{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.3995405283923488,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001997702641961744,
      "grad_norm": 8.875,
      "learning_rate": 0.0,
      "loss": 1.4044,
      "memory/device_reserved (GiB)": 86.98,
      "memory/max_active (GiB)": 76.6,
      "memory/max_allocated (GiB)": 76.6,
      "step": 1,
      "tokens_per_second_per_gpu": 12080.3,
      "total_tokens": 88833
    },
    {
      "epoch": 0.0003995405283923488,
      "grad_norm": 8.625,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 1.407,
      "memory/device_reserved (GiB)": 101.86,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 2,
      "tokens_per_second_per_gpu": 19147.53,
      "total_tokens": 180993
    },
    {
      "epoch": 0.0005993107925885232,
      "grad_norm": 9.1875,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 1.4311,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 3,
      "tokens_per_second_per_gpu": 18223.2,
      "total_tokens": 268993
    },
    {
      "epoch": 0.0007990810567846976,
      "grad_norm": 8.0,
      "learning_rate": 6.000000000000001e-07,
      "loss": 1.3511,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 4,
      "tokens_per_second_per_gpu": 18306.2,
      "total_tokens": 359437
    },
    {
      "epoch": 0.000998851320980872,
      "grad_norm": 8.625,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1.4297,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 5,
      "tokens_per_second_per_gpu": 20326.18,
      "total_tokens": 456114
    },
    {
      "epoch": 0.0011986215851770463,
      "grad_norm": 8.625,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.4184,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 6,
      "tokens_per_second_per_gpu": 18521.54,
      "total_tokens": 545509
    },
    {
      "epoch": 0.0013983918493732208,
      "grad_norm": 7.8125,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.3454,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 7,
      "tokens_per_second_per_gpu": 18747.26,
      "total_tokens": 639166
    },
    {
      "epoch": 0.0015981621135693952,
      "grad_norm": 8.6875,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 1.4629,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 8,
      "tokens_per_second_per_gpu": 19943.66,
      "total_tokens": 734740
    },
    {
      "epoch": 0.0017979323777655696,
      "grad_norm": 8.5625,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 1.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 9,
      "tokens_per_second_per_gpu": 18252.37,
      "total_tokens": 822714
    },
    {
      "epoch": 0.001997702641961744,
      "grad_norm": 8.25,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.3985,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 10,
      "tokens_per_second_per_gpu": 17564.17,
      "total_tokens": 908386
    },
    {
      "epoch": 0.0021974729061579185,
      "grad_norm": 8.0625,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.3653,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 11,
      "tokens_per_second_per_gpu": 18947.2,
      "total_tokens": 1001224
    },
    {
      "epoch": 0.0023972431703540927,
      "grad_norm": 8.25,
      "learning_rate": 2.2e-06,
      "loss": 1.3525,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 12,
      "tokens_per_second_per_gpu": 15991.21,
      "total_tokens": 1078798
    },
    {
      "epoch": 0.0025970134345502673,
      "grad_norm": 8.0,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.3854,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 13,
      "tokens_per_second_per_gpu": 19282.61,
      "total_tokens": 1172554
    },
    {
      "epoch": 0.0027967836987464415,
      "grad_norm": 7.03125,
      "learning_rate": 2.6e-06,
      "loss": 1.3296,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 14,
      "tokens_per_second_per_gpu": 19141.05,
      "total_tokens": 1266344
    },
    {
      "epoch": 0.002996553962942616,
      "grad_norm": 7.28125,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.3306,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 15,
      "tokens_per_second_per_gpu": 17741.01,
      "total_tokens": 1353101
    },
    {
      "epoch": 0.0031963242271387904,
      "grad_norm": 6.9375,
      "learning_rate": 3e-06,
      "loss": 1.3057,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 16,
      "tokens_per_second_per_gpu": 17461.64,
      "total_tokens": 1439123
    },
    {
      "epoch": 0.0033960944913349646,
      "grad_norm": 6.625,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.3333,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 17,
      "tokens_per_second_per_gpu": 18994.25,
      "total_tokens": 1531013
    },
    {
      "epoch": 0.0035958647555311392,
      "grad_norm": 6.65625,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.3206,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 18,
      "tokens_per_second_per_gpu": 17491.17,
      "total_tokens": 1615560
    },
    {
      "epoch": 0.0037956350197273134,
      "grad_norm": 6.09375,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.2949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 19,
      "tokens_per_second_per_gpu": 17298.22,
      "total_tokens": 1700965
    },
    {
      "epoch": 0.003995405283923488,
      "grad_norm": 6.0,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 1.3137,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 20,
      "tokens_per_second_per_gpu": 19409.75,
      "total_tokens": 1794883
    },
    {
      "epoch": 0.004195175548119662,
      "grad_norm": 5.4375,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.2392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 21,
      "tokens_per_second_per_gpu": 18253.96,
      "total_tokens": 1884703
    },
    {
      "epoch": 0.004394945812315837,
      "grad_norm": 4.90625,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.2102,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 22,
      "tokens_per_second_per_gpu": 18032.95,
      "total_tokens": 1973886
    },
    {
      "epoch": 0.0045947160765120116,
      "grad_norm": 4.71875,
      "learning_rate": 4.4e-06,
      "loss": 1.2464,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 23,
      "tokens_per_second_per_gpu": 18600.94,
      "total_tokens": 2064097
    },
    {
      "epoch": 0.004794486340708185,
      "grad_norm": 4.28125,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.2229,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 24,
      "tokens_per_second_per_gpu": 18997.25,
      "total_tokens": 2156067
    },
    {
      "epoch": 0.00499425660490436,
      "grad_norm": 4.0625,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.1683,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 25,
      "tokens_per_second_per_gpu": 18823.82,
      "total_tokens": 2246949
    },
    {
      "epoch": 0.005194026869100535,
      "grad_norm": 3.859375,
      "learning_rate": 5e-06,
      "loss": 1.2146,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 26,
      "tokens_per_second_per_gpu": 20230.9,
      "total_tokens": 2343942
    },
    {
      "epoch": 0.005393797133296708,
      "grad_norm": 3.328125,
      "learning_rate": 5.2e-06,
      "loss": 1.1309,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 27,
      "tokens_per_second_per_gpu": 18797.19,
      "total_tokens": 2435894
    },
    {
      "epoch": 0.005593567397492883,
      "grad_norm": 3.4375,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.0593,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 28,
      "tokens_per_second_per_gpu": 17280.36,
      "total_tokens": 2521341
    },
    {
      "epoch": 0.005793337661689058,
      "grad_norm": 3.234375,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.0656,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 29,
      "tokens_per_second_per_gpu": 16782.87,
      "total_tokens": 2603702
    },
    {
      "epoch": 0.005993107925885232,
      "grad_norm": 2.5625,
      "learning_rate": 5.8e-06,
      "loss": 1.0975,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 30,
      "tokens_per_second_per_gpu": 19317.94,
      "total_tokens": 2698698
    },
    {
      "epoch": 0.006192878190081406,
      "grad_norm": 2.359375,
      "learning_rate": 6e-06,
      "loss": 1.0129,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 31,
      "tokens_per_second_per_gpu": 17596.21,
      "total_tokens": 2786649
    },
    {
      "epoch": 0.006392648454277581,
      "grad_norm": 2.34375,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.0431,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 32,
      "tokens_per_second_per_gpu": 17675.83,
      "total_tokens": 2872421
    },
    {
      "epoch": 0.006592418718473755,
      "grad_norm": 1.9765625,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.9933,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 33,
      "tokens_per_second_per_gpu": 17434.75,
      "total_tokens": 2958573
    },
    {
      "epoch": 0.006792188982669929,
      "grad_norm": 1.75,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.9853,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 34,
      "tokens_per_second_per_gpu": 18139.99,
      "total_tokens": 3048323
    },
    {
      "epoch": 0.006991959246866104,
      "grad_norm": 1.65625,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.9892,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 35,
      "tokens_per_second_per_gpu": 19104.98,
      "total_tokens": 3141049
    },
    {
      "epoch": 0.0071917295110622784,
      "grad_norm": 1.5390625,
      "learning_rate": 7e-06,
      "loss": 1.0487,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 36,
      "tokens_per_second_per_gpu": 19222.43,
      "total_tokens": 3232249
    },
    {
      "epoch": 0.007391499775258453,
      "grad_norm": 1.4140625,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.0083,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 37,
      "tokens_per_second_per_gpu": 18372.25,
      "total_tokens": 3321456
    },
    {
      "epoch": 0.007591270039454627,
      "grad_norm": 1.4375,
      "learning_rate": 7.4e-06,
      "loss": 0.9817,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 38,
      "tokens_per_second_per_gpu": 18347.2,
      "total_tokens": 3411507
    },
    {
      "epoch": 0.0077910403036508015,
      "grad_norm": 1.296875,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.9812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 39,
      "tokens_per_second_per_gpu": 19981.65,
      "total_tokens": 3507274
    },
    {
      "epoch": 0.007990810567846975,
      "grad_norm": 1.203125,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.9041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 40,
      "tokens_per_second_per_gpu": 17566.12,
      "total_tokens": 3594794
    },
    {
      "epoch": 0.00819058083204315,
      "grad_norm": 1.1796875,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.9809,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 41,
      "tokens_per_second_per_gpu": 19390.88,
      "total_tokens": 3687994
    },
    {
      "epoch": 0.008390351096239325,
      "grad_norm": 1.0625,
      "learning_rate": 8.2e-06,
      "loss": 0.8908,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 42,
      "tokens_per_second_per_gpu": 18208.34,
      "total_tokens": 3777101
    },
    {
      "epoch": 0.008590121360435498,
      "grad_norm": 1.078125,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.8831,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 43,
      "tokens_per_second_per_gpu": 18401.34,
      "total_tokens": 3866806
    },
    {
      "epoch": 0.008789891624631674,
      "grad_norm": 1.015625,
      "learning_rate": 8.6e-06,
      "loss": 0.8951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 44,
      "tokens_per_second_per_gpu": 18568.36,
      "total_tokens": 3957260
    },
    {
      "epoch": 0.008989661888827848,
      "grad_norm": 0.9765625,
      "learning_rate": 8.8e-06,
      "loss": 0.9264,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 45,
      "tokens_per_second_per_gpu": 20672.2,
      "total_tokens": 4056792
    },
    {
      "epoch": 0.009189432153024023,
      "grad_norm": 1.0078125,
      "learning_rate": 9e-06,
      "loss": 0.8714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 46,
      "tokens_per_second_per_gpu": 17404.11,
      "total_tokens": 4142165
    },
    {
      "epoch": 0.009389202417220197,
      "grad_norm": 0.9453125,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.9091,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 47,
      "tokens_per_second_per_gpu": 19389.02,
      "total_tokens": 4236108
    },
    {
      "epoch": 0.00958897268141637,
      "grad_norm": 0.921875,
      "learning_rate": 9.4e-06,
      "loss": 0.868,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 48,
      "tokens_per_second_per_gpu": 18098.57,
      "total_tokens": 4323994
    },
    {
      "epoch": 0.009788742945612546,
      "grad_norm": 0.89453125,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.8818,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 49,
      "tokens_per_second_per_gpu": 18228.99,
      "total_tokens": 4412792
    },
    {
      "epoch": 0.00998851320980872,
      "grad_norm": 0.9140625,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.8499,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 50,
      "tokens_per_second_per_gpu": 18242.39,
      "total_tokens": 4501083
    },
    {
      "epoch": 0.010188283474004894,
      "grad_norm": 0.8828125,
      "learning_rate": 1e-05,
      "loss": 0.8366,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 51,
      "tokens_per_second_per_gpu": 18626.18,
      "total_tokens": 4593306
    },
    {
      "epoch": 0.01038805373820107,
      "grad_norm": 0.86328125,
      "learning_rate": 1.02e-05,
      "loss": 0.8403,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 52,
      "tokens_per_second_per_gpu": 18781.03,
      "total_tokens": 4687235
    },
    {
      "epoch": 0.010587824002397243,
      "grad_norm": 0.85546875,
      "learning_rate": 1.04e-05,
      "loss": 0.8384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 53,
      "tokens_per_second_per_gpu": 17929.45,
      "total_tokens": 4775601
    },
    {
      "epoch": 0.010787594266593417,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.8393,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 54,
      "tokens_per_second_per_gpu": 19513.24,
      "total_tokens": 4869394
    },
    {
      "epoch": 0.010987364530789592,
      "grad_norm": 0.86328125,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.8511,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 55,
      "tokens_per_second_per_gpu": 19830.01,
      "total_tokens": 4967591
    },
    {
      "epoch": 0.011187134794985766,
      "grad_norm": 0.8125,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.7989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 56,
      "tokens_per_second_per_gpu": 19199.44,
      "total_tokens": 5061747
    },
    {
      "epoch": 0.011386905059181942,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.7682,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 57,
      "tokens_per_second_per_gpu": 17734.06,
      "total_tokens": 5149611
    },
    {
      "epoch": 0.011586675323378115,
      "grad_norm": 0.83203125,
      "learning_rate": 1.14e-05,
      "loss": 0.7711,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 58,
      "tokens_per_second_per_gpu": 18234.8,
      "total_tokens": 5238688
    },
    {
      "epoch": 0.011786445587574289,
      "grad_norm": 0.80859375,
      "learning_rate": 1.16e-05,
      "loss": 0.7814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 59,
      "tokens_per_second_per_gpu": 18792.26,
      "total_tokens": 5330605
    },
    {
      "epoch": 0.011986215851770465,
      "grad_norm": 0.79296875,
      "learning_rate": 1.18e-05,
      "loss": 0.8027,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 60,
      "tokens_per_second_per_gpu": 19697.85,
      "total_tokens": 5427929
    },
    {
      "epoch": 0.012185986115966638,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2e-05,
      "loss": 0.7635,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 61,
      "tokens_per_second_per_gpu": 19182.0,
      "total_tokens": 5523093
    },
    {
      "epoch": 0.012385756380162812,
      "grad_norm": 0.80078125,
      "learning_rate": 1.22e-05,
      "loss": 0.7506,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 62,
      "tokens_per_second_per_gpu": 16852.6,
      "total_tokens": 5606507
    },
    {
      "epoch": 0.012585526644358988,
      "grad_norm": 0.796875,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.759,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 63,
      "tokens_per_second_per_gpu": 18976.55,
      "total_tokens": 5699726
    },
    {
      "epoch": 0.012785296908555161,
      "grad_norm": 0.96875,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.7482,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 64,
      "tokens_per_second_per_gpu": 16754.37,
      "total_tokens": 5780971
    },
    {
      "epoch": 0.012985067172751335,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.7236,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 65,
      "tokens_per_second_per_gpu": 16821.88,
      "total_tokens": 5864343
    },
    {
      "epoch": 0.01318483743694751,
      "grad_norm": 0.796875,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.7203,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 66,
      "tokens_per_second_per_gpu": 17414.68,
      "total_tokens": 5950850
    },
    {
      "epoch": 0.013384607701143685,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.7425,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 67,
      "tokens_per_second_per_gpu": 17312.44,
      "total_tokens": 6036301
    },
    {
      "epoch": 0.013584377965339858,
      "grad_norm": 0.75,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.7071,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 68,
      "tokens_per_second_per_gpu": 18182.24,
      "total_tokens": 6126732
    },
    {
      "epoch": 0.013784148229536034,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.6961,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 69,
      "tokens_per_second_per_gpu": 17064.71,
      "total_tokens": 6210299
    },
    {
      "epoch": 0.013983918493732208,
      "grad_norm": 0.76171875,
      "learning_rate": 1.38e-05,
      "loss": 0.739,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 70,
      "tokens_per_second_per_gpu": 19599.41,
      "total_tokens": 6305011
    },
    {
      "epoch": 0.014183688757928383,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4e-05,
      "loss": 0.7451,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 71,
      "tokens_per_second_per_gpu": 17564.68,
      "total_tokens": 6392193
    },
    {
      "epoch": 0.014383459022124557,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.7211,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 72,
      "tokens_per_second_per_gpu": 17038.95,
      "total_tokens": 6478117
    },
    {
      "epoch": 0.01458322928632073,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.736,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 73,
      "tokens_per_second_per_gpu": 19603.37,
      "total_tokens": 6573453
    },
    {
      "epoch": 0.014782999550516906,
      "grad_norm": 0.74609375,
      "learning_rate": 1.46e-05,
      "loss": 0.7468,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 74,
      "tokens_per_second_per_gpu": 19574.77,
      "total_tokens": 6669304
    },
    {
      "epoch": 0.01498276981471308,
      "grad_norm": 0.73828125,
      "learning_rate": 1.48e-05,
      "loss": 0.7226,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 75,
      "tokens_per_second_per_gpu": 18206.4,
      "total_tokens": 6758301
    },
    {
      "epoch": 0.015182540078909254,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7005,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 76,
      "tokens_per_second_per_gpu": 17624.09,
      "total_tokens": 6845361
    },
    {
      "epoch": 0.01538231034310543,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.7132,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 77,
      "tokens_per_second_per_gpu": 19094.66,
      "total_tokens": 6939316
    },
    {
      "epoch": 0.015582080607301603,
      "grad_norm": 0.71484375,
      "learning_rate": 1.54e-05,
      "loss": 0.7367,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 78,
      "tokens_per_second_per_gpu": 20579.31,
      "total_tokens": 7039667
    },
    {
      "epoch": 0.015781850871497777,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.7241,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 79,
      "tokens_per_second_per_gpu": 18615.85,
      "total_tokens": 7130105
    },
    {
      "epoch": 0.01598162113569395,
      "grad_norm": 0.765625,
      "learning_rate": 1.58e-05,
      "loss": 0.7146,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 80,
      "tokens_per_second_per_gpu": 18876.03,
      "total_tokens": 7221634
    },
    {
      "epoch": 0.016181391399890128,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.687,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 81,
      "tokens_per_second_per_gpu": 18742.21,
      "total_tokens": 7314475
    },
    {
      "epoch": 0.0163811616640863,
      "grad_norm": 0.734375,
      "learning_rate": 1.62e-05,
      "loss": 0.6846,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 82,
      "tokens_per_second_per_gpu": 18625.5,
      "total_tokens": 7404915
    },
    {
      "epoch": 0.016580931928282475,
      "grad_norm": 0.71484375,
      "learning_rate": 1.64e-05,
      "loss": 0.6552,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 83,
      "tokens_per_second_per_gpu": 19646.24,
      "total_tokens": 7500716
    },
    {
      "epoch": 0.01678070219247865,
      "grad_norm": 0.75,
      "learning_rate": 1.66e-05,
      "loss": 0.6898,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 84,
      "tokens_per_second_per_gpu": 18157.99,
      "total_tokens": 7591257
    },
    {
      "epoch": 0.016980472456674823,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.6556,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 85,
      "tokens_per_second_per_gpu": 16299.95,
      "total_tokens": 7671631
    },
    {
      "epoch": 0.017180242720870997,
      "grad_norm": 0.734375,
      "learning_rate": 1.7e-05,
      "loss": 0.7308,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 86,
      "tokens_per_second_per_gpu": 19570.82,
      "total_tokens": 7766388
    },
    {
      "epoch": 0.017380012985067174,
      "grad_norm": 0.73828125,
      "learning_rate": 1.72e-05,
      "loss": 0.6462,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 87,
      "tokens_per_second_per_gpu": 18751.13,
      "total_tokens": 7858610
    },
    {
      "epoch": 0.017579783249263348,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.707,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 88,
      "tokens_per_second_per_gpu": 19935.79,
      "total_tokens": 7953990
    },
    {
      "epoch": 0.01777955351345952,
      "grad_norm": 0.7734375,
      "learning_rate": 1.76e-05,
      "loss": 0.6655,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 89,
      "tokens_per_second_per_gpu": 17805.17,
      "total_tokens": 8041953
    },
    {
      "epoch": 0.017979323777655695,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.7043,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 90,
      "tokens_per_second_per_gpu": 18686.85,
      "total_tokens": 8133623
    },
    {
      "epoch": 0.01817909404185187,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8e-05,
      "loss": 0.6558,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 91,
      "tokens_per_second_per_gpu": 19017.24,
      "total_tokens": 8225616
    },
    {
      "epoch": 0.018378864306048046,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.6747,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 92,
      "tokens_per_second_per_gpu": 19018.61,
      "total_tokens": 8317907
    },
    {
      "epoch": 0.01857863457024422,
      "grad_norm": 0.8125,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.6396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 93,
      "tokens_per_second_per_gpu": 16652.39,
      "total_tokens": 8400755
    },
    {
      "epoch": 0.018778404834440394,
      "grad_norm": 0.7265625,
      "learning_rate": 1.86e-05,
      "loss": 0.6564,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 94,
      "tokens_per_second_per_gpu": 20544.06,
      "total_tokens": 8498995
    },
    {
      "epoch": 0.018978175098636568,
      "grad_norm": 0.73828125,
      "learning_rate": 1.88e-05,
      "loss": 0.6344,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 95,
      "tokens_per_second_per_gpu": 18474.09,
      "total_tokens": 8588703
    },
    {
      "epoch": 0.01917794536283274,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9e-05,
      "loss": 0.6396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 96,
      "tokens_per_second_per_gpu": 17247.57,
      "total_tokens": 8672973
    },
    {
      "epoch": 0.019377715627028915,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.6338,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 97,
      "tokens_per_second_per_gpu": 18226.98,
      "total_tokens": 8760894
    },
    {
      "epoch": 0.019577485891225092,
      "grad_norm": 0.73828125,
      "learning_rate": 1.94e-05,
      "loss": 0.593,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 98,
      "tokens_per_second_per_gpu": 17647.57,
      "total_tokens": 8848209
    },
    {
      "epoch": 0.019777256155421266,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.6413,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 99,
      "tokens_per_second_per_gpu": 17897.86,
      "total_tokens": 8935716
    },
    {
      "epoch": 0.01997702641961744,
      "grad_norm": 0.7421875,
      "learning_rate": 1.98e-05,
      "loss": 0.6586,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 100,
      "tokens_per_second_per_gpu": 17561.44,
      "total_tokens": 9022769
    },
    {
      "epoch": 0.020176796683813614,
      "grad_norm": 0.734375,
      "learning_rate": 2e-05,
      "loss": 0.6223,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 101,
      "tokens_per_second_per_gpu": 19191.06,
      "total_tokens": 9117162
    },
    {
      "epoch": 0.020376566948009787,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9999986330190926e-05,
      "loss": 0.6677,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 102,
      "tokens_per_second_per_gpu": 18980.93,
      "total_tokens": 9209783
    },
    {
      "epoch": 0.020576337212205965,
      "grad_norm": 0.828125,
      "learning_rate": 1.9999945320801072e-05,
      "loss": 0.6169,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 103,
      "tokens_per_second_per_gpu": 17383.64,
      "total_tokens": 9294875
    },
    {
      "epoch": 0.02077610747640214,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9999876971942557e-05,
      "loss": 0.6845,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 104,
      "tokens_per_second_per_gpu": 19237.07,
      "total_tokens": 9388693
    },
    {
      "epoch": 0.020975877740598312,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9999781283802247e-05,
      "loss": 0.5892,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 105,
      "tokens_per_second_per_gpu": 18754.06,
      "total_tokens": 9479728
    },
    {
      "epoch": 0.021175648004794486,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9999658256641746e-05,
      "loss": 0.6118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 106,
      "tokens_per_second_per_gpu": 19826.12,
      "total_tokens": 9575808
    },
    {
      "epoch": 0.02137541826899066,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9999507890797408e-05,
      "loss": 0.6137,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 107,
      "tokens_per_second_per_gpu": 17785.72,
      "total_tokens": 9662301
    },
    {
      "epoch": 0.021575188533186834,
      "grad_norm": 0.7578125,
      "learning_rate": 1.999933018668033e-05,
      "loss": 0.6548,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 108,
      "tokens_per_second_per_gpu": 18852.68,
      "total_tokens": 9754531
    },
    {
      "epoch": 0.02177495879738301,
      "grad_norm": 0.80078125,
      "learning_rate": 1.999912514477634e-05,
      "loss": 0.5825,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 109,
      "tokens_per_second_per_gpu": 16761.61,
      "total_tokens": 9836019
    },
    {
      "epoch": 0.021974729061579185,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9998892765646024e-05,
      "loss": 0.583,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 110,
      "tokens_per_second_per_gpu": 17573.08,
      "total_tokens": 9922022
    },
    {
      "epoch": 0.02217449932577536,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9998633049924693e-05,
      "loss": 0.5982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 111,
      "tokens_per_second_per_gpu": 17187.69,
      "total_tokens": 10006500
    },
    {
      "epoch": 0.022374269589971532,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9998345998322397e-05,
      "loss": 0.6243,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 112,
      "tokens_per_second_per_gpu": 19734.71,
      "total_tokens": 10101597
    },
    {
      "epoch": 0.022574039854167706,
      "grad_norm": 0.81640625,
      "learning_rate": 1.999803161162393e-05,
      "loss": 0.5827,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 113,
      "tokens_per_second_per_gpu": 18512.19,
      "total_tokens": 10191595
    },
    {
      "epoch": 0.022773810118363883,
      "grad_norm": 0.73046875,
      "learning_rate": 1.999768989068881e-05,
      "loss": 0.6023,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 114,
      "tokens_per_second_per_gpu": 19205.86,
      "total_tokens": 10286847
    },
    {
      "epoch": 0.022973580382560057,
      "grad_norm": 0.78515625,
      "learning_rate": 1.999732083645129e-05,
      "loss": 0.5982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 115,
      "tokens_per_second_per_gpu": 20817.88,
      "total_tokens": 10385941
    },
    {
      "epoch": 0.02317335064675623,
      "grad_norm": 0.7578125,
      "learning_rate": 1.999692444992035e-05,
      "loss": 0.58,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 116,
      "tokens_per_second_per_gpu": 17677.18,
      "total_tokens": 10475782
    },
    {
      "epoch": 0.023373120910952404,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9996500732179695e-05,
      "loss": 0.6219,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 117,
      "tokens_per_second_per_gpu": 18722.8,
      "total_tokens": 10567203
    },
    {
      "epoch": 0.023572891175148578,
      "grad_norm": 0.77734375,
      "learning_rate": 1.999604968438775e-05,
      "loss": 0.6274,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 118,
      "tokens_per_second_per_gpu": 19547.27,
      "total_tokens": 10662969
    },
    {
      "epoch": 0.023772661439344752,
      "grad_norm": 0.80078125,
      "learning_rate": 1.999557130777767e-05,
      "loss": 0.6301,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 119,
      "tokens_per_second_per_gpu": 18747.21,
      "total_tokens": 10756150
    },
    {
      "epoch": 0.02397243170354093,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9995065603657317e-05,
      "loss": 0.5914,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 120,
      "tokens_per_second_per_gpu": 17661.72,
      "total_tokens": 10844152
    },
    {
      "epoch": 0.024172201967737103,
      "grad_norm": 0.7890625,
      "learning_rate": 1.999453257340926e-05,
      "loss": 0.5804,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 121,
      "tokens_per_second_per_gpu": 16813.53,
      "total_tokens": 10927577
    },
    {
      "epoch": 0.024371972231933277,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9993972218490792e-05,
      "loss": 0.5867,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 122,
      "tokens_per_second_per_gpu": 18331.11,
      "total_tokens": 11017683
    },
    {
      "epoch": 0.02457174249612945,
      "grad_norm": 0.78125,
      "learning_rate": 1.9993384540433892e-05,
      "loss": 0.5744,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 123,
      "tokens_per_second_per_gpu": 17821.87,
      "total_tokens": 11106561
    },
    {
      "epoch": 0.024771512760325624,
      "grad_norm": 0.78515625,
      "learning_rate": 1.999276954084526e-05,
      "loss": 0.5982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 124,
      "tokens_per_second_per_gpu": 17769.5,
      "total_tokens": 11194195
    },
    {
      "epoch": 0.0249712830245218,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9992127221406276e-05,
      "loss": 0.5615,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 125,
      "tokens_per_second_per_gpu": 18496.38,
      "total_tokens": 11283289
    },
    {
      "epoch": 0.025171053288717975,
      "grad_norm": 0.77734375,
      "learning_rate": 1.999145758387301e-05,
      "loss": 0.586,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 126,
      "tokens_per_second_per_gpu": 17185.75,
      "total_tokens": 11367142
    },
    {
      "epoch": 0.02537082355291415,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9990760630076238e-05,
      "loss": 0.5811,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 127,
      "tokens_per_second_per_gpu": 17957.32,
      "total_tokens": 11455512
    },
    {
      "epoch": 0.025570593817110323,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9990036361921402e-05,
      "loss": 0.5343,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 128,
      "tokens_per_second_per_gpu": 17543.18,
      "total_tokens": 11541449
    },
    {
      "epoch": 0.025770364081306497,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9989284781388617e-05,
      "loss": 0.5921,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 129,
      "tokens_per_second_per_gpu": 19047.51,
      "total_tokens": 11635059
    },
    {
      "epoch": 0.02597013434550267,
      "grad_norm": 0.796875,
      "learning_rate": 1.998850589053268e-05,
      "loss": 0.57,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 130,
      "tokens_per_second_per_gpu": 17215.1,
      "total_tokens": 11719610
    },
    {
      "epoch": 0.026169904609698848,
      "grad_norm": 0.7265625,
      "learning_rate": 1.998769969148305e-05,
      "loss": 0.5655,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 131,
      "tokens_per_second_per_gpu": 19427.08,
      "total_tokens": 11814476
    },
    {
      "epoch": 0.02636967487389502,
      "grad_norm": 0.75,
      "learning_rate": 1.998686618644384e-05,
      "loss": 0.5963,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 132,
      "tokens_per_second_per_gpu": 20717.45,
      "total_tokens": 11915303
    },
    {
      "epoch": 0.026569445138091195,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9986005377693826e-05,
      "loss": 0.5657,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 133,
      "tokens_per_second_per_gpu": 17670.14,
      "total_tokens": 12001866
    },
    {
      "epoch": 0.02676921540228737,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9985117267586425e-05,
      "loss": 0.5564,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 134,
      "tokens_per_second_per_gpu": 16572.72,
      "total_tokens": 12084140
    },
    {
      "epoch": 0.026968985666483543,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9984201858549695e-05,
      "loss": 0.5899,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 135,
      "tokens_per_second_per_gpu": 20121.57,
      "total_tokens": 12181918
    },
    {
      "epoch": 0.027168755930679717,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9983259153086328e-05,
      "loss": 0.6035,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 136,
      "tokens_per_second_per_gpu": 19107.52,
      "total_tokens": 12274774
    },
    {
      "epoch": 0.027368526194875894,
      "grad_norm": 0.765625,
      "learning_rate": 1.9982289153773648e-05,
      "loss": 0.6114,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 137,
      "tokens_per_second_per_gpu": 19383.78,
      "total_tokens": 12369086
    },
    {
      "epoch": 0.027568296459072068,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9981291863263593e-05,
      "loss": 0.5384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 138,
      "tokens_per_second_per_gpu": 18735.75,
      "total_tokens": 12459681
    },
    {
      "epoch": 0.02776806672326824,
      "grad_norm": 0.703125,
      "learning_rate": 1.9980267284282718e-05,
      "loss": 0.5689,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 139,
      "tokens_per_second_per_gpu": 18553.72,
      "total_tokens": 12550215
    },
    {
      "epoch": 0.027967836987464415,
      "grad_norm": 0.73828125,
      "learning_rate": 1.997921541963218e-05,
      "loss": 0.5829,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 140,
      "tokens_per_second_per_gpu": 20713.08,
      "total_tokens": 12650418
    },
    {
      "epoch": 0.02816760725166059,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9978136272187745e-05,
      "loss": 0.5724,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 141,
      "tokens_per_second_per_gpu": 19243.32,
      "total_tokens": 12744486
    },
    {
      "epoch": 0.028367377515856766,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9977029844899757e-05,
      "loss": 0.6022,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 142,
      "tokens_per_second_per_gpu": 20122.13,
      "total_tokens": 12842353
    },
    {
      "epoch": 0.02856714778005294,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9975896140793144e-05,
      "loss": 0.537,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 143,
      "tokens_per_second_per_gpu": 18211.8,
      "total_tokens": 12930372
    },
    {
      "epoch": 0.028766918044249114,
      "grad_norm": 0.7578125,
      "learning_rate": 1.997473516296741e-05,
      "loss": 0.5558,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 144,
      "tokens_per_second_per_gpu": 18313.62,
      "total_tokens": 13021241
    },
    {
      "epoch": 0.028966688308445288,
      "grad_norm": 0.78125,
      "learning_rate": 1.9973546914596622e-05,
      "loss": 0.5592,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 145,
      "tokens_per_second_per_gpu": 17905.13,
      "total_tokens": 13109482
    },
    {
      "epoch": 0.02916645857264146,
      "grad_norm": 0.76171875,
      "learning_rate": 1.997233139892941e-05,
      "loss": 0.5805,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 146,
      "tokens_per_second_per_gpu": 18622.61,
      "total_tokens": 13201391
    },
    {
      "epoch": 0.029366228836837635,
      "grad_norm": 1.15625,
      "learning_rate": 1.997108861928895e-05,
      "loss": 0.5857,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 147,
      "tokens_per_second_per_gpu": 18686.08,
      "total_tokens": 13293892
    },
    {
      "epoch": 0.029565999101033812,
      "grad_norm": 0.75,
      "learning_rate": 1.9969818579072945e-05,
      "loss": 0.5786,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 148,
      "tokens_per_second_per_gpu": 20557.26,
      "total_tokens": 13393443
    },
    {
      "epoch": 0.029765769365229986,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9968521281753642e-05,
      "loss": 0.5749,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 149,
      "tokens_per_second_per_gpu": 19278.4,
      "total_tokens": 13486462
    },
    {
      "epoch": 0.02996553962942616,
      "grad_norm": 0.75,
      "learning_rate": 1.9967196730877805e-05,
      "loss": 0.5692,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 150,
      "tokens_per_second_per_gpu": 18959.54,
      "total_tokens": 13579070
    },
    {
      "epoch": 0.030165309893622334,
      "grad_norm": 0.76171875,
      "learning_rate": 1.99658449300667e-05,
      "loss": 0.559,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 151,
      "tokens_per_second_per_gpu": 18694.89,
      "total_tokens": 13670237
    },
    {
      "epoch": 0.030365080157818507,
      "grad_norm": 0.76953125,
      "learning_rate": 1.99644658830161e-05,
      "loss": 0.5599,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 152,
      "tokens_per_second_per_gpu": 18732.63,
      "total_tokens": 13762208
    },
    {
      "epoch": 0.030564850422014685,
      "grad_norm": 0.71875,
      "learning_rate": 1.996305959349627e-05,
      "loss": 0.5453,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 153,
      "tokens_per_second_per_gpu": 18334.7,
      "total_tokens": 13852296
    },
    {
      "epoch": 0.03076462068621086,
      "grad_norm": 0.76953125,
      "learning_rate": 1.996162606535195e-05,
      "loss": 0.5419,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 154,
      "tokens_per_second_per_gpu": 18688.64,
      "total_tokens": 13944542
    },
    {
      "epoch": 0.030964390950407032,
      "grad_norm": 0.73828125,
      "learning_rate": 1.996016530250235e-05,
      "loss": 0.512,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 155,
      "tokens_per_second_per_gpu": 18537.35,
      "total_tokens": 14036025
    },
    {
      "epoch": 0.031164161214603206,
      "grad_norm": 0.765625,
      "learning_rate": 1.995867730894114e-05,
      "loss": 0.5891,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 156,
      "tokens_per_second_per_gpu": 19367.41,
      "total_tokens": 14129766
    },
    {
      "epoch": 0.03136393147879938,
      "grad_norm": 0.734375,
      "learning_rate": 1.995716208873644e-05,
      "loss": 0.5023,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 157,
      "tokens_per_second_per_gpu": 17986.75,
      "total_tokens": 14218809
    },
    {
      "epoch": 0.031563701742995554,
      "grad_norm": 0.80859375,
      "learning_rate": 1.99556196460308e-05,
      "loss": 0.5582,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 158,
      "tokens_per_second_per_gpu": 18289.61,
      "total_tokens": 14308512
    },
    {
      "epoch": 0.03176347200719173,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9954049985041208e-05,
      "loss": 0.5602,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 159,
      "tokens_per_second_per_gpu": 19095.73,
      "total_tokens": 14402261
    },
    {
      "epoch": 0.0319632422713879,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9952453110059044e-05,
      "loss": 0.53,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 160,
      "tokens_per_second_per_gpu": 17428.99,
      "total_tokens": 14488127
    },
    {
      "epoch": 0.032163012535584075,
      "grad_norm": 0.78125,
      "learning_rate": 1.9950829025450116e-05,
      "loss": 0.5618,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 161,
      "tokens_per_second_per_gpu": 18766.0,
      "total_tokens": 14579841
    },
    {
      "epoch": 0.032362782799780256,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9949177735654602e-05,
      "loss": 0.541,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 162,
      "tokens_per_second_per_gpu": 17468.83,
      "total_tokens": 14665490
    },
    {
      "epoch": 0.03256255306397643,
      "grad_norm": 0.953125,
      "learning_rate": 1.994749924518707e-05,
      "loss": 0.5562,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 163,
      "tokens_per_second_per_gpu": 17050.48,
      "total_tokens": 14750657
    },
    {
      "epoch": 0.0327623233281726,
      "grad_norm": 0.734375,
      "learning_rate": 1.994579355863644e-05,
      "loss": 0.5622,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 164,
      "tokens_per_second_per_gpu": 19576.0,
      "total_tokens": 14843795
    },
    {
      "epoch": 0.03296209359236878,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9944060680666e-05,
      "loss": 0.5341,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 165,
      "tokens_per_second_per_gpu": 18987.33,
      "total_tokens": 14935021
    },
    {
      "epoch": 0.03316186385656495,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9942300616013378e-05,
      "loss": 0.5778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 166,
      "tokens_per_second_per_gpu": 19928.3,
      "total_tokens": 15031594
    },
    {
      "epoch": 0.033361634120761124,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9940513369490516e-05,
      "loss": 0.516,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 167,
      "tokens_per_second_per_gpu": 17935.25,
      "total_tokens": 15120321
    },
    {
      "epoch": 0.0335614043849573,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9938698945983676e-05,
      "loss": 0.5185,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 168,
      "tokens_per_second_per_gpu": 17234.01,
      "total_tokens": 15204132
    },
    {
      "epoch": 0.03376117464915347,
      "grad_norm": 0.76953125,
      "learning_rate": 1.993685735045343e-05,
      "loss": 0.5386,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 169,
      "tokens_per_second_per_gpu": 17444.21,
      "total_tokens": 15290200
    },
    {
      "epoch": 0.033960944913349646,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9934988587934624e-05,
      "loss": 0.5248,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 170,
      "tokens_per_second_per_gpu": 18833.13,
      "total_tokens": 15383316
    },
    {
      "epoch": 0.03416071517754582,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9933092663536384e-05,
      "loss": 0.4718,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 171,
      "tokens_per_second_per_gpu": 17169.59,
      "total_tokens": 15468208
    },
    {
      "epoch": 0.03436048544174199,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9931169582442096e-05,
      "loss": 0.5478,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 172,
      "tokens_per_second_per_gpu": 19532.52,
      "total_tokens": 15564114
    },
    {
      "epoch": 0.034560255705938174,
      "grad_norm": 0.71484375,
      "learning_rate": 1.9929219349909393e-05,
      "loss": 0.5404,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 173,
      "tokens_per_second_per_gpu": 19204.82,
      "total_tokens": 15657417
    },
    {
      "epoch": 0.03476002597013435,
      "grad_norm": 0.73828125,
      "learning_rate": 1.992724197127013e-05,
      "loss": 0.5047,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 174,
      "tokens_per_second_per_gpu": 18003.06,
      "total_tokens": 15745533
    },
    {
      "epoch": 0.03495979623433052,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9925237451930392e-05,
      "loss": 0.5294,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 175,
      "tokens_per_second_per_gpu": 18119.65,
      "total_tokens": 15835918
    },
    {
      "epoch": 0.035159566498526695,
      "grad_norm": 0.79296875,
      "learning_rate": 1.992320579737045e-05,
      "loss": 0.5178,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 176,
      "tokens_per_second_per_gpu": 18794.6,
      "total_tokens": 15927874
    },
    {
      "epoch": 0.03535933676272287,
      "grad_norm": 0.70703125,
      "learning_rate": 1.9921147013144782e-05,
      "loss": 0.531,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 177,
      "tokens_per_second_per_gpu": 19276.65,
      "total_tokens": 16023301
    },
    {
      "epoch": 0.03555910702691904,
      "grad_norm": 0.78125,
      "learning_rate": 1.9919061104882012e-05,
      "loss": 0.5326,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 178,
      "tokens_per_second_per_gpu": 16154.49,
      "total_tokens": 16102167
    },
    {
      "epoch": 0.03575887729111522,
      "grad_norm": 0.75390625,
      "learning_rate": 1.991694807828494e-05,
      "loss": 0.5307,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 179,
      "tokens_per_second_per_gpu": 19095.35,
      "total_tokens": 16194773
    },
    {
      "epoch": 0.03595864755531139,
      "grad_norm": 0.76953125,
      "learning_rate": 1.99148079391305e-05,
      "loss": 0.5018,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 180,
      "tokens_per_second_per_gpu": 18667.88,
      "total_tokens": 16286759
    },
    {
      "epoch": 0.036158417819507564,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9912640693269754e-05,
      "loss": 0.5255,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 181,
      "tokens_per_second_per_gpu": 18595.09,
      "total_tokens": 16378176
    },
    {
      "epoch": 0.03635818808370374,
      "grad_norm": 0.71875,
      "learning_rate": 1.991044634662786e-05,
      "loss": 0.5391,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 182,
      "tokens_per_second_per_gpu": 19169.52,
      "total_tokens": 16472899
    },
    {
      "epoch": 0.03655795834789991,
      "grad_norm": 0.7421875,
      "learning_rate": 1.990822490520409e-05,
      "loss": 0.5133,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 183,
      "tokens_per_second_per_gpu": 17061.43,
      "total_tokens": 16556259
    },
    {
      "epoch": 0.03675772861209609,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9905976375071773e-05,
      "loss": 0.5079,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 184,
      "tokens_per_second_per_gpu": 18900.1,
      "total_tokens": 16649863
    },
    {
      "epoch": 0.036957498876292266,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9903700762378303e-05,
      "loss": 0.5118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 185,
      "tokens_per_second_per_gpu": 17944.8,
      "total_tokens": 16738176
    },
    {
      "epoch": 0.03715726914048844,
      "grad_norm": 0.73046875,
      "learning_rate": 1.990139807334512e-05,
      "loss": 0.5161,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 186,
      "tokens_per_second_per_gpu": 19093.54,
      "total_tokens": 16830975
    },
    {
      "epoch": 0.037357039404684614,
      "grad_norm": 0.75390625,
      "learning_rate": 1.989906831426769e-05,
      "loss": 0.4813,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 187,
      "tokens_per_second_per_gpu": 16983.73,
      "total_tokens": 16915897
    },
    {
      "epoch": 0.03755680966888079,
      "grad_norm": 0.82421875,
      "learning_rate": 1.989671149151548e-05,
      "loss": 0.4745,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 188,
      "tokens_per_second_per_gpu": 16127.77,
      "total_tokens": 16995210
    },
    {
      "epoch": 0.03775657993307696,
      "grad_norm": 0.79296875,
      "learning_rate": 1.989432761153196e-05,
      "loss": 0.5425,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 189,
      "tokens_per_second_per_gpu": 19580.59,
      "total_tokens": 17089299
    },
    {
      "epoch": 0.037956350197273135,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9891916680834565e-05,
      "loss": 0.5239,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 190,
      "tokens_per_second_per_gpu": 19684.82,
      "total_tokens": 17183281
    },
    {
      "epoch": 0.03815612046146931,
      "grad_norm": 0.75,
      "learning_rate": 1.9889478706014687e-05,
      "loss": 0.5077,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 191,
      "tokens_per_second_per_gpu": 18583.5,
      "total_tokens": 17273403
    },
    {
      "epoch": 0.03835589072566548,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9887013693737653e-05,
      "loss": 0.518,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 192,
      "tokens_per_second_per_gpu": 16473.81,
      "total_tokens": 17353898
    },
    {
      "epoch": 0.038555660989861656,
      "grad_norm": 0.734375,
      "learning_rate": 1.9884521650742718e-05,
      "loss": 0.5316,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 193,
      "tokens_per_second_per_gpu": 20766.07,
      "total_tokens": 17454159
    },
    {
      "epoch": 0.03875543125405783,
      "grad_norm": 0.8125,
      "learning_rate": 1.9882002583843025e-05,
      "loss": 0.5092,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 194,
      "tokens_per_second_per_gpu": 18137.45,
      "total_tokens": 17542740
    },
    {
      "epoch": 0.03895520151825401,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9879456499925616e-05,
      "loss": 0.5313,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 195,
      "tokens_per_second_per_gpu": 17163.27,
      "total_tokens": 17626349
    },
    {
      "epoch": 0.039154971782450185,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 0.5259,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 196,
      "tokens_per_second_per_gpu": 18303.77,
      "total_tokens": 17715930
    },
    {
      "epoch": 0.03935474204664636,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9874283308955058e-05,
      "loss": 0.545,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 197,
      "tokens_per_second_per_gpu": 18807.98,
      "total_tokens": 17807788
    },
    {
      "epoch": 0.03955451231084253,
      "grad_norm": 0.78125,
      "learning_rate": 1.987165621604522e-05,
      "loss": 0.5163,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 198,
      "tokens_per_second_per_gpu": 17359.32,
      "total_tokens": 17892790
    },
    {
      "epoch": 0.039754282575038706,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9869002134404235e-05,
      "loss": 0.5076,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 199,
      "tokens_per_second_per_gpu": 17503.69,
      "total_tokens": 17977266
    },
    {
      "epoch": 0.03995405283923488,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9866321071288262e-05,
      "loss": 0.5306,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 200,
      "tokens_per_second_per_gpu": 18864.02,
      "total_tokens": 18068913
    },
    {
      "epoch": 0.040153823103431054,
      "grad_norm": 0.75,
      "learning_rate": 1.9863613034027224e-05,
      "loss": 0.5142,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 201,
      "tokens_per_second_per_gpu": 17886.53,
      "total_tokens": 18155666
    },
    {
      "epoch": 0.04035359336762723,
      "grad_norm": 0.734375,
      "learning_rate": 1.9860878030024792e-05,
      "loss": 0.4774,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 202,
      "tokens_per_second_per_gpu": 18049.28,
      "total_tokens": 18244991
    },
    {
      "epoch": 0.0405533636318234,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9858116066758362e-05,
      "loss": 0.5113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 203,
      "tokens_per_second_per_gpu": 18297.54,
      "total_tokens": 18334899
    },
    {
      "epoch": 0.040753133896019575,
      "grad_norm": 0.765625,
      "learning_rate": 1.985532715177904e-05,
      "loss": 0.5054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 204,
      "tokens_per_second_per_gpu": 18687.22,
      "total_tokens": 18426521
    },
    {
      "epoch": 0.04095290416021575,
      "grad_norm": 0.7578125,
      "learning_rate": 1.985251129271161e-05,
      "loss": 0.5321,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 205,
      "tokens_per_second_per_gpu": 19813.2,
      "total_tokens": 18523782
    },
    {
      "epoch": 0.04115267442441193,
      "grad_norm": 0.7734375,
      "learning_rate": 1.984966849725452e-05,
      "loss": 0.5167,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 206,
      "tokens_per_second_per_gpu": 18315.05,
      "total_tokens": 18613362
    },
    {
      "epoch": 0.0413524446886081,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9846798773179866e-05,
      "loss": 0.505,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 207,
      "tokens_per_second_per_gpu": 18236.78,
      "total_tokens": 18703756
    },
    {
      "epoch": 0.04155221495280428,
      "grad_norm": 0.76953125,
      "learning_rate": 1.984390212833337e-05,
      "loss": 0.4877,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 208,
      "tokens_per_second_per_gpu": 16810.04,
      "total_tokens": 18786807
    },
    {
      "epoch": 0.04175198521700045,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9840978570634338e-05,
      "loss": 0.5121,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 209,
      "tokens_per_second_per_gpu": 18951.32,
      "total_tokens": 18878778
    },
    {
      "epoch": 0.041951755481196625,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9838028108075673e-05,
      "loss": 0.5328,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 210,
      "tokens_per_second_per_gpu": 19385.0,
      "total_tokens": 18972299
    },
    {
      "epoch": 0.0421515257453928,
      "grad_norm": 0.734375,
      "learning_rate": 1.9835050748723826e-05,
      "loss": 0.5102,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 211,
      "tokens_per_second_per_gpu": 19441.08,
      "total_tokens": 19066992
    },
    {
      "epoch": 0.04235129600958897,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9832046500718782e-05,
      "loss": 0.5337,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 212,
      "tokens_per_second_per_gpu": 19633.32,
      "total_tokens": 19161635
    },
    {
      "epoch": 0.042551066273785146,
      "grad_norm": 0.75390625,
      "learning_rate": 1.982901537227404e-05,
      "loss": 0.5345,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 213,
      "tokens_per_second_per_gpu": 19816.74,
      "total_tokens": 19257403
    },
    {
      "epoch": 0.04275083653798132,
      "grad_norm": 0.8046875,
      "learning_rate": 1.982595737167659e-05,
      "loss": 0.4904,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 214,
      "tokens_per_second_per_gpu": 17158.18,
      "total_tokens": 19340967
    },
    {
      "epoch": 0.04295060680217749,
      "grad_norm": 0.76171875,
      "learning_rate": 1.982287250728689e-05,
      "loss": 0.5192,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 215,
      "tokens_per_second_per_gpu": 19970.95,
      "total_tokens": 19437852
    },
    {
      "epoch": 0.04315037706637367,
      "grad_norm": 0.71875,
      "learning_rate": 1.981976078753884e-05,
      "loss": 0.5115,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 216,
      "tokens_per_second_per_gpu": 18294.72,
      "total_tokens": 19529326
    },
    {
      "epoch": 0.04335014733056985,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9816622220939762e-05,
      "loss": 0.5465,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 217,
      "tokens_per_second_per_gpu": 18747.69,
      "total_tokens": 19621920
    },
    {
      "epoch": 0.04354991759476602,
      "grad_norm": 0.7734375,
      "learning_rate": 1.981345681607038e-05,
      "loss": 0.465,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 218,
      "tokens_per_second_per_gpu": 17541.5,
      "total_tokens": 19709675
    },
    {
      "epoch": 0.043749687858962195,
      "grad_norm": 0.765625,
      "learning_rate": 1.981026458158479e-05,
      "loss": 0.475,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 219,
      "tokens_per_second_per_gpu": 17165.51,
      "total_tokens": 19794275
    },
    {
      "epoch": 0.04394945812315837,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9807045526210435e-05,
      "loss": 0.498,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 220,
      "tokens_per_second_per_gpu": 18983.04,
      "total_tokens": 19887636
    },
    {
      "epoch": 0.04414922838735454,
      "grad_norm": 0.78125,
      "learning_rate": 1.9803799658748096e-05,
      "loss": 0.488,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 221,
      "tokens_per_second_per_gpu": 17399.45,
      "total_tokens": 19972837
    },
    {
      "epoch": 0.04434899865155072,
      "grad_norm": 0.8125,
      "learning_rate": 1.9800526988071843e-05,
      "loss": 0.476,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 222,
      "tokens_per_second_per_gpu": 17239.56,
      "total_tokens": 20059232
    },
    {
      "epoch": 0.04454876891574689,
      "grad_norm": 0.76171875,
      "learning_rate": 1.979722752312904e-05,
      "loss": 0.5296,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 223,
      "tokens_per_second_per_gpu": 18827.38,
      "total_tokens": 20150727
    },
    {
      "epoch": 0.044748539179943064,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9793901272940294e-05,
      "loss": 0.5141,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 224,
      "tokens_per_second_per_gpu": 18795.23,
      "total_tokens": 20241670
    },
    {
      "epoch": 0.04494830944413924,
      "grad_norm": 0.6875,
      "learning_rate": 1.9790548246599447e-05,
      "loss": 0.4856,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 225,
      "tokens_per_second_per_gpu": 19141.01,
      "total_tokens": 20335291
    },
    {
      "epoch": 0.04514807970833541,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9787168453273546e-05,
      "loss": 0.5215,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 226,
      "tokens_per_second_per_gpu": 19759.82,
      "total_tokens": 20430693
    },
    {
      "epoch": 0.045347849972531586,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9783761902202814e-05,
      "loss": 0.4853,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 227,
      "tokens_per_second_per_gpu": 19021.35,
      "total_tokens": 20523444
    },
    {
      "epoch": 0.045547620236727766,
      "grad_norm": 0.73046875,
      "learning_rate": 1.9780328602700636e-05,
      "loss": 0.4735,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 228,
      "tokens_per_second_per_gpu": 18011.42,
      "total_tokens": 20611507
    },
    {
      "epoch": 0.04574739050092394,
      "grad_norm": 0.70703125,
      "learning_rate": 1.9776868564153517e-05,
      "loss": 0.4907,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 229,
      "tokens_per_second_per_gpu": 18742.74,
      "total_tokens": 20702133
    },
    {
      "epoch": 0.045947160765120114,
      "grad_norm": 0.828125,
      "learning_rate": 1.9773381796021073e-05,
      "loss": 0.4906,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 230,
      "tokens_per_second_per_gpu": 18136.86,
      "total_tokens": 20790822
    },
    {
      "epoch": 0.04614693102931629,
      "grad_norm": 0.734375,
      "learning_rate": 1.9769868307835996e-05,
      "loss": 0.4978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 231,
      "tokens_per_second_per_gpu": 19228.87,
      "total_tokens": 20883986
    },
    {
      "epoch": 0.04634670129351246,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9766328109204025e-05,
      "loss": 0.5066,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 232,
      "tokens_per_second_per_gpu": 19023.95,
      "total_tokens": 20975561
    },
    {
      "epoch": 0.046546471557708635,
      "grad_norm": 0.75390625,
      "learning_rate": 1.976276120980393e-05,
      "loss": 0.4852,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 233,
      "tokens_per_second_per_gpu": 18056.42,
      "total_tokens": 21064452
    },
    {
      "epoch": 0.04674624182190481,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9759167619387474e-05,
      "loss": 0.4973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 234,
      "tokens_per_second_per_gpu": 18686.51,
      "total_tokens": 21154512
    },
    {
      "epoch": 0.04694601208610098,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9755547347779405e-05,
      "loss": 0.4683,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 235,
      "tokens_per_second_per_gpu": 17307.81,
      "total_tokens": 21240442
    },
    {
      "epoch": 0.047145782350297157,
      "grad_norm": 0.7734375,
      "learning_rate": 1.97519004048774e-05,
      "loss": 0.5236,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 236,
      "tokens_per_second_per_gpu": 19634.84,
      "total_tokens": 21334376
    },
    {
      "epoch": 0.04734555261449333,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9748226800652062e-05,
      "loss": 0.4865,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 237,
      "tokens_per_second_per_gpu": 19229.6,
      "total_tokens": 21428429
    },
    {
      "epoch": 0.047545322878689504,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9744526545146886e-05,
      "loss": 0.458,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 238,
      "tokens_per_second_per_gpu": 17981.99,
      "total_tokens": 21516053
    },
    {
      "epoch": 0.047745093142885685,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9740799648478233e-05,
      "loss": 0.5112,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 239,
      "tokens_per_second_per_gpu": 18306.99,
      "total_tokens": 21606119
    },
    {
      "epoch": 0.04794486340708186,
      "grad_norm": 0.73828125,
      "learning_rate": 1.973704612083529e-05,
      "loss": 0.4657,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 240,
      "tokens_per_second_per_gpu": 17214.23,
      "total_tokens": 21690429
    },
    {
      "epoch": 0.04814463367127803,
      "grad_norm": 0.7421875,
      "learning_rate": 1.973326597248006e-05,
      "loss": 0.5272,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 241,
      "tokens_per_second_per_gpu": 19522.27,
      "total_tokens": 21784804
    },
    {
      "epoch": 0.048344403935474206,
      "grad_norm": 0.75,
      "learning_rate": 1.9729459213747327e-05,
      "loss": 0.4966,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 242,
      "tokens_per_second_per_gpu": 20243.38,
      "total_tokens": 21880113
    },
    {
      "epoch": 0.04854417419967038,
      "grad_norm": 0.75390625,
      "learning_rate": 1.972562585504462e-05,
      "loss": 0.4884,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 243,
      "tokens_per_second_per_gpu": 18001.03,
      "total_tokens": 21967561
    },
    {
      "epoch": 0.048743944463866554,
      "grad_norm": 0.765625,
      "learning_rate": 1.97217659068522e-05,
      "loss": 0.5018,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 244,
      "tokens_per_second_per_gpu": 18874.06,
      "total_tokens": 22058505
    },
    {
      "epoch": 0.04894371472806273,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9717879379723012e-05,
      "loss": 0.5038,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 245,
      "tokens_per_second_per_gpu": 19849.72,
      "total_tokens": 22153686
    },
    {
      "epoch": 0.0491434849922589,
      "grad_norm": 0.75,
      "learning_rate": 1.9713966284282677e-05,
      "loss": 0.5032,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 246,
      "tokens_per_second_per_gpu": 17735.37,
      "total_tokens": 22240331
    },
    {
      "epoch": 0.049343255256455075,
      "grad_norm": 0.74609375,
      "learning_rate": 1.971002663122945e-05,
      "loss": 0.4723,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 247,
      "tokens_per_second_per_gpu": 18622.38,
      "total_tokens": 22332531
    },
    {
      "epoch": 0.04954302552065125,
      "grad_norm": 0.80859375,
      "learning_rate": 1.970606043133419e-05,
      "loss": 0.5114,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 248,
      "tokens_per_second_per_gpu": 18968.68,
      "total_tokens": 22424618
    },
    {
      "epoch": 0.04974279578484742,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9702067695440333e-05,
      "loss": 0.4947,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 249,
      "tokens_per_second_per_gpu": 18936.57,
      "total_tokens": 22516368
    },
    {
      "epoch": 0.0499425660490436,
      "grad_norm": 0.78125,
      "learning_rate": 1.969804843446387e-05,
      "loss": 0.4674,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 250,
      "tokens_per_second_per_gpu": 16634.25,
      "total_tokens": 22598669
    },
    {
      "epoch": 0.05014233631323978,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9694002659393306e-05,
      "loss": 0.4766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 251,
      "tokens_per_second_per_gpu": 19637.52,
      "total_tokens": 22694880
    },
    {
      "epoch": 0.05034210657743595,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9689930381289636e-05,
      "loss": 0.4639,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 252,
      "tokens_per_second_per_gpu": 17429.5,
      "total_tokens": 22780599
    },
    {
      "epoch": 0.050541876841632125,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9685831611286312e-05,
      "loss": 0.4883,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 253,
      "tokens_per_second_per_gpu": 19660.74,
      "total_tokens": 22875342
    },
    {
      "epoch": 0.0507416471058283,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9681706360589218e-05,
      "loss": 0.4952,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 254,
      "tokens_per_second_per_gpu": 18098.56,
      "total_tokens": 22963103
    },
    {
      "epoch": 0.05094141737002447,
      "grad_norm": 0.78125,
      "learning_rate": 1.9677554640476625e-05,
      "loss": 0.4999,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 255,
      "tokens_per_second_per_gpu": 18945.16,
      "total_tokens": 23055166
    },
    {
      "epoch": 0.051141187634220646,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9673376462299186e-05,
      "loss": 0.4749,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 256,
      "tokens_per_second_per_gpu": 17030.26,
      "total_tokens": 23138552
    },
    {
      "epoch": 0.05134095789841682,
      "grad_norm": 0.7578125,
      "learning_rate": 1.966917183747987e-05,
      "loss": 0.5124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 257,
      "tokens_per_second_per_gpu": 19023.64,
      "total_tokens": 23231758
    },
    {
      "epoch": 0.05154072816261299,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9664940777513973e-05,
      "loss": 0.5046,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 258,
      "tokens_per_second_per_gpu": 20036.35,
      "total_tokens": 23327833
    },
    {
      "epoch": 0.05174049842680917,
      "grad_norm": 0.71875,
      "learning_rate": 1.9660683293969042e-05,
      "loss": 0.4855,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 259,
      "tokens_per_second_per_gpu": 19402.85,
      "total_tokens": 23420039
    },
    {
      "epoch": 0.05194026869100534,
      "grad_norm": 0.8046875,
      "learning_rate": 1.965639939848488e-05,
      "loss": 0.4981,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 260,
      "tokens_per_second_per_gpu": 18497.03,
      "total_tokens": 23509432
    },
    {
      "epoch": 0.05214003895520152,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9652089102773487e-05,
      "loss": 0.5133,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 261,
      "tokens_per_second_per_gpu": 18415.12,
      "total_tokens": 23598540
    },
    {
      "epoch": 0.052339809219397695,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9647752418619055e-05,
      "loss": 0.4729,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 262,
      "tokens_per_second_per_gpu": 18349.6,
      "total_tokens": 23686989
    },
    {
      "epoch": 0.05253957948359387,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9643389357877907e-05,
      "loss": 0.4796,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 263,
      "tokens_per_second_per_gpu": 18309.69,
      "total_tokens": 23776963
    },
    {
      "epoch": 0.05273934974779004,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9638999932478487e-05,
      "loss": 0.5012,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 264,
      "tokens_per_second_per_gpu": 18802.69,
      "total_tokens": 23869316
    },
    {
      "epoch": 0.05293912001198622,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9634584154421316e-05,
      "loss": 0.4562,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 265,
      "tokens_per_second_per_gpu": 17157.57,
      "total_tokens": 23954773
    },
    {
      "epoch": 0.05313889027618239,
      "grad_norm": 0.76171875,
      "learning_rate": 1.963014203577896e-05,
      "loss": 0.4774,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 266,
      "tokens_per_second_per_gpu": 18691.75,
      "total_tokens": 24045859
    },
    {
      "epoch": 0.053338660540378564,
      "grad_norm": 0.75390625,
      "learning_rate": 1.962567358869601e-05,
      "loss": 0.4978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 267,
      "tokens_per_second_per_gpu": 19469.37,
      "total_tokens": 24139870
    },
    {
      "epoch": 0.05353843080457474,
      "grad_norm": 0.75390625,
      "learning_rate": 1.962117882538902e-05,
      "loss": 0.5154,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 268,
      "tokens_per_second_per_gpu": 20145.18,
      "total_tokens": 24236114
    },
    {
      "epoch": 0.05373820106877091,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9616657758146503e-05,
      "loss": 0.4689,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 269,
      "tokens_per_second_per_gpu": 18774.71,
      "total_tokens": 24327150
    },
    {
      "epoch": 0.053937971332967086,
      "grad_norm": 0.7421875,
      "learning_rate": 1.961211039932889e-05,
      "loss": 0.423,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 270,
      "tokens_per_second_per_gpu": 17083.67,
      "total_tokens": 24410190
    },
    {
      "epoch": 0.05413774159716326,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9607536761368484e-05,
      "loss": 0.4979,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 271,
      "tokens_per_second_per_gpu": 19270.86,
      "total_tokens": 24502320
    },
    {
      "epoch": 0.05433751186135943,
      "grad_norm": 2.65625,
      "learning_rate": 1.9602936856769432e-05,
      "loss": 0.4812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 272,
      "tokens_per_second_per_gpu": 17507.88,
      "total_tokens": 24587955
    },
    {
      "epoch": 0.054537282125555614,
      "grad_norm": 0.76953125,
      "learning_rate": 1.95983106981077e-05,
      "loss": 0.4809,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 273,
      "tokens_per_second_per_gpu": 19901.39,
      "total_tokens": 24684174
    },
    {
      "epoch": 0.05473705238975179,
      "grad_norm": 0.734375,
      "learning_rate": 1.9593658298031033e-05,
      "loss": 0.4715,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 274,
      "tokens_per_second_per_gpu": 18903.68,
      "total_tokens": 24777474
    },
    {
      "epoch": 0.05493682265394796,
      "grad_norm": 0.734375,
      "learning_rate": 1.9588979669258913e-05,
      "loss": 0.4942,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 275,
      "tokens_per_second_per_gpu": 20187.45,
      "total_tokens": 24876158
    },
    {
      "epoch": 0.055136592918144135,
      "grad_norm": 0.73046875,
      "learning_rate": 1.958427482458253e-05,
      "loss": 0.483,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 276,
      "tokens_per_second_per_gpu": 19364.38,
      "total_tokens": 24970279
    },
    {
      "epoch": 0.05533636318234031,
      "grad_norm": 0.7109375,
      "learning_rate": 1.957954377686475e-05,
      "loss": 0.456,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 277,
      "tokens_per_second_per_gpu": 18886.2,
      "total_tokens": 25062533
    },
    {
      "epoch": 0.05553613344653648,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9574786539040078e-05,
      "loss": 0.4631,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 278,
      "tokens_per_second_per_gpu": 17547.39,
      "total_tokens": 25147946
    },
    {
      "epoch": 0.05573590371073266,
      "grad_norm": 0.73046875,
      "learning_rate": 1.9570003124114622e-05,
      "loss": 0.4726,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 279,
      "tokens_per_second_per_gpu": 19175.27,
      "total_tokens": 25242866
    },
    {
      "epoch": 0.05593567397492883,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9565193545166052e-05,
      "loss": 0.4513,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 280,
      "tokens_per_second_per_gpu": 17616.94,
      "total_tokens": 25327149
    },
    {
      "epoch": 0.056135444239125004,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9560357815343577e-05,
      "loss": 0.4831,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 281,
      "tokens_per_second_per_gpu": 18642.59,
      "total_tokens": 25417750
    },
    {
      "epoch": 0.05633521450332118,
      "grad_norm": 2.4375,
      "learning_rate": 1.95554959478679e-05,
      "loss": 0.4508,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 282,
      "tokens_per_second_per_gpu": 17213.13,
      "total_tokens": 25501577
    },
    {
      "epoch": 0.05653498476751735,
      "grad_norm": 0.734375,
      "learning_rate": 1.955060795603117e-05,
      "loss": 0.4924,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 283,
      "tokens_per_second_per_gpu": 19673.0,
      "total_tokens": 25596861
    },
    {
      "epoch": 0.05673475503171353,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9545693853196985e-05,
      "loss": 0.4942,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 284,
      "tokens_per_second_per_gpu": 18701.89,
      "total_tokens": 25687800
    },
    {
      "epoch": 0.056934525295909706,
      "grad_norm": 0.75390625,
      "learning_rate": 1.95407536528003e-05,
      "loss": 0.4707,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 285,
      "tokens_per_second_per_gpu": 17044.8,
      "total_tokens": 25772206
    },
    {
      "epoch": 0.05713429556010588,
      "grad_norm": 0.734375,
      "learning_rate": 1.9535787368347444e-05,
      "loss": 0.474,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 286,
      "tokens_per_second_per_gpu": 18799.26,
      "total_tokens": 25863849
    },
    {
      "epoch": 0.057334065824302054,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9530795013416046e-05,
      "loss": 0.4657,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 287,
      "tokens_per_second_per_gpu": 18549.96,
      "total_tokens": 25954200
    },
    {
      "epoch": 0.05753383608849823,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9525776601655014e-05,
      "loss": 0.5117,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 288,
      "tokens_per_second_per_gpu": 19832.19,
      "total_tokens": 26048948
    },
    {
      "epoch": 0.0577336063526944,
      "grad_norm": 0.70703125,
      "learning_rate": 1.9520732146784493e-05,
      "loss": 0.4838,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 289,
      "tokens_per_second_per_gpu": 20344.69,
      "total_tokens": 26147269
    },
    {
      "epoch": 0.057933376616890575,
      "grad_norm": 0.73828125,
      "learning_rate": 1.951566166259583e-05,
      "loss": 0.4858,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 290,
      "tokens_per_second_per_gpu": 19096.01,
      "total_tokens": 26241318
    },
    {
      "epoch": 0.05813314688108675,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.5071,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 291,
      "tokens_per_second_per_gpu": 20222.41,
      "total_tokens": 26339613
    },
    {
      "epoch": 0.05833291714528292,
      "grad_norm": 0.7578125,
      "learning_rate": 1.950544266178525e-05,
      "loss": 0.4787,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 292,
      "tokens_per_second_per_gpu": 18311.94,
      "total_tokens": 26427549
    },
    {
      "epoch": 0.058532687409479096,
      "grad_norm": 0.75,
      "learning_rate": 1.9500294173101687e-05,
      "loss": 0.4364,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 293,
      "tokens_per_second_per_gpu": 18195.69,
      "total_tokens": 26514808
    },
    {
      "epoch": 0.05873245767367527,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9495119710976628e-05,
      "loss": 0.4819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 294,
      "tokens_per_second_per_gpu": 19433.08,
      "total_tokens": 26607406
    },
    {
      "epoch": 0.05893222793787145,
      "grad_norm": 0.734375,
      "learning_rate": 1.9489919289556844e-05,
      "loss": 0.4871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 295,
      "tokens_per_second_per_gpu": 19459.06,
      "total_tokens": 26701540
    },
    {
      "epoch": 0.059131998202067625,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9484692923060095e-05,
      "loss": 0.485,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 296,
      "tokens_per_second_per_gpu": 17767.29,
      "total_tokens": 26787153
    },
    {
      "epoch": 0.0593317684662638,
      "grad_norm": 0.734375,
      "learning_rate": 1.947944062577507e-05,
      "loss": 0.4922,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 297,
      "tokens_per_second_per_gpu": 19618.74,
      "total_tokens": 26881583
    },
    {
      "epoch": 0.05953153873045997,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9474162412061342e-05,
      "loss": 0.4757,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 298,
      "tokens_per_second_per_gpu": 17193.1,
      "total_tokens": 26964699
    },
    {
      "epoch": 0.059731308994656146,
      "grad_norm": 0.796875,
      "learning_rate": 1.946885829634935e-05,
      "loss": 0.4934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 299,
      "tokens_per_second_per_gpu": 17980.09,
      "total_tokens": 27053346
    },
    {
      "epoch": 0.05993107925885232,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9463528293140347e-05,
      "loss": 0.4447,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 300,
      "tokens_per_second_per_gpu": 18676.8,
      "total_tokens": 27142719
    },
    {
      "epoch": 0.060130849523048494,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9458172417006347e-05,
      "loss": 0.4637,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 301,
      "tokens_per_second_per_gpu": 18315.56,
      "total_tokens": 27233327
    },
    {
      "epoch": 0.06033061978724467,
      "grad_norm": 0.765625,
      "learning_rate": 1.9452790682590124e-05,
      "loss": 0.4507,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 302,
      "tokens_per_second_per_gpu": 18332.18,
      "total_tokens": 27321255
    },
    {
      "epoch": 0.06053039005144084,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9447383104605126e-05,
      "loss": 0.4937,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 303,
      "tokens_per_second_per_gpu": 19486.97,
      "total_tokens": 27415583
    },
    {
      "epoch": 0.060730160315637015,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9441949697835468e-05,
      "loss": 0.4647,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 304,
      "tokens_per_second_per_gpu": 19012.74,
      "total_tokens": 27506679
    },
    {
      "epoch": 0.06092993057983319,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9436490477135877e-05,
      "loss": 0.4387,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 305,
      "tokens_per_second_per_gpu": 17254.66,
      "total_tokens": 27591645
    },
    {
      "epoch": 0.06112970084402937,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9431005457431654e-05,
      "loss": 0.4757,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 306,
      "tokens_per_second_per_gpu": 19644.92,
      "total_tokens": 27685299
    },
    {
      "epoch": 0.06132947110822554,
      "grad_norm": 0.73046875,
      "learning_rate": 1.9425494653718632e-05,
      "loss": 0.4487,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 307,
      "tokens_per_second_per_gpu": 18733.86,
      "total_tokens": 27776771
    },
    {
      "epoch": 0.06152924137242172,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9419958081063136e-05,
      "loss": 0.4754,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 308,
      "tokens_per_second_per_gpu": 19040.6,
      "total_tokens": 27868858
    },
    {
      "epoch": 0.06172901163661789,
      "grad_norm": 0.76171875,
      "learning_rate": 1.941439575460195e-05,
      "loss": 0.4712,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 309,
      "tokens_per_second_per_gpu": 17499.37,
      "total_tokens": 27955552
    },
    {
      "epoch": 0.061928781900814064,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9408807689542257e-05,
      "loss": 0.4162,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 310,
      "tokens_per_second_per_gpu": 17170.67,
      "total_tokens": 28039699
    },
    {
      "epoch": 0.06212855216501024,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9403193901161614e-05,
      "loss": 0.4781,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 311,
      "tokens_per_second_per_gpu": 18837.09,
      "total_tokens": 28130081
    },
    {
      "epoch": 0.06232832242920641,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9397554404807904e-05,
      "loss": 0.5044,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 312,
      "tokens_per_second_per_gpu": 19388.56,
      "total_tokens": 28225060
    },
    {
      "epoch": 0.06252809269340259,
      "grad_norm": 0.78125,
      "learning_rate": 1.93918892158993e-05,
      "loss": 0.4719,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 313,
      "tokens_per_second_per_gpu": 18540.31,
      "total_tokens": 28318026
    },
    {
      "epoch": 0.06272786295759876,
      "grad_norm": 0.78125,
      "learning_rate": 1.9386198349924207e-05,
      "loss": 0.4547,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 314,
      "tokens_per_second_per_gpu": 18516.75,
      "total_tokens": 28407964
    },
    {
      "epoch": 0.06292763322179494,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9380481822441236e-05,
      "loss": 0.4402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 315,
      "tokens_per_second_per_gpu": 16812.28,
      "total_tokens": 28490050
    },
    {
      "epoch": 0.06312740348599111,
      "grad_norm": 0.734375,
      "learning_rate": 1.9374739649079155e-05,
      "loss": 0.477,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 316,
      "tokens_per_second_per_gpu": 19232.85,
      "total_tokens": 28583112
    },
    {
      "epoch": 0.06332717375018729,
      "grad_norm": 0.734375,
      "learning_rate": 1.9368971845536844e-05,
      "loss": 0.4835,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 317,
      "tokens_per_second_per_gpu": 18448.3,
      "total_tokens": 28672763
    },
    {
      "epoch": 0.06352694401438345,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9363178427583266e-05,
      "loss": 0.4733,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 318,
      "tokens_per_second_per_gpu": 19372.33,
      "total_tokens": 28766632
    },
    {
      "epoch": 0.06372671427857964,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9357359411057398e-05,
      "loss": 0.4881,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 319,
      "tokens_per_second_per_gpu": 19192.57,
      "total_tokens": 28859019
    },
    {
      "epoch": 0.0639264845427758,
      "grad_norm": 0.7890625,
      "learning_rate": 1.935151481186821e-05,
      "loss": 0.4587,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 320,
      "tokens_per_second_per_gpu": 19354.2,
      "total_tokens": 28952689
    },
    {
      "epoch": 0.06412625480697198,
      "grad_norm": 0.8046875,
      "learning_rate": 1.934564464599461e-05,
      "loss": 0.422,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 321,
      "tokens_per_second_per_gpu": 15869.5,
      "total_tokens": 29030761
    },
    {
      "epoch": 0.06432602507116815,
      "grad_norm": 0.734375,
      "learning_rate": 1.933974892948541e-05,
      "loss": 0.459,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 322,
      "tokens_per_second_per_gpu": 18816.63,
      "total_tokens": 29120675
    },
    {
      "epoch": 0.06452579533536433,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9333827678459277e-05,
      "loss": 0.461,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 323,
      "tokens_per_second_per_gpu": 17414.46,
      "total_tokens": 29205333
    },
    {
      "epoch": 0.06472556559956051,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9327880909104684e-05,
      "loss": 0.4593,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 324,
      "tokens_per_second_per_gpu": 18172.44,
      "total_tokens": 29294375
    },
    {
      "epoch": 0.06492533586375668,
      "grad_norm": 0.921875,
      "learning_rate": 1.9321908637679868e-05,
      "loss": 0.4502,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 325,
      "tokens_per_second_per_gpu": 19942.95,
      "total_tokens": 29390943
    },
    {
      "epoch": 0.06512510612795286,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9315910880512792e-05,
      "loss": 0.4997,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 326,
      "tokens_per_second_per_gpu": 20051.53,
      "total_tokens": 29487963
    },
    {
      "epoch": 0.06532487639214903,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9309887654001095e-05,
      "loss": 0.4768,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 327,
      "tokens_per_second_per_gpu": 21222.39,
      "total_tokens": 29589170
    },
    {
      "epoch": 0.0655246466563452,
      "grad_norm": 0.7109375,
      "learning_rate": 1.930383897461205e-05,
      "loss": 0.4843,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 328,
      "tokens_per_second_per_gpu": 18406.81,
      "total_tokens": 29680577
    },
    {
      "epoch": 0.06572441692054137,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9297764858882516e-05,
      "loss": 0.468,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 329,
      "tokens_per_second_per_gpu": 18344.59,
      "total_tokens": 29769698
    },
    {
      "epoch": 0.06592418718473755,
      "grad_norm": 0.75,
      "learning_rate": 1.9291665323418888e-05,
      "loss": 0.4523,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 330,
      "tokens_per_second_per_gpu": 17961.3,
      "total_tokens": 29857378
    },
    {
      "epoch": 0.06612395744893372,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9285540384897073e-05,
      "loss": 0.4355,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 331,
      "tokens_per_second_per_gpu": 18410.31,
      "total_tokens": 29946306
    },
    {
      "epoch": 0.0663237277131299,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9279390060062408e-05,
      "loss": 0.4553,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 332,
      "tokens_per_second_per_gpu": 18853.23,
      "total_tokens": 30038267
    },
    {
      "epoch": 0.06652349797732607,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9273214365729655e-05,
      "loss": 0.4574,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 333,
      "tokens_per_second_per_gpu": 18255.06,
      "total_tokens": 30127625
    },
    {
      "epoch": 0.06672326824152225,
      "grad_norm": 0.765625,
      "learning_rate": 1.9267013318782922e-05,
      "loss": 0.4485,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 334,
      "tokens_per_second_per_gpu": 18637.62,
      "total_tokens": 30218601
    },
    {
      "epoch": 0.06692303850571843,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9260786936175635e-05,
      "loss": 0.4535,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 335,
      "tokens_per_second_per_gpu": 18318.21,
      "total_tokens": 30308631
    },
    {
      "epoch": 0.0671228087699146,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9254535234930486e-05,
      "loss": 0.465,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 336,
      "tokens_per_second_per_gpu": 19810.49,
      "total_tokens": 30403625
    },
    {
      "epoch": 0.06732257903411078,
      "grad_norm": 0.765625,
      "learning_rate": 1.924825823213939e-05,
      "loss": 0.4568,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 337,
      "tokens_per_second_per_gpu": 19597.81,
      "total_tokens": 30498641
    },
    {
      "epoch": 0.06752234929830694,
      "grad_norm": 0.828125,
      "learning_rate": 1.924195594496343e-05,
      "loss": 0.427,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 338,
      "tokens_per_second_per_gpu": 16110.94,
      "total_tokens": 30578470
    },
    {
      "epoch": 0.06772211956250312,
      "grad_norm": 0.828125,
      "learning_rate": 1.923562839063282e-05,
      "loss": 0.4548,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 339,
      "tokens_per_second_per_gpu": 17550.34,
      "total_tokens": 30665187
    },
    {
      "epoch": 0.06792188982669929,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9229275586446853e-05,
      "loss": 0.4607,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 340,
      "tokens_per_second_per_gpu": 18631.78,
      "total_tokens": 30755522
    },
    {
      "epoch": 0.06812166009089547,
      "grad_norm": 0.765625,
      "learning_rate": 1.922289754977385e-05,
      "loss": 0.4647,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 341,
      "tokens_per_second_per_gpu": 18218.75,
      "total_tokens": 30844380
    },
    {
      "epoch": 0.06832143035509164,
      "grad_norm": 0.75390625,
      "learning_rate": 1.921649429805112e-05,
      "loss": 0.4631,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 342,
      "tokens_per_second_per_gpu": 19155.47,
      "total_tokens": 30937699
    },
    {
      "epoch": 0.06852120061928782,
      "grad_norm": 0.71875,
      "learning_rate": 1.9210065848784915e-05,
      "loss": 0.4902,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 343,
      "tokens_per_second_per_gpu": 19912.91,
      "total_tokens": 31034575
    },
    {
      "epoch": 0.06872097088348399,
      "grad_norm": 0.72265625,
      "learning_rate": 1.9203612219550357e-05,
      "loss": 0.454,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 344,
      "tokens_per_second_per_gpu": 19822.08,
      "total_tokens": 31132179
    },
    {
      "epoch": 0.06892074114768017,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9197133427991437e-05,
      "loss": 0.4589,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 345,
      "tokens_per_second_per_gpu": 17394.12,
      "total_tokens": 31217511
    },
    {
      "epoch": 0.06912051141187635,
      "grad_norm": 0.74609375,
      "learning_rate": 1.919062949182091e-05,
      "loss": 0.4694,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 346,
      "tokens_per_second_per_gpu": 18809.77,
      "total_tokens": 31310363
    },
    {
      "epoch": 0.06932028167607251,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9184100428820304e-05,
      "loss": 0.4512,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 347,
      "tokens_per_second_per_gpu": 20546.18,
      "total_tokens": 31408105
    },
    {
      "epoch": 0.0695200519402687,
      "grad_norm": 0.765625,
      "learning_rate": 1.9177546256839814e-05,
      "loss": 0.4194,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 348,
      "tokens_per_second_per_gpu": 16080.37,
      "total_tokens": 31486678
    },
    {
      "epoch": 0.06971982220446486,
      "grad_norm": 0.84375,
      "learning_rate": 1.9170966993798302e-05,
      "loss": 0.4689,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 349,
      "tokens_per_second_per_gpu": 18876.88,
      "total_tokens": 31578085
    },
    {
      "epoch": 0.06991959246866104,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9164362657683223e-05,
      "loss": 0.474,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 350,
      "tokens_per_second_per_gpu": 19889.06,
      "total_tokens": 31672429
    },
    {
      "epoch": 0.07011936273285721,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9157733266550577e-05,
      "loss": 0.4612,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 351,
      "tokens_per_second_per_gpu": 18279.43,
      "total_tokens": 31762205
    },
    {
      "epoch": 0.07031913299705339,
      "grad_norm": 0.71875,
      "learning_rate": 1.9151078838524867e-05,
      "loss": 0.412,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 352,
      "tokens_per_second_per_gpu": 17851.38,
      "total_tokens": 31849565
    },
    {
      "epoch": 0.07051890326124956,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9144399391799043e-05,
      "loss": 0.4253,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 353,
      "tokens_per_second_per_gpu": 16907.45,
      "total_tokens": 31931257
    },
    {
      "epoch": 0.07071867352544574,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9137694944634465e-05,
      "loss": 0.4292,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 354,
      "tokens_per_second_per_gpu": 18580.92,
      "total_tokens": 32020676
    },
    {
      "epoch": 0.0709184437896419,
      "grad_norm": 0.74609375,
      "learning_rate": 1.913096551536083e-05,
      "loss": 0.4686,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 355,
      "tokens_per_second_per_gpu": 17811.59,
      "total_tokens": 32107600
    },
    {
      "epoch": 0.07111821405383809,
      "grad_norm": 0.765625,
      "learning_rate": 1.9124211122376138e-05,
      "loss": 0.4543,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 356,
      "tokens_per_second_per_gpu": 19592.69,
      "total_tokens": 32202315
    },
    {
      "epoch": 0.07131798431803427,
      "grad_norm": 0.7421875,
      "learning_rate": 1.911743178414665e-05,
      "loss": 0.4185,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 357,
      "tokens_per_second_per_gpu": 17200.24,
      "total_tokens": 32286805
    },
    {
      "epoch": 0.07151775458223043,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9110627519206806e-05,
      "loss": 0.4673,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 358,
      "tokens_per_second_per_gpu": 19107.01,
      "total_tokens": 32379697
    },
    {
      "epoch": 0.07171752484642661,
      "grad_norm": 0.765625,
      "learning_rate": 1.9103798346159214e-05,
      "loss": 0.4562,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 359,
      "tokens_per_second_per_gpu": 18615.54,
      "total_tokens": 32470376
    },
    {
      "epoch": 0.07191729511062278,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9096944283674574e-05,
      "loss": 0.4343,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 360,
      "tokens_per_second_per_gpu": 17594.15,
      "total_tokens": 32557117
    },
    {
      "epoch": 0.07211706537481896,
      "grad_norm": 0.7890625,
      "learning_rate": 1.909006535049163e-05,
      "loss": 0.463,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 361,
      "tokens_per_second_per_gpu": 17704.48,
      "total_tokens": 32642986
    },
    {
      "epoch": 0.07231683563901513,
      "grad_norm": 0.734375,
      "learning_rate": 1.9083161565417115e-05,
      "loss": 0.4111,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 362,
      "tokens_per_second_per_gpu": 17251.99,
      "total_tokens": 32726348
    },
    {
      "epoch": 0.07251660590321131,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9076232947325724e-05,
      "loss": 0.4851,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 363,
      "tokens_per_second_per_gpu": 19515.12,
      "total_tokens": 32821377
    },
    {
      "epoch": 0.07271637616740748,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9069279515160027e-05,
      "loss": 0.4564,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 364,
      "tokens_per_second_per_gpu": 18735.82,
      "total_tokens": 32912104
    },
    {
      "epoch": 0.07291614643160366,
      "grad_norm": 0.78125,
      "learning_rate": 1.9062301287930448e-05,
      "loss": 0.4724,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 365,
      "tokens_per_second_per_gpu": 18227.05,
      "total_tokens": 33002672
    },
    {
      "epoch": 0.07311591669579982,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9055298284715192e-05,
      "loss": 0.4315,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 366,
      "tokens_per_second_per_gpu": 18151.94,
      "total_tokens": 33092420
    },
    {
      "epoch": 0.073315686959996,
      "grad_norm": 0.765625,
      "learning_rate": 1.9048270524660197e-05,
      "loss": 0.4147,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 367,
      "tokens_per_second_per_gpu": 17551.9,
      "total_tokens": 33178233
    },
    {
      "epoch": 0.07351545722419218,
      "grad_norm": 0.78125,
      "learning_rate": 1.9041218026979096e-05,
      "loss": 0.4409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 368,
      "tokens_per_second_per_gpu": 18533.84,
      "total_tokens": 33268228
    },
    {
      "epoch": 0.07371522748838835,
      "grad_norm": 0.77734375,
      "learning_rate": 1.903414081095315e-05,
      "loss": 0.4448,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 369,
      "tokens_per_second_per_gpu": 18382.29,
      "total_tokens": 33357686
    },
    {
      "epoch": 0.07391499775258453,
      "grad_norm": 0.7578125,
      "learning_rate": 1.902703889593119e-05,
      "loss": 0.4547,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 370,
      "tokens_per_second_per_gpu": 20030.89,
      "total_tokens": 33452722
    },
    {
      "epoch": 0.0741147680167807,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9019912301329593e-05,
      "loss": 0.4662,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 371,
      "tokens_per_second_per_gpu": 18866.38,
      "total_tokens": 33543831
    },
    {
      "epoch": 0.07431453828097688,
      "grad_norm": 0.7578125,
      "learning_rate": 1.901276104663218e-05,
      "loss": 0.4559,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 372,
      "tokens_per_second_per_gpu": 18398.79,
      "total_tokens": 33633472
    },
    {
      "epoch": 0.07451430854517305,
      "grad_norm": 0.703125,
      "learning_rate": 1.9005585151390224e-05,
      "loss": 0.4437,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 373,
      "tokens_per_second_per_gpu": 19233.21,
      "total_tokens": 33726158
    },
    {
      "epoch": 0.07471407880936923,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8998384635222337e-05,
      "loss": 0.4612,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 374,
      "tokens_per_second_per_gpu": 19099.25,
      "total_tokens": 33819103
    },
    {
      "epoch": 0.0749138490735654,
      "grad_norm": 0.71875,
      "learning_rate": 1.8991159517814463e-05,
      "loss": 0.4493,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 375,
      "tokens_per_second_per_gpu": 19527.58,
      "total_tokens": 33913338
    },
    {
      "epoch": 0.07511361933776158,
      "grad_norm": 0.6953125,
      "learning_rate": 1.898390981891979e-05,
      "loss": 0.4354,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 376,
      "tokens_per_second_per_gpu": 19007.22,
      "total_tokens": 34005425
    },
    {
      "epoch": 0.07531338960195774,
      "grad_norm": 0.765625,
      "learning_rate": 1.897663555835872e-05,
      "loss": 0.4728,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 377,
      "tokens_per_second_per_gpu": 19804.74,
      "total_tokens": 34101132
    },
    {
      "epoch": 0.07551315986615392,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8969336756018813e-05,
      "loss": 0.4598,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 378,
      "tokens_per_second_per_gpu": 17862.48,
      "total_tokens": 34187810
    },
    {
      "epoch": 0.0757129301303501,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8962013431854705e-05,
      "loss": 0.468,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 379,
      "tokens_per_second_per_gpu": 18205.99,
      "total_tokens": 34278318
    },
    {
      "epoch": 0.07591270039454627,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8954665605888086e-05,
      "loss": 0.4753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 380,
      "tokens_per_second_per_gpu": 19804.36,
      "total_tokens": 34376644
    },
    {
      "epoch": 0.07611247065874245,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8947293298207637e-05,
      "loss": 0.4226,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 381,
      "tokens_per_second_per_gpu": 16601.79,
      "total_tokens": 34458692
    },
    {
      "epoch": 0.07631224092293862,
      "grad_norm": 0.75,
      "learning_rate": 1.893989652896896e-05,
      "loss": 0.4754,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 382,
      "tokens_per_second_per_gpu": 19552.66,
      "total_tokens": 34553520
    },
    {
      "epoch": 0.0765120111871348,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8932475318394542e-05,
      "loss": 0.4443,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 383,
      "tokens_per_second_per_gpu": 18296.67,
      "total_tokens": 34640570
    },
    {
      "epoch": 0.07671178145133097,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8925029686773692e-05,
      "loss": 0.4478,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 384,
      "tokens_per_second_per_gpu": 19332.33,
      "total_tokens": 34734118
    },
    {
      "epoch": 0.07691155171552715,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8917559654462474e-05,
      "loss": 0.4662,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 385,
      "tokens_per_second_per_gpu": 19702.06,
      "total_tokens": 34830220
    },
    {
      "epoch": 0.07711132197972331,
      "grad_norm": 0.7734375,
      "learning_rate": 1.891006524188368e-05,
      "loss": 0.4422,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 386,
      "tokens_per_second_per_gpu": 18099.29,
      "total_tokens": 34919409
    },
    {
      "epoch": 0.0773110922439195,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8902546469526744e-05,
      "loss": 0.4603,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 387,
      "tokens_per_second_per_gpu": 20118.12,
      "total_tokens": 35015775
    },
    {
      "epoch": 0.07751086250811566,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8895003357947707e-05,
      "loss": 0.4307,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 388,
      "tokens_per_second_per_gpu": 18431.61,
      "total_tokens": 35106349
    },
    {
      "epoch": 0.07771063277231184,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8887435927769137e-05,
      "loss": 0.445,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 389,
      "tokens_per_second_per_gpu": 19391.85,
      "total_tokens": 35200487
    },
    {
      "epoch": 0.07791040303650802,
      "grad_norm": 0.859375,
      "learning_rate": 1.887984419968011e-05,
      "loss": 0.4551,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 390,
      "tokens_per_second_per_gpu": 19486.47,
      "total_tokens": 35294028
    },
    {
      "epoch": 0.07811017330070419,
      "grad_norm": 0.7109375,
      "learning_rate": 1.887222819443612e-05,
      "loss": 0.4177,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 391,
      "tokens_per_second_per_gpu": 18114.1,
      "total_tokens": 35381758
    },
    {
      "epoch": 0.07830994356490037,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8864587932859028e-05,
      "loss": 0.4316,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 392,
      "tokens_per_second_per_gpu": 18680.53,
      "total_tokens": 35473265
    },
    {
      "epoch": 0.07850971382909654,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8856923435837024e-05,
      "loss": 0.4708,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 393,
      "tokens_per_second_per_gpu": 19966.88,
      "total_tokens": 35569885
    },
    {
      "epoch": 0.07870948409329272,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8849234724324543e-05,
      "loss": 0.4401,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 394,
      "tokens_per_second_per_gpu": 20041.17,
      "total_tokens": 35664805
    },
    {
      "epoch": 0.07890925435748888,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8841521819342237e-05,
      "loss": 0.452,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 395,
      "tokens_per_second_per_gpu": 17064.04,
      "total_tokens": 35749944
    },
    {
      "epoch": 0.07910902462168506,
      "grad_norm": 0.7421875,
      "learning_rate": 1.883378474197689e-05,
      "loss": 0.465,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 396,
      "tokens_per_second_per_gpu": 18294.82,
      "total_tokens": 35839936
    },
    {
      "epoch": 0.07930879488588123,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8826023513381372e-05,
      "loss": 0.4426,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 397,
      "tokens_per_second_per_gpu": 18614.5,
      "total_tokens": 35930456
    },
    {
      "epoch": 0.07950856515007741,
      "grad_norm": 0.72265625,
      "learning_rate": 1.881823815477459e-05,
      "loss": 0.4324,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 398,
      "tokens_per_second_per_gpu": 18927.0,
      "total_tokens": 36022487
    },
    {
      "epoch": 0.07970833541427358,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8810428687441415e-05,
      "loss": 0.4436,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 399,
      "tokens_per_second_per_gpu": 18255.88,
      "total_tokens": 36110406
    },
    {
      "epoch": 0.07990810567846976,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8802595132732636e-05,
      "loss": 0.4409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 400,
      "tokens_per_second_per_gpu": 18284.79,
      "total_tokens": 36199208
    },
    {
      "epoch": 0.08010787594266594,
      "grad_norm": 0.75390625,
      "learning_rate": 1.879473751206489e-05,
      "loss": 0.4484,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 401,
      "tokens_per_second_per_gpu": 17659.79,
      "total_tokens": 36285159
    },
    {
      "epoch": 0.08030764620686211,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8786855846920615e-05,
      "loss": 0.4661,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 402,
      "tokens_per_second_per_gpu": 18452.81,
      "total_tokens": 36375461
    },
    {
      "epoch": 0.08050741647105829,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8778950158847976e-05,
      "loss": 0.4797,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 403,
      "tokens_per_second_per_gpu": 20207.15,
      "total_tokens": 36473161
    },
    {
      "epoch": 0.08070718673525445,
      "grad_norm": 0.765625,
      "learning_rate": 1.877102046946083e-05,
      "loss": 0.4557,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 404,
      "tokens_per_second_per_gpu": 19675.68,
      "total_tokens": 36568183
    },
    {
      "epoch": 0.08090695699945064,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8763066800438638e-05,
      "loss": 0.4212,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 405,
      "tokens_per_second_per_gpu": 18655.17,
      "total_tokens": 36657416
    },
    {
      "epoch": 0.0811067272636468,
      "grad_norm": 0.78125,
      "learning_rate": 1.875508917352643e-05,
      "loss": 0.427,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 406,
      "tokens_per_second_per_gpu": 17290.24,
      "total_tokens": 36742759
    },
    {
      "epoch": 0.08130649752784298,
      "grad_norm": 0.78125,
      "learning_rate": 1.8747087610534735e-05,
      "loss": 0.4312,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 407,
      "tokens_per_second_per_gpu": 18862.53,
      "total_tokens": 36833543
    },
    {
      "epoch": 0.08150626779203915,
      "grad_norm": 0.734375,
      "learning_rate": 1.8739062133339518e-05,
      "loss": 0.4153,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 408,
      "tokens_per_second_per_gpu": 17516.1,
      "total_tokens": 36920264
    },
    {
      "epoch": 0.08170603805623533,
      "grad_norm": 0.734375,
      "learning_rate": 1.8731012763882132e-05,
      "loss": 0.4716,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 409,
      "tokens_per_second_per_gpu": 19903.17,
      "total_tokens": 37017512
    },
    {
      "epoch": 0.0819058083204315,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8722939524169243e-05,
      "loss": 0.4455,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 410,
      "tokens_per_second_per_gpu": 19227.69,
      "total_tokens": 37111193
    },
    {
      "epoch": 0.08210557858462768,
      "grad_norm": 0.71875,
      "learning_rate": 1.8714842436272774e-05,
      "loss": 0.4673,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 411,
      "tokens_per_second_per_gpu": 19660.48,
      "total_tokens": 37206493
    },
    {
      "epoch": 0.08230534884882386,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8706721522329862e-05,
      "loss": 0.4779,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 412,
      "tokens_per_second_per_gpu": 18964.49,
      "total_tokens": 37297746
    },
    {
      "epoch": 0.08250511911302003,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8698576804542775e-05,
      "loss": 0.464,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 413,
      "tokens_per_second_per_gpu": 19165.29,
      "total_tokens": 37391906
    },
    {
      "epoch": 0.0827048893772162,
      "grad_norm": 0.8125,
      "learning_rate": 1.8690408305178858e-05,
      "loss": 0.4684,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 414,
      "tokens_per_second_per_gpu": 18010.82,
      "total_tokens": 37479544
    },
    {
      "epoch": 0.08290465964141237,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8682216046570477e-05,
      "loss": 0.4237,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 415,
      "tokens_per_second_per_gpu": 17127.05,
      "total_tokens": 37562434
    },
    {
      "epoch": 0.08310442990560855,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8674000051114953e-05,
      "loss": 0.4299,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 416,
      "tokens_per_second_per_gpu": 16720.61,
      "total_tokens": 37643783
    },
    {
      "epoch": 0.08330420016980472,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8665760341274505e-05,
      "loss": 0.4354,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 417,
      "tokens_per_second_per_gpu": 19094.44,
      "total_tokens": 37735797
    },
    {
      "epoch": 0.0835039704340009,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8657496939576187e-05,
      "loss": 0.4823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 418,
      "tokens_per_second_per_gpu": 19280.76,
      "total_tokens": 37830582
    },
    {
      "epoch": 0.08370374069819707,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8649209868611822e-05,
      "loss": 0.4612,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 419,
      "tokens_per_second_per_gpu": 18970.03,
      "total_tokens": 37922986
    },
    {
      "epoch": 0.08390351096239325,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8640899151037945e-05,
      "loss": 0.4475,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 420,
      "tokens_per_second_per_gpu": 17995.41,
      "total_tokens": 38009445
    },
    {
      "epoch": 0.08410328122658942,
      "grad_norm": 0.765625,
      "learning_rate": 1.863256480957574e-05,
      "loss": 0.4439,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 421,
      "tokens_per_second_per_gpu": 18779.83,
      "total_tokens": 38099722
    },
    {
      "epoch": 0.0843030514907856,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8624206867010982e-05,
      "loss": 0.4248,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 422,
      "tokens_per_second_per_gpu": 18780.03,
      "total_tokens": 38191632
    },
    {
      "epoch": 0.08450282175498178,
      "grad_norm": 0.82421875,
      "learning_rate": 1.861582534619396e-05,
      "loss": 0.413,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 423,
      "tokens_per_second_per_gpu": 16246.02,
      "total_tokens": 38270618
    },
    {
      "epoch": 0.08470259201917794,
      "grad_norm": 0.7265625,
      "learning_rate": 1.860742027003944e-05,
      "loss": 0.473,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 424,
      "tokens_per_second_per_gpu": 20166.67,
      "total_tokens": 38369114
    },
    {
      "epoch": 0.08490236228337412,
      "grad_norm": 0.77734375,
      "learning_rate": 1.859899166152657e-05,
      "loss": 0.4308,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 425,
      "tokens_per_second_per_gpu": 18619.51,
      "total_tokens": 38458330
    },
    {
      "epoch": 0.08510213254757029,
      "grad_norm": 0.703125,
      "learning_rate": 1.8590539543698852e-05,
      "loss": 0.4296,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 426,
      "tokens_per_second_per_gpu": 19298.7,
      "total_tokens": 38552471
    },
    {
      "epoch": 0.08530190281176647,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8582063939664052e-05,
      "loss": 0.4113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 427,
      "tokens_per_second_per_gpu": 18452.13,
      "total_tokens": 38643445
    },
    {
      "epoch": 0.08550167307596264,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8573564872594144e-05,
      "loss": 0.4254,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 428,
      "tokens_per_second_per_gpu": 17340.45,
      "total_tokens": 38728057
    },
    {
      "epoch": 0.08570144334015882,
      "grad_norm": 0.74609375,
      "learning_rate": 1.856504236572526e-05,
      "loss": 0.4204,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 429,
      "tokens_per_second_per_gpu": 17795.4,
      "total_tokens": 38813479
    },
    {
      "epoch": 0.08590121360435499,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8556496442357597e-05,
      "loss": 0.4538,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 430,
      "tokens_per_second_per_gpu": 19618.3,
      "total_tokens": 38907760
    },
    {
      "epoch": 0.08610098386855117,
      "grad_norm": 0.75,
      "learning_rate": 1.854792712585539e-05,
      "loss": 0.4244,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 431,
      "tokens_per_second_per_gpu": 17058.12,
      "total_tokens": 38991339
    },
    {
      "epoch": 0.08630075413274733,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8539334439646827e-05,
      "loss": 0.4427,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 432,
      "tokens_per_second_per_gpu": 19729.06,
      "total_tokens": 39087467
    },
    {
      "epoch": 0.08650052439694352,
      "grad_norm": 0.71875,
      "learning_rate": 1.8530718407223976e-05,
      "loss": 0.4127,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 433,
      "tokens_per_second_per_gpu": 17743.01,
      "total_tokens": 39173576
    },
    {
      "epoch": 0.0867002946611397,
      "grad_norm": 0.734375,
      "learning_rate": 1.8522079052142748e-05,
      "loss": 0.4602,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 434,
      "tokens_per_second_per_gpu": 19902.58,
      "total_tokens": 39269152
    },
    {
      "epoch": 0.08690006492533586,
      "grad_norm": 0.765625,
      "learning_rate": 1.8513416398022803e-05,
      "loss": 0.4357,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 435,
      "tokens_per_second_per_gpu": 18467.01,
      "total_tokens": 39358741
    },
    {
      "epoch": 0.08709983518953204,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8504730468547508e-05,
      "loss": 0.4114,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 436,
      "tokens_per_second_per_gpu": 18141.44,
      "total_tokens": 39446971
    },
    {
      "epoch": 0.08729960545372821,
      "grad_norm": 0.73828125,
      "learning_rate": 1.849602128746387e-05,
      "loss": 0.4206,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 437,
      "tokens_per_second_per_gpu": 17413.95,
      "total_tokens": 39531879
    },
    {
      "epoch": 0.08749937571792439,
      "grad_norm": 1.0078125,
      "learning_rate": 1.8487288878582447e-05,
      "loss": 0.4582,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 438,
      "tokens_per_second_per_gpu": 18633.87,
      "total_tokens": 39622256
    },
    {
      "epoch": 0.08769914598212056,
      "grad_norm": 0.76171875,
      "learning_rate": 1.847853326577732e-05,
      "loss": 0.4334,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 439,
      "tokens_per_second_per_gpu": 19427.5,
      "total_tokens": 39715924
    },
    {
      "epoch": 0.08789891624631674,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8469754472985994e-05,
      "loss": 0.4522,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 440,
      "tokens_per_second_per_gpu": 19013.41,
      "total_tokens": 39808110
    },
    {
      "epoch": 0.0880986865105129,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8460952524209355e-05,
      "loss": 0.4244,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 441,
      "tokens_per_second_per_gpu": 17656.86,
      "total_tokens": 39894219
    },
    {
      "epoch": 0.08829845677470909,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8452127443511597e-05,
      "loss": 0.4425,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 442,
      "tokens_per_second_per_gpu": 19419.62,
      "total_tokens": 39987487
    },
    {
      "epoch": 0.08849822703890525,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8443279255020153e-05,
      "loss": 0.4111,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 443,
      "tokens_per_second_per_gpu": 18191.33,
      "total_tokens": 40076790
    },
    {
      "epoch": 0.08869799730310143,
      "grad_norm": 0.78515625,
      "learning_rate": 1.843440798292563e-05,
      "loss": 0.4381,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 444,
      "tokens_per_second_per_gpu": 18053.51,
      "total_tokens": 40165096
    },
    {
      "epoch": 0.08889776756729761,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8425513651481748e-05,
      "loss": 0.436,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 445,
      "tokens_per_second_per_gpu": 18134.32,
      "total_tokens": 40253574
    },
    {
      "epoch": 0.08909753783149378,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8416596285005274e-05,
      "loss": 0.4302,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 446,
      "tokens_per_second_per_gpu": 18820.79,
      "total_tokens": 40345374
    },
    {
      "epoch": 0.08929730809568996,
      "grad_norm": 0.734375,
      "learning_rate": 1.8407655907875937e-05,
      "loss": 0.4524,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 447,
      "tokens_per_second_per_gpu": 19454.97,
      "total_tokens": 40440395
    },
    {
      "epoch": 0.08949707835988613,
      "grad_norm": 0.73046875,
      "learning_rate": 1.83986925445364e-05,
      "loss": 0.4427,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 448,
      "tokens_per_second_per_gpu": 18717.61,
      "total_tokens": 40531712
    },
    {
      "epoch": 0.08969684862408231,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8389706219492147e-05,
      "loss": 0.45,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 449,
      "tokens_per_second_per_gpu": 16469.27,
      "total_tokens": 40613429
    },
    {
      "epoch": 0.08989661888827848,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8380696957311447e-05,
      "loss": 0.4321,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 450,
      "tokens_per_second_per_gpu": 20118.31,
      "total_tokens": 40709871
    },
    {
      "epoch": 0.09009638915247466,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8371664782625287e-05,
      "loss": 0.435,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 451,
      "tokens_per_second_per_gpu": 18795.55,
      "total_tokens": 40801044
    },
    {
      "epoch": 0.09029615941667082,
      "grad_norm": 0.765625,
      "learning_rate": 1.836260972012728e-05,
      "loss": 0.4262,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 452,
      "tokens_per_second_per_gpu": 17295.97,
      "total_tokens": 40885158
    },
    {
      "epoch": 0.090495929680867,
      "grad_norm": 0.734375,
      "learning_rate": 1.8353531794573623e-05,
      "loss": 0.4382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 453,
      "tokens_per_second_per_gpu": 20211.93,
      "total_tokens": 40982884
    },
    {
      "epoch": 0.09069569994506317,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8344431030783022e-05,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 454,
      "tokens_per_second_per_gpu": 16817.32,
      "total_tokens": 41066086
    },
    {
      "epoch": 0.09089547020925935,
      "grad_norm": 0.73046875,
      "learning_rate": 1.833530745363661e-05,
      "loss": 0.4355,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 455,
      "tokens_per_second_per_gpu": 20488.34,
      "total_tokens": 41163509
    },
    {
      "epoch": 0.09109524047345553,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8326161088077905e-05,
      "loss": 0.4373,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 456,
      "tokens_per_second_per_gpu": 19447.56,
      "total_tokens": 41256862
    },
    {
      "epoch": 0.0912950107376517,
      "grad_norm": 0.734375,
      "learning_rate": 1.831699195911272e-05,
      "loss": 0.4723,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 457,
      "tokens_per_second_per_gpu": 21039.01,
      "total_tokens": 41358382
    },
    {
      "epoch": 0.09149478100184788,
      "grad_norm": 0.8203125,
      "learning_rate": 1.83078000918091e-05,
      "loss": 0.4292,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 458,
      "tokens_per_second_per_gpu": 17886.12,
      "total_tokens": 41443914
    },
    {
      "epoch": 0.09169455126604405,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8298585511297263e-05,
      "loss": 0.4135,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 459,
      "tokens_per_second_per_gpu": 17094.61,
      "total_tokens": 41527017
    },
    {
      "epoch": 0.09189432153024023,
      "grad_norm": 0.78125,
      "learning_rate": 1.8289348242769515e-05,
      "loss": 0.4425,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 460,
      "tokens_per_second_per_gpu": 17886.63,
      "total_tokens": 41612776
    },
    {
      "epoch": 0.0920940917944364,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8280088311480203e-05,
      "loss": 0.4329,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 461,
      "tokens_per_second_per_gpu": 18907.74,
      "total_tokens": 41703850
    },
    {
      "epoch": 0.09229386205863258,
      "grad_norm": 0.90234375,
      "learning_rate": 1.827080574274562e-05,
      "loss": 0.4469,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 462,
      "tokens_per_second_per_gpu": 19312.24,
      "total_tokens": 41798183
    },
    {
      "epoch": 0.09249363232282874,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8261500561943956e-05,
      "loss": 0.457,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 463,
      "tokens_per_second_per_gpu": 19838.61,
      "total_tokens": 41894753
    },
    {
      "epoch": 0.09269340258702492,
      "grad_norm": 0.70703125,
      "learning_rate": 1.8252172794515225e-05,
      "loss": 0.4346,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 464,
      "tokens_per_second_per_gpu": 19350.23,
      "total_tokens": 41987824
    },
    {
      "epoch": 0.09289317285122109,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8242822465961177e-05,
      "loss": 0.4994,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 465,
      "tokens_per_second_per_gpu": 19006.01,
      "total_tokens": 42080690
    },
    {
      "epoch": 0.09309294311541727,
      "grad_norm": 0.76171875,
      "learning_rate": 1.823344960184526e-05,
      "loss": 0.4338,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 466,
      "tokens_per_second_per_gpu": 18520.14,
      "total_tokens": 42171026
    },
    {
      "epoch": 0.09329271337961345,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8224054227792524e-05,
      "loss": 0.4538,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 467,
      "tokens_per_second_per_gpu": 19873.91,
      "total_tokens": 42267921
    },
    {
      "epoch": 0.09349248364380962,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8214636369489563e-05,
      "loss": 0.439,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 468,
      "tokens_per_second_per_gpu": 19250.81,
      "total_tokens": 42361852
    },
    {
      "epoch": 0.0936922539080058,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8205196052684445e-05,
      "loss": 0.4435,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 469,
      "tokens_per_second_per_gpu": 18492.74,
      "total_tokens": 42451040
    },
    {
      "epoch": 0.09389202417220197,
      "grad_norm": 0.75,
      "learning_rate": 1.8195733303186634e-05,
      "loss": 0.4455,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 470,
      "tokens_per_second_per_gpu": 19483.91,
      "total_tokens": 42544756
    },
    {
      "epoch": 0.09409179443639815,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8186248146866928e-05,
      "loss": 0.4459,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 471,
      "tokens_per_second_per_gpu": 17517.93,
      "total_tokens": 42631711
    },
    {
      "epoch": 0.09429156470059431,
      "grad_norm": 0.765625,
      "learning_rate": 1.8176740609657372e-05,
      "loss": 0.4017,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 472,
      "tokens_per_second_per_gpu": 17535.57,
      "total_tokens": 42716862
    },
    {
      "epoch": 0.0944913349647905,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8167210717551224e-05,
      "loss": 0.4251,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 473,
      "tokens_per_second_per_gpu": 18574.09,
      "total_tokens": 42806582
    },
    {
      "epoch": 0.09469110522898666,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8157658496602836e-05,
      "loss": 0.4265,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 474,
      "tokens_per_second_per_gpu": 18001.53,
      "total_tokens": 42895962
    },
    {
      "epoch": 0.09489087549318284,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8148083972927617e-05,
      "loss": 0.4355,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 475,
      "tokens_per_second_per_gpu": 18252.98,
      "total_tokens": 42985294
    },
    {
      "epoch": 0.09509064575737901,
      "grad_norm": 0.70703125,
      "learning_rate": 1.813848717270195e-05,
      "loss": 0.4578,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 476,
      "tokens_per_second_per_gpu": 19131.12,
      "total_tokens": 43080342
    },
    {
      "epoch": 0.09529041602157519,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8128868122163125e-05,
      "loss": 0.4309,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 477,
      "tokens_per_second_per_gpu": 18866.88,
      "total_tokens": 43172457
    },
    {
      "epoch": 0.09549018628577137,
      "grad_norm": 0.734375,
      "learning_rate": 1.8119226847609247e-05,
      "loss": 0.4366,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 478,
      "tokens_per_second_per_gpu": 18565.81,
      "total_tokens": 43262874
    },
    {
      "epoch": 0.09568995654996754,
      "grad_norm": 0.73828125,
      "learning_rate": 1.81095633753992e-05,
      "loss": 0.4647,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 479,
      "tokens_per_second_per_gpu": 20001.33,
      "total_tokens": 43358319
    },
    {
      "epoch": 0.09588972681416372,
      "grad_norm": 0.73828125,
      "learning_rate": 1.809987773195255e-05,
      "loss": 0.4295,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 480,
      "tokens_per_second_per_gpu": 18694.77,
      "total_tokens": 43448494
    },
    {
      "epoch": 0.09608949707835988,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.4476,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 481,
      "tokens_per_second_per_gpu": 19830.05,
      "total_tokens": 43543709
    },
    {
      "epoch": 0.09628926734255606,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8080440037330697e-05,
      "loss": 0.4222,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 482,
      "tokens_per_second_per_gpu": 16667.19,
      "total_tokens": 43626251
    },
    {
      "epoch": 0.09648903760675223,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8070688039297403e-05,
      "loss": 0.4291,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 483,
      "tokens_per_second_per_gpu": 16229.13,
      "total_tokens": 43705183
    },
    {
      "epoch": 0.09668880787094841,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8060913976311192e-05,
      "loss": 0.4281,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 484,
      "tokens_per_second_per_gpu": 19905.31,
      "total_tokens": 43801144
    },
    {
      "epoch": 0.09688857813514458,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8051117875093974e-05,
      "loss": 0.3972,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 485,
      "tokens_per_second_per_gpu": 16992.2,
      "total_tokens": 43883917
    },
    {
      "epoch": 0.09708834839934076,
      "grad_norm": 0.80078125,
      "learning_rate": 1.804129976242792e-05,
      "loss": 0.4097,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 486,
      "tokens_per_second_per_gpu": 17065.92,
      "total_tokens": 43967262
    },
    {
      "epoch": 0.09728811866353693,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8031459665155363e-05,
      "loss": 0.4215,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 487,
      "tokens_per_second_per_gpu": 18389.08,
      "total_tokens": 44056001
    },
    {
      "epoch": 0.09748788892773311,
      "grad_norm": 0.703125,
      "learning_rate": 1.802159761017877e-05,
      "loss": 0.4318,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 488,
      "tokens_per_second_per_gpu": 18725.55,
      "total_tokens": 44145662
    },
    {
      "epoch": 0.09768765919192929,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8011713624460608e-05,
      "loss": 0.452,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 489,
      "tokens_per_second_per_gpu": 19144.41,
      "total_tokens": 44239031
    },
    {
      "epoch": 0.09788742945612545,
      "grad_norm": 0.71484375,
      "learning_rate": 1.800180773502333e-05,
      "loss": 0.4275,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 490,
      "tokens_per_second_per_gpu": 18449.64,
      "total_tokens": 44330574
    },
    {
      "epoch": 0.09808719972032164,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7991879968949248e-05,
      "loss": 0.436,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 491,
      "tokens_per_second_per_gpu": 19979.05,
      "total_tokens": 44425601
    },
    {
      "epoch": 0.0982869699845178,
      "grad_norm": 0.734375,
      "learning_rate": 1.7981930353380502e-05,
      "loss": 0.4176,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 492,
      "tokens_per_second_per_gpu": 17901.35,
      "total_tokens": 44512930
    },
    {
      "epoch": 0.09848674024871398,
      "grad_norm": 0.703125,
      "learning_rate": 1.797195891551896e-05,
      "loss": 0.3952,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 493,
      "tokens_per_second_per_gpu": 17940.68,
      "total_tokens": 44599621
    },
    {
      "epoch": 0.09868651051291015,
      "grad_norm": 0.734375,
      "learning_rate": 1.796196568262615e-05,
      "loss": 0.4404,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 494,
      "tokens_per_second_per_gpu": 19264.65,
      "total_tokens": 44692261
    },
    {
      "epoch": 0.09888628077710633,
      "grad_norm": 0.7890625,
      "learning_rate": 1.795195068202319e-05,
      "loss": 0.4278,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 495,
      "tokens_per_second_per_gpu": 18076.99,
      "total_tokens": 44779354
    },
    {
      "epoch": 0.0990860510413025,
      "grad_norm": 0.734375,
      "learning_rate": 1.7941913941090712e-05,
      "loss": 0.4252,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 496,
      "tokens_per_second_per_gpu": 18602.28,
      "total_tokens": 44869736
    },
    {
      "epoch": 0.09928582130549868,
      "grad_norm": 0.75390625,
      "learning_rate": 1.793185548726878e-05,
      "loss": 0.4518,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 497,
      "tokens_per_second_per_gpu": 19099.19,
      "total_tokens": 44963646
    },
    {
      "epoch": 0.09948559156969485,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7921775348056822e-05,
      "loss": 0.3845,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 498,
      "tokens_per_second_per_gpu": 16508.16,
      "total_tokens": 45046570
    },
    {
      "epoch": 0.09968536183389103,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7911673551013553e-05,
      "loss": 0.4534,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 499,
      "tokens_per_second_per_gpu": 17870.7,
      "total_tokens": 45134425
    },
    {
      "epoch": 0.0998851320980872,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7901550123756906e-05,
      "loss": 0.4493,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 500,
      "tokens_per_second_per_gpu": 20954.15,
      "total_tokens": 45235592
    },
    {
      "epoch": 0.10008490236228337,
      "grad_norm": 0.7265625,
      "learning_rate": 1.789140509396394e-05,
      "loss": 0.4356,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 501,
      "tokens_per_second_per_gpu": 19823.72,
      "total_tokens": 45330454
    },
    {
      "epoch": 0.10028467262647955,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7881238489370776e-05,
      "loss": 0.4051,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 502,
      "tokens_per_second_per_gpu": 16256.05,
      "total_tokens": 45410084
    },
    {
      "epoch": 0.10048444289067572,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7871050337772527e-05,
      "loss": 0.4139,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 503,
      "tokens_per_second_per_gpu": 17571.93,
      "total_tokens": 45495898
    },
    {
      "epoch": 0.1006842131548719,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7860840667023212e-05,
      "loss": 0.4386,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 504,
      "tokens_per_second_per_gpu": 18145.54,
      "total_tokens": 45584602
    },
    {
      "epoch": 0.10088398341906807,
      "grad_norm": 0.72265625,
      "learning_rate": 1.785060950503568e-05,
      "loss": 0.4369,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 505,
      "tokens_per_second_per_gpu": 19255.48,
      "total_tokens": 45678219
    },
    {
      "epoch": 0.10108375368326425,
      "grad_norm": 0.7734375,
      "learning_rate": 1.784035687978153e-05,
      "loss": 0.4553,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 506,
      "tokens_per_second_per_gpu": 18722.62,
      "total_tokens": 45769353
    },
    {
      "epoch": 0.10128352394746042,
      "grad_norm": 0.76953125,
      "learning_rate": 1.783008281929106e-05,
      "loss": 0.4323,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 507,
      "tokens_per_second_per_gpu": 16529.92,
      "total_tokens": 45850767
    },
    {
      "epoch": 0.1014832942116566,
      "grad_norm": 0.703125,
      "learning_rate": 1.781978735165315e-05,
      "loss": 0.4323,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 508,
      "tokens_per_second_per_gpu": 19527.25,
      "total_tokens": 45946331
    },
    {
      "epoch": 0.10168306447585276,
      "grad_norm": 0.72265625,
      "learning_rate": 1.780947050501522e-05,
      "loss": 0.4659,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 509,
      "tokens_per_second_per_gpu": 20632.55,
      "total_tokens": 46046042
    },
    {
      "epoch": 0.10188283474004894,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7799132307583133e-05,
      "loss": 0.4157,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 510,
      "tokens_per_second_per_gpu": 16872.18,
      "total_tokens": 46129723
    },
    {
      "epoch": 0.10208260500424512,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7788772787621126e-05,
      "loss": 0.4221,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 511,
      "tokens_per_second_per_gpu": 18010.53,
      "total_tokens": 46216901
    },
    {
      "epoch": 0.10228237526844129,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7778391973451728e-05,
      "loss": 0.4209,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 512,
      "tokens_per_second_per_gpu": 18648.54,
      "total_tokens": 46307009
    },
    {
      "epoch": 0.10248214553263747,
      "grad_norm": 1.7109375,
      "learning_rate": 1.7767989893455696e-05,
      "loss": 0.4432,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 513,
      "tokens_per_second_per_gpu": 18950.16,
      "total_tokens": 46398522
    },
    {
      "epoch": 0.10268191579683364,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7757566576071916e-05,
      "loss": 0.4334,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 514,
      "tokens_per_second_per_gpu": 19348.56,
      "total_tokens": 46492828
    },
    {
      "epoch": 0.10288168606102982,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7747122049797336e-05,
      "loss": 0.4048,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 515,
      "tokens_per_second_per_gpu": 17455.66,
      "total_tokens": 46579099
    },
    {
      "epoch": 0.10308145632522599,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7736656343186897e-05,
      "loss": 0.4151,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 516,
      "tokens_per_second_per_gpu": 17575.15,
      "total_tokens": 46665463
    },
    {
      "epoch": 0.10328122658942217,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7726169484853438e-05,
      "loss": 0.4156,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 517,
      "tokens_per_second_per_gpu": 18758.96,
      "total_tokens": 46756408
    },
    {
      "epoch": 0.10348099685361833,
      "grad_norm": 0.74609375,
      "learning_rate": 1.771566150346763e-05,
      "loss": 0.4435,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 518,
      "tokens_per_second_per_gpu": 20156.35,
      "total_tokens": 46853073
    },
    {
      "epoch": 0.10368076711781452,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7705132427757895e-05,
      "loss": 0.4284,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 519,
      "tokens_per_second_per_gpu": 18743.09,
      "total_tokens": 46943243
    },
    {
      "epoch": 0.10388053738201068,
      "grad_norm": 0.765625,
      "learning_rate": 1.769458228651032e-05,
      "loss": 0.454,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 520,
      "tokens_per_second_per_gpu": 19357.57,
      "total_tokens": 47037608
    },
    {
      "epoch": 0.10408030764620686,
      "grad_norm": 0.71875,
      "learning_rate": 1.7684011108568593e-05,
      "loss": 0.4596,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 521,
      "tokens_per_second_per_gpu": 19707.94,
      "total_tokens": 47133207
    },
    {
      "epoch": 0.10428007791040304,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7673418922833904e-05,
      "loss": 0.3896,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 522,
      "tokens_per_second_per_gpu": 17268.67,
      "total_tokens": 47217287
    },
    {
      "epoch": 0.10447984817459921,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7662805758264894e-05,
      "loss": 0.4405,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 523,
      "tokens_per_second_per_gpu": 19096.68,
      "total_tokens": 47309169
    },
    {
      "epoch": 0.10467961843879539,
      "grad_norm": 0.76171875,
      "learning_rate": 1.765217164387754e-05,
      "loss": 0.4458,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 524,
      "tokens_per_second_per_gpu": 17501.19,
      "total_tokens": 47395117
    },
    {
      "epoch": 0.10487938870299156,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7641516608745114e-05,
      "loss": 0.4464,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 525,
      "tokens_per_second_per_gpu": 19754.87,
      "total_tokens": 47490847
    },
    {
      "epoch": 0.10507915896718774,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7630840681998068e-05,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 526,
      "tokens_per_second_per_gpu": 18140.82,
      "total_tokens": 47578584
    },
    {
      "epoch": 0.1052789292313839,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7620143892823977e-05,
      "loss": 0.4207,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 527,
      "tokens_per_second_per_gpu": 19673.73,
      "total_tokens": 47673035
    },
    {
      "epoch": 0.10547869949558009,
      "grad_norm": 0.7265625,
      "learning_rate": 1.760942627046746e-05,
      "loss": 0.4639,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 528,
      "tokens_per_second_per_gpu": 19598.92,
      "total_tokens": 47767630
    },
    {
      "epoch": 0.10567846975977625,
      "grad_norm": 0.7890625,
      "learning_rate": 1.759868784423009e-05,
      "loss": 0.4778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 529,
      "tokens_per_second_per_gpu": 19124.24,
      "total_tokens": 47860342
    },
    {
      "epoch": 0.10587824002397243,
      "grad_norm": 0.71875,
      "learning_rate": 1.75879286434703e-05,
      "loss": 0.4439,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 530,
      "tokens_per_second_per_gpu": 19128.86,
      "total_tokens": 47954112
    },
    {
      "epoch": 0.1060780102881686,
      "grad_norm": 0.74609375,
      "learning_rate": 1.757714869760335e-05,
      "loss": 0.4363,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 531,
      "tokens_per_second_per_gpu": 19947.27,
      "total_tokens": 48049547
    },
    {
      "epoch": 0.10627778055236478,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7566348036101187e-05,
      "loss": 0.4241,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 532,
      "tokens_per_second_per_gpu": 17903.34,
      "total_tokens": 48137012
    },
    {
      "epoch": 0.10647755081656095,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7555526688492418e-05,
      "loss": 0.432,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 533,
      "tokens_per_second_per_gpu": 18471.9,
      "total_tokens": 48225714
    },
    {
      "epoch": 0.10667732108075713,
      "grad_norm": 0.73046875,
      "learning_rate": 1.754468468436219e-05,
      "loss": 0.4292,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 534,
      "tokens_per_second_per_gpu": 18817.76,
      "total_tokens": 48317958
    },
    {
      "epoch": 0.10687709134495331,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7533822053352127e-05,
      "loss": 0.4392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 535,
      "tokens_per_second_per_gpu": 18995.79,
      "total_tokens": 48412348
    },
    {
      "epoch": 0.10707686160914948,
      "grad_norm": 0.8125,
      "learning_rate": 1.752293882516025e-05,
      "loss": 0.4481,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 536,
      "tokens_per_second_per_gpu": 18493.25,
      "total_tokens": 48503315
    },
    {
      "epoch": 0.10727663187334566,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7512035029540887e-05,
      "loss": 0.4392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 537,
      "tokens_per_second_per_gpu": 19339.67,
      "total_tokens": 48595844
    },
    {
      "epoch": 0.10747640213754182,
      "grad_norm": 0.765625,
      "learning_rate": 1.7501110696304598e-05,
      "loss": 0.4231,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 538,
      "tokens_per_second_per_gpu": 17774.79,
      "total_tokens": 48683322
    },
    {
      "epoch": 0.107676172401738,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7490165855318097e-05,
      "loss": 0.4386,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 539,
      "tokens_per_second_per_gpu": 18413.71,
      "total_tokens": 48771978
    },
    {
      "epoch": 0.10787594266593417,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7479200536504154e-05,
      "loss": 0.4344,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 540,
      "tokens_per_second_per_gpu": 19174.59,
      "total_tokens": 48864380
    },
    {
      "epoch": 0.10807571293013035,
      "grad_norm": 0.734375,
      "learning_rate": 1.7468214769841542e-05,
      "loss": 0.4261,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 541,
      "tokens_per_second_per_gpu": 17981.58,
      "total_tokens": 48951920
    },
    {
      "epoch": 0.10827548319432652,
      "grad_norm": 0.7109375,
      "learning_rate": 1.745720858536492e-05,
      "loss": 0.4286,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 542,
      "tokens_per_second_per_gpu": 21228.67,
      "total_tokens": 49052227
    },
    {
      "epoch": 0.1084752534585227,
      "grad_norm": 0.7265625,
      "learning_rate": 1.744618201316478e-05,
      "loss": 0.4177,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 543,
      "tokens_per_second_per_gpu": 18268.34,
      "total_tokens": 49141076
    },
    {
      "epoch": 0.10867502372271887,
      "grad_norm": 0.71875,
      "learning_rate": 1.7435135083387345e-05,
      "loss": 0.4183,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 544,
      "tokens_per_second_per_gpu": 19859.62,
      "total_tokens": 49236552
    },
    {
      "epoch": 0.10887479398691505,
      "grad_norm": 0.7421875,
      "learning_rate": 1.74240678262345e-05,
      "loss": 0.4072,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 545,
      "tokens_per_second_per_gpu": 16798.94,
      "total_tokens": 49318484
    },
    {
      "epoch": 0.10907456425111123,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7412980271963712e-05,
      "loss": 0.4066,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 546,
      "tokens_per_second_per_gpu": 18028.2,
      "total_tokens": 49405995
    },
    {
      "epoch": 0.1092743345153074,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7401872450887917e-05,
      "loss": 0.4114,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 547,
      "tokens_per_second_per_gpu": 17745.18,
      "total_tokens": 49492712
    },
    {
      "epoch": 0.10947410477950358,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7390744393375487e-05,
      "loss": 0.4299,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 548,
      "tokens_per_second_per_gpu": 19041.54,
      "total_tokens": 49584858
    },
    {
      "epoch": 0.10967387504369974,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7379596129850098e-05,
      "loss": 0.4268,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 549,
      "tokens_per_second_per_gpu": 18171.39,
      "total_tokens": 49673057
    },
    {
      "epoch": 0.10987364530789592,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7368427690790678e-05,
      "loss": 0.4586,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 550,
      "tokens_per_second_per_gpu": 20115.43,
      "total_tokens": 49769670
    },
    {
      "epoch": 0.11007341557209209,
      "grad_norm": 0.73046875,
      "learning_rate": 1.735723910673132e-05,
      "loss": 0.4574,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 551,
      "tokens_per_second_per_gpu": 18498.67,
      "total_tokens": 49860822
    },
    {
      "epoch": 0.11027318583628827,
      "grad_norm": 0.73828125,
      "learning_rate": 1.734603040826117e-05,
      "loss": 0.4702,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 552,
      "tokens_per_second_per_gpu": 19673.75,
      "total_tokens": 49956172
    },
    {
      "epoch": 0.11047295610048444,
      "grad_norm": 0.7109375,
      "learning_rate": 1.73348016260244e-05,
      "loss": 0.4261,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 553,
      "tokens_per_second_per_gpu": 18328.15,
      "total_tokens": 50045486
    },
    {
      "epoch": 0.11067272636468062,
      "grad_norm": 0.7265625,
      "learning_rate": 1.732355279072006e-05,
      "loss": 0.4369,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 554,
      "tokens_per_second_per_gpu": 19403.07,
      "total_tokens": 50140476
    },
    {
      "epoch": 0.11087249662887678,
      "grad_norm": 0.7421875,
      "learning_rate": 1.731228393310204e-05,
      "loss": 0.4345,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 555,
      "tokens_per_second_per_gpu": 18145.14,
      "total_tokens": 50229063
    },
    {
      "epoch": 0.11107226689307297,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7300995083978965e-05,
      "loss": 0.4233,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 556,
      "tokens_per_second_per_gpu": 19543.7,
      "total_tokens": 50325174
    },
    {
      "epoch": 0.11127203715726915,
      "grad_norm": 0.734375,
      "learning_rate": 1.7289686274214116e-05,
      "loss": 0.4647,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 557,
      "tokens_per_second_per_gpu": 19453.43,
      "total_tokens": 50419450
    },
    {
      "epoch": 0.11147180742146531,
      "grad_norm": 0.734375,
      "learning_rate": 1.7278357534725353e-05,
      "loss": 0.4489,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 558,
      "tokens_per_second_per_gpu": 18381.71,
      "total_tokens": 50509585
    },
    {
      "epoch": 0.1116715776856615,
      "grad_norm": 0.71875,
      "learning_rate": 1.726700889648501e-05,
      "loss": 0.4384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 559,
      "tokens_per_second_per_gpu": 18326.12,
      "total_tokens": 50599489
    },
    {
      "epoch": 0.11187134794985766,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7255640390519838e-05,
      "loss": 0.4121,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 560,
      "tokens_per_second_per_gpu": 19128.87,
      "total_tokens": 50691212
    },
    {
      "epoch": 0.11207111821405384,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7244252047910893e-05,
      "loss": 0.4207,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 561,
      "tokens_per_second_per_gpu": 19322.25,
      "total_tokens": 50782508
    },
    {
      "epoch": 0.11227088847825001,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7232843899793467e-05,
      "loss": 0.4452,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 562,
      "tokens_per_second_per_gpu": 18491.12,
      "total_tokens": 50873515
    },
    {
      "epoch": 0.11247065874244619,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7221415977357008e-05,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 563,
      "tokens_per_second_per_gpu": 18048.61,
      "total_tokens": 50960179
    },
    {
      "epoch": 0.11267042900664236,
      "grad_norm": 0.734375,
      "learning_rate": 1.7209968311845014e-05,
      "loss": 0.4211,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 564,
      "tokens_per_second_per_gpu": 17469.75,
      "total_tokens": 51045625
    },
    {
      "epoch": 0.11287019927083854,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7198500934554966e-05,
      "loss": 0.4371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 565,
      "tokens_per_second_per_gpu": 18457.88,
      "total_tokens": 51136316
    },
    {
      "epoch": 0.1130699695350347,
      "grad_norm": 0.73046875,
      "learning_rate": 1.718701387683824e-05,
      "loss": 0.4197,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 566,
      "tokens_per_second_per_gpu": 18260.89,
      "total_tokens": 51225028
    },
    {
      "epoch": 0.11326973979923088,
      "grad_norm": 0.73046875,
      "learning_rate": 1.717550717010001e-05,
      "loss": 0.4542,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 567,
      "tokens_per_second_per_gpu": 19723.45,
      "total_tokens": 51321212
    },
    {
      "epoch": 0.11346951006342706,
      "grad_norm": 0.76171875,
      "learning_rate": 1.716398084579917e-05,
      "loss": 0.4476,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 568,
      "tokens_per_second_per_gpu": 20293.65,
      "total_tokens": 51419538
    },
    {
      "epoch": 0.11366928032762323,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7152434935448257e-05,
      "loss": 0.4248,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 569,
      "tokens_per_second_per_gpu": 18630.76,
      "total_tokens": 51509706
    },
    {
      "epoch": 0.11386905059181941,
      "grad_norm": 0.78125,
      "learning_rate": 1.7140869470613344e-05,
      "loss": 0.3914,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 570,
      "tokens_per_second_per_gpu": 18011.95,
      "total_tokens": 51597312
    },
    {
      "epoch": 0.11406882085601558,
      "grad_norm": 0.75,
      "learning_rate": 1.7129284482913973e-05,
      "loss": 0.4325,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 571,
      "tokens_per_second_per_gpu": 18087.39,
      "total_tokens": 51685967
    },
    {
      "epoch": 0.11426859112021176,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7117680004023057e-05,
      "loss": 0.4227,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 572,
      "tokens_per_second_per_gpu": 18524.42,
      "total_tokens": 51775943
    },
    {
      "epoch": 0.11446836138440793,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7106056065666793e-05,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 573,
      "tokens_per_second_per_gpu": 17047.41,
      "total_tokens": 51858944
    },
    {
      "epoch": 0.11466813164860411,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7094412699624596e-05,
      "loss": 0.4321,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 574,
      "tokens_per_second_per_gpu": 18822.2,
      "total_tokens": 51950874
    },
    {
      "epoch": 0.11486790191280027,
      "grad_norm": 0.71875,
      "learning_rate": 1.7082749937728972e-05,
      "loss": 0.3655,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 575,
      "tokens_per_second_per_gpu": 16560.32,
      "total_tokens": 52033041
    },
    {
      "epoch": 0.11506767217699646,
      "grad_norm": 0.75,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.4396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 576,
      "tokens_per_second_per_gpu": 19659.22,
      "total_tokens": 52128535
    },
    {
      "epoch": 0.11526744244119262,
      "grad_norm": 0.73046875,
      "learning_rate": 1.705936635397259e-05,
      "loss": 0.4133,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 577,
      "tokens_per_second_per_gpu": 18921.22,
      "total_tokens": 52220909
    },
    {
      "epoch": 0.1154672127053888,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7047645596041654e-05,
      "loss": 0.4056,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 578,
      "tokens_per_second_per_gpu": 18352.47,
      "total_tokens": 52309231
    },
    {
      "epoch": 0.11566698296958498,
      "grad_norm": 0.73046875,
      "learning_rate": 1.703590557011677e-05,
      "loss": 0.4158,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 579,
      "tokens_per_second_per_gpu": 17747.56,
      "total_tokens": 52397148
    },
    {
      "epoch": 0.11586675323378115,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7024146308294725e-05,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 580,
      "tokens_per_second_per_gpu": 18640.89,
      "total_tokens": 52486907
    },
    {
      "epoch": 0.11606652349797733,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7012367842724887e-05,
      "loss": 0.4263,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 581,
      "tokens_per_second_per_gpu": 18632.36,
      "total_tokens": 52577320
    },
    {
      "epoch": 0.1162662937621735,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7000570205609136e-05,
      "loss": 0.4649,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 582,
      "tokens_per_second_per_gpu": 20136.77,
      "total_tokens": 52675774
    },
    {
      "epoch": 0.11646606402636968,
      "grad_norm": 0.78125,
      "learning_rate": 1.6988753429201756e-05,
      "loss": 0.4318,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 583,
      "tokens_per_second_per_gpu": 18917.05,
      "total_tokens": 52767207
    },
    {
      "epoch": 0.11666583429056585,
      "grad_norm": 0.6953125,
      "learning_rate": 1.697691754580937e-05,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 584,
      "tokens_per_second_per_gpu": 19631.44,
      "total_tokens": 52861508
    },
    {
      "epoch": 0.11686560455476203,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6965062587790823e-05,
      "loss": 0.4279,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 585,
      "tokens_per_second_per_gpu": 18810.94,
      "total_tokens": 52953106
    },
    {
      "epoch": 0.11706537481895819,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6953188587557122e-05,
      "loss": 0.3971,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 586,
      "tokens_per_second_per_gpu": 18937.38,
      "total_tokens": 53044807
    },
    {
      "epoch": 0.11726514508315437,
      "grad_norm": 0.7265625,
      "learning_rate": 1.694129557757133e-05,
      "loss": 0.4318,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 587,
      "tokens_per_second_per_gpu": 19397.72,
      "total_tokens": 53138442
    },
    {
      "epoch": 0.11746491534735054,
      "grad_norm": 0.828125,
      "learning_rate": 1.692938359034848e-05,
      "loss": 0.4212,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 588,
      "tokens_per_second_per_gpu": 19004.26,
      "total_tokens": 53230066
    },
    {
      "epoch": 0.11766468561154672,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6917452658455496e-05,
      "loss": 0.4189,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 589,
      "tokens_per_second_per_gpu": 18368.64,
      "total_tokens": 53319867
    },
    {
      "epoch": 0.1178644558757429,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6905502814511085e-05,
      "loss": 0.411,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 590,
      "tokens_per_second_per_gpu": 18572.14,
      "total_tokens": 53409036
    },
    {
      "epoch": 0.11806422613993907,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6893534091185658e-05,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 591,
      "tokens_per_second_per_gpu": 16608.35,
      "total_tokens": 53492102
    },
    {
      "epoch": 0.11826399640413525,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6881546521201263e-05,
      "loss": 0.4393,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 592,
      "tokens_per_second_per_gpu": 18689.0,
      "total_tokens": 53583749
    },
    {
      "epoch": 0.11846376666833142,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6869540137331445e-05,
      "loss": 0.3976,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 593,
      "tokens_per_second_per_gpu": 19054.5,
      "total_tokens": 53675554
    },
    {
      "epoch": 0.1186635369325276,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6857514972401207e-05,
      "loss": 0.4211,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 594,
      "tokens_per_second_per_gpu": 17957.39,
      "total_tokens": 53763981
    },
    {
      "epoch": 0.11886330719672376,
      "grad_norm": 0.7421875,
      "learning_rate": 1.684547105928689e-05,
      "loss": 0.4283,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 595,
      "tokens_per_second_per_gpu": 19763.61,
      "total_tokens": 53857905
    },
    {
      "epoch": 0.11906307746091994,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6833408430916085e-05,
      "loss": 0.4443,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 596,
      "tokens_per_second_per_gpu": 18629.89,
      "total_tokens": 53948173
    },
    {
      "epoch": 0.11926284772511611,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6821327120267567e-05,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 597,
      "tokens_per_second_per_gpu": 18030.77,
      "total_tokens": 54035722
    },
    {
      "epoch": 0.11946261798931229,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6809227160371173e-05,
      "loss": 0.4155,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 598,
      "tokens_per_second_per_gpu": 18332.31,
      "total_tokens": 54125075
    },
    {
      "epoch": 0.11966238825350846,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6797108584307732e-05,
      "loss": 0.4282,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 599,
      "tokens_per_second_per_gpu": 18725.17,
      "total_tokens": 54216789
    },
    {
      "epoch": 0.11986215851770464,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6784971425208967e-05,
      "loss": 0.4155,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 600,
      "tokens_per_second_per_gpu": 19050.18,
      "total_tokens": 54308457
    },
    {
      "epoch": 0.12006192878190082,
      "grad_norm": 0.734375,
      "learning_rate": 1.6772815716257414e-05,
      "loss": 0.4383,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 601,
      "tokens_per_second_per_gpu": 19978.95,
      "total_tokens": 54404146
    },
    {
      "epoch": 0.12026169904609699,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6760641490686307e-05,
      "loss": 0.4168,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 602,
      "tokens_per_second_per_gpu": 18336.05,
      "total_tokens": 54492769
    },
    {
      "epoch": 0.12046146931029317,
      "grad_norm": 0.7265625,
      "learning_rate": 1.674844878177952e-05,
      "loss": 0.4361,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 603,
      "tokens_per_second_per_gpu": 17825.42,
      "total_tokens": 54578931
    },
    {
      "epoch": 0.12066123957448933,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6736237622871453e-05,
      "loss": 0.3924,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 604,
      "tokens_per_second_per_gpu": 18955.06,
      "total_tokens": 54670975
    },
    {
      "epoch": 0.12086100983868552,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6724008047346946e-05,
      "loss": 0.4424,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 605,
      "tokens_per_second_per_gpu": 20463.94,
      "total_tokens": 54769448
    },
    {
      "epoch": 0.12106078010288168,
      "grad_norm": 0.734375,
      "learning_rate": 1.6711760088641197e-05,
      "loss": 0.4181,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 606,
      "tokens_per_second_per_gpu": 18074.29,
      "total_tokens": 54857374
    },
    {
      "epoch": 0.12126055036707786,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6699493780239652e-05,
      "loss": 0.3921,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 607,
      "tokens_per_second_per_gpu": 17476.64,
      "total_tokens": 54941484
    },
    {
      "epoch": 0.12146032063127403,
      "grad_norm": 0.72265625,
      "learning_rate": 1.668720915567793e-05,
      "loss": 0.4167,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 608,
      "tokens_per_second_per_gpu": 18693.55,
      "total_tokens": 55031605
    },
    {
      "epoch": 0.12166009089547021,
      "grad_norm": 0.703125,
      "learning_rate": 1.667490624854173e-05,
      "loss": 0.4179,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 609,
      "tokens_per_second_per_gpu": 19435.61,
      "total_tokens": 55126467
    },
    {
      "epoch": 0.12185986115966638,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6662585092466722e-05,
      "loss": 0.4446,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 610,
      "tokens_per_second_per_gpu": 19336.36,
      "total_tokens": 55221792
    },
    {
      "epoch": 0.12205963142386256,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6650245721138483e-05,
      "loss": 0.3856,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 611,
      "tokens_per_second_per_gpu": 17603.44,
      "total_tokens": 55306263
    },
    {
      "epoch": 0.12225940168805874,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6637888168292385e-05,
      "loss": 0.4077,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 612,
      "tokens_per_second_per_gpu": 17045.76,
      "total_tokens": 55389836
    },
    {
      "epoch": 0.1224591719522549,
      "grad_norm": 0.76171875,
      "learning_rate": 1.66255124677135e-05,
      "loss": 0.4541,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 613,
      "tokens_per_second_per_gpu": 19028.96,
      "total_tokens": 55480911
    },
    {
      "epoch": 0.12265894221645109,
      "grad_norm": 0.7421875,
      "learning_rate": 1.661311865323652e-05,
      "loss": 0.4276,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 614,
      "tokens_per_second_per_gpu": 17680.77,
      "total_tokens": 55569379
    },
    {
      "epoch": 0.12285871248064725,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6600706758745668e-05,
      "loss": 0.4168,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 615,
      "tokens_per_second_per_gpu": 19284.2,
      "total_tokens": 55664022
    },
    {
      "epoch": 0.12305848274484343,
      "grad_norm": 0.77734375,
      "learning_rate": 1.658827681817458e-05,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 616,
      "tokens_per_second_per_gpu": 17585.49,
      "total_tokens": 55748330
    },
    {
      "epoch": 0.1232582530090396,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6575828865506246e-05,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 617,
      "tokens_per_second_per_gpu": 19970.52,
      "total_tokens": 55844646
    },
    {
      "epoch": 0.12345802327323578,
      "grad_norm": 0.734375,
      "learning_rate": 1.6563362934772893e-05,
      "loss": 0.407,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 618,
      "tokens_per_second_per_gpu": 19372.72,
      "total_tokens": 55936545
    },
    {
      "epoch": 0.12365779353743195,
      "grad_norm": 0.78125,
      "learning_rate": 1.6550879060055897e-05,
      "loss": 0.41,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 619,
      "tokens_per_second_per_gpu": 18239.41,
      "total_tokens": 56025406
    },
    {
      "epoch": 0.12385756380162813,
      "grad_norm": 0.71875,
      "learning_rate": 1.6538377275485694e-05,
      "loss": 0.3884,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 620,
      "tokens_per_second_per_gpu": 18631.65,
      "total_tokens": 56115754
    },
    {
      "epoch": 0.1240573340658243,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6525857615241686e-05,
      "loss": 0.4134,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 621,
      "tokens_per_second_per_gpu": 17913.83,
      "total_tokens": 56201788
    },
    {
      "epoch": 0.12425710433002048,
      "grad_norm": 0.703125,
      "learning_rate": 1.651332011355215e-05,
      "loss": 0.3989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 622,
      "tokens_per_second_per_gpu": 17630.83,
      "total_tokens": 56287872
    },
    {
      "epoch": 0.12445687459421666,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6500764804694132e-05,
      "loss": 0.4194,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 623,
      "tokens_per_second_per_gpu": 18342.33,
      "total_tokens": 56375507
    },
    {
      "epoch": 0.12465664485841282,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6488191722993372e-05,
      "loss": 0.4335,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 624,
      "tokens_per_second_per_gpu": 19737.66,
      "total_tokens": 56469658
    },
    {
      "epoch": 0.124856415122609,
      "grad_norm": 0.75390625,
      "learning_rate": 1.647560090282419e-05,
      "loss": 0.4351,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 625,
      "tokens_per_second_per_gpu": 20651.74,
      "total_tokens": 56568267
    },
    {
      "epoch": 0.12505618538680519,
      "grad_norm": 0.74609375,
      "learning_rate": 1.646299237860941e-05,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 626,
      "tokens_per_second_per_gpu": 17901.29,
      "total_tokens": 56656162
    },
    {
      "epoch": 0.12525595565100134,
      "grad_norm": 0.734375,
      "learning_rate": 1.6450366184820256e-05,
      "loss": 0.3995,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 627,
      "tokens_per_second_per_gpu": 17940.05,
      "total_tokens": 56743224
    },
    {
      "epoch": 0.12545572591519752,
      "grad_norm": 0.77734375,
      "learning_rate": 1.643772235597626e-05,
      "loss": 0.4126,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 628,
      "tokens_per_second_per_gpu": 16794.15,
      "total_tokens": 56827009
    },
    {
      "epoch": 0.1256554961793937,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6425060926645168e-05,
      "loss": 0.4113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 629,
      "tokens_per_second_per_gpu": 16540.0,
      "total_tokens": 56908490
    },
    {
      "epoch": 0.12585526644358988,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6412381931442837e-05,
      "loss": 0.4285,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 630,
      "tokens_per_second_per_gpu": 18976.58,
      "total_tokens": 57000459
    },
    {
      "epoch": 0.12605503670778603,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6399685405033168e-05,
      "loss": 0.3882,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 631,
      "tokens_per_second_per_gpu": 16066.22,
      "total_tokens": 57079643
    },
    {
      "epoch": 0.12625480697198221,
      "grad_norm": 0.78125,
      "learning_rate": 1.638697138212797e-05,
      "loss": 0.4453,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 632,
      "tokens_per_second_per_gpu": 17823.37,
      "total_tokens": 57165509
    },
    {
      "epoch": 0.1264545772361784,
      "grad_norm": 0.72265625,
      "learning_rate": 1.63742398974869e-05,
      "loss": 0.3934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 633,
      "tokens_per_second_per_gpu": 18692.4,
      "total_tokens": 57254995
    },
    {
      "epoch": 0.12665434750037458,
      "grad_norm": 0.76171875,
      "learning_rate": 1.636149098591735e-05,
      "loss": 0.4272,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 634,
      "tokens_per_second_per_gpu": 18637.34,
      "total_tokens": 57345992
    },
    {
      "epoch": 0.12685411776457076,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6348724682274353e-05,
      "loss": 0.4164,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 635,
      "tokens_per_second_per_gpu": 17927.81,
      "total_tokens": 57432589
    },
    {
      "epoch": 0.1270538880287669,
      "grad_norm": 1.515625,
      "learning_rate": 1.6335941021460507e-05,
      "loss": 0.4283,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 636,
      "tokens_per_second_per_gpu": 20299.65,
      "total_tokens": 57528763
    },
    {
      "epoch": 0.1272536582929631,
      "grad_norm": 0.75,
      "learning_rate": 1.6323140038425842e-05,
      "loss": 0.4168,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 637,
      "tokens_per_second_per_gpu": 17966.28,
      "total_tokens": 57616257
    },
    {
      "epoch": 0.12745342855715927,
      "grad_norm": 0.75,
      "learning_rate": 1.6310321768167762e-05,
      "loss": 0.4012,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 638,
      "tokens_per_second_per_gpu": 18348.37,
      "total_tokens": 57706247
    },
    {
      "epoch": 0.12765319882135545,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6297486245730925e-05,
      "loss": 0.4284,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 639,
      "tokens_per_second_per_gpu": 19993.45,
      "total_tokens": 57804046
    },
    {
      "epoch": 0.1278529690855516,
      "grad_norm": 0.72265625,
      "learning_rate": 1.628463350620716e-05,
      "loss": 0.4143,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 640,
      "tokens_per_second_per_gpu": 19186.34,
      "total_tokens": 57898635
    },
    {
      "epoch": 0.12805273934974779,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6271763584735373e-05,
      "loss": 0.3977,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 641,
      "tokens_per_second_per_gpu": 18204.39,
      "total_tokens": 57986857
    },
    {
      "epoch": 0.12825250961394397,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6258876516501425e-05,
      "loss": 0.4371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 642,
      "tokens_per_second_per_gpu": 18272.56,
      "total_tokens": 58076622
    },
    {
      "epoch": 0.12845227987814015,
      "grad_norm": 0.7578125,
      "learning_rate": 1.624597233673808e-05,
      "loss": 0.4161,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 643,
      "tokens_per_second_per_gpu": 18755.32,
      "total_tokens": 58167171
    },
    {
      "epoch": 0.1286520501423363,
      "grad_norm": 0.734375,
      "learning_rate": 1.6233051080724867e-05,
      "loss": 0.4346,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 644,
      "tokens_per_second_per_gpu": 19100.14,
      "total_tokens": 58259729
    },
    {
      "epoch": 0.12885182040653248,
      "grad_norm": 0.7265625,
      "learning_rate": 1.622011278378801e-05,
      "loss": 0.3946,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 645,
      "tokens_per_second_per_gpu": 17022.07,
      "total_tokens": 58343279
    },
    {
      "epoch": 0.12905159067072866,
      "grad_norm": 0.71875,
      "learning_rate": 1.6207157481300315e-05,
      "loss": 0.4439,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 646,
      "tokens_per_second_per_gpu": 20252.07,
      "total_tokens": 58441703
    },
    {
      "epoch": 0.12925136093492484,
      "grad_norm": 0.68359375,
      "learning_rate": 1.6194185208681085e-05,
      "loss": 0.4225,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 647,
      "tokens_per_second_per_gpu": 19395.34,
      "total_tokens": 58536021
    },
    {
      "epoch": 0.12945113119912102,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6181196001396018e-05,
      "loss": 0.3982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 648,
      "tokens_per_second_per_gpu": 19012.75,
      "total_tokens": 58631069
    },
    {
      "epoch": 0.12965090146331718,
      "grad_norm": 0.734375,
      "learning_rate": 1.616818989495711e-05,
      "loss": 0.4239,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 649,
      "tokens_per_second_per_gpu": 18620.83,
      "total_tokens": 58721451
    },
    {
      "epoch": 0.12985067172751336,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6155166924922564e-05,
      "loss": 0.4379,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 650,
      "tokens_per_second_per_gpu": 17125.95,
      "total_tokens": 58804920
    },
    {
      "epoch": 0.13005044199170954,
      "grad_norm": 0.71875,
      "learning_rate": 1.6142127126896682e-05,
      "loss": 0.3768,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 651,
      "tokens_per_second_per_gpu": 17325.54,
      "total_tokens": 58889456
    },
    {
      "epoch": 0.13025021225590572,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6129070536529767e-05,
      "loss": 0.4328,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 652,
      "tokens_per_second_per_gpu": 19978.5,
      "total_tokens": 58985417
    },
    {
      "epoch": 0.13044998252010187,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6115997189518043e-05,
      "loss": 0.4278,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 653,
      "tokens_per_second_per_gpu": 19023.71,
      "total_tokens": 59077468
    },
    {
      "epoch": 0.13064975278429805,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6102907121603543e-05,
      "loss": 0.4102,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 654,
      "tokens_per_second_per_gpu": 19426.91,
      "total_tokens": 59170550
    },
    {
      "epoch": 0.13084952304849423,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6089800368574013e-05,
      "loss": 0.3938,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 655,
      "tokens_per_second_per_gpu": 17258.68,
      "total_tokens": 59254027
    },
    {
      "epoch": 0.1310492933126904,
      "grad_norm": 0.6875,
      "learning_rate": 1.6076676966262815e-05,
      "loss": 0.4,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 656,
      "tokens_per_second_per_gpu": 19151.96,
      "total_tokens": 59346943
    },
    {
      "epoch": 0.1312490635768866,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6063536950548825e-05,
      "loss": 0.4214,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 657,
      "tokens_per_second_per_gpu": 18349.07,
      "total_tokens": 59436313
    },
    {
      "epoch": 0.13144883384108275,
      "grad_norm": 0.7109375,
      "learning_rate": 1.605038035735635e-05,
      "loss": 0.3821,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 658,
      "tokens_per_second_per_gpu": 18059.7,
      "total_tokens": 59524242
    },
    {
      "epoch": 0.13164860410527893,
      "grad_norm": 0.7421875,
      "learning_rate": 1.603720722265501e-05,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 659,
      "tokens_per_second_per_gpu": 17108.29,
      "total_tokens": 59608030
    },
    {
      "epoch": 0.1318483743694751,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6024017582459652e-05,
      "loss": 0.3811,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 660,
      "tokens_per_second_per_gpu": 18275.32,
      "total_tokens": 59698185
    },
    {
      "epoch": 0.1320481446336713,
      "grad_norm": 0.734375,
      "learning_rate": 1.6010811472830253e-05,
      "loss": 0.4284,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 661,
      "tokens_per_second_per_gpu": 19399.98,
      "total_tokens": 59795206
    },
    {
      "epoch": 0.13224791489786744,
      "grad_norm": 0.77734375,
      "learning_rate": 1.599758892987181e-05,
      "loss": 0.4305,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 662,
      "tokens_per_second_per_gpu": 18335.21,
      "total_tokens": 59884050
    },
    {
      "epoch": 0.13244768516206362,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5984349989734248e-05,
      "loss": 0.4308,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 663,
      "tokens_per_second_per_gpu": 18098.79,
      "total_tokens": 59972177
    },
    {
      "epoch": 0.1326474554262598,
      "grad_norm": 0.69921875,
      "learning_rate": 1.597109468861232e-05,
      "loss": 0.4303,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 664,
      "tokens_per_second_per_gpu": 20199.6,
      "total_tokens": 60068363
    },
    {
      "epoch": 0.13284722569045598,
      "grad_norm": 0.734375,
      "learning_rate": 1.595782306274553e-05,
      "loss": 0.4096,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 665,
      "tokens_per_second_per_gpu": 19504.7,
      "total_tokens": 60161759
    },
    {
      "epoch": 0.13304699595465214,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5944535148417982e-05,
      "loss": 0.4375,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 666,
      "tokens_per_second_per_gpu": 17908.53,
      "total_tokens": 60249738
    },
    {
      "epoch": 0.13324676621884832,
      "grad_norm": 0.7265625,
      "learning_rate": 1.593123098195833e-05,
      "loss": 0.4157,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 667,
      "tokens_per_second_per_gpu": 18390.82,
      "total_tokens": 60340366
    },
    {
      "epoch": 0.1334465364830445,
      "grad_norm": 0.71875,
      "learning_rate": 1.5917910599739652e-05,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 668,
      "tokens_per_second_per_gpu": 19253.91,
      "total_tokens": 60433878
    },
    {
      "epoch": 0.13364630674724068,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5904574038179372e-05,
      "loss": 0.3731,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 669,
      "tokens_per_second_per_gpu": 17344.39,
      "total_tokens": 60518267
    },
    {
      "epoch": 0.13384607701143686,
      "grad_norm": 0.734375,
      "learning_rate": 1.589122133373914e-05,
      "loss": 0.4341,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 670,
      "tokens_per_second_per_gpu": 18995.93,
      "total_tokens": 60609328
    },
    {
      "epoch": 0.134045847275633,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 0.4287,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 671,
      "tokens_per_second_per_gpu": 20150.98,
      "total_tokens": 60705702
    },
    {
      "epoch": 0.1342456175398292,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5864467642285975e-05,
      "loss": 0.3911,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 672,
      "tokens_per_second_per_gpu": 17710.31,
      "total_tokens": 60792023
    },
    {
      "epoch": 0.13444538780402537,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5851066728416617e-05,
      "loss": 0.4251,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 673,
      "tokens_per_second_per_gpu": 17758.73,
      "total_tokens": 60878414
    },
    {
      "epoch": 0.13464515806822155,
      "grad_norm": 0.71875,
      "learning_rate": 1.5837649817954245e-05,
      "loss": 0.434,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 674,
      "tokens_per_second_per_gpu": 19413.63,
      "total_tokens": 60973398
    },
    {
      "epoch": 0.1348449283324177,
      "grad_norm": 0.734375,
      "learning_rate": 1.5824216947580183e-05,
      "loss": 0.417,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 675,
      "tokens_per_second_per_gpu": 19103.04,
      "total_tokens": 61066544
    },
    {
      "epoch": 0.1350446985966139,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5810768154019386e-05,
      "loss": 0.4069,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 676,
      "tokens_per_second_per_gpu": 17525.7,
      "total_tokens": 61152603
    },
    {
      "epoch": 0.13524446886081007,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5797303474040332e-05,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 677,
      "tokens_per_second_per_gpu": 18197.61,
      "total_tokens": 61239706
    },
    {
      "epoch": 0.13544423912500625,
      "grad_norm": 0.703125,
      "learning_rate": 1.5783822944454956e-05,
      "loss": 0.4025,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 678,
      "tokens_per_second_per_gpu": 18032.7,
      "total_tokens": 61327276
    },
    {
      "epoch": 0.13564400938920243,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5770326602118502e-05,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 679,
      "tokens_per_second_per_gpu": 19761.85,
      "total_tokens": 61421794
    },
    {
      "epoch": 0.13584377965339858,
      "grad_norm": 0.73046875,
      "learning_rate": 1.575681448392946e-05,
      "loss": 0.4047,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 680,
      "tokens_per_second_per_gpu": 18602.66,
      "total_tokens": 61510623
    },
    {
      "epoch": 0.13604354991759476,
      "grad_norm": 0.71875,
      "learning_rate": 1.5743286626829437e-05,
      "loss": 0.3856,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 681,
      "tokens_per_second_per_gpu": 18295.5,
      "total_tokens": 61598103
    },
    {
      "epoch": 0.13624332018179094,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5729743067803087e-05,
      "loss": 0.4152,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 682,
      "tokens_per_second_per_gpu": 18194.68,
      "total_tokens": 61687100
    },
    {
      "epoch": 0.13644309044598713,
      "grad_norm": 0.78125,
      "learning_rate": 1.5716183843877977e-05,
      "loss": 0.427,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 683,
      "tokens_per_second_per_gpu": 19216.2,
      "total_tokens": 61780032
    },
    {
      "epoch": 0.13664286071018328,
      "grad_norm": 0.73828125,
      "learning_rate": 1.570260899212451e-05,
      "loss": 0.392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 684,
      "tokens_per_second_per_gpu": 16768.42,
      "total_tokens": 61861936
    },
    {
      "epoch": 0.13684263097437946,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5689018549655815e-05,
      "loss": 0.4265,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 685,
      "tokens_per_second_per_gpu": 18192.79,
      "total_tokens": 61949633
    },
    {
      "epoch": 0.13704240123857564,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5675412553627638e-05,
      "loss": 0.4133,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 686,
      "tokens_per_second_per_gpu": 18813.16,
      "total_tokens": 62039839
    },
    {
      "epoch": 0.13724217150277182,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5661791041238258e-05,
      "loss": 0.4299,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 687,
      "tokens_per_second_per_gpu": 18138.4,
      "total_tokens": 62129166
    },
    {
      "epoch": 0.13744194176696797,
      "grad_norm": 0.7265625,
      "learning_rate": 1.564815404972836e-05,
      "loss": 0.4056,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 688,
      "tokens_per_second_per_gpu": 18738.97,
      "total_tokens": 62219379
    },
    {
      "epoch": 0.13764171203116415,
      "grad_norm": 0.734375,
      "learning_rate": 1.5634501616380967e-05,
      "loss": 0.4235,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 689,
      "tokens_per_second_per_gpu": 19840.34,
      "total_tokens": 62313571
    },
    {
      "epoch": 0.13784148229536033,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5620833778521306e-05,
      "loss": 0.4322,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 690,
      "tokens_per_second_per_gpu": 19067.92,
      "total_tokens": 62407824
    },
    {
      "epoch": 0.13804125255955652,
      "grad_norm": 0.72265625,
      "learning_rate": 1.560715057351673e-05,
      "loss": 0.4154,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 691,
      "tokens_per_second_per_gpu": 18330.62,
      "total_tokens": 62497397
    },
    {
      "epoch": 0.1382410228237527,
      "grad_norm": 0.72265625,
      "learning_rate": 1.559345203877659e-05,
      "loss": 0.4342,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 692,
      "tokens_per_second_per_gpu": 19840.16,
      "total_tokens": 62591760
    },
    {
      "epoch": 0.13844079308794885,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5579738211752165e-05,
      "loss": 0.4138,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 693,
      "tokens_per_second_per_gpu": 19649.55,
      "total_tokens": 62687740
    },
    {
      "epoch": 0.13864056335214503,
      "grad_norm": 0.7421875,
      "learning_rate": 1.556600912993653e-05,
      "loss": 0.3846,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 694,
      "tokens_per_second_per_gpu": 16997.87,
      "total_tokens": 62772123
    },
    {
      "epoch": 0.1388403336163412,
      "grad_norm": 0.6640625,
      "learning_rate": 1.555226483086447e-05,
      "loss": 0.397,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 695,
      "tokens_per_second_per_gpu": 19686.68,
      "total_tokens": 62867604
    },
    {
      "epoch": 0.1390401038805374,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5538505352112373e-05,
      "loss": 0.399,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 696,
      "tokens_per_second_per_gpu": 18106.54,
      "total_tokens": 62954608
    },
    {
      "epoch": 0.13923987414473354,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5524730731298136e-05,
      "loss": 0.4113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 697,
      "tokens_per_second_per_gpu": 19038.97,
      "total_tokens": 63047777
    },
    {
      "epoch": 0.13943964440892972,
      "grad_norm": 0.75,
      "learning_rate": 1.5510941006081032e-05,
      "loss": 0.4086,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 698,
      "tokens_per_second_per_gpu": 18104.8,
      "total_tokens": 63136462
    },
    {
      "epoch": 0.1396394146731259,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5497136214161662e-05,
      "loss": 0.405,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 699,
      "tokens_per_second_per_gpu": 19762.47,
      "total_tokens": 63230823
    },
    {
      "epoch": 0.1398391849373221,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5483316393281783e-05,
      "loss": 0.3783,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 700,
      "tokens_per_second_per_gpu": 16495.37,
      "total_tokens": 63312285
    },
    {
      "epoch": 0.14003895520151827,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5469481581224274e-05,
      "loss": 0.4032,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 701,
      "tokens_per_second_per_gpu": 17035.72,
      "total_tokens": 63398601
    },
    {
      "epoch": 0.14023872546571442,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5455631815812963e-05,
      "loss": 0.4096,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 702,
      "tokens_per_second_per_gpu": 18635.21,
      "total_tokens": 63489118
    },
    {
      "epoch": 0.1404384957299106,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5441767134912597e-05,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 703,
      "tokens_per_second_per_gpu": 18430.94,
      "total_tokens": 63579418
    },
    {
      "epoch": 0.14063826599410678,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5427887576428675e-05,
      "loss": 0.4046,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 704,
      "tokens_per_second_per_gpu": 20181.61,
      "total_tokens": 63674757
    },
    {
      "epoch": 0.14083803625830296,
      "grad_norm": 0.703125,
      "learning_rate": 1.541399317830738e-05,
      "loss": 0.3964,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 705,
      "tokens_per_second_per_gpu": 18134.12,
      "total_tokens": 63763070
    },
    {
      "epoch": 0.14103780652249912,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5400083978535475e-05,
      "loss": 0.4118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 706,
      "tokens_per_second_per_gpu": 17151.66,
      "total_tokens": 63846725
    },
    {
      "epoch": 0.1412375767866953,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5386160015140168e-05,
      "loss": 0.4316,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 707,
      "tokens_per_second_per_gpu": 21282.39,
      "total_tokens": 63948936
    },
    {
      "epoch": 0.14143734705089148,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5372221326189053e-05,
      "loss": 0.4142,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 708,
      "tokens_per_second_per_gpu": 19206.89,
      "total_tokens": 64043009
    },
    {
      "epoch": 0.14163711731508766,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5358267949789968e-05,
      "loss": 0.4384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 709,
      "tokens_per_second_per_gpu": 18361.88,
      "total_tokens": 64133090
    },
    {
      "epoch": 0.1418368875792838,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5344299924090915e-05,
      "loss": 0.4014,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 710,
      "tokens_per_second_per_gpu": 17963.13,
      "total_tokens": 64221128
    },
    {
      "epoch": 0.14203665784348,
      "grad_norm": 0.75,
      "learning_rate": 1.533031728727994e-05,
      "loss": 0.403,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 711,
      "tokens_per_second_per_gpu": 17556.13,
      "total_tokens": 64307391
    },
    {
      "epoch": 0.14223642810767617,
      "grad_norm": 0.70703125,
      "learning_rate": 1.531632007758504e-05,
      "loss": 0.4047,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 712,
      "tokens_per_second_per_gpu": 18307.71,
      "total_tokens": 64394972
    },
    {
      "epoch": 0.14243619837187235,
      "grad_norm": 0.73046875,
      "learning_rate": 1.530230833327405e-05,
      "loss": 0.4226,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 713,
      "tokens_per_second_per_gpu": 19951.17,
      "total_tokens": 64492602
    },
    {
      "epoch": 0.14263596863606853,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5288282092654544e-05,
      "loss": 0.4248,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 714,
      "tokens_per_second_per_gpu": 18022.97,
      "total_tokens": 64585867
    },
    {
      "epoch": 0.14283573890026469,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5274241394073733e-05,
      "loss": 0.4144,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 715,
      "tokens_per_second_per_gpu": 17461.39,
      "total_tokens": 64671629
    },
    {
      "epoch": 0.14303550916446087,
      "grad_norm": 0.73046875,
      "learning_rate": 1.526018627591834e-05,
      "loss": 0.4371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 716,
      "tokens_per_second_per_gpu": 19083.25,
      "total_tokens": 64764108
    },
    {
      "epoch": 0.14323527942865705,
      "grad_norm": 0.73046875,
      "learning_rate": 1.524611677661454e-05,
      "loss": 0.4086,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 717,
      "tokens_per_second_per_gpu": 19781.64,
      "total_tokens": 64856750
    },
    {
      "epoch": 0.14343504969285323,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5232032934627782e-05,
      "loss": 0.4021,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 718,
      "tokens_per_second_per_gpu": 19598.18,
      "total_tokens": 64950507
    },
    {
      "epoch": 0.14363481995704938,
      "grad_norm": 0.734375,
      "learning_rate": 1.5217934788462774e-05,
      "loss": 0.4336,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 719,
      "tokens_per_second_per_gpu": 19218.41,
      "total_tokens": 65045166
    },
    {
      "epoch": 0.14383459022124556,
      "grad_norm": 0.71875,
      "learning_rate": 1.52038223766633e-05,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 720,
      "tokens_per_second_per_gpu": 17756.11,
      "total_tokens": 65131775
    },
    {
      "epoch": 0.14403436048544174,
      "grad_norm": 0.75,
      "learning_rate": 1.5189695737812153e-05,
      "loss": 0.4097,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 721,
      "tokens_per_second_per_gpu": 17452.86,
      "total_tokens": 65218568
    },
    {
      "epoch": 0.14423413074963792,
      "grad_norm": 0.7265625,
      "learning_rate": 1.517555491053103e-05,
      "loss": 0.4093,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 722,
      "tokens_per_second_per_gpu": 18136.26,
      "total_tokens": 65307950
    },
    {
      "epoch": 0.1444339010138341,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5161399933480402e-05,
      "loss": 0.4266,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 723,
      "tokens_per_second_per_gpu": 20351.88,
      "total_tokens": 65405357
    },
    {
      "epoch": 0.14463367127803026,
      "grad_norm": 0.765625,
      "learning_rate": 1.5147230845359452e-05,
      "loss": 0.4118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 724,
      "tokens_per_second_per_gpu": 19001.43,
      "total_tokens": 65497035
    },
    {
      "epoch": 0.14483344154222644,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5133047684905916e-05,
      "loss": 0.4257,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 725,
      "tokens_per_second_per_gpu": 18535.47,
      "total_tokens": 65586798
    },
    {
      "epoch": 0.14503321180642262,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5118850490896012e-05,
      "loss": 0.3922,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 726,
      "tokens_per_second_per_gpu": 18034.29,
      "total_tokens": 65674473
    },
    {
      "epoch": 0.1452329820706188,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5104639302144327e-05,
      "loss": 0.3893,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 727,
      "tokens_per_second_per_gpu": 17444.1,
      "total_tokens": 65761733
    },
    {
      "epoch": 0.14543275233481495,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5090414157503715e-05,
      "loss": 0.4164,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 728,
      "tokens_per_second_per_gpu": 19187.32,
      "total_tokens": 65853962
    },
    {
      "epoch": 0.14563252259901113,
      "grad_norm": 0.828125,
      "learning_rate": 1.5076175095865171e-05,
      "loss": 0.3947,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 729,
      "tokens_per_second_per_gpu": 17580.51,
      "total_tokens": 65941149
    },
    {
      "epoch": 0.1458322928632073,
      "grad_norm": 1.0078125,
      "learning_rate": 1.5061922156157746e-05,
      "loss": 0.4175,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 730,
      "tokens_per_second_per_gpu": 19552.0,
      "total_tokens": 66036020
    },
    {
      "epoch": 0.1460320631274035,
      "grad_norm": 0.7265625,
      "learning_rate": 1.504765537734844e-05,
      "loss": 0.4191,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 731,
      "tokens_per_second_per_gpu": 18254.87,
      "total_tokens": 66125954
    },
    {
      "epoch": 0.14623183339159965,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5033374798442073e-05,
      "loss": 0.4093,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 732,
      "tokens_per_second_per_gpu": 19050.48,
      "total_tokens": 66217971
    },
    {
      "epoch": 0.14643160365579583,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5019080458481203e-05,
      "loss": 0.4027,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 733,
      "tokens_per_second_per_gpu": 18384.52,
      "total_tokens": 66306927
    },
    {
      "epoch": 0.146631373919992,
      "grad_norm": 0.72265625,
      "learning_rate": 1.500477239654602e-05,
      "loss": 0.3998,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 734,
      "tokens_per_second_per_gpu": 18471.3,
      "total_tokens": 66397700
    },
    {
      "epoch": 0.1468311441841882,
      "grad_norm": 0.765625,
      "learning_rate": 1.4990450651754207e-05,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 735,
      "tokens_per_second_per_gpu": 18698.62,
      "total_tokens": 66488687
    },
    {
      "epoch": 0.14703091444838437,
      "grad_norm": 0.71875,
      "learning_rate": 1.4976115263260876e-05,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 736,
      "tokens_per_second_per_gpu": 17900.52,
      "total_tokens": 66575085
    },
    {
      "epoch": 0.14723068471258052,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4961766270258422e-05,
      "loss": 0.416,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 737,
      "tokens_per_second_per_gpu": 19667.78,
      "total_tokens": 66671092
    },
    {
      "epoch": 0.1474304549767767,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4947403711976454e-05,
      "loss": 0.3956,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 738,
      "tokens_per_second_per_gpu": 18753.41,
      "total_tokens": 66763031
    },
    {
      "epoch": 0.14763022524097288,
      "grad_norm": 0.734375,
      "learning_rate": 1.4933027627681651e-05,
      "loss": 0.4106,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 739,
      "tokens_per_second_per_gpu": 17687.35,
      "total_tokens": 66850509
    },
    {
      "epoch": 0.14782999550516907,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4918638056677682e-05,
      "loss": 0.3978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 740,
      "tokens_per_second_per_gpu": 18592.92,
      "total_tokens": 66939624
    },
    {
      "epoch": 0.14802976576936522,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4904235038305084e-05,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 741,
      "tokens_per_second_per_gpu": 18128.12,
      "total_tokens": 67028949
    },
    {
      "epoch": 0.1482295360335614,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4889818611941155e-05,
      "loss": 0.391,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 742,
      "tokens_per_second_per_gpu": 18547.97,
      "total_tokens": 67118034
    },
    {
      "epoch": 0.14842930629775758,
      "grad_norm": 0.71875,
      "learning_rate": 1.4875388816999865e-05,
      "loss": 0.4335,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 743,
      "tokens_per_second_per_gpu": 19957.63,
      "total_tokens": 67214522
    },
    {
      "epoch": 0.14862907656195376,
      "grad_norm": 0.703125,
      "learning_rate": 1.4860945692931711e-05,
      "loss": 0.4479,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 744,
      "tokens_per_second_per_gpu": 21290.19,
      "total_tokens": 67317210
    },
    {
      "epoch": 0.14882884682614994,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4846489279223653e-05,
      "loss": 0.4127,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 745,
      "tokens_per_second_per_gpu": 20076.44,
      "total_tokens": 67413774
    },
    {
      "epoch": 0.1490286170903461,
      "grad_norm": 0.734375,
      "learning_rate": 1.4832019615398962e-05,
      "loss": 0.4192,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 746,
      "tokens_per_second_per_gpu": 18669.7,
      "total_tokens": 67503611
    },
    {
      "epoch": 0.14922838735454227,
      "grad_norm": 0.734375,
      "learning_rate": 1.4817536741017153e-05,
      "loss": 0.4095,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 747,
      "tokens_per_second_per_gpu": 18790.31,
      "total_tokens": 67594814
    },
    {
      "epoch": 0.14942815761873846,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4803040695673853e-05,
      "loss": 0.403,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 748,
      "tokens_per_second_per_gpu": 18127.84,
      "total_tokens": 67682834
    },
    {
      "epoch": 0.14962792788293464,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4788531519000696e-05,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 749,
      "tokens_per_second_per_gpu": 19080.1,
      "total_tokens": 67777324
    },
    {
      "epoch": 0.1498276981471308,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4774009250665215e-05,
      "loss": 0.3951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 750,
      "tokens_per_second_per_gpu": 18124.7,
      "total_tokens": 67865009
    },
    {
      "epoch": 0.15002746841132697,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4759473930370738e-05,
      "loss": 0.3719,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 751,
      "tokens_per_second_per_gpu": 18021.56,
      "total_tokens": 67951896
    },
    {
      "epoch": 0.15022723867552315,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4744925597856275e-05,
      "loss": 0.4179,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 752,
      "tokens_per_second_per_gpu": 19125.94,
      "total_tokens": 68044325
    },
    {
      "epoch": 0.15042700893971933,
      "grad_norm": 1.234375,
      "learning_rate": 1.473036429289641e-05,
      "loss": 0.4065,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 753,
      "tokens_per_second_per_gpu": 17658.82,
      "total_tokens": 68131313
    },
    {
      "epoch": 0.15062677920391548,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4715790055301197e-05,
      "loss": 0.3991,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 754,
      "tokens_per_second_per_gpu": 18758.23,
      "total_tokens": 68223358
    },
    {
      "epoch": 0.15082654946811166,
      "grad_norm": 0.70703125,
      "learning_rate": 1.470120292491605e-05,
      "loss": 0.4057,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 755,
      "tokens_per_second_per_gpu": 18887.03,
      "total_tokens": 68315774
    },
    {
      "epoch": 0.15102631973230785,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4686602941621618e-05,
      "loss": 0.4309,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 756,
      "tokens_per_second_per_gpu": 18813.38,
      "total_tokens": 68408483
    },
    {
      "epoch": 0.15122608999650403,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4671990145333697e-05,
      "loss": 0.4138,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 757,
      "tokens_per_second_per_gpu": 19178.73,
      "total_tokens": 68500626
    },
    {
      "epoch": 0.1514258602607002,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4657364576003122e-05,
      "loss": 0.4332,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 758,
      "tokens_per_second_per_gpu": 20069.12,
      "total_tokens": 68597376
    },
    {
      "epoch": 0.15162563052489636,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4642726273615639e-05,
      "loss": 0.4094,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 759,
      "tokens_per_second_per_gpu": 19709.21,
      "total_tokens": 68693357
    },
    {
      "epoch": 0.15182540078909254,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4628075278191803e-05,
      "loss": 0.3775,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 760,
      "tokens_per_second_per_gpu": 18881.67,
      "total_tokens": 68784373
    },
    {
      "epoch": 0.15202517105328872,
      "grad_norm": 0.71875,
      "learning_rate": 1.461341162978688e-05,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 761,
      "tokens_per_second_per_gpu": 19512.36,
      "total_tokens": 68878259
    },
    {
      "epoch": 0.1522249413174849,
      "grad_norm": 0.703125,
      "learning_rate": 1.4598735368490724e-05,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 762,
      "tokens_per_second_per_gpu": 18543.7,
      "total_tokens": 68969015
    },
    {
      "epoch": 0.15242471158168105,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4584046534427671e-05,
      "loss": 0.4177,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 763,
      "tokens_per_second_per_gpu": 19872.04,
      "total_tokens": 69064981
    },
    {
      "epoch": 0.15262448184587724,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4569345167756436e-05,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 764,
      "tokens_per_second_per_gpu": 19820.31,
      "total_tokens": 69159264
    },
    {
      "epoch": 0.15282425211007342,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4554631308669993e-05,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 765,
      "tokens_per_second_per_gpu": 18717.15,
      "total_tokens": 69250255
    },
    {
      "epoch": 0.1530240223742696,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 0.4243,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 766,
      "tokens_per_second_per_gpu": 18449.54,
      "total_tokens": 69340492
    },
    {
      "epoch": 0.15322379263846578,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4525166274194038e-05,
      "loss": 0.3994,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 767,
      "tokens_per_second_per_gpu": 18817.48,
      "total_tokens": 69432208
    },
    {
      "epoch": 0.15342356290266193,
      "grad_norm": 0.75,
      "learning_rate": 1.4510415179360806e-05,
      "loss": 0.4092,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 768,
      "tokens_per_second_per_gpu": 20049.64,
      "total_tokens": 69526596
    },
    {
      "epoch": 0.1536233331668581,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4495651753224706e-05,
      "loss": 0.4116,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 769,
      "tokens_per_second_per_gpu": 18881.29,
      "total_tokens": 69618365
    },
    {
      "epoch": 0.1538231034310543,
      "grad_norm": 0.94921875,
      "learning_rate": 1.4480876036148376e-05,
      "loss": 0.3978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 770,
      "tokens_per_second_per_gpu": 17494.47,
      "total_tokens": 69704830
    },
    {
      "epoch": 0.15402287369525047,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4466088068528068e-05,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 771,
      "tokens_per_second_per_gpu": 17981.93,
      "total_tokens": 69791923
    },
    {
      "epoch": 0.15422264395944663,
      "grad_norm": 0.734375,
      "learning_rate": 1.4451287890793513e-05,
      "loss": 0.3767,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 772,
      "tokens_per_second_per_gpu": 17240.16,
      "total_tokens": 69876481
    },
    {
      "epoch": 0.1544224142236428,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4436475543407843e-05,
      "loss": 0.3546,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 773,
      "tokens_per_second_per_gpu": 16311.66,
      "total_tokens": 69956304
    },
    {
      "epoch": 0.154622184487839,
      "grad_norm": 0.72265625,
      "learning_rate": 1.442165106686744e-05,
      "loss": 0.3696,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 774,
      "tokens_per_second_per_gpu": 17204.56,
      "total_tokens": 70040793
    },
    {
      "epoch": 0.15482195475203517,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4406814501701857e-05,
      "loss": 0.4134,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 775,
      "tokens_per_second_per_gpu": 19595.8,
      "total_tokens": 70135484
    },
    {
      "epoch": 0.15502172501623132,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4391965888473705e-05,
      "loss": 0.4025,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 776,
      "tokens_per_second_per_gpu": 18026.09,
      "total_tokens": 70222031
    },
    {
      "epoch": 0.1552214952804275,
      "grad_norm": 0.74609375,
      "learning_rate": 1.437710526777852e-05,
      "loss": 0.4156,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 777,
      "tokens_per_second_per_gpu": 17875.3,
      "total_tokens": 70309380
    },
    {
      "epoch": 0.15542126554462368,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4362232680244673e-05,
      "loss": 0.4247,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 778,
      "tokens_per_second_per_gpu": 19930.61,
      "total_tokens": 70405027
    },
    {
      "epoch": 0.15562103580881986,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4347348166533247e-05,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 779,
      "tokens_per_second_per_gpu": 16906.7,
      "total_tokens": 70487894
    },
    {
      "epoch": 0.15582080607301604,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4332451767337938e-05,
      "loss": 0.3945,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 780,
      "tokens_per_second_per_gpu": 18908.87,
      "total_tokens": 70581219
    },
    {
      "epoch": 0.1560205763372122,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4317543523384928e-05,
      "loss": 0.3983,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 781,
      "tokens_per_second_per_gpu": 18887.4,
      "total_tokens": 70672818
    },
    {
      "epoch": 0.15622034660140838,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4302623475432796e-05,
      "loss": 0.4119,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 782,
      "tokens_per_second_per_gpu": 18781.83,
      "total_tokens": 70765435
    },
    {
      "epoch": 0.15642011686560456,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4287691664272376e-05,
      "loss": 0.4188,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 783,
      "tokens_per_second_per_gpu": 19098.14,
      "total_tokens": 70857603
    },
    {
      "epoch": 0.15661988712980074,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4272748130726668e-05,
      "loss": 0.4131,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 784,
      "tokens_per_second_per_gpu": 20484.07,
      "total_tokens": 70956504
    },
    {
      "epoch": 0.1568196573939969,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4257792915650728e-05,
      "loss": 0.4094,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 785,
      "tokens_per_second_per_gpu": 18581.53,
      "total_tokens": 71047793
    },
    {
      "epoch": 0.15701942765819307,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4242826059931538e-05,
      "loss": 0.4035,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 786,
      "tokens_per_second_per_gpu": 18390.76,
      "total_tokens": 71139202
    },
    {
      "epoch": 0.15721919792238925,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4227847604487914e-05,
      "loss": 0.3971,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 787,
      "tokens_per_second_per_gpu": 19153.89,
      "total_tokens": 71234605
    },
    {
      "epoch": 0.15741896818658543,
      "grad_norm": 0.7421875,
      "learning_rate": 1.421285759027038e-05,
      "loss": 0.3874,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 788,
      "tokens_per_second_per_gpu": 17594.8,
      "total_tokens": 71319881
    },
    {
      "epoch": 0.15761873845078161,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4197856058261059e-05,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 789,
      "tokens_per_second_per_gpu": 18613.77,
      "total_tokens": 71409255
    },
    {
      "epoch": 0.15781850871497777,
      "grad_norm": 0.72265625,
      "learning_rate": 1.418284304947357e-05,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 790,
      "tokens_per_second_per_gpu": 19729.46,
      "total_tokens": 71504587
    },
    {
      "epoch": 0.15801827897917395,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4167818604952906e-05,
      "loss": 0.4139,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 791,
      "tokens_per_second_per_gpu": 19373.38,
      "total_tokens": 71597794
    },
    {
      "epoch": 0.15821804924337013,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4152782765775323e-05,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 792,
      "tokens_per_second_per_gpu": 19513.79,
      "total_tokens": 71691867
    },
    {
      "epoch": 0.1584178195075663,
      "grad_norm": 0.6875,
      "learning_rate": 1.4137735573048232e-05,
      "loss": 0.3924,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 793,
      "tokens_per_second_per_gpu": 18886.19,
      "total_tokens": 71783013
    },
    {
      "epoch": 0.15861758977176246,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4122677067910084e-05,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 794,
      "tokens_per_second_per_gpu": 17154.79,
      "total_tokens": 71868014
    },
    {
      "epoch": 0.15881736003595864,
      "grad_norm": 0.703125,
      "learning_rate": 1.4107607291530256e-05,
      "loss": 0.4132,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 795,
      "tokens_per_second_per_gpu": 19139.28,
      "total_tokens": 71960997
    },
    {
      "epoch": 0.15901713030015482,
      "grad_norm": 0.71875,
      "learning_rate": 1.409252628510894e-05,
      "loss": 0.415,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 796,
      "tokens_per_second_per_gpu": 20153.84,
      "total_tokens": 72055575
    },
    {
      "epoch": 0.159216900564351,
      "grad_norm": 0.69140625,
      "learning_rate": 1.4077434089877038e-05,
      "loss": 0.4194,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 797,
      "tokens_per_second_per_gpu": 19157.46,
      "total_tokens": 72148403
    },
    {
      "epoch": 0.15941667082854716,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4062330747096027e-05,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 798,
      "tokens_per_second_per_gpu": 17693.91,
      "total_tokens": 72234799
    },
    {
      "epoch": 0.15961644109274334,
      "grad_norm": 0.734375,
      "learning_rate": 1.4047216298057872e-05,
      "loss": 0.4043,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 799,
      "tokens_per_second_per_gpu": 19362.47,
      "total_tokens": 72327402
    },
    {
      "epoch": 0.15981621135693952,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4032090784084904e-05,
      "loss": 0.388,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 800,
      "tokens_per_second_per_gpu": 17615.52,
      "total_tokens": 72412746
    },
    {
      "epoch": 0.1600159816211357,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4016954246529697e-05,
      "loss": 0.4186,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 801,
      "tokens_per_second_per_gpu": 19399.97,
      "total_tokens": 72505018
    },
    {
      "epoch": 0.16021575188533188,
      "grad_norm": 0.734375,
      "learning_rate": 1.4001806726774965e-05,
      "loss": 0.4244,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 802,
      "tokens_per_second_per_gpu": 19294.81,
      "total_tokens": 72597803
    },
    {
      "epoch": 0.16041552214952803,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3986648266233452e-05,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 803,
      "tokens_per_second_per_gpu": 17424.39,
      "total_tokens": 72682758
    },
    {
      "epoch": 0.16061529241372421,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3971478906347806e-05,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 804,
      "tokens_per_second_per_gpu": 19246.74,
      "total_tokens": 72775864
    },
    {
      "epoch": 0.1608150626779204,
      "grad_norm": 0.703125,
      "learning_rate": 1.3956298688590484e-05,
      "loss": 0.4064,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 805,
      "tokens_per_second_per_gpu": 19326.97,
      "total_tokens": 72869428
    },
    {
      "epoch": 0.16101483294211658,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3941107654463619e-05,
      "loss": 0.4049,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 806,
      "tokens_per_second_per_gpu": 18374.03,
      "total_tokens": 72958650
    },
    {
      "epoch": 0.16121460320631273,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3925905845498915e-05,
      "loss": 0.3675,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 807,
      "tokens_per_second_per_gpu": 16446.66,
      "total_tokens": 73039252
    },
    {
      "epoch": 0.1614143734705089,
      "grad_norm": 0.7109375,
      "learning_rate": 1.391069330325754e-05,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 808,
      "tokens_per_second_per_gpu": 19260.52,
      "total_tokens": 73131381
    },
    {
      "epoch": 0.1616141437347051,
      "grad_norm": 0.75,
      "learning_rate": 1.3895470069330003e-05,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 809,
      "tokens_per_second_per_gpu": 18152.97,
      "total_tokens": 73218753
    },
    {
      "epoch": 0.16181391399890127,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3880236185336045e-05,
      "loss": 0.4229,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 810,
      "tokens_per_second_per_gpu": 19837.21,
      "total_tokens": 73314552
    },
    {
      "epoch": 0.16201368426309745,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3864991692924524e-05,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 811,
      "tokens_per_second_per_gpu": 19061.41,
      "total_tokens": 73406319
    },
    {
      "epoch": 0.1622134545272936,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3849736633773298e-05,
      "loss": 0.4083,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 812,
      "tokens_per_second_per_gpu": 17800.21,
      "total_tokens": 73492995
    },
    {
      "epoch": 0.16241322479148979,
      "grad_norm": 0.71875,
      "learning_rate": 1.3834471049589117e-05,
      "loss": 0.4311,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 813,
      "tokens_per_second_per_gpu": 19443.04,
      "total_tokens": 73587630
    },
    {
      "epoch": 0.16261299505568597,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3819194982107504e-05,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 814,
      "tokens_per_second_per_gpu": 19018.78,
      "total_tokens": 73679946
    },
    {
      "epoch": 0.16281276531988215,
      "grad_norm": 0.703125,
      "learning_rate": 1.3803908473092648e-05,
      "loss": 0.387,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 815,
      "tokens_per_second_per_gpu": 18616.66,
      "total_tokens": 73771386
    },
    {
      "epoch": 0.1630125355840783,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3788611564337277e-05,
      "loss": 0.3658,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 816,
      "tokens_per_second_per_gpu": 17425.78,
      "total_tokens": 73856683
    },
    {
      "epoch": 0.16321230584827448,
      "grad_norm": 0.71484375,
      "learning_rate": 1.377330429766256e-05,
      "loss": 0.3999,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 817,
      "tokens_per_second_per_gpu": 18915.38,
      "total_tokens": 73947866
    },
    {
      "epoch": 0.16341207611247066,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3757986714917972e-05,
      "loss": 0.4228,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 818,
      "tokens_per_second_per_gpu": 19902.29,
      "total_tokens": 74045338
    },
    {
      "epoch": 0.16361184637666684,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3742658857981204e-05,
      "loss": 0.3802,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 819,
      "tokens_per_second_per_gpu": 17318.45,
      "total_tokens": 74131628
    },
    {
      "epoch": 0.163811616640863,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3727320768758034e-05,
      "loss": 0.4009,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 820,
      "tokens_per_second_per_gpu": 19620.18,
      "total_tokens": 74227333
    },
    {
      "epoch": 0.16401138690505918,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3711972489182208e-05,
      "loss": 0.4251,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 821,
      "tokens_per_second_per_gpu": 19102.11,
      "total_tokens": 74320671
    },
    {
      "epoch": 0.16421115716925536,
      "grad_norm": 0.72265625,
      "learning_rate": 1.369661406121534e-05,
      "loss": 0.4261,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 822,
      "tokens_per_second_per_gpu": 20819.39,
      "total_tokens": 74422276
    },
    {
      "epoch": 0.16441092743345154,
      "grad_norm": 0.78125,
      "learning_rate": 1.3681245526846782e-05,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 823,
      "tokens_per_second_per_gpu": 16941.14,
      "total_tokens": 74505838
    },
    {
      "epoch": 0.16461069769764772,
      "grad_norm": 0.71484375,
      "learning_rate": 1.366586692809352e-05,
      "loss": 0.4089,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 824,
      "tokens_per_second_per_gpu": 19639.77,
      "total_tokens": 74601857
    },
    {
      "epoch": 0.16481046796184387,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3650478307000059e-05,
      "loss": 0.3714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 825,
      "tokens_per_second_per_gpu": 18843.08,
      "total_tokens": 74694909
    },
    {
      "epoch": 0.16501023822604005,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3635079705638298e-05,
      "loss": 0.419,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 826,
      "tokens_per_second_per_gpu": 20056.2,
      "total_tokens": 74789972
    },
    {
      "epoch": 0.16521000849023623,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3619671166107432e-05,
      "loss": 0.4109,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 827,
      "tokens_per_second_per_gpu": 18545.78,
      "total_tokens": 74878591
    },
    {
      "epoch": 0.1654097787544324,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3604252730533811e-05,
      "loss": 0.4191,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 828,
      "tokens_per_second_per_gpu": 20231.62,
      "total_tokens": 74977017
    },
    {
      "epoch": 0.16560954901862857,
      "grad_norm": 0.75,
      "learning_rate": 1.3588824441070852e-05,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 829,
      "tokens_per_second_per_gpu": 17170.54,
      "total_tokens": 75060237
    },
    {
      "epoch": 0.16580931928282475,
      "grad_norm": 0.7421875,
      "learning_rate": 1.357338633989891e-05,
      "loss": 0.41,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 830,
      "tokens_per_second_per_gpu": 17069.44,
      "total_tokens": 75145101
    },
    {
      "epoch": 0.16600908954702093,
      "grad_norm": 0.71875,
      "learning_rate": 1.3557938469225167e-05,
      "loss": 0.3541,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 831,
      "tokens_per_second_per_gpu": 17756.49,
      "total_tokens": 75230884
    },
    {
      "epoch": 0.1662088598112171,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3542480871283507e-05,
      "loss": 0.3886,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 832,
      "tokens_per_second_per_gpu": 17613.29,
      "total_tokens": 75316999
    },
    {
      "epoch": 0.1664086300754133,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3527013588334415e-05,
      "loss": 0.4038,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 833,
      "tokens_per_second_per_gpu": 18083.9,
      "total_tokens": 75404964
    },
    {
      "epoch": 0.16660840033960944,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3511536662664848e-05,
      "loss": 0.367,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 834,
      "tokens_per_second_per_gpu": 16966.15,
      "total_tokens": 75487943
    },
    {
      "epoch": 0.16680817060380562,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3496050136588135e-05,
      "loss": 0.3984,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 835,
      "tokens_per_second_per_gpu": 19142.12,
      "total_tokens": 75579927
    },
    {
      "epoch": 0.1670079408680018,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3480554052443847e-05,
      "loss": 0.4332,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 836,
      "tokens_per_second_per_gpu": 20051.07,
      "total_tokens": 75676493
    },
    {
      "epoch": 0.16720771113219798,
      "grad_norm": 0.75,
      "learning_rate": 1.3465048452597682e-05,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 837,
      "tokens_per_second_per_gpu": 17728.62,
      "total_tokens": 75763213
    },
    {
      "epoch": 0.16740748139639414,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3449533379441359e-05,
      "loss": 0.3786,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 838,
      "tokens_per_second_per_gpu": 17830.31,
      "total_tokens": 75850458
    },
    {
      "epoch": 0.16760725166059032,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3434008875392499e-05,
      "loss": 0.4006,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 839,
      "tokens_per_second_per_gpu": 19398.51,
      "total_tokens": 75943349
    },
    {
      "epoch": 0.1678070219247865,
      "grad_norm": 0.71875,
      "learning_rate": 1.3418474982894499e-05,
      "loss": 0.396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 840,
      "tokens_per_second_per_gpu": 19824.11,
      "total_tokens": 76038986
    },
    {
      "epoch": 0.16800679218898268,
      "grad_norm": 0.71875,
      "learning_rate": 1.3402931744416432e-05,
      "loss": 0.4134,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 841,
      "tokens_per_second_per_gpu": 18810.02,
      "total_tokens": 76131286
    },
    {
      "epoch": 0.16820656245317883,
      "grad_norm": 0.71875,
      "learning_rate": 1.3387379202452917e-05,
      "loss": 0.4062,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 842,
      "tokens_per_second_per_gpu": 17814.92,
      "total_tokens": 76217454
    },
    {
      "epoch": 0.168406332717375,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3371817399524006e-05,
      "loss": 0.4217,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 843,
      "tokens_per_second_per_gpu": 19823.49,
      "total_tokens": 76313247
    },
    {
      "epoch": 0.1686061029815712,
      "grad_norm": 0.7421875,
      "learning_rate": 1.335624637817508e-05,
      "loss": 0.4191,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 844,
      "tokens_per_second_per_gpu": 18905.75,
      "total_tokens": 76404392
    },
    {
      "epoch": 0.16880587324576737,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3340666180976713e-05,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 845,
      "tokens_per_second_per_gpu": 18967.16,
      "total_tokens": 76496804
    },
    {
      "epoch": 0.16900564350996355,
      "grad_norm": 0.7109375,
      "learning_rate": 1.332507685052457e-05,
      "loss": 0.3806,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 846,
      "tokens_per_second_per_gpu": 18652.04,
      "total_tokens": 76586333
    },
    {
      "epoch": 0.1692054137741597,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3309478429439284e-05,
      "loss": 0.3932,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 847,
      "tokens_per_second_per_gpu": 18553.12,
      "total_tokens": 76677666
    },
    {
      "epoch": 0.1694051840383559,
      "grad_norm": 0.703125,
      "learning_rate": 1.3293870960366346e-05,
      "loss": 0.4266,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 848,
      "tokens_per_second_per_gpu": 19293.28,
      "total_tokens": 76771218
    },
    {
      "epoch": 0.16960495430255207,
      "grad_norm": 0.703125,
      "learning_rate": 1.3278254485975977e-05,
      "loss": 0.3749,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 849,
      "tokens_per_second_per_gpu": 17919.49,
      "total_tokens": 76858192
    },
    {
      "epoch": 0.16980472456674825,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3262629048963028e-05,
      "loss": 0.3523,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 850,
      "tokens_per_second_per_gpu": 16521.09,
      "total_tokens": 76939904
    },
    {
      "epoch": 0.1700044948309444,
      "grad_norm": 0.703125,
      "learning_rate": 1.3246994692046837e-05,
      "loss": 0.4331,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 851,
      "tokens_per_second_per_gpu": 19812.69,
      "total_tokens": 77037029
    },
    {
      "epoch": 0.17020426509514058,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3231351457971144e-05,
      "loss": 0.4245,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 852,
      "tokens_per_second_per_gpu": 18224.33,
      "total_tokens": 77125915
    },
    {
      "epoch": 0.17040403535933676,
      "grad_norm": 0.71875,
      "learning_rate": 1.3215699389503956e-05,
      "loss": 0.4209,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 853,
      "tokens_per_second_per_gpu": 20420.07,
      "total_tokens": 77224596
    },
    {
      "epoch": 0.17060380562353294,
      "grad_norm": 0.6875,
      "learning_rate": 1.3200038529437427e-05,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 854,
      "tokens_per_second_per_gpu": 19820.4,
      "total_tokens": 77319374
    },
    {
      "epoch": 0.1708035758877291,
      "grad_norm": 0.88671875,
      "learning_rate": 1.3184368920587756e-05,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 855,
      "tokens_per_second_per_gpu": 17420.69,
      "total_tokens": 77404714
    },
    {
      "epoch": 0.17100334615192528,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3168690605795044e-05,
      "loss": 0.3642,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 856,
      "tokens_per_second_per_gpu": 17099.79,
      "total_tokens": 77488316
    },
    {
      "epoch": 0.17120311641612146,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3153003627923217e-05,
      "loss": 0.3801,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 857,
      "tokens_per_second_per_gpu": 17318.5,
      "total_tokens": 77572844
    },
    {
      "epoch": 0.17140288668031764,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3137308029859865e-05,
      "loss": 0.4204,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 858,
      "tokens_per_second_per_gpu": 17546.26,
      "total_tokens": 77659126
    },
    {
      "epoch": 0.17160265694451382,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3121603854516161e-05,
      "loss": 0.4019,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 859,
      "tokens_per_second_per_gpu": 20465.95,
      "total_tokens": 77757675
    },
    {
      "epoch": 0.17180242720870997,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3105891144826715e-05,
      "loss": 0.4053,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 860,
      "tokens_per_second_per_gpu": 19316.12,
      "total_tokens": 77850143
    },
    {
      "epoch": 0.17200219747290615,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.4276,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 861,
      "tokens_per_second_per_gpu": 20067.05,
      "total_tokens": 77947230
    },
    {
      "epoch": 0.17220196773710233,
      "grad_norm": 0.6796875,
      "learning_rate": 1.307444029426561e-05,
      "loss": 0.4187,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 862,
      "tokens_per_second_per_gpu": 19840.97,
      "total_tokens": 78044061
    },
    {
      "epoch": 0.17240173800129852,
      "grad_norm": 0.703125,
      "learning_rate": 1.3058702239379374e-05,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 863,
      "tokens_per_second_per_gpu": 18307.64,
      "total_tokens": 78133509
    },
    {
      "epoch": 0.17260150826549467,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3042955822118017e-05,
      "loss": 0.4127,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 864,
      "tokens_per_second_per_gpu": 19223.76,
      "total_tokens": 78227629
    },
    {
      "epoch": 0.17280127852969085,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3027201085531633e-05,
      "loss": 0.3908,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 865,
      "tokens_per_second_per_gpu": 16993.07,
      "total_tokens": 78310174
    },
    {
      "epoch": 0.17300104879388703,
      "grad_norm": 0.71875,
      "learning_rate": 1.3011438072693077e-05,
      "loss": 0.3855,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 866,
      "tokens_per_second_per_gpu": 17928.59,
      "total_tokens": 78396553
    },
    {
      "epoch": 0.1732008190580832,
      "grad_norm": 0.7109375,
      "learning_rate": 1.299566682669782e-05,
      "loss": 0.414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 867,
      "tokens_per_second_per_gpu": 18484.71,
      "total_tokens": 78486226
    },
    {
      "epoch": 0.1734005893222794,
      "grad_norm": 0.75,
      "learning_rate": 1.2979887390663847e-05,
      "loss": 0.4227,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 868,
      "tokens_per_second_per_gpu": 18619.89,
      "total_tokens": 78574993
    },
    {
      "epoch": 0.17360035958647554,
      "grad_norm": 0.71875,
      "learning_rate": 1.2964099807731539e-05,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 869,
      "tokens_per_second_per_gpu": 19102.41,
      "total_tokens": 78666803
    },
    {
      "epoch": 0.17380012985067173,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2948304121063538e-05,
      "loss": 0.4142,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 870,
      "tokens_per_second_per_gpu": 18422.96,
      "total_tokens": 78758685
    },
    {
      "epoch": 0.1739999001148679,
      "grad_norm": 0.71875,
      "learning_rate": 1.293250037384465e-05,
      "loss": 0.4169,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 871,
      "tokens_per_second_per_gpu": 20494.28,
      "total_tokens": 78857115
    },
    {
      "epoch": 0.1741996703790641,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2916688609281716e-05,
      "loss": 0.393,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 872,
      "tokens_per_second_per_gpu": 17586.25,
      "total_tokens": 78942992
    },
    {
      "epoch": 0.17439944064326024,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2900868870603502e-05,
      "loss": 0.3793,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 873,
      "tokens_per_second_per_gpu": 19465.81,
      "total_tokens": 79036354
    },
    {
      "epoch": 0.17459921090745642,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2885041201060561e-05,
      "loss": 0.4034,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 874,
      "tokens_per_second_per_gpu": 18836.59,
      "total_tokens": 79128309
    },
    {
      "epoch": 0.1747989811716526,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2869205643925142e-05,
      "loss": 0.3552,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 875,
      "tokens_per_second_per_gpu": 17401.03,
      "total_tokens": 79213267
    },
    {
      "epoch": 0.17499875143584878,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2853362242491054e-05,
      "loss": 0.395,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 876,
      "tokens_per_second_per_gpu": 19512.3,
      "total_tokens": 79309216
    },
    {
      "epoch": 0.17519852170004493,
      "grad_norm": 0.7109375,
      "learning_rate": 1.283751104007355e-05,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 877,
      "tokens_per_second_per_gpu": 18733.26,
      "total_tokens": 79401034
    },
    {
      "epoch": 0.17539829196424112,
      "grad_norm": 0.703125,
      "learning_rate": 1.2821652080009214e-05,
      "loss": 0.3985,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 878,
      "tokens_per_second_per_gpu": 18683.47,
      "total_tokens": 79492767
    },
    {
      "epoch": 0.1755980622284373,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2805785405655833e-05,
      "loss": 0.4188,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 879,
      "tokens_per_second_per_gpu": 20618.76,
      "total_tokens": 79591144
    },
    {
      "epoch": 0.17579783249263348,
      "grad_norm": 0.6875,
      "learning_rate": 1.2789911060392295e-05,
      "loss": 0.4249,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 880,
      "tokens_per_second_per_gpu": 19942.79,
      "total_tokens": 79688527
    },
    {
      "epoch": 0.17599760275682966,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2774029087618448e-05,
      "loss": 0.3708,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 881,
      "tokens_per_second_per_gpu": 19451.97,
      "total_tokens": 79781260
    },
    {
      "epoch": 0.1761973730210258,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2758139530755001e-05,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 882,
      "tokens_per_second_per_gpu": 17468.81,
      "total_tokens": 79866556
    },
    {
      "epoch": 0.176397143285222,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2742242433243396e-05,
      "loss": 0.3913,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 883,
      "tokens_per_second_per_gpu": 18931.14,
      "total_tokens": 79957402
    },
    {
      "epoch": 0.17659691354941817,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2726337838545692e-05,
      "loss": 0.4235,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 884,
      "tokens_per_second_per_gpu": 19037.58,
      "total_tokens": 80050617
    },
    {
      "epoch": 0.17679668381361435,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2710425790144445e-05,
      "loss": 0.3834,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 885,
      "tokens_per_second_per_gpu": 19262.57,
      "total_tokens": 80144396
    },
    {
      "epoch": 0.1769964540778105,
      "grad_norm": 0.6875,
      "learning_rate": 1.269450633154258e-05,
      "loss": 0.4,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 886,
      "tokens_per_second_per_gpu": 19249.77,
      "total_tokens": 80237536
    },
    {
      "epoch": 0.1771962243420067,
      "grad_norm": 0.6875,
      "learning_rate": 1.2678579506263299e-05,
      "loss": 0.411,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 887,
      "tokens_per_second_per_gpu": 19888.1,
      "total_tokens": 80333367
    },
    {
      "epoch": 0.17739599460620287,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2662645357849925e-05,
      "loss": 0.4196,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 888,
      "tokens_per_second_per_gpu": 18535.2,
      "total_tokens": 80423213
    },
    {
      "epoch": 0.17759576487039905,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2646703929865817e-05,
      "loss": 0.4046,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 889,
      "tokens_per_second_per_gpu": 18541.02,
      "total_tokens": 80513631
    },
    {
      "epoch": 0.17779553513459523,
      "grad_norm": 0.765625,
      "learning_rate": 1.263075526589423e-05,
      "loss": 0.4172,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 890,
      "tokens_per_second_per_gpu": 19513.87,
      "total_tokens": 80608753
    },
    {
      "epoch": 0.17799530539879138,
      "grad_norm": 0.7109375,
      "learning_rate": 1.26147994095382e-05,
      "loss": 0.4019,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 891,
      "tokens_per_second_per_gpu": 18937.46,
      "total_tokens": 80700013
    },
    {
      "epoch": 0.17819507566298756,
      "grad_norm": 0.6953125,
      "learning_rate": 1.259883640442043e-05,
      "loss": 0.3735,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 892,
      "tokens_per_second_per_gpu": 18523.64,
      "total_tokens": 80789519
    },
    {
      "epoch": 0.17839484592718374,
      "grad_norm": 0.75,
      "learning_rate": 1.2582866294183167e-05,
      "loss": 0.395,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 893,
      "tokens_per_second_per_gpu": 17665.29,
      "total_tokens": 80875392
    },
    {
      "epoch": 0.17859461619137992,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2566889122488084e-05,
      "loss": 0.3964,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 894,
      "tokens_per_second_per_gpu": 18933.66,
      "total_tokens": 80966297
    },
    {
      "epoch": 0.17879438645557608,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2550904933016152e-05,
      "loss": 0.4371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 895,
      "tokens_per_second_per_gpu": 18528.44,
      "total_tokens": 81055932
    },
    {
      "epoch": 0.17899415671977226,
      "grad_norm": 0.69921875,
      "learning_rate": 1.253491376946754e-05,
      "loss": 0.3954,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 896,
      "tokens_per_second_per_gpu": 18946.36,
      "total_tokens": 81148170
    },
    {
      "epoch": 0.17919392698396844,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2518915675561482e-05,
      "loss": 0.4227,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 897,
      "tokens_per_second_per_gpu": 18129.08,
      "total_tokens": 81236741
    },
    {
      "epoch": 0.17939369724816462,
      "grad_norm": 0.75,
      "learning_rate": 1.2502910695036149e-05,
      "loss": 0.3981,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 898,
      "tokens_per_second_per_gpu": 19690.78,
      "total_tokens": 81331709
    },
    {
      "epoch": 0.17959346751236077,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2486898871648552e-05,
      "loss": 0.3942,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 899,
      "tokens_per_second_per_gpu": 17869.38,
      "total_tokens": 81419062
    },
    {
      "epoch": 0.17979323777655695,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2470880249174397e-05,
      "loss": 0.3924,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 900,
      "tokens_per_second_per_gpu": 18631.18,
      "total_tokens": 81509870
    },
    {
      "epoch": 0.17999300804075313,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2454854871407993e-05,
      "loss": 0.3977,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 901,
      "tokens_per_second_per_gpu": 17449.88,
      "total_tokens": 81595445
    },
    {
      "epoch": 0.1801927783049493,
      "grad_norm": 0.7421875,
      "learning_rate": 1.243882278216211e-05,
      "loss": 0.4041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 902,
      "tokens_per_second_per_gpu": 17040.71,
      "total_tokens": 81679164
    },
    {
      "epoch": 0.1803925485691455,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2422784025267864e-05,
      "loss": 0.4175,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 903,
      "tokens_per_second_per_gpu": 19366.81,
      "total_tokens": 81773383
    },
    {
      "epoch": 0.18059231883334165,
      "grad_norm": 0.734375,
      "learning_rate": 1.2406738644574612e-05,
      "loss": 0.3796,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 904,
      "tokens_per_second_per_gpu": 17882.63,
      "total_tokens": 81860225
    },
    {
      "epoch": 0.18079208909753783,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2390686683949799e-05,
      "loss": 0.4329,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 905,
      "tokens_per_second_per_gpu": 20437.31,
      "total_tokens": 81958816
    },
    {
      "epoch": 0.180991859361734,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2374628187278888e-05,
      "loss": 0.4192,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 906,
      "tokens_per_second_per_gpu": 20000.29,
      "total_tokens": 82055912
    },
    {
      "epoch": 0.1811916296259302,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2358563198465184e-05,
      "loss": 0.4131,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 907,
      "tokens_per_second_per_gpu": 18809.01,
      "total_tokens": 82148094
    },
    {
      "epoch": 0.18139139989012634,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2342491761429761e-05,
      "loss": 0.3944,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 908,
      "tokens_per_second_per_gpu": 19103.14,
      "total_tokens": 82239179
    },
    {
      "epoch": 0.18159117015432252,
      "grad_norm": 0.765625,
      "learning_rate": 1.2326413920111304e-05,
      "loss": 0.4229,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 909,
      "tokens_per_second_per_gpu": 17635.79,
      "total_tokens": 82324190
    },
    {
      "epoch": 0.1817909404185187,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2310329718466023e-05,
      "loss": 0.3683,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 910,
      "tokens_per_second_per_gpu": 17463.19,
      "total_tokens": 82410364
    },
    {
      "epoch": 0.18199071068271488,
      "grad_norm": 0.84375,
      "learning_rate": 1.2294239200467516e-05,
      "loss": 0.3972,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 911,
      "tokens_per_second_per_gpu": 17629.18,
      "total_tokens": 82495282
    },
    {
      "epoch": 0.18219048094691107,
      "grad_norm": 0.79296875,
      "learning_rate": 1.227814241010664e-05,
      "loss": 0.411,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 912,
      "tokens_per_second_per_gpu": 19568.4,
      "total_tokens": 82590557
    },
    {
      "epoch": 0.18239025121110722,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2262039391391405e-05,
      "loss": 0.3838,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 913,
      "tokens_per_second_per_gpu": 18253.42,
      "total_tokens": 82679828
    },
    {
      "epoch": 0.1825900214753034,
      "grad_norm": 0.6796875,
      "learning_rate": 1.2245930188346849e-05,
      "loss": 0.3897,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 914,
      "tokens_per_second_per_gpu": 19783.82,
      "total_tokens": 82775006
    },
    {
      "epoch": 0.18278979173949958,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2229814845014918e-05,
      "loss": 0.3992,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 915,
      "tokens_per_second_per_gpu": 18843.07,
      "total_tokens": 82866827
    },
    {
      "epoch": 0.18298956200369576,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2213693405454345e-05,
      "loss": 0.3885,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 916,
      "tokens_per_second_per_gpu": 18476.11,
      "total_tokens": 82956412
    },
    {
      "epoch": 0.1831893322678919,
      "grad_norm": 0.6875,
      "learning_rate": 1.2197565913740531e-05,
      "loss": 0.4029,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 917,
      "tokens_per_second_per_gpu": 20133.4,
      "total_tokens": 83053357
    },
    {
      "epoch": 0.1833891025320881,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2181432413965428e-05,
      "loss": 0.4074,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 918,
      "tokens_per_second_per_gpu": 19008.57,
      "total_tokens": 83145828
    },
    {
      "epoch": 0.18358887279628427,
      "grad_norm": 0.7109375,
      "learning_rate": 1.21652929502374e-05,
      "loss": 0.3892,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 919,
      "tokens_per_second_per_gpu": 18318.51,
      "total_tokens": 83237582
    },
    {
      "epoch": 0.18378864306048046,
      "grad_norm": 0.703125,
      "learning_rate": 1.2149147566681126e-05,
      "loss": 0.3894,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 920,
      "tokens_per_second_per_gpu": 18402.88,
      "total_tokens": 83327880
    },
    {
      "epoch": 0.1839884133246766,
      "grad_norm": 0.74609375,
      "learning_rate": 1.213299630743747e-05,
      "loss": 0.3927,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 921,
      "tokens_per_second_per_gpu": 17722.88,
      "total_tokens": 83413938
    },
    {
      "epoch": 0.1841881835888728,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2116839216663358e-05,
      "loss": 0.3736,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 922,
      "tokens_per_second_per_gpu": 17571.24,
      "total_tokens": 83498806
    },
    {
      "epoch": 0.18438795385306897,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2100676338531661e-05,
      "loss": 0.3243,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 923,
      "tokens_per_second_per_gpu": 16528.71,
      "total_tokens": 83579748
    },
    {
      "epoch": 0.18458772411726515,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2084507717231064e-05,
      "loss": 0.4018,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 924,
      "tokens_per_second_per_gpu": 18692.64,
      "total_tokens": 83669186
    },
    {
      "epoch": 0.18478749438146133,
      "grad_norm": 0.6875,
      "learning_rate": 1.2068333396965968e-05,
      "loss": 0.4086,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 925,
      "tokens_per_second_per_gpu": 19772.04,
      "total_tokens": 83766147
    },
    {
      "epoch": 0.18498726464565748,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2052153421956343e-05,
      "loss": 0.3631,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 926,
      "tokens_per_second_per_gpu": 16380.97,
      "total_tokens": 83846857
    },
    {
      "epoch": 0.18518703490985366,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2035967836437625e-05,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 927,
      "tokens_per_second_per_gpu": 19617.09,
      "total_tokens": 83942259
    },
    {
      "epoch": 0.18538680517404985,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2019776684660584e-05,
      "loss": 0.3914,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 928,
      "tokens_per_second_per_gpu": 18234.87,
      "total_tokens": 84030777
    },
    {
      "epoch": 0.18558657543824603,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2003580010891214e-05,
      "loss": 0.3767,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 929,
      "tokens_per_second_per_gpu": 19702.87,
      "total_tokens": 84125441
    },
    {
      "epoch": 0.18578634570244218,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1987377859410598e-05,
      "loss": 0.4139,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 930,
      "tokens_per_second_per_gpu": 19260.11,
      "total_tokens": 84218470
    },
    {
      "epoch": 0.18598611596663836,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1971170274514802e-05,
      "loss": 0.4095,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 931,
      "tokens_per_second_per_gpu": 18758.45,
      "total_tokens": 84309542
    },
    {
      "epoch": 0.18618588623083454,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1954957300514749e-05,
      "loss": 0.3616,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 932,
      "tokens_per_second_per_gpu": 18230.28,
      "total_tokens": 84397975
    },
    {
      "epoch": 0.18638565649503072,
      "grad_norm": 0.71875,
      "learning_rate": 1.1938738981736084e-05,
      "loss": 0.3759,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 933,
      "tokens_per_second_per_gpu": 16823.95,
      "total_tokens": 84482034
    },
    {
      "epoch": 0.1865854267592269,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1922515362519072e-05,
      "loss": 0.4269,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 934,
      "tokens_per_second_per_gpu": 19616.77,
      "total_tokens": 84577531
    },
    {
      "epoch": 0.18678519702342306,
      "grad_norm": 0.71875,
      "learning_rate": 1.190628648721847e-05,
      "loss": 0.4134,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 935,
      "tokens_per_second_per_gpu": 19327.23,
      "total_tokens": 84671640
    },
    {
      "epoch": 0.18698496728761924,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1890052400203405e-05,
      "loss": 0.3818,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 936,
      "tokens_per_second_per_gpu": 18336.55,
      "total_tokens": 84762093
    },
    {
      "epoch": 0.18718473755181542,
      "grad_norm": 0.72265625,
      "learning_rate": 1.187381314585725e-05,
      "loss": 0.3846,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 937,
      "tokens_per_second_per_gpu": 17349.88,
      "total_tokens": 84847100
    },
    {
      "epoch": 0.1873845078160116,
      "grad_norm": 0.71484375,
      "learning_rate": 1.18575687685775e-05,
      "loss": 0.4267,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 938,
      "tokens_per_second_per_gpu": 19632.75,
      "total_tokens": 84942636
    },
    {
      "epoch": 0.18758427808020775,
      "grad_norm": 0.8515625,
      "learning_rate": 1.1841319312775672e-05,
      "loss": 0.3865,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 939,
      "tokens_per_second_per_gpu": 18202.15,
      "total_tokens": 85030618
    },
    {
      "epoch": 0.18778404834440393,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1825064822877153e-05,
      "loss": 0.3993,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 940,
      "tokens_per_second_per_gpu": 18634.73,
      "total_tokens": 85119900
    },
    {
      "epoch": 0.1879838186086001,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1808805343321102e-05,
      "loss": 0.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 941,
      "tokens_per_second_per_gpu": 18874.95,
      "total_tokens": 85213060
    },
    {
      "epoch": 0.1881835888727963,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1792540918560306e-05,
      "loss": 0.3954,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 942,
      "tokens_per_second_per_gpu": 19553.29,
      "total_tokens": 85305904
    },
    {
      "epoch": 0.18838335913699245,
      "grad_norm": 0.703125,
      "learning_rate": 1.1776271593061089e-05,
      "loss": 0.3822,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 943,
      "tokens_per_second_per_gpu": 17796.17,
      "total_tokens": 85393428
    },
    {
      "epoch": 0.18858312940118863,
      "grad_norm": 0.7265625,
      "learning_rate": 1.175999741130316e-05,
      "loss": 0.4062,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 944,
      "tokens_per_second_per_gpu": 19272.23,
      "total_tokens": 85487333
    },
    {
      "epoch": 0.1887828996653848,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1743718417779518e-05,
      "loss": 0.4148,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 945,
      "tokens_per_second_per_gpu": 17654.04,
      "total_tokens": 85573877
    },
    {
      "epoch": 0.188982669929581,
      "grad_norm": 0.93359375,
      "learning_rate": 1.1727434656996306e-05,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 946,
      "tokens_per_second_per_gpu": 16167.46,
      "total_tokens": 85653244
    },
    {
      "epoch": 0.18918244019377717,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1711146173472701e-05,
      "loss": 0.3841,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 947,
      "tokens_per_second_per_gpu": 17160.96,
      "total_tokens": 85737664
    },
    {
      "epoch": 0.18938221045797332,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1694853011740798e-05,
      "loss": 0.4002,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 948,
      "tokens_per_second_per_gpu": 18003.84,
      "total_tokens": 85826385
    },
    {
      "epoch": 0.1895819807221695,
      "grad_norm": 0.71875,
      "learning_rate": 1.1678555216345478e-05,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 949,
      "tokens_per_second_per_gpu": 18904.47,
      "total_tokens": 85917435
    },
    {
      "epoch": 0.18978175098636568,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1662252831844297e-05,
      "loss": 0.3949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 950,
      "tokens_per_second_per_gpu": 17467.6,
      "total_tokens": 86003587
    },
    {
      "epoch": 0.18998152125056186,
      "grad_norm": 0.74609375,
      "learning_rate": 1.164594590280734e-05,
      "loss": 0.385,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 951,
      "tokens_per_second_per_gpu": 18069.86,
      "total_tokens": 86089759
    },
    {
      "epoch": 0.19018129151475802,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1629634473817138e-05,
      "loss": 0.4092,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 952,
      "tokens_per_second_per_gpu": 18179.22,
      "total_tokens": 86176895
    },
    {
      "epoch": 0.1903810617789542,
      "grad_norm": 0.703125,
      "learning_rate": 1.1613318589468512e-05,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 953,
      "tokens_per_second_per_gpu": 18815.26,
      "total_tokens": 86267471
    },
    {
      "epoch": 0.19058083204315038,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1596998294368466e-05,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 954,
      "tokens_per_second_per_gpu": 18447.65,
      "total_tokens": 86357293
    },
    {
      "epoch": 0.19078060230734656,
      "grad_norm": 0.703125,
      "learning_rate": 1.1580673633136066e-05,
      "loss": 0.4116,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 955,
      "tokens_per_second_per_gpu": 18928.85,
      "total_tokens": 86449856
    },
    {
      "epoch": 0.19098037257154274,
      "grad_norm": 0.76953125,
      "learning_rate": 1.156434465040231e-05,
      "loss": 0.3921,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 956,
      "tokens_per_second_per_gpu": 16280.78,
      "total_tokens": 86529787
    },
    {
      "epoch": 0.1911801428357389,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1548011390810016e-05,
      "loss": 0.3961,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 957,
      "tokens_per_second_per_gpu": 18740.65,
      "total_tokens": 86620727
    },
    {
      "epoch": 0.19137991309993507,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1531673899013687e-05,
      "loss": 0.3728,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 958,
      "tokens_per_second_per_gpu": 17575.86,
      "total_tokens": 86706361
    },
    {
      "epoch": 0.19157968336413125,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1515332219679405e-05,
      "loss": 0.4056,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 959,
      "tokens_per_second_per_gpu": 18083.34,
      "total_tokens": 86794610
    },
    {
      "epoch": 0.19177945362832743,
      "grad_norm": 0.703125,
      "learning_rate": 1.14989863974847e-05,
      "loss": 0.3579,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 960,
      "tokens_per_second_per_gpu": 18168.32,
      "total_tokens": 86883555
    },
    {
      "epoch": 0.1919792238925236,
      "grad_norm": 0.6796875,
      "learning_rate": 1.148263647711842e-05,
      "loss": 0.4199,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 961,
      "tokens_per_second_per_gpu": 20845.0,
      "total_tokens": 86984653
    },
    {
      "epoch": 0.19217899415671977,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1466282503280631e-05,
      "loss": 0.4115,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 962,
      "tokens_per_second_per_gpu": 17959.79,
      "total_tokens": 87073323
    },
    {
      "epoch": 0.19237876442091595,
      "grad_norm": 0.69140625,
      "learning_rate": 1.144992452068246e-05,
      "loss": 0.3591,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 963,
      "tokens_per_second_per_gpu": 18675.13,
      "total_tokens": 87162272
    },
    {
      "epoch": 0.19257853468511213,
      "grad_norm": 0.75,
      "learning_rate": 1.143356257404602e-05,
      "loss": 0.3758,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 964,
      "tokens_per_second_per_gpu": 16151.03,
      "total_tokens": 87240906
    },
    {
      "epoch": 0.19277830494930828,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1417196708104244e-05,
      "loss": 0.4082,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 965,
      "tokens_per_second_per_gpu": 18104.97,
      "total_tokens": 87329796
    },
    {
      "epoch": 0.19297807521350446,
      "grad_norm": 0.69140625,
      "learning_rate": 1.140082696760078e-05,
      "loss": 0.4001,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 966,
      "tokens_per_second_per_gpu": 19046.66,
      "total_tokens": 87420762
    },
    {
      "epoch": 0.19317784547770064,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1384453397289876e-05,
      "loss": 0.377,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 967,
      "tokens_per_second_per_gpu": 16669.73,
      "total_tokens": 87502383
    },
    {
      "epoch": 0.19337761574189682,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1368076041936252e-05,
      "loss": 0.4414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 968,
      "tokens_per_second_per_gpu": 17951.44,
      "total_tokens": 87590529
    },
    {
      "epoch": 0.193577386006093,
      "grad_norm": 1.0625,
      "learning_rate": 1.135169494631497e-05,
      "loss": 0.4154,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 969,
      "tokens_per_second_per_gpu": 20549.85,
      "total_tokens": 87689021
    },
    {
      "epoch": 0.19377715627028916,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1335310155211319e-05,
      "loss": 0.4113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 970,
      "tokens_per_second_per_gpu": 17162.8,
      "total_tokens": 87773754
    },
    {
      "epoch": 0.19397692653448534,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1318921713420691e-05,
      "loss": 0.3807,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 971,
      "tokens_per_second_per_gpu": 18189.88,
      "total_tokens": 87863789
    },
    {
      "epoch": 0.19417669679868152,
      "grad_norm": 0.90234375,
      "learning_rate": 1.130252966574846e-05,
      "loss": 0.3965,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 972,
      "tokens_per_second_per_gpu": 17876.18,
      "total_tokens": 87951641
    },
    {
      "epoch": 0.1943764670628777,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1286134057009862e-05,
      "loss": 0.3734,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 973,
      "tokens_per_second_per_gpu": 17120.89,
      "total_tokens": 88035416
    },
    {
      "epoch": 0.19457623732707385,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1269734932029864e-05,
      "loss": 0.3685,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 974,
      "tokens_per_second_per_gpu": 18006.71,
      "total_tokens": 88122870
    },
    {
      "epoch": 0.19477600759127003,
      "grad_norm": 0.765625,
      "learning_rate": 1.1253332335643043e-05,
      "loss": 0.4098,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 975,
      "tokens_per_second_per_gpu": 20212.25,
      "total_tokens": 88219310
    },
    {
      "epoch": 0.19497577785546621,
      "grad_norm": 0.70703125,
      "learning_rate": 1.123692631269348e-05,
      "loss": 0.4019,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 976,
      "tokens_per_second_per_gpu": 19672.81,
      "total_tokens": 88314420
    },
    {
      "epoch": 0.1951755481196624,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1220516908034602e-05,
      "loss": 0.3999,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 977,
      "tokens_per_second_per_gpu": 18643.12,
      "total_tokens": 88404130
    },
    {
      "epoch": 0.19537531838385858,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1204104166529111e-05,
      "loss": 0.4035,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 978,
      "tokens_per_second_per_gpu": 18680.25,
      "total_tokens": 88495075
    },
    {
      "epoch": 0.19557508864805473,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1187688133048801e-05,
      "loss": 0.3932,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 979,
      "tokens_per_second_per_gpu": 19726.0,
      "total_tokens": 88591258
    },
    {
      "epoch": 0.1957748589122509,
      "grad_norm": 0.703125,
      "learning_rate": 1.1171268852474492e-05,
      "loss": 0.3681,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 980,
      "tokens_per_second_per_gpu": 18835.37,
      "total_tokens": 88681207
    },
    {
      "epoch": 0.1959746291764471,
      "grad_norm": 0.734375,
      "learning_rate": 1.1154846369695864e-05,
      "loss": 0.357,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 981,
      "tokens_per_second_per_gpu": 16827.74,
      "total_tokens": 88762038
    },
    {
      "epoch": 0.19617439944064327,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1138420729611355e-05,
      "loss": 0.4254,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 982,
      "tokens_per_second_per_gpu": 20578.94,
      "total_tokens": 88861435
    },
    {
      "epoch": 0.19637416970483942,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1121991977128046e-05,
      "loss": 0.3835,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 983,
      "tokens_per_second_per_gpu": 18728.24,
      "total_tokens": 88952834
    },
    {
      "epoch": 0.1965739399690356,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1105560157161512e-05,
      "loss": 0.4052,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 984,
      "tokens_per_second_per_gpu": 18298.9,
      "total_tokens": 89041916
    },
    {
      "epoch": 0.19677371023323179,
      "grad_norm": 0.703125,
      "learning_rate": 1.1089125314635727e-05,
      "loss": 0.3745,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 985,
      "tokens_per_second_per_gpu": 17811.69,
      "total_tokens": 89129190
    },
    {
      "epoch": 0.19697348049742797,
      "grad_norm": 0.6875,
      "learning_rate": 1.107268749448292e-05,
      "loss": 0.4282,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 986,
      "tokens_per_second_per_gpu": 18311.77,
      "total_tokens": 89219784
    },
    {
      "epoch": 0.19717325076162412,
      "grad_norm": 0.72265625,
      "learning_rate": 1.105624674164346e-05,
      "loss": 0.3502,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 987,
      "tokens_per_second_per_gpu": 16318.44,
      "total_tokens": 89299389
    },
    {
      "epoch": 0.1973730210258203,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1039803101065742e-05,
      "loss": 0.4236,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 988,
      "tokens_per_second_per_gpu": 19177.83,
      "total_tokens": 89394282
    },
    {
      "epoch": 0.19757279129001648,
      "grad_norm": 0.703125,
      "learning_rate": 1.1023356617706051e-05,
      "loss": 0.3799,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 989,
      "tokens_per_second_per_gpu": 17666.48,
      "total_tokens": 89481121
    },
    {
      "epoch": 0.19777256155421266,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1006907336528445e-05,
      "loss": 0.4273,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 990,
      "tokens_per_second_per_gpu": 19956.34,
      "total_tokens": 89579295
    },
    {
      "epoch": 0.19797233181840884,
      "grad_norm": 0.70703125,
      "learning_rate": 1.099045530250463e-05,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 991,
      "tokens_per_second_per_gpu": 18717.0,
      "total_tokens": 89670527
    },
    {
      "epoch": 0.198172102082605,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0974000560613835e-05,
      "loss": 0.3695,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 992,
      "tokens_per_second_per_gpu": 17274.06,
      "total_tokens": 89755091
    },
    {
      "epoch": 0.19837187234680118,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0957543155842703e-05,
      "loss": 0.406,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 993,
      "tokens_per_second_per_gpu": 19089.87,
      "total_tokens": 89847083
    },
    {
      "epoch": 0.19857164261099736,
      "grad_norm": 0.734375,
      "learning_rate": 1.0941083133185146e-05,
      "loss": 0.3911,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 994,
      "tokens_per_second_per_gpu": 18325.98,
      "total_tokens": 89936182
    },
    {
      "epoch": 0.19877141287519354,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0924620537642237e-05,
      "loss": 0.3752,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 995,
      "tokens_per_second_per_gpu": 17054.16,
      "total_tokens": 90019450
    },
    {
      "epoch": 0.1989711831393897,
      "grad_norm": 0.703125,
      "learning_rate": 1.0908155414222083e-05,
      "loss": 0.3961,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 996,
      "tokens_per_second_per_gpu": 18727.13,
      "total_tokens": 90111654
    },
    {
      "epoch": 0.19917095340358587,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0891687807939707e-05,
      "loss": 0.4198,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 997,
      "tokens_per_second_per_gpu": 20649.65,
      "total_tokens": 90211988
    },
    {
      "epoch": 0.19937072366778205,
      "grad_norm": 0.734375,
      "learning_rate": 1.0875217763816912e-05,
      "loss": 0.4017,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 998,
      "tokens_per_second_per_gpu": 17460.07,
      "total_tokens": 90296186
    },
    {
      "epoch": 0.19957049393197823,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0858745326882172e-05,
      "loss": 0.3975,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 999,
      "tokens_per_second_per_gpu": 17897.31,
      "total_tokens": 90381957
    },
    {
      "epoch": 0.1997702641961744,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0842270542170498e-05,
      "loss": 0.399,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1000,
      "tokens_per_second_per_gpu": 18230.47,
      "total_tokens": 90470651
    },
    {
      "epoch": 0.19997003446037057,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0825793454723325e-05,
      "loss": 0.3959,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1001,
      "tokens_per_second_per_gpu": 18074.57,
      "total_tokens": 90558554
    },
    {
      "epoch": 0.20016980472456675,
      "grad_norm": 0.6875,
      "learning_rate": 1.0809314109588377e-05,
      "loss": 0.3946,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1002,
      "tokens_per_second_per_gpu": 18104.11,
      "total_tokens": 90648152
    },
    {
      "epoch": 0.20036957498876293,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0792832551819558e-05,
      "loss": 0.3691,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1003,
      "tokens_per_second_per_gpu": 18560.73,
      "total_tokens": 90737134
    },
    {
      "epoch": 0.2005693452529591,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0776348826476821e-05,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1004,
      "tokens_per_second_per_gpu": 17565.51,
      "total_tokens": 90822641
    },
    {
      "epoch": 0.20076911551715526,
      "grad_norm": 0.6875,
      "learning_rate": 1.0759862978626032e-05,
      "loss": 0.4007,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1005,
      "tokens_per_second_per_gpu": 19407.65,
      "total_tokens": 90916574
    },
    {
      "epoch": 0.20096888578135144,
      "grad_norm": 0.734375,
      "learning_rate": 1.0743375053338879e-05,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1006,
      "tokens_per_second_per_gpu": 17394.7,
      "total_tokens": 91002117
    },
    {
      "epoch": 0.20116865604554762,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0726885095692712e-05,
      "loss": 0.3636,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1007,
      "tokens_per_second_per_gpu": 18508.72,
      "total_tokens": 91092712
    },
    {
      "epoch": 0.2013684263097438,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0710393150770451e-05,
      "loss": 0.3685,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1008,
      "tokens_per_second_per_gpu": 17918.64,
      "total_tokens": 91180316
    },
    {
      "epoch": 0.20156819657393996,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0693899263660442e-05,
      "loss": 0.3661,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1009,
      "tokens_per_second_per_gpu": 18697.9,
      "total_tokens": 91270551
    },
    {
      "epoch": 0.20176796683813614,
      "grad_norm": 0.71875,
      "learning_rate": 1.067740347945634e-05,
      "loss": 0.4155,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1010,
      "tokens_per_second_per_gpu": 18625.56,
      "total_tokens": 91361937
    },
    {
      "epoch": 0.20196773710233232,
      "grad_norm": 0.75,
      "learning_rate": 1.0660905843256995e-05,
      "loss": 0.4007,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1011,
      "tokens_per_second_per_gpu": 16921.78,
      "total_tokens": 91444694
    },
    {
      "epoch": 0.2021675073665285,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0644406400166309e-05,
      "loss": 0.401,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1012,
      "tokens_per_second_per_gpu": 17571.29,
      "total_tokens": 91531387
    },
    {
      "epoch": 0.20236727763072468,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0627905195293135e-05,
      "loss": 0.3991,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1013,
      "tokens_per_second_per_gpu": 18640.23,
      "total_tokens": 91621991
    },
    {
      "epoch": 0.20256704789492083,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0611402273751132e-05,
      "loss": 0.4144,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1014,
      "tokens_per_second_per_gpu": 19603.33,
      "total_tokens": 91717974
    },
    {
      "epoch": 0.202766818159117,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0594897680658657e-05,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1015,
      "tokens_per_second_per_gpu": 17163.2,
      "total_tokens": 91801185
    },
    {
      "epoch": 0.2029665884233132,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0578391461138642e-05,
      "loss": 0.3843,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1016,
      "tokens_per_second_per_gpu": 17988.85,
      "total_tokens": 91888162
    },
    {
      "epoch": 0.20316635868750937,
      "grad_norm": 0.734375,
      "learning_rate": 1.0561883660318456e-05,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1017,
      "tokens_per_second_per_gpu": 18320.37,
      "total_tokens": 91976772
    },
    {
      "epoch": 0.20336612895170553,
      "grad_norm": 0.6953125,
      "learning_rate": 1.05453743233298e-05,
      "loss": 0.3787,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1018,
      "tokens_per_second_per_gpu": 17901.83,
      "total_tokens": 92064833
    },
    {
      "epoch": 0.2035658992159017,
      "grad_norm": 0.71875,
      "learning_rate": 1.0528863495308568e-05,
      "loss": 0.4144,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1019,
      "tokens_per_second_per_gpu": 18246.59,
      "total_tokens": 92154796
    },
    {
      "epoch": 0.2037656694800979,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0512351221394731e-05,
      "loss": 0.4285,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1020,
      "tokens_per_second_per_gpu": 20700.7,
      "total_tokens": 92254218
    },
    {
      "epoch": 0.20396543974429407,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0495837546732224e-05,
      "loss": 0.4065,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1021,
      "tokens_per_second_per_gpu": 18245.59,
      "total_tokens": 92342256
    },
    {
      "epoch": 0.20416521000849025,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0479322516468797e-05,
      "loss": 0.3697,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1022,
      "tokens_per_second_per_gpu": 19118.07,
      "total_tokens": 92433442
    },
    {
      "epoch": 0.2043649802726864,
      "grad_norm": 0.703125,
      "learning_rate": 1.046280617575591e-05,
      "loss": 0.3714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1023,
      "tokens_per_second_per_gpu": 17398.92,
      "total_tokens": 92518307
    },
    {
      "epoch": 0.20456475053688258,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0446288569748613e-05,
      "loss": 0.3769,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1024,
      "tokens_per_second_per_gpu": 18349.46,
      "total_tokens": 92607584
    },
    {
      "epoch": 0.20476452080107876,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0429769743605406e-05,
      "loss": 0.3876,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1025,
      "tokens_per_second_per_gpu": 20057.71,
      "total_tokens": 92703084
    },
    {
      "epoch": 0.20496429106527495,
      "grad_norm": 0.703125,
      "learning_rate": 1.0413249742488132e-05,
      "loss": 0.3905,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1026,
      "tokens_per_second_per_gpu": 18196.56,
      "total_tokens": 92792601
    },
    {
      "epoch": 0.2051640613294711,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0396728611561843e-05,
      "loss": 0.4001,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1027,
      "tokens_per_second_per_gpu": 18999.72,
      "total_tokens": 92883973
    },
    {
      "epoch": 0.20536383159366728,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0380206395994678e-05,
      "loss": 0.415,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1028,
      "tokens_per_second_per_gpu": 19480.11,
      "total_tokens": 92978048
    },
    {
      "epoch": 0.20556360185786346,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0363683140957745e-05,
      "loss": 0.3553,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1029,
      "tokens_per_second_per_gpu": 17704.89,
      "total_tokens": 93064018
    },
    {
      "epoch": 0.20576337212205964,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0347158891624992e-05,
      "loss": 0.3875,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1030,
      "tokens_per_second_per_gpu": 18009.41,
      "total_tokens": 93152140
    },
    {
      "epoch": 0.2059631423862558,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0330633693173083e-05,
      "loss": 0.3997,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1031,
      "tokens_per_second_per_gpu": 19032.06,
      "total_tokens": 93243696
    },
    {
      "epoch": 0.20616291265045197,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0314107590781284e-05,
      "loss": 0.4207,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1032,
      "tokens_per_second_per_gpu": 18946.76,
      "total_tokens": 93336333
    },
    {
      "epoch": 0.20636268291464815,
      "grad_norm": 0.703125,
      "learning_rate": 1.0297580629631324e-05,
      "loss": 0.3653,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1033,
      "tokens_per_second_per_gpu": 17510.01,
      "total_tokens": 93422905
    },
    {
      "epoch": 0.20656245317884434,
      "grad_norm": 0.7109375,
      "learning_rate": 1.028105285490729e-05,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1034,
      "tokens_per_second_per_gpu": 18879.67,
      "total_tokens": 93515411
    },
    {
      "epoch": 0.20676222344304052,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0264524311795478e-05,
      "loss": 0.3486,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1035,
      "tokens_per_second_per_gpu": 17345.78,
      "total_tokens": 93599856
    },
    {
      "epoch": 0.20696199370723667,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0247995045484303e-05,
      "loss": 0.3757,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1036,
      "tokens_per_second_per_gpu": 19024.6,
      "total_tokens": 93692209
    },
    {
      "epoch": 0.20716176397143285,
      "grad_norm": 0.7265625,
      "learning_rate": 1.023146510116414e-05,
      "loss": 0.331,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1037,
      "tokens_per_second_per_gpu": 16516.62,
      "total_tokens": 93773040
    },
    {
      "epoch": 0.20736153423562903,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0214934524027228e-05,
      "loss": 0.3984,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1038,
      "tokens_per_second_per_gpu": 17856.72,
      "total_tokens": 93862439
    },
    {
      "epoch": 0.2075613044998252,
      "grad_norm": 0.71875,
      "learning_rate": 1.0198403359267538e-05,
      "loss": 0.3788,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1039,
      "tokens_per_second_per_gpu": 17475.52,
      "total_tokens": 93947322
    },
    {
      "epoch": 0.20776107476402136,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0181871652080637e-05,
      "loss": 0.3922,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1040,
      "tokens_per_second_per_gpu": 18438.53,
      "total_tokens": 94037175
    },
    {
      "epoch": 0.20796084502821754,
      "grad_norm": 0.75,
      "learning_rate": 1.0165339447663586e-05,
      "loss": 0.3667,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1041,
      "tokens_per_second_per_gpu": 16152.35,
      "total_tokens": 94116974
    },
    {
      "epoch": 0.20816061529241373,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0148806791214797e-05,
      "loss": 0.4227,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1042,
      "tokens_per_second_per_gpu": 19688.46,
      "total_tokens": 94213598
    },
    {
      "epoch": 0.2083603855566099,
      "grad_norm": 0.75,
      "learning_rate": 1.0132273727933925e-05,
      "loss": 0.3778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1043,
      "tokens_per_second_per_gpu": 17944.31,
      "total_tokens": 94299622
    },
    {
      "epoch": 0.2085601558208061,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0115740303021729e-05,
      "loss": 0.3698,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1044,
      "tokens_per_second_per_gpu": 16853.93,
      "total_tokens": 94383327
    },
    {
      "epoch": 0.20875992608500224,
      "grad_norm": 0.71875,
      "learning_rate": 1.0099206561679964e-05,
      "loss": 0.4182,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1045,
      "tokens_per_second_per_gpu": 19831.73,
      "total_tokens": 94479250
    },
    {
      "epoch": 0.20895969634919842,
      "grad_norm": 0.703125,
      "learning_rate": 1.008267254911125e-05,
      "loss": 0.3814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1046,
      "tokens_per_second_per_gpu": 18356.56,
      "total_tokens": 94569867
    },
    {
      "epoch": 0.2091594666133946,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0066138310518942e-05,
      "loss": 0.3834,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1047,
      "tokens_per_second_per_gpu": 17942.2,
      "total_tokens": 94657608
    },
    {
      "epoch": 0.20935923687759078,
      "grad_norm": 0.73046875,
      "learning_rate": 1.004960389110702e-05,
      "loss": 0.3875,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1048,
      "tokens_per_second_per_gpu": 18893.29,
      "total_tokens": 94750674
    },
    {
      "epoch": 0.20955900714178693,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0033069336079952e-05,
      "loss": 0.4055,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1049,
      "tokens_per_second_per_gpu": 18537.76,
      "total_tokens": 94840916
    },
    {
      "epoch": 0.20975877740598312,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0016534690642584e-05,
      "loss": 0.4051,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1050,
      "tokens_per_second_per_gpu": 19769.48,
      "total_tokens": 94935168
    },
    {
      "epoch": 0.2099585476701793,
      "grad_norm": 0.71875,
      "learning_rate": 1e-05,
      "loss": 0.3913,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1051,
      "tokens_per_second_per_gpu": 17659.43,
      "total_tokens": 95022986
    },
    {
      "epoch": 0.21015831793437548,
      "grad_norm": 0.7265625,
      "learning_rate": 9.983465309357418e-06,
      "loss": 0.4118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1052,
      "tokens_per_second_per_gpu": 20009.33,
      "total_tokens": 95118583
    },
    {
      "epoch": 0.21035808819857163,
      "grad_norm": 0.68359375,
      "learning_rate": 9.96693066392005e-06,
      "loss": 0.3836,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1053,
      "tokens_per_second_per_gpu": 18146.46,
      "total_tokens": 95208033
    },
    {
      "epoch": 0.2105578584627678,
      "grad_norm": 0.73046875,
      "learning_rate": 9.950396108892983e-06,
      "loss": 0.4072,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1054,
      "tokens_per_second_per_gpu": 17937.71,
      "total_tokens": 95295564
    },
    {
      "epoch": 0.210757628726964,
      "grad_norm": 0.67578125,
      "learning_rate": 9.93386168948106e-06,
      "loss": 0.3876,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1055,
      "tokens_per_second_per_gpu": 18486.6,
      "total_tokens": 95387109
    },
    {
      "epoch": 0.21095739899116017,
      "grad_norm": 0.703125,
      "learning_rate": 9.917327450888751e-06,
      "loss": 0.4071,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1056,
      "tokens_per_second_per_gpu": 19378.84,
      "total_tokens": 95481580
    },
    {
      "epoch": 0.21115716925535635,
      "grad_norm": 0.69921875,
      "learning_rate": 9.900793438320037e-06,
      "loss": 0.3927,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1057,
      "tokens_per_second_per_gpu": 19796.93,
      "total_tokens": 95575732
    },
    {
      "epoch": 0.2113569395195525,
      "grad_norm": 0.69921875,
      "learning_rate": 9.884259696978275e-06,
      "loss": 0.4291,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1058,
      "tokens_per_second_per_gpu": 19887.8,
      "total_tokens": 95672127
    },
    {
      "epoch": 0.2115567097837487,
      "grad_norm": 0.67578125,
      "learning_rate": 9.86772627206608e-06,
      "loss": 0.4002,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1059,
      "tokens_per_second_per_gpu": 20185.02,
      "total_tokens": 95770067
    },
    {
      "epoch": 0.21175648004794487,
      "grad_norm": 0.6953125,
      "learning_rate": 9.851193208785206e-06,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1060,
      "tokens_per_second_per_gpu": 19302.24,
      "total_tokens": 95864370
    },
    {
      "epoch": 0.21195625031214105,
      "grad_norm": 0.74609375,
      "learning_rate": 9.834660552336415e-06,
      "loss": 0.3948,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1061,
      "tokens_per_second_per_gpu": 16927.69,
      "total_tokens": 95946988
    },
    {
      "epoch": 0.2121560205763372,
      "grad_norm": 0.66796875,
      "learning_rate": 9.818128347919364e-06,
      "loss": 0.3784,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1062,
      "tokens_per_second_per_gpu": 18958.29,
      "total_tokens": 96037932
    },
    {
      "epoch": 0.21235579084053338,
      "grad_norm": 0.72265625,
      "learning_rate": 9.801596640732466e-06,
      "loss": 0.3702,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1063,
      "tokens_per_second_per_gpu": 17024.17,
      "total_tokens": 96121588
    },
    {
      "epoch": 0.21255556110472956,
      "grad_norm": 0.703125,
      "learning_rate": 9.785065475972773e-06,
      "loss": 0.3673,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1064,
      "tokens_per_second_per_gpu": 18230.72,
      "total_tokens": 96209846
    },
    {
      "epoch": 0.21275533136892574,
      "grad_norm": 0.75390625,
      "learning_rate": 9.768534898835864e-06,
      "loss": 0.4095,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1065,
      "tokens_per_second_per_gpu": 17544.93,
      "total_tokens": 96295921
    },
    {
      "epoch": 0.2129551016331219,
      "grad_norm": 0.67578125,
      "learning_rate": 9.7520049545157e-06,
      "loss": 0.377,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1066,
      "tokens_per_second_per_gpu": 19406.71,
      "total_tokens": 96388618
    },
    {
      "epoch": 0.21315487189731808,
      "grad_norm": 0.6953125,
      "learning_rate": 9.735475688204522e-06,
      "loss": 0.3818,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1067,
      "tokens_per_second_per_gpu": 17758.0,
      "total_tokens": 96475205
    },
    {
      "epoch": 0.21335464216151426,
      "grad_norm": 0.7265625,
      "learning_rate": 9.718947145092715e-06,
      "loss": 0.4002,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1068,
      "tokens_per_second_per_gpu": 19417.58,
      "total_tokens": 96568837
    },
    {
      "epoch": 0.21355441242571044,
      "grad_norm": 0.69140625,
      "learning_rate": 9.702419370368677e-06,
      "loss": 0.4303,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1069,
      "tokens_per_second_per_gpu": 19831.87,
      "total_tokens": 96666297
    },
    {
      "epoch": 0.21375418268990662,
      "grad_norm": 0.69921875,
      "learning_rate": 9.685892409218718e-06,
      "loss": 0.4011,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1070,
      "tokens_per_second_per_gpu": 20150.63,
      "total_tokens": 96763636
    },
    {
      "epoch": 0.21395395295410277,
      "grad_norm": 0.7109375,
      "learning_rate": 9.669366306826919e-06,
      "loss": 0.3979,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1071,
      "tokens_per_second_per_gpu": 18417.03,
      "total_tokens": 96853284
    },
    {
      "epoch": 0.21415372321829895,
      "grad_norm": 0.70703125,
      "learning_rate": 9.652841108375013e-06,
      "loss": 0.409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1072,
      "tokens_per_second_per_gpu": 17849.87,
      "total_tokens": 96940631
    },
    {
      "epoch": 0.21435349348249513,
      "grad_norm": 0.73046875,
      "learning_rate": 9.636316859042258e-06,
      "loss": 0.3889,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1073,
      "tokens_per_second_per_gpu": 18006.54,
      "total_tokens": 97026019
    },
    {
      "epoch": 0.21455326374669131,
      "grad_norm": 0.73046875,
      "learning_rate": 9.619793604005323e-06,
      "loss": 0.371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1074,
      "tokens_per_second_per_gpu": 16441.5,
      "total_tokens": 97106779
    },
    {
      "epoch": 0.21475303401088747,
      "grad_norm": 0.7109375,
      "learning_rate": 9.603271388438158e-06,
      "loss": 0.3765,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1075,
      "tokens_per_second_per_gpu": 17702.57,
      "total_tokens": 97192869
    },
    {
      "epoch": 0.21495280427508365,
      "grad_norm": 0.7265625,
      "learning_rate": 9.586750257511868e-06,
      "loss": 0.3881,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1076,
      "tokens_per_second_per_gpu": 18859.46,
      "total_tokens": 97284370
    },
    {
      "epoch": 0.21515257453927983,
      "grad_norm": 0.72265625,
      "learning_rate": 9.570230256394595e-06,
      "loss": 0.3907,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1077,
      "tokens_per_second_per_gpu": 18552.45,
      "total_tokens": 97374206
    },
    {
      "epoch": 0.215352344803476,
      "grad_norm": 0.70703125,
      "learning_rate": 9.553711430251388e-06,
      "loss": 0.3892,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1078,
      "tokens_per_second_per_gpu": 17409.27,
      "total_tokens": 97458582
    },
    {
      "epoch": 0.2155521150676722,
      "grad_norm": 0.7265625,
      "learning_rate": 9.537193824244092e-06,
      "loss": 0.4015,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1079,
      "tokens_per_second_per_gpu": 18327.67,
      "total_tokens": 97547689
    },
    {
      "epoch": 0.21575188533186834,
      "grad_norm": 0.7109375,
      "learning_rate": 9.520677483531205e-06,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1080,
      "tokens_per_second_per_gpu": 17948.36,
      "total_tokens": 97635849
    },
    {
      "epoch": 0.21595165559606452,
      "grad_norm": 0.71875,
      "learning_rate": 9.504162453267776e-06,
      "loss": 0.3814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1081,
      "tokens_per_second_per_gpu": 18263.95,
      "total_tokens": 97723821
    },
    {
      "epoch": 0.2161514258602607,
      "grad_norm": 0.71875,
      "learning_rate": 9.48764877860527e-06,
      "loss": 0.3946,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1082,
      "tokens_per_second_per_gpu": 18333.99,
      "total_tokens": 97813140
    },
    {
      "epoch": 0.21635119612445688,
      "grad_norm": 0.70703125,
      "learning_rate": 9.471136504691436e-06,
      "loss": 0.4313,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1083,
      "tokens_per_second_per_gpu": 19770.12,
      "total_tokens": 97910797
    },
    {
      "epoch": 0.21655096638865304,
      "grad_norm": 0.69140625,
      "learning_rate": 9.454625676670203e-06,
      "loss": 0.3804,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1084,
      "tokens_per_second_per_gpu": 18576.25,
      "total_tokens": 98000656
    },
    {
      "epoch": 0.21675073665284922,
      "grad_norm": 0.734375,
      "learning_rate": 9.438116339681546e-06,
      "loss": 0.3896,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1085,
      "tokens_per_second_per_gpu": 17426.85,
      "total_tokens": 98084953
    },
    {
      "epoch": 0.2169505069170454,
      "grad_norm": 0.734375,
      "learning_rate": 9.421608538861361e-06,
      "loss": 0.3777,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1086,
      "tokens_per_second_per_gpu": 16854.8,
      "total_tokens": 98167221
    },
    {
      "epoch": 0.21715027718124158,
      "grad_norm": 0.6875,
      "learning_rate": 9.405102319341345e-06,
      "loss": 0.3989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1087,
      "tokens_per_second_per_gpu": 19054.55,
      "total_tokens": 98260818
    },
    {
      "epoch": 0.21735004744543773,
      "grad_norm": 0.6953125,
      "learning_rate": 9.38859772624887e-06,
      "loss": 0.3748,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1088,
      "tokens_per_second_per_gpu": 17754.42,
      "total_tokens": 98348546
    },
    {
      "epoch": 0.2175498177096339,
      "grad_norm": 0.71875,
      "learning_rate": 9.372094804706867e-06,
      "loss": 0.3978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1089,
      "tokens_per_second_per_gpu": 17325.76,
      "total_tokens": 98434081
    },
    {
      "epoch": 0.2177495879738301,
      "grad_norm": 0.71484375,
      "learning_rate": 9.35559359983369e-06,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1090,
      "tokens_per_second_per_gpu": 17480.14,
      "total_tokens": 98520574
    },
    {
      "epoch": 0.21794935823802628,
      "grad_norm": 0.72265625,
      "learning_rate": 9.339094156743007e-06,
      "loss": 0.3949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1091,
      "tokens_per_second_per_gpu": 18101.03,
      "total_tokens": 98609082
    },
    {
      "epoch": 0.21814912850222246,
      "grad_norm": 0.7109375,
      "learning_rate": 9.322596520543661e-06,
      "loss": 0.3653,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1092,
      "tokens_per_second_per_gpu": 17431.66,
      "total_tokens": 98694341
    },
    {
      "epoch": 0.2183488987664186,
      "grad_norm": 0.7109375,
      "learning_rate": 9.30610073633956e-06,
      "loss": 0.3844,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1093,
      "tokens_per_second_per_gpu": 16673.14,
      "total_tokens": 98777179
    },
    {
      "epoch": 0.2185486690306148,
      "grad_norm": 0.72265625,
      "learning_rate": 9.28960684922955e-06,
      "loss": 0.391,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1094,
      "tokens_per_second_per_gpu": 17625.56,
      "total_tokens": 98863310
    },
    {
      "epoch": 0.21874843929481097,
      "grad_norm": 0.7109375,
      "learning_rate": 9.27311490430729e-06,
      "loss": 0.4195,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1095,
      "tokens_per_second_per_gpu": 18974.54,
      "total_tokens": 98956051
    },
    {
      "epoch": 0.21894820955900715,
      "grad_norm": 0.7265625,
      "learning_rate": 9.256624946661126e-06,
      "loss": 0.3909,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1096,
      "tokens_per_second_per_gpu": 17036.78,
      "total_tokens": 99040690
    },
    {
      "epoch": 0.2191479798232033,
      "grad_norm": 0.72265625,
      "learning_rate": 9.24013702137397e-06,
      "loss": 0.3918,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1097,
      "tokens_per_second_per_gpu": 17987.39,
      "total_tokens": 99128200
    },
    {
      "epoch": 0.21934775008739948,
      "grad_norm": 0.6875,
      "learning_rate": 9.223651173523182e-06,
      "loss": 0.3883,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1098,
      "tokens_per_second_per_gpu": 18964.45,
      "total_tokens": 99220247
    },
    {
      "epoch": 0.21954752035159567,
      "grad_norm": 0.71875,
      "learning_rate": 9.20716744818044e-06,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1099,
      "tokens_per_second_per_gpu": 17644.81,
      "total_tokens": 99307358
    },
    {
      "epoch": 0.21974729061579185,
      "grad_norm": 0.703125,
      "learning_rate": 9.190685890411628e-06,
      "loss": 0.3947,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1100,
      "tokens_per_second_per_gpu": 17331.11,
      "total_tokens": 99392906
    },
    {
      "epoch": 0.21994706087998803,
      "grad_norm": 0.68359375,
      "learning_rate": 9.174206545276678e-06,
      "loss": 0.3935,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1101,
      "tokens_per_second_per_gpu": 20228.18,
      "total_tokens": 99491279
    },
    {
      "epoch": 0.22014683114418418,
      "grad_norm": 0.69140625,
      "learning_rate": 9.157729457829504e-06,
      "loss": 0.4115,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1102,
      "tokens_per_second_per_gpu": 18808.58,
      "total_tokens": 99583026
    },
    {
      "epoch": 0.22034660140838036,
      "grad_norm": 0.6953125,
      "learning_rate": 9.14125467311783e-06,
      "loss": 0.4082,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1103,
      "tokens_per_second_per_gpu": 18921.2,
      "total_tokens": 99675362
    },
    {
      "epoch": 0.22054637167257654,
      "grad_norm": 0.69921875,
      "learning_rate": 9.124782236183088e-06,
      "loss": 0.3736,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1104,
      "tokens_per_second_per_gpu": 17196.27,
      "total_tokens": 99760953
    },
    {
      "epoch": 0.22074614193677272,
      "grad_norm": 0.70703125,
      "learning_rate": 9.108312192060298e-06,
      "loss": 0.3702,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1105,
      "tokens_per_second_per_gpu": 17348.98,
      "total_tokens": 99846349
    },
    {
      "epoch": 0.22094591220096887,
      "grad_norm": 0.70703125,
      "learning_rate": 9.091844585777919e-06,
      "loss": 0.3705,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1106,
      "tokens_per_second_per_gpu": 17986.4,
      "total_tokens": 99934128
    },
    {
      "epoch": 0.22114568246516506,
      "grad_norm": 0.71484375,
      "learning_rate": 9.075379462357766e-06,
      "loss": 0.392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1107,
      "tokens_per_second_per_gpu": 18271.73,
      "total_tokens": 100023228
    },
    {
      "epoch": 0.22134545272936124,
      "grad_norm": 0.703125,
      "learning_rate": 9.058916866814857e-06,
      "loss": 0.3786,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1108,
      "tokens_per_second_per_gpu": 17790.65,
      "total_tokens": 100110144
    },
    {
      "epoch": 0.22154522299355742,
      "grad_norm": 0.70703125,
      "learning_rate": 9.042456844157299e-06,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1109,
      "tokens_per_second_per_gpu": 19390.42,
      "total_tokens": 100204722
    },
    {
      "epoch": 0.22174499325775357,
      "grad_norm": 0.7109375,
      "learning_rate": 9.02599943938617e-06,
      "loss": 0.3913,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1110,
      "tokens_per_second_per_gpu": 17881.35,
      "total_tokens": 100290834
    },
    {
      "epoch": 0.22194476352194975,
      "grad_norm": 0.67578125,
      "learning_rate": 9.009544697495373e-06,
      "loss": 0.3884,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1111,
      "tokens_per_second_per_gpu": 18800.82,
      "total_tokens": 100382427
    },
    {
      "epoch": 0.22214453378614593,
      "grad_norm": 0.6953125,
      "learning_rate": 8.993092663471556e-06,
      "loss": 0.4091,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1112,
      "tokens_per_second_per_gpu": 19143.26,
      "total_tokens": 100474811
    },
    {
      "epoch": 0.2223443040503421,
      "grad_norm": 0.7109375,
      "learning_rate": 8.97664338229395e-06,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1113,
      "tokens_per_second_per_gpu": 17509.86,
      "total_tokens": 100560312
    },
    {
      "epoch": 0.2225440743145383,
      "grad_norm": 0.703125,
      "learning_rate": 8.960196898934264e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1114,
      "tokens_per_second_per_gpu": 18398.87,
      "total_tokens": 100650076
    },
    {
      "epoch": 0.22274384457873445,
      "grad_norm": 0.703125,
      "learning_rate": 8.943753258356545e-06,
      "loss": 0.3934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1115,
      "tokens_per_second_per_gpu": 19218.1,
      "total_tokens": 100743434
    },
    {
      "epoch": 0.22294361484293063,
      "grad_norm": 0.75,
      "learning_rate": 8.927312505517086e-06,
      "loss": 0.4079,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1116,
      "tokens_per_second_per_gpu": 17061.85,
      "total_tokens": 100827601
    },
    {
      "epoch": 0.2231433851071268,
      "grad_norm": 0.71875,
      "learning_rate": 8.910874685364275e-06,
      "loss": 0.4163,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1117,
      "tokens_per_second_per_gpu": 18235.66,
      "total_tokens": 100917285
    },
    {
      "epoch": 0.223343155371323,
      "grad_norm": 0.7109375,
      "learning_rate": 8.894439842838488e-06,
      "loss": 0.3852,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1118,
      "tokens_per_second_per_gpu": 19736.9,
      "total_tokens": 101012860
    },
    {
      "epoch": 0.22354292563551914,
      "grad_norm": 0.6953125,
      "learning_rate": 8.878008022871959e-06,
      "loss": 0.3997,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1119,
      "tokens_per_second_per_gpu": 19240.95,
      "total_tokens": 101105085
    },
    {
      "epoch": 0.22374269589971532,
      "grad_norm": 0.69140625,
      "learning_rate": 8.861579270388649e-06,
      "loss": 0.3885,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1120,
      "tokens_per_second_per_gpu": 19035.66,
      "total_tokens": 101196877
    },
    {
      "epoch": 0.2239424661639115,
      "grad_norm": 0.68359375,
      "learning_rate": 8.84515363030414e-06,
      "loss": 0.3971,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1121,
      "tokens_per_second_per_gpu": 19703.72,
      "total_tokens": 101292121
    },
    {
      "epoch": 0.22414223642810768,
      "grad_norm": 0.70703125,
      "learning_rate": 8.82873114752551e-06,
      "loss": 0.4025,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1122,
      "tokens_per_second_per_gpu": 19115.02,
      "total_tokens": 101385037
    },
    {
      "epoch": 0.22434200669230386,
      "grad_norm": 0.73828125,
      "learning_rate": 8.812311866951198e-06,
      "loss": 0.3986,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1123,
      "tokens_per_second_per_gpu": 18567.64,
      "total_tokens": 101475485
    },
    {
      "epoch": 0.22454177695650002,
      "grad_norm": 0.73828125,
      "learning_rate": 8.795895833470896e-06,
      "loss": 0.3971,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1124,
      "tokens_per_second_per_gpu": 17526.61,
      "total_tokens": 101561556
    },
    {
      "epoch": 0.2247415472206962,
      "grad_norm": 0.69140625,
      "learning_rate": 8.7794830919654e-06,
      "loss": 0.3755,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1125,
      "tokens_per_second_per_gpu": 17544.39,
      "total_tokens": 101646227
    },
    {
      "epoch": 0.22494131748489238,
      "grad_norm": 0.71875,
      "learning_rate": 8.763073687306523e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1126,
      "tokens_per_second_per_gpu": 17443.86,
      "total_tokens": 101730566
    },
    {
      "epoch": 0.22514108774908856,
      "grad_norm": 0.6484375,
      "learning_rate": 8.746667664356957e-06,
      "loss": 0.3895,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1127,
      "tokens_per_second_per_gpu": 20284.63,
      "total_tokens": 101829376
    },
    {
      "epoch": 0.2253408580132847,
      "grad_norm": 0.72265625,
      "learning_rate": 8.730265067970141e-06,
      "loss": 0.4012,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1128,
      "tokens_per_second_per_gpu": 17934.13,
      "total_tokens": 101915973
    },
    {
      "epoch": 0.2255406282774809,
      "grad_norm": 0.72265625,
      "learning_rate": 8.713865942990143e-06,
      "loss": 0.3799,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1129,
      "tokens_per_second_per_gpu": 18096.17,
      "total_tokens": 102002753
    },
    {
      "epoch": 0.22574039854167707,
      "grad_norm": 0.6796875,
      "learning_rate": 8.697470334251543e-06,
      "loss": 0.4147,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1130,
      "tokens_per_second_per_gpu": 19923.33,
      "total_tokens": 102101681
    },
    {
      "epoch": 0.22594016880587325,
      "grad_norm": 0.703125,
      "learning_rate": 8.68107828657931e-06,
      "loss": 0.39,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1131,
      "tokens_per_second_per_gpu": 18940.99,
      "total_tokens": 102192949
    },
    {
      "epoch": 0.2261399390700694,
      "grad_norm": 0.703125,
      "learning_rate": 8.664689844788683e-06,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1132,
      "tokens_per_second_per_gpu": 18221.03,
      "total_tokens": 102281099
    },
    {
      "epoch": 0.2263397093342656,
      "grad_norm": 0.734375,
      "learning_rate": 8.648305053685035e-06,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1133,
      "tokens_per_second_per_gpu": 18738.94,
      "total_tokens": 102372985
    },
    {
      "epoch": 0.22653947959846177,
      "grad_norm": 0.70703125,
      "learning_rate": 8.63192395806375e-06,
      "loss": 0.3836,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1134,
      "tokens_per_second_per_gpu": 18112.34,
      "total_tokens": 102460764
    },
    {
      "epoch": 0.22673924986265795,
      "grad_norm": 0.7109375,
      "learning_rate": 8.615546602710126e-06,
      "loss": 0.3877,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1135,
      "tokens_per_second_per_gpu": 18461.2,
      "total_tokens": 102551054
    },
    {
      "epoch": 0.22693902012685413,
      "grad_norm": 0.69921875,
      "learning_rate": 8.599173032399222e-06,
      "loss": 0.3783,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1136,
      "tokens_per_second_per_gpu": 19077.81,
      "total_tokens": 102643670
    },
    {
      "epoch": 0.22713879039105028,
      "grad_norm": 0.7265625,
      "learning_rate": 8.582803291895758e-06,
      "loss": 0.4016,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1137,
      "tokens_per_second_per_gpu": 18748.06,
      "total_tokens": 102734551
    },
    {
      "epoch": 0.22733856065524646,
      "grad_norm": 0.6796875,
      "learning_rate": 8.566437425953985e-06,
      "loss": 0.4045,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1138,
      "tokens_per_second_per_gpu": 20453.92,
      "total_tokens": 102831918
    },
    {
      "epoch": 0.22753833091944264,
      "grad_norm": 0.703125,
      "learning_rate": 8.550075479317542e-06,
      "loss": 0.3572,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1139,
      "tokens_per_second_per_gpu": 18047.68,
      "total_tokens": 102919126
    },
    {
      "epoch": 0.22773810118363882,
      "grad_norm": 0.703125,
      "learning_rate": 8.533717496719374e-06,
      "loss": 0.4047,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1140,
      "tokens_per_second_per_gpu": 18741.08,
      "total_tokens": 103009565
    },
    {
      "epoch": 0.22793787144783498,
      "grad_norm": 0.71484375,
      "learning_rate": 8.51736352288158e-06,
      "loss": 0.3809,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1141,
      "tokens_per_second_per_gpu": 18028.84,
      "total_tokens": 103098075
    },
    {
      "epoch": 0.22813764171203116,
      "grad_norm": 0.6953125,
      "learning_rate": 8.501013602515306e-06,
      "loss": 0.3864,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1142,
      "tokens_per_second_per_gpu": 18799.61,
      "total_tokens": 103187920
    },
    {
      "epoch": 0.22833741197622734,
      "grad_norm": 0.6953125,
      "learning_rate": 8.484667780320598e-06,
      "loss": 0.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1143,
      "tokens_per_second_per_gpu": 19375.74,
      "total_tokens": 103280265
    },
    {
      "epoch": 0.22853718224042352,
      "grad_norm": 0.71875,
      "learning_rate": 8.468326100986317e-06,
      "loss": 0.3729,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1144,
      "tokens_per_second_per_gpu": 16514.46,
      "total_tokens": 103360167
    },
    {
      "epoch": 0.2287369525046197,
      "grad_norm": 0.6796875,
      "learning_rate": 8.451988609189987e-06,
      "loss": 0.412,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1145,
      "tokens_per_second_per_gpu": 18640.91,
      "total_tokens": 103451681
    },
    {
      "epoch": 0.22893672276881585,
      "grad_norm": 0.703125,
      "learning_rate": 8.43565534959769e-06,
      "loss": 0.4356,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1146,
      "tokens_per_second_per_gpu": 20238.9,
      "total_tokens": 103551291
    },
    {
      "epoch": 0.22913649303301203,
      "grad_norm": 0.72265625,
      "learning_rate": 8.419326366863937e-06,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1147,
      "tokens_per_second_per_gpu": 17661.07,
      "total_tokens": 103639039
    },
    {
      "epoch": 0.22933626329720821,
      "grad_norm": 0.71875,
      "learning_rate": 8.403001705631537e-06,
      "loss": 0.3963,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1148,
      "tokens_per_second_per_gpu": 17550.27,
      "total_tokens": 103724138
    },
    {
      "epoch": 0.2295360335614044,
      "grad_norm": 0.7109375,
      "learning_rate": 8.386681410531493e-06,
      "loss": 0.3628,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1149,
      "tokens_per_second_per_gpu": 17430.12,
      "total_tokens": 103808723
    },
    {
      "epoch": 0.22973580382560055,
      "grad_norm": 0.8671875,
      "learning_rate": 8.370365526182862e-06,
      "loss": 0.4117,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1150,
      "tokens_per_second_per_gpu": 20409.51,
      "total_tokens": 103908291
    },
    {
      "epoch": 0.22993557408979673,
      "grad_norm": 0.703125,
      "learning_rate": 8.35405409719266e-06,
      "loss": 0.4023,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1151,
      "tokens_per_second_per_gpu": 19120.72,
      "total_tokens": 104001012
    },
    {
      "epoch": 0.2301353443539929,
      "grad_norm": 0.69140625,
      "learning_rate": 8.33774716815571e-06,
      "loss": 0.4014,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1152,
      "tokens_per_second_per_gpu": 19263.81,
      "total_tokens": 104095058
    },
    {
      "epoch": 0.2303351146181891,
      "grad_norm": 0.6953125,
      "learning_rate": 8.321444783654524e-06,
      "loss": 0.3635,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1153,
      "tokens_per_second_per_gpu": 16905.23,
      "total_tokens": 104177634
    },
    {
      "epoch": 0.23053488488238524,
      "grad_norm": 0.703125,
      "learning_rate": 8.305146988259205e-06,
      "loss": 0.4173,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1154,
      "tokens_per_second_per_gpu": 19048.38,
      "total_tokens": 104270924
    },
    {
      "epoch": 0.23073465514658142,
      "grad_norm": 0.6953125,
      "learning_rate": 8.2888538265273e-06,
      "loss": 0.4076,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1155,
      "tokens_per_second_per_gpu": 19528.34,
      "total_tokens": 104365404
    },
    {
      "epoch": 0.2309344254107776,
      "grad_norm": 0.68359375,
      "learning_rate": 8.2725653430037e-06,
      "loss": 0.4036,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1156,
      "tokens_per_second_per_gpu": 19002.06,
      "total_tokens": 104457855
    },
    {
      "epoch": 0.23113419567497379,
      "grad_norm": 0.7265625,
      "learning_rate": 8.256281582220486e-06,
      "loss": 0.347,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1157,
      "tokens_per_second_per_gpu": 17901.24,
      "total_tokens": 104544361
    },
    {
      "epoch": 0.23133396593916997,
      "grad_norm": 0.72265625,
      "learning_rate": 8.24000258869684e-06,
      "loss": 0.4159,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1158,
      "tokens_per_second_per_gpu": 18436.85,
      "total_tokens": 104634271
    },
    {
      "epoch": 0.23153373620336612,
      "grad_norm": 0.69140625,
      "learning_rate": 8.223728406938914e-06,
      "loss": 0.3897,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1159,
      "tokens_per_second_per_gpu": 18112.75,
      "total_tokens": 104721928
    },
    {
      "epoch": 0.2317335064675623,
      "grad_norm": 0.71875,
      "learning_rate": 8.207459081439696e-06,
      "loss": 0.4009,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1160,
      "tokens_per_second_per_gpu": 18299.69,
      "total_tokens": 104811313
    },
    {
      "epoch": 0.23193327673175848,
      "grad_norm": 0.69140625,
      "learning_rate": 8.191194656678905e-06,
      "loss": 0.4066,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1161,
      "tokens_per_second_per_gpu": 20114.35,
      "total_tokens": 104907141
    },
    {
      "epoch": 0.23213304699595466,
      "grad_norm": 0.68359375,
      "learning_rate": 8.174935177122849e-06,
      "loss": 0.3904,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1162,
      "tokens_per_second_per_gpu": 20408.41,
      "total_tokens": 105003038
    },
    {
      "epoch": 0.23233281726015081,
      "grad_norm": 0.66796875,
      "learning_rate": 8.15868068722433e-06,
      "loss": 0.3733,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1163,
      "tokens_per_second_per_gpu": 19129.23,
      "total_tokens": 105095449
    },
    {
      "epoch": 0.232532587524347,
      "grad_norm": 0.671875,
      "learning_rate": 8.142431231422501e-06,
      "loss": 0.4111,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1164,
      "tokens_per_second_per_gpu": 20050.85,
      "total_tokens": 105190152
    },
    {
      "epoch": 0.23273235778854318,
      "grad_norm": 0.70703125,
      "learning_rate": 8.126186854142752e-06,
      "loss": 0.3933,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1165,
      "tokens_per_second_per_gpu": 18672.92,
      "total_tokens": 105281121
    },
    {
      "epoch": 0.23293212805273936,
      "grad_norm": 0.71484375,
      "learning_rate": 8.109947599796599e-06,
      "loss": 0.3683,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1166,
      "tokens_per_second_per_gpu": 17956.3,
      "total_tokens": 105370006
    },
    {
      "epoch": 0.23313189831693554,
      "grad_norm": 0.71484375,
      "learning_rate": 8.093713512781534e-06,
      "loss": 0.3977,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1167,
      "tokens_per_second_per_gpu": 20267.27,
      "total_tokens": 105467234
    },
    {
      "epoch": 0.2333316685811317,
      "grad_norm": 0.70703125,
      "learning_rate": 8.077484637480931e-06,
      "loss": 0.3916,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1168,
      "tokens_per_second_per_gpu": 17620.06,
      "total_tokens": 105553540
    },
    {
      "epoch": 0.23353143884532787,
      "grad_norm": 0.703125,
      "learning_rate": 8.06126101826392e-06,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1169,
      "tokens_per_second_per_gpu": 18146.43,
      "total_tokens": 105641165
    },
    {
      "epoch": 0.23373120910952405,
      "grad_norm": 0.70703125,
      "learning_rate": 8.045042699485256e-06,
      "loss": 0.4022,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1170,
      "tokens_per_second_per_gpu": 19304.39,
      "total_tokens": 105735040
    },
    {
      "epoch": 0.23393097937372023,
      "grad_norm": 0.703125,
      "learning_rate": 8.0288297254852e-06,
      "loss": 0.3759,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1171,
      "tokens_per_second_per_gpu": 18391.33,
      "total_tokens": 105823946
    },
    {
      "epoch": 0.23413074963791639,
      "grad_norm": 0.67578125,
      "learning_rate": 8.012622140589404e-06,
      "loss": 0.4022,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1172,
      "tokens_per_second_per_gpu": 20336.86,
      "total_tokens": 105922580
    },
    {
      "epoch": 0.23433051990211257,
      "grad_norm": 0.70703125,
      "learning_rate": 7.99641998910879e-06,
      "loss": 0.3956,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1173,
      "tokens_per_second_per_gpu": 18151.07,
      "total_tokens": 106011176
    },
    {
      "epoch": 0.23453029016630875,
      "grad_norm": 0.66796875,
      "learning_rate": 7.980223315339417e-06,
      "loss": 0.3918,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1174,
      "tokens_per_second_per_gpu": 19303.37,
      "total_tokens": 106105858
    },
    {
      "epoch": 0.23473006043050493,
      "grad_norm": 0.6796875,
      "learning_rate": 7.964032163562378e-06,
      "loss": 0.3718,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1175,
      "tokens_per_second_per_gpu": 19381.32,
      "total_tokens": 106199337
    },
    {
      "epoch": 0.23492983069470108,
      "grad_norm": 0.69140625,
      "learning_rate": 7.947846578043658e-06,
      "loss": 0.3981,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1176,
      "tokens_per_second_per_gpu": 19639.57,
      "total_tokens": 106294402
    },
    {
      "epoch": 0.23512960095889726,
      "grad_norm": 0.6875,
      "learning_rate": 7.931666603034034e-06,
      "loss": 0.3957,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1177,
      "tokens_per_second_per_gpu": 20065.09,
      "total_tokens": 106392906
    },
    {
      "epoch": 0.23532937122309344,
      "grad_norm": 0.71484375,
      "learning_rate": 7.915492282768938e-06,
      "loss": 0.3882,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1178,
      "tokens_per_second_per_gpu": 18202.86,
      "total_tokens": 106482193
    },
    {
      "epoch": 0.23552914148728962,
      "grad_norm": 0.671875,
      "learning_rate": 7.899323661468344e-06,
      "loss": 0.3724,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1179,
      "tokens_per_second_per_gpu": 19131.03,
      "total_tokens": 106573917
    },
    {
      "epoch": 0.2357289117514858,
      "grad_norm": 0.73046875,
      "learning_rate": 7.883160783336646e-06,
      "loss": 0.4081,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1180,
      "tokens_per_second_per_gpu": 18864.02,
      "total_tokens": 106664948
    },
    {
      "epoch": 0.23592868201568196,
      "grad_norm": 0.72265625,
      "learning_rate": 7.867003692562533e-06,
      "loss": 0.414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1181,
      "tokens_per_second_per_gpu": 19868.13,
      "total_tokens": 106759967
    },
    {
      "epoch": 0.23612845227987814,
      "grad_norm": 0.74609375,
      "learning_rate": 7.850852433318878e-06,
      "loss": 0.412,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1182,
      "tokens_per_second_per_gpu": 18770.39,
      "total_tokens": 106850537
    },
    {
      "epoch": 0.23632822254407432,
      "grad_norm": 0.72265625,
      "learning_rate": 7.834707049762605e-06,
      "loss": 0.4111,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1183,
      "tokens_per_second_per_gpu": 19135.95,
      "total_tokens": 106943800
    },
    {
      "epoch": 0.2365279928082705,
      "grad_norm": 0.68359375,
      "learning_rate": 7.818567586034578e-06,
      "loss": 0.3813,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1184,
      "tokens_per_second_per_gpu": 18829.63,
      "total_tokens": 107034808
    },
    {
      "epoch": 0.23672776307246665,
      "grad_norm": 0.6953125,
      "learning_rate": 7.80243408625947e-06,
      "loss": 0.3578,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1185,
      "tokens_per_second_per_gpu": 17574.33,
      "total_tokens": 107120202
    },
    {
      "epoch": 0.23692753333666283,
      "grad_norm": 0.68359375,
      "learning_rate": 7.786306594545658e-06,
      "loss": 0.3729,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1186,
      "tokens_per_second_per_gpu": 18770.82,
      "total_tokens": 107210986
    },
    {
      "epoch": 0.237127303600859,
      "grad_norm": 0.69921875,
      "learning_rate": 7.770185154985086e-06,
      "loss": 0.3982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1187,
      "tokens_per_second_per_gpu": 19098.06,
      "total_tokens": 107304319
    },
    {
      "epoch": 0.2373270738650552,
      "grad_norm": 0.6875,
      "learning_rate": 7.754069811653155e-06,
      "loss": 0.4184,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1188,
      "tokens_per_second_per_gpu": 19785.62,
      "total_tokens": 107400868
    },
    {
      "epoch": 0.23752684412925137,
      "grad_norm": 0.6953125,
      "learning_rate": 7.7379606086086e-06,
      "loss": 0.3886,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1189,
      "tokens_per_second_per_gpu": 19425.63,
      "total_tokens": 107494589
    },
    {
      "epoch": 0.23772661439344753,
      "grad_norm": 0.76171875,
      "learning_rate": 7.721857589893364e-06,
      "loss": 0.3819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1190,
      "tokens_per_second_per_gpu": 16689.33,
      "total_tokens": 107574318
    },
    {
      "epoch": 0.2379263846576437,
      "grad_norm": 0.703125,
      "learning_rate": 7.705760799532485e-06,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1191,
      "tokens_per_second_per_gpu": 19068.76,
      "total_tokens": 107668067
    },
    {
      "epoch": 0.2381261549218399,
      "grad_norm": 0.6796875,
      "learning_rate": 7.689670281533977e-06,
      "loss": 0.4173,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1192,
      "tokens_per_second_per_gpu": 19890.07,
      "total_tokens": 107765359
    },
    {
      "epoch": 0.23832592518603607,
      "grad_norm": 0.82421875,
      "learning_rate": 7.673586079888699e-06,
      "loss": 0.4272,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1193,
      "tokens_per_second_per_gpu": 19470.72,
      "total_tokens": 107863000
    },
    {
      "epoch": 0.23852569545023222,
      "grad_norm": 0.69921875,
      "learning_rate": 7.657508238570246e-06,
      "loss": 0.3861,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1194,
      "tokens_per_second_per_gpu": 18600.43,
      "total_tokens": 107951854
    },
    {
      "epoch": 0.2387254657144284,
      "grad_norm": 0.6796875,
      "learning_rate": 7.641436801534818e-06,
      "loss": 0.3938,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1195,
      "tokens_per_second_per_gpu": 19095.78,
      "total_tokens": 108043788
    },
    {
      "epoch": 0.23892523597862458,
      "grad_norm": 0.69921875,
      "learning_rate": 7.625371812721115e-06,
      "loss": 0.4162,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1196,
      "tokens_per_second_per_gpu": 19071.54,
      "total_tokens": 108135892
    },
    {
      "epoch": 0.23912500624282076,
      "grad_norm": 0.69921875,
      "learning_rate": 7.6093133160502e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1197,
      "tokens_per_second_per_gpu": 18878.3,
      "total_tokens": 108227514
    },
    {
      "epoch": 0.23932477650701692,
      "grad_norm": 0.70703125,
      "learning_rate": 7.593261355425394e-06,
      "loss": 0.4081,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1198,
      "tokens_per_second_per_gpu": 18961.11,
      "total_tokens": 108318907
    },
    {
      "epoch": 0.2395245467712131,
      "grad_norm": 0.69140625,
      "learning_rate": 7.577215974732139e-06,
      "loss": 0.3778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1199,
      "tokens_per_second_per_gpu": 18768.7,
      "total_tokens": 108409143
    },
    {
      "epoch": 0.23972431703540928,
      "grad_norm": 0.703125,
      "learning_rate": 7.561177217837894e-06,
      "loss": 0.3662,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1200,
      "tokens_per_second_per_gpu": 17107.45,
      "total_tokens": 108493364
    },
    {
      "epoch": 0.23992408729960546,
      "grad_norm": 0.7109375,
      "learning_rate": 7.545145128592009e-06,
      "loss": 0.4365,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1201,
      "tokens_per_second_per_gpu": 21286.65,
      "total_tokens": 108596282
    },
    {
      "epoch": 0.24012385756380164,
      "grad_norm": 0.69140625,
      "learning_rate": 7.529119750825605e-06,
      "loss": 0.3956,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1202,
      "tokens_per_second_per_gpu": 18692.71,
      "total_tokens": 108687457
    },
    {
      "epoch": 0.2403236278279978,
      "grad_norm": 0.6875,
      "learning_rate": 7.513101128351454e-06,
      "loss": 0.424,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1203,
      "tokens_per_second_per_gpu": 20395.69,
      "total_tokens": 108787283
    },
    {
      "epoch": 0.24052339809219397,
      "grad_norm": 0.765625,
      "learning_rate": 7.497089304963854e-06,
      "loss": 0.3596,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1204,
      "tokens_per_second_per_gpu": 17494.5,
      "total_tokens": 108872135
    },
    {
      "epoch": 0.24072316835639015,
      "grad_norm": 0.703125,
      "learning_rate": 7.481084324438521e-06,
      "loss": 0.3767,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1205,
      "tokens_per_second_per_gpu": 18619.9,
      "total_tokens": 108961991
    },
    {
      "epoch": 0.24092293862058634,
      "grad_norm": 0.72265625,
      "learning_rate": 7.46508623053246e-06,
      "loss": 0.361,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1206,
      "tokens_per_second_per_gpu": 16248.62,
      "total_tokens": 109041614
    },
    {
      "epoch": 0.2411227088847825,
      "grad_norm": 0.69921875,
      "learning_rate": 7.449095066983849e-06,
      "loss": 0.39,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1207,
      "tokens_per_second_per_gpu": 17369.18,
      "total_tokens": 109126708
    },
    {
      "epoch": 0.24132247914897867,
      "grad_norm": 0.6796875,
      "learning_rate": 7.433110877511922e-06,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1208,
      "tokens_per_second_per_gpu": 19928.53,
      "total_tokens": 109221677
    },
    {
      "epoch": 0.24152224941317485,
      "grad_norm": 0.703125,
      "learning_rate": 7.4171337058168365e-06,
      "loss": 0.4046,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1209,
      "tokens_per_second_per_gpu": 19613.27,
      "total_tokens": 109314917
    },
    {
      "epoch": 0.24172201967737103,
      "grad_norm": 0.703125,
      "learning_rate": 7.401163595579573e-06,
      "loss": 0.3759,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1210,
      "tokens_per_second_per_gpu": 17718.94,
      "total_tokens": 109401603
    },
    {
      "epoch": 0.2419217899415672,
      "grad_norm": 0.70703125,
      "learning_rate": 7.385200590461803e-06,
      "loss": 0.3802,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1211,
      "tokens_per_second_per_gpu": 18518.37,
      "total_tokens": 109490692
    },
    {
      "epoch": 0.24212156020576336,
      "grad_norm": 0.68359375,
      "learning_rate": 7.369244734105774e-06,
      "loss": 0.3741,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1212,
      "tokens_per_second_per_gpu": 19250.22,
      "total_tokens": 109584568
    },
    {
      "epoch": 0.24232133046995954,
      "grad_norm": 0.7109375,
      "learning_rate": 7.353296070134186e-06,
      "loss": 0.3891,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1213,
      "tokens_per_second_per_gpu": 17940.75,
      "total_tokens": 109671116
    },
    {
      "epoch": 0.24252110073415573,
      "grad_norm": 0.7265625,
      "learning_rate": 7.337354642150077e-06,
      "loss": 0.4079,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1214,
      "tokens_per_second_per_gpu": 18044.95,
      "total_tokens": 109758332
    },
    {
      "epoch": 0.2427208709983519,
      "grad_norm": 0.703125,
      "learning_rate": 7.321420493736705e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1215,
      "tokens_per_second_per_gpu": 17451.4,
      "total_tokens": 109844087
    },
    {
      "epoch": 0.24292064126254806,
      "grad_norm": 0.734375,
      "learning_rate": 7.305493668457421e-06,
      "loss": 0.4097,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1216,
      "tokens_per_second_per_gpu": 18805.29,
      "total_tokens": 109935034
    },
    {
      "epoch": 0.24312041152674424,
      "grad_norm": 0.6953125,
      "learning_rate": 7.28957420985556e-06,
      "loss": 0.4265,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1217,
      "tokens_per_second_per_gpu": 20162.91,
      "total_tokens": 110032467
    },
    {
      "epoch": 0.24332018179094042,
      "grad_norm": 0.70703125,
      "learning_rate": 7.273662161454309e-06,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1218,
      "tokens_per_second_per_gpu": 17662.02,
      "total_tokens": 110119229
    },
    {
      "epoch": 0.2435199520551366,
      "grad_norm": 0.71484375,
      "learning_rate": 7.257757566756604e-06,
      "loss": 0.3946,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1219,
      "tokens_per_second_per_gpu": 19600.59,
      "total_tokens": 110212922
    },
    {
      "epoch": 0.24371972231933275,
      "grad_norm": 0.6796875,
      "learning_rate": 7.2418604692450015e-06,
      "loss": 0.3864,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1220,
      "tokens_per_second_per_gpu": 19016.35,
      "total_tokens": 110305855
    },
    {
      "epoch": 0.24391949258352894,
      "grad_norm": 0.69921875,
      "learning_rate": 7.225970912381557e-06,
      "loss": 0.3688,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1221,
      "tokens_per_second_per_gpu": 19042.34,
      "total_tokens": 110396493
    },
    {
      "epoch": 0.24411926284772512,
      "grad_norm": 0.7109375,
      "learning_rate": 7.210088939607709e-06,
      "loss": 0.3824,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1222,
      "tokens_per_second_per_gpu": 17895.91,
      "total_tokens": 110482877
    },
    {
      "epoch": 0.2443190331119213,
      "grad_norm": 0.69140625,
      "learning_rate": 7.194214594344169e-06,
      "loss": 0.3861,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1223,
      "tokens_per_second_per_gpu": 18743.81,
      "total_tokens": 110573554
    },
    {
      "epoch": 0.24451880337611748,
      "grad_norm": 0.703125,
      "learning_rate": 7.178347919990789e-06,
      "loss": 0.41,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1224,
      "tokens_per_second_per_gpu": 20547.15,
      "total_tokens": 110671589
    },
    {
      "epoch": 0.24471857364031363,
      "grad_norm": 0.7109375,
      "learning_rate": 7.16248895992645e-06,
      "loss": 0.3967,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1225,
      "tokens_per_second_per_gpu": 18389.82,
      "total_tokens": 110759989
    },
    {
      "epoch": 0.2449183439045098,
      "grad_norm": 0.7265625,
      "learning_rate": 7.14663775750895e-06,
      "loss": 0.4033,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1226,
      "tokens_per_second_per_gpu": 18281.25,
      "total_tokens": 110848432
    },
    {
      "epoch": 0.245118114168706,
      "grad_norm": 0.734375,
      "learning_rate": 7.13079435607486e-06,
      "loss": 0.3875,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1227,
      "tokens_per_second_per_gpu": 17482.03,
      "total_tokens": 110934887
    },
    {
      "epoch": 0.24531788443290217,
      "grad_norm": 0.70703125,
      "learning_rate": 7.114958798939442e-06,
      "loss": 0.3713,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1228,
      "tokens_per_second_per_gpu": 18082.61,
      "total_tokens": 111022368
    },
    {
      "epoch": 0.24551765469709833,
      "grad_norm": 0.73046875,
      "learning_rate": 7.099131129396501e-06,
      "loss": 0.3935,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1229,
      "tokens_per_second_per_gpu": 19165.69,
      "total_tokens": 111115185
    },
    {
      "epoch": 0.2457174249612945,
      "grad_norm": 0.7578125,
      "learning_rate": 7.083311390718285e-06,
      "loss": 0.3772,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1230,
      "tokens_per_second_per_gpu": 18137.93,
      "total_tokens": 111203180
    },
    {
      "epoch": 0.2459171952254907,
      "grad_norm": 0.70703125,
      "learning_rate": 7.067499626155354e-06,
      "loss": 0.4229,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1231,
      "tokens_per_second_per_gpu": 17916.08,
      "total_tokens": 111291869
    },
    {
      "epoch": 0.24611696548968687,
      "grad_norm": 0.9765625,
      "learning_rate": 7.051695878936466e-06,
      "loss": 0.3994,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1232,
      "tokens_per_second_per_gpu": 19510.59,
      "total_tokens": 111385951
    },
    {
      "epoch": 0.24631673575388305,
      "grad_norm": 0.71875,
      "learning_rate": 7.035900192268464e-06,
      "loss": 0.3787,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1233,
      "tokens_per_second_per_gpu": 18351.92,
      "total_tokens": 111475919
    },
    {
      "epoch": 0.2465165060180792,
      "grad_norm": 0.7265625,
      "learning_rate": 7.020112609336154e-06,
      "loss": 0.3712,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1234,
      "tokens_per_second_per_gpu": 16980.35,
      "total_tokens": 111559112
    },
    {
      "epoch": 0.24671627628227538,
      "grad_norm": 0.671875,
      "learning_rate": 7.004333173302185e-06,
      "loss": 0.3772,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1235,
      "tokens_per_second_per_gpu": 19252.3,
      "total_tokens": 111651015
    },
    {
      "epoch": 0.24691604654647156,
      "grad_norm": 0.6875,
      "learning_rate": 6.988561927306927e-06,
      "loss": 0.3381,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1236,
      "tokens_per_second_per_gpu": 18152.71,
      "total_tokens": 111737884
    },
    {
      "epoch": 0.24711581681066774,
      "grad_norm": 0.6953125,
      "learning_rate": 6.972798914468369e-06,
      "loss": 0.3926,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1237,
      "tokens_per_second_per_gpu": 19928.84,
      "total_tokens": 111832427
    },
    {
      "epoch": 0.2473155870748639,
      "grad_norm": 0.6953125,
      "learning_rate": 6.957044177881987e-06,
      "loss": 0.3846,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1238,
      "tokens_per_second_per_gpu": 17172.32,
      "total_tokens": 111916911
    },
    {
      "epoch": 0.24751535733906008,
      "grad_norm": 0.6640625,
      "learning_rate": 6.941297760620626e-06,
      "loss": 0.367,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1239,
      "tokens_per_second_per_gpu": 18569.71,
      "total_tokens": 112006780
    },
    {
      "epoch": 0.24771512760325626,
      "grad_norm": 0.71875,
      "learning_rate": 6.925559705734394e-06,
      "loss": 0.3858,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1240,
      "tokens_per_second_per_gpu": 16981.85,
      "total_tokens": 112089096
    },
    {
      "epoch": 0.24791489786745244,
      "grad_norm": 0.703125,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.4001,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1241,
      "tokens_per_second_per_gpu": 18736.28,
      "total_tokens": 112180651
    },
    {
      "epoch": 0.2481146681316486,
      "grad_norm": 0.6640625,
      "learning_rate": 6.89410885517329e-06,
      "loss": 0.3822,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1242,
      "tokens_per_second_per_gpu": 19295.7,
      "total_tokens": 112274126
    },
    {
      "epoch": 0.24831443839584477,
      "grad_norm": 0.69140625,
      "learning_rate": 6.878396145483841e-06,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1243,
      "tokens_per_second_per_gpu": 18566.05,
      "total_tokens": 112363650
    },
    {
      "epoch": 0.24851420866004095,
      "grad_norm": 0.71484375,
      "learning_rate": 6.8626919701401355e-06,
      "loss": 0.3841,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1244,
      "tokens_per_second_per_gpu": 17699.93,
      "total_tokens": 112449866
    },
    {
      "epoch": 0.24871397892423713,
      "grad_norm": 0.7109375,
      "learning_rate": 6.846996372076786e-06,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1245,
      "tokens_per_second_per_gpu": 17828.4,
      "total_tokens": 112537180
    },
    {
      "epoch": 0.24891374918843331,
      "grad_norm": 0.72265625,
      "learning_rate": 6.831309394204957e-06,
      "loss": 0.3701,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1246,
      "tokens_per_second_per_gpu": 19282.7,
      "total_tokens": 112630540
    },
    {
      "epoch": 0.24911351945262947,
      "grad_norm": 0.71484375,
      "learning_rate": 6.815631079412249e-06,
      "loss": 0.3998,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1247,
      "tokens_per_second_per_gpu": 18574.48,
      "total_tokens": 112720645
    },
    {
      "epoch": 0.24931328971682565,
      "grad_norm": 0.7265625,
      "learning_rate": 6.7999614705625725e-06,
      "loss": 0.4219,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1248,
      "tokens_per_second_per_gpu": 18929.62,
      "total_tokens": 112812550
    },
    {
      "epoch": 0.24951305998102183,
      "grad_norm": 0.71875,
      "learning_rate": 6.784300610496049e-06,
      "loss": 0.3833,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1249,
      "tokens_per_second_per_gpu": 17577.33,
      "total_tokens": 112897935
    },
    {
      "epoch": 0.249712830245218,
      "grad_norm": 0.69921875,
      "learning_rate": 6.7686485420288576e-06,
      "loss": 0.3867,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1250,
      "tokens_per_second_per_gpu": 17590.09,
      "total_tokens": 112985083
    },
    {
      "epoch": 0.24991260050941416,
      "grad_norm": 0.703125,
      "learning_rate": 6.7530053079531664e-06,
      "loss": 0.4061,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1251,
      "tokens_per_second_per_gpu": 19854.98,
      "total_tokens": 113082133
    },
    {
      "epoch": 0.25011237077361037,
      "grad_norm": 0.69921875,
      "learning_rate": 6.737370951036977e-06,
      "loss": 0.3768,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1252,
      "tokens_per_second_per_gpu": 18214.03,
      "total_tokens": 113170026
    },
    {
      "epoch": 0.2503121410378065,
      "grad_norm": 0.73046875,
      "learning_rate": 6.721745514024023e-06,
      "loss": 0.3742,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1253,
      "tokens_per_second_per_gpu": 16418.72,
      "total_tokens": 113250058
    },
    {
      "epoch": 0.2505119113020027,
      "grad_norm": 0.69921875,
      "learning_rate": 6.706129039633659e-06,
      "loss": 0.3608,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1254,
      "tokens_per_second_per_gpu": 16345.27,
      "total_tokens": 113330448
    },
    {
      "epoch": 0.2507116815661989,
      "grad_norm": 0.71875,
      "learning_rate": 6.690521570560717e-06,
      "loss": 0.3714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1255,
      "tokens_per_second_per_gpu": 17268.78,
      "total_tokens": 113415162
    },
    {
      "epoch": 0.25091145183039504,
      "grad_norm": 0.71484375,
      "learning_rate": 6.674923149475433e-06,
      "loss": 0.3637,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1256,
      "tokens_per_second_per_gpu": 17429.94,
      "total_tokens": 113500583
    },
    {
      "epoch": 0.25111122209459125,
      "grad_norm": 0.703125,
      "learning_rate": 6.659333819023291e-06,
      "loss": 0.3589,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1257,
      "tokens_per_second_per_gpu": 16233.32,
      "total_tokens": 113581358
    },
    {
      "epoch": 0.2513109923587874,
      "grad_norm": 0.6875,
      "learning_rate": 6.643753621824922e-06,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1258,
      "tokens_per_second_per_gpu": 18648.99,
      "total_tokens": 113671097
    },
    {
      "epoch": 0.25151076262298355,
      "grad_norm": 0.73828125,
      "learning_rate": 6.6281826004759985e-06,
      "loss": 0.3867,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1259,
      "tokens_per_second_per_gpu": 18413.35,
      "total_tokens": 113760506
    },
    {
      "epoch": 0.25171053288717976,
      "grad_norm": 0.68359375,
      "learning_rate": 6.612620797547087e-06,
      "loss": 0.4056,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1260,
      "tokens_per_second_per_gpu": 19081.08,
      "total_tokens": 113854296
    },
    {
      "epoch": 0.2519103031513759,
      "grad_norm": 0.71484375,
      "learning_rate": 6.59706825558357e-06,
      "loss": 0.3985,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1261,
      "tokens_per_second_per_gpu": 18889.58,
      "total_tokens": 113946914
    },
    {
      "epoch": 0.25211007341557207,
      "grad_norm": 0.6875,
      "learning_rate": 6.581525017105501e-06,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1262,
      "tokens_per_second_per_gpu": 18912.26,
      "total_tokens": 114040193
    },
    {
      "epoch": 0.2523098436797683,
      "grad_norm": 0.69140625,
      "learning_rate": 6.565991124607507e-06,
      "loss": 0.3927,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1263,
      "tokens_per_second_per_gpu": 19241.07,
      "total_tokens": 114135089
    },
    {
      "epoch": 0.25250961394396443,
      "grad_norm": 0.72265625,
      "learning_rate": 6.5504666205586435e-06,
      "loss": 0.4025,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1264,
      "tokens_per_second_per_gpu": 19304.85,
      "total_tokens": 114228875
    },
    {
      "epoch": 0.25270938420816064,
      "grad_norm": 0.703125,
      "learning_rate": 6.534951547402322e-06,
      "loss": 0.3997,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1265,
      "tokens_per_second_per_gpu": 18129.42,
      "total_tokens": 114318557
    },
    {
      "epoch": 0.2529091544723568,
      "grad_norm": 0.70703125,
      "learning_rate": 6.519445947556156e-06,
      "loss": 0.3441,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1266,
      "tokens_per_second_per_gpu": 16667.84,
      "total_tokens": 114401302
    },
    {
      "epoch": 0.25310892473655294,
      "grad_norm": 0.6875,
      "learning_rate": 6.503949863411866e-06,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1267,
      "tokens_per_second_per_gpu": 18248.85,
      "total_tokens": 114490536
    },
    {
      "epoch": 0.25330869500074915,
      "grad_norm": 0.70703125,
      "learning_rate": 6.488463337335157e-06,
      "loss": 0.4116,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1268,
      "tokens_per_second_per_gpu": 18794.03,
      "total_tokens": 114581847
    },
    {
      "epoch": 0.2535084652649453,
      "grad_norm": 0.72265625,
      "learning_rate": 6.47298641166559e-06,
      "loss": 0.3858,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1269,
      "tokens_per_second_per_gpu": 19215.58,
      "total_tokens": 114673842
    },
    {
      "epoch": 0.2537082355291415,
      "grad_norm": 0.75,
      "learning_rate": 6.457519128716496e-06,
      "loss": 0.3823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1270,
      "tokens_per_second_per_gpu": 17354.04,
      "total_tokens": 114758170
    },
    {
      "epoch": 0.25390800579333767,
      "grad_norm": 0.703125,
      "learning_rate": 6.442061530774835e-06,
      "loss": 0.3752,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1271,
      "tokens_per_second_per_gpu": 17665.72,
      "total_tokens": 114845808
    },
    {
      "epoch": 0.2541077760575338,
      "grad_norm": 0.6953125,
      "learning_rate": 6.426613660101091e-06,
      "loss": 0.4068,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1272,
      "tokens_per_second_per_gpu": 19916.63,
      "total_tokens": 114942637
    },
    {
      "epoch": 0.25430754632173,
      "grad_norm": 0.69140625,
      "learning_rate": 6.411175558929152e-06,
      "loss": 0.3627,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1273,
      "tokens_per_second_per_gpu": 17604.44,
      "total_tokens": 115028235
    },
    {
      "epoch": 0.2545073165859262,
      "grad_norm": 0.69140625,
      "learning_rate": 6.395747269466192e-06,
      "loss": 0.3962,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1274,
      "tokens_per_second_per_gpu": 19419.41,
      "total_tokens": 115122237
    },
    {
      "epoch": 0.25470708685012233,
      "grad_norm": 0.68359375,
      "learning_rate": 6.380328833892571e-06,
      "loss": 0.4064,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1275,
      "tokens_per_second_per_gpu": 19899.88,
      "total_tokens": 115217764
    },
    {
      "epoch": 0.25490685711431854,
      "grad_norm": 0.7109375,
      "learning_rate": 6.364920294361701e-06,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1276,
      "tokens_per_second_per_gpu": 18779.39,
      "total_tokens": 115307896
    },
    {
      "epoch": 0.2551066273785147,
      "grad_norm": 0.66015625,
      "learning_rate": 6.349521692999945e-06,
      "loss": 0.433,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1277,
      "tokens_per_second_per_gpu": 20683.78,
      "total_tokens": 115408992
    },
    {
      "epoch": 0.2553063976427109,
      "grad_norm": 0.73046875,
      "learning_rate": 6.334133071906483e-06,
      "loss": 0.3775,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1278,
      "tokens_per_second_per_gpu": 18086.95,
      "total_tokens": 115496350
    },
    {
      "epoch": 0.25550616790690706,
      "grad_norm": 0.66796875,
      "learning_rate": 6.318754473153221e-06,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1279,
      "tokens_per_second_per_gpu": 20052.02,
      "total_tokens": 115593135
    },
    {
      "epoch": 0.2557059381711032,
      "grad_norm": 0.69140625,
      "learning_rate": 6.303385938784663e-06,
      "loss": 0.3617,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1280,
      "tokens_per_second_per_gpu": 18744.21,
      "total_tokens": 115683030
    },
    {
      "epoch": 0.2559057084352994,
      "grad_norm": 0.69921875,
      "learning_rate": 6.2880275108177915e-06,
      "loss": 0.351,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1281,
      "tokens_per_second_per_gpu": 17847.54,
      "total_tokens": 115768458
    },
    {
      "epoch": 0.25610547869949557,
      "grad_norm": 0.6875,
      "learning_rate": 6.27267923124197e-06,
      "loss": 0.3945,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1282,
      "tokens_per_second_per_gpu": 19498.51,
      "total_tokens": 115862515
    },
    {
      "epoch": 0.2563052489636918,
      "grad_norm": 0.69921875,
      "learning_rate": 6.257341142018798e-06,
      "loss": 0.4019,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1283,
      "tokens_per_second_per_gpu": 17969.33,
      "total_tokens": 115951239
    },
    {
      "epoch": 0.25650501922788793,
      "grad_norm": 0.69921875,
      "learning_rate": 6.2420132850820296e-06,
      "loss": 0.3722,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1284,
      "tokens_per_second_per_gpu": 17824.52,
      "total_tokens": 116037883
    },
    {
      "epoch": 0.2567047894920841,
      "grad_norm": 0.69921875,
      "learning_rate": 6.2266957023374434e-06,
      "loss": 0.3746,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1285,
      "tokens_per_second_per_gpu": 17943.15,
      "total_tokens": 116125963
    },
    {
      "epoch": 0.2569045597562803,
      "grad_norm": 0.70703125,
      "learning_rate": 6.211388435662722e-06,
      "loss": 0.3343,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1286,
      "tokens_per_second_per_gpu": 16099.92,
      "total_tokens": 116205104
    },
    {
      "epoch": 0.25710433002047645,
      "grad_norm": 0.73046875,
      "learning_rate": 6.196091526907355e-06,
      "loss": 0.3606,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1287,
      "tokens_per_second_per_gpu": 18009.09,
      "total_tokens": 116293479
    },
    {
      "epoch": 0.2573041002846726,
      "grad_norm": 0.7109375,
      "learning_rate": 6.180805017892498e-06,
      "loss": 0.3877,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1288,
      "tokens_per_second_per_gpu": 18010.49,
      "total_tokens": 116380791
    },
    {
      "epoch": 0.2575038705488688,
      "grad_norm": 0.71484375,
      "learning_rate": 6.165528950410884e-06,
      "loss": 0.3761,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1289,
      "tokens_per_second_per_gpu": 18194.29,
      "total_tokens": 116468964
    },
    {
      "epoch": 0.25770364081306496,
      "grad_norm": 0.7421875,
      "learning_rate": 6.150263366226703e-06,
      "loss": 0.3826,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1290,
      "tokens_per_second_per_gpu": 18175.01,
      "total_tokens": 116558495
    },
    {
      "epoch": 0.25790341107726117,
      "grad_norm": 0.7265625,
      "learning_rate": 6.13500830707548e-06,
      "loss": 0.3855,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1291,
      "tokens_per_second_per_gpu": 18177.76,
      "total_tokens": 116645941
    },
    {
      "epoch": 0.2581031813414573,
      "grad_norm": 0.70703125,
      "learning_rate": 6.119763814663958e-06,
      "loss": 0.4093,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1292,
      "tokens_per_second_per_gpu": 19877.8,
      "total_tokens": 116742668
    },
    {
      "epoch": 0.2583029516056535,
      "grad_norm": 0.84765625,
      "learning_rate": 6.10452993067e-06,
      "loss": 0.3914,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1293,
      "tokens_per_second_per_gpu": 17741.56,
      "total_tokens": 116829781
    },
    {
      "epoch": 0.2585027218698497,
      "grad_norm": 0.72265625,
      "learning_rate": 6.089306696742461e-06,
      "loss": 0.3604,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1294,
      "tokens_per_second_per_gpu": 16687.46,
      "total_tokens": 116912032
    },
    {
      "epoch": 0.25870249213404584,
      "grad_norm": 0.6796875,
      "learning_rate": 6.074094154501087e-06,
      "loss": 0.3958,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1295,
      "tokens_per_second_per_gpu": 18968.03,
      "total_tokens": 117005879
    },
    {
      "epoch": 0.25890226239824204,
      "grad_norm": 0.69921875,
      "learning_rate": 6.058892345536387e-06,
      "loss": 0.3802,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1296,
      "tokens_per_second_per_gpu": 18862.33,
      "total_tokens": 117096015
    },
    {
      "epoch": 0.2591020326624382,
      "grad_norm": 0.73828125,
      "learning_rate": 6.04370131140952e-06,
      "loss": 0.3873,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1297,
      "tokens_per_second_per_gpu": 17132.99,
      "total_tokens": 117178620
    },
    {
      "epoch": 0.25930180292663435,
      "grad_norm": 0.71484375,
      "learning_rate": 6.028521093652195e-06,
      "loss": 0.3895,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1298,
      "tokens_per_second_per_gpu": 17725.99,
      "total_tokens": 117265087
    },
    {
      "epoch": 0.25950157319083056,
      "grad_norm": 0.71484375,
      "learning_rate": 6.0133517337665504e-06,
      "loss": 0.3914,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1299,
      "tokens_per_second_per_gpu": 19134.83,
      "total_tokens": 117357820
    },
    {
      "epoch": 0.2597013434550267,
      "grad_norm": 0.70703125,
      "learning_rate": 5.9981932732250356e-06,
      "loss": 0.3994,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1300,
      "tokens_per_second_per_gpu": 18771.97,
      "total_tokens": 117447478
    },
    {
      "epoch": 0.2599011137192229,
      "grad_norm": 0.76953125,
      "learning_rate": 5.983045753470308e-06,
      "loss": 0.4067,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1301,
      "tokens_per_second_per_gpu": 17762.5,
      "total_tokens": 117533432
    },
    {
      "epoch": 0.2601008839834191,
      "grad_norm": 0.734375,
      "learning_rate": 5.9679092159151e-06,
      "loss": 0.3915,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1302,
      "tokens_per_second_per_gpu": 18233.51,
      "total_tokens": 117621623
    },
    {
      "epoch": 0.2603006542476152,
      "grad_norm": 0.68359375,
      "learning_rate": 5.952783701942129e-06,
      "loss": 0.3783,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1303,
      "tokens_per_second_per_gpu": 18227.46,
      "total_tokens": 117709425
    },
    {
      "epoch": 0.26050042451181143,
      "grad_norm": 0.68359375,
      "learning_rate": 5.937669252903974e-06,
      "loss": 0.3883,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1304,
      "tokens_per_second_per_gpu": 19646.77,
      "total_tokens": 117804473
    },
    {
      "epoch": 0.2607001947760076,
      "grad_norm": 0.70703125,
      "learning_rate": 5.922565910122967e-06,
      "loss": 0.38,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1305,
      "tokens_per_second_per_gpu": 16877.95,
      "total_tokens": 117887466
    },
    {
      "epoch": 0.26089996504020374,
      "grad_norm": 0.73046875,
      "learning_rate": 5.907473714891061e-06,
      "loss": 0.4219,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1306,
      "tokens_per_second_per_gpu": 18898.72,
      "total_tokens": 117978170
    },
    {
      "epoch": 0.26109973530439995,
      "grad_norm": 0.69140625,
      "learning_rate": 5.8923927084697475e-06,
      "loss": 0.3961,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1307,
      "tokens_per_second_per_gpu": 20106.52,
      "total_tokens": 118076065
    },
    {
      "epoch": 0.2612995055685961,
      "grad_norm": 0.7109375,
      "learning_rate": 5.8773229320899195e-06,
      "loss": 0.393,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1308,
      "tokens_per_second_per_gpu": 19582.05,
      "total_tokens": 118172348
    },
    {
      "epoch": 0.2614992758327923,
      "grad_norm": 0.671875,
      "learning_rate": 5.862264426951768e-06,
      "loss": 0.3812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1309,
      "tokens_per_second_per_gpu": 19150.68,
      "total_tokens": 118265305
    },
    {
      "epoch": 0.26169904609698846,
      "grad_norm": 0.69140625,
      "learning_rate": 5.8472172342246816e-06,
      "loss": 0.3673,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1310,
      "tokens_per_second_per_gpu": 18842.93,
      "total_tokens": 118355964
    },
    {
      "epoch": 0.2618988163611846,
      "grad_norm": 0.69140625,
      "learning_rate": 5.832181395047099e-06,
      "loss": 0.406,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1311,
      "tokens_per_second_per_gpu": 18911.14,
      "total_tokens": 118449186
    },
    {
      "epoch": 0.2620985866253808,
      "grad_norm": 0.69921875,
      "learning_rate": 5.817156950526432e-06,
      "loss": 0.4171,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1312,
      "tokens_per_second_per_gpu": 19351.3,
      "total_tokens": 118543212
    },
    {
      "epoch": 0.262298356889577,
      "grad_norm": 0.9140625,
      "learning_rate": 5.802143941738945e-06,
      "loss": 0.3672,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1313,
      "tokens_per_second_per_gpu": 16411.29,
      "total_tokens": 118624224
    },
    {
      "epoch": 0.2624981271537732,
      "grad_norm": 0.68359375,
      "learning_rate": 5.787142409729622e-06,
      "loss": 0.3484,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1314,
      "tokens_per_second_per_gpu": 17817.59,
      "total_tokens": 118711617
    },
    {
      "epoch": 0.26269789741796934,
      "grad_norm": 0.69140625,
      "learning_rate": 5.772152395512087e-06,
      "loss": 0.4104,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1315,
      "tokens_per_second_per_gpu": 19718.17,
      "total_tokens": 118807975
    },
    {
      "epoch": 0.2628976676821655,
      "grad_norm": 0.7109375,
      "learning_rate": 5.7571739400684644e-06,
      "loss": 0.389,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1316,
      "tokens_per_second_per_gpu": 17271.73,
      "total_tokens": 118893305
    },
    {
      "epoch": 0.2630974379463617,
      "grad_norm": 0.73046875,
      "learning_rate": 5.742207084349274e-06,
      "loss": 0.413,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1317,
      "tokens_per_second_per_gpu": 19147.18,
      "total_tokens": 118984629
    },
    {
      "epoch": 0.26329720821055785,
      "grad_norm": 0.7265625,
      "learning_rate": 5.727251869273333e-06,
      "loss": 0.3945,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1318,
      "tokens_per_second_per_gpu": 18631.17,
      "total_tokens": 119074071
    },
    {
      "epoch": 0.263496978474754,
      "grad_norm": 0.890625,
      "learning_rate": 5.712308335727629e-06,
      "loss": 0.4164,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1319,
      "tokens_per_second_per_gpu": 19038.23,
      "total_tokens": 119166518
    },
    {
      "epoch": 0.2636967487389502,
      "grad_norm": 0.6953125,
      "learning_rate": 5.697376524567205e-06,
      "loss": 0.3802,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1320,
      "tokens_per_second_per_gpu": 19218.23,
      "total_tokens": 119259660
    },
    {
      "epoch": 0.26389651900314637,
      "grad_norm": 0.703125,
      "learning_rate": 5.6824564766150724e-06,
      "loss": 0.4048,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1321,
      "tokens_per_second_per_gpu": 19931.25,
      "total_tokens": 119356240
    },
    {
      "epoch": 0.2640962892673426,
      "grad_norm": 0.69140625,
      "learning_rate": 5.667548232662064e-06,
      "loss": 0.3817,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1322,
      "tokens_per_second_per_gpu": 18917.51,
      "total_tokens": 119447583
    },
    {
      "epoch": 0.26429605953153873,
      "grad_norm": 0.80078125,
      "learning_rate": 5.652651833466756e-06,
      "loss": 0.3874,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1323,
      "tokens_per_second_per_gpu": 18881.23,
      "total_tokens": 119538571
    },
    {
      "epoch": 0.2644958297957349,
      "grad_norm": 0.67578125,
      "learning_rate": 5.6377673197553325e-06,
      "loss": 0.4082,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1324,
      "tokens_per_second_per_gpu": 21832.87,
      "total_tokens": 119640688
    },
    {
      "epoch": 0.2646956000599311,
      "grad_norm": 0.69921875,
      "learning_rate": 5.622894732221482e-06,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1325,
      "tokens_per_second_per_gpu": 19426.6,
      "total_tokens": 119734584
    },
    {
      "epoch": 0.26489537032412724,
      "grad_norm": 0.74609375,
      "learning_rate": 5.608034111526298e-06,
      "loss": 0.4027,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1326,
      "tokens_per_second_per_gpu": 16721.44,
      "total_tokens": 119817576
    },
    {
      "epoch": 0.26509514058832345,
      "grad_norm": 1.125,
      "learning_rate": 5.593185498298142e-06,
      "loss": 0.3735,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1327,
      "tokens_per_second_per_gpu": 17244.51,
      "total_tokens": 119903014
    },
    {
      "epoch": 0.2652949108525196,
      "grad_norm": 0.70703125,
      "learning_rate": 5.578348933132565e-06,
      "loss": 0.3986,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1328,
      "tokens_per_second_per_gpu": 19647.68,
      "total_tokens": 119998929
    },
    {
      "epoch": 0.26549468111671576,
      "grad_norm": 0.66015625,
      "learning_rate": 5.563524456592163e-06,
      "loss": 0.3641,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1329,
      "tokens_per_second_per_gpu": 18918.07,
      "total_tokens": 120091778
    },
    {
      "epoch": 0.26569445138091197,
      "grad_norm": 0.71484375,
      "learning_rate": 5.548712109206487e-06,
      "loss": 0.4187,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1330,
      "tokens_per_second_per_gpu": 17785.66,
      "total_tokens": 120178716
    },
    {
      "epoch": 0.2658942216451081,
      "grad_norm": 0.6953125,
      "learning_rate": 5.533911931471936e-06,
      "loss": 0.4114,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1331,
      "tokens_per_second_per_gpu": 19826.45,
      "total_tokens": 120274199
    },
    {
      "epoch": 0.2660939919093043,
      "grad_norm": 0.7109375,
      "learning_rate": 5.519123963851624e-06,
      "loss": 0.3688,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1332,
      "tokens_per_second_per_gpu": 17147.48,
      "total_tokens": 120358231
    },
    {
      "epoch": 0.2662937621735005,
      "grad_norm": 0.79296875,
      "learning_rate": 5.504348246775299e-06,
      "loss": 0.3911,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1333,
      "tokens_per_second_per_gpu": 16940.83,
      "total_tokens": 120441501
    },
    {
      "epoch": 0.26649353243769663,
      "grad_norm": 0.71875,
      "learning_rate": 5.489584820639195e-06,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1334,
      "tokens_per_second_per_gpu": 20087.4,
      "total_tokens": 120537188
    },
    {
      "epoch": 0.26669330270189284,
      "grad_norm": 0.71484375,
      "learning_rate": 5.474833725805963e-06,
      "loss": 0.4201,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1335,
      "tokens_per_second_per_gpu": 19940.46,
      "total_tokens": 120633116
    },
    {
      "epoch": 0.266893072966089,
      "grad_norm": 0.70703125,
      "learning_rate": 5.460095002604533e-06,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1336,
      "tokens_per_second_per_gpu": 18205.32,
      "total_tokens": 120721963
    },
    {
      "epoch": 0.26709284323028515,
      "grad_norm": 0.703125,
      "learning_rate": 5.445368691330008e-06,
      "loss": 0.3783,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1337,
      "tokens_per_second_per_gpu": 18753.57,
      "total_tokens": 120812184
    },
    {
      "epoch": 0.26729261349448136,
      "grad_norm": 0.69921875,
      "learning_rate": 5.430654832243569e-06,
      "loss": 0.4005,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1338,
      "tokens_per_second_per_gpu": 18829.77,
      "total_tokens": 120904580
    },
    {
      "epoch": 0.2674923837586775,
      "grad_norm": 0.67578125,
      "learning_rate": 5.415953465572332e-06,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1339,
      "tokens_per_second_per_gpu": 18986.72,
      "total_tokens": 120998163
    },
    {
      "epoch": 0.2676921540228737,
      "grad_norm": 0.7265625,
      "learning_rate": 5.401264631509277e-06,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1340,
      "tokens_per_second_per_gpu": 18122.03,
      "total_tokens": 121085672
    },
    {
      "epoch": 0.26789192428706987,
      "grad_norm": 0.6875,
      "learning_rate": 5.386588370213124e-06,
      "loss": 0.3952,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1341,
      "tokens_per_second_per_gpu": 19134.91,
      "total_tokens": 121178501
    },
    {
      "epoch": 0.268091694551266,
      "grad_norm": 0.6953125,
      "learning_rate": 5.371924721808197e-06,
      "loss": 0.3807,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1342,
      "tokens_per_second_per_gpu": 17726.59,
      "total_tokens": 121265734
    },
    {
      "epoch": 0.26829146481546223,
      "grad_norm": 0.7109375,
      "learning_rate": 5.357273726384368e-06,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1343,
      "tokens_per_second_per_gpu": 18075.41,
      "total_tokens": 121355604
    },
    {
      "epoch": 0.2684912350796584,
      "grad_norm": 0.70703125,
      "learning_rate": 5.3426354239968804e-06,
      "loss": 0.3884,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1344,
      "tokens_per_second_per_gpu": 18534.98,
      "total_tokens": 121444650
    },
    {
      "epoch": 0.2686910053438546,
      "grad_norm": 0.671875,
      "learning_rate": 5.328009854666303e-06,
      "loss": 0.41,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1345,
      "tokens_per_second_per_gpu": 19401.52,
      "total_tokens": 121540030
    },
    {
      "epoch": 0.26889077560805075,
      "grad_norm": 0.69140625,
      "learning_rate": 5.3133970583783865e-06,
      "loss": 0.3798,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1346,
      "tokens_per_second_per_gpu": 18638.39,
      "total_tokens": 121630152
    },
    {
      "epoch": 0.2690905458722469,
      "grad_norm": 0.67578125,
      "learning_rate": 5.298797075083956e-06,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1347,
      "tokens_per_second_per_gpu": 19034.84,
      "total_tokens": 121722697
    },
    {
      "epoch": 0.2692903161364431,
      "grad_norm": 0.69921875,
      "learning_rate": 5.284209944698806e-06,
      "loss": 0.3927,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1348,
      "tokens_per_second_per_gpu": 18661.79,
      "total_tokens": 121813698
    },
    {
      "epoch": 0.26949008640063926,
      "grad_norm": 0.68359375,
      "learning_rate": 5.269635707103593e-06,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1349,
      "tokens_per_second_per_gpu": 19107.35,
      "total_tokens": 121906171
    },
    {
      "epoch": 0.2696898566648354,
      "grad_norm": 0.73046875,
      "learning_rate": 5.255074402143727e-06,
      "loss": 0.3678,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1350,
      "tokens_per_second_per_gpu": 17265.48,
      "total_tokens": 121988745
    },
    {
      "epoch": 0.2698896269290316,
      "grad_norm": 0.734375,
      "learning_rate": 5.240526069629265e-06,
      "loss": 0.415,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1351,
      "tokens_per_second_per_gpu": 18116.8,
      "total_tokens": 122077407
    },
    {
      "epoch": 0.2700893971932278,
      "grad_norm": 0.7109375,
      "learning_rate": 5.225990749334789e-06,
      "loss": 0.4169,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1352,
      "tokens_per_second_per_gpu": 19678.35,
      "total_tokens": 122171397
    },
    {
      "epoch": 0.270289167457424,
      "grad_norm": 0.6796875,
      "learning_rate": 5.211468480999304e-06,
      "loss": 0.3877,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1353,
      "tokens_per_second_per_gpu": 19511.7,
      "total_tokens": 122264431
    },
    {
      "epoch": 0.27048893772162014,
      "grad_norm": 0.703125,
      "learning_rate": 5.196959304326148e-06,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1354,
      "tokens_per_second_per_gpu": 18269.22,
      "total_tokens": 122353625
    },
    {
      "epoch": 0.2706887079858163,
      "grad_norm": 0.66015625,
      "learning_rate": 5.1824632589828465e-06,
      "loss": 0.3965,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1355,
      "tokens_per_second_per_gpu": 19653.79,
      "total_tokens": 122449636
    },
    {
      "epoch": 0.2708884782500125,
      "grad_norm": 0.66796875,
      "learning_rate": 5.167980384601041e-06,
      "loss": 0.3673,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1356,
      "tokens_per_second_per_gpu": 19063.39,
      "total_tokens": 122542012
    },
    {
      "epoch": 0.27108824851420865,
      "grad_norm": 0.6875,
      "learning_rate": 5.153510720776354e-06,
      "loss": 0.3701,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1357,
      "tokens_per_second_per_gpu": 18042.69,
      "total_tokens": 122630637
    },
    {
      "epoch": 0.27128801877840486,
      "grad_norm": 0.6796875,
      "learning_rate": 5.139054307068289e-06,
      "loss": 0.3818,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1358,
      "tokens_per_second_per_gpu": 18587.56,
      "total_tokens": 122719900
    },
    {
      "epoch": 0.271487789042601,
      "grad_norm": 0.68359375,
      "learning_rate": 5.124611183000138e-06,
      "loss": 0.4045,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1359,
      "tokens_per_second_per_gpu": 19336.16,
      "total_tokens": 122813379
    },
    {
      "epoch": 0.27168755930679717,
      "grad_norm": 0.6796875,
      "learning_rate": 5.110181388058842e-06,
      "loss": 0.397,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1360,
      "tokens_per_second_per_gpu": 18154.89,
      "total_tokens": 122903189
    },
    {
      "epoch": 0.2718873295709934,
      "grad_norm": 0.6953125,
      "learning_rate": 5.095764961694923e-06,
      "loss": 0.404,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1361,
      "tokens_per_second_per_gpu": 19857.56,
      "total_tokens": 122997997
    },
    {
      "epoch": 0.2720870998351895,
      "grad_norm": 0.71875,
      "learning_rate": 5.081361943322323e-06,
      "loss": 0.4021,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1362,
      "tokens_per_second_per_gpu": 18675.68,
      "total_tokens": 123087275
    },
    {
      "epoch": 0.2722868700993857,
      "grad_norm": 0.69921875,
      "learning_rate": 5.066972372318351e-06,
      "loss": 0.4049,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1363,
      "tokens_per_second_per_gpu": 19753.94,
      "total_tokens": 123182430
    },
    {
      "epoch": 0.2724866403635819,
      "grad_norm": 0.71875,
      "learning_rate": 5.05259628802355e-06,
      "loss": 0.3977,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1364,
      "tokens_per_second_per_gpu": 19594.57,
      "total_tokens": 123276116
    },
    {
      "epoch": 0.27268641062777804,
      "grad_norm": 0.71875,
      "learning_rate": 5.0382337297415775e-06,
      "loss": 0.3814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1365,
      "tokens_per_second_per_gpu": 17457.59,
      "total_tokens": 123361602
    },
    {
      "epoch": 0.27288618089197425,
      "grad_norm": 0.69140625,
      "learning_rate": 5.023884736739132e-06,
      "loss": 0.4148,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1366,
      "tokens_per_second_per_gpu": 20752.28,
      "total_tokens": 123462700
    },
    {
      "epoch": 0.2730859511561704,
      "grad_norm": 0.69140625,
      "learning_rate": 5.009549348245795e-06,
      "loss": 0.3652,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1367,
      "tokens_per_second_per_gpu": 18368.91,
      "total_tokens": 123552335
    },
    {
      "epoch": 0.27328572142036656,
      "grad_norm": 0.703125,
      "learning_rate": 4.995227603453981e-06,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1368,
      "tokens_per_second_per_gpu": 19694.15,
      "total_tokens": 123649437
    },
    {
      "epoch": 0.27348549168456276,
      "grad_norm": 0.68359375,
      "learning_rate": 4.980919541518796e-06,
      "loss": 0.3839,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1369,
      "tokens_per_second_per_gpu": 18358.73,
      "total_tokens": 123738004
    },
    {
      "epoch": 0.2736852619487589,
      "grad_norm": 0.79296875,
      "learning_rate": 4.966625201557931e-06,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1370,
      "tokens_per_second_per_gpu": 18521.59,
      "total_tokens": 123826748
    },
    {
      "epoch": 0.2738850322129551,
      "grad_norm": 0.66796875,
      "learning_rate": 4.952344622651566e-06,
      "loss": 0.3986,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1371,
      "tokens_per_second_per_gpu": 20131.26,
      "total_tokens": 123924212
    },
    {
      "epoch": 0.2740848024771513,
      "grad_norm": 0.703125,
      "learning_rate": 4.938077843842255e-06,
      "loss": 0.386,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1372,
      "tokens_per_second_per_gpu": 18573.69,
      "total_tokens": 124014803
    },
    {
      "epoch": 0.27428457274134743,
      "grad_norm": 0.703125,
      "learning_rate": 4.92382490413483e-06,
      "loss": 0.4007,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1373,
      "tokens_per_second_per_gpu": 18730.23,
      "total_tokens": 124105869
    },
    {
      "epoch": 0.27448434300554364,
      "grad_norm": 0.68359375,
      "learning_rate": 4.909585842496287e-06,
      "loss": 0.3951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1374,
      "tokens_per_second_per_gpu": 18379.51,
      "total_tokens": 124195579
    },
    {
      "epoch": 0.2746841132697398,
      "grad_norm": 0.6796875,
      "learning_rate": 4.895360697855674e-06,
      "loss": 0.4074,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1375,
      "tokens_per_second_per_gpu": 20722.7,
      "total_tokens": 124294207
    },
    {
      "epoch": 0.27488388353393595,
      "grad_norm": 1.2734375,
      "learning_rate": 4.881149509103993e-06,
      "loss": 0.3768,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1376,
      "tokens_per_second_per_gpu": 18565.29,
      "total_tokens": 124384357
    },
    {
      "epoch": 0.27508365379813216,
      "grad_norm": 0.703125,
      "learning_rate": 4.866952315094088e-06,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1377,
      "tokens_per_second_per_gpu": 18532.94,
      "total_tokens": 124474654
    },
    {
      "epoch": 0.2752834240623283,
      "grad_norm": 0.65625,
      "learning_rate": 4.8527691546405476e-06,
      "loss": 0.3814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1378,
      "tokens_per_second_per_gpu": 20103.55,
      "total_tokens": 124572253
    },
    {
      "epoch": 0.2754831943265245,
      "grad_norm": 0.7265625,
      "learning_rate": 4.838600066519597e-06,
      "loss": 0.4048,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1379,
      "tokens_per_second_per_gpu": 18313.21,
      "total_tokens": 124660458
    },
    {
      "epoch": 0.27568296459072067,
      "grad_norm": 0.6875,
      "learning_rate": 4.824445089468975e-06,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1380,
      "tokens_per_second_per_gpu": 19452.76,
      "total_tokens": 124753952
    },
    {
      "epoch": 0.2758827348549168,
      "grad_norm": 0.7265625,
      "learning_rate": 4.8103042621878515e-06,
      "loss": 0.4041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1381,
      "tokens_per_second_per_gpu": 17613.42,
      "total_tokens": 124839133
    },
    {
      "epoch": 0.27608250511911303,
      "grad_norm": 0.71484375,
      "learning_rate": 4.796177623336703e-06,
      "loss": 0.3625,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1382,
      "tokens_per_second_per_gpu": 17420.6,
      "total_tokens": 124925247
    },
    {
      "epoch": 0.2762822753833092,
      "grad_norm": 0.7265625,
      "learning_rate": 4.782065211537226e-06,
      "loss": 0.4069,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1383,
      "tokens_per_second_per_gpu": 17978.9,
      "total_tokens": 125012633
    },
    {
      "epoch": 0.2764820456475054,
      "grad_norm": 0.671875,
      "learning_rate": 4.767967065372221e-06,
      "loss": 0.3891,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1384,
      "tokens_per_second_per_gpu": 19551.34,
      "total_tokens": 125106747
    },
    {
      "epoch": 0.27668181591170155,
      "grad_norm": 0.6875,
      "learning_rate": 4.753883223385467e-06,
      "loss": 0.3795,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1385,
      "tokens_per_second_per_gpu": 18918.63,
      "total_tokens": 125199758
    },
    {
      "epoch": 0.2768815861758977,
      "grad_norm": 0.83203125,
      "learning_rate": 4.739813724081661e-06,
      "loss": 0.3835,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1386,
      "tokens_per_second_per_gpu": 17941.04,
      "total_tokens": 125288801
    },
    {
      "epoch": 0.2770813564400939,
      "grad_norm": 0.734375,
      "learning_rate": 4.7257586059262706e-06,
      "loss": 0.3688,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1387,
      "tokens_per_second_per_gpu": 17431.64,
      "total_tokens": 125371986
    },
    {
      "epoch": 0.27728112670429006,
      "grad_norm": 0.71875,
      "learning_rate": 4.711717907345456e-06,
      "loss": 0.397,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1388,
      "tokens_per_second_per_gpu": 17940.81,
      "total_tokens": 125459050
    },
    {
      "epoch": 0.27748089696848627,
      "grad_norm": 0.7421875,
      "learning_rate": 4.6976916667259555e-06,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1389,
      "tokens_per_second_per_gpu": 17379.37,
      "total_tokens": 125543606
    },
    {
      "epoch": 0.2776806672326824,
      "grad_norm": 0.73828125,
      "learning_rate": 4.683679922414964e-06,
      "loss": 0.3974,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1390,
      "tokens_per_second_per_gpu": 18168.0,
      "total_tokens": 125633859
    },
    {
      "epoch": 0.2778804374968786,
      "grad_norm": 0.68359375,
      "learning_rate": 4.669682712720065e-06,
      "loss": 0.37,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1391,
      "tokens_per_second_per_gpu": 18605.72,
      "total_tokens": 125722934
    },
    {
      "epoch": 0.2780802077610748,
      "grad_norm": 0.67578125,
      "learning_rate": 4.655700075909088e-06,
      "loss": 0.376,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1392,
      "tokens_per_second_per_gpu": 18636.4,
      "total_tokens": 125812949
    },
    {
      "epoch": 0.27827997802527094,
      "grad_norm": 0.83203125,
      "learning_rate": 4.641732050210032e-06,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1393,
      "tokens_per_second_per_gpu": 17748.12,
      "total_tokens": 125898420
    },
    {
      "epoch": 0.2784797482894671,
      "grad_norm": 0.7578125,
      "learning_rate": 4.627778673810953e-06,
      "loss": 0.3979,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1394,
      "tokens_per_second_per_gpu": 18975.71,
      "total_tokens": 125990881
    },
    {
      "epoch": 0.2786795185536633,
      "grad_norm": 0.67578125,
      "learning_rate": 4.613839984859835e-06,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1395,
      "tokens_per_second_per_gpu": 19263.09,
      "total_tokens": 126084476
    },
    {
      "epoch": 0.27887928881785945,
      "grad_norm": 0.67578125,
      "learning_rate": 4.599916021464531e-06,
      "loss": 0.383,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1396,
      "tokens_per_second_per_gpu": 18611.76,
      "total_tokens": 126175413
    },
    {
      "epoch": 0.27907905908205566,
      "grad_norm": 0.671875,
      "learning_rate": 4.58600682169262e-06,
      "loss": 0.3647,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1397,
      "tokens_per_second_per_gpu": 19172.74,
      "total_tokens": 126267781
    },
    {
      "epoch": 0.2792788293462518,
      "grad_norm": 0.765625,
      "learning_rate": 4.57211242357133e-06,
      "loss": 0.3831,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1398,
      "tokens_per_second_per_gpu": 18356.83,
      "total_tokens": 126356685
    },
    {
      "epoch": 0.27947859961044796,
      "grad_norm": 0.671875,
      "learning_rate": 4.5582328650874095e-06,
      "loss": 0.3931,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1399,
      "tokens_per_second_per_gpu": 19440.05,
      "total_tokens": 126451008
    },
    {
      "epoch": 0.2796783698746442,
      "grad_norm": 0.70703125,
      "learning_rate": 4.544368184187039e-06,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1400,
      "tokens_per_second_per_gpu": 19308.54,
      "total_tokens": 126544466
    },
    {
      "epoch": 0.2798781401388403,
      "grad_norm": 0.7109375,
      "learning_rate": 4.530518418775734e-06,
      "loss": 0.397,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1401,
      "tokens_per_second_per_gpu": 18941.97,
      "total_tokens": 126637418
    },
    {
      "epoch": 0.28007791040303653,
      "grad_norm": 0.7109375,
      "learning_rate": 4.5166836067182165e-06,
      "loss": 0.4058,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1402,
      "tokens_per_second_per_gpu": 18751.79,
      "total_tokens": 126727180
    },
    {
      "epoch": 0.2802776806672327,
      "grad_norm": 0.68359375,
      "learning_rate": 4.5028637858383415e-06,
      "loss": 0.3881,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1403,
      "tokens_per_second_per_gpu": 19138.5,
      "total_tokens": 126819503
    },
    {
      "epoch": 0.28047745093142884,
      "grad_norm": 0.7109375,
      "learning_rate": 4.489058993918969e-06,
      "loss": 0.3809,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1404,
      "tokens_per_second_per_gpu": 18800.62,
      "total_tokens": 126910706
    },
    {
      "epoch": 0.28067722119562505,
      "grad_norm": 0.69921875,
      "learning_rate": 4.4752692687018685e-06,
      "loss": 0.3765,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1405,
      "tokens_per_second_per_gpu": 17012.49,
      "total_tokens": 126995494
    },
    {
      "epoch": 0.2808769914598212,
      "grad_norm": 0.69140625,
      "learning_rate": 4.4614946478876305e-06,
      "loss": 0.3616,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1406,
      "tokens_per_second_per_gpu": 18025.85,
      "total_tokens": 127081971
    },
    {
      "epoch": 0.28107676172401735,
      "grad_norm": 0.8671875,
      "learning_rate": 4.447735169135533e-06,
      "loss": 0.348,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1407,
      "tokens_per_second_per_gpu": 14323.13,
      "total_tokens": 127142882
    },
    {
      "epoch": 0.28127653198821356,
      "grad_norm": 0.703125,
      "learning_rate": 4.4339908700634745e-06,
      "loss": 0.3777,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1408,
      "tokens_per_second_per_gpu": 18633.82,
      "total_tokens": 127233089
    },
    {
      "epoch": 0.2814763022524097,
      "grad_norm": 0.65625,
      "learning_rate": 4.420261788247841e-06,
      "loss": 0.4009,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1409,
      "tokens_per_second_per_gpu": 20113.93,
      "total_tokens": 127331256
    },
    {
      "epoch": 0.2816760725166059,
      "grad_norm": 0.703125,
      "learning_rate": 4.406547961223412e-06,
      "loss": 0.3942,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1410,
      "tokens_per_second_per_gpu": 19161.11,
      "total_tokens": 127424788
    },
    {
      "epoch": 0.2818758427808021,
      "grad_norm": 0.7421875,
      "learning_rate": 4.392849426483275e-06,
      "loss": 0.3855,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1411,
      "tokens_per_second_per_gpu": 18427.49,
      "total_tokens": 127514304
    },
    {
      "epoch": 0.28207561304499823,
      "grad_norm": 0.703125,
      "learning_rate": 4.379166221478697e-06,
      "loss": 0.3714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1412,
      "tokens_per_second_per_gpu": 17337.1,
      "total_tokens": 127598340
    },
    {
      "epoch": 0.28227538330919444,
      "grad_norm": 0.69921875,
      "learning_rate": 4.365498383619036e-06,
      "loss": 0.3961,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1413,
      "tokens_per_second_per_gpu": 19649.49,
      "total_tokens": 127691899
    },
    {
      "epoch": 0.2824751535733906,
      "grad_norm": 0.6875,
      "learning_rate": 4.351845950271644e-06,
      "loss": 0.3619,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1414,
      "tokens_per_second_per_gpu": 17911.18,
      "total_tokens": 127780361
    },
    {
      "epoch": 0.2826749238375868,
      "grad_norm": 0.74609375,
      "learning_rate": 4.338208958761747e-06,
      "loss": 0.3971,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1415,
      "tokens_per_second_per_gpu": 19765.14,
      "total_tokens": 127875562
    },
    {
      "epoch": 0.28287469410178295,
      "grad_norm": 0.7109375,
      "learning_rate": 4.324587446372365e-06,
      "loss": 0.4142,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1416,
      "tokens_per_second_per_gpu": 19470.33,
      "total_tokens": 127973346
    },
    {
      "epoch": 0.2830744643659791,
      "grad_norm": 1.109375,
      "learning_rate": 4.3109814503441894e-06,
      "loss": 0.3923,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1417,
      "tokens_per_second_per_gpu": 18539.71,
      "total_tokens": 128064288
    },
    {
      "epoch": 0.2832742346301753,
      "grad_norm": 0.6796875,
      "learning_rate": 4.29739100787549e-06,
      "loss": 0.379,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1418,
      "tokens_per_second_per_gpu": 18771.45,
      "total_tokens": 128155606
    },
    {
      "epoch": 0.28347400489437147,
      "grad_norm": 0.66015625,
      "learning_rate": 4.2838161561220245e-06,
      "loss": 0.384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1419,
      "tokens_per_second_per_gpu": 19927.89,
      "total_tokens": 128251165
    },
    {
      "epoch": 0.2836737751585676,
      "grad_norm": 0.71484375,
      "learning_rate": 4.2702569321969145e-06,
      "loss": 0.4054,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1420,
      "tokens_per_second_per_gpu": 17690.37,
      "total_tokens": 128337239
    },
    {
      "epoch": 0.28387354542276383,
      "grad_norm": 0.69921875,
      "learning_rate": 4.256713373170565e-06,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1421,
      "tokens_per_second_per_gpu": 17651.57,
      "total_tokens": 128425232
    },
    {
      "epoch": 0.28407331568696,
      "grad_norm": 0.6640625,
      "learning_rate": 4.243185516070546e-06,
      "loss": 0.3731,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1422,
      "tokens_per_second_per_gpu": 18743.0,
      "total_tokens": 128517262
    },
    {
      "epoch": 0.2842730859511562,
      "grad_norm": 0.6875,
      "learning_rate": 4.2296733978815e-06,
      "loss": 0.3859,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1423,
      "tokens_per_second_per_gpu": 18624.61,
      "total_tokens": 128607507
    },
    {
      "epoch": 0.28447285621535234,
      "grad_norm": 0.70703125,
      "learning_rate": 4.216177055545048e-06,
      "loss": 0.3882,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1424,
      "tokens_per_second_per_gpu": 18121.26,
      "total_tokens": 128694458
    },
    {
      "epoch": 0.2846726264795485,
      "grad_norm": 0.6875,
      "learning_rate": 4.202696525959667e-06,
      "loss": 0.3972,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1425,
      "tokens_per_second_per_gpu": 19403.15,
      "total_tokens": 128789214
    },
    {
      "epoch": 0.2848723967437447,
      "grad_norm": 0.67578125,
      "learning_rate": 4.189231845980618e-06,
      "loss": 0.389,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1426,
      "tokens_per_second_per_gpu": 19473.97,
      "total_tokens": 128884477
    },
    {
      "epoch": 0.28507216700794086,
      "grad_norm": 0.921875,
      "learning_rate": 4.17578305241982e-06,
      "loss": 0.3959,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1427,
      "tokens_per_second_per_gpu": 18567.36,
      "total_tokens": 128975462
    },
    {
      "epoch": 0.28527193727213707,
      "grad_norm": 0.6796875,
      "learning_rate": 4.162350182045754e-06,
      "loss": 0.4201,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1428,
      "tokens_per_second_per_gpu": 19497.87,
      "total_tokens": 129071677
    },
    {
      "epoch": 0.2854717075363332,
      "grad_norm": 0.6875,
      "learning_rate": 4.148933271583385e-06,
      "loss": 0.4187,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1429,
      "tokens_per_second_per_gpu": 19837.66,
      "total_tokens": 129168300
    },
    {
      "epoch": 0.28567147780052937,
      "grad_norm": 0.6875,
      "learning_rate": 4.135532357714026e-06,
      "loss": 0.4032,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1430,
      "tokens_per_second_per_gpu": 19524.23,
      "total_tokens": 129262577
    },
    {
      "epoch": 0.2858712480647256,
      "grad_norm": 0.734375,
      "learning_rate": 4.12214747707527e-06,
      "loss": 0.405,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1431,
      "tokens_per_second_per_gpu": 18541.21,
      "total_tokens": 129353142
    },
    {
      "epoch": 0.28607101832892173,
      "grad_norm": 0.74609375,
      "learning_rate": 4.1087786662608665e-06,
      "loss": 0.3828,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1432,
      "tokens_per_second_per_gpu": 18235.11,
      "total_tokens": 129442944
    },
    {
      "epoch": 0.28627078859311794,
      "grad_norm": 0.703125,
      "learning_rate": 4.0954259618206295e-06,
      "loss": 0.384,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1433,
      "tokens_per_second_per_gpu": 18311.72,
      "total_tokens": 129530253
    },
    {
      "epoch": 0.2864705588573141,
      "grad_norm": 0.69921875,
      "learning_rate": 4.082089400260351e-06,
      "loss": 0.4169,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1434,
      "tokens_per_second_per_gpu": 19976.04,
      "total_tokens": 129628194
    },
    {
      "epoch": 0.28667032912151025,
      "grad_norm": 0.70703125,
      "learning_rate": 4.068769018041674e-06,
      "loss": 0.3484,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1435,
      "tokens_per_second_per_gpu": 17668.69,
      "total_tokens": 129714746
    },
    {
      "epoch": 0.28687009938570646,
      "grad_norm": 2.40625,
      "learning_rate": 4.055464851582022e-06,
      "loss": 0.3714,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1436,
      "tokens_per_second_per_gpu": 17712.9,
      "total_tokens": 129801264
    },
    {
      "epoch": 0.2870698696499026,
      "grad_norm": 0.66796875,
      "learning_rate": 4.042176937254474e-06,
      "loss": 0.3656,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1437,
      "tokens_per_second_per_gpu": 19358.28,
      "total_tokens": 129893848
    },
    {
      "epoch": 0.28726963991409876,
      "grad_norm": 0.69140625,
      "learning_rate": 4.028905311387678e-06,
      "loss": 0.3823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1438,
      "tokens_per_second_per_gpu": 18424.07,
      "total_tokens": 129983766
    },
    {
      "epoch": 0.28746941017829497,
      "grad_norm": 0.7109375,
      "learning_rate": 4.015650010265757e-06,
      "loss": 0.4036,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1439,
      "tokens_per_second_per_gpu": 19189.88,
      "total_tokens": 130078550
    },
    {
      "epoch": 0.2876691804424911,
      "grad_norm": 0.69140625,
      "learning_rate": 4.002411070128197e-06,
      "loss": 0.3831,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1440,
      "tokens_per_second_per_gpu": 19457.11,
      "total_tokens": 130171761
    },
    {
      "epoch": 0.28786895070668733,
      "grad_norm": 0.68359375,
      "learning_rate": 3.989188527169749e-06,
      "loss": 0.3733,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1441,
      "tokens_per_second_per_gpu": 18332.14,
      "total_tokens": 130261670
    },
    {
      "epoch": 0.2880687209708835,
      "grad_norm": 0.67578125,
      "learning_rate": 3.975982417540351e-06,
      "loss": 0.3967,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1442,
      "tokens_per_second_per_gpu": 19395.36,
      "total_tokens": 130356933
    },
    {
      "epoch": 0.28826849123507964,
      "grad_norm": 0.703125,
      "learning_rate": 3.962792777344992e-06,
      "loss": 0.4166,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1443,
      "tokens_per_second_per_gpu": 19715.86,
      "total_tokens": 130451974
    },
    {
      "epoch": 0.28846826149927585,
      "grad_norm": 0.71875,
      "learning_rate": 3.949619642643654e-06,
      "loss": 0.3658,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1444,
      "tokens_per_second_per_gpu": 17741.46,
      "total_tokens": 130537231
    },
    {
      "epoch": 0.288668031763472,
      "grad_norm": 0.70703125,
      "learning_rate": 3.936463049451179e-06,
      "loss": 0.3935,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1445,
      "tokens_per_second_per_gpu": 18135.77,
      "total_tokens": 130625976
    },
    {
      "epoch": 0.2888678020276682,
      "grad_norm": 0.75,
      "learning_rate": 3.923323033737188e-06,
      "loss": 0.3989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1446,
      "tokens_per_second_per_gpu": 17355.43,
      "total_tokens": 130710202
    },
    {
      "epoch": 0.28906757229186436,
      "grad_norm": 0.67578125,
      "learning_rate": 3.910199631425989e-06,
      "loss": 0.4015,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1447,
      "tokens_per_second_per_gpu": 19922.2,
      "total_tokens": 130806481
    },
    {
      "epoch": 0.2892673425560605,
      "grad_norm": 0.70703125,
      "learning_rate": 3.8970928783964564e-06,
      "loss": 0.3727,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1448,
      "tokens_per_second_per_gpu": 17955.31,
      "total_tokens": 130894818
    },
    {
      "epoch": 0.2894671128202567,
      "grad_norm": 0.69140625,
      "learning_rate": 3.884002810481959e-06,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1449,
      "tokens_per_second_per_gpu": 18930.44,
      "total_tokens": 130987186
    },
    {
      "epoch": 0.2896668830844529,
      "grad_norm": 0.72265625,
      "learning_rate": 3.8709294634702374e-06,
      "loss": 0.3668,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1450,
      "tokens_per_second_per_gpu": 18512.27,
      "total_tokens": 131076323
    },
    {
      "epoch": 0.28986665334864903,
      "grad_norm": 0.671875,
      "learning_rate": 3.857872873103322e-06,
      "loss": 0.4123,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1451,
      "tokens_per_second_per_gpu": 21145.98,
      "total_tokens": 131176049
    },
    {
      "epoch": 0.29006642361284524,
      "grad_norm": 0.71484375,
      "learning_rate": 3.844833075077439e-06,
      "loss": 0.3874,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1452,
      "tokens_per_second_per_gpu": 17997.39,
      "total_tokens": 131263858
    },
    {
      "epoch": 0.2902661938770414,
      "grad_norm": 0.7109375,
      "learning_rate": 3.83181010504289e-06,
      "loss": 0.4014,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1453,
      "tokens_per_second_per_gpu": 18723.94,
      "total_tokens": 131355975
    },
    {
      "epoch": 0.2904659641412376,
      "grad_norm": 0.67578125,
      "learning_rate": 3.818803998603986e-06,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1454,
      "tokens_per_second_per_gpu": 19256.88,
      "total_tokens": 131448172
    },
    {
      "epoch": 0.29066573440543375,
      "grad_norm": 0.68359375,
      "learning_rate": 3.8058147913189215e-06,
      "loss": 0.3995,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1455,
      "tokens_per_second_per_gpu": 19194.29,
      "total_tokens": 131541078
    },
    {
      "epoch": 0.2908655046696299,
      "grad_norm": 0.7109375,
      "learning_rate": 3.792842518699689e-06,
      "loss": 0.3889,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1456,
      "tokens_per_second_per_gpu": 18729.96,
      "total_tokens": 131631665
    },
    {
      "epoch": 0.2910652749338261,
      "grad_norm": 0.6953125,
      "learning_rate": 3.7798872162119948e-06,
      "loss": 0.3822,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1457,
      "tokens_per_second_per_gpu": 19966.53,
      "total_tokens": 131728032
    },
    {
      "epoch": 0.29126504519802227,
      "grad_norm": 0.7109375,
      "learning_rate": 3.766948919275133e-06,
      "loss": 0.4083,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1458,
      "tokens_per_second_per_gpu": 18477.96,
      "total_tokens": 131817725
    },
    {
      "epoch": 0.2914648154622185,
      "grad_norm": 0.703125,
      "learning_rate": 3.754027663261922e-06,
      "loss": 0.3967,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1459,
      "tokens_per_second_per_gpu": 18728.85,
      "total_tokens": 131909633
    },
    {
      "epoch": 0.2916645857264146,
      "grad_norm": 0.6953125,
      "learning_rate": 3.7411234834985765e-06,
      "loss": 0.4064,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1460,
      "tokens_per_second_per_gpu": 20081.13,
      "total_tokens": 132005823
    },
    {
      "epoch": 0.2918643559906108,
      "grad_norm": 0.73046875,
      "learning_rate": 3.72823641526463e-06,
      "loss": 0.3763,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1461,
      "tokens_per_second_per_gpu": 17846.93,
      "total_tokens": 132092188
    },
    {
      "epoch": 0.292064126254807,
      "grad_norm": 0.71875,
      "learning_rate": 3.71536649379284e-06,
      "loss": 0.3482,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1462,
      "tokens_per_second_per_gpu": 16008.09,
      "total_tokens": 132170931
    },
    {
      "epoch": 0.29226389651900314,
      "grad_norm": 0.6640625,
      "learning_rate": 3.702513754269076e-06,
      "loss": 0.3703,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1463,
      "tokens_per_second_per_gpu": 19378.37,
      "total_tokens": 132263704
    },
    {
      "epoch": 0.2924636667831993,
      "grad_norm": 0.68359375,
      "learning_rate": 3.689678231832241e-06,
      "loss": 0.4033,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1464,
      "tokens_per_second_per_gpu": 19784.41,
      "total_tokens": 132360429
    },
    {
      "epoch": 0.2926634370473955,
      "grad_norm": 0.67578125,
      "learning_rate": 3.676859961574162e-06,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1465,
      "tokens_per_second_per_gpu": 18995.97,
      "total_tokens": 132453469
    },
    {
      "epoch": 0.29286320731159166,
      "grad_norm": 0.69921875,
      "learning_rate": 3.6640589785394955e-06,
      "loss": 0.3781,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1466,
      "tokens_per_second_per_gpu": 17099.59,
      "total_tokens": 132540960
    },
    {
      "epoch": 0.29306297757578786,
      "grad_norm": 0.70703125,
      "learning_rate": 3.651275317725648e-06,
      "loss": 0.4289,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1467,
      "tokens_per_second_per_gpu": 17481.25,
      "total_tokens": 132627839
    },
    {
      "epoch": 0.293262747839984,
      "grad_norm": 0.66015625,
      "learning_rate": 3.6385090140826563e-06,
      "loss": 0.3533,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1468,
      "tokens_per_second_per_gpu": 17308.87,
      "total_tokens": 132713065
    },
    {
      "epoch": 0.29346251810418017,
      "grad_norm": 0.6953125,
      "learning_rate": 3.625760102513103e-06,
      "loss": 0.3949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1469,
      "tokens_per_second_per_gpu": 18908.88,
      "total_tokens": 132804477
    },
    {
      "epoch": 0.2936622883683764,
      "grad_norm": 0.6875,
      "learning_rate": 3.6130286178720343e-06,
      "loss": 0.3813,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1470,
      "tokens_per_second_per_gpu": 17692.89,
      "total_tokens": 132890341
    },
    {
      "epoch": 0.29386205863257253,
      "grad_norm": 0.72265625,
      "learning_rate": 3.6003145949668338e-06,
      "loss": 0.3604,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1471,
      "tokens_per_second_per_gpu": 17893.18,
      "total_tokens": 132977129
    },
    {
      "epoch": 0.29406182889676874,
      "grad_norm": 0.74609375,
      "learning_rate": 3.587618068557164e-06,
      "loss": 0.3924,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1472,
      "tokens_per_second_per_gpu": 19870.57,
      "total_tokens": 133071377
    },
    {
      "epoch": 0.2942615991609649,
      "grad_norm": 0.703125,
      "learning_rate": 3.5749390733548382e-06,
      "loss": 0.3856,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1473,
      "tokens_per_second_per_gpu": 17613.33,
      "total_tokens": 133158434
    },
    {
      "epoch": 0.29446136942516105,
      "grad_norm": 0.6640625,
      "learning_rate": 3.5622776440237415e-06,
      "loss": 0.3787,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1474,
      "tokens_per_second_per_gpu": 18793.13,
      "total_tokens": 133250180
    },
    {
      "epoch": 0.29466113968935725,
      "grad_norm": 0.734375,
      "learning_rate": 3.549633815179746e-06,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1475,
      "tokens_per_second_per_gpu": 18166.37,
      "total_tokens": 133338475
    },
    {
      "epoch": 0.2948609099535534,
      "grad_norm": 0.7109375,
      "learning_rate": 3.5370076213905904e-06,
      "loss": 0.3738,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1476,
      "tokens_per_second_per_gpu": 18060.3,
      "total_tokens": 133426511
    },
    {
      "epoch": 0.29506068021774956,
      "grad_norm": 0.6875,
      "learning_rate": 3.5243990971758124e-06,
      "loss": 0.4039,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1477,
      "tokens_per_second_per_gpu": 19324.8,
      "total_tokens": 133517993
    },
    {
      "epoch": 0.29526045048194577,
      "grad_norm": 0.6875,
      "learning_rate": 3.511808277006632e-06,
      "loss": 0.3943,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1478,
      "tokens_per_second_per_gpu": 20295.24,
      "total_tokens": 133615005
    },
    {
      "epoch": 0.2954602207461419,
      "grad_norm": 0.67578125,
      "learning_rate": 3.499235195305868e-06,
      "loss": 0.3823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1479,
      "tokens_per_second_per_gpu": 19079.48,
      "total_tokens": 133706554
    },
    {
      "epoch": 0.29565999101033813,
      "grad_norm": 0.70703125,
      "learning_rate": 3.4866798864478523e-06,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1480,
      "tokens_per_second_per_gpu": 18548.92,
      "total_tokens": 133797045
    },
    {
      "epoch": 0.2958597612745343,
      "grad_norm": 0.70703125,
      "learning_rate": 3.4741423847583134e-06,
      "loss": 0.4091,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1481,
      "tokens_per_second_per_gpu": 19705.9,
      "total_tokens": 133892521
    },
    {
      "epoch": 0.29605953153873044,
      "grad_norm": 0.6796875,
      "learning_rate": 3.4616227245143085e-06,
      "loss": 0.4089,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1482,
      "tokens_per_second_per_gpu": 19258.45,
      "total_tokens": 133988061
    },
    {
      "epoch": 0.29625930180292664,
      "grad_norm": 0.70703125,
      "learning_rate": 3.449120939944107e-06,
      "loss": 0.3804,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1483,
      "tokens_per_second_per_gpu": 18549.44,
      "total_tokens": 134077661
    },
    {
      "epoch": 0.2964590720671228,
      "grad_norm": 0.69921875,
      "learning_rate": 3.4366370652271085e-06,
      "loss": 0.3861,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1484,
      "tokens_per_second_per_gpu": 18200.15,
      "total_tokens": 134165438
    },
    {
      "epoch": 0.296658842331319,
      "grad_norm": 0.734375,
      "learning_rate": 3.4241711344937557e-06,
      "loss": 0.3949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1485,
      "tokens_per_second_per_gpu": 17455.84,
      "total_tokens": 134249113
    },
    {
      "epoch": 0.29685861259551516,
      "grad_norm": 0.70703125,
      "learning_rate": 3.4117231818254205e-06,
      "loss": 0.4043,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1486,
      "tokens_per_second_per_gpu": 19586.54,
      "total_tokens": 134343639
    },
    {
      "epoch": 0.2970583828597113,
      "grad_norm": 0.70703125,
      "learning_rate": 3.3992932412543358e-06,
      "loss": 0.3661,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1487,
      "tokens_per_second_per_gpu": 17665.2,
      "total_tokens": 134429419
    },
    {
      "epoch": 0.2972581531239075,
      "grad_norm": 0.6953125,
      "learning_rate": 3.3868813467634833e-06,
      "loss": 0.3745,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1488,
      "tokens_per_second_per_gpu": 18761.96,
      "total_tokens": 134519407
    },
    {
      "epoch": 0.2974579233881037,
      "grad_norm": 0.71484375,
      "learning_rate": 3.3744875322865035e-06,
      "loss": 0.3937,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1489,
      "tokens_per_second_per_gpu": 18478.1,
      "total_tokens": 134607856
    },
    {
      "epoch": 0.2976576936522999,
      "grad_norm": 0.7109375,
      "learning_rate": 3.3621118317076183e-06,
      "loss": 0.3662,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1490,
      "tokens_per_second_per_gpu": 17467.75,
      "total_tokens": 134693997
    },
    {
      "epoch": 0.29785746391649603,
      "grad_norm": 0.734375,
      "learning_rate": 3.349754278861517e-06,
      "loss": 0.3908,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1491,
      "tokens_per_second_per_gpu": 18889.1,
      "total_tokens": 134785313
    },
    {
      "epoch": 0.2980572341806922,
      "grad_norm": 0.69140625,
      "learning_rate": 3.3374149075332796e-06,
      "loss": 0.3574,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1492,
      "tokens_per_second_per_gpu": 17230.46,
      "total_tokens": 134870872
    },
    {
      "epoch": 0.2982570044448884,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3250937514582758e-06,
      "loss": 0.3885,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1493,
      "tokens_per_second_per_gpu": 19147.28,
      "total_tokens": 134963315
    },
    {
      "epoch": 0.29845677470908455,
      "grad_norm": 0.6796875,
      "learning_rate": 3.3127908443220713e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1494,
      "tokens_per_second_per_gpu": 18954.57,
      "total_tokens": 135055241
    },
    {
      "epoch": 0.2986565449732807,
      "grad_norm": 0.67578125,
      "learning_rate": 3.300506219760351e-06,
      "loss": 0.3734,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1495,
      "tokens_per_second_per_gpu": 19479.05,
      "total_tokens": 135148246
    },
    {
      "epoch": 0.2988563152374769,
      "grad_norm": 0.66796875,
      "learning_rate": 3.288239911358807e-06,
      "loss": 0.4045,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1496,
      "tokens_per_second_per_gpu": 19928.26,
      "total_tokens": 135245192
    },
    {
      "epoch": 0.29905608550167306,
      "grad_norm": 0.71484375,
      "learning_rate": 3.2759919526530536e-06,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1497,
      "tokens_per_second_per_gpu": 18567.43,
      "total_tokens": 135333959
    },
    {
      "epoch": 0.29925585576586927,
      "grad_norm": 0.69921875,
      "learning_rate": 3.2637623771285497e-06,
      "loss": 0.4042,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1498,
      "tokens_per_second_per_gpu": 19251.96,
      "total_tokens": 135426819
    },
    {
      "epoch": 0.2994556260300654,
      "grad_norm": 0.703125,
      "learning_rate": 3.25155121822048e-06,
      "loss": 0.3934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1499,
      "tokens_per_second_per_gpu": 19094.29,
      "total_tokens": 135519758
    },
    {
      "epoch": 0.2996553962942616,
      "grad_norm": 0.69921875,
      "learning_rate": 3.239358509313695e-06,
      "loss": 0.3495,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1500,
      "tokens_per_second_per_gpu": 16922.41,
      "total_tokens": 135602419
    },
    {
      "epoch": 0.2998551665584578,
      "grad_norm": 0.71875,
      "learning_rate": 3.2271842837425917e-06,
      "loss": 0.4059,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1501,
      "tokens_per_second_per_gpu": 18537.64,
      "total_tokens": 135692127
    },
    {
      "epoch": 0.30005493682265394,
      "grad_norm": 0.72265625,
      "learning_rate": 3.2150285747910324e-06,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1502,
      "tokens_per_second_per_gpu": 18553.1,
      "total_tokens": 135780712
    },
    {
      "epoch": 0.30025470708685015,
      "grad_norm": 0.6796875,
      "learning_rate": 3.202891415692271e-06,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1503,
      "tokens_per_second_per_gpu": 20021.2,
      "total_tokens": 135876043
    },
    {
      "epoch": 0.3004544773510463,
      "grad_norm": 0.6796875,
      "learning_rate": 3.1907728396288274e-06,
      "loss": 0.3958,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1504,
      "tokens_per_second_per_gpu": 19481.24,
      "total_tokens": 135971102
    },
    {
      "epoch": 0.30065424761524245,
      "grad_norm": 0.6953125,
      "learning_rate": 3.178672879732435e-06,
      "loss": 0.4049,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1505,
      "tokens_per_second_per_gpu": 20180.52,
      "total_tokens": 136068160
    },
    {
      "epoch": 0.30085401787943866,
      "grad_norm": 0.86328125,
      "learning_rate": 3.1665915690839165e-06,
      "loss": 0.3947,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1506,
      "tokens_per_second_per_gpu": 17880.79,
      "total_tokens": 136156155
    },
    {
      "epoch": 0.3010537881436348,
      "grad_norm": 0.71484375,
      "learning_rate": 3.1545289407131128e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1507,
      "tokens_per_second_per_gpu": 18948.1,
      "total_tokens": 136247217
    },
    {
      "epoch": 0.30125355840783097,
      "grad_norm": 0.68359375,
      "learning_rate": 3.142485027598795e-06,
      "loss": 0.4585,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1508,
      "tokens_per_second_per_gpu": 20277.99,
      "total_tokens": 136347251
    },
    {
      "epoch": 0.3014533286720272,
      "grad_norm": 0.765625,
      "learning_rate": 3.1304598626685544e-06,
      "loss": 0.3855,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1509,
      "tokens_per_second_per_gpu": 18913.61,
      "total_tokens": 136439897
    },
    {
      "epoch": 0.30165309893622333,
      "grad_norm": 0.71875,
      "learning_rate": 3.118453478798743e-06,
      "loss": 0.3774,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1510,
      "tokens_per_second_per_gpu": 19360.82,
      "total_tokens": 136531630
    },
    {
      "epoch": 0.30185286920041954,
      "grad_norm": 0.72265625,
      "learning_rate": 3.1064659088143424e-06,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1511,
      "tokens_per_second_per_gpu": 17489.87,
      "total_tokens": 136616904
    },
    {
      "epoch": 0.3020526394646157,
      "grad_norm": 0.6796875,
      "learning_rate": 3.0944971854889193e-06,
      "loss": 0.3906,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1512,
      "tokens_per_second_per_gpu": 18873.03,
      "total_tokens": 136708975
    },
    {
      "epoch": 0.30225240972881184,
      "grad_norm": 0.6875,
      "learning_rate": 3.0825473415445073e-06,
      "loss": 0.3848,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1513,
      "tokens_per_second_per_gpu": 18507.27,
      "total_tokens": 136799015
    },
    {
      "epoch": 0.30245217999300805,
      "grad_norm": 0.703125,
      "learning_rate": 3.070616409651519e-06,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1514,
      "tokens_per_second_per_gpu": 19145.43,
      "total_tokens": 136891726
    },
    {
      "epoch": 0.3026519502572042,
      "grad_norm": 0.703125,
      "learning_rate": 3.0587044224286743e-06,
      "loss": 0.3996,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1515,
      "tokens_per_second_per_gpu": 19188.98,
      "total_tokens": 136984179
    },
    {
      "epoch": 0.3028517205214004,
      "grad_norm": 0.66796875,
      "learning_rate": 3.0468114124428806e-06,
      "loss": 0.3823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1516,
      "tokens_per_second_per_gpu": 19160.09,
      "total_tokens": 137077706
    },
    {
      "epoch": 0.30305149078559657,
      "grad_norm": 0.6875,
      "learning_rate": 3.034937412209178e-06,
      "loss": 0.3972,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1517,
      "tokens_per_second_per_gpu": 19402.11,
      "total_tokens": 137170732
    },
    {
      "epoch": 0.3032512610497927,
      "grad_norm": 0.6953125,
      "learning_rate": 3.0230824541906332e-06,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1518,
      "tokens_per_second_per_gpu": 17984.95,
      "total_tokens": 137259145
    },
    {
      "epoch": 0.30345103131398893,
      "grad_norm": 0.69140625,
      "learning_rate": 3.011246570798242e-06,
      "loss": 0.3822,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1519,
      "tokens_per_second_per_gpu": 18756.1,
      "total_tokens": 137349634
    },
    {
      "epoch": 0.3036508015781851,
      "grad_norm": 0.67578125,
      "learning_rate": 2.999429794390869e-06,
      "loss": 0.407,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1520,
      "tokens_per_second_per_gpu": 19076.15,
      "total_tokens": 137442677
    },
    {
      "epoch": 0.30385057184238123,
      "grad_norm": 0.68359375,
      "learning_rate": 2.9876321572751143e-06,
      "loss": 0.4026,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1521,
      "tokens_per_second_per_gpu": 19973.37,
      "total_tokens": 137539907
    },
    {
      "epoch": 0.30405034210657744,
      "grad_norm": 0.6640625,
      "learning_rate": 2.975853691705276e-06,
      "loss": 0.3836,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1522,
      "tokens_per_second_per_gpu": 19210.9,
      "total_tokens": 137634561
    },
    {
      "epoch": 0.3042501123707736,
      "grad_norm": 0.6875,
      "learning_rate": 2.9640944298832306e-06,
      "loss": 0.4125,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1523,
      "tokens_per_second_per_gpu": 19510.26,
      "total_tokens": 137729399
    },
    {
      "epoch": 0.3044498826349698,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9523544039583498e-06,
      "loss": 0.3915,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1524,
      "tokens_per_second_per_gpu": 18728.47,
      "total_tokens": 137820836
    },
    {
      "epoch": 0.30464965289916596,
      "grad_norm": 0.65625,
      "learning_rate": 2.9406336460274144e-06,
      "loss": 0.3984,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1525,
      "tokens_per_second_per_gpu": 21072.21,
      "total_tokens": 137921558
    },
    {
      "epoch": 0.3048494231633621,
      "grad_norm": 0.671875,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.3887,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1526,
      "tokens_per_second_per_gpu": 19246.88,
      "total_tokens": 138014984
    },
    {
      "epoch": 0.3050491934275583,
      "grad_norm": 0.69921875,
      "learning_rate": 2.9172500622710263e-06,
      "loss": 0.399,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1527,
      "tokens_per_second_per_gpu": 18430.83,
      "total_tokens": 138103759
    },
    {
      "epoch": 0.30524896369175447,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9055873003754066e-06,
      "loss": 0.3801,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1528,
      "tokens_per_second_per_gpu": 19260.75,
      "total_tokens": 138196449
    },
    {
      "epoch": 0.3054487339559507,
      "grad_norm": 0.71484375,
      "learning_rate": 2.8939439343332086e-06,
      "loss": 0.3902,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1529,
      "tokens_per_second_per_gpu": 19601.42,
      "total_tokens": 138291022
    },
    {
      "epoch": 0.30564850422014683,
      "grad_norm": 0.67578125,
      "learning_rate": 2.882319995976949e-06,
      "loss": 0.3692,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1530,
      "tokens_per_second_per_gpu": 18268.63,
      "total_tokens": 138381719
    },
    {
      "epoch": 0.305848274484343,
      "grad_norm": 0.74609375,
      "learning_rate": 2.8707155170860303e-06,
      "loss": 0.3575,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1531,
      "tokens_per_second_per_gpu": 16508.19,
      "total_tokens": 138462581
    },
    {
      "epoch": 0.3060480447485392,
      "grad_norm": 0.73046875,
      "learning_rate": 2.8591305293866557e-06,
      "loss": 0.3737,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1532,
      "tokens_per_second_per_gpu": 17865.94,
      "total_tokens": 138548928
    },
    {
      "epoch": 0.30624781501273535,
      "grad_norm": 0.6953125,
      "learning_rate": 2.847565064551747e-06,
      "loss": 0.4348,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1533,
      "tokens_per_second_per_gpu": 19426.46,
      "total_tokens": 138643802
    },
    {
      "epoch": 0.30644758527693156,
      "grad_norm": 0.66015625,
      "learning_rate": 2.836019154200831e-06,
      "loss": 0.4136,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1534,
      "tokens_per_second_per_gpu": 20288.14,
      "total_tokens": 138742019
    },
    {
      "epoch": 0.3066473555411277,
      "grad_norm": 0.71875,
      "learning_rate": 2.824492829899994e-06,
      "loss": 0.4025,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1535,
      "tokens_per_second_per_gpu": 18276.13,
      "total_tokens": 138831467
    },
    {
      "epoch": 0.30684712580532386,
      "grad_norm": 0.7421875,
      "learning_rate": 2.812986123161762e-06,
      "loss": 0.3624,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1536,
      "tokens_per_second_per_gpu": 16720.45,
      "total_tokens": 138913526
    },
    {
      "epoch": 0.30704689606952007,
      "grad_norm": 0.73828125,
      "learning_rate": 2.8014990654450325e-06,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1537,
      "tokens_per_second_per_gpu": 20445.83,
      "total_tokens": 139012082
    },
    {
      "epoch": 0.3072466663337162,
      "grad_norm": 0.6875,
      "learning_rate": 2.7900316881549915e-06,
      "loss": 0.4162,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1538,
      "tokens_per_second_per_gpu": 19879.27,
      "total_tokens": 139107725
    },
    {
      "epoch": 0.3074464365979124,
      "grad_norm": 0.7109375,
      "learning_rate": 2.778584022642996e-06,
      "loss": 0.3922,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1539,
      "tokens_per_second_per_gpu": 18367.97,
      "total_tokens": 139196701
    },
    {
      "epoch": 0.3076462068621086,
      "grad_norm": 0.68359375,
      "learning_rate": 2.7671561002065362e-06,
      "loss": 0.3905,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1540,
      "tokens_per_second_per_gpu": 18823.51,
      "total_tokens": 139288867
    },
    {
      "epoch": 0.30784597712630474,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7557479520891104e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1541,
      "tokens_per_second_per_gpu": 18908.75,
      "total_tokens": 139380776
    },
    {
      "epoch": 0.30804574739050095,
      "grad_norm": 0.6796875,
      "learning_rate": 2.744359609480163e-06,
      "loss": 0.4171,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1542,
      "tokens_per_second_per_gpu": 20557.38,
      "total_tokens": 139479162
    },
    {
      "epoch": 0.3082455176546971,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7329911035149934e-06,
      "loss": 0.3876,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1543,
      "tokens_per_second_per_gpu": 19160.03,
      "total_tokens": 139572435
    },
    {
      "epoch": 0.30844528791889325,
      "grad_norm": 0.69921875,
      "learning_rate": 2.7216424652746497e-06,
      "loss": 0.3858,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1544,
      "tokens_per_second_per_gpu": 20244.91,
      "total_tokens": 139670265
    },
    {
      "epoch": 0.30864505818308946,
      "grad_norm": 0.70703125,
      "learning_rate": 2.7103137257858867e-06,
      "loss": 0.3778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1545,
      "tokens_per_second_per_gpu": 17473.45,
      "total_tokens": 139754875
    },
    {
      "epoch": 0.3088448284472856,
      "grad_norm": 0.703125,
      "learning_rate": 2.6990049160210386e-06,
      "loss": 0.3903,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1546,
      "tokens_per_second_per_gpu": 18139.58,
      "total_tokens": 139845250
    },
    {
      "epoch": 0.3090445987114818,
      "grad_norm": 0.67578125,
      "learning_rate": 2.687716066897964e-06,
      "loss": 0.3964,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1547,
      "tokens_per_second_per_gpu": 20999.02,
      "total_tokens": 139945371
    },
    {
      "epoch": 0.309244368975678,
      "grad_norm": 0.703125,
      "learning_rate": 2.6764472092799443e-06,
      "loss": 0.4041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1548,
      "tokens_per_second_per_gpu": 19340.37,
      "total_tokens": 140038653
    },
    {
      "epoch": 0.3094441392398741,
      "grad_norm": 0.72265625,
      "learning_rate": 2.6651983739756026e-06,
      "loss": 0.4164,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1549,
      "tokens_per_second_per_gpu": 18111.62,
      "total_tokens": 140127203
    },
    {
      "epoch": 0.30964390950407034,
      "grad_norm": 0.72265625,
      "learning_rate": 2.65396959173883e-06,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1550,
      "tokens_per_second_per_gpu": 18267.73,
      "total_tokens": 140215425
    },
    {
      "epoch": 0.3098436797682665,
      "grad_norm": 0.703125,
      "learning_rate": 2.642760893268684e-06,
      "loss": 0.405,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1551,
      "tokens_per_second_per_gpu": 19054.35,
      "total_tokens": 140309155
    },
    {
      "epoch": 0.31004345003246264,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6315723092093228e-06,
      "loss": 0.3785,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1552,
      "tokens_per_second_per_gpu": 18335.88,
      "total_tokens": 140398574
    },
    {
      "epoch": 0.31024322029665885,
      "grad_norm": 0.671875,
      "learning_rate": 2.6204038701499056e-06,
      "loss": 0.3578,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1553,
      "tokens_per_second_per_gpu": 18526.48,
      "total_tokens": 140488511
    },
    {
      "epoch": 0.310442990560855,
      "grad_norm": 0.69921875,
      "learning_rate": 2.6092556066245155e-06,
      "loss": 0.4035,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1554,
      "tokens_per_second_per_gpu": 18539.61,
      "total_tokens": 140578598
    },
    {
      "epoch": 0.3106427608250512,
      "grad_norm": 0.703125,
      "learning_rate": 2.5981275491120837e-06,
      "loss": 0.3912,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1555,
      "tokens_per_second_per_gpu": 18309.56,
      "total_tokens": 140666812
    },
    {
      "epoch": 0.31084253108924736,
      "grad_norm": 0.703125,
      "learning_rate": 2.587019728036292e-06,
      "loss": 0.3897,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1556,
      "tokens_per_second_per_gpu": 18125.62,
      "total_tokens": 140754359
    },
    {
      "epoch": 0.3110423013534435,
      "grad_norm": 0.703125,
      "learning_rate": 2.5759321737655017e-06,
      "loss": 0.3951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1557,
      "tokens_per_second_per_gpu": 17888.97,
      "total_tokens": 140842168
    },
    {
      "epoch": 0.3112420716176397,
      "grad_norm": 0.70703125,
      "learning_rate": 2.5648649166126606e-06,
      "loss": 0.414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1558,
      "tokens_per_second_per_gpu": 20266.24,
      "total_tokens": 140938393
    },
    {
      "epoch": 0.3114418418818359,
      "grad_norm": 0.70703125,
      "learning_rate": 2.553817986835225e-06,
      "loss": 0.4189,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1559,
      "tokens_per_second_per_gpu": 19366.71,
      "total_tokens": 141031986
    },
    {
      "epoch": 0.3116416121460321,
      "grad_norm": 0.71875,
      "learning_rate": 2.542791414635085e-06,
      "loss": 0.4256,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1560,
      "tokens_per_second_per_gpu": 18493.21,
      "total_tokens": 141123011
    },
    {
      "epoch": 0.31184138241022824,
      "grad_norm": 0.68359375,
      "learning_rate": 2.5317852301584642e-06,
      "loss": 0.3527,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1561,
      "tokens_per_second_per_gpu": 16779.27,
      "total_tokens": 141205346
    },
    {
      "epoch": 0.3120411526744244,
      "grad_norm": 0.7265625,
      "learning_rate": 2.5207994634958475e-06,
      "loss": 0.3782,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1562,
      "tokens_per_second_per_gpu": 17089.51,
      "total_tokens": 141288880
    },
    {
      "epoch": 0.3122409229386206,
      "grad_norm": 0.703125,
      "learning_rate": 2.5098341446819097e-06,
      "loss": 0.4041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1563,
      "tokens_per_second_per_gpu": 19572.67,
      "total_tokens": 141382462
    },
    {
      "epoch": 0.31244069320281675,
      "grad_norm": 0.7109375,
      "learning_rate": 2.4988893036954045e-06,
      "loss": 0.43,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1564,
      "tokens_per_second_per_gpu": 18559.75,
      "total_tokens": 141474840
    },
    {
      "epoch": 0.3126404634670129,
      "grad_norm": 0.78515625,
      "learning_rate": 2.487964970459118e-06,
      "loss": 0.3931,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1565,
      "tokens_per_second_per_gpu": 19481.73,
      "total_tokens": 141569422
    },
    {
      "epoch": 0.3128402337312091,
      "grad_norm": 0.7265625,
      "learning_rate": 2.4770611748397556e-06,
      "loss": 0.4108,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1566,
      "tokens_per_second_per_gpu": 19141.6,
      "total_tokens": 141661891
    },
    {
      "epoch": 0.31304000399540527,
      "grad_norm": 0.7109375,
      "learning_rate": 2.466177946647874e-06,
      "loss": 0.4082,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1567,
      "tokens_per_second_per_gpu": 18225.69,
      "total_tokens": 141750440
    },
    {
      "epoch": 0.3132397742596015,
      "grad_norm": 0.703125,
      "learning_rate": 2.455315315637813e-06,
      "loss": 0.3761,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1568,
      "tokens_per_second_per_gpu": 17953.77,
      "total_tokens": 141838209
    },
    {
      "epoch": 0.31343954452379763,
      "grad_norm": 0.6875,
      "learning_rate": 2.4444733115075823e-06,
      "loss": 0.3804,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1569,
      "tokens_per_second_per_gpu": 18650.74,
      "total_tokens": 141926856
    },
    {
      "epoch": 0.3136393147879938,
      "grad_norm": 0.6875,
      "learning_rate": 2.4336519638988134e-06,
      "loss": 0.3824,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1570,
      "tokens_per_second_per_gpu": 18437.98,
      "total_tokens": 142017624
    },
    {
      "epoch": 0.31383908505219,
      "grad_norm": 0.703125,
      "learning_rate": 2.422851302396655e-06,
      "loss": 0.4179,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1571,
      "tokens_per_second_per_gpu": 18764.53,
      "total_tokens": 142111047
    },
    {
      "epoch": 0.31403885531638615,
      "grad_norm": 0.69921875,
      "learning_rate": 2.4120713565297005e-06,
      "loss": 0.3778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1572,
      "tokens_per_second_per_gpu": 18021.75,
      "total_tokens": 142199954
    },
    {
      "epoch": 0.31423862558058235,
      "grad_norm": 0.7265625,
      "learning_rate": 2.401312155769916e-06,
      "loss": 0.4151,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1573,
      "tokens_per_second_per_gpu": 18289.06,
      "total_tokens": 142290017
    },
    {
      "epoch": 0.3144383958447785,
      "grad_norm": 0.69140625,
      "learning_rate": 2.3905737295325393e-06,
      "loss": 0.4106,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1574,
      "tokens_per_second_per_gpu": 19549.09,
      "total_tokens": 142385213
    },
    {
      "epoch": 0.31463816610897466,
      "grad_norm": 0.70703125,
      "learning_rate": 2.379856107176024e-06,
      "loss": 0.373,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1575,
      "tokens_per_second_per_gpu": 18253.4,
      "total_tokens": 142474813
    },
    {
      "epoch": 0.31483793637317087,
      "grad_norm": 0.75,
      "learning_rate": 2.369159318001937e-06,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1576,
      "tokens_per_second_per_gpu": 17093.29,
      "total_tokens": 142559100
    },
    {
      "epoch": 0.315037706637367,
      "grad_norm": 0.72265625,
      "learning_rate": 2.3584833912548887e-06,
      "loss": 0.3919,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1577,
      "tokens_per_second_per_gpu": 17893.79,
      "total_tokens": 142647568
    },
    {
      "epoch": 0.31523747690156323,
      "grad_norm": 0.7265625,
      "learning_rate": 2.347828356122461e-06,
      "loss": 0.4104,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1578,
      "tokens_per_second_per_gpu": 18264.98,
      "total_tokens": 142737277
    },
    {
      "epoch": 0.3154372471657594,
      "grad_norm": 0.6953125,
      "learning_rate": 2.3371942417351077e-06,
      "loss": 0.3615,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1579,
      "tokens_per_second_per_gpu": 17502.97,
      "total_tokens": 142823193
    },
    {
      "epoch": 0.31563701742995554,
      "grad_norm": 0.69140625,
      "learning_rate": 2.326581077166097e-06,
      "loss": 0.38,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1580,
      "tokens_per_second_per_gpu": 18222.04,
      "total_tokens": 142912448
    },
    {
      "epoch": 0.31583678769415174,
      "grad_norm": 0.72265625,
      "learning_rate": 2.315988891431412e-06,
      "loss": 0.3846,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1581,
      "tokens_per_second_per_gpu": 16963.97,
      "total_tokens": 142995277
    },
    {
      "epoch": 0.3160365579583479,
      "grad_norm": 0.703125,
      "learning_rate": 2.3054177134896818e-06,
      "loss": 0.3872,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1582,
      "tokens_per_second_per_gpu": 18669.98,
      "total_tokens": 143085204
    },
    {
      "epoch": 0.31623632822254405,
      "grad_norm": 0.69140625,
      "learning_rate": 2.2948675722421086e-06,
      "loss": 0.414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1583,
      "tokens_per_second_per_gpu": 19789.18,
      "total_tokens": 143180597
    },
    {
      "epoch": 0.31643609848674026,
      "grad_norm": 0.6875,
      "learning_rate": 2.284338496532371e-06,
      "loss": 0.4315,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1584,
      "tokens_per_second_per_gpu": 20248.4,
      "total_tokens": 143278575
    },
    {
      "epoch": 0.3166358687509364,
      "grad_norm": 0.6875,
      "learning_rate": 2.2738305151465646e-06,
      "loss": 0.4001,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1585,
      "tokens_per_second_per_gpu": 20898.31,
      "total_tokens": 143378114
    },
    {
      "epoch": 0.3168356390151326,
      "grad_norm": 0.6875,
      "learning_rate": 2.263343656813107e-06,
      "loss": 0.3872,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1586,
      "tokens_per_second_per_gpu": 18112.86,
      "total_tokens": 143467141
    },
    {
      "epoch": 0.3170354092793288,
      "grad_norm": 0.70703125,
      "learning_rate": 2.2528779502026652e-06,
      "loss": 0.3797,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1587,
      "tokens_per_second_per_gpu": 17228.63,
      "total_tokens": 143552497
    },
    {
      "epoch": 0.3172351795435249,
      "grad_norm": 0.71875,
      "learning_rate": 2.2424334239280878e-06,
      "loss": 0.379,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1588,
      "tokens_per_second_per_gpu": 16746.53,
      "total_tokens": 143633858
    },
    {
      "epoch": 0.31743494980772113,
      "grad_norm": 0.68359375,
      "learning_rate": 2.2320101065443055e-06,
      "loss": 0.3632,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1589,
      "tokens_per_second_per_gpu": 18268.92,
      "total_tokens": 143723254
    },
    {
      "epoch": 0.3176347200719173,
      "grad_norm": 0.6875,
      "learning_rate": 2.221608026548271e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1590,
      "tokens_per_second_per_gpu": 17943.36,
      "total_tokens": 143810674
    },
    {
      "epoch": 0.3178344903361135,
      "grad_norm": 0.68359375,
      "learning_rate": 2.211227212378877e-06,
      "loss": 0.3986,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1591,
      "tokens_per_second_per_gpu": 19253.62,
      "total_tokens": 143904474
    },
    {
      "epoch": 0.31803426060030965,
      "grad_norm": 0.6953125,
      "learning_rate": 2.200867692416868e-06,
      "loss": 0.3798,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1592,
      "tokens_per_second_per_gpu": 19630.64,
      "total_tokens": 143998823
    },
    {
      "epoch": 0.3182340308645058,
      "grad_norm": 0.703125,
      "learning_rate": 2.190529494984782e-06,
      "loss": 0.4111,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1593,
      "tokens_per_second_per_gpu": 18908.0,
      "total_tokens": 144089248
    },
    {
      "epoch": 0.318433801128702,
      "grad_norm": 0.72265625,
      "learning_rate": 2.1802126483468524e-06,
      "loss": 0.3626,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1594,
      "tokens_per_second_per_gpu": 17124.94,
      "total_tokens": 144173784
    },
    {
      "epoch": 0.31863357139289816,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1699171807089414e-06,
      "loss": 0.3901,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1595,
      "tokens_per_second_per_gpu": 18495.25,
      "total_tokens": 144264180
    },
    {
      "epoch": 0.3188333416570943,
      "grad_norm": 0.6875,
      "learning_rate": 2.1596431202184707e-06,
      "loss": 0.373,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1596,
      "tokens_per_second_per_gpu": 18304.96,
      "total_tokens": 144352792
    },
    {
      "epoch": 0.3190331119212905,
      "grad_norm": 0.7109375,
      "learning_rate": 2.149390494964323e-06,
      "loss": 0.3805,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1597,
      "tokens_per_second_per_gpu": 17258.31,
      "total_tokens": 144438114
    },
    {
      "epoch": 0.3192328821854867,
      "grad_norm": 0.703125,
      "learning_rate": 2.13915933297679e-06,
      "loss": 0.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1598,
      "tokens_per_second_per_gpu": 17860.1,
      "total_tokens": 144525865
    },
    {
      "epoch": 0.3194326524496829,
      "grad_norm": 0.7109375,
      "learning_rate": 2.1289496622274754e-06,
      "loss": 0.4213,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1599,
      "tokens_per_second_per_gpu": 20151.82,
      "total_tokens": 144623582
    },
    {
      "epoch": 0.31963242271387904,
      "grad_norm": 0.6953125,
      "learning_rate": 2.118761510629226e-06,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1600,
      "tokens_per_second_per_gpu": 18334.73,
      "total_tokens": 144715347
    },
    {
      "epoch": 0.3198321929780752,
      "grad_norm": 0.6953125,
      "learning_rate": 2.1085949060360654e-06,
      "loss": 0.3938,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1601,
      "tokens_per_second_per_gpu": 17869.12,
      "total_tokens": 144803581
    },
    {
      "epoch": 0.3200319632422714,
      "grad_norm": 0.70703125,
      "learning_rate": 2.098449876243096e-06,
      "loss": 0.3999,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1602,
      "tokens_per_second_per_gpu": 18626.82,
      "total_tokens": 144891945
    },
    {
      "epoch": 0.32023173350646755,
      "grad_norm": 0.69921875,
      "learning_rate": 2.0883264489864476e-06,
      "loss": 0.3709,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1603,
      "tokens_per_second_per_gpu": 18331.31,
      "total_tokens": 144982137
    },
    {
      "epoch": 0.32043150377066376,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0782246519431814e-06,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1604,
      "tokens_per_second_per_gpu": 18995.37,
      "total_tokens": 145073765
    },
    {
      "epoch": 0.3206312740348599,
      "grad_norm": 0.6953125,
      "learning_rate": 2.0681445127312218e-06,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1605,
      "tokens_per_second_per_gpu": 19947.68,
      "total_tokens": 145169628
    },
    {
      "epoch": 0.32083104429905607,
      "grad_norm": 0.75,
      "learning_rate": 2.0580860589092897e-06,
      "loss": 0.3934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1606,
      "tokens_per_second_per_gpu": 18072.75,
      "total_tokens": 145258519
    },
    {
      "epoch": 0.3210308145632523,
      "grad_norm": 0.6796875,
      "learning_rate": 2.048049317976809e-06,
      "loss": 0.3665,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1607,
      "tokens_per_second_per_gpu": 18500.97,
      "total_tokens": 145347839
    },
    {
      "epoch": 0.32123058482744843,
      "grad_norm": 0.6953125,
      "learning_rate": 2.0380343173738516e-06,
      "loss": 0.3923,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1608,
      "tokens_per_second_per_gpu": 17981.28,
      "total_tokens": 145435172
    },
    {
      "epoch": 0.3214303550916446,
      "grad_norm": 0.71875,
      "learning_rate": 2.0280410844810426e-06,
      "loss": 0.3861,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1609,
      "tokens_per_second_per_gpu": 18268.66,
      "total_tokens": 145522918
    },
    {
      "epoch": 0.3216301253558408,
      "grad_norm": 0.67578125,
      "learning_rate": 2.0180696466194983e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1610,
      "tokens_per_second_per_gpu": 19167.41,
      "total_tokens": 145616842
    },
    {
      "epoch": 0.32182989562003694,
      "grad_norm": 0.66796875,
      "learning_rate": 2.008120031050753e-06,
      "loss": 0.3908,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1611,
      "tokens_per_second_per_gpu": 19826.63,
      "total_tokens": 145713609
    },
    {
      "epoch": 0.32202966588423315,
      "grad_norm": 0.6953125,
      "learning_rate": 1.9981922649766714e-06,
      "loss": 0.3891,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1612,
      "tokens_per_second_per_gpu": 18328.84,
      "total_tokens": 145803028
    },
    {
      "epoch": 0.3222294361484293,
      "grad_norm": 0.6953125,
      "learning_rate": 1.988286375539391e-06,
      "loss": 0.3699,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1613,
      "tokens_per_second_per_gpu": 17380.07,
      "total_tokens": 145887152
    },
    {
      "epoch": 0.32242920641262546,
      "grad_norm": 0.703125,
      "learning_rate": 1.9784023898212345e-06,
      "loss": 0.3852,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1614,
      "tokens_per_second_per_gpu": 17520.97,
      "total_tokens": 145973979
    },
    {
      "epoch": 0.32262897667682167,
      "grad_norm": 0.67578125,
      "learning_rate": 1.9685403348446374e-06,
      "loss": 0.3852,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1615,
      "tokens_per_second_per_gpu": 19894.5,
      "total_tokens": 146068539
    },
    {
      "epoch": 0.3228287469410178,
      "grad_norm": 0.69140625,
      "learning_rate": 1.9587002375720864e-06,
      "loss": 0.3983,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1616,
      "tokens_per_second_per_gpu": 18632.93,
      "total_tokens": 146158554
    },
    {
      "epoch": 0.323028517205214,
      "grad_norm": 0.68359375,
      "learning_rate": 1.9488821249060297e-06,
      "loss": 0.4049,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1617,
      "tokens_per_second_per_gpu": 20180.48,
      "total_tokens": 146257392
    },
    {
      "epoch": 0.3232282874694102,
      "grad_norm": 0.68359375,
      "learning_rate": 1.9390860236888097e-06,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1618,
      "tokens_per_second_per_gpu": 19450.69,
      "total_tokens": 146351651
    },
    {
      "epoch": 0.32342805773360633,
      "grad_norm": 0.72265625,
      "learning_rate": 1.929311960702599e-06,
      "loss": 0.3538,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1619,
      "tokens_per_second_per_gpu": 16636.56,
      "total_tokens": 146432535
    },
    {
      "epoch": 0.32362782799780254,
      "grad_norm": 0.6875,
      "learning_rate": 1.919559962669305e-06,
      "loss": 0.383,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1620,
      "tokens_per_second_per_gpu": 18936.58,
      "total_tokens": 146523772
    },
    {
      "epoch": 0.3238275982619987,
      "grad_norm": 0.67578125,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.3633,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1621,
      "tokens_per_second_per_gpu": 18781.49,
      "total_tokens": 146613922
    },
    {
      "epoch": 0.3240273685261949,
      "grad_norm": 0.6796875,
      "learning_rate": 1.9001222680474518e-06,
      "loss": 0.3594,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1622,
      "tokens_per_second_per_gpu": 18358.0,
      "total_tokens": 146702596
    },
    {
      "epoch": 0.32422713879039106,
      "grad_norm": 0.66015625,
      "learning_rate": 1.8904366246007998e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1623,
      "tokens_per_second_per_gpu": 19695.55,
      "total_tokens": 146797094
    },
    {
      "epoch": 0.3244269090545872,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8807731523907556e-06,
      "loss": 0.3721,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1624,
      "tokens_per_second_per_gpu": 19420.69,
      "total_tokens": 146890646
    },
    {
      "epoch": 0.3246266793187834,
      "grad_norm": 0.6875,
      "learning_rate": 1.8711318778368792e-06,
      "loss": 0.368,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1625,
      "tokens_per_second_per_gpu": 17930.38,
      "total_tokens": 146978138
    },
    {
      "epoch": 0.32482644958297957,
      "grad_norm": 0.7265625,
      "learning_rate": 1.861512827298051e-06,
      "loss": 0.372,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1626,
      "tokens_per_second_per_gpu": 16742.25,
      "total_tokens": 147061039
    },
    {
      "epoch": 0.3250262198471757,
      "grad_norm": 0.703125,
      "learning_rate": 1.8519160270723858e-06,
      "loss": 0.3815,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1627,
      "tokens_per_second_per_gpu": 18219.94,
      "total_tokens": 147150304
    },
    {
      "epoch": 0.32522599011137193,
      "grad_norm": 0.66015625,
      "learning_rate": 1.842341503397166e-06,
      "loss": 0.3899,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1628,
      "tokens_per_second_per_gpu": 19011.34,
      "total_tokens": 147243511
    },
    {
      "epoch": 0.3254257603755681,
      "grad_norm": 0.68359375,
      "learning_rate": 1.8327892824487792e-06,
      "loss": 0.3889,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1629,
      "tokens_per_second_per_gpu": 19984.49,
      "total_tokens": 147340316
    },
    {
      "epoch": 0.3256255306397643,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8232593903426266e-06,
      "loss": 0.3646,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1630,
      "tokens_per_second_per_gpu": 17001.06,
      "total_tokens": 147422885
    },
    {
      "epoch": 0.32582530090396045,
      "grad_norm": 0.703125,
      "learning_rate": 1.8137518531330768e-06,
      "loss": 0.4033,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1631,
      "tokens_per_second_per_gpu": 17677.62,
      "total_tokens": 147511035
    },
    {
      "epoch": 0.3260250711681566,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8042666968133682e-06,
      "loss": 0.4136,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1632,
      "tokens_per_second_per_gpu": 19583.75,
      "total_tokens": 147607663
    },
    {
      "epoch": 0.3262248414323528,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7948039473155553e-06,
      "loss": 0.3904,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1633,
      "tokens_per_second_per_gpu": 18411.9,
      "total_tokens": 147695949
    },
    {
      "epoch": 0.32642461169654896,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7853636305104382e-06,
      "loss": 0.4158,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1634,
      "tokens_per_second_per_gpu": 19631.79,
      "total_tokens": 147791520
    },
    {
      "epoch": 0.32662438196074517,
      "grad_norm": 0.69140625,
      "learning_rate": 1.775945772207477e-06,
      "loss": 0.3898,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1635,
      "tokens_per_second_per_gpu": 19083.48,
      "total_tokens": 147884487
    },
    {
      "epoch": 0.3268241522249413,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7665503981547428e-06,
      "loss": 0.3819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1636,
      "tokens_per_second_per_gpu": 16528.8,
      "total_tokens": 147964945
    },
    {
      "epoch": 0.3270239224891375,
      "grad_norm": 0.6796875,
      "learning_rate": 1.7571775340388275e-06,
      "loss": 0.3964,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1637,
      "tokens_per_second_per_gpu": 19215.74,
      "total_tokens": 148057986
    },
    {
      "epoch": 0.3272236927533337,
      "grad_norm": 0.71484375,
      "learning_rate": 1.747827205484779e-06,
      "loss": 0.368,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1638,
      "tokens_per_second_per_gpu": 17707.63,
      "total_tokens": 148142960
    },
    {
      "epoch": 0.32742346301752984,
      "grad_norm": 0.73046875,
      "learning_rate": 1.738499438056045e-06,
      "loss": 0.3777,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1639,
      "tokens_per_second_per_gpu": 19589.95,
      "total_tokens": 148237313
    },
    {
      "epoch": 0.327623233281726,
      "grad_norm": 0.703125,
      "learning_rate": 1.7291942572543806e-06,
      "loss": 0.3966,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1640,
      "tokens_per_second_per_gpu": 18921.21,
      "total_tokens": 148329331
    },
    {
      "epoch": 0.3278230035459222,
      "grad_norm": 0.6875,
      "learning_rate": 1.7199116885197996e-06,
      "loss": 0.3849,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1641,
      "tokens_per_second_per_gpu": 17975.13,
      "total_tokens": 148416877
    },
    {
      "epoch": 0.32802277381011835,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7106517572304883e-06,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1642,
      "tokens_per_second_per_gpu": 20087.22,
      "total_tokens": 148511698
    },
    {
      "epoch": 0.32822254407431456,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7014144887027406e-06,
      "loss": 0.392,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1643,
      "tokens_per_second_per_gpu": 17811.09,
      "total_tokens": 148597916
    },
    {
      "epoch": 0.3284223143385107,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6921999081909036e-06,
      "loss": 0.3905,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1644,
      "tokens_per_second_per_gpu": 18708.08,
      "total_tokens": 148689442
    },
    {
      "epoch": 0.32862208460270687,
      "grad_norm": 0.66796875,
      "learning_rate": 1.6830080408872852e-06,
      "loss": 0.3806,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1645,
      "tokens_per_second_per_gpu": 18524.28,
      "total_tokens": 148779943
    },
    {
      "epoch": 0.3288218548669031,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6738389119220966e-06,
      "loss": 0.363,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1646,
      "tokens_per_second_per_gpu": 16978.68,
      "total_tokens": 148863738
    },
    {
      "epoch": 0.3290216251310992,
      "grad_norm": 0.796875,
      "learning_rate": 1.6646925463633924e-06,
      "loss": 0.3978,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1647,
      "tokens_per_second_per_gpu": 17607.11,
      "total_tokens": 148948698
    },
    {
      "epoch": 0.32922139539529544,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6555689692169796e-06,
      "loss": 0.3785,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1648,
      "tokens_per_second_per_gpu": 19360.72,
      "total_tokens": 149041397
    },
    {
      "epoch": 0.3294211656594916,
      "grad_norm": 0.69140625,
      "learning_rate": 1.646468205426377e-06,
      "loss": 0.3959,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1649,
      "tokens_per_second_per_gpu": 17589.48,
      "total_tokens": 149128023
    },
    {
      "epoch": 0.32962093592368774,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6373902798727226e-06,
      "loss": 0.3905,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1650,
      "tokens_per_second_per_gpu": 18586.35,
      "total_tokens": 149217805
    },
    {
      "epoch": 0.32982070618788395,
      "grad_norm": 0.71875,
      "learning_rate": 1.6283352173747148e-06,
      "loss": 0.3865,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1651,
      "tokens_per_second_per_gpu": 17859.64,
      "total_tokens": 149304184
    },
    {
      "epoch": 0.3300204764520801,
      "grad_norm": 0.6875,
      "learning_rate": 1.6193030426885526e-06,
      "loss": 0.3762,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1652,
      "tokens_per_second_per_gpu": 17911.29,
      "total_tokens": 149390620
    },
    {
      "epoch": 0.33022024671627626,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6102937805078544e-06,
      "loss": 0.4095,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1653,
      "tokens_per_second_per_gpu": 17974.76,
      "total_tokens": 149479060
    },
    {
      "epoch": 0.33042001698047246,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6013074554636022e-06,
      "loss": 0.3992,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1654,
      "tokens_per_second_per_gpu": 18478.31,
      "total_tokens": 149571004
    },
    {
      "epoch": 0.3306197872446686,
      "grad_norm": 0.703125,
      "learning_rate": 1.5923440921240641e-06,
      "loss": 0.4134,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1655,
      "tokens_per_second_per_gpu": 18843.85,
      "total_tokens": 149662833
    },
    {
      "epoch": 0.3308195575088648,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5834037149947291e-06,
      "loss": 0.3707,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1656,
      "tokens_per_second_per_gpu": 18793.67,
      "total_tokens": 149753241
    },
    {
      "epoch": 0.331019327773061,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5744863485182537e-06,
      "loss": 0.4107,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1657,
      "tokens_per_second_per_gpu": 20075.44,
      "total_tokens": 149850899
    },
    {
      "epoch": 0.33121909803725713,
      "grad_norm": 0.6796875,
      "learning_rate": 1.565592017074371e-06,
      "loss": 0.3774,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1658,
      "tokens_per_second_per_gpu": 19497.93,
      "total_tokens": 149944729
    },
    {
      "epoch": 0.33141886830145334,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5567207449798517e-06,
      "loss": 0.375,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1659,
      "tokens_per_second_per_gpu": 17721.8,
      "total_tokens": 150030587
    },
    {
      "epoch": 0.3316186385656495,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5478725564884045e-06,
      "loss": 0.3826,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1660,
      "tokens_per_second_per_gpu": 19535.99,
      "total_tokens": 150123851
    },
    {
      "epoch": 0.3318184088298457,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5390474757906449e-06,
      "loss": 0.4064,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1661,
      "tokens_per_second_per_gpu": 18044.22,
      "total_tokens": 150212524
    },
    {
      "epoch": 0.33201817909404185,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5302455270140083e-06,
      "loss": 0.3854,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1662,
      "tokens_per_second_per_gpu": 18197.76,
      "total_tokens": 150301347
    },
    {
      "epoch": 0.332217949358238,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5214667342226818e-06,
      "loss": 0.3819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1663,
      "tokens_per_second_per_gpu": 19625.1,
      "total_tokens": 150396463
    },
    {
      "epoch": 0.3324177196224342,
      "grad_norm": 0.6953125,
      "learning_rate": 1.512711121417556e-06,
      "loss": 0.3736,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1664,
      "tokens_per_second_per_gpu": 17708.52,
      "total_tokens": 150482430
    },
    {
      "epoch": 0.33261748988663037,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5039787125361327e-06,
      "loss": 0.3712,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1665,
      "tokens_per_second_per_gpu": 17792.81,
      "total_tokens": 150568925
    },
    {
      "epoch": 0.3328172601508266,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4952695314524912e-06,
      "loss": 0.3801,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1666,
      "tokens_per_second_per_gpu": 17591.18,
      "total_tokens": 150654913
    },
    {
      "epoch": 0.33301703041502273,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4865836019771995e-06,
      "loss": 0.4013,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1667,
      "tokens_per_second_per_gpu": 19254.33,
      "total_tokens": 150748758
    },
    {
      "epoch": 0.3332168006792189,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4779209478572542e-06,
      "loss": 0.4,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1668,
      "tokens_per_second_per_gpu": 18785.99,
      "total_tokens": 150840413
    },
    {
      "epoch": 0.3334165709434151,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4692815927760274e-06,
      "loss": 0.4156,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1669,
      "tokens_per_second_per_gpu": 19268.93,
      "total_tokens": 150933925
    },
    {
      "epoch": 0.33361634120761124,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4606655603531762e-06,
      "loss": 0.3911,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1670,
      "tokens_per_second_per_gpu": 19271.79,
      "total_tokens": 151027259
    },
    {
      "epoch": 0.3338161114718074,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4520728741446087e-06,
      "loss": 0.395,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1671,
      "tokens_per_second_per_gpu": 17822.44,
      "total_tokens": 151114502
    },
    {
      "epoch": 0.3340158817360036,
      "grad_norm": 0.6953125,
      "learning_rate": 1.443503557642405e-06,
      "loss": 0.3984,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1672,
      "tokens_per_second_per_gpu": 19658.97,
      "total_tokens": 151208127
    },
    {
      "epoch": 0.33421565200019976,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4349576342747462e-06,
      "loss": 0.3749,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1673,
      "tokens_per_second_per_gpu": 18752.49,
      "total_tokens": 151298861
    },
    {
      "epoch": 0.33441542226439597,
      "grad_norm": 0.6875,
      "learning_rate": 1.4264351274058584e-06,
      "loss": 0.3837,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1674,
      "tokens_per_second_per_gpu": 19209.23,
      "total_tokens": 151391333
    },
    {
      "epoch": 0.3346151925285921,
      "grad_norm": 0.6875,
      "learning_rate": 1.4179360603359504e-06,
      "loss": 0.3964,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1675,
      "tokens_per_second_per_gpu": 18073.46,
      "total_tokens": 151480396
    },
    {
      "epoch": 0.3348149627927883,
      "grad_norm": 0.7734375,
      "learning_rate": 1.409460456301147e-06,
      "loss": 0.3812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1676,
      "tokens_per_second_per_gpu": 17736.34,
      "total_tokens": 151567181
    },
    {
      "epoch": 0.3350147330569845,
      "grad_norm": 0.68359375,
      "learning_rate": 1.4010083384734308e-06,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1677,
      "tokens_per_second_per_gpu": 19971.63,
      "total_tokens": 151663119
    },
    {
      "epoch": 0.33521450332118063,
      "grad_norm": 0.6875,
      "learning_rate": 1.3925797299605649e-06,
      "loss": 0.3579,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1678,
      "tokens_per_second_per_gpu": 17777.51,
      "total_tokens": 151749338
    },
    {
      "epoch": 0.33541427358537684,
      "grad_norm": 0.68359375,
      "learning_rate": 1.384174653806044e-06,
      "loss": 0.4029,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1679,
      "tokens_per_second_per_gpu": 19720.3,
      "total_tokens": 151844990
    },
    {
      "epoch": 0.335614043849573,
      "grad_norm": 0.67578125,
      "learning_rate": 1.3757931329890218e-06,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1680,
      "tokens_per_second_per_gpu": 19147.6,
      "total_tokens": 151939228
    },
    {
      "epoch": 0.33581381411376915,
      "grad_norm": 0.6796875,
      "learning_rate": 1.367435190424261e-06,
      "loss": 0.4015,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1681,
      "tokens_per_second_per_gpu": 19511.21,
      "total_tokens": 152033999
    },
    {
      "epoch": 0.33601358437796536,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3591008489620572e-06,
      "loss": 0.3896,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1682,
      "tokens_per_second_per_gpu": 17518.25,
      "total_tokens": 152119881
    },
    {
      "epoch": 0.3362133546421615,
      "grad_norm": 0.69921875,
      "learning_rate": 1.350790131388181e-06,
      "loss": 0.4066,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1683,
      "tokens_per_second_per_gpu": 19492.49,
      "total_tokens": 152214663
    },
    {
      "epoch": 0.33641312490635766,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3425030604238154e-06,
      "loss": 0.402,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1684,
      "tokens_per_second_per_gpu": 18423.59,
      "total_tokens": 152304764
    },
    {
      "epoch": 0.33661289517055387,
      "grad_norm": 0.6875,
      "learning_rate": 1.3342396587254957e-06,
      "loss": 0.3637,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1685,
      "tokens_per_second_per_gpu": 18990.3,
      "total_tokens": 152397203
    },
    {
      "epoch": 0.33681266543475,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3259999488850473e-06,
      "loss": 0.4035,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1686,
      "tokens_per_second_per_gpu": 18257.77,
      "total_tokens": 152486314
    },
    {
      "epoch": 0.33701243569894623,
      "grad_norm": 0.66796875,
      "learning_rate": 1.317783953429528e-06,
      "loss": 0.3878,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1687,
      "tokens_per_second_per_gpu": 19580.09,
      "total_tokens": 152580882
    },
    {
      "epoch": 0.3372122059631424,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3095916948211452e-06,
      "loss": 0.3874,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1688,
      "tokens_per_second_per_gpu": 18631.6,
      "total_tokens": 152672292
    },
    {
      "epoch": 0.33741197622733854,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3014231954572287e-06,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1689,
      "tokens_per_second_per_gpu": 17134.37,
      "total_tokens": 152757034
    },
    {
      "epoch": 0.33761174649153475,
      "grad_norm": 0.703125,
      "learning_rate": 1.293278477670139e-06,
      "loss": 0.3784,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1690,
      "tokens_per_second_per_gpu": 19050.79,
      "total_tokens": 152848856
    },
    {
      "epoch": 0.3378115167557309,
      "grad_norm": 0.6875,
      "learning_rate": 1.2851575637272262e-06,
      "loss": 0.3472,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1691,
      "tokens_per_second_per_gpu": 17419.65,
      "total_tokens": 152933101
    },
    {
      "epoch": 0.3380112870199271,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2770604758307637e-06,
      "loss": 0.3866,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1692,
      "tokens_per_second_per_gpu": 18183.51,
      "total_tokens": 153021486
    },
    {
      "epoch": 0.33821105728412326,
      "grad_norm": 0.77734375,
      "learning_rate": 1.26898723611787e-06,
      "loss": 0.4122,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1693,
      "tokens_per_second_per_gpu": 19101.09,
      "total_tokens": 153114184
    },
    {
      "epoch": 0.3384108275483194,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2609378666604833e-06,
      "loss": 0.3906,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1694,
      "tokens_per_second_per_gpu": 18186.36,
      "total_tokens": 153202571
    },
    {
      "epoch": 0.3386105978125156,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2529123894652661e-06,
      "loss": 0.3362,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1695,
      "tokens_per_second_per_gpu": 17036.92,
      "total_tokens": 153286443
    },
    {
      "epoch": 0.3388103680767118,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2449108264735721e-06,
      "loss": 0.3708,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1696,
      "tokens_per_second_per_gpu": 18482.26,
      "total_tokens": 153375287
    },
    {
      "epoch": 0.33901013834090793,
      "grad_norm": 0.6796875,
      "learning_rate": 1.2369331995613664e-06,
      "loss": 0.3974,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1697,
      "tokens_per_second_per_gpu": 19223.82,
      "total_tokens": 153468172
    },
    {
      "epoch": 0.33920990860510414,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2289795305391728e-06,
      "loss": 0.3836,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1698,
      "tokens_per_second_per_gpu": 19867.48,
      "total_tokens": 153563523
    },
    {
      "epoch": 0.3394096788693003,
      "grad_norm": 0.6875,
      "learning_rate": 1.2210498411520256e-06,
      "loss": 0.3726,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1699,
      "tokens_per_second_per_gpu": 18136.32,
      "total_tokens": 153654053
    },
    {
      "epoch": 0.3396094491334965,
      "grad_norm": 0.6875,
      "learning_rate": 1.2131441530793863e-06,
      "loss": 0.3874,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1700,
      "tokens_per_second_per_gpu": 18286.83,
      "total_tokens": 153744463
    },
    {
      "epoch": 0.33980921939769265,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2052624879351105e-06,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1701,
      "tokens_per_second_per_gpu": 17692.24,
      "total_tokens": 153831648
    },
    {
      "epoch": 0.3400089896618888,
      "grad_norm": 0.6875,
      "learning_rate": 1.1974048672673655e-06,
      "loss": 0.3821,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1702,
      "tokens_per_second_per_gpu": 19374.94,
      "total_tokens": 153924311
    },
    {
      "epoch": 0.340208759926085,
      "grad_norm": 0.66796875,
      "learning_rate": 1.1895713125585851e-06,
      "loss": 0.4041,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1703,
      "tokens_per_second_per_gpu": 19480.87,
      "total_tokens": 154019640
    },
    {
      "epoch": 0.34040853019028117,
      "grad_norm": 0.6953125,
      "learning_rate": 1.181761845225411e-06,
      "loss": 0.3949,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1704,
      "tokens_per_second_per_gpu": 19874.9,
      "total_tokens": 154113008
    },
    {
      "epoch": 0.3406083004544774,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1739764866186309e-06,
      "loss": 0.3931,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1705,
      "tokens_per_second_per_gpu": 19514.75,
      "total_tokens": 154207148
    },
    {
      "epoch": 0.34080807071867353,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1662152580231145e-06,
      "loss": 0.3558,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1706,
      "tokens_per_second_per_gpu": 18304.75,
      "total_tokens": 154295110
    },
    {
      "epoch": 0.3410078409828697,
      "grad_norm": 0.734375,
      "learning_rate": 1.158478180657766e-06,
      "loss": 0.3928,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1707,
      "tokens_per_second_per_gpu": 16992.1,
      "total_tokens": 154378032
    },
    {
      "epoch": 0.3412076112470659,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1507652756754573e-06,
      "loss": 0.388,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1708,
      "tokens_per_second_per_gpu": 18419.96,
      "total_tokens": 154468893
    },
    {
      "epoch": 0.34140738151126204,
      "grad_norm": 0.73828125,
      "learning_rate": 1.143076564162977e-06,
      "loss": 0.374,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1709,
      "tokens_per_second_per_gpu": 16587.19,
      "total_tokens": 154550189
    },
    {
      "epoch": 0.3416071517754582,
      "grad_norm": 0.703125,
      "learning_rate": 1.135412067140974e-06,
      "loss": 0.4009,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1710,
      "tokens_per_second_per_gpu": 18614.65,
      "total_tokens": 154642228
    },
    {
      "epoch": 0.3418069220396544,
      "grad_norm": 0.70703125,
      "learning_rate": 1.127771805563882e-06,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1711,
      "tokens_per_second_per_gpu": 17894.95,
      "total_tokens": 154729251
    },
    {
      "epoch": 0.34200669230385056,
      "grad_norm": 0.7265625,
      "learning_rate": 1.120155800319891e-06,
      "loss": 0.3629,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1712,
      "tokens_per_second_per_gpu": 15909.49,
      "total_tokens": 154807996
    },
    {
      "epoch": 0.34220646256804677,
      "grad_norm": 0.6953125,
      "learning_rate": 1.112564072230863e-06,
      "loss": 0.3772,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1713,
      "tokens_per_second_per_gpu": 17899.38,
      "total_tokens": 154895414
    },
    {
      "epoch": 0.3424062328322429,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1049966420522961e-06,
      "loss": 0.3745,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1714,
      "tokens_per_second_per_gpu": 17323.67,
      "total_tokens": 154980925
    },
    {
      "epoch": 0.34260600309643907,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0974535304732581e-06,
      "loss": 0.3742,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1715,
      "tokens_per_second_per_gpu": 18622.09,
      "total_tokens": 155071967
    },
    {
      "epoch": 0.3428057733606353,
      "grad_norm": 0.6875,
      "learning_rate": 1.0899347581163222e-06,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1716,
      "tokens_per_second_per_gpu": 19394.71,
      "total_tokens": 155165374
    },
    {
      "epoch": 0.34300554362483143,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0824403455375287e-06,
      "loss": 0.3965,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1717,
      "tokens_per_second_per_gpu": 18893.11,
      "total_tokens": 155257635
    },
    {
      "epoch": 0.34320531388902764,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0749703132263122e-06,
      "loss": 0.3916,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1718,
      "tokens_per_second_per_gpu": 18913.91,
      "total_tokens": 155350176
    },
    {
      "epoch": 0.3434050841532238,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0675246816054585e-06,
      "loss": 0.3697,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1719,
      "tokens_per_second_per_gpu": 18772.2,
      "total_tokens": 155441453
    },
    {
      "epoch": 0.34360485441741995,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0601034710310431e-06,
      "loss": 0.4382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1720,
      "tokens_per_second_per_gpu": 19958.29,
      "total_tokens": 155539891
    },
    {
      "epoch": 0.34380462468161616,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0527067017923654e-06,
      "loss": 0.3871,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1721,
      "tokens_per_second_per_gpu": 17940.29,
      "total_tokens": 155627594
    },
    {
      "epoch": 0.3440043949458123,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0453343941119153e-06,
      "loss": 0.3966,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1722,
      "tokens_per_second_per_gpu": 18538.19,
      "total_tokens": 155717068
    },
    {
      "epoch": 0.3442041652100085,
      "grad_norm": 0.67578125,
      "learning_rate": 1.037986568145297e-06,
      "loss": 0.3898,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1723,
      "tokens_per_second_per_gpu": 19389.16,
      "total_tokens": 155810626
    },
    {
      "epoch": 0.34440393547420467,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0306632439811893e-06,
      "loss": 0.385,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1724,
      "tokens_per_second_per_gpu": 18775.98,
      "total_tokens": 155901764
    },
    {
      "epoch": 0.3446037057384008,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0233644416412792e-06,
      "loss": 0.3764,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1725,
      "tokens_per_second_per_gpu": 17815.37,
      "total_tokens": 155988666
    },
    {
      "epoch": 0.34480347600259703,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0160901810802114e-06,
      "loss": 0.3825,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1726,
      "tokens_per_second_per_gpu": 17254.65,
      "total_tokens": 156073261
    },
    {
      "epoch": 0.3450032462667932,
      "grad_norm": 0.703125,
      "learning_rate": 1.0088404821855414e-06,
      "loss": 0.3796,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1727,
      "tokens_per_second_per_gpu": 17234.51,
      "total_tokens": 156157089
    },
    {
      "epoch": 0.34520301653098934,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0016153647776638e-06,
      "loss": 0.3875,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1728,
      "tokens_per_second_per_gpu": 18063.24,
      "total_tokens": 156244786
    },
    {
      "epoch": 0.34540278679518555,
      "grad_norm": 0.69921875,
      "learning_rate": 9.944148486097793e-07,
      "loss": 0.3699,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1729,
      "tokens_per_second_per_gpu": 17793.76,
      "total_tokens": 156332687
    },
    {
      "epoch": 0.3456025570593817,
      "grad_norm": 0.65625,
      "learning_rate": 9.872389533678207e-07,
      "loss": 0.3885,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1730,
      "tokens_per_second_per_gpu": 20503.93,
      "total_tokens": 156430470
    },
    {
      "epoch": 0.3458023273235779,
      "grad_norm": 0.703125,
      "learning_rate": 9.800876986704111e-07,
      "loss": 0.3646,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1731,
      "tokens_per_second_per_gpu": 16442.28,
      "total_tokens": 156511832
    },
    {
      "epoch": 0.34600209758777406,
      "grad_norm": 0.69140625,
      "learning_rate": 9.729611040688102e-07,
      "loss": 0.4045,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1732,
      "tokens_per_second_per_gpu": 19834.46,
      "total_tokens": 156607137
    },
    {
      "epoch": 0.3462018678519702,
      "grad_norm": 0.71484375,
      "learning_rate": 9.658591890468515e-07,
      "loss": 0.4021,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1733,
      "tokens_per_second_per_gpu": 18456.88,
      "total_tokens": 156697010
    },
    {
      "epoch": 0.3464016381161664,
      "grad_norm": 0.68359375,
      "learning_rate": 9.587819730209047e-07,
      "loss": 0.3814,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1734,
      "tokens_per_second_per_gpu": 17971.9,
      "total_tokens": 156785270
    },
    {
      "epoch": 0.3466014083803626,
      "grad_norm": 0.69921875,
      "learning_rate": 9.517294753398066e-07,
      "loss": 0.3641,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1735,
      "tokens_per_second_per_gpu": 18041.68,
      "total_tokens": 156872402
    },
    {
      "epoch": 0.3468011786445588,
      "grad_norm": 0.6953125,
      "learning_rate": 9.447017152848126e-07,
      "loss": 0.3953,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1736,
      "tokens_per_second_per_gpu": 18629.22,
      "total_tokens": 156963175
    },
    {
      "epoch": 0.34700094890875494,
      "grad_norm": 0.703125,
      "learning_rate": 9.376987120695547e-07,
      "loss": 0.3753,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1737,
      "tokens_per_second_per_gpu": 17348.91,
      "total_tokens": 157048649
    },
    {
      "epoch": 0.3472007191729511,
      "grad_norm": 0.69140625,
      "learning_rate": 9.307204848399754e-07,
      "loss": 0.401,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1738,
      "tokens_per_second_per_gpu": 20144.21,
      "total_tokens": 157146345
    },
    {
      "epoch": 0.3474004894371473,
      "grad_norm": 0.71484375,
      "learning_rate": 9.237670526742793e-07,
      "loss": 0.3531,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1739,
      "tokens_per_second_per_gpu": 16542.15,
      "total_tokens": 157226886
    },
    {
      "epoch": 0.34760025970134345,
      "grad_norm": 0.66796875,
      "learning_rate": 9.168384345828885e-07,
      "loss": 0.3654,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1740,
      "tokens_per_second_per_gpu": 18553.7,
      "total_tokens": 157318358
    },
    {
      "epoch": 0.3478000299655396,
      "grad_norm": 0.67578125,
      "learning_rate": 9.09934649508375e-07,
      "loss": 0.3473,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1741,
      "tokens_per_second_per_gpu": 17743.0,
      "total_tokens": 157404764
    },
    {
      "epoch": 0.3479998002297358,
      "grad_norm": 0.71875,
      "learning_rate": 9.030557163254283e-07,
      "loss": 0.3765,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1742,
      "tokens_per_second_per_gpu": 16504.68,
      "total_tokens": 157486823
    },
    {
      "epoch": 0.34819957049393196,
      "grad_norm": 0.67578125,
      "learning_rate": 8.96201653840788e-07,
      "loss": 0.4042,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1743,
      "tokens_per_second_per_gpu": 19282.34,
      "total_tokens": 157580449
    },
    {
      "epoch": 0.3483993407581282,
      "grad_norm": 0.72265625,
      "learning_rate": 8.893724807931969e-07,
      "loss": 0.3915,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1744,
      "tokens_per_second_per_gpu": 19005.51,
      "total_tokens": 157672173
    },
    {
      "epoch": 0.3485991110223243,
      "grad_norm": 0.6875,
      "learning_rate": 8.825682158533555e-07,
      "loss": 0.369,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1745,
      "tokens_per_second_per_gpu": 17255.47,
      "total_tokens": 157757135
    },
    {
      "epoch": 0.3487988812865205,
      "grad_norm": 0.6875,
      "learning_rate": 8.757888776238621e-07,
      "loss": 0.3869,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1746,
      "tokens_per_second_per_gpu": 19348.89,
      "total_tokens": 157851655
    },
    {
      "epoch": 0.3489986515507167,
      "grad_norm": 0.69140625,
      "learning_rate": 8.69034484639173e-07,
      "loss": 0.4288,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1747,
      "tokens_per_second_per_gpu": 19563.45,
      "total_tokens": 157946738
    },
    {
      "epoch": 0.34919842181491284,
      "grad_norm": 0.6875,
      "learning_rate": 8.623050553655365e-07,
      "loss": 0.4002,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1748,
      "tokens_per_second_per_gpu": 18132.08,
      "total_tokens": 158034825
    },
    {
      "epoch": 0.34939819207910905,
      "grad_norm": 0.6796875,
      "learning_rate": 8.55600608200956e-07,
      "loss": 0.3969,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1749,
      "tokens_per_second_per_gpu": 19761.92,
      "total_tokens": 158131017
    },
    {
      "epoch": 0.3495979623433052,
      "grad_norm": 0.6953125,
      "learning_rate": 8.489211614751359e-07,
      "loss": 0.3819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1750,
      "tokens_per_second_per_gpu": 17712.83,
      "total_tokens": 158218501
    },
    {
      "epoch": 0.34979773260750135,
      "grad_norm": 0.69140625,
      "learning_rate": 8.42266733449425e-07,
      "loss": 0.413,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1751,
      "tokens_per_second_per_gpu": 20591.96,
      "total_tokens": 158317208
    },
    {
      "epoch": 0.34999750287169756,
      "grad_norm": 0.72265625,
      "learning_rate": 8.356373423167807e-07,
      "loss": 0.3678,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1752,
      "tokens_per_second_per_gpu": 18860.99,
      "total_tokens": 158408983
    },
    {
      "epoch": 0.3501972731358937,
      "grad_norm": 0.69921875,
      "learning_rate": 8.290330062017015e-07,
      "loss": 0.3915,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1753,
      "tokens_per_second_per_gpu": 18468.1,
      "total_tokens": 158499292
    },
    {
      "epoch": 0.35039704340008987,
      "grad_norm": 0.69140625,
      "learning_rate": 8.224537431601886e-07,
      "loss": 0.4129,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1754,
      "tokens_per_second_per_gpu": 19755.88,
      "total_tokens": 158594035
    },
    {
      "epoch": 0.3505968136642861,
      "grad_norm": 0.72265625,
      "learning_rate": 8.158995711797002e-07,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1755,
      "tokens_per_second_per_gpu": 17225.56,
      "total_tokens": 158678109
    },
    {
      "epoch": 0.35079658392848223,
      "grad_norm": 0.69921875,
      "learning_rate": 8.093705081790892e-07,
      "loss": 0.3943,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1756,
      "tokens_per_second_per_gpu": 17955.52,
      "total_tokens": 158764744
    },
    {
      "epoch": 0.35099635419267844,
      "grad_norm": 0.6953125,
      "learning_rate": 8.02866572008566e-07,
      "loss": 0.4288,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1757,
      "tokens_per_second_per_gpu": 19918.67,
      "total_tokens": 158860575
    },
    {
      "epoch": 0.3511961244568746,
      "grad_norm": 0.671875,
      "learning_rate": 7.963877804496445e-07,
      "loss": 0.3766,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1758,
      "tokens_per_second_per_gpu": 19521.95,
      "total_tokens": 158953334
    },
    {
      "epoch": 0.35139589472107075,
      "grad_norm": 0.6875,
      "learning_rate": 7.899341512150894e-07,
      "loss": 0.3586,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1759,
      "tokens_per_second_per_gpu": 17728.66,
      "total_tokens": 159039572
    },
    {
      "epoch": 0.35159566498526695,
      "grad_norm": 0.7265625,
      "learning_rate": 7.835057019488823e-07,
      "loss": 0.4038,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1760,
      "tokens_per_second_per_gpu": 19524.09,
      "total_tokens": 159132884
    },
    {
      "epoch": 0.3517954352494631,
      "grad_norm": 0.69140625,
      "learning_rate": 7.771024502261526e-07,
      "loss": 0.3857,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1761,
      "tokens_per_second_per_gpu": 19028.98,
      "total_tokens": 159226327
    },
    {
      "epoch": 0.3519952055136593,
      "grad_norm": 0.671875,
      "learning_rate": 7.7072441355315e-07,
      "loss": 0.3894,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1762,
      "tokens_per_second_per_gpu": 19448.27,
      "total_tokens": 159321164
    },
    {
      "epoch": 0.35219497577785547,
      "grad_norm": 0.68359375,
      "learning_rate": 7.643716093671827e-07,
      "loss": 0.3811,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1763,
      "tokens_per_second_per_gpu": 20145.74,
      "total_tokens": 159417370
    },
    {
      "epoch": 0.3523947460420516,
      "grad_norm": 0.6953125,
      "learning_rate": 7.580440550365709e-07,
      "loss": 0.3862,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1764,
      "tokens_per_second_per_gpu": 18418.03,
      "total_tokens": 159508237
    },
    {
      "epoch": 0.35259451630624783,
      "grad_norm": 0.6953125,
      "learning_rate": 7.517417678606121e-07,
      "loss": 0.3995,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1765,
      "tokens_per_second_per_gpu": 19766.01,
      "total_tokens": 159603135
    },
    {
      "epoch": 0.352794286570444,
      "grad_norm": 0.69921875,
      "learning_rate": 7.454647650695157e-07,
      "loss": 0.3657,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1766,
      "tokens_per_second_per_gpu": 17870.97,
      "total_tokens": 159689732
    },
    {
      "epoch": 0.3529940568346402,
      "grad_norm": 0.69140625,
      "learning_rate": 7.392130638243667e-07,
      "loss": 0.4001,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1767,
      "tokens_per_second_per_gpu": 19498.96,
      "total_tokens": 159784001
    },
    {
      "epoch": 0.35319382709883634,
      "grad_norm": 0.68359375,
      "learning_rate": 7.329866812170805e-07,
      "loss": 0.4026,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1768,
      "tokens_per_second_per_gpu": 19301.26,
      "total_tokens": 159877302
    },
    {
      "epoch": 0.3533935973630325,
      "grad_norm": 0.6953125,
      "learning_rate": 7.267856342703461e-07,
      "loss": 0.3829,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1769,
      "tokens_per_second_per_gpu": 17857.62,
      "total_tokens": 159963556
    },
    {
      "epoch": 0.3535933676272287,
      "grad_norm": 0.6875,
      "learning_rate": 7.206099399375921e-07,
      "loss": 0.4068,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1770,
      "tokens_per_second_per_gpu": 17644.76,
      "total_tokens": 160051249
    },
    {
      "epoch": 0.35379313789142486,
      "grad_norm": 0.7109375,
      "learning_rate": 7.144596151029304e-07,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1771,
      "tokens_per_second_per_gpu": 18406.51,
      "total_tokens": 160140948
    },
    {
      "epoch": 0.353992908155621,
      "grad_norm": 0.6875,
      "learning_rate": 7.083346765811117e-07,
      "loss": 0.4133,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1772,
      "tokens_per_second_per_gpu": 20019.71,
      "total_tokens": 160238392
    },
    {
      "epoch": 0.3541926784198172,
      "grad_norm": 0.671875,
      "learning_rate": 7.022351411174866e-07,
      "loss": 0.3991,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1773,
      "tokens_per_second_per_gpu": 19219.42,
      "total_tokens": 160332463
    },
    {
      "epoch": 0.3543924486840134,
      "grad_norm": 0.69140625,
      "learning_rate": 6.961610253879503e-07,
      "loss": 0.3576,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1774,
      "tokens_per_second_per_gpu": 17251.64,
      "total_tokens": 160417905
    },
    {
      "epoch": 0.3545922189482096,
      "grad_norm": 0.6796875,
      "learning_rate": 6.901123459989068e-07,
      "loss": 0.4105,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1775,
      "tokens_per_second_per_gpu": 20819.61,
      "total_tokens": 160517581
    },
    {
      "epoch": 0.35479198921240573,
      "grad_norm": 0.7421875,
      "learning_rate": 6.840891194872112e-07,
      "loss": 0.4044,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1776,
      "tokens_per_second_per_gpu": 18153.09,
      "total_tokens": 160605102
    },
    {
      "epoch": 0.3549917594766019,
      "grad_norm": 0.6953125,
      "learning_rate": 6.780913623201346e-07,
      "loss": 0.3836,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1777,
      "tokens_per_second_per_gpu": 18048.27,
      "total_tokens": 160694063
    },
    {
      "epoch": 0.3551915297407981,
      "grad_norm": 0.69921875,
      "learning_rate": 6.72119090895319e-07,
      "loss": 0.3787,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1778,
      "tokens_per_second_per_gpu": 18091.42,
      "total_tokens": 160781168
    },
    {
      "epoch": 0.35539130000499425,
      "grad_norm": 0.703125,
      "learning_rate": 6.661723215407223e-07,
      "loss": 0.371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1779,
      "tokens_per_second_per_gpu": 18160.46,
      "total_tokens": 160870327
    },
    {
      "epoch": 0.35559107026919046,
      "grad_norm": 0.703125,
      "learning_rate": 6.602510705145893e-07,
      "loss": 0.3589,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1780,
      "tokens_per_second_per_gpu": 17178.27,
      "total_tokens": 160954404
    },
    {
      "epoch": 0.3557908405333866,
      "grad_norm": 0.6953125,
      "learning_rate": 6.543553540053926e-07,
      "loss": 0.4092,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1781,
      "tokens_per_second_per_gpu": 19472.52,
      "total_tokens": 161047370
    },
    {
      "epoch": 0.35599061079758276,
      "grad_norm": 0.6796875,
      "learning_rate": 6.484851881317933e-07,
      "loss": 0.409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1782,
      "tokens_per_second_per_gpu": 19781.79,
      "total_tokens": 161142310
    },
    {
      "epoch": 0.35619038106177897,
      "grad_norm": 0.6875,
      "learning_rate": 6.426405889426046e-07,
      "loss": 0.3925,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1783,
      "tokens_per_second_per_gpu": 20283.7,
      "total_tokens": 161239555
    },
    {
      "epoch": 0.3563901513259751,
      "grad_norm": 0.6875,
      "learning_rate": 6.368215724167337e-07,
      "loss": 0.4201,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1784,
      "tokens_per_second_per_gpu": 20387.32,
      "total_tokens": 161338295
    },
    {
      "epoch": 0.3565899215901713,
      "grad_norm": 0.7109375,
      "learning_rate": 6.310281544631547e-07,
      "loss": 0.4021,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1785,
      "tokens_per_second_per_gpu": 19789.66,
      "total_tokens": 161434835
    },
    {
      "epoch": 0.3567896918543675,
      "grad_norm": 0.734375,
      "learning_rate": 6.252603509208466e-07,
      "loss": 0.4031,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1786,
      "tokens_per_second_per_gpu": 19592.64,
      "total_tokens": 161528129
    },
    {
      "epoch": 0.35698946211856364,
      "grad_norm": 0.734375,
      "learning_rate": 6.195181775587655e-07,
      "loss": 0.3734,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1787,
      "tokens_per_second_per_gpu": 16783.28,
      "total_tokens": 161610746
    },
    {
      "epoch": 0.35718923238275985,
      "grad_norm": 0.7109375,
      "learning_rate": 6.138016500757948e-07,
      "loss": 0.3697,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1788,
      "tokens_per_second_per_gpu": 18655.36,
      "total_tokens": 161701657
    },
    {
      "epoch": 0.357389002646956,
      "grad_norm": 0.6796875,
      "learning_rate": 6.081107841007006e-07,
      "loss": 0.3982,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1789,
      "tokens_per_second_per_gpu": 18852.02,
      "total_tokens": 161794146
    },
    {
      "epoch": 0.35758877291115215,
      "grad_norm": 0.69921875,
      "learning_rate": 6.024455951920949e-07,
      "loss": 0.3972,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1790,
      "tokens_per_second_per_gpu": 19723.46,
      "total_tokens": 161887838
    },
    {
      "epoch": 0.35778854317534836,
      "grad_norm": 0.6640625,
      "learning_rate": 5.968060988383884e-07,
      "loss": 0.3651,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1791,
      "tokens_per_second_per_gpu": 19001.15,
      "total_tokens": 161979691
    },
    {
      "epoch": 0.3579883134395445,
      "grad_norm": 0.71875,
      "learning_rate": 5.911923104577455e-07,
      "loss": 0.3824,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1792,
      "tokens_per_second_per_gpu": 17818.18,
      "total_tokens": 162068210
    },
    {
      "epoch": 0.3581880837037407,
      "grad_norm": 0.69140625,
      "learning_rate": 5.856042453980526e-07,
      "loss": 0.4157,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1793,
      "tokens_per_second_per_gpu": 19191.03,
      "total_tokens": 162160936
    },
    {
      "epoch": 0.3583878539679369,
      "grad_norm": 0.6953125,
      "learning_rate": 5.800419189368655e-07,
      "loss": 0.3776,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1794,
      "tokens_per_second_per_gpu": 18042.46,
      "total_tokens": 162248710
    },
    {
      "epoch": 0.35858762423213303,
      "grad_norm": 0.6875,
      "learning_rate": 5.745053462813699e-07,
      "loss": 0.3834,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1795,
      "tokens_per_second_per_gpu": 17770.84,
      "total_tokens": 162336939
    },
    {
      "epoch": 0.35878739449632924,
      "grad_norm": 0.69140625,
      "learning_rate": 5.689945425683474e-07,
      "loss": 0.4056,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1796,
      "tokens_per_second_per_gpu": 17450.28,
      "total_tokens": 162424365
    },
    {
      "epoch": 0.3589871647605254,
      "grad_norm": 0.6953125,
      "learning_rate": 5.63509522864123e-07,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1797,
      "tokens_per_second_per_gpu": 18272.99,
      "total_tokens": 162513885
    },
    {
      "epoch": 0.35918693502472154,
      "grad_norm": 0.69921875,
      "learning_rate": 5.580503021645323e-07,
      "loss": 0.3897,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1798,
      "tokens_per_second_per_gpu": 18187.29,
      "total_tokens": 162603284
    },
    {
      "epoch": 0.35938670528891775,
      "grad_norm": 0.71875,
      "learning_rate": 5.526168953948752e-07,
      "loss": 0.3886,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1799,
      "tokens_per_second_per_gpu": 17438.39,
      "total_tokens": 162688355
    },
    {
      "epoch": 0.3595864755531139,
      "grad_norm": 0.703125,
      "learning_rate": 5.47209317409878e-07,
      "loss": 0.3887,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1800,
      "tokens_per_second_per_gpu": 18507.47,
      "total_tokens": 162777055
    },
    {
      "epoch": 0.3597862458173101,
      "grad_norm": 0.6953125,
      "learning_rate": 5.418275829936537e-07,
      "loss": 0.387,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1801,
      "tokens_per_second_per_gpu": 18334.22,
      "total_tokens": 162865312
    },
    {
      "epoch": 0.35998601608150627,
      "grad_norm": 0.68359375,
      "learning_rate": 5.36471706859657e-07,
      "loss": 0.4149,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1802,
      "tokens_per_second_per_gpu": 19449.28,
      "total_tokens": 162960680
    },
    {
      "epoch": 0.3601857863457024,
      "grad_norm": 0.72265625,
      "learning_rate": 5.311417036506516e-07,
      "loss": 0.3531,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1803,
      "tokens_per_second_per_gpu": 16983.69,
      "total_tokens": 163043382
    },
    {
      "epoch": 0.3603855566098986,
      "grad_norm": 0.6953125,
      "learning_rate": 5.258375879386601e-07,
      "loss": 0.4252,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1804,
      "tokens_per_second_per_gpu": 18338.64,
      "total_tokens": 163133373
    },
    {
      "epoch": 0.3605853268740948,
      "grad_norm": 0.67578125,
      "learning_rate": 5.205593742249326e-07,
      "loss": 0.4073,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1805,
      "tokens_per_second_per_gpu": 19802.62,
      "total_tokens": 163228212
    },
    {
      "epoch": 0.360785097138291,
      "grad_norm": 0.6796875,
      "learning_rate": 5.15307076939906e-07,
      "loss": 0.3947,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1806,
      "tokens_per_second_per_gpu": 19347.48,
      "total_tokens": 163322776
    },
    {
      "epoch": 0.36098486740248714,
      "grad_norm": 0.703125,
      "learning_rate": 5.100807104431571e-07,
      "loss": 0.3987,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1807,
      "tokens_per_second_per_gpu": 18475.25,
      "total_tokens": 163414315
    },
    {
      "epoch": 0.3611846376666833,
      "grad_norm": 0.70703125,
      "learning_rate": 5.048802890233773e-07,
      "loss": 0.3735,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1808,
      "tokens_per_second_per_gpu": 17771.92,
      "total_tokens": 163500361
    },
    {
      "epoch": 0.3613844079308795,
      "grad_norm": 0.78125,
      "learning_rate": 4.997058268983135e-07,
      "loss": 0.3604,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1809,
      "tokens_per_second_per_gpu": 17612.31,
      "total_tokens": 163585477
    },
    {
      "epoch": 0.36158417819507566,
      "grad_norm": 0.71484375,
      "learning_rate": 4.945573382147517e-07,
      "loss": 0.3737,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1810,
      "tokens_per_second_per_gpu": 17610.41,
      "total_tokens": 163671372
    },
    {
      "epoch": 0.36178394845927186,
      "grad_norm": 0.69140625,
      "learning_rate": 4.894348370484648e-07,
      "loss": 0.3945,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1811,
      "tokens_per_second_per_gpu": 19305.75,
      "total_tokens": 163765056
    },
    {
      "epoch": 0.361983718723468,
      "grad_norm": 0.72265625,
      "learning_rate": 4.84338337404171e-07,
      "loss": 0.3804,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1812,
      "tokens_per_second_per_gpu": 16962.26,
      "total_tokens": 163848646
    },
    {
      "epoch": 0.36218348898766417,
      "grad_norm": 0.67578125,
      "learning_rate": 4.792678532155115e-07,
      "loss": 0.3943,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1813,
      "tokens_per_second_per_gpu": 20623.75,
      "total_tokens": 163947383
    },
    {
      "epoch": 0.3623832592518604,
      "grad_norm": 0.69140625,
      "learning_rate": 4.7422339834499065e-07,
      "loss": 0.394,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1814,
      "tokens_per_second_per_gpu": 19299.6,
      "total_tokens": 164040140
    },
    {
      "epoch": 0.36258302951605653,
      "grad_norm": 0.74609375,
      "learning_rate": 4.692049865839565e-07,
      "loss": 0.3825,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1815,
      "tokens_per_second_per_gpu": 17185.35,
      "total_tokens": 164125318
    },
    {
      "epoch": 0.3627827997802527,
      "grad_norm": 0.77734375,
      "learning_rate": 4.642126316525586e-07,
      "loss": 0.3838,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1816,
      "tokens_per_second_per_gpu": 19100.18,
      "total_tokens": 164218546
    },
    {
      "epoch": 0.3629825700444489,
      "grad_norm": 1.15625,
      "learning_rate": 4.592463471997022e-07,
      "loss": 0.3805,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1817,
      "tokens_per_second_per_gpu": 19193.72,
      "total_tokens": 164312483
    },
    {
      "epoch": 0.36318234030864505,
      "grad_norm": 0.68359375,
      "learning_rate": 4.543061468030208e-07,
      "loss": 0.3912,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1818,
      "tokens_per_second_per_gpu": 19092.09,
      "total_tokens": 164405281
    },
    {
      "epoch": 0.36338211057284125,
      "grad_norm": 0.8828125,
      "learning_rate": 4.493920439688315e-07,
      "loss": 0.4037,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1819,
      "tokens_per_second_per_gpu": 18613.69,
      "total_tokens": 164497214
    },
    {
      "epoch": 0.3635818808370374,
      "grad_norm": 0.67578125,
      "learning_rate": 4.4450405213210424e-07,
      "loss": 0.403,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1820,
      "tokens_per_second_per_gpu": 19233.73,
      "total_tokens": 164590864
    },
    {
      "epoch": 0.36378165110123356,
      "grad_norm": 0.70703125,
      "learning_rate": 4.396421846564236e-07,
      "loss": 0.4055,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1821,
      "tokens_per_second_per_gpu": 17606.8,
      "total_tokens": 164676535
    },
    {
      "epoch": 0.36398142136542977,
      "grad_norm": 0.70703125,
      "learning_rate": 4.348064548339492e-07,
      "loss": 0.3881,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1822,
      "tokens_per_second_per_gpu": 17523.76,
      "total_tokens": 164762648
    },
    {
      "epoch": 0.3641811916296259,
      "grad_norm": 0.70703125,
      "learning_rate": 4.299968758853812e-07,
      "loss": 0.3937,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1823,
      "tokens_per_second_per_gpu": 19013.26,
      "total_tokens": 164852796
    },
    {
      "epoch": 0.36438096189382213,
      "grad_norm": 0.69921875,
      "learning_rate": 4.2521346095992453e-07,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1824,
      "tokens_per_second_per_gpu": 19271.76,
      "total_tokens": 164945648
    },
    {
      "epoch": 0.3645807321580183,
      "grad_norm": 0.671875,
      "learning_rate": 4.204562231352516e-07,
      "loss": 0.3823,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1825,
      "tokens_per_second_per_gpu": 18717.6,
      "total_tokens": 165036825
    },
    {
      "epoch": 0.36478050242221444,
      "grad_norm": 0.67578125,
      "learning_rate": 4.1572517541747294e-07,
      "loss": 0.4179,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1826,
      "tokens_per_second_per_gpu": 19761.39,
      "total_tokens": 165132620
    },
    {
      "epoch": 0.36498027268641065,
      "grad_norm": 0.68359375,
      "learning_rate": 4.1102033074108984e-07,
      "loss": 0.3821,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1827,
      "tokens_per_second_per_gpu": 18292.4,
      "total_tokens": 165222233
    },
    {
      "epoch": 0.3651800429506068,
      "grad_norm": 1.1015625,
      "learning_rate": 4.0634170196896747e-07,
      "loss": 0.3889,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1828,
      "tokens_per_second_per_gpu": 18870.24,
      "total_tokens": 165313280
    },
    {
      "epoch": 0.36537981321480295,
      "grad_norm": 0.72265625,
      "learning_rate": 4.016893018922996e-07,
      "loss": 0.3822,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1829,
      "tokens_per_second_per_gpu": 17095.19,
      "total_tokens": 165397519
    },
    {
      "epoch": 0.36557958347899916,
      "grad_norm": 0.6875,
      "learning_rate": 3.9706314323056936e-07,
      "loss": 0.4112,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1830,
      "tokens_per_second_per_gpu": 19884.26,
      "total_tokens": 165494727
    },
    {
      "epoch": 0.3657793537431953,
      "grad_norm": 0.6953125,
      "learning_rate": 3.924632386315186e-07,
      "loss": 0.3724,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1831,
      "tokens_per_second_per_gpu": 17566.26,
      "total_tokens": 165580361
    },
    {
      "epoch": 0.3659791240073915,
      "grad_norm": 0.67578125,
      "learning_rate": 3.878896006711108e-07,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1832,
      "tokens_per_second_per_gpu": 20417.86,
      "total_tokens": 165678569
    },
    {
      "epoch": 0.3661788942715877,
      "grad_norm": 0.6875,
      "learning_rate": 3.83342241853496e-07,
      "loss": 0.3738,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1833,
      "tokens_per_second_per_gpu": 19691.62,
      "total_tokens": 165771878
    },
    {
      "epoch": 0.3663786645357838,
      "grad_norm": 0.69921875,
      "learning_rate": 3.788211746109827e-07,
      "loss": 0.3819,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1834,
      "tokens_per_second_per_gpu": 18441.73,
      "total_tokens": 165862064
    },
    {
      "epoch": 0.36657843479998004,
      "grad_norm": 0.69140625,
      "learning_rate": 3.743264113039924e-07,
      "loss": 0.3856,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1835,
      "tokens_per_second_per_gpu": 19244.65,
      "total_tokens": 165953703
    },
    {
      "epoch": 0.3667782050641762,
      "grad_norm": 0.71875,
      "learning_rate": 3.698579642210398e-07,
      "loss": 0.3896,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1836,
      "tokens_per_second_per_gpu": 16853.5,
      "total_tokens": 166036374
    },
    {
      "epoch": 0.3669779753283724,
      "grad_norm": 0.703125,
      "learning_rate": 3.6541584557868604e-07,
      "loss": 0.3879,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1837,
      "tokens_per_second_per_gpu": 17244.25,
      "total_tokens": 166121095
    },
    {
      "epoch": 0.36717774559256855,
      "grad_norm": 0.69140625,
      "learning_rate": 3.6100006752151416e-07,
      "loss": 0.3726,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1838,
      "tokens_per_second_per_gpu": 18064.37,
      "total_tokens": 166208228
    },
    {
      "epoch": 0.3673775158567647,
      "grad_norm": 0.73046875,
      "learning_rate": 3.5661064212209494e-07,
      "loss": 0.3708,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1839,
      "tokens_per_second_per_gpu": 17875.47,
      "total_tokens": 166293652
    },
    {
      "epoch": 0.3675772861209609,
      "grad_norm": 0.6875,
      "learning_rate": 3.522475813809467e-07,
      "loss": 0.3842,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1840,
      "tokens_per_second_per_gpu": 18272.44,
      "total_tokens": 166382002
    },
    {
      "epoch": 0.36777705638515706,
      "grad_norm": 0.7109375,
      "learning_rate": 3.4791089722651437e-07,
      "loss": 0.3763,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1841,
      "tokens_per_second_per_gpu": 17658.02,
      "total_tokens": 166467236
    },
    {
      "epoch": 0.3679768266493532,
      "grad_norm": 0.703125,
      "learning_rate": 3.4360060151512276e-07,
      "loss": 0.414,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1842,
      "tokens_per_second_per_gpu": 18595.62,
      "total_tokens": 166557327
    },
    {
      "epoch": 0.3681765969135494,
      "grad_norm": 0.6953125,
      "learning_rate": 3.393167060309588e-07,
      "loss": 0.396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1843,
      "tokens_per_second_per_gpu": 18781.74,
      "total_tokens": 166647867
    },
    {
      "epoch": 0.3683763671777456,
      "grad_norm": 0.68359375,
      "learning_rate": 3.3505922248602827e-07,
      "loss": 0.3741,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1844,
      "tokens_per_second_per_gpu": 18527.38,
      "total_tokens": 166737081
    },
    {
      "epoch": 0.3685761374419418,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3082816252012927e-07,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1845,
      "tokens_per_second_per_gpu": 19061.56,
      "total_tokens": 166830586
    },
    {
      "epoch": 0.36877590770613794,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2662353770081755e-07,
      "loss": 0.3869,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1846,
      "tokens_per_second_per_gpu": 18702.31,
      "total_tokens": 166921050
    },
    {
      "epoch": 0.3689756779703341,
      "grad_norm": 0.6875,
      "learning_rate": 3.2244535952337565e-07,
      "loss": 0.378,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1847,
      "tokens_per_second_per_gpu": 18008.17,
      "total_tokens": 167009206
    },
    {
      "epoch": 0.3691754482345303,
      "grad_norm": 0.6953125,
      "learning_rate": 3.182936394107838e-07,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1848,
      "tokens_per_second_per_gpu": 17846.08,
      "total_tokens": 167097605
    },
    {
      "epoch": 0.36937521849872645,
      "grad_norm": 0.74609375,
      "learning_rate": 3.1416838871368925e-07,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1849,
      "tokens_per_second_per_gpu": 19220.86,
      "total_tokens": 167189502
    },
    {
      "epoch": 0.36957498876292266,
      "grad_norm": 0.69921875,
      "learning_rate": 3.100696187103658e-07,
      "loss": 0.3888,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1850,
      "tokens_per_second_per_gpu": 18462.9,
      "total_tokens": 167279472
    },
    {
      "epoch": 0.3697747590271188,
      "grad_norm": 0.7265625,
      "learning_rate": 3.059973406066963e-07,
      "loss": 0.3788,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1851,
      "tokens_per_second_per_gpu": 17996.23,
      "total_tokens": 167365049
    },
    {
      "epoch": 0.36997452929131497,
      "grad_norm": 0.69140625,
      "learning_rate": 3.019515655361327e-07,
      "loss": 0.366,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1852,
      "tokens_per_second_per_gpu": 18288.2,
      "total_tokens": 167455299
    },
    {
      "epoch": 0.3701742995555112,
      "grad_norm": 0.7265625,
      "learning_rate": 2.9793230455966936e-07,
      "loss": 0.3645,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1853,
      "tokens_per_second_per_gpu": 17064.63,
      "total_tokens": 167538682
    },
    {
      "epoch": 0.37037406981970733,
      "grad_norm": 0.6796875,
      "learning_rate": 2.9393956866581307e-07,
      "loss": 0.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1854,
      "tokens_per_second_per_gpu": 19711.44,
      "total_tokens": 167632724
    },
    {
      "epoch": 0.37057384008390354,
      "grad_norm": 0.6875,
      "learning_rate": 2.8997336877055194e-07,
      "loss": 0.3775,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1855,
      "tokens_per_second_per_gpu": 18169.54,
      "total_tokens": 167721916
    },
    {
      "epoch": 0.3707736103480997,
      "grad_norm": 0.96875,
      "learning_rate": 2.860337157173243e-07,
      "loss": 0.353,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1856,
      "tokens_per_second_per_gpu": 17571.96,
      "total_tokens": 167807716
    },
    {
      "epoch": 0.37097338061229584,
      "grad_norm": 0.703125,
      "learning_rate": 2.8212062027698995e-07,
      "loss": 0.3951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1857,
      "tokens_per_second_per_gpu": 18691.87,
      "total_tokens": 167898020
    },
    {
      "epoch": 0.37117315087649205,
      "grad_norm": 0.71875,
      "learning_rate": 2.7823409314780227e-07,
      "loss": 0.3884,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1858,
      "tokens_per_second_per_gpu": 17830.49,
      "total_tokens": 167984993
    },
    {
      "epoch": 0.3713729211406882,
      "grad_norm": 0.73828125,
      "learning_rate": 2.7437414495538274e-07,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1859,
      "tokens_per_second_per_gpu": 17475.73,
      "total_tokens": 168069733
    },
    {
      "epoch": 0.37157269140488436,
      "grad_norm": 0.69921875,
      "learning_rate": 2.705407862526765e-07,
      "loss": 0.3821,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1860,
      "tokens_per_second_per_gpu": 18949.63,
      "total_tokens": 168161284
    },
    {
      "epoch": 0.37177246166908057,
      "grad_norm": 0.7109375,
      "learning_rate": 2.667340275199426e-07,
      "loss": 0.3951,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1861,
      "tokens_per_second_per_gpu": 18918.42,
      "total_tokens": 168252961
    },
    {
      "epoch": 0.3719722319332767,
      "grad_norm": 0.68359375,
      "learning_rate": 2.629538791647124e-07,
      "loss": 0.3812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1862,
      "tokens_per_second_per_gpu": 18700.83,
      "total_tokens": 168344305
    },
    {
      "epoch": 0.37217200219747293,
      "grad_norm": 0.70703125,
      "learning_rate": 2.592003515217689e-07,
      "loss": 0.4034,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1863,
      "tokens_per_second_per_gpu": 19936.93,
      "total_tokens": 168439380
    },
    {
      "epoch": 0.3723717724616691,
      "grad_norm": 0.72265625,
      "learning_rate": 2.554734548531157e-07,
      "loss": 0.366,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1864,
      "tokens_per_second_per_gpu": 19093.61,
      "total_tokens": 168530332
    },
    {
      "epoch": 0.37257154272586523,
      "grad_norm": 0.70703125,
      "learning_rate": 2.5177319934793995e-07,
      "loss": 0.3854,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1865,
      "tokens_per_second_per_gpu": 18667.83,
      "total_tokens": 168621641
    },
    {
      "epoch": 0.37277131299006144,
      "grad_norm": 0.765625,
      "learning_rate": 2.4809959512260285e-07,
      "loss": 0.3917,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1866,
      "tokens_per_second_per_gpu": 20269.71,
      "total_tokens": 168719533
    },
    {
      "epoch": 0.3729710832542576,
      "grad_norm": 0.71875,
      "learning_rate": 2.44452652220597e-07,
      "loss": 0.3901,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1867,
      "tokens_per_second_per_gpu": 17002.57,
      "total_tokens": 168803623
    },
    {
      "epoch": 0.3731708535184538,
      "grad_norm": 0.69921875,
      "learning_rate": 2.4083238061252565e-07,
      "loss": 0.3979,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1868,
      "tokens_per_second_per_gpu": 19566.6,
      "total_tokens": 168899298
    },
    {
      "epoch": 0.37337062378264996,
      "grad_norm": 0.6953125,
      "learning_rate": 2.3723879019607376e-07,
      "loss": 0.3948,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1869,
      "tokens_per_second_per_gpu": 19707.12,
      "total_tokens": 168993711
    },
    {
      "epoch": 0.3735703940468461,
      "grad_norm": 0.7109375,
      "learning_rate": 2.3367189079597785e-07,
      "loss": 0.3665,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1870,
      "tokens_per_second_per_gpu": 17127.74,
      "total_tokens": 169077908
    },
    {
      "epoch": 0.3737701643110423,
      "grad_norm": 0.67578125,
      "learning_rate": 2.3013169216400732e-07,
      "loss": 0.3933,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1871,
      "tokens_per_second_per_gpu": 18675.53,
      "total_tokens": 169169803
    },
    {
      "epoch": 0.37396993457523847,
      "grad_norm": 0.734375,
      "learning_rate": 2.2661820397892886e-07,
      "loss": 0.3865,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1872,
      "tokens_per_second_per_gpu": 16803.35,
      "total_tokens": 169252424
    },
    {
      "epoch": 0.3741697048394346,
      "grad_norm": 0.72265625,
      "learning_rate": 2.2313143584648423e-07,
      "loss": 0.36,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1873,
      "tokens_per_second_per_gpu": 16908.41,
      "total_tokens": 169334667
    },
    {
      "epoch": 0.37436947510363083,
      "grad_norm": 0.7265625,
      "learning_rate": 2.1967139729936693e-07,
      "loss": 0.3882,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1874,
      "tokens_per_second_per_gpu": 17549.44,
      "total_tokens": 169421469
    },
    {
      "epoch": 0.374569245367827,
      "grad_norm": 0.73046875,
      "learning_rate": 2.1623809779718674e-07,
      "loss": 0.4237,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1875,
      "tokens_per_second_per_gpu": 19728.92,
      "total_tokens": 169516357
    },
    {
      "epoch": 0.3747690156320232,
      "grad_norm": 0.70703125,
      "learning_rate": 2.1283154672645522e-07,
      "loss": 0.4213,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1876,
      "tokens_per_second_per_gpu": 20000.03,
      "total_tokens": 169612851
    },
    {
      "epoch": 0.37496878589621935,
      "grad_norm": 0.6796875,
      "learning_rate": 2.0945175340055356e-07,
      "loss": 0.4039,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1877,
      "tokens_per_second_per_gpu": 18733.09,
      "total_tokens": 169705405
    },
    {
      "epoch": 0.3751685561604155,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0609872705970702e-07,
      "loss": 0.3849,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1878,
      "tokens_per_second_per_gpu": 17949.86,
      "total_tokens": 169792538
    },
    {
      "epoch": 0.3753683264246117,
      "grad_norm": 0.703125,
      "learning_rate": 2.0277247687096157e-07,
      "loss": 0.3847,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1879,
      "tokens_per_second_per_gpu": 18073.29,
      "total_tokens": 169880309
    },
    {
      "epoch": 0.37556809668880786,
      "grad_norm": 0.69921875,
      "learning_rate": 1.9947301192815738e-07,
      "loss": 0.4042,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1880,
      "tokens_per_second_per_gpu": 18850.28,
      "total_tokens": 169971380
    },
    {
      "epoch": 0.37576786695300407,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9620034125190645e-07,
      "loss": 0.4137,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1881,
      "tokens_per_second_per_gpu": 19652.37,
      "total_tokens": 170066071
    },
    {
      "epoch": 0.3759676372172002,
      "grad_norm": 0.69921875,
      "learning_rate": 1.9295447378956612e-07,
      "loss": 0.3954,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1882,
      "tokens_per_second_per_gpu": 19500.74,
      "total_tokens": 170159968
    },
    {
      "epoch": 0.3761674074813964,
      "grad_norm": 0.6875,
      "learning_rate": 1.8973541841521336e-07,
      "loss": 0.4036,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1883,
      "tokens_per_second_per_gpu": 19564.16,
      "total_tokens": 170256041
    },
    {
      "epoch": 0.3763671777455926,
      "grad_norm": 0.69921875,
      "learning_rate": 1.865431839296228e-07,
      "loss": 0.4103,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1884,
      "tokens_per_second_per_gpu": 18370.84,
      "total_tokens": 170346381
    },
    {
      "epoch": 0.37656694800978874,
      "grad_norm": 0.6875,
      "learning_rate": 1.833777790602398e-07,
      "loss": 0.4068,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1885,
      "tokens_per_second_per_gpu": 19355.84,
      "total_tokens": 170439406
    },
    {
      "epoch": 0.3767667182739849,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8023921246116405e-07,
      "loss": 0.3941,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1886,
      "tokens_per_second_per_gpu": 19559.63,
      "total_tokens": 170532326
    },
    {
      "epoch": 0.3769664885381811,
      "grad_norm": 0.67578125,
      "learning_rate": 1.7712749271311392e-07,
      "loss": 0.409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1887,
      "tokens_per_second_per_gpu": 19033.39,
      "total_tokens": 170624845
    },
    {
      "epoch": 0.37716625880237725,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7404262832341313e-07,
      "loss": 0.4308,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1888,
      "tokens_per_second_per_gpu": 20036.71,
      "total_tokens": 170722265
    },
    {
      "epoch": 0.37736602906657346,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7098462772596302e-07,
      "loss": 0.3767,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1889,
      "tokens_per_second_per_gpu": 18043.31,
      "total_tokens": 170809788
    },
    {
      "epoch": 0.3775657993307696,
      "grad_norm": 0.69140625,
      "learning_rate": 1.679534992812215e-07,
      "loss": 0.4263,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1890,
      "tokens_per_second_per_gpu": 19260.37,
      "total_tokens": 170905377
    },
    {
      "epoch": 0.37776556959496577,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6494925127617632e-07,
      "loss": 0.3692,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1891,
      "tokens_per_second_per_gpu": 16946.28,
      "total_tokens": 170989714
    },
    {
      "epoch": 0.377965339859162,
      "grad_norm": 0.69140625,
      "learning_rate": 1.619718919243285e-07,
      "loss": 0.3958,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1892,
      "tokens_per_second_per_gpu": 19612.03,
      "total_tokens": 171084441
    },
    {
      "epoch": 0.37816511012335813,
      "grad_norm": 0.703125,
      "learning_rate": 1.5902142936566333e-07,
      "loss": 0.3933,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1893,
      "tokens_per_second_per_gpu": 18842.93,
      "total_tokens": 171175227
    },
    {
      "epoch": 0.37836488038755434,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5609787166663392e-07,
      "loss": 0.3563,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1894,
      "tokens_per_second_per_gpu": 16580.86,
      "total_tokens": 171257029
    },
    {
      "epoch": 0.3785646506517505,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5320122682013438e-07,
      "loss": 0.3578,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1895,
      "tokens_per_second_per_gpu": 19243.43,
      "total_tokens": 171348305
    },
    {
      "epoch": 0.37876442091594664,
      "grad_norm": 0.703125,
      "learning_rate": 1.5033150274548325e-07,
      "loss": 0.3761,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1896,
      "tokens_per_second_per_gpu": 17106.5,
      "total_tokens": 171430765
    },
    {
      "epoch": 0.37896419118014285,
      "grad_norm": 0.671875,
      "learning_rate": 1.474887072883935e-07,
      "loss": 0.3792,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1897,
      "tokens_per_second_per_gpu": 19254.11,
      "total_tokens": 171524178
    },
    {
      "epoch": 0.379163961444339,
      "grad_norm": 0.703125,
      "learning_rate": 1.446728482209603e-07,
      "loss": 0.3566,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1898,
      "tokens_per_second_per_gpu": 17517.2,
      "total_tokens": 171609983
    },
    {
      "epoch": 0.37936373170853516,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4188393324163663e-07,
      "loss": 0.4065,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1899,
      "tokens_per_second_per_gpu": 18327.44,
      "total_tokens": 171700246
    },
    {
      "epoch": 0.37956350197273137,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3912196997520776e-07,
      "loss": 0.3991,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1900,
      "tokens_per_second_per_gpu": 18633.18,
      "total_tokens": 171791308
    },
    {
      "epoch": 0.3797632722369275,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3638696597277678e-07,
      "loss": 0.3828,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1901,
      "tokens_per_second_per_gpu": 17639.71,
      "total_tokens": 171876402
    },
    {
      "epoch": 0.3799630425011237,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3367892871173904e-07,
      "loss": 0.3778,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1902,
      "tokens_per_second_per_gpu": 18263.68,
      "total_tokens": 171965799
    },
    {
      "epoch": 0.3801628127653199,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3099786559576555e-07,
      "loss": 0.3785,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1903,
      "tokens_per_second_per_gpu": 17456.44,
      "total_tokens": 172052045
    },
    {
      "epoch": 0.38036258302951603,
      "grad_norm": 0.71875,
      "learning_rate": 1.2834378395478187e-07,
      "loss": 0.4053,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1904,
      "tokens_per_second_per_gpu": 18229.79,
      "total_tokens": 172140477
    },
    {
      "epoch": 0.38056235329371224,
      "grad_norm": 0.71875,
      "learning_rate": 1.2571669104494254e-07,
      "loss": 0.4104,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1905,
      "tokens_per_second_per_gpu": 18608.79,
      "total_tokens": 172230666
    },
    {
      "epoch": 0.3807621235579084,
      "grad_norm": 0.7109375,
      "learning_rate": 1.231165940486234e-07,
      "loss": 0.387,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1906,
      "tokens_per_second_per_gpu": 20217.9,
      "total_tokens": 172326516
    },
    {
      "epoch": 0.3809618938221046,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2054350007438708e-07,
      "loss": 0.3875,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1907,
      "tokens_per_second_per_gpu": 19060.55,
      "total_tokens": 172419319
    },
    {
      "epoch": 0.38116166408630076,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1799741615697524e-07,
      "loss": 0.3791,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1908,
      "tokens_per_second_per_gpu": 19059.85,
      "total_tokens": 172511773
    },
    {
      "epoch": 0.3813614343504969,
      "grad_norm": 0.6875,
      "learning_rate": 1.154783492572853e-07,
      "loss": 0.3988,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1909,
      "tokens_per_second_per_gpu": 18478.34,
      "total_tokens": 172601910
    },
    {
      "epoch": 0.3815612046146931,
      "grad_norm": 0.6953125,
      "learning_rate": 1.129863062623482e-07,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1910,
      "tokens_per_second_per_gpu": 18894.66,
      "total_tokens": 172693505
    },
    {
      "epoch": 0.38176097487888927,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1052129398531508e-07,
      "loss": 0.3409,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1911,
      "tokens_per_second_per_gpu": 17697.06,
      "total_tokens": 172779692
    },
    {
      "epoch": 0.3819607451430855,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0808331916543624e-07,
      "loss": 0.3709,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1912,
      "tokens_per_second_per_gpu": 18560.17,
      "total_tokens": 172870944
    },
    {
      "epoch": 0.38216051540728163,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0567238846803995e-07,
      "loss": 0.3821,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1913,
      "tokens_per_second_per_gpu": 18447.18,
      "total_tokens": 172961402
    },
    {
      "epoch": 0.3823602856714778,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0328850848452032e-07,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1914,
      "tokens_per_second_per_gpu": 19778.66,
      "total_tokens": 173055926
    },
    {
      "epoch": 0.382560055935674,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0093168573231393e-07,
      "loss": 0.3913,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1915,
      "tokens_per_second_per_gpu": 19386.32,
      "total_tokens": 173149934
    },
    {
      "epoch": 0.38275982619987015,
      "grad_norm": 0.68359375,
      "learning_rate": 9.86019266548821e-08,
      "loss": 0.3809,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1916,
      "tokens_per_second_per_gpu": 19058.32,
      "total_tokens": 173239878
    },
    {
      "epoch": 0.3829595964640663,
      "grad_norm": 0.71875,
      "learning_rate": 9.629923762170091e-08,
      "loss": 0.3744,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1917,
      "tokens_per_second_per_gpu": 17145.76,
      "total_tokens": 173322215
    },
    {
      "epoch": 0.3831593667282625,
      "grad_norm": 0.703125,
      "learning_rate": 9.402362492823002e-08,
      "loss": 0.4034,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1918,
      "tokens_per_second_per_gpu": 18822.7,
      "total_tokens": 173413622
    },
    {
      "epoch": 0.38335913699245866,
      "grad_norm": 0.71484375,
      "learning_rate": 9.177509479591174e-08,
      "loss": 0.3621,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1919,
      "tokens_per_second_per_gpu": 17323.23,
      "total_tokens": 173497442
    },
    {
      "epoch": 0.38355890725665487,
      "grad_norm": 1.0390625,
      "learning_rate": 8.955365337213861e-08,
      "loss": 0.3902,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1920,
      "tokens_per_second_per_gpu": 18815.5,
      "total_tokens": 173587201
    },
    {
      "epoch": 0.383758677520851,
      "grad_norm": 0.703125,
      "learning_rate": 8.735930673024806e-08,
      "loss": 0.3648,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1921,
      "tokens_per_second_per_gpu": 17384.47,
      "total_tokens": 173671964
    },
    {
      "epoch": 0.3839584477850472,
      "grad_norm": 0.70703125,
      "learning_rate": 8.519206086950005e-08,
      "loss": 0.3693,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1922,
      "tokens_per_second_per_gpu": 17286.21,
      "total_tokens": 173755944
    },
    {
      "epoch": 0.3841582180492434,
      "grad_norm": 0.69921875,
      "learning_rate": 8.305192171506049e-08,
      "loss": 0.3746,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1923,
      "tokens_per_second_per_gpu": 17744.39,
      "total_tokens": 173843045
    },
    {
      "epoch": 0.38435798831343954,
      "grad_norm": 0.70703125,
      "learning_rate": 8.093889511799014e-08,
      "loss": 0.3661,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1924,
      "tokens_per_second_per_gpu": 17706.98,
      "total_tokens": 173929193
    },
    {
      "epoch": 0.38455775857763574,
      "grad_norm": 0.7109375,
      "learning_rate": 7.885298685522235e-08,
      "loss": 0.3744,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1925,
      "tokens_per_second_per_gpu": 19261.13,
      "total_tokens": 174020664
    },
    {
      "epoch": 0.3847575288418319,
      "grad_norm": 0.703125,
      "learning_rate": 7.679420262954984e-08,
      "loss": 0.432,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1926,
      "tokens_per_second_per_gpu": 19388.32,
      "total_tokens": 174116243
    },
    {
      "epoch": 0.38495729910602805,
      "grad_norm": 0.70703125,
      "learning_rate": 7.476254806961014e-08,
      "loss": 0.3705,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1927,
      "tokens_per_second_per_gpu": 17298.88,
      "total_tokens": 174201040
    },
    {
      "epoch": 0.38515706937022426,
      "grad_norm": 0.6875,
      "learning_rate": 7.275802872987015e-08,
      "loss": 0.382,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1928,
      "tokens_per_second_per_gpu": 18849.0,
      "total_tokens": 174293279
    },
    {
      "epoch": 0.3853568396344204,
      "grad_norm": 0.70703125,
      "learning_rate": 7.078065009060941e-08,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1929,
      "tokens_per_second_per_gpu": 18604.87,
      "total_tokens": 174382017
    },
    {
      "epoch": 0.38555660989861656,
      "grad_norm": 0.7109375,
      "learning_rate": 6.883041755790576e-08,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1930,
      "tokens_per_second_per_gpu": 17621.06,
      "total_tokens": 174468551
    },
    {
      "epoch": 0.3857563801628128,
      "grad_norm": 0.6796875,
      "learning_rate": 6.690733646361858e-08,
      "loss": 0.4156,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1931,
      "tokens_per_second_per_gpu": 19937.41,
      "total_tokens": 174563904
    },
    {
      "epoch": 0.3859561504270089,
      "grad_norm": 0.69140625,
      "learning_rate": 6.501141206537887e-08,
      "loss": 0.4124,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1932,
      "tokens_per_second_per_gpu": 19000.45,
      "total_tokens": 174656483
    },
    {
      "epoch": 0.38615592069120513,
      "grad_norm": 0.69140625,
      "learning_rate": 6.314264954657257e-08,
      "loss": 0.3784,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1933,
      "tokens_per_second_per_gpu": 19343.2,
      "total_tokens": 174747367
    },
    {
      "epoch": 0.3863556909554013,
      "grad_norm": 0.66015625,
      "learning_rate": 6.130105401632502e-08,
      "loss": 0.3805,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1934,
      "tokens_per_second_per_gpu": 19624.99,
      "total_tokens": 174842325
    },
    {
      "epoch": 0.38655546121959744,
      "grad_norm": 0.6875,
      "learning_rate": 5.948663050948767e-08,
      "loss": 0.4169,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1935,
      "tokens_per_second_per_gpu": 19656.08,
      "total_tokens": 174937517
    },
    {
      "epoch": 0.38675523148379365,
      "grad_norm": 0.66796875,
      "learning_rate": 5.769938398662356e-08,
      "loss": 0.413,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1936,
      "tokens_per_second_per_gpu": 19463.42,
      "total_tokens": 175034821
    },
    {
      "epoch": 0.3869550017479898,
      "grad_norm": 0.6796875,
      "learning_rate": 5.5939319333998546e-08,
      "loss": 0.407,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1937,
      "tokens_per_second_per_gpu": 20243.23,
      "total_tokens": 175133339
    },
    {
      "epoch": 0.387154772012186,
      "grad_norm": 0.69921875,
      "learning_rate": 5.420644136356123e-08,
      "loss": 0.354,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1938,
      "tokens_per_second_per_gpu": 16834.51,
      "total_tokens": 175216404
    },
    {
      "epoch": 0.38735454227638216,
      "grad_norm": 0.71875,
      "learning_rate": 5.250075481293526e-08,
      "loss": 0.4174,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1939,
      "tokens_per_second_per_gpu": 19235.25,
      "total_tokens": 175309444
    },
    {
      "epoch": 0.3875543125405783,
      "grad_norm": 0.67578125,
      "learning_rate": 5.0822264345400384e-08,
      "loss": 0.4168,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1940,
      "tokens_per_second_per_gpu": 19734.06,
      "total_tokens": 175405964
    },
    {
      "epoch": 0.3877540828047745,
      "grad_norm": 0.69140625,
      "learning_rate": 4.9170974549885844e-08,
      "loss": 0.3843,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1941,
      "tokens_per_second_per_gpu": 17482.66,
      "total_tokens": 175493821
    },
    {
      "epoch": 0.3879538530689707,
      "grad_norm": 0.6796875,
      "learning_rate": 4.754688994095591e-08,
      "loss": 0.4108,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1942,
      "tokens_per_second_per_gpu": 19405.5,
      "total_tokens": 175588089
    },
    {
      "epoch": 0.38815362333316683,
      "grad_norm": 0.69140625,
      "learning_rate": 4.595001495879547e-08,
      "loss": 0.3833,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1943,
      "tokens_per_second_per_gpu": 18914.48,
      "total_tokens": 175680265
    },
    {
      "epoch": 0.38835339359736304,
      "grad_norm": 0.68359375,
      "learning_rate": 4.438035396920004e-08,
      "loss": 0.421,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1944,
      "tokens_per_second_per_gpu": 20027.02,
      "total_tokens": 175778526
    },
    {
      "epoch": 0.3885531638615592,
      "grad_norm": 0.6953125,
      "learning_rate": 4.2837911263562406e-08,
      "loss": 0.4024,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1945,
      "tokens_per_second_per_gpu": 17875.83,
      "total_tokens": 175865471
    },
    {
      "epoch": 0.3887529341257554,
      "grad_norm": 0.73046875,
      "learning_rate": 4.132269105886155e-08,
      "loss": 0.3812,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1946,
      "tokens_per_second_per_gpu": 18687.4,
      "total_tokens": 175955683
    },
    {
      "epoch": 0.38895270438995155,
      "grad_norm": 0.7109375,
      "learning_rate": 3.9834697497652673e-08,
      "loss": 0.4053,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1947,
      "tokens_per_second_per_gpu": 20347.16,
      "total_tokens": 176053816
    },
    {
      "epoch": 0.3891524746541477,
      "grad_norm": 0.69140625,
      "learning_rate": 3.837393464805273e-08,
      "loss": 0.4113,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1948,
      "tokens_per_second_per_gpu": 19446.14,
      "total_tokens": 176146995
    },
    {
      "epoch": 0.3893522449183439,
      "grad_norm": 0.6796875,
      "learning_rate": 3.694040650373154e-08,
      "loss": 0.3742,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1949,
      "tokens_per_second_per_gpu": 17895.97,
      "total_tokens": 176236147
    },
    {
      "epoch": 0.38955201518254007,
      "grad_norm": 0.68359375,
      "learning_rate": 3.553411698390075e-08,
      "loss": 0.3739,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1950,
      "tokens_per_second_per_gpu": 18088.06,
      "total_tokens": 176323960
    },
    {
      "epoch": 0.3897517854467363,
      "grad_norm": 0.67578125,
      "learning_rate": 3.4155069933301535e-08,
      "loss": 0.4084,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1951,
      "tokens_per_second_per_gpu": 20068.42,
      "total_tokens": 176419545
    },
    {
      "epoch": 0.38995155571093243,
      "grad_norm": 0.67578125,
      "learning_rate": 3.280326912219689e-08,
      "loss": 0.3705,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1952,
      "tokens_per_second_per_gpu": 17995.41,
      "total_tokens": 176506865
    },
    {
      "epoch": 0.3901513259751286,
      "grad_norm": 0.71484375,
      "learning_rate": 3.147871824635717e-08,
      "loss": 0.3911,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1953,
      "tokens_per_second_per_gpu": 18445.65,
      "total_tokens": 176597052
    },
    {
      "epoch": 0.3903510962393248,
      "grad_norm": 0.70703125,
      "learning_rate": 3.018142092705567e-08,
      "loss": 0.398,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1954,
      "tokens_per_second_per_gpu": 18372.97,
      "total_tokens": 176686310
    },
    {
      "epoch": 0.39055086650352094,
      "grad_norm": 0.68359375,
      "learning_rate": 2.8911380711051928e-08,
      "loss": 0.3939,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1955,
      "tokens_per_second_per_gpu": 19616.24,
      "total_tokens": 176780756
    },
    {
      "epoch": 0.39075063676771715,
      "grad_norm": 0.6796875,
      "learning_rate": 2.766860107058844e-08,
      "loss": 0.3944,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1956,
      "tokens_per_second_per_gpu": 18291.31,
      "total_tokens": 176870763
    },
    {
      "epoch": 0.3909504070319133,
      "grad_norm": 1.6015625,
      "learning_rate": 2.645308540337843e-08,
      "loss": 0.4371,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1957,
      "tokens_per_second_per_gpu": 19597.81,
      "total_tokens": 176969066
    },
    {
      "epoch": 0.39115017729610946,
      "grad_norm": 0.6796875,
      "learning_rate": 2.5264837032592527e-08,
      "loss": 0.4118,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1958,
      "tokens_per_second_per_gpu": 20354.89,
      "total_tokens": 177068584
    },
    {
      "epoch": 0.39134994756030567,
      "grad_norm": 0.6640625,
      "learning_rate": 2.4103859206857648e-08,
      "loss": 0.4098,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1959,
      "tokens_per_second_per_gpu": 21004.39,
      "total_tokens": 177169220
    },
    {
      "epoch": 0.3915497178245018,
      "grad_norm": 0.69921875,
      "learning_rate": 2.2970155100244807e-08,
      "loss": 0.3934,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1960,
      "tokens_per_second_per_gpu": 17898.7,
      "total_tokens": 177256362
    },
    {
      "epoch": 0.391749488088698,
      "grad_norm": 0.6953125,
      "learning_rate": 2.1863727812254653e-08,
      "loss": 0.4051,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1961,
      "tokens_per_second_per_gpu": 19132.09,
      "total_tokens": 177348530
    },
    {
      "epoch": 0.3919492583528942,
      "grad_norm": 1.65625,
      "learning_rate": 2.0784580367818607e-08,
      "loss": 0.4137,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1962,
      "tokens_per_second_per_gpu": 19480.33,
      "total_tokens": 177443192
    },
    {
      "epoch": 0.39214902861709033,
      "grad_norm": 0.734375,
      "learning_rate": 1.973271571728441e-08,
      "loss": 0.4071,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1963,
      "tokens_per_second_per_gpu": 17717.86,
      "total_tokens": 177530512
    },
    {
      "epoch": 0.39234879888128654,
      "grad_norm": 0.6875,
      "learning_rate": 1.8708136736409476e-08,
      "loss": 0.3954,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1964,
      "tokens_per_second_per_gpu": 18266.5,
      "total_tokens": 177618854
    },
    {
      "epoch": 0.3925485691454827,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7710846226355327e-08,
      "loss": 0.3887,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1965,
      "tokens_per_second_per_gpu": 19366.23,
      "total_tokens": 177713212
    },
    {
      "epoch": 0.39274833940967885,
      "grad_norm": 0.6875,
      "learning_rate": 1.674084691367428e-08,
      "loss": 0.3998,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1966,
      "tokens_per_second_per_gpu": 19924.87,
      "total_tokens": 177809546
    },
    {
      "epoch": 0.39294810967387506,
      "grad_norm": 0.67578125,
      "learning_rate": 1.5798141450307225e-08,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1967,
      "tokens_per_second_per_gpu": 19727.84,
      "total_tokens": 177904440
    },
    {
      "epoch": 0.3931478799380712,
      "grad_norm": 0.66796875,
      "learning_rate": 1.4882732413576961e-08,
      "loss": 0.4184,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1968,
      "tokens_per_second_per_gpu": 20051.46,
      "total_tokens": 178002939
    },
    {
      "epoch": 0.3933476502022674,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3994622306173766e-08,
      "loss": 0.3973,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1969,
      "tokens_per_second_per_gpu": 19307.12,
      "total_tokens": 178095882
    },
    {
      "epoch": 0.39354742046646357,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3133813556159835e-08,
      "loss": 0.399,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1970,
      "tokens_per_second_per_gpu": 18636.67,
      "total_tokens": 178186265
    },
    {
      "epoch": 0.3937471907306597,
      "grad_norm": 0.671875,
      "learning_rate": 1.230030851695263e-08,
      "loss": 0.4102,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1971,
      "tokens_per_second_per_gpu": 19970.75,
      "total_tokens": 178284066
    },
    {
      "epoch": 0.39394696099485593,
      "grad_norm": 0.6875,
      "learning_rate": 1.1494109467321545e-08,
      "loss": 0.3929,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1972,
      "tokens_per_second_per_gpu": 18080.21,
      "total_tokens": 178373059
    },
    {
      "epoch": 0.3941467312590521,
      "grad_norm": 0.6484375,
      "learning_rate": 1.0715218611384582e-08,
      "loss": 0.3487,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1973,
      "tokens_per_second_per_gpu": 18107.46,
      "total_tokens": 178462577
    },
    {
      "epoch": 0.39434650152324824,
      "grad_norm": 0.70703125,
      "learning_rate": 9.963638078600567e-09,
      "loss": 0.3908,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1974,
      "tokens_per_second_per_gpu": 18851.74,
      "total_tokens": 178552907
    },
    {
      "epoch": 0.39454627178744445,
      "grad_norm": 0.6953125,
      "learning_rate": 9.23936992376251e-09,
      "loss": 0.4106,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1975,
      "tokens_per_second_per_gpu": 18813.45,
      "total_tokens": 178647475
    },
    {
      "epoch": 0.3947460420516406,
      "grad_norm": 0.69140625,
      "learning_rate": 8.542416126989805e-09,
      "loss": 0.39,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1976,
      "tokens_per_second_per_gpu": 18072.03,
      "total_tokens": 178734580
    },
    {
      "epoch": 0.3949458123158368,
      "grad_norm": 0.66796875,
      "learning_rate": 7.872778593728258e-09,
      "loss": 0.3751,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1977,
      "tokens_per_second_per_gpu": 17771.57,
      "total_tokens": 178822026
    },
    {
      "epoch": 0.39514558258003296,
      "grad_norm": 0.6796875,
      "learning_rate": 7.230459154742298e-09,
      "loss": 0.3885,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1978,
      "tokens_per_second_per_gpu": 19663.45,
      "total_tokens": 178916775
    },
    {
      "epoch": 0.3953453528442291,
      "grad_norm": 0.67578125,
      "learning_rate": 6.615459566108318e-09,
      "loss": 0.38,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1979,
      "tokens_per_second_per_gpu": 18458.06,
      "total_tokens": 179008067
    },
    {
      "epoch": 0.3955451231084253,
      "grad_norm": 0.71875,
      "learning_rate": 6.02778150921024e-09,
      "loss": 0.376,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1980,
      "tokens_per_second_per_gpu": 17789.7,
      "total_tokens": 179094215
    },
    {
      "epoch": 0.3957448933726215,
      "grad_norm": 0.69140625,
      "learning_rate": 5.467426590739511e-09,
      "loss": 0.4055,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1981,
      "tokens_per_second_per_gpu": 18779.17,
      "total_tokens": 179186628
    },
    {
      "epoch": 0.3959446636368177,
      "grad_norm": 0.72265625,
      "learning_rate": 4.9343963426840006e-09,
      "loss": 0.3634,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1982,
      "tokens_per_second_per_gpu": 16630.51,
      "total_tokens": 179268560
    },
    {
      "epoch": 0.39614443390101384,
      "grad_norm": 0.69140625,
      "learning_rate": 4.428692222329112e-09,
      "loss": 0.3858,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1983,
      "tokens_per_second_per_gpu": 17749.38,
      "total_tokens": 179354760
    },
    {
      "epoch": 0.39634420416521,
      "grad_norm": 0.69140625,
      "learning_rate": 3.950315612248901e-09,
      "loss": 0.396,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1984,
      "tokens_per_second_per_gpu": 19222.54,
      "total_tokens": 179448168
    },
    {
      "epoch": 0.3965439744294062,
      "grad_norm": 0.70703125,
      "learning_rate": 3.499267820307184e-09,
      "loss": 0.3989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1985,
      "tokens_per_second_per_gpu": 18562.21,
      "total_tokens": 179537487
    },
    {
      "epoch": 0.39674374469360235,
      "grad_norm": 0.7109375,
      "learning_rate": 3.0755500796531e-09,
      "loss": 0.3889,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1986,
      "tokens_per_second_per_gpu": 19360.04,
      "total_tokens": 179631266
    },
    {
      "epoch": 0.3969435149577985,
      "grad_norm": 0.703125,
      "learning_rate": 2.679163548712227e-09,
      "loss": 0.4044,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1987,
      "tokens_per_second_per_gpu": 18921.43,
      "total_tokens": 179723006
    },
    {
      "epoch": 0.3971432852219947,
      "grad_norm": 0.703125,
      "learning_rate": 2.3101093111921323e-09,
      "loss": 0.3852,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1988,
      "tokens_per_second_per_gpu": 17377.98,
      "total_tokens": 179808368
    },
    {
      "epoch": 0.39734305548619087,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9683883760723834e-09,
      "loss": 0.3744,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1989,
      "tokens_per_second_per_gpu": 16488.2,
      "total_tokens": 179890457
    },
    {
      "epoch": 0.3975428257503871,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6540016776045442e-09,
      "loss": 0.3863,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1990,
      "tokens_per_second_per_gpu": 19005.78,
      "total_tokens": 179982811
    },
    {
      "epoch": 0.3977425960145832,
      "grad_norm": 0.6875,
      "learning_rate": 1.3669500753099586e-09,
      "loss": 0.3486,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1991,
      "tokens_per_second_per_gpu": 18440.46,
      "total_tokens": 180071820
    },
    {
      "epoch": 0.3979423662787794,
      "grad_norm": 0.734375,
      "learning_rate": 1.1072343539775266e-09,
      "loss": 0.3857,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1992,
      "tokens_per_second_per_gpu": 17896.25,
      "total_tokens": 180158607
    },
    {
      "epoch": 0.3981421365429756,
      "grad_norm": 0.6875,
      "learning_rate": 8.748552236603758e-10,
      "loss": 0.3956,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1993,
      "tokens_per_second_per_gpu": 18176.41,
      "total_tokens": 180247388
    },
    {
      "epoch": 0.39834190680717174,
      "grad_norm": 0.67578125,
      "learning_rate": 6.698133196725298e-10,
      "loss": 0.4143,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1994,
      "tokens_per_second_per_gpu": 20961.08,
      "total_tokens": 180348547
    },
    {
      "epoch": 0.39854167707136795,
      "grad_norm": 0.70703125,
      "learning_rate": 4.921092025911289e-10,
      "loss": 0.386,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1995,
      "tokens_per_second_per_gpu": 18703.8,
      "total_tokens": 180439985
    },
    {
      "epoch": 0.3987414473355641,
      "grad_norm": 0.70703125,
      "learning_rate": 3.4174335825420955e-10,
      "loss": 0.3569,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1996,
      "tokens_per_second_per_gpu": 17405.72,
      "total_tokens": 180523703
    },
    {
      "epoch": 0.39894121759976026,
      "grad_norm": 0.7265625,
      "learning_rate": 2.1871619775404308e-10,
      "loss": 0.3989,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1997,
      "tokens_per_second_per_gpu": 18004.88,
      "total_tokens": 180611805
    },
    {
      "epoch": 0.39914098786395646,
      "grad_norm": 0.6796875,
      "learning_rate": 1.2302805744268674e-10,
      "loss": 0.4109,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1998,
      "tokens_per_second_per_gpu": 19842.41,
      "total_tokens": 180708918
    },
    {
      "epoch": 0.3993407581281526,
      "grad_norm": 0.7109375,
      "learning_rate": 5.467919892865326e-11,
      "loss": 0.3736,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 1999,
      "tokens_per_second_per_gpu": 18293.3,
      "total_tokens": 180797281
    },
    {
      "epoch": 0.3995405283923488,
      "grad_norm": 0.69921875,
      "learning_rate": 1.366980907580029e-11,
      "loss": 0.3977,
      "memory/device_reserved (GiB)": 101.87,
      "memory/max_active (GiB)": 91.58,
      "memory/max_allocated (GiB)": 91.58,
      "step": 2000,
      "tokens_per_second_per_gpu": 17910.55,
      "total_tokens": 180885523
    }
  ],
  "logging_steps": 1,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.714305740189991e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}