss": 0.77,
      "step": 4021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8727769575780905,
      "learning_rate": 2.56660640946246e-06,
      "loss": 0.8214,
      "step": 4022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6052270590512155,
      "learning_rate": 2.5624399493633257e-06,
      "loss": 0.3495,
      "step": 4023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9094304939640173,
      "learning_rate": 2.558276376781086e-06,
      "loss": 0.771,
      "step": 4024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9092951488984455,
      "learning_rate": 2.55411569333218e-06,
      "loss": 0.8349,
      "step": 4025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8627217948325921,
      "learning_rate": 2.5499579006319365e-06,
      "loss": 0.7722,
      "step": 4026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9516137326104612,
      "learning_rate": 2.5458030002945457e-06,
      "loss": 0.7865,
      "step": 4027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9134080841292606,
      "learning_rate": 2.5416509939330836e-06,
      "loss": 0.7844,
      "step": 4028
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.00311326612108,
      "learning_rate": 2.537501883159509e-06,
      "loss": 0.8169,
      "step": 4029
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9756040825602661,
      "learning_rate": 2.5333556695846384e-06,
      "loss": 0.7983,
      "step": 4030
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8012845559835642,
      "learning_rate": 2.5292123548181847e-06,
      "loss": 0.7904,
      "step": 4031
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0068415502647072,
      "learning_rate": 2.525071940468722e-06,
      "loss": 0.8411,
      "step": 4032
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.957148410598195,
      "learning_rate": 2.520934428143701e-06,
      "loss": 0.8206,
      "step": 4033
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9028764172769097,
      "learning_rate": 2.5167998194494468e-06,
      "loss": 0.748,
      "step": 4034
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8784176697658944,
      "learning_rate": 2.5126681159911558e-06,
      "loss": 0.7605,
      "step": 4035
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8723557888425821,
      "learning_rate": 2.5085393193729e-06,
      "loss": 0.8147,
      "step": 4036
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9436676393461698,
      "learning_rate": 2.5044134311976156e-06,
      "loss": 0.7824,
      "step": 4037
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9638948696252382,
      "learning_rate": 2.5002904530671236e-06,
      "loss": 0.724,
      "step": 4038
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8893360643970675,
      "learning_rate": 2.4961703865820974e-06,
      "loss": 0.8078,
      "step": 4039
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9009637980665037,
      "learning_rate": 2.492053233342091e-06,
      "loss": 0.8065,
      "step": 4040
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9500887157734325,
      "learning_rate": 2.487938994945527e-06,
      "loss": 0.7497,
      "step": 4041
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8831144161851008,
      "learning_rate": 2.4838276729896884e-06,
      "loss": 0.7593,
      "step": 4042
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9333504970675036,
      "learning_rate": 2.479719269070743e-06,
      "loss": 0.8035,
      "step": 4043
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9105954661970656,
      "learning_rate": 2.4756137847837025e-06,
      "loss": 0.8084,
      "step": 4044
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9010151254845133,
      "learning_rate": 2.4715112217224657e-06,
      "loss": 0.7766,
      "step": 4045
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9698619109409733,
      "learning_rate": 2.467411581479786e-06,
      "loss": 0.81,
      "step": 4046
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9114968933452496,
      "learning_rate": 2.463314865647286e-06,
      "loss": 0.7173,
      "step": 4047
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9634114161833938,
      "learning_rate": 2.45922107581545e-06,
      "loss": 0.7498,
      "step": 4048
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.982022408301079,
      "learning_rate": 2.4551302135736287e-06,
      "loss": 0.863,
      "step": 4049
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9084898770289908,
      "learning_rate": 2.4510422805100366e-06,
      "loss": 0.7857,
      "step": 4050
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8607663622526159,
      "learning_rate": 2.446957278211746e-06,
      "loss": 0.714,
      "step": 4051
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0500388046297113,
      "learning_rate": 2.4428752082647044e-06,
      "loss": 0.8099,
      "step": 4052
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9217335067201039,
      "learning_rate": 2.438796072253704e-06,
      "loss": 0.7721,
      "step": 4053
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9475324852118293,
      "learning_rate": 2.4347198717624054e-06,
      "loss": 0.7508,
      "step": 4054
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9413977136363414,
      "learning_rate": 2.4306466083733392e-06,
      "loss": 0.7904,
      "step": 4055
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9369688305027141,
      "learning_rate": 2.426576283667873e-06,
      "loss": 0.8406,
      "step": 4056
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5969812410434074,
      "learning_rate": 2.422508899226258e-06,
      "loss": 0.3362,
      "step": 4057
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9440712055962592,
      "learning_rate": 2.418444456627589e-06,
      "loss": 0.7517,
      "step": 4058
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6336913746225199,
      "learning_rate": 2.4143829574498224e-06,
      "loss": 0.3255,
      "step": 4059
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8926865414817148,
      "learning_rate": 2.4103244032697717e-06,
      "loss": 0.8105,
      "step": 4060
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9554650698130467,
      "learning_rate": 2.406268795663108e-06,
      "loss": 0.799,
      "step": 4061
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8969052808082565,
      "learning_rate": 2.4022161362043574e-06,
      "loss": 0.7946,
      "step": 4062
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9480881799176505,
      "learning_rate": 2.3981664264669025e-06,
      "loss": 0.8299,
      "step": 4063
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9454572474183917,
      "learning_rate": 2.3941196680229794e-06,
      "loss": 0.7995,
      "step": 4064
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9464639685603982,
      "learning_rate": 2.3900758624436772e-06,
      "loss": 0.8183,
      "step": 4065
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9170720392667422,
      "learning_rate": 2.3860350112989473e-06,
      "loss": 0.8276,
      "step": 4066
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8773624240532395,
      "learning_rate": 2.3819971161575807e-06,
      "loss": 0.7389,
      "step": 4067
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8988696409221573,
      "learning_rate": 2.3779621785872252e-06,
      "loss": 0.7989,
      "step": 4068
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.891341384799849,
      "learning_rate": 2.3739302001543918e-06,
      "loss": 0.7367,
      "step": 4069
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.916418292960584,
      "learning_rate": 2.3699011824244234e-06,
      "loss": 0.7927,
      "step": 4070
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9626302261827461,
      "learning_rate": 2.365875126961531e-06,
      "loss": 0.7623,
      "step": 4071
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8584031277421008,
      "learning_rate": 2.3618520353287644e-06,
      "loss": 0.786,
      "step": 4072
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0089324660448709,
      "learning_rate": 2.3578319090880263e-06,
      "loss": 0.8043,
      "step": 4073
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8814411162649073,
      "learning_rate": 2.3538147498000695e-06,
      "loss": 0.8036,
      "step": 4074
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9413076755745972,
      "learning_rate": 2.349800559024492e-06,
      "loss": 0.7261,
      "step": 4075
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8927979601596819,
      "learning_rate": 2.3457893383197415e-06,
      "loss": 0.7848,
      "step": 4076
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9476850802544236,
      "learning_rate": 2.3417810892431104e-06,
      "loss": 0.7344,
      "step": 4077
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9327542860809677,
      "learning_rate": 2.3377758133507455e-06,
      "loss": 0.7855,
      "step": 4078
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9700992062476465,
      "learning_rate": 2.3337735121976247e-06,
      "loss": 0.8425,
      "step": 4079
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9086197173226288,
      "learning_rate": 2.32977418733758e-06,
      "loss": 0.8231,
      "step": 4080
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9351646751607481,
      "learning_rate": 2.3257778403232954e-06,
      "loss": 0.7496,
      "step": 4081
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9671010764620571,
      "learning_rate": 2.321784472706279e-06,
      "loss": 0.7866,
      "step": 4082
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0072901569077233,
      "learning_rate": 2.317794086036901e-06,
      "loss": 0.8049,
      "step": 4083
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.857131427088822,
      "learning_rate": 2.3138066818643647e-06,
      "loss": 0.8031,
      "step": 4084
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8820923876437998,
      "learning_rate": 2.3098222617367184e-06,
      "loss": 0.7079,
      "step": 4085
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9187563117554114,
      "learning_rate": 2.30584082720085e-06,
      "loss": 0.7752,
      "step": 4086
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.956891096566894,
      "learning_rate": 2.301862379802492e-06,
      "loss": 0.8251,
      "step": 4087
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.00336668134055,
      "learning_rate": 2.297886921086211e-06,
      "loss": 0.7565,
      "step": 4088
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9629514933221662,
      "learning_rate": 2.2939144525954194e-06,
      "loss": 0.7868,
      "step": 4089
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9174194354971649,
      "learning_rate": 2.2899449758723657e-06,
      "loss": 0.766,
      "step": 4090
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9586393201459138,
      "learning_rate": 2.285978492458134e-06,
      "loss": 0.7059,
      "step": 4091
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8665788568280236,
      "learning_rate": 2.282015003892659e-06,
      "loss": 0.7929,
      "step": 4092
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8644675117608589,
      "learning_rate": 2.2780545117146947e-06,
      "loss": 0.7652,
      "step": 4093
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.884462249556066,
      "learning_rate": 2.2740970174618405e-06,
      "loss": 0.7704,
      "step": 4094
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9613175861628851,
      "learning_rate": 2.270142522670541e-06,
      "loss": 0.7513,
      "step": 4095
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0009264397283633,
      "learning_rate": 2.2661910288760545e-06,
      "loss": 0.7685,
      "step": 4096
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9463658200815155,
      "learning_rate": 2.262242537612497e-06,
      "loss": 0.8149,
      "step": 4097
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9065732965036861,
      "learning_rate": 2.258297050412804e-06,
      "loss": 0.7375,
      "step": 4098
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8895751720654287,
      "learning_rate": 2.254354568808752e-06,
      "loss": 0.8117,
      "step": 4099
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9279790462802493,
      "learning_rate": 2.2504150943309455e-06,
      "loss": 0.7112,
      "step": 4100
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9010877399509154,
      "learning_rate": 2.246478628508827e-06,
      "loss": 0.7505,
      "step": 4101
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8781334639990658,
      "learning_rate": 2.242545172870665e-06,
      "loss": 0.8002,
      "step": 4102
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.687583788018246,
      "learning_rate": 2.238614728943561e-06,
      "loss": 0.293,
      "step": 4103
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0201277175584795,
      "learning_rate": 2.2346872982534584e-06,
      "loss": 0.7284,
      "step": 4104
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9669391166202116,
      "learning_rate": 2.2307628823251083e-06,
      "loss": 0.7963,
      "step": 4105
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5922544967534045,
      "learning_rate": 2.2268414826821117e-06,
      "loss": 0.3232,
      "step": 4106
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9305508160413591,
      "learning_rate": 2.222923100846893e-06,
      "loss": 0.8237,
      "step": 4107
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8774793182306662,
      "learning_rate": 2.2190077383406938e-06,
      "loss": 0.7654,
      "step": 4108
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9089649668544914,
      "learning_rate": 2.2150953966835996e-06,
      "loss": 0.7818,
      "step": 4109
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9300853419955135,
      "learning_rate": 2.211186077394516e-06,
      "loss": 0.7471,
      "step": 4110
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9306107297628496,
      "learning_rate": 2.207279781991173e-06,
      "loss": 0.7728,
      "step": 4111
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8957766997089358,
      "learning_rate": 2.2033765119901294e-06,
      "loss": 0.7767,
      "step": 4112
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9930624251089387,
      "learning_rate": 2.1994762689067705e-06,
      "loss": 0.7847,
      "step": 4113
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8695014536791572,
      "learning_rate": 2.1955790542553036e-06,
      "loss": 0.7761,
      "step": 4114
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8885240175048227,
      "learning_rate": 2.1916848695487615e-06,
      "loss": 0.7736,
      "step": 4115
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8876849072647026,
      "learning_rate": 2.1877937162990015e-06,
      "loss": 0.7781,
      "step": 4116
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9734643986113716,
      "learning_rate": 2.1839055960167e-06,
      "loss": 0.7528,
      "step": 4117
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9036553109660933,
      "learning_rate": 2.180020510211367e-06,
      "loss": 0.7251,
      "step": 4118
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.944295521604042,
      "learning_rate": 2.1761384603913203e-06,
      "loss": 0.8312,
      "step": 4119
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8930325868010598,
      "learning_rate": 2.172259448063704e-06,
      "loss": 0.841,
      "step": 4120
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.953143174724169,
      "learning_rate": 2.1683834747344913e-06,
      "loss": 0.8558,
      "step": 4121
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9511455616943899,
      "learning_rate": 2.1645105419084587e-06,
      "loss": 0.7433,
      "step": 4122
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8878852009817365,
      "learning_rate": 2.160640651089221e-06,
      "loss": 0.8255,
      "step": 4123
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9201659843756552,
      "learning_rate": 2.1567738037791998e-06,
      "loss": 0.8097,
      "step": 4124
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9716391411185906,
      "learning_rate": 2.152910001479638e-06,
      "loss": 0.8081,
      "step": 4125
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8858157941406769,
      "learning_rate": 2.1490492456905964e-06,
      "loss": 0.7902,
      "step": 4126
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8858497094616478,
      "learning_rate": 2.1451915379109546e-06,
      "loss": 0.7978,
      "step": 4127
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8778607224161487,
      "learning_rate": 2.141336879638406e-06,
      "loss": 0.8234,
      "step": 4128
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9381193740103095,
      "learning_rate": 2.1374852723694595e-06,
      "loss": 0.7937,
      "step": 4129
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9247141600412696,
      "learning_rate": 2.133636717599451e-06,
      "loss": 0.8125,
      "step": 4130
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8382973839610621,
      "learning_rate": 2.1297912168225086e-06,
      "loss": 0.7432,
      "step": 4131
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9015399373659163,
      "learning_rate": 2.1259487715316e-06,
      "loss": 0.7732,
      "step": 4132
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6267005795761172,
      "learning_rate": 2.1221093832184903e-06,
      "loss": 0.277,
      "step": 4133
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9514749025186865,
      "learning_rate": 2.118273053373757e-06,
      "loss": 0.8003,
      "step": 4134
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5883194450153946,
      "learning_rate": 2.1144397834868034e-06,
      "loss": 0.3203,
      "step": 4135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.973097723401237,
      "learning_rate": 2.1106095750458332e-06,
      "loss": 0.7636,
      "step": 4136
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8568154066046462,
      "learning_rate": 2.106782429537866e-06,
      "loss": 0.8202,
      "step": 4137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9067767752116617,
      "learning_rate": 2.1029583484487315e-06,
      "loss": 0.8023,
      "step": 4138
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9314081193789681,
      "learning_rate": 2.0991373332630683e-06,
      "loss": 0.7951,
      "step": 4139
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.01717963465603,
      "learning_rate": 2.0953193854643274e-06,
      "loss": 0.769,
      "step": 4140
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8632891527036423,
      "learning_rate": 2.0915045065347673e-06,
      "loss": 0.7779,
      "step": 4141
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8826030098509128,
      "learning_rate": 2.0876926979554545e-06,
      "loss": 0.7862,
      "step": 4142
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0003026581337486,
      "learning_rate": 2.0838839612062633e-06,
      "loss": 0.7579,
      "step": 4143
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8813476400031937,
      "learning_rate": 2.080078297765884e-06,
      "loss": 0.8161,
      "step": 4144
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9357923861731673,
      "learning_rate": 2.0762757091117937e-06,
      "loss": 0.8443,
      "step": 4145
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9145168286939215,
      "learning_rate": 2.0724761967202987e-06,
      "loss": 0.8091,
      "step": 4146
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7991484156441965,
      "learning_rate": 2.0686797620664987e-06,
      "loss": 0.7753,
      "step": 4147
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9293400037147097,
      "learning_rate": 2.0648864066242937e-06,
      "loss": 0.8116,
      "step": 4148
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9269873648322695,
      "learning_rate": 2.0610961318664013e-06,
      "loss": 0.8144,
      "step": 4149
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8905467891285217,
      "learning_rate": 2.0573089392643362e-06,
      "loss": 0.7828,
      "step": 4150
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9737863260600879,
      "learning_rate": 2.0535248302884147e-06,
      "loss": 0.7724,
      "step": 4151
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9061521912073187,
      "learning_rate": 2.0497438064077603e-06,
      "loss": 0.7884,
      "step": 4152
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9786256889586016,
      "learning_rate": 2.045965869090295e-06,
      "loss": 0.7958,
      "step": 4153
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9297269146141532,
      "learning_rate": 2.0421910198027452e-06,
      "loss": 0.7725,
      "step": 4154
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.002816261630375,
      "learning_rate": 2.0384192600106335e-06,
      "loss": 0.7422,
      "step": 4155
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8292803384852188,
      "learning_rate": 2.0346505911782956e-06,
      "loss": 0.8153,
      "step": 4156
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9044306584785352,
      "learning_rate": 2.0308850147688484e-06,
      "loss": 0.6963,
      "step": 4157
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9069601513167944,
      "learning_rate": 2.0271225322442255e-06,
      "loss": 0.7987,
      "step": 4158
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9460764244741151,
      "learning_rate": 2.0233631450651525e-06,
      "loss": 0.7913,
      "step": 4159
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.893060909831015,
      "learning_rate": 2.019606854691145e-06,
      "loss": 0.778,
      "step": 4160
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9211889914227526,
      "learning_rate": 2.0158536625805325e-06,
      "loss": 0.7817,
      "step": 4161
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8866218226923815,
      "learning_rate": 2.01210357019043e-06,
      "loss": 0.7737,
      "step": 4162
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9509814559393819,
      "learning_rate": 2.008356578976752e-06,
      "loss": 0.7566,
      "step": 4163
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9273571006463455,
      "learning_rate": 2.004612690394212e-06,
      "loss": 0.8072,
      "step": 4164
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.991382161042055,
      "learning_rate": 2.0008719058963144e-06,
      "loss": 0.7911,
      "step": 4165
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0456698480189377,
      "learning_rate": 1.997134226935361e-06,
      "loss": 0.8183,
      "step": 4166
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9025920298251603,
      "learning_rate": 1.9933996549624468e-06,
      "loss": 0.7933,
      "step": 4167
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0050242994479184,
      "learning_rate": 1.9896681914274616e-06,
      "loss": 0.7877,
      "step": 4168
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9874695056419966,
      "learning_rate": 1.9859398377790872e-06,
      "loss": 0.7661,
      "step": 4169
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6160022331912641,
      "learning_rate": 1.982214595464804e-06,
      "loss": 0.3415,
      "step": 4170
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9237998330268188,
      "learning_rate": 1.97849246593087e-06,
      "loss": 0.7954,
      "step": 4171
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.917983734992662,
      "learning_rate": 1.9747734506223525e-06,
      "loss": 0.776,
      "step": 4172
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5940971773154293,
      "learning_rate": 1.9710575509831008e-06,
      "loss": 0.35,
      "step": 4173
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9540274146131092,
      "learning_rate": 1.967344768455747e-06,
      "loss": 0.8541,
      "step": 4174
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9468762681519007,
      "learning_rate": 1.9636351044817292e-06,
      "loss": 0.7562,
      "step": 4175
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8942695449845464,
      "learning_rate": 1.9599285605012643e-06,
      "loss": 0.8252,
      "step": 4176
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9793980733653549,
      "learning_rate": 1.9562251379533593e-06,
      "loss": 0.7737,
      "step": 4177
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9726698109196565,
      "learning_rate": 1.952524838275811e-06,
      "loss": 0.8066,
      "step": 4178
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9012311776499032,
      "learning_rate": 1.9488276629052026e-06,
      "loss": 0.7516,
      "step": 4179
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6085534149041898,
      "learning_rate": 1.945133613276907e-06,
      "loss": 0.3002,
      "step": 4180
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9282685824836415,
      "learning_rate": 1.941442690825076e-06,
      "loss": 0.7514,
      "step": 4181
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9303635640468793,
      "learning_rate": 1.937754896982663e-06,
      "loss": 0.7812,
      "step": 4182
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8721905610190306,
      "learning_rate": 1.9340702331813842e-06,
      "loss": 0.7183,
      "step": 4183
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9707960336411953,
      "learning_rate": 1.9303887008517618e-06,
      "loss": 0.7667,
      "step": 4184
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9323872606093531,
      "learning_rate": 1.9267103014230935e-06,
      "loss": 0.7741,
      "step": 4185
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8874789296601138,
      "learning_rate": 1.923035036323452e-06,
      "loss": 0.76,
      "step": 4186
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9170425986945627,
      "learning_rate": 1.91936290697971e-06,
      "loss": 0.7788,
      "step": 4187
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.890038871383186,
      "learning_rate": 1.9156939148175125e-06,
      "loss": 0.8354,
      "step": 4188
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8878036347427131,
      "learning_rate": 1.9120280612612873e-06,
      "loss": 0.7422,
      "step": 4189
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.881669111751836,
      "learning_rate": 1.9083653477342467e-06,
      "loss": 0.7766,
      "step": 4190
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9308203789086117,
      "learning_rate": 1.904705775658381e-06,
      "loss": 0.7821,
      "step": 4191
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9338437173699001,
      "learning_rate": 1.9010493464544621e-06,
      "loss": 0.8041,
      "step": 4192
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9317509126811548,
      "learning_rate": 1.8973960615420416e-06,
      "loss": 0.8016,
      "step": 4193
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9221697488662778,
      "learning_rate": 1.8937459223394517e-06,
      "loss": 0.7936,
      "step": 4194
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8783832306484317,
      "learning_rate": 1.8900989302637985e-06,
      "loss": 0.7875,
      "step": 4195
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.869510230224188,
      "learning_rate": 1.8864550867309771e-06,
      "loss": 0.7642,
      "step": 4196
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9600755704578353,
      "learning_rate": 1.8828143931556442e-06,
      "loss": 0.8106,
      "step": 4197
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.861799440406029,
      "learning_rate": 1.8791768509512487e-06,
      "loss": 0.7648,
      "step": 4198
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9130907609373587,
      "learning_rate": 1.875542461530011e-06,
      "loss": 0.776,
      "step": 4199
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9552743473460065,
      "learning_rate": 1.871911226302917e-06,
      "loss": 0.8048,
      "step": 4200
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9879031348539385,
      "learning_rate": 1.868283146679747e-06,
      "loss": 0.7764,
      "step": 4201
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9592166459593632,
      "learning_rate": 1.8646582240690414e-06,
      "loss": 0.7411,
      "step": 4202
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8968482484134126,
      "learning_rate": 1.8610364598781227e-06,
      "loss": 0.694,
      "step": 4203
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0013595925164556,
      "learning_rate": 1.8574178555130818e-06,
      "loss": 0.8536,
      "step": 4204
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8535287366935842,
      "learning_rate": 1.8538024123787868e-06,
      "loss": 0.8427,
      "step": 4205
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9337663194244114,
      "learning_rate": 1.8501901318788773e-06,
      "loss": 0.8251,
      "step": 4206
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9297751025541904,
      "learning_rate": 1.8465810154157626e-06,
      "loss": 0.7623,
      "step": 4207
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9758136660560746,
      "learning_rate": 1.8429750643906331e-06,
      "loss": 0.7348,
      "step": 4208
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8552082779207348,
      "learning_rate": 1.8393722802034331e-06,
      "loss": 0.7839,
      "step": 4209
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.96239067507966,
      "learning_rate": 1.835772664252895e-06,
      "loss": 0.7695,
      "step": 4210
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8898350642107336,
      "learning_rate": 1.832176217936511e-06,
      "loss": 0.7518,
      "step": 4211
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9767780504759784,
      "learning_rate": 1.8285829426505453e-06,
      "loss": 0.8304,
      "step": 4212
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9148802828152164,
      "learning_rate": 1.8249928397900351e-06,
      "loss": 0.8258,
      "step": 4213
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9664554665040052,
      "learning_rate": 1.8214059107487726e-06,
      "loss": 0.823,
      "step": 4214
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9511017757635767,
      "learning_rate": 1.8178221569193343e-06,
      "loss": 0.7877,
      "step": 4215
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8802131581515157,
      "learning_rate": 1.8142415796930568e-06,
      "loss": 0.8,
      "step": 4216
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6207711542986102,
      "learning_rate": 1.8106641804600411e-06,
      "loss": 0.3134,
      "step": 4217
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9094837067880043,
      "learning_rate": 1.8070899606091586e-06,
      "loss": 0.7914,
      "step": 4218
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9258170666687436,
      "learning_rate": 1.8035189215280423e-06,
      "loss": 0.7654,
      "step": 4219
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9266197955503541,
      "learning_rate": 1.799951064603095e-06,
      "loss": 0.7674,
      "step": 4220
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9279200099636339,
      "learning_rate": 1.7963863912194768e-06,
      "loss": 0.843,
      "step": 4221
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8910091462102546,
      "learning_rate": 1.7928249027611255e-06,
      "loss": 0.7915,
      "step": 4222
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6566443198818888,
      "learning_rate": 1.789266600610724e-06,
      "loss": 0.3393,
      "step": 4223
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9984132594388154,
      "learning_rate": 1.7857114861497337e-06,
      "loss": 0.7674,
      "step": 4224
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6204604340303165,
      "learning_rate": 1.782159560758373e-06,
      "loss": 0.3481,
      "step": 4225
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9869573578358385,
      "learning_rate": 1.7786108258156154e-06,
      "loss": 0.8675,
      "step": 4226
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9735163295669216,
      "learning_rate": 1.7750652826992077e-06,
      "loss": 0.7453,
      "step": 4227
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9759248503063459,
      "learning_rate": 1.7715229327856498e-06,
      "loss": 0.7986,
      "step": 4228
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9225875789772675,
      "learning_rate": 1.7679837774502052e-06,
      "loss": 0.7567,
      "step": 4229
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.871879402822697,
      "learning_rate": 1.7644478180668945e-06,
      "loss": 0.8075,
      "step": 4230
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8551783221125279,
      "learning_rate": 1.7609150560084986e-06,
      "loss": 0.7227,
      "step": 4231
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0307816789692175,
      "learning_rate": 1.7573854926465582e-06,
      "loss": 0.8007,
      "step": 4232
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0747536712380885,
      "learning_rate": 1.7538591293513685e-06,
      "loss": 0.7793,
      "step": 4233
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9376763218979033,
      "learning_rate": 1.7503359674919929e-06,
      "loss": 0.7991,
      "step": 4234
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8900080842668259,
      "learning_rate": 1.746816008436234e-06,
      "loss": 0.8259,
      "step": 4235
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.011424655449596,
      "learning_rate": 1.7432992535506687e-06,
      "loss": 0.7662,
      "step": 4236
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.930310910339362,
      "learning_rate": 1.7397857042006194e-06,
      "loss": 0.7895,
      "step": 4237
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5829974648621259,
      "learning_rate": 1.736275361750167e-06,
      "loss": 0.33,
      "step": 4238
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9868947106107392,
      "learning_rate": 1.7327682275621506e-06,
      "loss": 0.798,
      "step": 4239
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0268675140893342,
      "learning_rate": 1.7292643029981525e-06,
      "loss": 0.8306,
      "step": 4240
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9504472030139467,
      "learning_rate": 1.7257635894185232e-06,
      "loss": 0.7945,
      "step": 4241
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0472838842980021,
      "learning_rate": 1.7222660881823594e-06,
      "loss": 0.8037,
      "step": 4242
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9103394038078558,
      "learning_rate": 1.7187718006475117e-06,
      "loss": 0.7585,
      "step": 4243
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6274787830471201,
      "learning_rate": 1.7152807281705809e-06,
      "loss": 0.3102,
      "step": 4244
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9981001434509197,
      "learning_rate": 1.7117928721069233e-06,
      "loss": 0.8355,
      "step": 4245
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6227927926801685,
      "learning_rate": 1.708308233810644e-06,
      "loss": 0.3176,
      "step": 4246
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.835417530494462,
      "learning_rate": 1.704826814634597e-06,
      "loss": 0.8105,
      "step": 4247
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9412236809535742,
      "learning_rate": 1.701348615930397e-06,
      "loss": 0.7987,
      "step": 4248
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8843831261555931,
      "learning_rate": 1.6978736390483896e-06,
      "loss": 0.7526,
      "step": 4249
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8140602805305069,
      "learning_rate": 1.6944018853376898e-06,
      "loss": 0.7454,
      "step": 4250
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9778708760344361,
      "learning_rate": 1.6909333561461471e-06,
      "loss": 0.7722,
      "step": 4251
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9637212113742789,
      "learning_rate": 1.6874680528203657e-06,
      "loss": 0.8346,
      "step": 4252
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9642856548486072,
      "learning_rate": 1.6840059767056949e-06,
      "loss": 0.814,
      "step": 4253
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.905896085157836,
      "learning_rate": 1.6805471291462316e-06,
      "loss": 0.7496,
      "step": 4254
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9734371825980394,
      "learning_rate": 1.6770915114848197e-06,
      "loss": 0.7641,
      "step": 4255
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8532593996395553,
      "learning_rate": 1.67363912506305e-06,
      "loss": 0.7697,
      "step": 4256
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9026846158001232,
      "learning_rate": 1.6701899712212565e-06,
      "loss": 0.7758,
      "step": 4257
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5889777353375228,
      "learning_rate": 1.66674405129852e-06,
      "loss": 0.32,
      "step": 4258
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.994020078246345,
      "learning_rate": 1.6633013666326636e-06,
      "loss": 0.8152,
      "step": 4259
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9216641643645339,
      "learning_rate": 1.6598619185602616e-06,
      "loss": 0.8273,
      "step": 4260
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9496272436031576,
      "learning_rate": 1.656425708416617e-06,
      "loss": 0.7833,
      "step": 4261
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9214040848591896,
      "learning_rate": 1.6529927375357957e-06,
      "loss": 0.799,
      "step": 4262
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.961919336135079,
      "learning_rate": 1.6495630072505841e-06,
      "loss": 0.8164,
      "step": 4263
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9623040828205482,
      "learning_rate": 1.6461365188925304e-06,
      "loss": 0.8193,
      "step": 4264
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9057558400099577,
      "learning_rate": 1.642713273791914e-06,
      "loss": 0.759,
      "step": 4265
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9144845392195674,
      "learning_rate": 1.6392932732777489e-06,
      "loss": 0.7377,
      "step": 4266
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8715390993545358,
      "learning_rate": 1.6358765186778057e-06,
      "loss": 0.7768,
      "step": 4267
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.837073902998664,
      "learning_rate": 1.6324630113185835e-06,
      "loss": 0.7897,
      "step": 4268
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8979174195789245,
      "learning_rate": 1.629052752525323e-06,
      "loss": 0.7758,
      "step": 4269
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8672712114886296,
      "learning_rate": 1.625645743622003e-06,
      "loss": 0.8108,
      "step": 4270
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0987056896622003,
      "learning_rate": 1.6222419859313443e-06,
      "loss": 0.8435,
      "step": 4271
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9316780639619258,
      "learning_rate": 1.6188414807747999e-06,
      "loss": 0.785,
      "step": 4272
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8554455988616709,
      "learning_rate": 1.6154442294725636e-06,
      "loss": 0.8302,
      "step": 4273
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0143336362133568,
      "learning_rate": 1.6120502333435695e-06,
      "loss": 0.8493,
      "step": 4274
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8994513922243665,
      "learning_rate": 1.6086594937054767e-06,
      "loss": 0.8389,
      "step": 4275
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8953544109352097,
      "learning_rate": 1.6052720118746923e-06,
      "loss": 0.7805,
      "step": 4276
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8725574587594171,
      "learning_rate": 1.6018877891663521e-06,
      "loss": 0.7604,
      "step": 4277
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8324507184269773,
      "learning_rate": 1.5985068268943283e-06,
      "loss": 0.7641,
      "step": 4278
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9591624826318664,
      "learning_rate": 1.5951291263712255e-06,
      "loss": 0.7924,
      "step": 4279
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8089716439751189,
      "learning_rate": 1.5917546889083834e-06,
      "loss": 0.7129,
      "step": 4280
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.884251284254815,
      "learning_rate": 1.5883835158158767e-06,
      "loss": 0.7746,
      "step": 4281
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8737010013422353,
      "learning_rate": 1.5850156084025091e-06,
      "loss": 0.7788,
      "step": 4282
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5567175017358135,
      "learning_rate": 1.5816509679758185e-06,
      "loss": 0.357,
      "step": 4283
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9235502813433227,
      "learning_rate": 1.578289595842074e-06,
      "loss": 0.7521,
      "step": 4284
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5914621366427504,
      "learning_rate": 1.5749314933062754e-06,
      "loss": 0.3224,
      "step": 4285
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8650266798202878,
      "learning_rate": 1.5715766616721584e-06,
      "loss": 0.7991,
      "step": 4286
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8826378829238718,
      "learning_rate": 1.5682251022421757e-06,
      "loss": 0.7535,
      "step": 4287
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8501279884133042,
      "learning_rate": 1.5648768163175277e-06,
      "loss": 0.7632,
      "step": 4288
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0159707402489984,
      "learning_rate": 1.5615318051981243e-06,
      "loss": 0.8357,
      "step": 4289
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9100306822353319,
      "learning_rate": 1.5581900701826226e-06,
      "loss": 0.7404,
      "step": 4290
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8724519946545426,
      "learning_rate": 1.5548516125683976e-06,
      "loss": 0.828,
      "step": 4291
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8713502687968242,
      "learning_rate": 1.5515164336515465e-06,
      "loss": 0.7434,
      "step": 4292
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8572520706214971,
      "learning_rate": 1.5481845347269077e-06,
      "loss": 0.7733,
      "step": 4293
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9180806573671245,
      "learning_rate": 1.5448559170880373e-06,
      "loss": 0.7633,
      "step": 4294
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9684416247434195,
      "learning_rate": 1.5415305820272198e-06,
      "loss": 0.8176,
      "step": 4295
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.927565482351307,
      "learning_rate": 1.5382085308354633e-06,
      "loss": 0.8003,
      "step": 4296
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9241397190545066,
      "learning_rate": 1.534889764802503e-06,
      "loss": 0.72,
      "step": 4297
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8907851389701109,
      "learning_rate": 1.5315742852167992e-06,
      "loss": 0.8026,
      "step": 4298
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9301027659797347,
      "learning_rate": 1.5282620933655312e-06,
      "loss": 0.8051,
      "step": 4299
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0474960515314096,
      "learning_rate": 1.5249531905346138e-06,
      "loss": 0.8344,
      "step": 4300
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8962985531301771,
      "learning_rate": 1.521647578008667e-06,
      "loss": 0.7787,
      "step": 4301
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9264374821011067,
      "learning_rate": 1.5183452570710522e-06,
      "loss": 0.7576,
      "step": 4302
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.840875027783641,
      "learning_rate": 1.5150462290038392e-06,
      "loss": 0.8029,
      "step": 4303
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.866970348394614,
      "learning_rate": 1.511750495087827e-06,
      "loss": 0.7511,
      "step": 4304
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9909324353396016,
      "learning_rate": 1.5084580566025309e-06,
      "loss": 0.8116,
      "step": 4305
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9090927509614657,
      "learning_rate": 1.5051689148261895e-06,
      "loss": 0.7448,
      "step": 4306
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8873986830792759,
      "learning_rate": 1.5018830710357612e-06,
      "loss": 0.7899,
      "step": 4307
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9532433341343033,
      "learning_rate": 1.4986005265069204e-06,
      "loss": 0.7493,
      "step": 4308
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9680973085084961,
      "learning_rate": 1.4953212825140728e-06,
      "loss": 0.8044,
      "step": 4309
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0484870398430226,
      "learning_rate": 1.4920453403303249e-06,
      "loss": 0.7978,
      "step": 4310
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9343027397758058,
      "learning_rate": 1.4887727012275112e-06,
      "loss": 0.7827,
      "step": 4311
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8709794715794785,
      "learning_rate": 1.4855033664761898e-06,
      "loss": 0.7825,
      "step": 4312
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.848550234992214,
      "learning_rate": 1.48223733734562e-06,
      "loss": 0.7468,
      "step": 4313
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8840000247233776,
      "learning_rate": 1.4789746151037942e-06,
      "loss": 0.7509,
      "step": 4314
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9642076381415672,
      "learning_rate": 1.475715201017407e-06,
      "loss": 0.8031,
      "step": 4315
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9289713586190181,
      "learning_rate": 1.4724590963518803e-06,
      "loss": 0.8239,
      "step": 4316
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.938174585316232,
      "learning_rate": 1.4692063023713444e-06,
      "loss": 0.8034,
      "step": 4317
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9069770064519574,
      "learning_rate": 1.4659568203386464e-06,
      "loss": 0.7514,
      "step": 4318
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9711680176931206,
      "learning_rate": 1.4627106515153456e-06,
      "loss": 0.7932,
      "step": 4319
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9565944069968948,
      "learning_rate": 1.4594677971617178e-06,
      "loss": 0.7702,
      "step": 4320
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9091084103553192,
      "learning_rate": 1.4562282585367493e-06,
      "loss": 0.8055,
      "step": 4321
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9274497269332194,
      "learning_rate": 1.452992036898142e-06,
      "loss": 0.7212,
      "step": 4322
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9397244909103273,
      "learning_rate": 1.4497591335023087e-06,
      "loss": 0.7915,
      "step": 4323
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8812324681127575,
      "learning_rate": 1.446529549604373e-06,
      "loss": 0.7504,
      "step": 4324
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9957036431370326,
      "learning_rate": 1.4433032864581687e-06,
      "loss": 0.7991,
      "step": 4325
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9124084445402763,
      "learning_rate": 1.4400803453162482e-06,
      "loss": 0.7802,
      "step": 4326
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8779770772839546,
      "learning_rate": 1.4368607274298596e-06,
      "loss": 0.8168,
      "step": 4327
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8767530484786039,
      "learning_rate": 1.4336444340489775e-06,
      "loss": 0.7838,
      "step": 4328
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9148230487265289,
      "learning_rate": 1.430431466422273e-06,
      "loss": 0.7804,
      "step": 4329
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9039003226721379,
      "learning_rate": 1.4272218257971327e-06,
      "loss": 0.8312,
      "step": 4330
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.911384141164536,
      "learning_rate": 1.4240155134196499e-06,
      "loss": 0.8362,
      "step": 4331
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9068127339917342,
      "learning_rate": 1.4208125305346232e-06,
      "loss": 0.8071,
      "step": 4332
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9943388269875496,
      "learning_rate": 1.4176128783855636e-06,
      "loss": 0.7868,
      "step": 4333
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9073902720003054,
      "learning_rate": 1.4144165582146819e-06,
      "loss": 0.787,
      "step": 4334
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9528540520265462,
      "learning_rate": 1.4112235712629063e-06,
      "loss": 0.794,
      "step": 4335
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8983252945057786,
      "learning_rate": 1.40803391876986e-06,
      "loss": 0.8144,
      "step": 4336
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.019202148871244,
      "learning_rate": 1.4048476019738756e-06,
      "loss": 0.8216,
      "step": 4337
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9069201690647171,
      "learning_rate": 1.4016646221119912e-06,
      "loss": 0.7387,
      "step": 4338
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9493841991787596,
      "learning_rate": 1.3984849804199485e-06,
      "loss": 0.7389,
      "step": 4339
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9065292345440988,
      "learning_rate": 1.395308678132199e-06,
      "loss": 0.8055,
      "step": 4340
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9417048928824572,
      "learning_rate": 1.392135716481885e-06,
      "loss": 0.7648,
      "step": 4341
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8973084191491382,
      "learning_rate": 1.3889660967008656e-06,
      "loss": 0.7616,
      "step": 4342
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8740197370759852,
      "learning_rate": 1.3857998200196943e-06,
      "loss": 0.7736,
      "step": 4343
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9410623178382123,
      "learning_rate": 1.3826368876676278e-06,
      "loss": 0.8034,
      "step": 4344
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8467409900672285,
      "learning_rate": 1.379477300872626e-06,
      "loss": 0.8145,
      "step": 4345
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8937422197312135,
      "learning_rate": 1.3763210608613497e-06,
      "loss": 0.7814,
      "step": 4346
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8795896115995728,
      "learning_rate": 1.3731681688591593e-06,
      "loss": 0.7988,
      "step": 4347
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9703530647247699,
      "learning_rate": 1.370018626090116e-06,
      "loss": 0.7613,
      "step": 4348
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0302811159055694,
      "learning_rate": 1.3668724337769823e-06,
      "loss": 0.8391,
      "step": 4349
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9814407830110098,
      "learning_rate": 1.3637295931412153e-06,
      "loss": 0.835,
      "step": 4350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8997050108825332,
      "learning_rate": 1.3605901054029746e-06,
      "loss": 0.8195,
      "step": 4351
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8789999494674401,
      "learning_rate": 1.3574539717811231e-06,
      "loss": 0.7927,
      "step": 4352
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9026442866909381,
      "learning_rate": 1.3543211934932065e-06,
      "loss": 0.7995,
      "step": 4353
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9684775696688712,
      "learning_rate": 1.3511917717554846e-06,
      "loss": 0.7827,
      "step": 4354
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9508937623954643,
      "learning_rate": 1.348065707782904e-06,
      "loss": 0.7651,
      "step": 4355
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8486242073015678,
      "learning_rate": 1.3449430027891096e-06,
      "loss": 0.784,
      "step": 4356
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9388074633880222,
      "learning_rate": 1.3418236579864452e-06,
      "loss": 0.8654,
      "step": 4357
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9556832873127648,
      "learning_rate": 1.338707674585945e-06,
      "loss": 0.7988,
      "step": 4358
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0027479700794861,
      "learning_rate": 1.3355950537973438e-06,
      "loss": 0.8422,
      "step": 4359
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9744791841461294,
      "learning_rate": 1.332485796829065e-06,
      "loss": 0.8475,
      "step": 4360
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5745269399839653,
      "learning_rate": 1.329379904888235e-06,
      "loss": 0.3188,
      "step": 4361
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8614258582326286,
      "learning_rate": 1.3262773791806617e-06,
      "loss": 0.7981,
      "step": 4362
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.916078460247318,
      "learning_rate": 1.3231782209108546e-06,
      "loss": 0.7753,
      "step": 4363
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9518479719225845,
      "learning_rate": 1.3200824312820137e-06,
      "loss": 0.8148,
      "step": 4364
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.918807284380159,
      "learning_rate": 1.3169900114960298e-06,
      "loss": 0.8132,
      "step": 4365
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9777605020790532,
      "learning_rate": 1.3139009627534927e-06,
      "loss": 0.8658,
      "step": 4366
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8179466082456519,
      "learning_rate": 1.3108152862536683e-06,
      "loss": 0.7916,
      "step": 4367
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9444440796520173,
      "learning_rate": 1.3077329831945295e-06,
      "loss": 0.7821,
      "step": 4368
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.023469404367568,
      "learning_rate": 1.3046540547727305e-06,
      "loss": 0.7483,
      "step": 4369
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9620566961876149,
      "learning_rate": 1.3015785021836159e-06,
      "loss": 0.8508,
      "step": 4370
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8996375483984788,
      "learning_rate": 1.2985063266212229e-06,
      "loss": 0.7893,
      "step": 4371
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.997432366407894,
      "learning_rate": 1.295437529278275e-06,
      "loss": 0.7662,
      "step": 4372
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5788072267052482,
      "learning_rate": 1.2923721113461852e-06,
      "loss": 0.3405,
      "step": 4373
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.891152687377223,
      "learning_rate": 1.2893100740150522e-06,
      "loss": 0.7618,
      "step": 4374
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9139814515382549,
      "learning_rate": 1.2862514184736695e-06,
      "loss": 0.7559,
      "step": 4375
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9762553120503225,
      "learning_rate": 1.2831961459095088e-06,
      "loss": 0.7696,
      "step": 4376
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9566948041414332,
      "learning_rate": 1.2801442575087296e-06,
      "loss": 0.709,
      "step": 4377
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9220072708969668,
      "learning_rate": 1.2770957544561868e-06,
      "loss": 0.8005,
      "step": 4378
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8946560731986778,
      "learning_rate": 1.274050637935408e-06,
      "loss": 0.7932,
      "step": 4379
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9181580918930831,
      "learning_rate": 1.2710089091286148e-06,
      "loss": 0.8362,
      "step": 4380
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9815998705347999,
      "learning_rate": 1.2679705692167122e-06,
      "loss": 0.7558,
      "step": 4381
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8561187334758205,
      "learning_rate": 1.2649356193792873e-06,
      "loss": 0.7776,
      "step": 4382
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8951918525329731,
      "learning_rate": 1.261904060794612e-06,
      "loss": 0.7629,
      "step": 4383
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9856326377643929,
      "learning_rate": 1.2588758946396417e-06,
      "loss": 0.8154,
      "step": 4384
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9785594769748144,
      "learning_rate": 1.2558511220900138e-06,
      "loss": 0.821,
      "step": 4385
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9635168270236434,
      "learning_rate": 1.2528297443200489e-06,
      "loss": 0.7841,
      "step": 4386
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.624785638627191,
      "learning_rate": 1.2498117625027562e-06,
      "loss": 0.3098,
      "step": 4387
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9944407265630326,
      "learning_rate": 1.246797177809812e-06,
      "loss": 0.783,
      "step": 4388
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9453731499675523,
      "learning_rate": 1.2437859914115847e-06,
      "loss": 0.7784,
      "step": 4389
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9199185483967512,
      "learning_rate": 1.2407782044771211e-06,
      "loss": 0.7792,
      "step": 4390
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8604058873402738,
      "learning_rate": 1.237773818174146e-06,
      "loss": 0.7039,
      "step": 4391
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9177006794234251,
      "learning_rate": 1.23477283366907e-06,
      "loss": 0.7663,
      "step": 4392
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9246780930461935,
      "learning_rate": 1.2317752521269722e-06,
      "loss": 0.6816,
      "step": 4393
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.007386953935457,
      "learning_rate": 1.2287810747116224e-06,
      "loss": 0.7944,
      "step": 4394
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.016806569096264,
      "learning_rate": 1.225790302585461e-06,
      "loss": 0.6862,
      "step": 4395
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9471319383247364,
      "learning_rate": 1.2228029369096094e-06,
      "loss": 0.8162,
      "step": 4396
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.949065571609412,
      "learning_rate": 1.2198189788438652e-06,
      "loss": 0.7714,
      "step": 4397
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8492013993109456,
      "learning_rate": 1.216838429546704e-06,
      "loss": 0.7158,
      "step": 4398
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.856386514883304,
      "learning_rate": 1.2138612901752777e-06,
      "loss": 0.769,
      "step": 4399
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6466667959291544,
      "learning_rate": 1.2108875618854122e-06,
      "loss": 0.3148,
      "step": 4400
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8467345456453694,
      "learning_rate": 1.2079172458316168e-06,
      "loss": 0.7544,
      "step": 4401
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0110136645383703,
      "learning_rate": 1.204950343167065e-06,
      "loss": 0.8245,
      "step": 4402
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9698769278987116,
      "learning_rate": 1.2019868550436099e-06,
      "loss": 0.8374,
      "step": 4403
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9998975173676492,
      "learning_rate": 1.1990267826117874e-06,
      "loss": 0.8129,
      "step": 4404
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8771250546697489,
      "learning_rate": 1.1960701270207885e-06,
      "loss": 0.7594,
      "step": 4405
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9631023366470297,
      "learning_rate": 1.1931168894184974e-06,
      "loss": 0.7796,
      "step": 4406
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8970615244820546,
      "learning_rate": 1.19016707095146e-06,
      "loss": 0.8039,
      "step": 4407
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.959668645424729,
      "learning_rate": 1.187220672764897e-06,
      "loss": 0.7751,
      "step": 4408
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8386793854097985,
      "learning_rate": 1.1842776960027014e-06,
      "loss": 0.7881,
      "step": 4409
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9284698374435386,
      "learning_rate": 1.1813381418074388e-06,
      "loss": 0.7894,
      "step": 4410
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.611316898562027,
      "learning_rate": 1.1784020113203453e-06,
      "loss": 0.3547,
      "step": 4411
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9259502678110607,
      "learning_rate": 1.1754693056813272e-06,
      "loss": 0.7153,
      "step": 4412
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8873481429635391,
      "learning_rate": 1.172540026028962e-06,
      "loss": 0.7502,
      "step": 4413
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8893258747131457,
      "learning_rate": 1.169614173500494e-06,
      "loss": 0.7214,
      "step": 4414
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9248231789795627,
      "learning_rate": 1.1666917492318486e-06,
      "loss": 0.7383,
      "step": 4415
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9315716877409536,
      "learning_rate": 1.1637727543576027e-06,
      "loss": 0.7867,
      "step": 4416
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8821715112995467,
      "learning_rate": 1.1608571900110122e-06,
      "loss": 0.7637,
      "step": 4417
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9642229341624187,
      "learning_rate": 1.1579450573240058e-06,
      "loss": 0.7539,
      "step": 4418
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.927544243553698,
      "learning_rate": 1.1550363574271638e-06,
      "loss": 0.811,
      "step": 4419
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.912674968586832,
      "learning_rate": 1.1521310914497518e-06,
      "loss": 0.7952,
      "step": 4420
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9966023422967517,
      "learning_rate": 1.149229260519691e-06,
      "loss": 0.7675,
      "step": 4421
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9356608825306343,
      "learning_rate": 1.1463308657635718e-06,
      "loss": 0.8409,
      "step": 4422
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.931916797613099,
      "learning_rate": 1.1434359083066515e-06,
      "loss": 0.8043,
      "step": 4423
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9671479348793389,
      "learning_rate": 1.140544389272853e-06,
      "loss": 0.8057,
      "step": 4424
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9931479482506265,
      "learning_rate": 1.1376563097847616e-06,
      "loss": 0.7283,
      "step": 4425
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8916933045396369,
      "learning_rate": 1.1347716709636282e-06,
      "loss": 0.7638,
      "step": 4426
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8841607017145052,
      "learning_rate": 1.1318904739293745e-06,
      "loss": 0.7265,
      "step": 4427
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9745224720925167,
      "learning_rate": 1.129012719800575e-06,
      "loss": 0.7757,
      "step": 4428
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9050331106638091,
      "learning_rate": 1.1261384096944728e-06,
      "loss": 0.7507,
      "step": 4429
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9637152387542668,
      "learning_rate": 1.1232675447269803e-06,
      "loss": 0.7341,
      "step": 4430
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8768462772314612,
      "learning_rate": 1.1204001260126574e-06,
      "loss": 0.7626,
      "step": 4431
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0384282545123487,
      "learning_rate": 1.1175361546647413e-06,
      "loss": 0.7529,
      "step": 4432
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.875771413444998,
      "learning_rate": 1.1146756317951224e-06,
      "loss": 0.7599,
      "step": 4433
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9877451138117046,
      "learning_rate": 1.1118185585143536e-06,
      "loss": 0.7977,
      "step": 4434
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9680401427824781,
      "learning_rate": 1.1089649359316501e-06,
      "loss": 0.7861,
      "step": 4435
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8488125291162694,
      "learning_rate": 1.1061147651548855e-06,
      "loss": 0.8011,
      "step": 4436
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9548618325513104,
      "learning_rate": 1.1032680472905932e-06,
      "loss": 0.7494,
      "step": 4437
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0391960929824169,
      "learning_rate": 1.1004247834439697e-06,
      "loss": 0.7933,
      "step": 4438
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9375303725448659,
      "learning_rate": 1.097584974718866e-06,
      "loss": 0.7941,
      "step": 4439
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9118764791088955,
      "learning_rate": 1.0947486222177928e-06,
      "loss": 0.7901,
      "step": 4440
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9303093468897271,
      "learning_rate": 1.0919157270419257e-06,
      "loss": 0.7459,
      "step": 4441
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8437580555568817,
      "learning_rate": 1.0890862902910849e-06,
      "loss": 0.7908,
      "step": 4442
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9327260627906601,
      "learning_rate": 1.0862603130637562e-06,
      "loss": 0.7649,
      "step": 4443
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9337856920151959,
      "learning_rate": 1.0834377964570863e-06,
      "loss": 0.7264,
      "step": 4444
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9118197932787658,
      "learning_rate": 1.0806187415668668e-06,
      "loss": 0.7698,
      "step": 4445
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9099123250839759,
      "learning_rate": 1.0778031494875574e-06,
      "loss": 0.7596,
      "step": 4446
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9372319623121994,
      "learning_rate": 1.0749910213122649e-06,
      "loss": 0.7354,
      "step": 4447
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9028034419726554,
      "learning_rate": 1.072182358132755e-06,
      "loss": 0.7323,
      "step": 4448
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8907401178858971,
      "learning_rate": 1.0693771610394477e-06,
      "loss": 0.7712,
      "step": 4449
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0548650646191111,
      "learning_rate": 1.066575431121417e-06,
      "loss": 0.81,
      "step": 4450
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8984100261146734,
      "learning_rate": 1.06377716946639e-06,
      "loss": 0.7902,
      "step": 4451
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9385082347945755,
      "learning_rate": 1.0609823771607487e-06,
      "loss": 0.7795,
      "step": 4452
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8615242620949355,
      "learning_rate": 1.0581910552895302e-06,
      "loss": 0.8036,
      "step": 4453
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9066525962146907,
      "learning_rate": 1.055403204936416e-06,
      "loss": 0.8034,
      "step": 4454
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9767568324566025,
      "learning_rate": 1.0526188271837512e-06,
      "loss": 0.786,
      "step": 4455
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9305159048321721,
      "learning_rate": 1.0498379231125278e-06,
      "loss": 0.8094,
      "step": 4456
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9120232251169853,
      "learning_rate": 1.047060493802381e-06,
      "loss": 0.8035,
      "step": 4457
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9145895356044104,
      "learning_rate": 1.0442865403316117e-06,
      "loss": 0.7272,
      "step": 4458
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6576139187497894,
      "learning_rate": 1.0415160637771604e-06,
      "loss": 0.3692,
      "step": 4459
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9096247148417228,
      "learning_rate": 1.0387490652146236e-06,
      "loss": 0.8566,
      "step": 4460
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9403318390641534,
      "learning_rate": 1.0359855457182455e-06,
      "loss": 0.7455,
      "step": 4461
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9404582668430166,
      "learning_rate": 1.0332255063609177e-06,
      "loss": 0.7953,
      "step": 4462
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9157699104677942,
      "learning_rate": 1.0304689482141839e-06,
      "loss": 0.8136,
      "step": 4463
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.908270366804635,
      "learning_rate": 1.027715872348234e-06,
      "loss": 0.7616,
      "step": 4464
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9044831168606051,
      "learning_rate": 1.0249662798319072e-06,
      "loss": 0.7643,
      "step": 4465
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0543204779366204,
      "learning_rate": 1.0222201717326885e-06,
      "loss": 0.7658,
      "step": 4466
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9272423005139472,
      "learning_rate": 1.0194775491167164e-06,
      "loss": 0.7727,
      "step": 4467
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9171548381496207,
      "learning_rate": 1.0167384130487667e-06,
      "loss": 0.7955,
      "step": 4468
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9973266842441105,
      "learning_rate": 1.0140027645922656e-06,
      "loss": 0.7257,
      "step": 4469
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0032620218315025,
      "learning_rate": 1.0112706048092924e-06,
      "loss": 0.7971,
      "step": 4470
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8902488098830124,
      "learning_rate": 1.0085419347605575e-06,
      "loss": 0.7914,
      "step": 4471
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.860538853059453,
      "learning_rate": 1.00581675550543e-06,
      "loss": 0.7853,
      "step": 4472
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9055263017988606,
      "learning_rate": 1.003095068101917e-06,
      "loss": 0.7174,
      "step": 4473
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9839496989952973,
      "learning_rate": 1.0003768736066722e-06,
      "loss": 0.7421,
      "step": 4474
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9470963996715289,
      "learning_rate": 9.976621730749892e-07,
      "loss": 0.7983,
      "step": 4475
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.042332484174126,
      "learning_rate": 9.949509675608115e-07,
      "loss": 0.7735,
      "step": 4476
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8902016282283364,
      "learning_rate": 9.922432581167207e-07,
      "loss": 0.8131,
      "step": 4477
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9553042467866402,
      "learning_rate": 9.895390457939414e-07,
      "loss": 0.7968,
      "step": 4478
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8846868041488697,
      "learning_rate": 9.86838331642348e-07,
      "loss": 0.7665,
      "step": 4479
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0348644577461374,
      "learning_rate": 9.84141116710442e-07,
      "loss": 0.7809,
      "step": 4480
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9482251630083754,
      "learning_rate": 9.814474020453824e-07,
      "loss": 0.6908,
      "step": 4481
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9706087222430025,
      "learning_rate": 9.787571886929604e-07,
      "loss": 0.829,
      "step": 4482
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8678284779326667,
      "learning_rate": 9.76070477697605e-07,
      "loss": 0.747,
      "step": 4483
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9439922244001654,
      "learning_rate": 9.733872701023938e-07,
      "loss": 0.7981,
      "step": 4484
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9606497562018557,
      "learning_rate": 9.707075669490407e-07,
      "loss": 0.7956,
      "step": 4485
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.845114603660183,
      "learning_rate": 9.680313692778976e-07,
      "loss": 0.7301,
      "step": 4486
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9627853445052444,
      "learning_rate": 9.653586781279567e-07,
      "loss": 0.7414,
      "step": 4487
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9812637353291883,
      "learning_rate": 9.626894945368492e-07,
      "loss": 0.7611,
      "step": 4488
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9430560779262724,
      "learning_rate": 9.600238195408428e-07,
      "loss": 0.7991,
      "step": 4489
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9396439215310333,
      "learning_rate": 9.573616541748464e-07,
      "loss": 0.7687,
      "step": 4490
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8607619306987945,
      "learning_rate": 9.547029994724023e-07,
      "loss": 0.7335,
      "step": 4491
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9921938368452268,
      "learning_rate": 9.520478564656898e-07,
      "loss": 0.7411,
      "step": 4492
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.915248479334553,
      "learning_rate": 9.49396226185535e-07,
      "loss": 0.7718,
      "step": 4493
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8929969661270961,
      "learning_rate": 9.467481096613829e-07,
      "loss": 0.7729,
      "step": 4494
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.051329233863551,
      "learning_rate": 9.441035079213267e-07,
      "loss": 0.8262,
      "step": 4495
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9228112586620948,
      "learning_rate": 9.414624219920953e-07,
      "loss": 0.7783,
      "step": 4496
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8595921790627419,
      "learning_rate": 9.38824852899043e-07,
      "loss": 0.7387,
      "step": 4497
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9212212496106614,
      "learning_rate": 9.361908016661703e-07,
      "loss": 0.7856,
      "step": 4498
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6004100028204391,
      "learning_rate": 9.335602693161039e-07,
      "loss": 0.3157,
      "step": 4499
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8947996518747799,
      "learning_rate": 9.309332568701079e-07,
      "loss": 0.8001,
      "step": 4500
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8935462578652081,
      "learning_rate": 9.283097653480788e-07,
      "loss": 0.7647,
      "step": 4501
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9218315254685184,
      "learning_rate": 9.256897957685463e-07,
      "loss": 0.8501,
      "step": 4502
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.879929223381953,
      "learning_rate": 9.230733491486721e-07,
      "loss": 0.7648,
      "step": 4503
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8857721345490582,
      "learning_rate": 9.204604265042505e-07,
      "loss": 0.7884,
      "step": 4504
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8775651625042505,
      "learning_rate": 9.178510288497123e-07,
      "loss": 0.7228,
      "step": 4505
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0279386256708374,
      "learning_rate": 9.15245157198108e-07,
      "loss": 0.7876,
      "step": 4506
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9425985521722622,
      "learning_rate": 9.126428125611342e-07,
      "loss": 0.7992,
      "step": 4507
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8407110342318475,
      "learning_rate": 9.10043995949108e-07,
      "loss": 0.712,
      "step": 4508
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9588246902090991,
      "learning_rate": 9.074487083709759e-07,
      "loss": 0.8286,
      "step": 4509
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.929265188695774,
      "learning_rate": 9.04856950834323e-07,
      "loss": 0.8325,
      "step": 4510
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9581541650528325,
      "learning_rate": 9.022687243453554e-07,
      "loss": 0.7765,
      "step": 4511
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9427704753704729,
      "learning_rate": 8.996840299089149e-07,
      "loss": 0.7592,
      "step": 4512
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8895424835092584,
      "learning_rate": 8.971028685284655e-07,
      "loss": 0.7826,
      "step": 4513
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8885106337379941,
      "learning_rate": 8.945252412061056e-07,
      "loss": 0.77,
      "step": 4514
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0315063698173472,
      "learning_rate": 8.91951148942557e-07,
      "loss": 0.825,
      "step": 4515
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9396911422924439,
      "learning_rate": 8.893805927371724e-07,
      "loss": 0.7984,
      "step": 4516
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8972678749959699,
      "learning_rate": 8.868135735879291e-07,
      "loss": 0.806,
      "step": 4517
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9351572878572095,
      "learning_rate": 8.842500924914299e-07,
      "loss": 0.8399,
      "step": 4518
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.92798363424012,
      "learning_rate": 8.816901504429143e-07,
      "loss": 0.7633,
      "step": 4519
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8941696252151938,
      "learning_rate": 8.791337484362305e-07,
      "loss": 0.7669,
      "step": 4520
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9072379028383514,
      "learning_rate": 8.765808874638682e-07,
      "loss": 0.7498,
      "step": 4521
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8415859609827878,
      "learning_rate": 8.740315685169364e-07,
      "loss": 0.7809,
      "step": 4522
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9331266000571716,
      "learning_rate": 8.714857925851617e-07,
      "loss": 0.7719,
      "step": 4523
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9400270611266179,
      "learning_rate": 8.689435606569086e-07,
      "loss": 0.8473,
      "step": 4524
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.878588228439865,
      "learning_rate": 8.664048737191566e-07,
      "loss": 0.8116,
      "step": 4525
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9364563083307438,
      "learning_rate": 8.638697327575108e-07,
      "loss": 0.8119,
      "step": 4526
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9272281106126047,
      "learning_rate": 8.613381387562015e-07,
      "loss": 0.8127,
      "step": 4527
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.915813536861963,
      "learning_rate": 8.588100926980802e-07,
      "loss": 0.7646,
      "step": 4528
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9907680112619857,
      "learning_rate": 8.56285595564621e-07,
      "loss": 0.8027,
      "step": 4529
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8783927213307087,
      "learning_rate": 8.537646483359185e-07,
      "loss": 0.6981,
      "step": 4530
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9000277566375604,
      "learning_rate": 8.512472519906978e-07,
      "loss": 0.8277,
      "step": 4531
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0254837710217162,
      "learning_rate": 8.487334075062914e-07,
      "loss": 0.8184,
      "step": 4532
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8671593496386645,
      "learning_rate": 8.462231158586654e-07,
      "loss": 0.7487,
      "step": 4533
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6181743871058648,
      "learning_rate": 8.437163780224011e-07,
      "loss": 0.3176,
      "step": 4534
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9201208557986177,
      "learning_rate": 8.412131949706958e-07,
      "loss": 0.8324,
      "step": 4535
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9885610134520626,
      "learning_rate": 8.387135676753755e-07,
      "loss": 0.7707,
      "step": 4536
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9740277310137743,
      "learning_rate": 8.362174971068804e-07,
      "loss": 0.8013,
      "step": 4537
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8870723225302317,
      "learning_rate": 8.337249842342721e-07,
      "loss": 0.7597,
      "step": 4538
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9097007611435771,
      "learning_rate": 8.312360300252287e-07,
      "loss": 0.7732,
      "step": 4539
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9289302564455305,
      "learning_rate": 8.287506354460484e-07,
      "loss": 0.809,
      "step": 4540
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9080938934965168,
      "learning_rate": 8.26268801461646e-07,
      "loss": 0.7811,
      "step": 4541
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9365090278575023,
      "learning_rate": 8.237905290355563e-07,
      "loss": 0.769,
      "step": 4542
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9125179935411829,
      "learning_rate": 8.213158191299297e-07,
      "loss": 0.7785,
      "step": 4543
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.004662938943116,
      "learning_rate": 8.188446727055311e-07,
      "loss": 0.7348,
      "step": 4544
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9619486153878815,
      "learning_rate": 8.163770907217506e-07,
      "loss": 0.7955,
      "step": 4545
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9572961528212552,
      "learning_rate": 8.139130741365819e-07,
      "loss": 0.7905,
      "step": 4546
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8673610243812715,
      "learning_rate": 8.114526239066456e-07,
      "loss": 0.7995,
      "step": 4547
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9380298752476245,
      "learning_rate": 8.08995740987173e-07,
      "loss": 0.7675,
      "step": 4548
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8721306963890196,
      "learning_rate": 8.065424263320054e-07,
      "loss": 0.7583,
      "step": 4549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.852587333385394,
      "learning_rate": 8.040926808936112e-07,
      "loss": 0.7568,
      "step": 4550
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8718909380028156,
      "learning_rate": 8.016465056230616e-07,
      "loss": 0.7883,
      "step": 4551
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9169643746241978,
      "learning_rate": 7.99203901470047e-07,
      "loss": 0.7843,
      "step": 4552
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9698589108344413,
      "learning_rate": 7.967648693828712e-07,
      "loss": 0.7477,
      "step": 4553
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8218851715008519,
      "learning_rate": 7.943294103084487e-07,
      "loss": 0.7464,
      "step": 4554
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8899550576901735,
      "learning_rate": 7.9189752519231e-07,
      "loss": 0.7151,
      "step": 4555
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8889649175253955,
      "learning_rate": 7.894692149785954e-07,
      "loss": 0.7626,
      "step": 4556
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8691114986018353,
      "learning_rate": 7.870444806100619e-07,
      "loss": 0.7698,
      "step": 4557
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.890012188317992,
      "learning_rate": 7.846233230280698e-07,
      "loss": 0.7464,
      "step": 4558
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9622410508915916,
      "learning_rate": 7.822057431725994e-07,
      "loss": 0.7895,
      "step": 4559
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.904556882644464,
      "learning_rate": 7.797917419822377e-07,
      "loss": 0.7669,
      "step": 4560
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0196269927158965,
      "learning_rate": 7.773813203941827e-07,
      "loss": 0.7942,
      "step": 4561
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8797275496724145,
      "learning_rate": 7.749744793442448e-07,
      "loss": 0.7799,
      "step": 4562
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9452601263861672,
      "learning_rate": 7.725712197668378e-07,
      "loss": 0.7471,
      "step": 4563
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8887738978856351,
      "learning_rate": 7.701715425949952e-07,
      "loss": 0.7639,
      "step": 4564
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8749254201020912,
      "learning_rate": 7.677754487603517e-07,
      "loss": 0.7285,
      "step": 4565
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9375291880643531,
      "learning_rate": 7.653829391931533e-07,
      "loss": 0.6632,
      "step": 4566
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9152997056883132,
      "learning_rate": 7.629940148222559e-07,
      "loss": 0.7147,
      "step": 4567
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.847729984674189,
      "learning_rate": 7.606086765751209e-07,
      "loss": 0.7458,
      "step": 4568
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9865404651126207,
      "learning_rate": 7.582269253778185e-07,
      "loss": 0.8169,
      "step": 4569
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9724732306031515,
      "learning_rate": 7.55848762155027e-07,
      "loss": 0.7786,
      "step": 4570
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8464744143313666,
      "learning_rate": 7.534741878300333e-07,
      "loss": 0.7475,
      "step": 4571
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8585763450991851,
      "learning_rate": 7.511032033247256e-07,
      "loss": 0.7725,
      "step": 4572
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9978632826899884,
      "learning_rate": 7.487358095596031e-07,
      "loss": 0.8545,
      "step": 4573
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9106644626972529,
      "learning_rate": 7.463720074537728e-07,
      "loss": 0.822,
      "step": 4574
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8899924069206102,
      "learning_rate": 7.440117979249362e-07,
      "loss": 0.7963,
      "step": 4575
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9323720672294622,
      "learning_rate": 7.416551818894158e-07,
      "loss": 0.8207,
      "step": 4576
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8354773332757944,
      "learning_rate": 7.393021602621264e-07,
      "loss": 0.7751,
      "step": 4577
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9041314025160673,
      "learning_rate": 7.369527339565951e-07,
      "loss": 0.7792,
      "step": 4578
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8948472410500361,
      "learning_rate": 7.346069038849469e-07,
      "loss": 0.7465,
      "step": 4579
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8597456498514192,
      "learning_rate": 7.322646709579173e-07,
      "loss": 0.7321,
      "step": 4580
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9626109007572164,
      "learning_rate": 7.299260360848382e-07,
      "loss": 0.7925,
      "step": 4581
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9359491365454459,
      "learning_rate": 7.275910001736497e-07,
      "loss": 0.7784,
      "step": 4582
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.932999297573922,
      "learning_rate": 7.252595641308957e-07,
      "loss": 0.8232,
      "step": 4583
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9592630925528166,
      "learning_rate": 7.229317288617144e-07,
      "loss": 0.7811,
      "step": 4584
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9372350556791639,
      "learning_rate": 7.20607495269856e-07,
      "loss": 0.7104,
      "step": 4585
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8887260903861193,
      "learning_rate": 7.182868642576679e-07,
      "loss": 0.7407,
      "step": 4586
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9559937878783779,
      "learning_rate": 7.15969836726097e-07,
      "loss": 0.7738,
      "step": 4587
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9082138205739344,
      "learning_rate": 7.13656413574696e-07,
      "loss": 0.7409,
      "step": 4588
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8818271254736044,
      "learning_rate": 7.113465957016097e-07,
      "loss": 0.8091,
      "step": 4589
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8791549464539143,
      "learning_rate": 7.090403840035942e-07,
      "loss": 0.7684,
      "step": 4590
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9357241765422598,
      "learning_rate": 7.067377793759999e-07,
      "loss": 0.7734,
      "step": 4591
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9475909979444586,
      "learning_rate": 7.044387827127752e-07,
      "loss": 0.795,
      "step": 4592
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8957819415084309,
      "learning_rate": 7.021433949064704e-07,
      "loss": 0.7711,
      "step": 4593
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9445193015546554,
      "learning_rate": 6.99851616848235e-07,
      "loss": 0.7779,
      "step": 4594
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.022148199264715,
      "learning_rate": 6.975634494278149e-07,
      "loss": 0.7431,
      "step": 4595
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9245132568167672,
      "learning_rate": 6.952788935335541e-07,
      "loss": 0.7604,
      "step": 4596
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8845932188066742,
      "learning_rate": 6.92997950052402e-07,
      "loss": 0.7836,
      "step": 4597
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8842827291498436,
      "learning_rate": 6.907206198698912e-07,
      "loss": 0.7812,
      "step": 4598
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9080331502309776,
      "learning_rate": 6.884469038701646e-07,
      "loss": 0.7656,
      "step": 4599
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9551599446551602,
      "learning_rate": 6.861768029359595e-07,
      "loss": 0.7895,
      "step": 4600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8845639635836527,
      "learning_rate": 6.839103179485995e-07,
      "loss": 0.7995,
      "step": 4601
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8966843125877966,
      "learning_rate": 6.816474497880177e-07,
      "loss": 0.7796,
      "step": 4602
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9025193205911222,
      "learning_rate": 6.793881993327366e-07,
      "loss": 0.7608,
      "step": 4603
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9161827630350536,
      "learning_rate": 6.77132567459875e-07,
      "loss": 0.7672,
      "step": 4604
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.909449555589964,
      "learning_rate": 6.748805550451453e-07,
      "loss": 0.7845,
      "step": 4605
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9299701467129678,
      "learning_rate": 6.726321629628585e-07,
      "loss": 0.8152,
      "step": 4606
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9442024370819656,
      "learning_rate": 6.703873920859161e-07,
      "loss": 0.7593,
      "step": 4607
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8676051049060763,
      "learning_rate": 6.681462432858154e-07,
      "loss": 0.7914,
      "step": 4608
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5850836405441265,
      "learning_rate": 6.659087174326506e-07,
      "loss": 0.3259,
      "step": 4609
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9337978363173466,
      "learning_rate": 6.636748153951e-07,
      "loss": 0.7805,
      "step": 4610
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6110505448495223,
      "learning_rate": 6.614445380404478e-07,
      "loss": 0.3425,
      "step": 4611
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8738790621936945,
      "learning_rate": 6.592178862345622e-07,
      "loss": 0.7308,
      "step": 4612
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9807426419872222,
      "learning_rate": 6.569948608419041e-07,
      "loss": 0.7954,
      "step": 4613
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9615667689236103,
      "learning_rate": 6.547754627255332e-07,
      "loss": 0.7797,
      "step": 4614
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9351505492041804,
      "learning_rate": 6.52559692747089e-07,
      "loss": 0.7965,
      "step": 4615
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0168243160952424,
      "learning_rate": 6.503475517668168e-07,
      "loss": 0.7782,
      "step": 4616
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9408083693758162,
      "learning_rate": 6.481390406435417e-07,
      "loss": 0.7638,
      "step": 4617
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0468739473547555,
      "learning_rate": 6.459341602346858e-07,
      "loss": 0.8049,
      "step": 4618
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5803955900018661,
      "learning_rate": 6.437329113962576e-07,
      "loss": 0.3574,
      "step": 4619
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9038467339928642,
      "learning_rate": 6.415352949828601e-07,
      "loss": 0.7867,
      "step": 4620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9840305381806581,
      "learning_rate": 6.393413118476821e-07,
      "loss": 0.8051,
      "step": 4621
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9010339701696835,
      "learning_rate": 6.371509628425021e-07,
      "loss": 0.7751,
      "step": 4622
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9156012924847075,
      "learning_rate": 6.349642488176943e-07,
      "loss": 0.7791,
      "step": 4623
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9597499762102785,
      "learning_rate": 6.327811706222097e-07,
      "loss": 0.785,
      "step": 4624
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9301115831113196,
      "learning_rate": 6.306017291035981e-07,
      "loss": 0.7796,
      "step": 4625
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9186613632830745,
      "learning_rate": 6.284259251079939e-07,
      "loss": 0.7374,
      "step": 4626
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8997171509477017,
      "learning_rate": 6.262537594801177e-07,
      "loss": 0.7662,
      "step": 4627
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9643660838110291,
      "learning_rate": 6.240852330632796e-07,
      "loss": 0.7714,
      "step": 4628
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8931476301629024,
      "learning_rate": 6.219203466993762e-07,
      "loss": 0.7803,
      "step": 4629
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8533847018811133,
      "learning_rate": 6.197591012288918e-07,
      "loss": 0.7586,
      "step": 4630
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0181149263048979,
      "learning_rate": 6.17601497490895e-07,
      "loss": 0.7726,
      "step": 4631
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.914592375626657,
      "learning_rate": 6.154475363230417e-07,
      "loss": 0.8093,
      "step": 4632
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.032415303962788,
      "learning_rate": 6.132972185615749e-07,
      "loss": 0.7976,
      "step": 4633
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.896408504166702,
      "learning_rate": 6.111505450413202e-07,
      "loss": 0.8334,
      "step": 4634
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9490987596151738,
      "learning_rate": 6.090075165956943e-07,
      "loss": 0.7829,
      "step": 4635
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8747099610971997,
      "learning_rate": 6.068681340566896e-07,
      "loss": 0.7363,
      "step": 4636
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9082699663935254,
      "learning_rate": 6.047323982548924e-07,
      "loss": 0.7803,
      "step": 4637
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9401138303571317,
      "learning_rate": 6.026003100194633e-07,
      "loss": 0.8229,
      "step": 4638
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9625431213786401,
      "learning_rate": 6.004718701781575e-07,
      "loss": 0.7797,
      "step": 4639
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8546140966645465,
      "learning_rate": 5.983470795573088e-07,
      "loss": 0.7756,
      "step": 4640
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9734709159334023,
      "learning_rate": 5.962259389818292e-07,
      "loss": 0.7759,
      "step": 4641
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6360285862082997,
      "learning_rate": 5.941084492752236e-07,
      "loss": 0.3235,
      "step": 4642
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8970112687159878,
      "learning_rate": 5.91994611259572e-07,
      "loss": 0.7702,
      "step": 4643
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9095107337851629,
      "learning_rate": 5.898844257555392e-07,
      "loss": 0.8042,
      "step": 4644
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.973010905389458,
      "learning_rate": 5.87777893582372e-07,
      "loss": 0.7482,
      "step": 4645
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9022012397805347,
      "learning_rate": 5.856750155578983e-07,
      "loss": 0.7793,
      "step": 4646
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9322984315473137,
      "learning_rate": 5.835757924985286e-07,
      "loss": 0.8008,
      "step": 4647
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9563843199765101,
      "learning_rate": 5.81480225219252e-07,
      "loss": 0.7771,
      "step": 4648
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8801431578997486,
      "learning_rate": 5.793883145336443e-07,
      "loss": 0.7762,
      "step": 4649
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9369483303119786,
      "learning_rate": 5.773000612538505e-07,
      "loss": 0.8208,
      "step": 4650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.926609234913799,
      "learning_rate": 5.752154661906085e-07,
      "loss": 0.7935,
      "step": 4651
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9584351614737223,
      "learning_rate": 5.731345301532265e-07,
      "loss": 0.7902,
      "step": 4652
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8939825094313234,
      "learning_rate": 5.710572539495962e-07,
      "loss": 0.7397,
      "step": 4653
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9754756534853958,
      "learning_rate": 5.68983638386188e-07,
      "loss": 0.8332,
      "step": 4654
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8817074124668909,
      "learning_rate": 5.669136842680512e-07,
      "loss": 0.7976,
      "step": 4655
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9057003200290427,
      "learning_rate": 5.648473923988129e-07,
      "loss": 0.7326,
      "step": 4656
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8631098708487901,
      "learning_rate": 5.627847635806771e-07,
      "loss": 0.7204,
      "step": 4657
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9064291615599737,
      "learning_rate": 5.607257986144321e-07,
      "loss": 0.7534,
      "step": 4658
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8548051498723044,
      "learning_rate": 5.58670498299434e-07,
      "loss": 0.7967,
      "step": 4659
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8969860595050584,
      "learning_rate": 5.566188634336212e-07,
      "loss": 0.7643,
      "step": 4660
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9778406390633492,
      "learning_rate": 5.545708948135142e-07,
      "loss": 0.7684,
      "step": 4661
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9932035058083903,
      "learning_rate": 5.525265932341984e-07,
      "loss": 0.8049,
      "step": 4662
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8732366739609658,
      "learning_rate": 5.504859594893475e-07,
      "loss": 0.8137,
      "step": 4663
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9399639401893006,
      "learning_rate": 5.484489943712013e-07,
      "loss": 0.7203,
      "step": 4664
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9057741831361419,
      "learning_rate": 5.464156986705826e-07,
      "loss": 0.7999,
      "step": 4665
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8920253326186462,
      "learning_rate": 5.443860731768869e-07,
      "loss": 0.8138,
      "step": 4666
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8914199036374576,
      "learning_rate": 5.423601186780836e-07,
      "loss": 0.7842,
      "step": 4667
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8501501313676257,
      "learning_rate": 5.403378359607181e-07,
      "loss": 0.7815,
      "step": 4668
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9368806852228799,
      "learning_rate": 5.383192258099113e-07,
      "loss": 0.7851,
      "step": 4669
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9581175629718838,
      "learning_rate": 5.36304289009355e-07,
      "loss": 0.7592,
      "step": 4670
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8784481514404464,
      "learning_rate": 5.342930263413193e-07,
      "loss": 0.726,
      "step": 4671
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8939936298473287,
      "learning_rate": 5.322854385866439e-07,
      "loss": 0.7277,
      "step": 4672
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9616302939416792,
      "learning_rate": 5.302815265247452e-07,
      "loss": 0.828,
      "step": 4673
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9465244200315954,
      "learning_rate": 5.282812909336077e-07,
      "loss": 0.7292,
      "step": 4674
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9321639132832595,
      "learning_rate": 5.262847325897968e-07,
      "loss": 0.7576,
      "step": 4675
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9142697755947302,
      "learning_rate": 5.242918522684392e-07,
      "loss": 0.8022,
      "step": 4676
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9033819577086369,
      "learning_rate": 5.22302650743245e-07,
      "loss": 0.8085,
      "step": 4677
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9064319151570717,
      "learning_rate": 5.203171287864872e-07,
      "loss": 0.77,
      "step": 4678
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.628579757074144,
      "learning_rate": 5.183352871690162e-07,
      "loss": 0.3481,
      "step": 4679
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9506510740996696,
      "learning_rate": 5.163571266602485e-07,
      "loss": 0.7748,
      "step": 4680
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8823574420899475,
      "learning_rate": 5.143826480281778e-07,
      "loss": 0.7536,
      "step": 4681
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.98779296073593,
      "learning_rate": 5.124118520393606e-07,
      "loss": 0.8184,
      "step": 4682
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9400395194779139,
      "learning_rate": 5.104447394589295e-07,
      "loss": 0.7475,
      "step": 4683
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9057242437152196,
      "learning_rate": 5.084813110505871e-07,
      "loss": 0.8081,
      "step": 4684
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9590397129776954,
      "learning_rate": 5.065215675766023e-07,
      "loss": 0.7844,
      "step": 4685
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8641710139454776,
      "learning_rate": 5.045655097978131e-07,
      "loss": 0.7651,
      "step": 4686
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.96095447917801,
      "learning_rate": 5.026131384736321e-07,
      "loss": 0.7762,
      "step": 4687
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8671582712153479,
      "learning_rate": 5.006644543620342e-07,
      "loss": 0.8324,
      "step": 4688
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9204105971661632,
      "learning_rate": 4.987194582195687e-07,
      "loss": 0.7668,
      "step": 4689
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8656380881557763,
      "learning_rate": 4.967781508013459e-07,
      "loss": 0.7768,
      "step": 4690
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8837060446764569,
      "learning_rate": 4.948405328610506e-07,
      "loss": 0.8019,
      "step": 4691
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9402874990621554,
      "learning_rate": 4.929066051509346e-07,
      "loss": 0.8314,
      "step": 4692
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9578784244987527,
      "learning_rate": 4.909763684218116e-07,
      "loss": 0.718,
      "step": 4693
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9325103070644498,
      "learning_rate": 4.890498234230689e-07,
      "loss": 0.7737,
      "step": 4694
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9223507829327242,
      "learning_rate": 4.871269709026561e-07,
      "loss": 0.7916,
      "step": 4695
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9675984108877012,
      "learning_rate": 4.852078116070902e-07,
      "loss": 0.7326,
      "step": 4696
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9150906513148653,
      "learning_rate": 4.832923462814565e-07,
      "loss": 0.7798,
      "step": 4697
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8786270222570394,
      "learning_rate": 4.813805756694035e-07,
      "loss": 0.7553,
      "step": 4698
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0495766547298693,
      "learning_rate": 4.794725005131462e-07,
      "loss": 0.7704,
      "step": 4699
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9522233490941134,
      "learning_rate": 4.775681215534656e-07,
      "loss": 0.7625,
      "step": 4700
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9794876089255288,
      "learning_rate": 4.7566743952970894e-07,
      "loss": 0.7536,
      "step": 4701
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5795163337566122,
      "learning_rate": 4.7377045517978173e-07,
      "loss": 0.3043,
      "step": 4702
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9326872243846868,
      "learning_rate": 4.7187716924016355e-07,
      "loss": 0.7947,
      "step": 4703
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9257489121549488,
      "learning_rate": 4.6998758244588995e-07,
      "loss": 0.7465,
      "step": 4704
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.924065996045471,
      "learning_rate": 4.6810169553056616e-07,
      "loss": 0.7125,
      "step": 4705
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9336725417710035,
      "learning_rate": 4.662195092263566e-07,
      "loss": 0.7574,
      "step": 4706
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9390723285462199,
      "learning_rate": 4.643410242639912e-07,
      "loss": 0.7735,
      "step": 4707
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.839183470194505,
      "learning_rate": 4.6246624137276206e-07,
      "loss": 0.8256,
      "step": 4708
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9330030728943056,
      "learning_rate": 4.605951612805237e-07,
      "loss": 0.7861,
      "step": 4709
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0163770644438395,
      "learning_rate": 4.587277847136984e-07,
      "loss": 0.7581,
      "step": 4710
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9602679130950977,
      "learning_rate": 4.568641123972606e-07,
      "loss": 0.7898,
      "step": 4711
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9383194099141793,
      "learning_rate": 4.550041450547549e-07,
      "loss": 0.7636,
      "step": 4712
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.040148693449511,
      "learning_rate": 4.5314788340828365e-07,
      "loss": 0.7771,
      "step": 4713
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.88820064234978,
      "learning_rate": 4.512953281785104e-07,
      "loss": 0.7999,
      "step": 4714
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.908572804804285,
      "learning_rate": 4.494464800846654e-07,
      "loss": 0.771,
      "step": 4715
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.916880801426457,
      "learning_rate": 4.476013398445289e-07,
      "loss": 0.7151,
      "step": 4716
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8735030133958231,
      "learning_rate": 4.4575990817445234e-07,
      "loss": 0.8461,
      "step": 4717
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9591331191770144,
      "learning_rate": 4.4392218578934164e-07,
      "loss": 0.8056,
      "step": 4718
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8773762189683936,
      "learning_rate": 4.4208817340266385e-07,
      "loss": 0.8018,
      "step": 4719
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9085744255176551,
      "learning_rate": 4.4025787172644495e-07,
      "loss": 0.8272,
      "step": 4720
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9208751889931941,
      "learning_rate": 4.384312814712721e-07,
      "loss": 0.7424,
      "step": 4721
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9423461076773054,
      "learning_rate": 4.366084033462914e-07,
      "loss": 0.8028,
      "step": 4722
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9289535369559498,
      "learning_rate": 4.3478923805920335e-07,
      "loss": 0.766,
      "step": 4723
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8855491033958495,
      "learning_rate": 4.329737863162753e-07,
      "loss": 0.7238,
      "step": 4724
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8642321904743331,
      "learning_rate": 4.311620488223256e-07,
      "loss": 0.7374,
      "step": 4725
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9124477958912743,
      "learning_rate": 4.2935402628073166e-07,
      "loss": 0.8033,
      "step": 4726
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9374309157954804,
      "learning_rate": 4.27549719393433e-07,
      "loss": 0.7865,
      "step": 4727
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8988829280827303,
      "learning_rate": 4.2574912886092166e-07,
      "loss": 0.7743,
      "step": 4728
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9370042791662625,
      "learning_rate": 4.239522553822495e-07,
      "loss": 0.8106,
      "step": 4729
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8876627500623752,
      "learning_rate": 4.221590996550251e-07,
      "loss": 0.7501,
      "step": 4730
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9589000841367601,
      "learning_rate": 4.203696623754139e-07,
      "loss": 0.7649,
      "step": 4731
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9850996330312567,
      "learning_rate": 4.1858394423813563e-07,
      "loss": 0.7833,
      "step": 4732
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9310665474909108,
      "learning_rate": 4.1680194593646696e-07,
      "loss": 0.7441,
      "step": 4733
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8926975923633537,
      "learning_rate": 4.1502366816224327e-07,
      "loss": 0.799,
      "step": 4734
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9379064337066567,
      "learning_rate": 4.1324911160585014e-07,
      "loss": 0.8177,
      "step": 4735
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9729629230864646,
      "learning_rate": 4.1147827695623643e-07,
      "loss": 0.7989,
      "step": 4736
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9569221570643163,
      "learning_rate": 4.097111649008967e-07,
      "loss": 0.7989,
      "step": 4737
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9020421963541617,
      "learning_rate": 4.0794777612588543e-07,
      "loss": 0.7726,
      "step": 4738
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9113323089349866,
      "learning_rate": 4.061881113158117e-07,
      "loss": 0.7929,
      "step": 4739
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8591759024354791,
      "learning_rate": 4.044321711538368e-07,
      "loss": 0.7903,
      "step": 4740
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9515094843316331,
      "learning_rate": 4.02679956321681e-07,
      "loss": 0.7715,
      "step": 4741
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.95869094135673,
      "learning_rate": 4.00931467499609e-07,
      "loss": 0.824,
      "step": 4742
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9635045084011988,
      "learning_rate": 3.9918670536644776e-07,
      "loss": 0.8262,
      "step": 4743
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9423945077852308,
      "learning_rate": 3.974456705995733e-07,
      "loss": 0.7912,
      "step": 4744
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9390860316685089,
      "learning_rate": 3.9570836387491487e-07,
      "loss": 0.754,
      "step": 4745
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9698996092961482,
      "learning_rate": 3.9397478586695513e-07,
      "loss": 0.787,
      "step": 4746
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8966053833510496,
      "learning_rate": 3.9224493724872915e-07,
      "loss": 0.7977,
      "step": 4747
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9268852334462795,
      "learning_rate": 3.90518818691823e-07,
      "loss": 0.7651,
      "step": 4748
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9258323043716806,
      "learning_rate": 3.8879643086637384e-07,
      "loss": 0.7964,
      "step": 4749
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9507774101993193,
      "learning_rate": 3.8707777444107697e-07,
      "loss": 0.8015,
      "step": 4750
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9066328903899241,
      "learning_rate": 3.8536285008316854e-07,
      "loss": 0.8727,
      "step": 4751
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9418477494955587,
      "learning_rate": 3.8365165845844266e-07,
      "loss": 0.8201,
      "step": 4752
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8779360114040625,
      "learning_rate": 3.819442002312457e-07,
      "loss": 0.7231,
      "step": 4753
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9766002042167473,
      "learning_rate": 3.8024047606446736e-07,
      "loss": 0.7654,
      "step": 4754
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8828775326018253,
      "learning_rate": 3.785404866195552e-07,
      "loss": 0.7474,
      "step": 4755
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8784042368457501,
      "learning_rate": 3.768442325565036e-07,
      "loss": 0.8255,
      "step": 4756
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.880393880214939,
      "learning_rate": 3.751517145338546e-07,
      "loss": 0.7818,
      "step": 4757
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8716849542803146,
      "learning_rate": 3.7346293320870363e-07,
      "loss": 0.7889,
      "step": 4758
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8917174294038599,
      "learning_rate": 3.717778892366941e-07,
      "loss": 0.7831,
      "step": 4759
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5968765534557354,
      "learning_rate": 3.700965832720171e-07,
      "loss": 0.3165,
      "step": 4760
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6138898809070941,
      "learning_rate": 3.684190159674117e-07,
      "loss": 0.3097,
      "step": 4761
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.905050057254636,
      "learning_rate": 3.6674518797417236e-07,
      "loss": 0.7617,
      "step": 4762
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9512997799262071,
      "learning_rate": 3.6507509994213155e-07,
      "loss": 0.7755,
      "step": 4763
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9162488778420596,
      "learning_rate": 3.6340875251967946e-07,
      "loss": 0.8098,
      "step": 4764
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.92582661846012,
      "learning_rate": 3.617461463537464e-07,
      "loss": 0.7799,
      "step": 4765
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8471216771783022,
      "learning_rate": 3.6008728208981157e-07,
      "loss": 0.8068,
      "step": 4766
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9390415796573396,
      "learning_rate": 3.5843216037190873e-07,
      "loss": 0.8174,
      "step": 4767
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9236439532404452,
      "learning_rate": 3.5678078184260834e-07,
      "loss": 0.7652,
      "step": 4768
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5640186950850259,
      "learning_rate": 3.5513314714303524e-07,
      "loss": 0.3258,
      "step": 4769
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9023261723932474,
      "learning_rate": 3.5348925691285675e-07,
      "loss": 0.795,
      "step": 4770
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9347496878275234,
      "learning_rate": 3.518491117902878e-07,
      "loss": 0.8022,
      "step": 4771
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9641472665353439,
      "learning_rate": 3.502127124120891e-07,
      "loss": 0.7847,
      "step": 4772
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9147094753340166,
      "learning_rate": 3.48580059413568e-07,
      "loss": 0.8448,
      "step": 4773
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8464030880421806,
      "learning_rate": 3.4695115342857524e-07,
      "loss": 0.7849,
      "step": 4774
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8507542553122873,
      "learning_rate": 3.4532599508950826e-07,
      "loss": 0.7992,
      "step": 4775
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9128592944247272,
      "learning_rate": 3.437045850273113e-07,
      "loss": 0.7542,
      "step": 4776
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9107870395669527,
      "learning_rate": 3.420869238714708e-07,
      "loss": 0.8074,
      "step": 4777
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9846497539695789,
      "learning_rate": 3.404730122500155e-07,
      "loss": 0.7431,
      "step": 4778
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.866671074961014,
      "learning_rate": 3.3886285078952753e-07,
      "loss": 0.753,
      "step": 4779
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.633917441943657,
      "learning_rate": 3.3725644011512125e-07,
      "loss": 0.33,
      "step": 4780
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9482221293120382,
      "learning_rate": 3.356537808504634e-07,
      "loss": 0.7461,
      "step": 4781
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9230140516494525,
      "learning_rate": 3.3405487361776177e-07,
      "loss": 0.7988,
      "step": 4782
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9129295544144559,
      "learning_rate": 3.3245971903776654e-07,
      "loss": 0.747,
      "step": 4783
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9740903562701312,
      "learning_rate": 3.308683177297711e-07,
      "loss": 0.816,
      "step": 4784
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8917940938319727,
      "learning_rate": 3.292806703116125e-07,
      "loss": 0.7856,
      "step": 4785
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9049410611059923,
      "learning_rate": 3.2769677739966975e-07,
      "loss": 0.751,
      "step": 4786
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9747966721840919,
      "learning_rate": 3.2611663960886665e-07,
      "loss": 0.7633,
      "step": 4787
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9199906202514679,
      "learning_rate": 3.245402575526646e-07,
      "loss": 0.7846,
      "step": 4788
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8932292624453594,
      "learning_rate": 3.2296763184306965e-07,
      "loss": 0.8032,
      "step": 4789
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.902416337151841,
      "learning_rate": 3.2139876309063233e-07,
      "loss": 0.7262,
      "step": 4790
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5885397056222346,
      "learning_rate": 3.198336519044376e-07,
      "loss": 0.3335,
      "step": 4791
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8588659065483569,
      "learning_rate": 3.182722988921161e-07,
      "loss": 0.7289,
      "step": 4792
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9493010898775455,
      "learning_rate": 3.167147046598418e-07,
      "loss": 0.7697,
      "step": 4793
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9670986110473208,
      "learning_rate": 3.151608698123232e-07,
      "loss": 0.8209,
      "step": 4794
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9143699230657668,
      "learning_rate": 3.1361079495281443e-07,
      "loss": 0.7612,
      "step": 4795
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8922739445778446,
      "learning_rate": 3.1206448068310635e-07,
      "loss": 0.7373,
      "step": 4796
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9243722786417597,
      "learning_rate": 3.1052192760353316e-07,
      "loss": 0.8307,
      "step": 4797
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8778355243497509,
      "learning_rate": 3.0898313631296586e-07,
      "loss": 0.8046,
      "step": 4798
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8905527772868063,
      "learning_rate": 3.0744810740881646e-07,
      "loss": 0.7481,
      "step": 4799
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9013724383762289,
      "learning_rate": 3.0591684148703617e-07,
      "loss": 0.8025,
      "step": 4800
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9183983585055764,
      "learning_rate": 3.043893391421149e-07,
      "loss": 0.7275,
      "step": 4801
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9146655085080667,
      "learning_rate": 3.0286560096708275e-07,
      "loss": 0.8076,
      "step": 4802
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9429802042493395,
      "learning_rate": 3.013456275535054e-07,
      "loss": 0.7714,
      "step": 4803
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9237005079336532,
      "learning_rate": 2.998294194914897e-07,
      "loss": 0.7197,
      "step": 4804
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9306875733270873,
      "learning_rate": 2.983169773696815e-07,
      "loss": 0.7665,
      "step": 4805
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9311246347730396,
      "learning_rate": 2.968083017752599e-07,
      "loss": 0.8732,
      "step": 4806
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8520511578009792,
      "learning_rate": 2.953033932939464e-07,
      "loss": 0.7297,
      "step": 4807
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8307151564518581,
      "learning_rate": 2.938022525099982e-07,
      "loss": 0.8077,
      "step": 4808
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0059633473250882,
      "learning_rate": 2.9230488000621003e-07,
      "loss": 0.8035,
      "step": 4809
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0074113637319255,
      "learning_rate": 2.908112763639137e-07,
      "loss": 0.7736,
      "step": 4810
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6177684595668484,
      "learning_rate": 2.8932144216297643e-07,
      "loss": 0.3022,
      "step": 4811
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0003104615052922,
      "learning_rate": 2.878353779818044e-07,
      "loss": 0.7704,
      "step": 4812
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9931222145862691,
      "learning_rate": 2.863530843973372e-07,
      "loss": 0.8089,
      "step": 4813
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9230616567528289,
      "learning_rate": 2.848745619850546e-07,
      "loss": 0.7536,
      "step": 4814
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.927261208285002,
      "learning_rate": 2.833998113189662e-07,
      "loss": 0.8157,
      "step": 4815
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8951245216166556,
      "learning_rate": 2.8192883297162634e-07,
      "loss": 0.7981,
      "step": 4816
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9716211591278833,
      "learning_rate": 2.804616275141148e-07,
      "loss": 0.7955,
      "step": 4817
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.907413092388851,
      "learning_rate": 2.7899819551605256e-07,
      "loss": 0.7897,
      "step": 4818
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9609714922184491,
      "learning_rate": 2.7753853754559634e-07,
      "loss": 0.7945,
      "step": 4819
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9825087475950315,
      "learning_rate": 2.760826541694328e-07,
      "loss": 0.8164,
      "step": 4820
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9408750083788276,
      "learning_rate": 2.746305459527876e-07,
      "loss": 0.7474,
      "step": 4821
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6173600135117122,
      "learning_rate": 2.7318221345941865e-07,
      "loss": 0.3503,
      "step": 4822
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8965645240814942,
      "learning_rate": 2.717376572516184e-07,
      "loss": 0.7598,
      "step": 4823
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9369597639258023,
      "learning_rate": 2.7029687789021377e-07,
      "loss": 0.6989,
      "step": 4824
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9309683511942738,
      "learning_rate": 2.688598759345651e-07,
      "loss": 0.797,
      "step": 4825
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8681056534811176,
      "learning_rate": 2.67426651942565e-07,
      "loss": 0.771,
      "step": 4826
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9056112692726552,
      "learning_rate": 2.659972064706406e-07,
      "loss": 0.8033,
      "step": 4827
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8905687778435131,
      "learning_rate": 2.645715400737536e-07,
      "loss": 0.7677,
      "step": 4828
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9220649249061457,
      "learning_rate": 2.631496533053934e-07,
      "loss": 0.8084,
      "step": 4829
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.585625222159573,
      "learning_rate": 2.6173154671758847e-07,
      "loss": 0.3199,
      "step": 4830
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6416988277303491,
      "learning_rate": 2.603172208608962e-07,
      "loss": 0.3246,
      "step": 4831
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.046167704971511,
      "learning_rate": 2.589066762844039e-07,
      "loss": 0.8104,
      "step": 4832
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0108341939947183,
      "learning_rate": 2.57499913535737e-07,
      "loss": 0.7722,
      "step": 4833
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.869223484565972,
      "learning_rate": 2.5609693316104745e-07,
      "loss": 0.7602,
      "step": 4834
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8934253180766373,
      "learning_rate": 2.5469773570502063e-07,
      "loss": 0.79,
      "step": 4835
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6059543955758662,
      "learning_rate": 2.5330232171087433e-07,
      "loss": 0.3706,
      "step": 4836
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9099779703957844,
      "learning_rate": 2.51910691720354e-07,
      "loss": 0.7869,
      "step": 4837
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.079873052619521,
      "learning_rate": 2.5052284627374077e-07,
      "loss": 0.7765,
      "step": 4838
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9379431433443478,
      "learning_rate": 2.491387859098426e-07,
      "loss": 0.8263,
      "step": 4839
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.85839269292209,
      "learning_rate": 2.477585111659997e-07,
      "loss": 0.7401,
      "step": 4840
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9461586556518309,
      "learning_rate": 2.463820225780811e-07,
      "loss": 0.7865,
      "step": 4841
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0276298020458245,
      "learning_rate": 2.4500932068049046e-07,
      "loss": 0.7668,
      "step": 4842
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9771722470721841,
      "learning_rate": 2.4364040600615477e-07,
      "loss": 0.7449,
      "step": 4843
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9070918570377606,
      "learning_rate": 2.422752790865346e-07,
      "loss": 0.7369,
      "step": 4844
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8760209334128096,
      "learning_rate": 2.409139404516203e-07,
      "loss": 0.7708,
      "step": 4845
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8605533648959167,
      "learning_rate": 2.3955639062992696e-07,
      "loss": 0.7618,
      "step": 4846
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.996968449789136,
      "learning_rate": 2.3820263014850741e-07,
      "loss": 0.8001,
      "step": 4847
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9229803141844137,
      "learning_rate": 2.3685265953293345e-07,
      "loss": 0.7944,
      "step": 4848
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8701483017724038,
      "learning_rate": 2.3550647930731362e-07,
      "loss": 0.7453,
      "step": 4849
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9702373571860322,
      "learning_rate": 2.3416408999427876e-07,
      "loss": 0.7768,
      "step": 4850
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.943392190550045,
      "learning_rate": 2.3282549211499307e-07,
      "loss": 0.8242,
      "step": 4851
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9332773739199632,
      "learning_rate": 2.3149068618914417e-07,
      "loss": 0.7846,
      "step": 4852
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8481653029226726,
      "learning_rate": 2.3015967273494867e-07,
      "loss": 0.7837,
      "step": 4853
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6501781904974716,
      "learning_rate": 2.2883245226915652e-07,
      "loss": 0.3404,
      "step": 4854
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8484566642081376,
      "learning_rate": 2.2750902530703667e-07,
      "loss": 0.7791,
      "step": 4855
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.941536379091822,
      "learning_rate": 2.2618939236238924e-07,
      "loss": 0.7367,
      "step": 4856
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8664515762064858,
      "learning_rate": 2.2487355394754328e-07,
      "loss": 0.8065,
      "step": 4857
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.915836288459647,
      "learning_rate": 2.2356151057334908e-07,
      "loss": 0.7717,
      "step": 4858
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6410123937445171,
      "learning_rate": 2.2225326274919135e-07,
      "loss": 0.3247,
      "step": 4859
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8545692744224567,
      "learning_rate": 2.209488109829727e-07,
      "loss": 0.7748,
      "step": 4860
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0171383945947614,
      "learning_rate": 2.196481557811303e-07,
      "loss": 0.7274,
      "step": 4861
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9509516026215543,
      "learning_rate": 2.1835129764861907e-07,
      "loss": 0.8146,
      "step": 4862
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.85956958686427,
      "learning_rate": 2.1705823708892737e-07,
      "loss": 0.7904,
      "step": 4863
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.938141130105045,
      "learning_rate": 2.1576897460406477e-07,
      "loss": 0.7454,
      "step": 4864
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0140775373459614,
      "learning_rate": 2.144835106945664e-07,
      "loss": 0.7375,
      "step": 4865
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9836268166902826,
      "learning_rate": 2.1320184585949532e-07,
      "loss": 0.8299,
      "step": 4866
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9030565233663558,
      "learning_rate": 2.119239805964357e-07,
      "loss": 0.7951,
      "step": 4867
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6073456525509062,
      "learning_rate": 2.106499154015018e-07,
      "loss": 0.3339,
      "step": 4868
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9245843725607273,
      "learning_rate": 2.0937965076932576e-07,
      "loss": 0.7845,
      "step": 4869
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0000816493256852,
      "learning_rate": 2.0811318719307194e-07,
      "loss": 0.7916,
      "step": 4870
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8572951371974193,
      "learning_rate": 2.0685052516442373e-07,
      "loss": 0.7413,
      "step": 4871
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8847860381053654,
      "learning_rate": 2.0559166517358787e-07,
      "loss": 0.7564,
      "step": 4872
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9200652600724947,
      "learning_rate": 2.0433660770930009e-07,
      "loss": 0.776,
      "step": 4873
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9688526597156719,
      "learning_rate": 2.0308535325881616e-07,
      "loss": 0.7884,
      "step": 4874
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0066258485786095,
      "learning_rate": 2.0183790230791532e-07,
      "loss": 0.7828,
      "step": 4875
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8971626391252006,
      "learning_rate": 2.0059425534090128e-07,
      "loss": 0.7478,
      "step": 4876
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9508648519818205,
      "learning_rate": 1.9935441284059998e-07,
      "loss": 0.8245,
      "step": 4877
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9875477117713682,
      "learning_rate": 1.981183752883631e-07,
      "loss": 0.8031,
      "step": 4878
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0064654244353521,
      "learning_rate": 1.9688614316406006e-07,
      "loss": 0.7745,
      "step": 4879
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9993263078715522,
      "learning_rate": 1.9565771694608937e-07,
      "loss": 0.7732,
      "step": 4880
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9003748223899826,
      "learning_rate": 1.9443309711136393e-07,
      "loss": 0.7733,
      "step": 4881
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.972808263597637,
      "learning_rate": 1.9321228413532788e-07,
      "loss": 0.7345,
      "step": 4882
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9357978305274911,
      "learning_rate": 1.9199527849194098e-07,
      "loss": 0.7517,
      "step": 4883
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6221434640579327,
      "learning_rate": 1.907820806536842e-07,
      "loss": 0.3356,
      "step": 4884
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9070941700898004,
      "learning_rate": 1.895726910915663e-07,
      "loss": 0.75,
      "step": 4885
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.940138044596096,
      "learning_rate": 1.883671102751128e-07,
      "loss": 0.7879,
      "step": 4886
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5949342909836688,
      "learning_rate": 1.8716533867237153e-07,
      "loss": 0.333,
      "step": 4887
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8521060298561715,
      "learning_rate": 1.859673767499115e-07,
      "loss": 0.8271,
      "step": 4888
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6034622423395362,
      "learning_rate": 1.847732249728218e-07,
      "loss": 0.3084,
      "step": 4889
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9762541935146989,
      "learning_rate": 1.83582883804716e-07,
      "loss": 0.8051,
      "step": 4890
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8601199984815302,
      "learning_rate": 1.8239635370772223e-07,
      "loss": 0.7391,
      "step": 4891
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6146257237597436,
      "learning_rate": 1.8121363514249534e-07,
      "loss": 0.3282,
      "step": 4892
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9816434815549672,
      "learning_rate": 1.8003472856820469e-07,
      "loss": 0.7535,
      "step": 4893
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9376807992166003,
      "learning_rate": 1.7885963444254528e-07,
      "loss": 0.7738,
      "step": 4894
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8463719265401027,
      "learning_rate": 1.7768835322172552e-07,
      "loss": 0.7739,
      "step": 4895
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9523627152561239,
      "learning_rate": 1.7652088536048052e-07,
      "loss": 0.8236,
      "step": 4896
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9554349081983543,
      "learning_rate": 1.7535723131206106e-07,
      "loss": 0.787,
      "step": 4897
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0405868363792723,
      "learning_rate": 1.7419739152823468e-07,
      "loss": 0.8352,
      "step": 4898
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8909066412683441,
      "learning_rate": 1.7304136645929448e-07,
      "loss": 0.8098,
      "step": 4899
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9312739070361425,
      "learning_rate": 1.7188915655404814e-07,
      "loss": 0.7609,
      "step": 4900
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9695804145836611,
      "learning_rate": 1.707407622598223e-07,
      "loss": 0.744,
      "step": 4901
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9588405968474955,
      "learning_rate": 1.695961840224636e-07,
      "loss": 0.7863,
      "step": 4902
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9626112096303906,
      "learning_rate": 1.6845542228633772e-07,
      "loss": 0.7578,
      "step": 4903
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9787938202927955,
      "learning_rate": 1.6731847749432705e-07,
      "loss": 0.7717,
      "step": 4904
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9079479228138541,
      "learning_rate": 1.6618535008783075e-07,
      "loss": 0.7873,
      "step": 4905
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8691490016222846,
      "learning_rate": 1.6505604050677249e-07,
      "loss": 0.8249,
      "step": 4906
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.940803196424483,
      "learning_rate": 1.6393054918958373e-07,
      "loss": 0.7818,
      "step": 4907
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.836995706865723,
      "learning_rate": 1.6280887657322276e-07,
      "loss": 0.7106,
      "step": 4908
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9618099519379624,
      "learning_rate": 1.616910230931612e-07,
      "loss": 0.7833,
      "step": 4909
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9253167751993969,
      "learning_rate": 1.6057698918338526e-07,
      "loss": 0.8094,
      "step": 4910
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9468141487644405,
      "learning_rate": 1.5946677527640563e-07,
      "loss": 0.7716,
      "step": 4911
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9526284681605207,
      "learning_rate": 1.5836038180324198e-07,
      "loss": 0.7596,
      "step": 4912
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6025920594215467,
      "learning_rate": 1.5725780919343624e-07,
      "loss": 0.3525,
      "step": 4913
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.91104786604393,
      "learning_rate": 1.561590578750438e-07,
      "loss": 0.8573,
      "step": 4914
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5816536855240042,
      "learning_rate": 1.55064128274639e-07,
      "loss": 0.3503,
      "step": 4915
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9204844709138739,
      "learning_rate": 1.5397302081731069e-07,
      "loss": 0.7892,
      "step": 4916
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8865682481032985,
      "learning_rate": 1.5288573592666445e-07,
      "loss": 0.7969,
      "step": 4917
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9074980275225286,
      "learning_rate": 1.518022740248215e-07,
      "loss": 0.7639,
      "step": 4918
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9306201335280249,
      "learning_rate": 1.5072263553241872e-07,
      "loss": 0.74,
      "step": 4919
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0434778247799985,
      "learning_rate": 1.4964682086861082e-07,
      "loss": 0.7919,
      "step": 4920
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0169346309120055,
      "learning_rate": 1.4857483045106258e-07,
      "loss": 0.8382,
      "step": 4921
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9544594917177128,
      "learning_rate": 1.475066646959611e-07,
      "loss": 0.8575,
      "step": 4922
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9206446799132814,
      "learning_rate": 1.4644232401800352e-07,
      "loss": 0.7571,
      "step": 4923
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8742061301514521,
      "learning_rate": 1.4538180883040264e-07,
      "loss": 0.7824,
      "step": 4924
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.878144804475178,
      "learning_rate": 1.4432511954488915e-07,
      "loss": 0.7726,
      "step": 4925
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9878660470571944,
      "learning_rate": 1.4327225657170485e-07,
      "loss": 0.8412,
      "step": 4926
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9877352655257098,
      "learning_rate": 1.4222322031960723e-07,
      "loss": 0.7666,
      "step": 4927
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9432755390078552,
      "learning_rate": 1.411780111958694e-07,
      "loss": 0.738,
      "step": 4928
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0078222155102685,
      "learning_rate": 1.4013662960627562e-07,
      "loss": 0.8069,
      "step": 4929
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0058907575711504,
      "learning_rate": 1.3909907595512806e-07,
      "loss": 0.7617,
      "step": 4930
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.910584591230057,
      "learning_rate": 1.3806535064524006e-07,
      "loss": 0.7686,
      "step": 4931
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9943483331001703,
      "learning_rate": 1.3703545407793951e-07,
      "loss": 0.8279,
      "step": 4932
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9905776442517158,
      "learning_rate": 1.360093866530665e-07,
      "loss": 0.8557,
      "step": 4933
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8370824457223636,
      "learning_rate": 1.34987148768978e-07,
      "loss": 0.7766,
      "step": 4934
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9249581116055176,
      "learning_rate": 1.3396874082253986e-07,
      "loss": 0.7989,
      "step": 4935
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9067760814260938,
      "learning_rate": 1.3295416320913357e-07,
      "loss": 0.789,
      "step": 4936
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.950818517090922,
      "learning_rate": 1.3194341632265518e-07,
      "loss": 0.7985,
      "step": 4937
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8982984888817513,
      "learning_rate": 1.3093650055550855e-07,
      "loss": 0.8304,
      "step": 4938
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8826000136717677,
      "learning_rate": 1.2993341629861432e-07,
      "loss": 0.7875,
      "step": 4939
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9539923570025424,
      "learning_rate": 1.2893416394140323e-07,
      "loss": 0.7576,
      "step": 4940
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8970337173618682,
      "learning_rate": 1.279387438718216e-07,
      "loss": 0.8008,
      "step": 4941
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9034733515053621,
      "learning_rate": 1.269471564763247e-07,
      "loss": 0.7639,
      "step": 4942
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9152760809384445,
      "learning_rate": 1.2595940213988024e-07,
      "loss": 0.7631,
      "step": 4943
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.895887435072524,
      "learning_rate": 1.2497548124597026e-07,
      "loss": 0.7885,
      "step": 4944
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8484726595300383,
      "learning_rate": 1.2399539417658368e-07,
      "loss": 0.7683,
      "step": 4945
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9659603113134985,
      "learning_rate": 1.2301914131222726e-07,
      "loss": 0.7651,
      "step": 4946
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8753677056155772,
      "learning_rate": 1.2204672303191335e-07,
      "loss": 0.7214,
      "step": 4947
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9660043948411958,
      "learning_rate": 1.2107813971317106e-07,
      "loss": 0.7237,
      "step": 4948
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8688384245971285,
      "learning_rate": 1.201133917320363e-07,
      "loss": 0.7703,
      "step": 4949
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8796317974788602,
      "learning_rate": 1.1915247946305498e-07,
      "loss": 0.8149,
      "step": 4950
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.920379493212735,
      "learning_rate": 1.1819540327929092e-07,
      "loss": 0.7646,
      "step": 4951
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9592338326661909,
      "learning_rate": 1.1724216355231022e-07,
      "loss": 0.7853,
      "step": 4952
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9458451848987313,
      "learning_rate": 1.1629276065219575e-07,
      "loss": 0.7611,
      "step": 4953
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9253410115514694,
      "learning_rate": 1.1534719494753821e-07,
      "loss": 0.8315,
      "step": 4954
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8608643961183893,
      "learning_rate": 1.144054668054373e-07,
      "loss": 0.7653,
      "step": 4955
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9076584846704987,
      "learning_rate": 1.1346757659150498e-07,
      "loss": 0.7968,
      "step": 4956
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9088138010294727,
      "learning_rate": 1.1253352466986334e-07,
      "loss": 0.7826,
      "step": 4957
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.88175826889745,
      "learning_rate": 1.116033114031434e-07,
      "loss": 0.8014,
      "step": 4958
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8515561318487509,
      "learning_rate": 1.1067693715248406e-07,
      "loss": 0.7604,
      "step": 4959
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8979561816086096,
      "learning_rate": 1.0975440227753764e-07,
      "loss": 0.7877,
      "step": 4960
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9290220796631459,
      "learning_rate": 1.0883570713646318e-07,
      "loss": 0.79,
      "step": 4961
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9369304275335939,
      "learning_rate": 1.0792085208593095e-07,
      "loss": 0.7731,
      "step": 4962
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9591889614766889,
      "learning_rate": 1.0700983748111792e-07,
      "loss": 0.7525,
      "step": 4963
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9712030901285176,
      "learning_rate": 1.061026636757101e-07,
      "loss": 0.796,
      "step": 4964
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8640178663880093,
      "learning_rate": 1.0519933102190682e-07,
      "loss": 0.7777,
      "step": 4965
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9071401510261392,
      "learning_rate": 1.0429983987041092e-07,
      "loss": 0.7997,
      "step": 4966
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9542736400507469,
      "learning_rate": 1.0340419057043527e-07,
      "loss": 0.784,
      "step": 4967
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9712106679787051,
      "learning_rate": 1.0251238346970393e-07,
      "loss": 0.8072,
      "step": 4968
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8555276353211276,
      "learning_rate": 1.0162441891444441e-07,
      "loss": 0.7934,
      "step": 4969
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9102184442985857,
      "learning_rate": 1.007402972493976e-07,
      "loss": 0.7562,
      "step": 4970
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9184492326098113,
      "learning_rate": 9.986001881780783e-08,
      "loss": 0.7606,
      "step": 4971
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6300204431629686,
      "learning_rate": 9.898358396143171e-08,
      "loss": 0.3105,
      "step": 4972
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.848733399288125,
      "learning_rate": 9.811099302052928e-08,
      "loss": 0.7522,
      "step": 4973
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.967510588385886,
      "learning_rate": 9.72422463338718e-08,
      "loss": 0.7549,
      "step": 4974
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8451371566301178,
      "learning_rate": 9.637734423873612e-08,
      "loss": 0.737,
      "step": 4975
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.035977350575363,
      "learning_rate": 9.55162870709081e-08,
      "loss": 0.7555,
      "step": 4976
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9908100099533663,
      "learning_rate": 9.465907516467698e-08,
      "loss": 0.8071,
      "step": 4977
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6071121098295148,
      "learning_rate": 9.380570885284546e-08,
      "loss": 0.3314,
      "step": 4978
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8613489470430357,
      "learning_rate": 9.295618846671739e-08,
      "loss": 0.8278,
      "step": 4979
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0059663037527398,
      "learning_rate": 9.211051433610674e-08,
      "loss": 0.8131,
      "step": 4980
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9013026662301753,
      "learning_rate": 9.126868678933198e-08,
      "loss": 0.7707,
      "step": 4981
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8454578969983151,
      "learning_rate": 9.04307061532217e-08,
      "loss": 0.7468,
      "step": 4982
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9166940194424742,
      "learning_rate": 8.959657275310674e-08,
      "loss": 0.7829,
      "step": 4983
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9731457219066313,
      "learning_rate": 8.876628691282918e-08,
      "loss": 0.8109,
      "step": 4984
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8703989111769811,
      "learning_rate": 8.793984895473117e-08,
      "loss": 0.7454,
      "step": 4985
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9730450385052238,
      "learning_rate": 8.711725919966718e-08,
      "loss": 0.7764,
      "step": 4986
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8681495628283215,
      "learning_rate": 8.629851796699284e-08,
      "loss": 0.6986,
      "step": 4987
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9098710843339488,
      "learning_rate": 8.54836255745728e-08,
      "loss": 0.7511,
      "step": 4988
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9894503479514625,
      "learning_rate": 8.467258233877728e-08,
      "loss": 0.7443,
      "step": 4989
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0084184831913225,
      "learning_rate": 8.386538857447779e-08,
      "loss": 0.8182,
      "step": 4990
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9237989684175514,
      "learning_rate": 8.306204459505807e-08,
      "loss": 0.7473,
      "step": 4991
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8743707503572291,
      "learning_rate": 8.226255071240308e-08,
      "loss": 0.7673,
      "step": 4992
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9199349545198949,
      "learning_rate": 8.146690723690342e-08,
      "loss": 0.7936,
      "step": 4993
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8504555474437374,
      "learning_rate": 8.067511447745535e-08,
      "loss": 0.8183,
      "step": 4994
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8580652181209343,
      "learning_rate": 7.988717274146074e-08,
      "loss": 0.7323,
      "step": 4995
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9628209759648824,
      "learning_rate": 7.910308233482488e-08,
      "loss": 0.8041,
      "step": 4996
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9683362267990885,
      "learning_rate": 7.832284356195764e-08,
      "loss": 0.7855,
      "step": 4997
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.910990242495492,
      "learning_rate": 7.754645672577776e-08,
      "loss": 0.7517,
      "step": 4998
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9102603789857442,
      "learning_rate": 7.677392212770196e-08,
      "loss": 0.7143,
      "step": 4999
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9510287358532217,
      "learning_rate": 7.600524006765808e-08,
      "loss": 0.7586,
      "step": 5000
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8626224827570422,
      "learning_rate": 7.524041084407185e-08,
      "loss": 0.7122,
      "step": 5001
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9660416793934643,
      "learning_rate": 7.447943475387797e-08,
      "loss": 0.7949,
      "step": 5002
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9271790277815928,
      "learning_rate": 7.372231209251346e-08,
      "loss": 0.7815,
      "step": 5003
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0362818321591012,
      "learning_rate": 7.296904315391873e-08,
      "loss": 0.8143,
      "step": 5004
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9500922909762664,
      "learning_rate": 7.221962823053874e-08,
      "loss": 0.7872,
      "step": 5005
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8907242929276079,
      "learning_rate": 7.147406761332298e-08,
      "loss": 0.7418,
      "step": 5006
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9107328515266581,
      "learning_rate": 7.073236159172325e-08,
      "loss": 0.7646,
      "step": 5007
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9026832865500347,
      "learning_rate": 6.999451045369587e-08,
      "loss": 0.7678,
      "step": 5008
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.966316060595893,
      "learning_rate": 6.926051448569948e-08,
      "loss": 0.8191,
      "step": 5009
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9468387161625105,
      "learning_rate": 6.853037397269724e-08,
      "loss": 0.7301,
      "step": 5010
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9219426001750979,
      "learning_rate": 6.78040891981524e-08,
      "loss": 0.8011,
      "step": 5011
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9504660048549808,
      "learning_rate": 6.70816604440383e-08,
      "loss": 0.8408,
      "step": 5012
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8959150310143531,
      "learning_rate": 6.63630879908217e-08,
      "loss": 0.8083,
      "step": 5013
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.901740898053628,
      "learning_rate": 6.564837211748054e-08,
      "loss": 0.7547,
      "step": 5014
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9562563035794599,
      "learning_rate": 6.493751310149177e-08,
      "loss": 0.8067,
      "step": 5015
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9213010062532642,
      "learning_rate": 6.42305112188335e-08,
      "loss": 0.8166,
      "step": 5016
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.950057882307086,
      "learning_rate": 6.352736674398951e-08,
      "loss": 0.8031,
      "step": 5017
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9023567736499117,
      "learning_rate": 6.282807994994477e-08,
      "loss": 0.7667,
      "step": 5018
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8932694606553941,
      "learning_rate": 6.213265110818656e-08,
      "loss": 0.7933,
      "step": 5019
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8694803287964843,
      "learning_rate": 6.144108048870335e-08,
      "loss": 0.7778,
      "step": 5020
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9664234650413493,
      "learning_rate": 6.075336835998813e-08,
      "loss": 0.8017,
      "step": 5021
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6662873805464257,
      "learning_rate": 6.00695149890329e-08,
      "loss": 0.3378,
      "step": 5022
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0022019684712586,
      "learning_rate": 5.938952064133419e-08,
      "loss": 0.7764,
      "step": 5023
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9515286663536127,
      "learning_rate": 5.871338558088857e-08,
      "loss": 0.8084,
      "step": 5024
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0044316427285434,
      "learning_rate": 5.8041110070194976e-08,
      "loss": 0.7854,
      "step": 5025
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8686934585233136,
      "learning_rate": 5.7372694370254614e-08,
      "loss": 0.8198,
      "step": 5026
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9904890485339377,
      "learning_rate": 5.67081387405688e-08,
      "loss": 0.7147,
      "step": 5027
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9807520472311677,
      "learning_rate": 5.6047443439141146e-08,
      "loss": 0.7798,
      "step": 5028
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9728433526477133,
      "learning_rate": 5.539060872247537e-08,
      "loss": 0.782,
      "step": 5029
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.848464601810097,
      "learning_rate": 5.47376348455797e-08,
      "loss": 0.7826,
      "step": 5030
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5958101088574197,
      "learning_rate": 5.408852206195914e-08,
      "loss": 0.3294,
      "step": 5031
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9649586653760838,
      "learning_rate": 5.344327062362098e-08,
      "loss": 0.7411,
      "step": 5032
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5781320254520818,
      "learning_rate": 5.2801880781075954e-08,
      "loss": 0.2987,
      "step": 5033
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9193909871221462,
      "learning_rate": 5.216435278333376e-08,
      "loss": 0.7557,
      "step": 5034
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8854723642136058,
      "learning_rate": 5.153068687790197e-08,
      "loss": 0.7937,
      "step": 5035
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9909883146964272,
      "learning_rate": 5.0900883310794903e-08,
      "loss": 0.796,
      "step": 5036
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9584645232337159,
      "learning_rate": 5.0274942326521414e-08,
      "loss": 0.76,
      "step": 5037
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9602625110975455,
      "learning_rate": 4.9652864168096e-08,
      "loss": 0.7376,
      "step": 5038
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9996514064360443,
      "learning_rate": 4.9034649077027706e-08,
      "loss": 0.7496,
      "step": 5039
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9194018935351912,
      "learning_rate": 4.84202972933312e-08,
      "loss": 0.7912,
      "step": 5040
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9942883656222246,
      "learning_rate": 4.7809809055517906e-08,
      "loss": 0.8289,
      "step": 5041
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9610934191586014,
      "learning_rate": 4.720318460060047e-08,
      "loss": 0.8153,
      "step": 5042
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9022025266432937,
      "learning_rate": 4.6600424164091606e-08,
      "loss": 0.7652,
      "step": 5043
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9286608367837792,
      "learning_rate": 4.6001527980004125e-08,
      "loss": 0.7318,
      "step": 5044
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8881884162409105,
      "learning_rate": 4.54064962808487e-08,
      "loss": 0.7703,
      "step": 5045
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8704547518952894,
      "learning_rate": 4.4815329297639434e-08,
      "loss": 0.7874,
      "step": 5046
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9281845592896073,
      "learning_rate": 4.422802725988606e-08,
      "loss": 0.7505,
      "step": 5047
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0124243672237427,
      "learning_rate": 4.364459039559843e-08,
      "loss": 0.7704,
      "step": 5048
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9003374670130316,
      "learning_rate": 4.3065018931289784e-08,
      "loss": 0.8089,
      "step": 5049
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9200264701433798,
      "learning_rate": 4.248931309196791e-08,
      "loss": 0.7798,
      "step": 5050
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9472196958654725,
      "learning_rate": 4.1917473101140696e-08,
      "loss": 0.7371,
      "step": 5051
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8538550939182987,
      "learning_rate": 4.134949918081832e-08,
      "loss": 0.7496,
      "step": 5052
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8884811956595531,
      "learning_rate": 4.0785391551506626e-08,
      "loss": 0.7554,
      "step": 5053
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9632018062335925,
      "learning_rate": 4.022515043221154e-08,
      "loss": 0.7559,
      "step": 5054
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9488758573583196,
      "learning_rate": 3.966877604043795e-08,
      "loss": 0.7628,
      "step": 5055
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.006638519586886,
      "learning_rate": 3.9116268592189755e-08,
      "loss": 0.7794,
      "step": 5056
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8892024513022649,
      "learning_rate": 3.8567628301969806e-08,
      "loss": 0.8114,
      "step": 5057
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8751887303932526,
      "learning_rate": 3.802285538277772e-08,
      "loss": 0.8029,
      "step": 5058
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8912206462299512,
      "learning_rate": 3.748195004611543e-08,
      "loss": 0.7354,
      "step": 5059
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9442236787006377,
      "learning_rate": 3.69449125019794e-08,
      "loss": 0.7923,
      "step": 5060
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8885644346045019,
      "learning_rate": 3.6411742958866184e-08,
      "loss": 0.7383,
      "step": 5061
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0220226603903957,
      "learning_rate": 3.588244162377019e-08,
      "loss": 0.8047,
      "step": 5062
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6044484649970755,
      "learning_rate": 3.5357008702185945e-08,
      "loss": 0.3213,
      "step": 5063
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6035195505992906,
      "learning_rate": 3.483544439810249e-08,
      "loss": 0.3936,
      "step": 5064
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8503846352952841,
      "learning_rate": 3.4317748914011187e-08,
      "loss": 0.7386,
      "step": 5065
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8787880828681719,
      "learning_rate": 3.3803922450897917e-08,
      "loss": 0.7361,
      "step": 5066
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.907810137180579,
      "learning_rate": 3.329396520824757e-08,
      "loss": 0.7806,
      "step": 5067
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9578189147169288,
      "learning_rate": 3.2787877384045095e-08,
      "loss": 0.725,
      "step": 5068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9433596736402019,
      "learning_rate": 3.228565917476889e-08,
      "loss": 0.7993,
      "step": 5069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9645721683398218,
      "learning_rate": 3.178731077539743e-08,
      "loss": 0.8045,
      "step": 5070
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8679067999683483,
      "learning_rate": 3.129283237940928e-08,
      "loss": 0.7914,
      "step": 5071
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9751134411752432,
      "learning_rate": 3.080222417877421e-08,
      "loss": 0.7513,
      "step": 5072
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9367568284114622,
      "learning_rate": 3.031548636396764e-08,
      "loss": 0.7758,
      "step": 5073
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.020602360365608,
      "learning_rate": 2.983261912395397e-08,
      "loss": 0.7911,
      "step": 5074
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9997097101377811,
      "learning_rate": 2.9353622646199898e-08,
      "loss": 0.7784,
      "step": 5075
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8616109105926225,
      "learning_rate": 2.8878497116671124e-08,
      "loss": 0.7441,
      "step": 5076
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0029630206260236,
      "learning_rate": 2.8407242719823424e-08,
      "loss": 0.823,
      "step": 5077
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9148307143764053,
      "learning_rate": 2.7939859638617118e-08,
      "loss": 0.7951,
      "step": 5078
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9318130228565908,
      "learning_rate": 2.7476348054504832e-08,
      "loss": 0.7603,
      "step": 5079
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9351587223246965,
      "learning_rate": 2.7016708147439285e-08,
      "loss": 0.7734,
      "step": 5080
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9768623900099198,
      "learning_rate": 2.6560940095866626e-08,
      "loss": 0.8205,
      "step": 5081
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5832388385585285,
      "learning_rate": 2.6109044076733092e-08,
      "loss": 0.3391,
      "step": 5082
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0288162156480367,
      "learning_rate": 2.5661020265479452e-08,
      "loss": 0.7966,
      "step": 5083
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9241882315703519,
      "learning_rate": 2.5216868836043242e-08,
      "loss": 0.7764,
      "step": 5084
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9310341595240639,
      "learning_rate": 2.4776589960862074e-08,
      "loss": 0.7912,
      "step": 5085
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9452903583392599,
      "learning_rate": 2.434018381086589e-08,
      "loss": 0.7865,
      "step": 5086
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8673144233669848,
      "learning_rate": 2.3907650555481387e-08,
      "loss": 0.7515,
      "step": 5087
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9552187616170927,
      "learning_rate": 2.3478990362634235e-08,
      "loss": 0.7748,
      "step": 5088
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9468119460270613,
      "learning_rate": 2.3054203398743537e-08,
      "loss": 0.7967,
      "step": 5089
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0244340103723706,
      "learning_rate": 2.263328982872959e-08,
      "loss": 0.8403,
      "step": 5090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9544249042105117,
      "learning_rate": 2.221624981600168e-08,
      "loss": 0.7823,
      "step": 5091
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9966297138251075,
      "learning_rate": 2.1803083522471402e-08,
      "loss": 0.8359,
      "step": 5092
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9737431337896618,
      "learning_rate": 2.1393791108542672e-08,
      "loss": 0.8068,
      "step": 5093
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.891202798161864,
      "learning_rate": 2.098837273311838e-08,
      "loss": 0.7929,
      "step": 5094
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8913768118664277,
      "learning_rate": 2.058682855359595e-08,
      "loss": 0.7976,
      "step": 5095
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8846957605979979,
      "learning_rate": 2.0189158725867353e-08,
      "loss": 0.7538,
      "step": 5096
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8559566719219804,
      "learning_rate": 1.979536340432131e-08,
      "loss": 0.7444,
      "step": 5097
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8658379162281634,
      "learning_rate": 1.9405442741844415e-08,
      "loss": 0.7007,
      "step": 5098
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.006534037666525,
      "learning_rate": 1.9019396889816688e-08,
      "loss": 0.7555,
      "step": 5099
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.943673028149766,
      "learning_rate": 1.8637225998114904e-08,
      "loss": 0.7572,
      "step": 5100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9512758239817569,
      "learning_rate": 1.825893021510927e-08,
      "loss": 0.768,
      "step": 5101
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9400762873639721,
      "learning_rate": 1.7884509687668972e-08,
      "loss": 0.8087,
      "step": 5102
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.93564056084685,
      "learning_rate": 1.7513964561156617e-08,
      "loss": 0.7836,
      "step": 5103
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.937754727318476,
      "learning_rate": 1.714729497942935e-08,
      "loss": 0.7878,
      "step": 5104
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9565719665330287,
      "learning_rate": 1.6784501084843307e-08,
      "loss": 0.8063,
      "step": 5105
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9528335506522491,
      "learning_rate": 1.6425583018244706e-08,
      "loss": 0.7808,
      "step": 5106
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9786735978689565,
      "learning_rate": 1.607054091897986e-08,
      "loss": 0.6928,
      "step": 5107
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8475735570195716,
      "learning_rate": 1.57193749248874e-08,
      "loss": 0.7831,
      "step": 5108
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8944931653350722,
      "learning_rate": 1.537208517230271e-08,
      "loss": 0.8039,
      "step": 5109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8800696498506794,
      "learning_rate": 1.5028671796055715e-08,
      "loss": 0.794,
      "step": 5110
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9609674508672396,
      "learning_rate": 1.4689134929470884e-08,
      "loss": 0.8013,
      "step": 5111
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.939095287714589,
      "learning_rate": 1.435347470436832e-08,
      "loss": 0.773,
      "step": 5112
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.919904300320353,
      "learning_rate": 1.4021691251062675e-08,
      "loss": 0.7638,
      "step": 5113
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9662350939400373,
      "learning_rate": 1.3693784698363133e-08,
      "loss": 0.7187,
      "step": 5114
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8970800093096211,
      "learning_rate": 1.3369755173575639e-08,
      "loss": 0.7529,
      "step": 5115
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9322852151768934,
      "learning_rate": 1.3049602802498451e-08,
      "loss": 0.8244,
      "step": 5116
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9225464606655747,
      "learning_rate": 1.273332770942659e-08,
      "loss": 0.7963,
      "step": 5117
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6384150956035214,
      "learning_rate": 1.2420930017148503e-08,
      "loss": 0.3329,
      "step": 5118
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0056637214905293,
      "learning_rate": 1.2112409846947171e-08,
      "loss": 0.7812,
      "step": 5119
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9133939542257811,
      "learning_rate": 1.1807767318602337e-08,
      "loss": 0.7793,
      "step": 5120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9201037191505707,
      "learning_rate": 1.150700255038606e-08,
      "loss": 0.7673,
      "step": 5121
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9476147438462665,
      "learning_rate": 1.1210115659063825e-08,
      "loss": 0.7466,
      "step": 5122
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8733767494571113,
      "learning_rate": 1.0917106759900097e-08,
      "loss": 0.732,
      "step": 5123
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9535135533444326,
      "learning_rate": 1.0627975966649439e-08,
      "loss": 0.7928,
      "step": 5124
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9019327948940158,
      "learning_rate": 1.034272339156206e-08,
      "loss": 0.7962,
      "step": 5125
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8985172305025607,
      "learning_rate": 1.0061349145383814e-08,
      "loss": 0.7601,
      "step": 5126
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9054284356578993,
      "learning_rate": 9.783853337353987e-09,
      "loss": 0.8107,
      "step": 5127
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5744009790118957,
      "learning_rate": 9.510236075205292e-09,
      "loss": 0.3293,
      "step": 5128
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8889243109289029,
      "learning_rate": 9.240497465164978e-09,
      "loss": 0.7509,
      "step": 5129
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.967469800375464,
      "learning_rate": 8.974637611955939e-09,
      "loss": 0.8269,
      "step": 5130
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.899904228393208,
      "learning_rate": 8.712656618793391e-09,
      "loss": 0.7699,
      "step": 5131
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9583490775630612,
      "learning_rate": 8.454554587388198e-09,
      "loss": 0.7423,
      "step": 5132
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9507210170157205,
      "learning_rate": 8.200331617943535e-09,
      "loss": 0.8253,
      "step": 5133
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9119979063542955,
      "learning_rate": 7.949987809158232e-09,
      "loss": 0.7957,
      "step": 5134
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5999698819095963,
      "learning_rate": 7.703523258223433e-09,
      "loss": 0.3323,
      "step": 5135
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9125571332489604,
      "learning_rate": 7.460938060825929e-09,
      "loss": 0.7841,
      "step": 5136
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8468595692341249,
      "learning_rate": 7.222232311145938e-09,
      "loss": 0.8129,
      "step": 5137
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9147798099445484,
      "learning_rate": 6.987406101855998e-09,
      "loss": 0.7647,
      "step": 5138
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9831081104347278,
      "learning_rate": 6.756459524125403e-09,
      "loss": 0.8181,
      "step": 5139
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9707165950765443,
      "learning_rate": 6.5293926676135434e-09,
      "loss": 0.8618,
      "step": 5140
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9177015154003823,
      "learning_rate": 6.306205620477679e-09,
      "loss": 0.7814,
      "step": 5141
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9549147216480882,
      "learning_rate": 6.086898469365166e-09,
      "loss": 0.7776,
      "step": 5142
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.904510061349586,
      "learning_rate": 5.871471299419007e-09,
      "loss": 0.8094,
      "step": 5143
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9487217571675315,
      "learning_rate": 5.6599241942767445e-09,
      "loss": 0.7919,
      "step": 5144
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9663504789747849,
      "learning_rate": 5.452257236066017e-09,
      "loss": 0.7685,
      "step": 5145
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5899693946564348,
      "learning_rate": 5.248470505412328e-09,
      "loss": 0.3275,
      "step": 5146
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9502964737784096,
      "learning_rate": 5.0485640814312844e-09,
      "loss": 0.8314,
      "step": 5147
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9869233927048849,
      "learning_rate": 4.8525380417330234e-09,
      "loss": 0.8025,
      "step": 5148
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8903816337937716,
      "learning_rate": 4.660392462424446e-09,
      "loss": 0.777,
      "step": 5149
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.96910506649411,
      "learning_rate": 4.472127418099215e-09,
      "loss": 0.7915,
      "step": 5150
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9842525709507112,
      "learning_rate": 4.287742981851084e-09,
      "loss": 0.7756,
      "step": 5151
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8481425622698692,
      "learning_rate": 4.1072392252639034e-09,
      "loss": 0.7528,
      "step": 5152
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9368545548020248,
      "learning_rate": 3.930616218414951e-09,
      "loss": 0.7879,
      "step": 5153
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8929688901386961,
      "learning_rate": 3.757874029874931e-09,
      "loss": 0.8327,
      "step": 5154
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9386928715443122,
      "learning_rate": 3.5890127267090844e-09,
      "loss": 0.8066,
      "step": 5155
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8987132938420193,
      "learning_rate": 3.424032374476083e-09,
      "loss": 0.7435,
      "step": 5156
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9188435229053308,
      "learning_rate": 3.2629330372246915e-09,
      "loss": 0.8048,
      "step": 5157
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.924344084935355,
      "learning_rate": 3.105714777501545e-09,
      "loss": 0.7469,
      "step": 5158
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9066754933343186,
      "learning_rate": 2.9523776563422644e-09,
      "loss": 0.7686,
      "step": 5159
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9334741599607584,
      "learning_rate": 2.802921733278119e-09,
      "loss": 0.782,
      "step": 5160
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.976141219465834,
      "learning_rate": 2.657347066333804e-09,
      "loss": 0.7963,
      "step": 5161
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9608707333306407,
      "learning_rate": 2.5156537120263335e-09,
      "loss": 0.8099,
      "step": 5162
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9084370972495627,
      "learning_rate": 2.3778417253650376e-09,
      "loss": 0.8213,
      "step": 5163
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9285927841767859,
      "learning_rate": 2.2439111598537844e-09,
      "loss": 0.7989,
      "step": 5164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9480905953078294,
      "learning_rate": 2.113862067488759e-09,
      "loss": 0.8418,
      "step": 5165
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8262040410137912,
      "learning_rate": 1.987694498760684e-09,
      "loss": 0.7587,
      "step": 5166
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9788578490517359,
      "learning_rate": 1.865408502650379e-09,
      "loss": 0.7813,
      "step": 5167
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9653344757307686,
      "learning_rate": 1.747004126635421e-09,
      "loss": 0.7693,
      "step": 5168
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8629673964740447,
      "learning_rate": 1.6324814166823744e-09,
      "loss": 0.7859,
      "step": 5169
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8800283085730147,
      "learning_rate": 1.5218404172545609e-09,
      "loss": 0.7892,
      "step": 5170
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9877468475260579,
      "learning_rate": 1.415081171305399e-09,
      "loss": 0.7797,
      "step": 5171
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9431161023836341,
      "learning_rate": 1.3122037202828452e-09,
      "loss": 0.8275,
      "step": 5172
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8986229999181022,
      "learning_rate": 1.2132081041282829e-09,
      "loss": 0.8158,
      "step": 5173
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.983870449239283,
      "learning_rate": 1.1180943612754124e-09,
      "loss": 0.7862,
      "step": 5174
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9869401472899687,
      "learning_rate": 1.026862528649142e-09,
      "loss": 0.7994,
      "step": 5175
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9829225243284404,
      "learning_rate": 9.39512641668916e-10,
      "loss": 0.7374,
      "step": 5176
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9558566646241304,
      "learning_rate": 8.560447342487177e-10,
      "loss": 0.7653,
      "step": 5177
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8820386208252872,
      "learning_rate": 7.764588387915161e-10,
      "loss": 0.7696,
      "step": 5178
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9112632028004433,
      "learning_rate": 7.007549861970387e-10,
      "loss": 0.7031,
      "step": 5179
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0367284374208465,
      "learning_rate": 6.289332058551089e-10,
      "loss": 0.8306,
      "step": 5180
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8731881421801555,
      "learning_rate": 5.609935256500887e-10,
      "loss": 0.7919,
      "step": 5181
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9562688754165123,
      "learning_rate": 4.969359719586563e-10,
      "loss": 0.7407,
      "step": 5182
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9365916284491022,
      "learning_rate": 4.3676056964869764e-10,
      "loss": 0.7812,
      "step": 5183
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8705033463313663,
      "learning_rate": 3.804673420837457e-10,
      "loss": 0.7915,
      "step": 5184
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8640111600888389,
      "learning_rate": 3.2805631111743064e-10,
      "loss": 0.7918,
      "step": 5185
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8136543616849451,
      "learning_rate": 2.795274971001405e-10,
      "loss": 0.7336,
      "step": 5186
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9108737795064036,
      "learning_rate": 2.3488091886902933e-10,
      "loss": 0.7745,
      "step": 5187
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.913993863820868,
      "learning_rate": 1.941165937602296e-10,
      "loss": 0.8198,
      "step": 5188
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8658599803768934,
      "learning_rate": 1.5723453759886042e-10,
      "loss": 0.852,
      "step": 5189
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9426690719351505,
      "learning_rate": 1.2423476470346808e-10,
      "loss": 0.7571,
      "step": 5190
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.952984964844677,
      "learning_rate": 9.511728788602625e-11,
      "loss": 0.8067,
      "step": 5191
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9295809745114999,
      "learning_rate": 6.988211845082582e-11,
      "loss": 0.8261,
      "step": 5192
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0005382445777,
      "learning_rate": 4.852926619447473e-11,
      "loss": 0.839,
      "step": 5193
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9522561217816633,
      "learning_rate": 3.105873940811854e-11,
      "loss": 0.8171,
      "step": 5194
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0046022981012765,
      "learning_rate": 1.7470544874109706e-11,
      "loss": 0.7454,
      "step": 5195
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9224900643951546,
      "learning_rate": 7.764687866007592e-12,
      "loss": 0.7823,
      "step": 5196
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2685547945038689,
      "learning_rate": 1.9411721552398123e-12,
      "loss": 0.5522,
      "step": 5197
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7566593647144272,
      "learning_rate": 0.0,
      "loss": 0.4772,
      "step": 5198
    },
    {
      "epoch": 1.0,
      "step": 5198,
      "total_flos": 1719199234555904.0,
      "train_loss": 0.028819141547557894,
      "train_runtime": 2434.9615,
      "train_samples_per_second": 273.227,
      "train_steps_per_second": 2.135
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5198,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 1719199234555904.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}