{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.9672,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.666666666666667e-07,
      "loss": 1.101,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.0874,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 1.0302,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.0754,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.0552,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 1.056,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.666666666666667e-06,
      "loss": 1.0807,
      "step": 8
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-06,
      "loss": 0.9727,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.0609,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 1.0426,
      "step": 11
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.9514,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.946,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.9148,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-06,
      "loss": 0.9166,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.8649,
      "step": 16
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.9305,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 6e-06,
      "loss": 0.8224,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.8195,
      "step": 19
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.8794,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 7e-06,
      "loss": 0.7984,
      "step": 21
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.8589,
      "step": 22
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.7756,
      "step": 23
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.8005,
      "step": 24
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.8899,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.8451,
      "step": 26
    },
    {
      "epoch": 0.01,
      "learning_rate": 9e-06,
      "loss": 0.8091,
      "step": 27
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.8062,
      "step": 28
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.8387,
      "step": 29
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 0.774,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0333333333333335e-05,
      "loss": 0.7554,
      "step": 31
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.8233,
      "step": 32
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.7762,
      "step": 33
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1333333333333334e-05,
      "loss": 0.8007,
      "step": 34
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.7668,
      "step": 35
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2e-05,
      "loss": 0.7904,
      "step": 36
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2333333333333334e-05,
      "loss": 0.824,
      "step": 37
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2666666666666667e-05,
      "loss": 0.7715,
      "step": 38
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.7424,
      "step": 39
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.7362,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.3666666666666667e-05,
      "loss": 0.7583,
      "step": 41
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.4e-05,
      "loss": 0.8013,
      "step": 42
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.4333333333333334e-05,
      "loss": 0.7942,
      "step": 43
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.7419,
      "step": 44
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7636,
      "step": 45
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5333333333333334e-05,
      "loss": 0.8152,
      "step": 46
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5666666666666667e-05,
      "loss": 0.7442,
      "step": 47
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.7432,
      "step": 48
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.6333333333333335e-05,
      "loss": 0.7055,
      "step": 49
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.7479,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7e-05,
      "loss": 0.7984,
      "step": 51
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.7365,
      "step": 52
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7666666666666668e-05,
      "loss": 0.7525,
      "step": 53
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.8e-05,
      "loss": 0.7407,
      "step": 54
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.798,
      "step": 55
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.7416,
      "step": 56
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9e-05,
      "loss": 0.8083,
      "step": 57
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9333333333333333e-05,
      "loss": 0.7662,
      "step": 58
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9666666666666666e-05,
      "loss": 0.7259,
      "step": 59
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 0.7616,
      "step": 60
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999986888082895e-05,
      "loss": 0.6824,
      "step": 61
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999994755236596e-05,
      "loss": 0.7889,
      "step": 62
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999881992952353e-05,
      "loss": 0.7583,
      "step": 63
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999979021001399e-05,
      "loss": 0.7744,
      "step": 64
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999672203791564e-05,
      "loss": 0.756,
      "step": 65
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999952797459453e-05,
      "loss": 0.7075,
      "step": 66
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999357522801125e-05,
      "loss": 0.7573,
      "step": 67
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999916084885832e-05,
      "loss": 0.6942,
      "step": 68
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999893795328188e-05,
      "loss": 0.7428,
      "step": 69
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9998688836656322e-05,
      "loss": 0.7451,
      "step": 70
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9998413499634927e-05,
      "loss": 0.7405,
      "step": 71
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9998111942939727e-05,
      "loss": 0.7407,
      "step": 72
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9997784167361526e-05,
      "loss": 0.7318,
      "step": 73
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9997430173759876e-05,
      "loss": 0.7698,
      "step": 74
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999704996306308e-05,
      "loss": 0.7822,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9996643536268202e-05,
      "loss": 0.8002,
      "step": 76
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9996210894441047e-05,
      "loss": 0.7528,
      "step": 77
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9995752038716166e-05,
      "loss": 0.7819,
      "step": 78
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9995266970296856e-05,
      "loss": 0.7695,
      "step": 79
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9994755690455154e-05,
      "loss": 0.7493,
      "step": 80
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9994218200531823e-05,
      "loss": 0.7173,
      "step": 81
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999365450193638e-05,
      "loss": 0.7805,
      "step": 82
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999306459614705e-05,
      "loss": 0.7364,
      "step": 83
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.99924484847108e-05,
      "loss": 0.7128,
      "step": 84
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9991806169243302e-05,
      "loss": 0.7227,
      "step": 85
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9991137651428957e-05,
      "loss": 0.8038,
      "step": 86
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999044293302088e-05,
      "loss": 0.8071,
      "step": 87
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.998972201584088e-05,
      "loss": 0.7258,
      "step": 88
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9988974901779482e-05,
      "loss": 0.7094,
      "step": 89
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.998820159279591e-05,
      "loss": 0.7975,
      "step": 90
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998740209091807e-05,
      "loss": 0.7445,
      "step": 91
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9986576398242566e-05,
      "loss": 0.7527,
      "step": 92
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998572451693468e-05,
      "loss": 0.7194,
      "step": 93
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998484644922837e-05,
      "loss": 0.8186,
      "step": 94
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9983942197426272e-05,
      "loss": 0.7771,
      "step": 95
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9983011763899674e-05,
      "loss": 0.7875,
      "step": 96
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998205515108853e-05,
      "loss": 0.767,
      "step": 97
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998107236150145e-05,
      "loss": 0.7356,
      "step": 98
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9980063397715685e-05,
      "loss": 0.7307,
      "step": 99
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.997902826237712e-05,
      "loss": 0.712,
      "step": 100
    },
    {
      "epoch": 0.05,
      "eval_code_gate_load": [
        199.4,
        191.5,
        181.9,
        145.3,
        177.5,
        178.75,
        187.95,
        177.7
      ],
      "eval_code_loss": 0.5201171636581421,
      "eval_code_runtime": 1.7839,
      "eval_code_samples_per_second": 560.56,
      "eval_code_steps_per_second": 35.315,
      "step": 100
    },
    {
      "epoch": 0.05,
      "eval_orca_gate_load": [
        500.1,
        352.35,
        397.35,
        384.15,
        349.8,
        443.1,
        371.2,
        344.25
      ],
      "eval_orca_loss": 0.7696288824081421,
      "eval_orca_runtime": 2.0215,
      "eval_orca_samples_per_second": 494.677,
      "eval_orca_steps_per_second": 31.165,
      "step": 100
    },
    {
      "epoch": 0.05,
      "eval_math_gate_load": [
        320.6,
        221.3,
        228.4,
        217.5,
        248.3,
        243.2,
        282.25,
        271.15
      ],
      "eval_math_loss": 0.716601550579071,
      "eval_math_runtime": 1.8689,
      "eval_math_samples_per_second": 535.077,
      "eval_math_steps_per_second": 33.71,
      "step": 100
    },
    {
      "epoch": 0.05,
      "eval_sharegpt_gate_load": [
        1524.7,
        1115.1,
        1297.45,
        1170.1,
        1142.65,
        1383.85,
        1207.25,
        1070.5
      ],
      "eval_sharegpt_loss": 0.7251952886581421,
      "eval_sharegpt_runtime": 2.996,
      "eval_sharegpt_samples_per_second": 333.775,
      "eval_sharegpt_steps_per_second": 21.028,
      "step": 100
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9977966958200276e-05,
      "loss": 0.7711,
      "step": 101
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.997687948796831e-05,
      "loss": 0.7064,
      "step": 102
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9975765854532974e-05,
      "loss": 0.7409,
      "step": 103
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.997462606081465e-05,
      "loss": 0.7148,
      "step": 104
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9973460109802306e-05,
      "loss": 0.7689,
      "step": 105
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.997226800455352e-05,
      "loss": 0.7637,
      "step": 106
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9971049748194448e-05,
      "loss": 0.7224,
      "step": 107
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9969805343919822e-05,
      "loss": 0.7147,
      "step": 108
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9968534794992947e-05,
      "loss": 0.6927,
      "step": 109
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9967238104745695e-05,
      "loss": 0.7395,
      "step": 110
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.996591527657848e-05,
      "loss": 0.6984,
      "step": 111
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9964566313960265e-05,
      "loss": 0.7243,
      "step": 112
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9963191220428552e-05,
      "loss": 0.8089,
      "step": 113
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9961789999589357e-05,
      "loss": 0.7965,
      "step": 114
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.996036265511722e-05,
      "loss": 0.7219,
      "step": 115
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995890919075519e-05,
      "loss": 0.7092,
      "step": 116
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9957429610314797e-05,
      "loss": 0.7492,
      "step": 117
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995592391767608e-05,
      "loss": 0.7885,
      "step": 118
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995439211678754e-05,
      "loss": 0.7533,
      "step": 119
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995283421166614e-05,
      "loss": 0.6916,
      "step": 120
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995125020639731e-05,
      "loss": 0.6982,
      "step": 121
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.994964010513492e-05,
      "loss": 0.7155,
      "step": 122
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9948003912101274e-05,
      "loss": 0.7385,
      "step": 123
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9946341631587086e-05,
      "loss": 0.7199,
      "step": 124
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9944653267951507e-05,
      "loss": 0.68,
      "step": 125
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9942938825622064e-05,
      "loss": 0.8073,
      "step": 126
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.994119830909469e-05,
      "loss": 0.705,
      "step": 127
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9939431722933678e-05,
      "loss": 0.7802,
      "step": 128
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9937639071771704e-05,
      "loss": 0.7184,
      "step": 129
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.993582036030978e-05,
      "loss": 0.7413,
      "step": 130
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9933975593317263e-05,
      "loss": 0.7346,
      "step": 131
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9932104775631847e-05,
      "loss": 0.7115,
      "step": 132
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.993020791215953e-05,
      "loss": 0.713,
      "step": 133
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.992828500787461e-05,
      "loss": 0.7238,
      "step": 134
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9926336067819686e-05,
      "loss": 0.7186,
      "step": 135
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9924361097105624e-05,
      "loss": 0.7158,
      "step": 136
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9922360100911553e-05,
      "loss": 0.7106,
      "step": 137
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.992033308448486e-05,
      "loss": 0.7274,
      "step": 138
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9918280053141144e-05,
      "loss": 0.738,
      "step": 139
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9916201012264255e-05,
      "loss": 0.7267,
      "step": 140
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9914095967306224e-05,
      "loss": 0.734,
      "step": 141
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9911964923787295e-05,
      "loss": 0.7932,
      "step": 142
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.990980788729588e-05,
      "loss": 0.7568,
      "step": 143
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.990762486348855e-05,
      "loss": 0.7244,
      "step": 144
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9905415858090036e-05,
      "loss": 0.7349,
      "step": 145
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9903180876893195e-05,
      "loss": 0.674,
      "step": 146
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9900919925759e-05,
      "loss": 0.7401,
      "step": 147
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.989863301061654e-05,
      "loss": 0.7278,
      "step": 148
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9896320137462984e-05,
      "loss": 0.7734,
      "step": 149
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9893981312363563e-05,
      "loss": 0.6926,
      "step": 150
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.989161654145158e-05,
      "loss": 0.7339,
      "step": 151
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9889225830928365e-05,
      "loss": 0.7335,
      "step": 152
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9886809187063285e-05,
      "loss": 0.6829,
      "step": 153
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9884366616193707e-05,
      "loss": 0.6921,
      "step": 154
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.988189812472498e-05,
      "loss": 0.7695,
      "step": 155
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.987940371913044e-05,
      "loss": 0.7079,
      "step": 156
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 0.7568,
      "step": 157
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.987433719179702e-05,
      "loss": 0.7084,
      "step": 158
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.987176508334451e-05,
      "loss": 0.7419,
      "step": 159
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9869167087338908e-05,
      "loss": 0.7072,
      "step": 160
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9866543210593154e-05,
      "loss": 0.769,
      "step": 161
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.986389345998806e-05,
      "loss": 0.7518,
      "step": 162
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.986121784247229e-05,
      "loss": 0.7457,
      "step": 163
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9858516365062334e-05,
      "loss": 0.7448,
      "step": 164
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9855789034842504e-05,
      "loss": 0.7529,
      "step": 165
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9853035858964907e-05,
      "loss": 0.6531,
      "step": 166
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9850256844649422e-05,
      "loss": 0.7162,
      "step": 167
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9847451999183692e-05,
      "loss": 0.6711,
      "step": 168
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.98446213299231e-05,
      "loss": 0.687,
      "step": 169
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9841764844290744e-05,
      "loss": 0.7153,
      "step": 170
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9838882549777426e-05,
      "loss": 0.7152,
      "step": 171
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9835974453941623e-05,
      "loss": 0.7557,
      "step": 172
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.983304056440948e-05,
      "loss": 0.7054,
      "step": 173
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.983008088887478e-05,
      "loss": 0.6988,
      "step": 174
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9827095435098926e-05,
      "loss": 0.6721,
      "step": 175
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9824084210910924e-05,
      "loss": 0.7319,
      "step": 176
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9821047224207362e-05,
      "loss": 0.6729,
      "step": 177
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9817984482952378e-05,
      "loss": 0.7491,
      "step": 178
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9814895995177653e-05,
      "loss": 0.7494,
      "step": 179
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9811781768982392e-05,
      "loss": 0.7351,
      "step": 180
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9808641812533286e-05,
      "loss": 0.7003,
      "step": 181
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.980547613406451e-05,
      "loss": 0.7291,
      "step": 182
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9802284741877674e-05,
      "loss": 0.7574,
      "step": 183
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9799067644341844e-05,
      "loss": 0.6798,
      "step": 184
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9795824849893483e-05,
      "loss": 0.7454,
      "step": 185
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9792556367036432e-05,
      "loss": 0.6889,
      "step": 186
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9789262204341918e-05,
      "loss": 0.7128,
      "step": 187
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.978594237044849e-05,
      "loss": 0.6948,
      "step": 188
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9782596874062028e-05,
      "loss": 0.733,
      "step": 189
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.977922572395571e-05,
      "loss": 0.7165,
      "step": 190
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9775828928969976e-05,
      "loss": 0.7471,
      "step": 191
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.977240649801253e-05,
      "loss": 0.7039,
      "step": 192
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.97689584400583e-05,
      "loss": 0.6682,
      "step": 193
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9765484764149413e-05,
      "loss": 0.7194,
      "step": 194
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.976198547939518e-05,
      "loss": 0.6918,
      "step": 195
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9758460594972068e-05,
      "loss": 0.7341,
      "step": 196
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9754910120123675e-05,
      "loss": 0.7172,
      "step": 197
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9751334064160708e-05,
      "loss": 0.6916,
      "step": 198
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9747732436460955e-05,
      "loss": 0.7046,
      "step": 199
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9744105246469264e-05,
      "loss": 0.7297,
      "step": 200
    },
    {
      "epoch": 0.1,
      "eval_code_gate_load": [
        202.15,
        184.0,
        181.85,
        156.2,
        180.4,
        177.6,
        186.5,
        171.3
      ],
      "eval_code_loss": 0.5015624761581421,
      "eval_code_runtime": 1.8076,
      "eval_code_samples_per_second": 553.219,
      "eval_code_steps_per_second": 34.853,
      "step": 200
    },
    {
      "epoch": 0.1,
      "eval_orca_gate_load": [
        500.1,
        348.05,
        396.7,
        392.0,
        354.8,
        437.65,
        368.05,
        344.95
      ],
      "eval_orca_loss": 0.7518554925918579,
      "eval_orca_runtime": 2.0082,
      "eval_orca_samples_per_second": 497.958,
      "eval_orca_steps_per_second": 31.371,
      "step": 200
    },
    {
      "epoch": 0.1,
      "eval_math_gate_load": [
        319.05,
        220.35,
        222.35,
        231.65,
        255.05,
        239.45,
        280.9,
        263.9
      ],
      "eval_math_loss": 0.6756836175918579,
      "eval_math_runtime": 1.8563,
      "eval_math_samples_per_second": 538.698,
      "eval_math_steps_per_second": 33.938,
      "step": 200
    },
    {
      "epoch": 0.1,
      "eval_sharegpt_gate_load": [
        1522.6,
        1108.8,
        1288.45,
        1197.4,
        1139.0,
        1373.8,
        1198.4,
        1083.15
      ],
      "eval_sharegpt_loss": 0.7212890386581421,
      "eval_sharegpt_runtime": 3.0067,
      "eval_sharegpt_samples_per_second": 332.594,
      "eval_sharegpt_steps_per_second": 20.953,
      "step": 200
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9740452503697518e-05,
      "loss": 0.7521,
      "step": 201
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9736774217724614e-05,
      "loss": 0.716,
      "step": 202
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9733070398196423e-05,
      "loss": 0.7111,
      "step": 203
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9729341054825783e-05,
      "loss": 0.7261,
      "step": 204
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.972558619739246e-05,
      "loss": 0.7397,
      "step": 205
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.972180583574313e-05,
      "loss": 0.7313,
      "step": 206
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9717999979791356e-05,
      "loss": 0.6935,
      "step": 207
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9714168639517543e-05,
      "loss": 0.7706,
      "step": 208
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9710311824968942e-05,
      "loss": 0.7282,
      "step": 209
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9706429546259592e-05,
      "loss": 0.6551,
      "step": 210
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9702521813570322e-05,
      "loss": 0.6824,
      "step": 211
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9698588637148705e-05,
      "loss": 0.7006,
      "step": 212
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9694630027309035e-05,
      "loss": 0.734,
      "step": 213
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9690645994432307e-05,
      "loss": 0.6711,
      "step": 214
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9686636548966177e-05,
      "loss": 0.7231,
      "step": 215
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9682601701424958e-05,
      "loss": 0.7045,
      "step": 216
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9678541462389564e-05,
      "loss": 0.7036,
      "step": 217
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9674455842507494e-05,
      "loss": 0.6919,
      "step": 218
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9670344852492814e-05,
      "loss": 0.7436,
      "step": 219
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9666208503126115e-05,
      "loss": 0.7014,
      "step": 220
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.966204680525449e-05,
      "loss": 0.7137,
      "step": 221
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9657859769791506e-05,
      "loss": 0.6578,
      "step": 222
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.965364740771718e-05,
      "loss": 0.6613,
      "step": 223
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9649409730077934e-05,
      "loss": 0.748,
      "step": 224
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.964514674798659e-05,
      "loss": 0.7396,
      "step": 225
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9640858472622316e-05,
      "loss": 0.6985,
      "step": 226
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.963654491523062e-05,
      "loss": 0.7084,
      "step": 227
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9632206087123296e-05,
      "loss": 0.6876,
      "step": 228
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9627841999678422e-05,
      "loss": 0.719,
      "step": 229
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9623452664340305e-05,
      "loss": 0.6693,
      "step": 230
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9619038092619465e-05,
      "loss": 0.722,
      "step": 231
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9614598296092603e-05,
      "loss": 0.7495,
      "step": 232
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9610133286402565e-05,
      "loss": 0.653,
      "step": 233
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9605643075258323e-05,
      "loss": 0.7022,
      "step": 234
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.960112767443493e-05,
      "loss": 0.7374,
      "step": 235
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9596587095773496e-05,
      "loss": 0.6933,
      "step": 236
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9592021351181163e-05,
      "loss": 0.6685,
      "step": 237
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.958743045263106e-05,
      "loss": 0.7344,
      "step": 238
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9582814412162288e-05,
      "loss": 0.6979,
      "step": 239
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.957817324187987e-05,
      "loss": 0.6684,
      "step": 240
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.957350695395474e-05,
      "loss": 0.7081,
      "step": 241
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.956881556062369e-05,
      "loss": 0.7117,
      "step": 242
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.956409907418935e-05,
      "loss": 0.6398,
      "step": 243
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9559357507020163e-05,
      "loss": 0.7447,
      "step": 244
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.955459087155033e-05,
      "loss": 0.7159,
      "step": 245
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9549799180279793e-05,
      "loss": 0.7141,
      "step": 246
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9544982445774217e-05,
      "loss": 0.7247,
      "step": 247
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9540140680664915e-05,
      "loss": 0.6986,
      "step": 248
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9535273897648857e-05,
      "loss": 0.7617,
      "step": 249
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.953038210948861e-05,
      "loss": 0.7067,
      "step": 250
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9525465329012322e-05,
      "loss": 0.7115,
      "step": 251
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.952052356911368e-05,
      "loss": 0.7017,
      "step": 252
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9515556842751863e-05,
      "loss": 0.6714,
      "step": 253
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.6627,
      "step": 254
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9505548542802805e-05,
      "loss": 0.6834,
      "step": 255
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.950050699546116e-05,
      "loss": 0.7105,
      "step": 256
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.949544053414748e-05,
      "loss": 0.7134,
      "step": 257
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9490349172147964e-05,
      "loss": 0.7188,
      "step": 258
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9485232922814117e-05,
      "loss": 0.6334,
      "step": 259
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9480091799562706e-05,
      "loss": 0.7144,
      "step": 260
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.947492581587573e-05,
      "loss": 0.6746,
      "step": 261
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9469734985300373e-05,
      "loss": 0.778,
      "step": 262
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9464519321448988e-05,
      "loss": 0.7379,
      "step": 263
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9459278837999048e-05,
      "loss": 0.699,
      "step": 264
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9454013548693103e-05,
      "loss": 0.7559,
      "step": 265
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9448723467338765e-05,
      "loss": 0.6769,
      "step": 266
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.944340860780865e-05,
      "loss": 0.709,
      "step": 267
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9438068984040366e-05,
      "loss": 0.7177,
      "step": 268
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9432704610036448e-05,
      "loss": 0.7849,
      "step": 269
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9427315499864345e-05,
      "loss": 0.7355,
      "step": 270
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9421901667656364e-05,
      "loss": 0.686,
      "step": 271
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9416463127609655e-05,
      "loss": 0.6962,
      "step": 272
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9410999893986157e-05,
      "loss": 0.7009,
      "step": 273
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9405511981112553e-05,
      "loss": 0.6892,
      "step": 274
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9399999403380266e-05,
      "loss": 0.6725,
      "step": 275
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9394462175245382e-05,
      "loss": 0.75,
      "step": 276
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9388900311228636e-05,
      "loss": 0.7397,
      "step": 277
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9383313825915372e-05,
      "loss": 0.6471,
      "step": 278
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9377702733955493e-05,
      "loss": 0.7021,
      "step": 279
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.937206705006344e-05,
      "loss": 0.6346,
      "step": 280
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9366406789018127e-05,
      "loss": 0.7311,
      "step": 281
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9360721965662934e-05,
      "loss": 0.6929,
      "step": 282
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9355012594905645e-05,
      "loss": 0.705,
      "step": 283
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9349278691718426e-05,
      "loss": 0.7271,
      "step": 284
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9343520271137764e-05,
      "loss": 0.7556,
      "step": 285
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9337737348264448e-05,
      "loss": 0.6825,
      "step": 286
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9331929938263515e-05,
      "loss": 0.6625,
      "step": 287
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9326098056364224e-05,
      "loss": 0.6577,
      "step": 288
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9320241717860007e-05,
      "loss": 0.6815,
      "step": 289
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9314360938108427e-05,
      "loss": 0.7224,
      "step": 290
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.930845573253114e-05,
      "loss": 0.6794,
      "step": 291
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9302526116613863e-05,
      "loss": 0.7068,
      "step": 292
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9296572105906323e-05,
      "loss": 0.759,
      "step": 293
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9290593716022218e-05,
      "loss": 0.7061,
      "step": 294
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.928459096263918e-05,
      "loss": 0.7049,
      "step": 295
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9278563861498726e-05,
      "loss": 0.7267,
      "step": 296
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.927251242840623e-05,
      "loss": 0.7339,
      "step": 297
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9266436679230866e-05,
      "loss": 0.7487,
      "step": 298
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.926033662990558e-05,
      "loss": 0.7184,
      "step": 299
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9254212296427043e-05,
      "loss": 0.7435,
      "step": 300
    },
    {
      "epoch": 0.15,
      "eval_code_gate_load": [
        206.0,
        187.95,
        190.05,
        161.5,
        166.75,
        174.0,
        185.1,
        168.65
      ],
      "eval_code_loss": 0.5054687261581421,
      "eval_code_runtime": 1.7765,
      "eval_code_samples_per_second": 562.908,
      "eval_code_steps_per_second": 35.463,
      "step": 300
    },
    {
      "epoch": 0.15,
      "eval_orca_gate_load": [
        508.25,
        351.75,
        403.15,
        398.9,
        341.1,
        425.3,
        372.7,
        341.15
      ],
      "eval_orca_loss": 0.74560546875,
      "eval_orca_runtime": 1.9987,
      "eval_orca_samples_per_second": 500.33,
      "eval_orca_steps_per_second": 31.521,
      "step": 300
    },
    {
      "epoch": 0.15,
      "eval_math_gate_load": [
        324.1,
        225.55,
        232.0,
        231.1,
        244.15,
        235.65,
        281.2,
        258.95
      ],
      "eval_math_loss": 0.660449206829071,
      "eval_math_runtime": 1.8524,
      "eval_math_samples_per_second": 539.839,
      "eval_math_steps_per_second": 34.01,
      "step": 300
    },
    {
      "epoch": 0.15,
      "eval_sharegpt_gate_load": [
        1528.15,
        1114.7,
        1318.0,
        1217.15,
        1114.7,
        1348.95,
        1194.85,
        1075.1
      ],
      "eval_sharegpt_loss": 0.7177734375,
      "eval_sharegpt_runtime": 3.0081,
      "eval_sharegpt_samples_per_second": 332.432,
      "eval_sharegpt_steps_per_second": 20.943,
      "step": 300
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9248063694855603e-05,
      "loss": 0.7142,
      "step": 301
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.924189084131525e-05,
      "loss": 0.6985,
      "step": 302
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.923569375199357e-05,
      "loss": 0.7139,
      "step": 303
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.922947244314172e-05,
      "loss": 0.7499,
      "step": 304
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.922322693107434e-05,
      "loss": 0.6487,
      "step": 305
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9216957232169567e-05,
      "loss": 0.6963,
      "step": 306
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9210663362868956e-05,
      "loss": 0.6527,
      "step": 307
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9204345339677442e-05,
      "loss": 0.6664,
      "step": 308
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9198003179163308e-05,
      "loss": 0.7128,
      "step": 309
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9191636897958123e-05,
      "loss": 0.7186,
      "step": 310
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9185246512756727e-05,
      "loss": 0.6744,
      "step": 311
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9178832040317153e-05,
      "loss": 0.671,
      "step": 312
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.917239349746061e-05,
      "loss": 0.7735,
      "step": 313
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.916593090107143e-05,
      "loss": 0.6821,
      "step": 314
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9159444268097012e-05,
      "loss": 0.6755,
      "step": 315
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.91529336155478e-05,
      "loss": 0.6606,
      "step": 316
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9146398960497213e-05,
      "loss": 0.7263,
      "step": 317
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.913984032008163e-05,
      "loss": 0.6973,
      "step": 318
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9133257711500318e-05,
      "loss": 0.6823,
      "step": 319
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9126651152015404e-05,
      "loss": 0.6701,
      "step": 320
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9120020658951814e-05,
      "loss": 0.6863,
      "step": 321
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.911336624969725e-05,
      "loss": 0.7225,
      "step": 322
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.910668794170212e-05,
      "loss": 0.7344,
      "step": 323
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9099985752479505e-05,
      "loss": 0.743,
      "step": 324
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9093259699605125e-05,
      "loss": 0.7427,
      "step": 325
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.908650980071726e-05,
      "loss": 0.6911,
      "step": 326
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9079736073516735e-05,
      "loss": 0.6815,
      "step": 327
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9072938535766864e-05,
      "loss": 0.681,
      "step": 328
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9066117205293393e-05,
      "loss": 0.6892,
      "step": 329
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.905927209998447e-05,
      "loss": 0.7275,
      "step": 330
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.905240323779058e-05,
      "loss": 0.7578,
      "step": 331
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.904551063672452e-05,
      "loss": 0.6842,
      "step": 332
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9038594314861328e-05,
      "loss": 0.7046,
      "step": 333
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9031654290338256e-05,
      "loss": 0.686,
      "step": 334
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.90246905813547e-05,
      "loss": 0.6564,
      "step": 335
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9017703206172187e-05,
      "loss": 0.6683,
      "step": 336
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9010692183114285e-05,
      "loss": 0.7419,
      "step": 337
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.900365753056659e-05,
      "loss": 0.6519,
      "step": 338
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8996599266976658e-05,
      "loss": 0.6392,
      "step": 339
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8989517410853956e-05,
      "loss": 0.6843,
      "step": 340
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.898241198076983e-05,
      "loss": 0.6899,
      "step": 341
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8975282995357448e-05,
      "loss": 0.7023,
      "step": 342
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8968130473311732e-05,
      "loss": 0.6331,
      "step": 343
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.896095443338935e-05,
      "loss": 0.659,
      "step": 344
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8953754894408617e-05,
      "loss": 0.6294,
      "step": 345
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8946531875249496e-05,
      "loss": 0.6617,
      "step": 346
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.89392853948535e-05,
      "loss": 0.6872,
      "step": 347
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8932015472223692e-05,
      "loss": 0.7242,
      "step": 348
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.892472212642459e-05,
      "loss": 0.6774,
      "step": 349
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8917405376582144e-05,
      "loss": 0.6537,
      "step": 350
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.891006524188368e-05,
      "loss": 0.7253,
      "step": 351
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8902701741577844e-05,
      "loss": 0.6531,
      "step": 352
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.889531489497455e-05,
      "loss": 0.6655,
      "step": 353
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8887904721444955e-05,
      "loss": 0.7832,
      "step": 354
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8880471240421365e-05,
      "loss": 0.6488,
      "step": 355
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8873014471397225e-05,
      "loss": 0.7323,
      "step": 356
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8865534433927034e-05,
      "loss": 0.6627,
      "step": 357
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8858031147626326e-05,
      "loss": 0.6513,
      "step": 358
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.885050463217159e-05,
      "loss": 0.6807,
      "step": 359
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8842954907300236e-05,
      "loss": 0.6281,
      "step": 360
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.883538199281054e-05,
      "loss": 0.6593,
      "step": 361
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8827785908561585e-05,
      "loss": 0.6379,
      "step": 362
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8820166674473217e-05,
      "loss": 0.6934,
      "step": 363
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.881252431052599e-05,
      "loss": 0.6419,
      "step": 364
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.880485883676111e-05,
      "loss": 0.7072,
      "step": 365
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.879717027328039e-05,
      "loss": 0.7061,
      "step": 366
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8789458640246193e-05,
      "loss": 0.668,
      "step": 367
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8781723957881374e-05,
      "loss": 0.6409,
      "step": 368
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8773966246469238e-05,
      "loss": 0.6655,
      "step": 369
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.876618552635348e-05,
      "loss": 0.7031,
      "step": 370
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8758381817938126e-05,
      "loss": 0.6492,
      "step": 371
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.87505551416875e-05,
      "loss": 0.7515,
      "step": 372
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.874270551812614e-05,
      "loss": 0.6949,
      "step": 373
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8734832967838775e-05,
      "loss": 0.7355,
      "step": 374
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8726937511470247e-05,
      "loss": 0.6748,
      "step": 375
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.871901916972547e-05,
      "loss": 0.6932,
      "step": 376
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8711077963369377e-05,
      "loss": 0.7317,
      "step": 377
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8703113913226847e-05,
      "loss": 0.7287,
      "step": 378
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8695127040182678e-05,
      "loss": 0.7347,
      "step": 379
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8687117365181514e-05,
      "loss": 0.665,
      "step": 380
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.867908490922779e-05,
      "loss": 0.6309,
      "step": 381
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.867102969338569e-05,
      "loss": 0.6849,
      "step": 382
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8662951738779077e-05,
      "loss": 0.7277,
      "step": 383
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.865485106659145e-05,
      "loss": 0.7113,
      "step": 384
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8646727698065865e-05,
      "loss": 0.6555,
      "step": 385
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.863858165450492e-05,
      "loss": 0.6328,
      "step": 386
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.863041295727066e-05,
      "loss": 0.755,
      "step": 387
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.862222162778454e-05,
      "loss": 0.72,
      "step": 388
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8614007687527374e-05,
      "loss": 0.7177,
      "step": 389
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8605771158039253e-05,
      "loss": 0.7039,
      "step": 390
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8597512060919523e-05,
      "loss": 0.6861,
      "step": 391
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.85892304178267e-05,
      "loss": 0.672,
      "step": 392
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8580926250478425e-05,
      "loss": 0.6745,
      "step": 393
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8572599580651415e-05,
      "loss": 0.6572,
      "step": 394
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8564250430181387e-05,
      "loss": 0.6489,
      "step": 395
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8555878820963014e-05,
      "loss": 0.7114,
      "step": 396
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8547484774949865e-05,
      "loss": 0.7313,
      "step": 397
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8539068314154355e-05,
      "loss": 0.7268,
      "step": 398
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8530629460647658e-05,
      "loss": 0.698,
      "step": 399
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8522168236559693e-05,
      "loss": 0.7727,
      "step": 400
    },
    {
      "epoch": 0.2,
      "eval_code_gate_load": [
        202.85,
        191.5,
        178.8,
        158.95,
        174.1,
        173.95,
        189.7,
        170.15
      ],
      "eval_code_loss": 0.4867187440395355,
      "eval_code_runtime": 1.7854,
      "eval_code_samples_per_second": 560.106,
      "eval_code_steps_per_second": 35.287,
      "step": 400
    },
    {
      "epoch": 0.2,
      "eval_orca_gate_load": [
        501.6,
        354.85,
        398.45,
        395.2,
        353.6,
        425.0,
        373.85,
        339.75
      ],
      "eval_orca_loss": 0.735644519329071,
      "eval_orca_runtime": 2.0136,
      "eval_orca_samples_per_second": 496.616,
      "eval_orca_steps_per_second": 31.287,
      "step": 400
    },
    {
      "epoch": 0.2,
      "eval_math_gate_load": [
        320.85,
        237.0,
        216.35,
        231.35,
        252.05,
        238.55,
        277.05,
        259.5
      ],
      "eval_math_loss": 0.6509765386581421,
      "eval_math_runtime": 1.8609,
      "eval_math_samples_per_second": 537.377,
      "eval_math_steps_per_second": 33.855,
      "step": 400
    },
    {
      "epoch": 0.2,
      "eval_sharegpt_gate_load": [
        1520.85,
        1130.25,
        1282.2,
        1212.05,
        1141.8,
        1334.7,
        1205.7,
        1084.05
      ],
      "eval_sharegpt_loss": 0.712109386920929,
      "eval_sharegpt_runtime": 2.9955,
      "eval_sharegpt_samples_per_second": 333.837,
      "eval_sharegpt_steps_per_second": 21.032,
      "step": 400
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8513684664079033e-05,
      "loss": 0.736,
      "step": 401
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8505178765452853e-05,
      "loss": 0.7126,
      "step": 402
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8496650562986888e-05,
      "loss": 0.6468,
      "step": 403
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8488100079045345e-05,
      "loss": 0.76,
      "step": 404
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.847952733605088e-05,
      "loss": 0.641,
      "step": 405
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.847093235648451e-05,
      "loss": 0.656,
      "step": 406
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8462315162885563e-05,
      "loss": 0.7218,
      "step": 407
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8453675777851627e-05,
      "loss": 0.6604,
      "step": 408
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8445014224038485e-05,
      "loss": 0.7317,
      "step": 409
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8436330524160048e-05,
      "loss": 0.7086,
      "step": 410
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8427624700988308e-05,
      "loss": 0.7205,
      "step": 411
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8418896777353272e-05,
      "loss": 0.7507,
      "step": 412
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.84101467761429e-05,
      "loss": 0.6566,
      "step": 413
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8401374720303054e-05,
      "loss": 0.6488,
      "step": 414
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8392580632837423e-05,
      "loss": 0.6858,
      "step": 415
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8383764536807486e-05,
      "loss": 0.6794,
      "step": 416
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.837492645533241e-05,
      "loss": 0.7179,
      "step": 417
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.836606641158905e-05,
      "loss": 0.6611,
      "step": 418
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.835718442881183e-05,
      "loss": 0.6555,
      "step": 419
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8348280530292712e-05,
      "loss": 0.6509,
      "step": 420
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8339354739381138e-05,
      "loss": 0.6516,
      "step": 421
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.833040707948395e-05,
      "loss": 0.7115,
      "step": 422
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8321437574065347e-05,
      "loss": 0.6653,
      "step": 423
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.831244624664681e-05,
      "loss": 0.6984,
      "step": 424
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8303433120807043e-05,
      "loss": 0.6648,
      "step": 425
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.829439822018192e-05,
      "loss": 0.6358,
      "step": 426
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8285341568464416e-05,
      "loss": 0.7535,
      "step": 427
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.827626318940454e-05,
      "loss": 0.6195,
      "step": 428
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8267163106809288e-05,
      "loss": 0.6978,
      "step": 429
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8258041344542567e-05,
      "loss": 0.6611,
      "step": 430
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.824889792652513e-05,
      "loss": 0.6458,
      "step": 431
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8239732876734525e-05,
      "loss": 0.68,
      "step": 432
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8230546219205032e-05,
      "loss": 0.6779,
      "step": 433
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.822133797802758e-05,
      "loss": 0.719,
      "step": 434
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8212108177349722e-05,
      "loss": 0.6448,
      "step": 435
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8202856841375517e-05,
      "loss": 0.7534,
      "step": 436
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.819358399436553e-05,
      "loss": 0.6638,
      "step": 437
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8184289660636715e-05,
      "loss": 0.7415,
      "step": 438
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.817497386456238e-05,
      "loss": 0.6746,
      "step": 439
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.816563663057211e-05,
      "loss": 0.6855,
      "step": 440
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.815627798315172e-05,
      "loss": 0.7072,
      "step": 441
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8146897946843162e-05,
      "loss": 0.7095,
      "step": 442
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.81374965462445e-05,
      "loss": 0.6492,
      "step": 443
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.81280738060098e-05,
      "loss": 0.6364,
      "step": 444
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8118629750849106e-05,
      "loss": 0.7134,
      "step": 445
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.810916440552835e-05,
      "loss": 0.6878,
      "step": 446
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8099677794869297e-05,
      "loss": 0.6927,
      "step": 447
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.6148,
      "step": 448
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.808064087710212e-05,
      "loss": 0.6622,
      "step": 449
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8071090619916095e-05,
      "loss": 0.7079,
      "step": 450
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8061519197235835e-05,
      "loss": 0.6797,
      "step": 451
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8051926634161282e-05,
      "loss": 0.6875,
      "step": 452
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.804231295584782e-05,
      "loss": 0.6777,
      "step": 453
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8032678187506187e-05,
      "loss": 0.6696,
      "step": 454
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.802302235440245e-05,
      "loss": 0.7006,
      "step": 455
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8013345481857903e-05,
      "loss": 0.6464,
      "step": 456
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8003647595249016e-05,
      "loss": 0.7272,
      "step": 457
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.799392872000736e-05,
      "loss": 0.6675,
      "step": 458
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7984188881619563e-05,
      "loss": 0.6513,
      "step": 459
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.797442810562721e-05,
      "loss": 0.707,
      "step": 460
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.79646464176268e-05,
      "loss": 0.6863,
      "step": 461
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7954843843269665e-05,
      "loss": 0.6823,
      "step": 462
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.794502040826192e-05,
      "loss": 0.6794,
      "step": 463
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.793517613836437e-05,
      "loss": 0.6876,
      "step": 464
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7925311059392472e-05,
      "loss": 0.6972,
      "step": 465
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7915425197216246e-05,
      "loss": 0.6594,
      "step": 466
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7905518577760207e-05,
      "loss": 0.6748,
      "step": 467
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7895591227003316e-05,
      "loss": 0.7027,
      "step": 468
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.788564317097889e-05,
      "loss": 0.6533,
      "step": 469
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7875674435774546e-05,
      "loss": 0.7014,
      "step": 470
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.786568504753213e-05,
      "loss": 0.6529,
      "step": 471
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7855675032447648e-05,
      "loss": 0.6857,
      "step": 472
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.78456444167712e-05,
      "loss": 0.6052,
      "step": 473
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7835593226806902e-05,
      "loss": 0.6513,
      "step": 474
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7825521488912833e-05,
      "loss": 0.6751,
      "step": 475
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7815429229500946e-05,
      "loss": 0.6228,
      "step": 476
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7805316475037016e-05,
      "loss": 0.7381,
      "step": 477
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7795183252040568e-05,
      "loss": 0.7245,
      "step": 478
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7785029587084793e-05,
      "loss": 0.6885,
      "step": 479
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7774855506796497e-05,
      "loss": 0.7099,
      "step": 480
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7764661037856013e-05,
      "loss": 0.7003,
      "step": 481
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7754446206997152e-05,
      "loss": 0.6168,
      "step": 482
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.774421104100712e-05,
      "loss": 0.6982,
      "step": 483
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7733955566726438e-05,
      "loss": 0.705,
      "step": 484
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7723679811048904e-05,
      "loss": 0.6737,
      "step": 485
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.771338380092148e-05,
      "loss": 0.649,
      "step": 486
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7703067563344252e-05,
      "loss": 0.6895,
      "step": 487
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7692731125370355e-05,
      "loss": 0.724,
      "step": 488
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.768237451410589e-05,
      "loss": 0.6591,
      "step": 489
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.767199775670986e-05,
      "loss": 0.6185,
      "step": 490
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7661600880394113e-05,
      "loss": 0.634,
      "step": 491
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7651183912423228e-05,
      "loss": 0.7616,
      "step": 492
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7640746880114505e-05,
      "loss": 0.6675,
      "step": 493
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7630289810837836e-05,
      "loss": 0.6559,
      "step": 494
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7619812732015664e-05,
      "loss": 0.6797,
      "step": 495
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7609315671122912e-05,
      "loss": 0.6389,
      "step": 496
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.75987986556869e-05,
      "loss": 0.6548,
      "step": 497
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.758826171328727e-05,
      "loss": 0.641,
      "step": 498
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7577704871555924e-05,
      "loss": 0.6575,
      "step": 499
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7567128158176955e-05,
      "loss": 0.7441,
      "step": 500
    },
    {
      "epoch": 0.25,
      "eval_code_gate_load": [
        211.7,
        175.45,
        182.4,
        153.8,
        175.7,
        182.45,
        190.25,
        168.25
      ],
      "eval_code_loss": 0.48564451932907104,
      "eval_code_runtime": 1.7819,
      "eval_code_samples_per_second": 561.202,
      "eval_code_steps_per_second": 35.356,
      "step": 500
    },
    {
      "epoch": 0.25,
      "eval_orca_gate_load": [
        510.85,
        348.25,
        399.85,
        394.75,
        347.0,
        427.2,
        372.4,
        342.0
      ],
      "eval_orca_loss": 0.7276366949081421,
      "eval_orca_runtime": 2.0064,
      "eval_orca_samples_per_second": 498.405,
      "eval_orca_steps_per_second": 31.399,
      "step": 500
    },
    {
      "epoch": 0.25,
      "eval_math_gate_load": [
        323.9,
        211.65,
        228.5,
        240.05,
        256.55,
        243.35,
        275.65,
        253.05
      ],
      "eval_math_loss": 0.6341797113418579,
      "eval_math_runtime": 1.8482,
      "eval_math_samples_per_second": 541.069,
      "eval_math_steps_per_second": 34.087,
      "step": 500
    },
    {
      "epoch": 0.25,
      "eval_sharegpt_gate_load": [
        1535.15,
        1108.85,
        1287.05,
        1205.95,
        1139.5,
        1344.0,
        1206.15,
        1084.95
      ],
      "eval_sharegpt_loss": 0.702929675579071,
      "eval_sharegpt_runtime": 3.0148,
      "eval_sharegpt_samples_per_second": 331.693,
      "eval_sharegpt_steps_per_second": 20.897,
      "step": 500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7556531600886554e-05,
      "loss": 0.6686,
      "step": 501
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7545915227472967e-05,
      "loss": 0.6791,
      "step": 502
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.753527906577638e-05,
      "loss": 0.6089,
      "step": 503
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7524623143688905e-05,
      "loss": 0.6985,
      "step": 504
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7513947489154443e-05,
      "loss": 0.6266,
      "step": 505
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7503252130168657e-05,
      "loss": 0.653,
      "step": 506
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.749253709477888e-05,
      "loss": 0.6769,
      "step": 507
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.748180241108404e-05,
      "loss": 0.7099,
      "step": 508
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.74710481072346e-05,
      "loss": 0.6426,
      "step": 509
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7460274211432463e-05,
      "loss": 0.6726,
      "step": 510
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7449480751930915e-05,
      "loss": 0.6101,
      "step": 511
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7438667757034547e-05,
      "loss": 0.7545,
      "step": 512
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7427835255099173e-05,
      "loss": 0.6404,
      "step": 513
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7416983274531777e-05,
      "loss": 0.6997,
      "step": 514
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.74061118437904e-05,
      "loss": 0.636,
      "step": 515
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.739522099138411e-05,
      "loss": 0.6581,
      "step": 516
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7384310745872896e-05,
      "loss": 0.6526,
      "step": 517
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7373381135867605e-05,
      "loss": 0.6856,
      "step": 518
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7362432190029862e-05,
      "loss": 0.6896,
      "step": 519
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7351463937072008e-05,
      "loss": 0.6557,
      "step": 520
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7340476405757e-05,
      "loss": 0.5884,
      "step": 521
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.732946962489836e-05,
      "loss": 0.7313,
      "step": 522
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7318443623360092e-05,
      "loss": 0.6795,
      "step": 523
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7307398430056595e-05,
      "loss": 0.6634,
      "step": 524
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7296334073952606e-05,
      "loss": 0.6827,
      "step": 525
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.72852505840631e-05,
      "loss": 0.5858,
      "step": 526
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7274147989453246e-05,
      "loss": 0.6678,
      "step": 527
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.72630263192383e-05,
      "loss": 0.6171,
      "step": 528
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7251885602583547e-05,
      "loss": 0.7119,
      "step": 529
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7240725868704218e-05,
      "loss": 0.6961,
      "step": 530
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.722954714686541e-05,
      "loss": 0.6365,
      "step": 531
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7218349466382024e-05,
      "loss": 0.6798,
      "step": 532
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7207132856618668e-05,
      "loss": 0.6779,
      "step": 533
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.719589734698959e-05,
      "loss": 0.7134,
      "step": 534
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.718464296695861e-05,
      "loss": 0.6633,
      "step": 535
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7173369746039026e-05,
      "loss": 0.6756,
      "step": 536
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7162077713793547e-05,
      "loss": 0.6724,
      "step": 537
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7150766899834205e-05,
      "loss": 0.7333,
      "step": 538
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7139437333822303e-05,
      "loss": 0.719,
      "step": 539
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7128089045468294e-05,
      "loss": 0.6325,
      "step": 540
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.711672206453175e-05,
      "loss": 0.6804,
      "step": 541
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7105336420821247e-05,
      "loss": 0.6392,
      "step": 542
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.709393214419431e-05,
      "loss": 0.6337,
      "step": 543
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7082509264557333e-05,
      "loss": 0.6137,
      "step": 544
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.6187,
      "step": 545
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.705960781612262e-05,
      "loss": 0.6155,
      "step": 546
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7048129307381266e-05,
      "loss": 0.7153,
      "step": 547
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7036632315742464e-05,
      "loss": 0.6237,
      "step": 548
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7025116871355737e-05,
      "loss": 0.6675,
      "step": 549
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7013583004418994e-05,
      "loss": 0.6524,
      "step": 550
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7002030745178455e-05,
      "loss": 0.6712,
      "step": 551
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6990460123928577e-05,
      "loss": 0.7206,
      "step": 552
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6978871171011963e-05,
      "loss": 0.6446,
      "step": 553
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.696726391681929e-05,
      "loss": 0.7012,
      "step": 554
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.695563839178923e-05,
      "loss": 0.6295,
      "step": 555
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6943994626408365e-05,
      "loss": 0.6708,
      "step": 556
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6932332651211115e-05,
      "loss": 0.6714,
      "step": 557
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.692065249677965e-05,
      "loss": 0.6664,
      "step": 558
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6908954193743816e-05,
      "loss": 0.6241,
      "step": 559
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6897237772781046e-05,
      "loss": 0.5995,
      "step": 560
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6885503264616282e-05,
      "loss": 0.6553,
      "step": 561
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6873750700021917e-05,
      "loss": 0.6247,
      "step": 562
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.686198010981767e-05,
      "loss": 0.6658,
      "step": 563
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6850191524870548e-05,
      "loss": 0.6658,
      "step": 564
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6838384976094738e-05,
      "loss": 0.5976,
      "step": 565
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.682656049445154e-05,
      "loss": 0.6962,
      "step": 566
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6814718110949274e-05,
      "loss": 0.618,
      "step": 567
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6802857856643214e-05,
      "loss": 0.6911,
      "step": 568
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6790979762635497e-05,
      "loss": 0.6263,
      "step": 569
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6779083860075032e-05,
      "loss": 0.7031,
      "step": 570
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6767170180157442e-05,
      "loss": 0.6945,
      "step": 571
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6755238754124965e-05,
      "loss": 0.5995,
      "step": 572
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.674328961326637e-05,
      "loss": 0.6473,
      "step": 573
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6731322788916892e-05,
      "loss": 0.6429,
      "step": 574
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6719338312458123e-05,
      "loss": 0.6871,
      "step": 575
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6707336215317968e-05,
      "loss": 0.7102,
      "step": 576
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6695316528970517e-05,
      "loss": 0.6381,
      "step": 577
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6683279284936004e-05,
      "loss": 0.6506,
      "step": 578
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6671224514780692e-05,
      "loss": 0.6909,
      "step": 579
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.665915225011681e-05,
      "loss": 0.6609,
      "step": 580
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6647062522602474e-05,
      "loss": 0.6844,
      "step": 581
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6634955363941573e-05,
      "loss": 0.639,
      "step": 582
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.662283080588373e-05,
      "loss": 0.6377,
      "step": 583
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6610688880224178e-05,
      "loss": 0.6264,
      "step": 584
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.65985296188037e-05,
      "loss": 0.5948,
      "step": 585
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6586353053508548e-05,
      "loss": 0.6641,
      "step": 586
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.657415921627034e-05,
      "loss": 0.6331,
      "step": 587
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6561948139065997e-05,
      "loss": 0.6469,
      "step": 588
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.654971985391764e-05,
      "loss": 0.6639,
      "step": 589
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6537474392892527e-05,
      "loss": 0.6445,
      "step": 590
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6525211788102946e-05,
      "loss": 0.6451,
      "step": 591
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6512932071706153e-05,
      "loss": 0.6608,
      "step": 592
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6500635275904274e-05,
      "loss": 0.6746,
      "step": 593
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6488321432944218e-05,
      "loss": 0.6803,
      "step": 594
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6475990575117603e-05,
      "loss": 0.6298,
      "step": 595
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.646364273476067e-05,
      "loss": 0.6742,
      "step": 596
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6451277944254186e-05,
      "loss": 0.6271,
      "step": 597
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6438896236023374e-05,
      "loss": 0.6675,
      "step": 598
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6426497642537826e-05,
      "loss": 0.5843,
      "step": 599
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6414082196311402e-05,
      "loss": 0.6476,
      "step": 600
    },
    {
      "epoch": 0.3,
      "eval_code_gate_load": [
        205.25,
        186.05,
        177.25,
        157.05,
        177.4,
        178.15,
        188.5,
        170.35
      ],
      "eval_code_loss": 0.47382813692092896,
      "eval_code_runtime": 1.7893,
      "eval_code_samples_per_second": 558.879,
      "eval_code_steps_per_second": 35.209,
      "step": 600
    },
    {
      "epoch": 0.3,
      "eval_orca_gate_load": [
        503.95,
        352.45,
        398.2,
        397.5,
        349.4,
        424.3,
        374.4,
        342.1
      ],
      "eval_orca_loss": 0.7177734375,
      "eval_orca_runtime": 2.0043,
      "eval_orca_samples_per_second": 498.93,
      "eval_orca_steps_per_second": 31.433,
      "step": 600
    },
    {
      "epoch": 0.3,
      "eval_math_gate_load": [
        314.1,
        223.2,
        230.75,
        237.9,
        259.05,
        238.65,
        276.2,
        252.85
      ],
      "eval_math_loss": 0.45869141817092896,
      "eval_math_runtime": 1.8415,
      "eval_math_samples_per_second": 543.045,
      "eval_math_steps_per_second": 34.212,
      "step": 600
    },
    {
      "epoch": 0.3,
      "eval_sharegpt_gate_load": [
        1523.65,
        1118.3,
        1287.95,
        1209.65,
        1143.35,
        1350.6,
        1193.3,
        1084.8
      ],
      "eval_sharegpt_loss": 0.6988281011581421,
      "eval_sharegpt_runtime": 2.9965,
      "eval_sharegpt_samples_per_second": 333.722,
      "eval_sharegpt_steps_per_second": 21.024,
      "step": 600
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.640164992990216e-05,
      "loss": 0.6594,
      "step": 601
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.638920087591228e-05,
      "loss": 0.6526,
      "step": 602
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.637673506698794e-05,
      "loss": 0.6331,
      "step": 603
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6364252535819284e-05,
      "loss": 0.6393,
      "step": 604
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6351753315140285e-05,
      "loss": 0.6594,
      "step": 605
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.63392374377287e-05,
      "loss": 0.6047,
      "step": 606
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6326704936405953e-05,
      "loss": 0.6231,
      "step": 607
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6314155844037074e-05,
      "loss": 0.646,
      "step": 608
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6301590193530585e-05,
      "loss": 0.6644,
      "step": 609
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6289008017838447e-05,
      "loss": 0.6033,
      "step": 610
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6276409349955945e-05,
      "loss": 0.6628,
      "step": 611
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.626379422292162e-05,
      "loss": 0.6536,
      "step": 612
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6251162669817172e-05,
      "loss": 0.6094,
      "step": 613
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6238514723767372e-05,
      "loss": 0.6263,
      "step": 614
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.622585041793999e-05,
      "loss": 0.5747,
      "step": 615
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6213169785545688e-05,
      "loss": 0.5773,
      "step": 616
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6200472859837946e-05,
      "loss": 0.6433,
      "step": 617
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6187759674112972e-05,
      "loss": 0.63,
      "step": 618
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6175030261709615e-05,
      "loss": 0.6301,
      "step": 619
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6162284656009276e-05,
      "loss": 0.6516,
      "step": 620
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6149522890435815e-05,
      "loss": 0.6746,
      "step": 621
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6136744998455477e-05,
      "loss": 0.5876,
      "step": 622
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6123951013576796e-05,
      "loss": 0.6151,
      "step": 623
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6111140969350504e-05,
      "loss": 0.6078,
      "step": 624
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6098314899369446e-05,
      "loss": 0.564,
      "step": 625
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6085472837268504e-05,
      "loss": 0.6233,
      "step": 626
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.607261481672448e-05,
      "loss": 0.6358,
      "step": 627
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6059740871456035e-05,
      "loss": 0.6307,
      "step": 628
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6046851035223594e-05,
      "loss": 0.6252,
      "step": 629
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.603394534182925e-05,
      "loss": 0.6263,
      "step": 630
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6021023825116672e-05,
      "loss": 0.6303,
      "step": 631
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6008086518971037e-05,
      "loss": 0.6099,
      "step": 632
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.599513345731892e-05,
      "loss": 0.6102,
      "step": 633
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.598216467412822e-05,
      "loss": 0.6181,
      "step": 634
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5969180203408052e-05,
      "loss": 0.6137,
      "step": 635
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5956180079208684e-05,
      "loss": 0.6133,
      "step": 636
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5943164335621418e-05,
      "loss": 0.6424,
      "step": 637
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.593013300677853e-05,
      "loss": 0.6082,
      "step": 638
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.591708612685316e-05,
      "loss": 0.575,
      "step": 639
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5904023730059227e-05,
      "loss": 0.6683,
      "step": 640
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5890945850651347e-05,
      "loss": 0.6528,
      "step": 641
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 0.6372,
      "step": 642
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.586474378121511e-05,
      "loss": 0.6389,
      "step": 643
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5851619659898623e-05,
      "loss": 0.6056,
      "step": 644
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5838480193391753e-05,
      "loss": 0.5766,
      "step": 645
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.582532541615122e-05,
      "loss": 0.6306,
      "step": 646
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5812155362673895e-05,
      "loss": 0.6092,
      "step": 647
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.57989700674967e-05,
      "loss": 0.6193,
      "step": 648
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5785769565196543e-05,
      "loss": 0.693,
      "step": 649
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5772553890390196e-05,
      "loss": 0.6228,
      "step": 650
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5759323077734233e-05,
      "loss": 0.6016,
      "step": 651
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5746077161924905e-05,
      "loss": 0.6191,
      "step": 652
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5732816177698097e-05,
      "loss": 0.5908,
      "step": 653
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5719540159829185e-05,
      "loss": 0.6295,
      "step": 654
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5706249143132982e-05,
      "loss": 0.5966,
      "step": 655
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5692943162463628e-05,
      "loss": 0.571,
      "step": 656
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5679622252714507e-05,
      "loss": 0.6001,
      "step": 657
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5666286448818152e-05,
      "loss": 0.6675,
      "step": 658
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.565293578574615e-05,
      "loss": 0.5932,
      "step": 659
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5639570298509067e-05,
      "loss": 0.6454,
      "step": 660
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5626190022156328e-05,
      "loss": 0.6011,
      "step": 661
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5612794991776147e-05,
      "loss": 0.5868,
      "step": 662
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5599385242495437e-05,
      "loss": 0.5421,
      "step": 663
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5585960809479698e-05,
      "loss": 0.6044,
      "step": 664
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5572521727932937e-05,
      "loss": 0.5905,
      "step": 665
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5559068033097583e-05,
      "loss": 0.6102,
      "step": 666
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.554559976025438e-05,
      "loss": 0.5814,
      "step": 667
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5532116944722308e-05,
      "loss": 0.6065,
      "step": 668
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5518619621858474e-05,
      "loss": 0.5764,
      "step": 669
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5505107827058038e-05,
      "loss": 0.5827,
      "step": 670
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5491581595754102e-05,
      "loss": 0.5963,
      "step": 671
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.547804096341763e-05,
      "loss": 0.5861,
      "step": 672
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.546448596555736e-05,
      "loss": 0.6071,
      "step": 673
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5450916637719683e-05,
      "loss": 0.6061,
      "step": 674
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5437333015488586e-05,
      "loss": 0.6383,
      "step": 675
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5423735134485537e-05,
      "loss": 0.5513,
      "step": 676
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5410123030369387e-05,
      "loss": 0.6036,
      "step": 677
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5396496738836292e-05,
      "loss": 0.6438,
      "step": 678
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5382856295619622e-05,
      "loss": 0.6068,
      "step": 679
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.536920173648984e-05,
      "loss": 0.6384,
      "step": 680
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.535553309725444e-05,
      "loss": 0.5917,
      "step": 681
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5341850413757834e-05,
      "loss": 0.5726,
      "step": 682
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.532815372188126e-05,
      "loss": 0.5571,
      "step": 683
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5314443057542703e-05,
      "loss": 0.6066,
      "step": 684
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.530071845669678e-05,
      "loss": 0.5798,
      "step": 685
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5286979955334655e-05,
      "loss": 0.6412,
      "step": 686
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5273227589483945e-05,
      "loss": 0.5704,
      "step": 687
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5259461395208628e-05,
      "loss": 0.6293,
      "step": 688
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5245681408608946e-05,
      "loss": 0.541,
      "step": 689
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.52318876658213e-05,
      "loss": 0.6042,
      "step": 690
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5218080203018181e-05,
      "loss": 0.5809,
      "step": 691
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5204259056408046e-05,
      "loss": 0.5862,
      "step": 692
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5190424262235241e-05,
      "loss": 0.5623,
      "step": 693
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5176575856779904e-05,
      "loss": 0.5797,
      "step": 694
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.516271387635786e-05,
      "loss": 0.6085,
      "step": 695
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5148838357320537e-05,
      "loss": 0.5822,
      "step": 696
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5134949336054866e-05,
      "loss": 0.5458,
      "step": 697
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.512104684898319e-05,
      "loss": 0.6001,
      "step": 698
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5107130932563151e-05,
      "loss": 0.5995,
      "step": 699
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5093201623287631e-05,
      "loss": 0.6145,
      "step": 700
    },
    {
      "epoch": 0.35,
      "eval_code_gate_load": [
        209.05,
        177.55,
        174.45,
        152.3,
        174.4,
        187.35,
        191.25,
        173.65
      ],
      "eval_code_loss": 0.34501951932907104,
      "eval_code_runtime": 1.7766,
      "eval_code_samples_per_second": 562.88,
      "eval_code_steps_per_second": 35.461,
      "step": 700
    },
    {
      "epoch": 0.35,
      "eval_orca_gate_load": [
        509.4,
        346.8,
        397.0,
        390.9,
        346.15,
        430.65,
        368.2,
        353.2
      ],
      "eval_orca_loss": 0.4732421934604645,
      "eval_orca_runtime": 1.9952,
      "eval_orca_samples_per_second": 501.201,
      "eval_orca_steps_per_second": 31.576,
      "step": 700
    },
    {
      "epoch": 0.35,
      "eval_math_gate_load": [
        328.55,
        214.9,
        226.85,
        225.9,
        251.3,
        251.0,
        278.75,
        255.45
      ],
      "eval_math_loss": 0.45625001192092896,
      "eval_math_runtime": 1.8447,
      "eval_math_samples_per_second": 542.082,
      "eval_math_steps_per_second": 34.151,
      "step": 700
    },
    {
      "epoch": 0.35,
      "eval_sharegpt_gate_load": [
        1548.45,
        1089.65,
        1285.75,
        1191.1,
        1132.35,
        1374.15,
        1192.4,
        1097.75
      ],
      "eval_sharegpt_loss": 0.689746081829071,
      "eval_sharegpt_runtime": 2.9874,
      "eval_sharegpt_samples_per_second": 334.74,
      "eval_sharegpt_steps_per_second": 21.089,
      "step": 700
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.507925895768461e-05,
      "loss": 0.5587,
      "step": 701
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5065302972317108e-05,
      "loss": 0.5865,
      "step": 702
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5051333703783069e-05,
      "loss": 0.5186,
      "step": 703
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5037351188715265e-05,
      "loss": 0.6164,
      "step": 704
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5023355463781221e-05,
      "loss": 0.586,
      "step": 705
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5009346565683088e-05,
      "loss": 0.5101,
      "step": 706
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4995324531157569e-05,
      "loss": 0.5553,
      "step": 707
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4981289396975818e-05,
      "loss": 0.5443,
      "step": 708
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4967241199943332e-05,
      "loss": 0.589,
      "step": 709
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4953179976899878e-05,
      "loss": 0.6207,
      "step": 710
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4939105764719369e-05,
      "loss": 0.5548,
      "step": 711
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4925018600309784e-05,
      "loss": 0.5938,
      "step": 712
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4910918520613074e-05,
      "loss": 0.5599,
      "step": 713
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4896805562605052e-05,
      "loss": 0.5786,
      "step": 714
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4882679763295307e-05,
      "loss": 0.5052,
      "step": 715
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4868541159727097e-05,
      "loss": 0.5806,
      "step": 716
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4854389788977266e-05,
      "loss": 0.5824,
      "step": 717
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4840225688156132e-05,
      "loss": 0.5878,
      "step": 718
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4826048894407396e-05,
      "loss": 0.5256,
      "step": 719
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4811859444908053e-05,
      "loss": 0.5338,
      "step": 720
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4797657376868273e-05,
      "loss": 0.5115,
      "step": 721
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4783442727531328e-05,
      "loss": 0.5406,
      "step": 722
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4769215534173476e-05,
      "loss": 0.5402,
      "step": 723
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4754975834103877e-05,
      "loss": 0.5703,
      "step": 724
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4740723664664483e-05,
      "loss": 0.5609,
      "step": 725
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4726459063229946e-05,
      "loss": 0.5399,
      "step": 726
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4712182067207516e-05,
      "loss": 0.5649,
      "step": 727
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4697892714036959e-05,
      "loss": 0.5274,
      "step": 728
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4683591041190433e-05,
      "loss": 0.5253,
      "step": 729
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4669277086172406e-05,
      "loss": 0.4835,
      "step": 730
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4654950886519563e-05,
      "loss": 0.5794,
      "step": 731
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4640612479800686e-05,
      "loss": 0.521,
      "step": 732
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4626261903616579e-05,
      "loss": 0.5273,
      "step": 733
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4611899195599952e-05,
      "loss": 0.5263,
      "step": 734
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4597524393415336e-05,
      "loss": 0.5614,
      "step": 735
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4583137534758968e-05,
      "loss": 0.5693,
      "step": 736
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4568738657358715e-05,
      "loss": 0.5616,
      "step": 737
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.455432779897395e-05,
      "loss": 0.5403,
      "step": 738
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 0.5085,
      "step": 739
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4525470290445392e-05,
      "loss": 0.5168,
      "step": 740
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4511023715977048e-05,
      "loss": 0.587,
      "step": 741
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4496565311874902e-05,
      "loss": 0.5324,
      "step": 742
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4482095116054421e-05,
      "loss": 0.4944,
      "step": 743
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4467613166462024e-05,
      "loss": 0.5339,
      "step": 744
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4453119501074924e-05,
      "loss": 0.5517,
      "step": 745
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4438614157901073e-05,
      "loss": 0.5189,
      "step": 746
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4424097174979038e-05,
      "loss": 0.535,
      "step": 747
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4409568590377918e-05,
      "loss": 0.5303,
      "step": 748
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4395028442197231e-05,
      "loss": 0.5561,
      "step": 749
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4380476768566825e-05,
      "loss": 0.5305,
      "step": 750
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4365913607646762e-05,
      "loss": 0.496,
      "step": 751
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4351338997627233e-05,
      "loss": 0.5268,
      "step": 752
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.433675297672846e-05,
      "loss": 0.5,
      "step": 753
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4322155583200577e-05,
      "loss": 0.517,
      "step": 754
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4307546855323549e-05,
      "loss": 0.5244,
      "step": 755
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.429292683140706e-05,
      "loss": 0.4792,
      "step": 756
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4278295549790419e-05,
      "loss": 0.5154,
      "step": 757
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4263653048842461e-05,
      "loss": 0.4767,
      "step": 758
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.424899936696143e-05,
      "loss": 0.4858,
      "step": 759
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4234334542574906e-05,
      "loss": 0.4936,
      "step": 760
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4219658614139674e-05,
      "loss": 0.4957,
      "step": 761
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4204971620141648e-05,
      "loss": 0.5789,
      "step": 762
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4190273599095761e-05,
      "loss": 0.5306,
      "step": 763
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4175564589545853e-05,
      "loss": 0.4988,
      "step": 764
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4160844630064596e-05,
      "loss": 0.5283,
      "step": 765
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4146113759253362e-05,
      "loss": 0.5026,
      "step": 766
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4131372015742141e-05,
      "loss": 0.5237,
      "step": 767
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.411661943818944e-05,
      "loss": 0.5602,
      "step": 768
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4101856065282174e-05,
      "loss": 0.528,
      "step": 769
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4087081935735565e-05,
      "loss": 0.5042,
      "step": 770
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4072297088293043e-05,
      "loss": 0.4794,
      "step": 771
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4057501561726157e-05,
      "loss": 0.5531,
      "step": 772
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4042695394834435e-05,
      "loss": 0.4915,
      "step": 773
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4027878626445339e-05,
      "loss": 0.4861,
      "step": 774
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4013051295414108e-05,
      "loss": 0.4889,
      "step": 775
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3998213440623691e-05,
      "loss": 0.5035,
      "step": 776
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3983365100984633e-05,
      "loss": 0.5034,
      "step": 777
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3968506315434973e-05,
      "loss": 0.4847,
      "step": 778
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3953637122940147e-05,
      "loss": 0.504,
      "step": 779
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3938757562492873e-05,
      "loss": 0.483,
      "step": 780
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3923867673113067e-05,
      "loss": 0.5039,
      "step": 781
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.390896749384773e-05,
      "loss": 0.4807,
      "step": 782
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3894057063770841e-05,
      "loss": 0.513,
      "step": 783
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3879136421983265e-05,
      "loss": 0.4903,
      "step": 784
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3864205607612648e-05,
      "loss": 0.5104,
      "step": 785
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3849264659813314e-05,
      "loss": 0.4922,
      "step": 786
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3834313617766146e-05,
      "loss": 0.5198,
      "step": 787
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3819352520678519e-05,
      "loss": 0.4577,
      "step": 788
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.380438140778416e-05,
      "loss": 0.4697,
      "step": 789
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.378940031834307e-05,
      "loss": 0.4832,
      "step": 790
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3774409291641407e-05,
      "loss": 0.4664,
      "step": 791
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3759408366991391e-05,
      "loss": 0.492,
      "step": 792
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3744397583731204e-05,
      "loss": 0.496,
      "step": 793
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3729376981224869e-05,
      "loss": 0.498,
      "step": 794
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3714346598862168e-05,
      "loss": 0.5533,
      "step": 795
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3699306476058523e-05,
      "loss": 0.4858,
      "step": 796
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3684256652254906e-05,
      "loss": 0.511,
      "step": 797
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3669197166917723e-05,
      "loss": 0.5326,
      "step": 798
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.365412805953872e-05,
      "loss": 0.5154,
      "step": 799
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3639049369634878e-05,
      "loss": 0.5222,
      "step": 800
    },
    {
      "epoch": 0.4,
      "eval_code_gate_load": [
        212.75,
        176.15,
        177.3,
        159.4,
        177.95,
        178.35,
        186.4,
        171.7
      ],
      "eval_code_loss": 0.35273438692092896,
      "eval_code_runtime": 1.7754,
      "eval_code_samples_per_second": 563.24,
      "eval_code_steps_per_second": 35.484,
      "step": 800
    },
    {
      "epoch": 0.4,
      "eval_orca_gate_load": [
        506.35,
        346.1,
        403.4,
        400.65,
        347.1,
        420.75,
        365.05,
        352.9
      ],
      "eval_orca_loss": 0.45512694120407104,
      "eval_orca_runtime": 2.0061,
      "eval_orca_samples_per_second": 498.484,
      "eval_orca_steps_per_second": 31.405,
      "step": 800
    },
    {
      "epoch": 0.4,
      "eval_math_gate_load": [
        326.15,
        213.85,
        229.75,
        231.9,
        256.15,
        245.8,
        275.0,
        254.1
      ],
      "eval_math_loss": 0.4581542909145355,
      "eval_math_runtime": 1.8544,
      "eval_math_samples_per_second": 539.26,
      "eval_math_steps_per_second": 33.973,
      "step": 800
    },
    {
      "epoch": 0.4,
      "eval_sharegpt_gate_load": [
        1533.15,
        1088.05,
        1310.2,
        1211.85,
        1135.75,
        1350.35,
        1168.4,
        1113.85
      ],
      "eval_sharegpt_loss": 0.5342773199081421,
      "eval_sharegpt_runtime": 2.9927,
      "eval_sharegpt_samples_per_second": 334.144,
      "eval_sharegpt_steps_per_second": 21.051,
      "step": 800
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3623961136748296e-05,
      "loss": 0.5299,
      "step": 801
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3608863400446113e-05,
      "loss": 0.485,
      "step": 802
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3593756200320373e-05,
      "loss": 0.4949,
      "step": 803
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.357863957598796e-05,
      "loss": 0.5395,
      "step": 804
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.356351356709045e-05,
      "loss": 0.5138,
      "step": 805
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3548378213294042e-05,
      "loss": 0.5286,
      "step": 806
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3533233554289433e-05,
      "loss": 0.5311,
      "step": 807
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3518079629791725e-05,
      "loss": 0.5093,
      "step": 808
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3502916479540327e-05,
      "loss": 0.4513,
      "step": 809
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3487744143298822e-05,
      "loss": 0.4592,
      "step": 810
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3472562660854902e-05,
      "loss": 0.4641,
      "step": 811
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.345737207202023e-05,
      "loss": 0.5504,
      "step": 812
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3442172416630355e-05,
      "loss": 0.5057,
      "step": 813
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3426963734544601e-05,
      "loss": 0.4988,
      "step": 814
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3411746065645961e-05,
      "loss": 0.4449,
      "step": 815
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3396519449841006e-05,
      "loss": 0.4707,
      "step": 816
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3381283927059751e-05,
      "loss": 0.476,
      "step": 817
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3366039537255589e-05,
      "loss": 0.5192,
      "step": 818
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3350786320405145e-05,
      "loss": 0.494,
      "step": 819
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3335524316508208e-05,
      "loss": 0.4846,
      "step": 820
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3320253565587602e-05,
      "loss": 0.4849,
      "step": 821
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3304974107689088e-05,
      "loss": 0.5147,
      "step": 822
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.328968598288127e-05,
      "loss": 0.5149,
      "step": 823
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3274389231255466e-05,
      "loss": 0.5002,
      "step": 824
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3259083892925633e-05,
      "loss": 0.5237,
      "step": 825
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3243770008028225e-05,
      "loss": 0.4745,
      "step": 826
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3228447616722128e-05,
      "loss": 0.5389,
      "step": 827
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3213116759188525e-05,
      "loss": 0.4719,
      "step": 828
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.31977774756308e-05,
      "loss": 0.5096,
      "step": 829
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3182429806274442e-05,
      "loss": 0.5167,
      "step": 830
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3167073791366915e-05,
      "loss": 0.494,
      "step": 831
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3151709471177589e-05,
      "loss": 0.5393,
      "step": 832
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3136336885997591e-05,
      "loss": 0.5502,
      "step": 833
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3120956076139746e-05,
      "loss": 0.4955,
      "step": 834
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3105567081938423e-05,
      "loss": 0.5306,
      "step": 835
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.4821,
      "step": 836
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3074764701950095e-05,
      "loss": 0.4851,
      "step": 837
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.305935139693874e-05,
      "loss": 0.5209,
      "step": 838
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3043930069134998e-05,
      "loss": 0.4816,
      "step": 839
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3028500758979507e-05,
      "loss": 0.4333,
      "step": 840
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3013063506933838e-05,
      "loss": 0.472,
      "step": 841
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.299761835348038e-05,
      "loss": 0.4837,
      "step": 842
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2982165339122248e-05,
      "loss": 0.5189,
      "step": 843
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.296670450438317e-05,
      "loss": 0.5207,
      "step": 844
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2951235889807386e-05,
      "loss": 0.4656,
      "step": 845
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2935759535959528e-05,
      "loss": 0.537,
      "step": 846
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2920275483424538e-05,
      "loss": 0.5239,
      "step": 847
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2904783772807534e-05,
      "loss": 0.4885,
      "step": 848
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2889284444733722e-05,
      "loss": 0.5081,
      "step": 849
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2873777539848284e-05,
      "loss": 0.4915,
      "step": 850
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2858263098816265e-05,
      "loss": 0.4385,
      "step": 851
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2842741162322487e-05,
      "loss": 0.5104,
      "step": 852
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.282721177107141e-05,
      "loss": 0.5064,
      "step": 853
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2811674965787058e-05,
      "loss": 0.4651,
      "step": 854
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.279613078721289e-05,
      "loss": 0.5029,
      "step": 855
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2780579276111702e-05,
      "loss": 0.5042,
      "step": 856
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.276502047326552e-05,
      "loss": 0.5048,
      "step": 857
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2749454419475486e-05,
      "loss": 0.4466,
      "step": 858
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.273388115556177e-05,
      "loss": 0.5513,
      "step": 859
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2718300722363431e-05,
      "loss": 0.5231,
      "step": 860
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2702713160738344e-05,
      "loss": 0.4755,
      "step": 861
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2687118511563075e-05,
      "loss": 0.5436,
      "step": 862
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2671516815732767e-05,
      "loss": 0.4676,
      "step": 863
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2655908114161053e-05,
      "loss": 0.5156,
      "step": 864
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2640292447779932e-05,
      "loss": 0.4917,
      "step": 865
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2624669857539669e-05,
      "loss": 0.5152,
      "step": 866
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2609040384408685e-05,
      "loss": 0.4448,
      "step": 867
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2593404069373452e-05,
      "loss": 0.4678,
      "step": 868
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2577760953438382e-05,
      "loss": 0.4526,
      "step": 869
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2562111077625723e-05,
      "loss": 0.4869,
      "step": 870
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2546454482975454e-05,
      "loss": 0.4524,
      "step": 871
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2530791210545163e-05,
      "loss": 0.4979,
      "step": 872
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.251512130140996e-05,
      "loss": 0.4931,
      "step": 873
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2499444796662354e-05,
      "loss": 0.4952,
      "step": 874
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.248376173741215e-05,
      "loss": 0.476,
      "step": 875
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2468072164786342e-05,
      "loss": 0.5432,
      "step": 876
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2452376119929009e-05,
      "loss": 0.5156,
      "step": 877
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2436673644001196e-05,
      "loss": 0.4924,
      "step": 878
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2420964778180815e-05,
      "loss": 0.4994,
      "step": 879
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2405249563662539e-05,
      "loss": 0.4219,
      "step": 880
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2389528041657679e-05,
      "loss": 0.4999,
      "step": 881
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.23738002533941e-05,
      "loss": 0.5012,
      "step": 882
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2358066240116092e-05,
      "loss": 0.4539,
      "step": 883
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2342326043084268e-05,
      "loss": 0.4784,
      "step": 884
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2326579703575464e-05,
      "loss": 0.5036,
      "step": 885
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2310827262882614e-05,
      "loss": 0.5389,
      "step": 886
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2295068762314661e-05,
      "loss": 0.5221,
      "step": 887
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2279304243196438e-05,
      "loss": 0.4863,
      "step": 888
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2263533746868552e-05,
      "loss": 0.47,
      "step": 889
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2247757314687296e-05,
      "loss": 0.5165,
      "step": 890
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2231974988024522e-05,
      "loss": 0.4943,
      "step": 891
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2216186808267544e-05,
      "loss": 0.4758,
      "step": 892
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2200392816819022e-05,
      "loss": 0.4999,
      "step": 893
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2184593055096853e-05,
      "loss": 0.5106,
      "step": 894
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2168787564534078e-05,
      "loss": 0.476,
      "step": 895
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.215297638657875e-05,
      "loss": 0.5206,
      "step": 896
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2137159562693839e-05,
      "loss": 0.4682,
      "step": 897
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2121337134357121e-05,
      "loss": 0.5406,
      "step": 898
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2105509143061072e-05,
      "loss": 0.4805,
      "step": 899
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2089675630312755e-05,
      "loss": 0.5203,
      "step": 900
    },
    {
      "epoch": 0.45,
      "eval_code_gate_load": [
        205.5,
        177.4,
        183.4,
        157.6,
        179.75,
        179.4,
        190.55,
        166.4
      ],
      "eval_code_loss": 0.35786134004592896,
      "eval_code_runtime": 1.7808,
      "eval_code_samples_per_second": 561.533,
      "eval_code_steps_per_second": 35.377,
      "step": 900
    },
    {
      "epoch": 0.45,
      "eval_orca_gate_load": [
        499.1,
        348.55,
        403.15,
        395.45,
        351.2,
        424.25,
        368.4,
        352.2
      ],
      "eval_orca_loss": 0.46074217557907104,
      "eval_orca_runtime": 1.999,
      "eval_orca_samples_per_second": 500.239,
      "eval_orca_steps_per_second": 31.515,
      "step": 900
    },
    {
      "epoch": 0.45,
      "eval_math_gate_load": [
        316.75,
        225.15,
        231.5,
        227.4,
        256.35,
        248.25,
        277.4,
        249.9
      ],
      "eval_math_loss": 0.4561523497104645,
      "eval_math_runtime": 1.8388,
      "eval_math_samples_per_second": 543.841,
      "eval_math_steps_per_second": 34.262,
      "step": 900
    },
    {
      "epoch": 0.45,
      "eval_sharegpt_gate_load": [
        1524.45,
        1103.4,
        1313.4,
        1203.9,
        1138.05,
        1354.3,
        1174.25,
        1099.85
      ],
      "eval_sharegpt_loss": 0.5375000238418579,
      "eval_sharegpt_runtime": 2.9992,
      "eval_sharegpt_samples_per_second": 333.421,
      "eval_sharegpt_steps_per_second": 21.006,
      "step": 900
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2073836637633705e-05,
      "loss": 0.4337,
      "step": 901
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2057992206559837e-05,
      "loss": 0.4969,
      "step": 902
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.204214237864133e-05,
      "loss": 0.5045,
      "step": 903
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2026287195442503e-05,
      "loss": 0.4796,
      "step": 904
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2010426698541728e-05,
      "loss": 0.5411,
      "step": 905
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.199456092953131e-05,
      "loss": 0.5126,
      "step": 906
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.197868993001738e-05,
      "loss": 0.5036,
      "step": 907
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1962813741619777e-05,
      "loss": 0.4752,
      "step": 908
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.194693240597196e-05,
      "loss": 0.5218,
      "step": 909
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1931045964720882e-05,
      "loss": 0.4636,
      "step": 910
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1915154459526876e-05,
      "loss": 0.5427,
      "step": 911
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.189925793206357e-05,
      "loss": 0.5337,
      "step": 912
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.188335642401775e-05,
      "loss": 0.5242,
      "step": 913
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1867449977089264e-05,
      "loss": 0.4671,
      "step": 914
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1851538632990922e-05,
      "loss": 0.5247,
      "step": 915
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1835622433448361e-05,
      "loss": 0.4585,
      "step": 916
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.181970142019997e-05,
      "loss": 0.5068,
      "step": 917
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1803775634996735e-05,
      "loss": 0.5094,
      "step": 918
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1787845119602184e-05,
      "loss": 0.5033,
      "step": 919
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.177190991579223e-05,
      "loss": 0.5635,
      "step": 920
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1755970065355087e-05,
      "loss": 0.4865,
      "step": 921
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.174002561009116e-05,
      "loss": 0.5134,
      "step": 922
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1724076591812919e-05,
      "loss": 0.5193,
      "step": 923
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1708123052344803e-05,
      "loss": 0.4283,
      "step": 924
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1692165033523117e-05,
      "loss": 0.5168,
      "step": 925
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1676202577195901e-05,
      "loss": 0.5103,
      "step": 926
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1660235725222835e-05,
      "loss": 0.4611,
      "step": 927
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.164426451947513e-05,
      "loss": 0.5215,
      "step": 928
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1628289001835405e-05,
      "loss": 0.5178,
      "step": 929
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1612309214197599e-05,
      "loss": 0.4999,
      "step": 930
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1596325198466841e-05,
      "loss": 0.5103,
      "step": 931
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1580336996559343e-05,
      "loss": 0.4791,
      "step": 932
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.156434465040231e-05,
      "loss": 0.4663,
      "step": 933
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1548348201933799e-05,
      "loss": 0.5155,
      "step": 934
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1532347693102632e-05,
      "loss": 0.4827,
      "step": 935
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.151634316586828e-05,
      "loss": 0.4985,
      "step": 936
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.150033466220075e-05,
      "loss": 0.4664,
      "step": 937
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1484322224080474e-05,
      "loss": 0.4885,
      "step": 938
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1468305893498204e-05,
      "loss": 0.4893,
      "step": 939
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1452285712454905e-05,
      "loss": 0.5168,
      "step": 940
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1436261722961627e-05,
      "loss": 0.5049,
      "step": 941
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1420233967039423e-05,
      "loss": 0.464,
      "step": 942
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1404202486719205e-05,
      "loss": 0.4962,
      "step": 943
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.138816732404167e-05,
      "loss": 0.5134,
      "step": 944
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1372128521057155e-05,
      "loss": 0.5082,
      "step": 945
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1356086119825553e-05,
      "loss": 0.4526,
      "step": 946
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1340040162416197e-05,
      "loss": 0.5121,
      "step": 947
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1323990690907734e-05,
      "loss": 0.4691,
      "step": 948
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1307937747388034e-05,
      "loss": 0.4597,
      "step": 949
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1291881373954066e-05,
      "loss": 0.4535,
      "step": 950
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1275821612711803e-05,
      "loss": 0.4899,
      "step": 951
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1259758505776092e-05,
      "loss": 0.5067,
      "step": 952
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1243692095270565e-05,
      "loss": 0.463,
      "step": 953
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1227622423327501e-05,
      "loss": 0.5201,
      "step": 954
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1211549532087749e-05,
      "loss": 0.4925,
      "step": 955
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.119547346370059e-05,
      "loss": 0.5388,
      "step": 956
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1179394260323639e-05,
      "loss": 0.4614,
      "step": 957
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1163311964122733e-05,
      "loss": 0.4782,
      "step": 958
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.114722661727182e-05,
      "loss": 0.4644,
      "step": 959
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1131138261952845e-05,
      "loss": 0.4268,
      "step": 960
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1115046940355643e-05,
      "loss": 0.4545,
      "step": 961
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.109895269467783e-05,
      "loss": 0.4529,
      "step": 962
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1082855567124693e-05,
      "loss": 0.5314,
      "step": 963
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1066755599909065e-05,
      "loss": 0.4394,
      "step": 964
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.105065283525124e-05,
      "loss": 0.4943,
      "step": 965
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1034547315378838e-05,
      "loss": 0.4995,
      "step": 966
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.101843908252671e-05,
      "loss": 0.5093,
      "step": 967
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1002328178936813e-05,
      "loss": 0.4655,
      "step": 968
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0986214646858115e-05,
      "loss": 0.47,
      "step": 969
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0970098528546482e-05,
      "loss": 0.5542,
      "step": 970
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0953979866264549e-05,
      "loss": 0.4962,
      "step": 971
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0937858702281631e-05,
      "loss": 0.4945,
      "step": 972
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0921735078873599e-05,
      "loss": 0.4754,
      "step": 973
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.090560903832278e-05,
      "loss": 0.4328,
      "step": 974
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.088948062291783e-05,
      "loss": 0.4566,
      "step": 975
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.087334987495364e-05,
      "loss": 0.4559,
      "step": 976
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0857216836731221e-05,
      "loss": 0.5433,
      "step": 977
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0841081550557577e-05,
      "loss": 0.4789,
      "step": 978
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0824944058745623e-05,
      "loss": 0.4684,
      "step": 979
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0808804403614044e-05,
      "loss": 0.4446,
      "step": 980
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0792662627487207e-05,
      "loss": 0.4864,
      "step": 981
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0776518772695035e-05,
      "loss": 0.4218,
      "step": 982
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0760372881572904e-05,
      "loss": 0.4892,
      "step": 983
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0744224996461541e-05,
      "loss": 0.4257,
      "step": 984
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0728075159706881e-05,
      "loss": 0.5109,
      "step": 985
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0711923413659995e-05,
      "loss": 0.4799,
      "step": 986
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.069576980067695e-05,
      "loss": 0.4825,
      "step": 987
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0679614363118718e-05,
      "loss": 0.4711,
      "step": 988
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0663457143351044e-05,
      "loss": 0.528,
      "step": 989
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0647298183744359e-05,
      "loss": 0.4973,
      "step": 990
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0631137526673647e-05,
      "loss": 0.5144,
      "step": 991
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.061497521451835e-05,
      "loss": 0.5067,
      "step": 992
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0598811289662243e-05,
      "loss": 0.5282,
      "step": 993
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0582645794493337e-05,
      "loss": 0.4935,
      "step": 994
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0566478771403763e-05,
      "loss": 0.4667,
      "step": 995
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.055031026278965e-05,
      "loss": 0.4847,
      "step": 996
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0534140311051026e-05,
      "loss": 0.4582,
      "step": 997
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0517968958591705e-05,
      "loss": 0.502,
      "step": 998
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0501796247819176e-05,
      "loss": 0.4594,
      "step": 999
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0485622221144485e-05,
      "loss": 0.49,
      "step": 1000
    },
    {
      "epoch": 0.5,
      "eval_code_gate_load": [
        209.5,
        175.9,
        182.2,
        159.9,
        172.05,
        178.75,
        190.7,
        171.0
      ],
      "eval_code_loss": 0.35546875,
      "eval_code_runtime": 1.7833,
      "eval_code_samples_per_second": 560.749,
      "eval_code_steps_per_second": 35.327,
      "step": 1000
    },
    {
      "epoch": 0.5,
      "eval_orca_gate_load": [
        501.25,
        343.4,
        403.4,
        400.1,
        346.3,
        423.95,
        366.8,
        357.1
      ],
      "eval_orca_loss": 0.452880859375,
      "eval_orca_runtime": 2.0036,
      "eval_orca_samples_per_second": 499.113,
      "eval_orca_steps_per_second": 31.444,
      "step": 1000
    },
    {
      "epoch": 0.5,
      "eval_math_gate_load": [
        322.65,
        220.6,
        229.0,
        235.25,
        246.65,
        247.5,
        277.5,
        253.55
      ],
      "eval_math_loss": 0.4518066346645355,
      "eval_math_runtime": 1.847,
      "eval_math_samples_per_second": 541.417,
      "eval_math_steps_per_second": 34.109,
      "step": 1000
    },
    {
      "epoch": 0.5,
      "eval_sharegpt_gate_load": [
        1521.05,
        1084.45,
        1307.85,
        1203.0,
        1140.15,
        1363.05,
        1179.85,
        1112.2
      ],
      "eval_sharegpt_loss": 0.5404297113418579,
      "eval_sharegpt_runtime": 3.0138,
      "eval_sharegpt_samples_per_second": 331.805,
      "eval_sharegpt_steps_per_second": 20.904,
      "step": 1000
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.046944692098213e-05,
      "loss": 0.487,
      "step": 1001
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0453270389749956e-05,
      "loss": 0.4748,
      "step": 1002
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0437092669869025e-05,
      "loss": 0.4349,
      "step": 1003
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0420913803763522e-05,
      "loss": 0.4664,
      "step": 1004
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0404733833860639e-05,
      "loss": 0.4951,
      "step": 1005
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0388552802590461e-05,
      "loss": 0.4668,
      "step": 1006
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0372370752385854e-05,
      "loss": 0.4626,
      "step": 1007
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0356187725682359e-05,
      "loss": 0.4278,
      "step": 1008
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0340003764918078e-05,
      "loss": 0.5119,
      "step": 1009
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0323818912533561e-05,
      "loss": 0.4633,
      "step": 1010
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0307633210971697e-05,
      "loss": 0.461,
      "step": 1011
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0291446702677598e-05,
      "loss": 0.5217,
      "step": 1012
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0275259430098502e-05,
      "loss": 0.4573,
      "step": 1013
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0259071435683636e-05,
      "loss": 0.5034,
      "step": 1014
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0242882761884132e-05,
      "loss": 0.4356,
      "step": 1015
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.02266934511529e-05,
      "loss": 0.4805,
      "step": 1016
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0210503545944522e-05,
      "loss": 0.4934,
      "step": 1017
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0194313088715135e-05,
      "loss": 0.4913,
      "step": 1018
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0178122121922324e-05,
      "loss": 0.4323,
      "step": 1019
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0161930688025018e-05,
      "loss": 0.4759,
      "step": 1020
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0145738829483354e-05,
      "loss": 0.4667,
      "step": 1021
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0129546588758605e-05,
      "loss": 0.4452,
      "step": 1022
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0113354008313025e-05,
      "loss": 0.438,
      "step": 1023
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0097161130609774e-05,
      "loss": 0.4632,
      "step": 1024
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0080967998112787e-05,
      "loss": 0.4721,
      "step": 1025
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0064774653286662e-05,
      "loss": 0.4787,
      "step": 1026
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0048581138596563e-05,
      "loss": 0.479,
      "step": 1027
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.003238749650809e-05,
      "loss": 0.4629,
      "step": 1028
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.001619376948718e-05,
      "loss": 0.4641,
      "step": 1029
    },
    {
      "epoch": 0.52,
      "learning_rate": 1e-05,
      "loss": 0.4645,
      "step": 1030
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.98380623051282e-06,
      "loss": 0.4603,
      "step": 1031
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.967612503491915e-06,
      "loss": 0.4245,
      "step": 1032
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.95141886140344e-06,
      "loss": 0.5036,
      "step": 1033
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.935225346713341e-06,
      "loss": 0.4646,
      "step": 1034
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.919032001887215e-06,
      "loss": 0.4433,
      "step": 1035
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.90283886939023e-06,
      "loss": 0.4786,
      "step": 1036
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.886645991686977e-06,
      "loss": 0.5202,
      "step": 1037
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.870453411241399e-06,
      "loss": 0.4868,
      "step": 1038
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.854261170516648e-06,
      "loss": 0.4462,
      "step": 1039
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.838069311974986e-06,
      "loss": 0.4161,
      "step": 1040
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.821877878077678e-06,
      "loss": 0.4759,
      "step": 1041
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.805686911284867e-06,
      "loss": 0.5145,
      "step": 1042
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.789496454055482e-06,
      "loss": 0.4849,
      "step": 1043
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.773306548847102e-06,
      "loss": 0.4507,
      "step": 1044
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.757117238115871e-06,
      "loss": 0.4427,
      "step": 1045
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.740928564316369e-06,
      "loss": 0.4745,
      "step": 1046
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.724740569901503e-06,
      "loss": 0.4943,
      "step": 1047
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.708553297322407e-06,
      "loss": 0.4544,
      "step": 1048
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.692366789028308e-06,
      "loss": 0.4622,
      "step": 1049
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.676181087466444e-06,
      "loss": 0.4576,
      "step": 1050
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.659996235081926e-06,
      "loss": 0.4764,
      "step": 1051
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.643812274317644e-06,
      "loss": 0.5082,
      "step": 1052
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.627629247614151e-06,
      "loss": 0.4467,
      "step": 1053
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.611447197409544e-06,
      "loss": 0.4574,
      "step": 1054
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.595266166139366e-06,
      "loss": 0.4667,
      "step": 1055
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.579086196236483e-06,
      "loss": 0.4444,
      "step": 1056
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.562907330130981e-06,
      "loss": 0.4719,
      "step": 1057
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.54672961025005e-06,
      "loss": 0.4315,
      "step": 1058
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.530553079017872e-06,
      "loss": 0.5312,
      "step": 1059
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.514377778855521e-06,
      "loss": 0.4925,
      "step": 1060
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.498203752180827e-06,
      "loss": 0.4775,
      "step": 1061
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.482031041408296e-06,
      "loss": 0.4874,
      "step": 1062
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.465859688948977e-06,
      "loss": 0.5414,
      "step": 1063
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.449689737210352e-06,
      "loss": 0.479,
      "step": 1064
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.433521228596237e-06,
      "loss": 0.4297,
      "step": 1065
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.417354205506663e-06,
      "loss": 0.5173,
      "step": 1066
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.401188710337757e-06,
      "loss": 0.4412,
      "step": 1067
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.385024785481653e-06,
      "loss": 0.4954,
      "step": 1068
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.368862473326355e-06,
      "loss": 0.4635,
      "step": 1069
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.352701816255643e-06,
      "loss": 0.5658,
      "step": 1070
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.336542856648958e-06,
      "loss": 0.478,
      "step": 1071
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.320385636881283e-06,
      "loss": 0.4725,
      "step": 1072
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.30423019932305e-06,
      "loss": 0.5015,
      "step": 1073
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.288076586340005e-06,
      "loss": 0.5177,
      "step": 1074
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.27192484029312e-06,
      "loss": 0.4999,
      "step": 1075
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.255775003538462e-06,
      "loss": 0.4933,
      "step": 1076
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.239627118427098e-06,
      "loss": 0.4579,
      "step": 1077
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.22348122730497e-06,
      "loss": 0.4491,
      "step": 1078
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.207337372512797e-06,
      "loss": 0.4519,
      "step": 1079
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.19119559638596e-06,
      "loss": 0.4515,
      "step": 1080
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.17505594125438e-06,
      "loss": 0.4672,
      "step": 1081
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.158918449442425e-06,
      "loss": 0.4807,
      "step": 1082
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.142783163268782e-06,
      "loss": 0.5425,
      "step": 1083
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.126650125046361e-06,
      "loss": 0.4717,
      "step": 1084
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.110519377082174e-06,
      "loss": 0.5154,
      "step": 1085
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.094390961677223e-06,
      "loss": 0.4954,
      "step": 1086
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.078264921126405e-06,
      "loss": 0.43,
      "step": 1087
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.062141297718372e-06,
      "loss": 0.4305,
      "step": 1088
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.046020133735455e-06,
      "loss": 0.409,
      "step": 1089
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.02990147145352e-06,
      "loss": 0.4316,
      "step": 1090
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.013785353141887e-06,
      "loss": 0.4189,
      "step": 1091
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.99767182106319e-06,
      "loss": 0.5003,
      "step": 1092
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.981560917473292e-06,
      "loss": 0.4231,
      "step": 1093
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.965452684621164e-06,
      "loss": 0.4326,
      "step": 1094
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.949347164748761e-06,
      "loss": 0.4282,
      "step": 1095
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.933244400090937e-06,
      "loss": 0.4409,
      "step": 1096
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.91714443287531e-06,
      "loss": 0.5219,
      "step": 1097
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.901047305322172e-06,
      "loss": 0.4599,
      "step": 1098
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.88495305964436e-06,
      "loss": 0.4843,
      "step": 1099
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.868861738047158e-06,
      "loss": 0.4487,
      "step": 1100
    },
    {
      "epoch": 0.55,
      "eval_code_gate_load": [
        204.55,
        178.2,
        181.25,
        158.8,
        174.15,
        180.85,
        194.9,
        167.3
      ],
      "eval_code_loss": 0.3551269471645355,
      "eval_code_runtime": 1.8096,
      "eval_code_samples_per_second": 552.622,
      "eval_code_steps_per_second": 34.815,
      "step": 1100
    },
    {
      "epoch": 0.55,
      "eval_orca_gate_load": [
        497.9,
        347.05,
        405.0,
        400.25,
        349.7,
        422.7,
        367.5,
        352.2
      ],
      "eval_orca_loss": 0.45805662870407104,
      "eval_orca_runtime": 1.9956,
      "eval_orca_samples_per_second": 501.094,
      "eval_orca_steps_per_second": 31.569,
      "step": 1100
    },
    {
      "epoch": 0.55,
      "eval_math_gate_load": [
        309.9,
        222.05,
        237.3,
        231.3,
        253.5,
        249.15,
        277.95,
        251.55
      ],
      "eval_math_loss": 0.32890623807907104,
      "eval_math_runtime": 1.861,
      "eval_math_samples_per_second": 537.357,
      "eval_math_steps_per_second": 33.853,
      "step": 1100
    },
    {
      "epoch": 0.55,
      "eval_sharegpt_gate_load": [
        1510.0,
        1101.7,
        1304.1,
        1211.55,
        1140.2,
        1358.85,
        1183.65,
        1101.55
      ],
      "eval_sharegpt_loss": 0.539843738079071,
      "eval_sharegpt_runtime": 3.0143,
      "eval_sharegpt_samples_per_second": 331.753,
      "eval_sharegpt_steps_per_second": 20.9,
      "step": 1100
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.852773382728184e-06,
      "loss": 0.4398,
      "step": 1101
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.836688035877268e-06,
      "loss": 0.4278,
      "step": 1102
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.820605739676363e-06,
      "loss": 0.4476,
      "step": 1103
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.804526536299413e-06,
      "loss": 0.4136,
      "step": 1104
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.788450467912254e-06,
      "loss": 0.445,
      "step": 1105
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.772377576672502e-06,
      "loss": 0.4633,
      "step": 1106
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.75630790472944e-06,
      "loss": 0.4523,
      "step": 1107
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.740241494223911e-06,
      "loss": 0.4346,
      "step": 1108
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.724178387288202e-06,
      "loss": 0.4706,
      "step": 1109
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.708118626045939e-06,
      "loss": 0.4377,
      "step": 1110
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.692062252611973e-06,
      "loss": 0.4326,
      "step": 1111
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.676009309092273e-06,
      "loss": 0.4979,
      "step": 1112
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.659959837583808e-06,
      "loss": 0.4885,
      "step": 1113
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.643913880174449e-06,
      "loss": 0.424,
      "step": 1114
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.62787147894285e-06,
      "loss": 0.4862,
      "step": 1115
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.611832675958335e-06,
      "loss": 0.4256,
      "step": 1116
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.595797513280799e-06,
      "loss": 0.4172,
      "step": 1117
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.579766032960582e-06,
      "loss": 0.4389,
      "step": 1118
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.563738277038376e-06,
      "loss": 0.421,
      "step": 1119
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.5477142875451e-06,
      "loss": 0.4555,
      "step": 1120
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.531694106501796e-06,
      "loss": 0.4244,
      "step": 1121
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.515677775919528e-06,
      "loss": 0.4084,
      "step": 1122
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.499665337799254e-06,
      "loss": 0.4406,
      "step": 1123
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.48365683413172e-06,
      "loss": 0.4274,
      "step": 1124
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.46765230689737e-06,
      "loss": 0.3716,
      "step": 1125
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.451651798066203e-06,
      "loss": 0.459,
      "step": 1126
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.43565534959769e-06,
      "loss": 0.4536,
      "step": 1127
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.419663003440657e-06,
      "loss": 0.4558,
      "step": 1128
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.40367480153316e-06,
      "loss": 0.4123,
      "step": 1129
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.387690785802403e-06,
      "loss": 0.4408,
      "step": 1130
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.371710998164595e-06,
      "loss": 0.481,
      "step": 1131
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.355735480524874e-06,
      "loss": 0.3822,
      "step": 1132
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.339764274777165e-06,
      "loss": 0.4389,
      "step": 1133
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.3237974228041e-06,
      "loss": 0.403,
      "step": 1134
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.307834966476885e-06,
      "loss": 0.4676,
      "step": 1135
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.291876947655197e-06,
      "loss": 0.4541,
      "step": 1136
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.275923408187086e-06,
      "loss": 0.4605,
      "step": 1137
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.259974389908842e-06,
      "loss": 0.4276,
      "step": 1138
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.244029934644916e-06,
      "loss": 0.4232,
      "step": 1139
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.228090084207773e-06,
      "loss": 0.4284,
      "step": 1140
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.212154880397817e-06,
      "loss": 0.3999,
      "step": 1141
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.196224365003267e-06,
      "loss": 0.4149,
      "step": 1142
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.180298579800034e-06,
      "loss": 0.5071,
      "step": 1143
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.16437756655164e-06,
      "loss": 0.4423,
      "step": 1144
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.148461367009081e-06,
      "loss": 0.4525,
      "step": 1145
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.132550022910737e-06,
      "loss": 0.4204,
      "step": 1146
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.116643575982254e-06,
      "loss": 0.4082,
      "step": 1147
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.100742067936432e-06,
      "loss": 0.4174,
      "step": 1148
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.084845540473127e-06,
      "loss": 0.4162,
      "step": 1149
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.068954035279121e-06,
      "loss": 0.4527,
      "step": 1150
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.053067594028044e-06,
      "loss": 0.4525,
      "step": 1151
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.037186258380226e-06,
      "loss": 0.4433,
      "step": 1152
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.021310069982624e-06,
      "loss": 0.5178,
      "step": 1153
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.005439070468692e-06,
      "loss": 0.4018,
      "step": 1154
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.989573301458274e-06,
      "loss": 0.4235,
      "step": 1155
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.9737128045575e-06,
      "loss": 0.4846,
      "step": 1156
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.957857621358674e-06,
      "loss": 0.4281,
      "step": 1157
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.942007793440165e-06,
      "loss": 0.4087,
      "step": 1158
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.9261633623663e-06,
      "loss": 0.4426,
      "step": 1159
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.91032436968725e-06,
      "loss": 0.4638,
      "step": 1160
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.894490856938931e-06,
      "loss": 0.4776,
      "step": 1161
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.87866286564288e-06,
      "loss": 0.4447,
      "step": 1162
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.862840437306165e-06,
      "loss": 0.3879,
      "step": 1163
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.847023613421251e-06,
      "loss": 0.4486,
      "step": 1164
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.831212435465925e-06,
      "loss": 0.4983,
      "step": 1165
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.815406944903148e-06,
      "loss": 0.4229,
      "step": 1166
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.799607183180981e-06,
      "loss": 0.4204,
      "step": 1167
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.78381319173246e-06,
      "loss": 0.3881,
      "step": 1168
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.768025011975481e-06,
      "loss": 0.4176,
      "step": 1169
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.752242685312709e-06,
      "loss": 0.4191,
      "step": 1170
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.736466253131451e-06,
      "loss": 0.4608,
      "step": 1171
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.720695756803569e-06,
      "loss": 0.446,
      "step": 1172
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.704931237685342e-06,
      "loss": 0.3986,
      "step": 1173
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.689172737117389e-06,
      "loss": 0.438,
      "step": 1174
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.673420296424541e-06,
      "loss": 0.4122,
      "step": 1175
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.657673956915735e-06,
      "loss": 0.4458,
      "step": 1176
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.641933759883913e-06,
      "loss": 0.4381,
      "step": 1177
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.6261997466059035e-06,
      "loss": 0.4505,
      "step": 1178
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.610471958342326e-06,
      "loss": 0.4123,
      "step": 1179
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.594750436337467e-06,
      "loss": 0.4655,
      "step": 1180
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.579035221819188e-06,
      "loss": 0.4205,
      "step": 1181
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.5633263559988035e-06,
      "loss": 0.4072,
      "step": 1182
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.547623880070992e-06,
      "loss": 0.4442,
      "step": 1183
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.531927835213657e-06,
      "loss": 0.4164,
      "step": 1184
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.516238262587851e-06,
      "loss": 0.4247,
      "step": 1185
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.500555203337647e-06,
      "loss": 0.4238,
      "step": 1186
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.48487869859004e-06,
      "loss": 0.4493,
      "step": 1187
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.469208789454838e-06,
      "loss": 0.4132,
      "step": 1188
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.4535455170245476e-06,
      "loss": 0.4364,
      "step": 1189
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.4378889223742766e-06,
      "loss": 0.4417,
      "step": 1190
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.422239046561619e-06,
      "loss": 0.3977,
      "step": 1191
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.40659593062655e-06,
      "loss": 0.4288,
      "step": 1192
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.390959615591315e-06,
      "loss": 0.4328,
      "step": 1193
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.375330142460331e-06,
      "loss": 0.3894,
      "step": 1194
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.35970755222007e-06,
      "loss": 0.4406,
      "step": 1195
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.344091885838949e-06,
      "loss": 0.4512,
      "step": 1196
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.328483184267236e-06,
      "loss": 0.3937,
      "step": 1197
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.312881488436928e-06,
      "loss": 0.4402,
      "step": 1198
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.297286839261659e-06,
      "loss": 0.4219,
      "step": 1199
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.2816992776365714e-06,
      "loss": 0.4174,
      "step": 1200
    },
    {
      "epoch": 0.6,
      "eval_code_gate_load": [
        207.85,
        176.65,
        176.75,
        155.35,
        175.4,
        184.75,
        195.35,
        167.9
      ],
      "eval_code_loss": 0.26713865995407104,
      "eval_code_runtime": 1.7771,
      "eval_code_samples_per_second": 562.706,
      "eval_code_steps_per_second": 35.45,
      "step": 1200
    },
    {
      "epoch": 0.6,
      "eval_orca_gate_load": [
        500.65,
        345.65,
        405.4,
        401.1,
        345.55,
        420.1,
        369.5,
        354.35
      ],
      "eval_orca_loss": 0.4488769471645355,
      "eval_orca_runtime": 2.0075,
      "eval_orca_samples_per_second": 498.135,
      "eval_orca_steps_per_second": 31.383,
      "step": 1200
    },
    {
      "epoch": 0.6,
      "eval_math_gate_load": [
        324.85,
        216.9,
        232.55,
        225.15,
        251.7,
        253.15,
        278.7,
        249.7
      ],
      "eval_math_loss": 0.30810546875,
      "eval_math_runtime": 1.8431,
      "eval_math_samples_per_second": 542.566,
      "eval_math_steps_per_second": 34.182,
      "step": 1200
    },
    {
      "epoch": 0.6,
      "eval_sharegpt_gate_load": [
        1528.05,
        1087.15,
        1303.2,
        1211.5,
        1131.85,
        1353.75,
        1187.35,
        1108.75
      ],
      "eval_sharegpt_loss": 0.537792980670929,
      "eval_sharegpt_runtime": 2.9986,
      "eval_sharegpt_samples_per_second": 333.485,
      "eval_sharegpt_steps_per_second": 21.01,
      "step": 1200
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.2661188444382345e-06,
      "loss": 0.3947,
      "step": 1201
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.250545580524515e-06,
      "loss": 0.446,
      "step": 1202
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.234979526734482e-06,
      "loss": 0.4238,
      "step": 1203
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.219420723888301e-06,
      "loss": 0.4102,
      "step": 1204
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.203869212787112e-06,
      "loss": 0.4027,
      "step": 1205
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.188325034212944e-06,
      "loss": 0.3781,
      "step": 1206
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.1727882289285915e-06,
      "loss": 0.3997,
      "step": 1207
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.157258837677514e-06,
      "loss": 0.4436,
      "step": 1208
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.1417369011837355e-06,
      "loss": 0.4706,
      "step": 1209
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.126222460151719e-06,
      "loss": 0.4107,
      "step": 1210
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.110715555266281e-06,
      "loss": 0.3836,
      "step": 1211
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.095216227192467e-06,
      "loss": 0.4006,
      "step": 1212
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.0797245165754654e-06,
      "loss": 0.4064,
      "step": 1213
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.064240464040472e-06,
      "loss": 0.3926,
      "step": 1214
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.048764110192618e-06,
      "loss": 0.4111,
      "step": 1215
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.033295495616834e-06,
      "loss": 0.379,
      "step": 1216
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.017834660877756e-06,
      "loss": 0.3834,
      "step": 1217
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.002381646519625e-06,
      "loss": 0.4477,
      "step": 1218
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.986936493066165e-06,
      "loss": 0.4177,
      "step": 1219
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.971499241020495e-06,
      "loss": 0.4571,
      "step": 1220
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.956069930865005e-06,
      "loss": 0.4066,
      "step": 1221
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.940648603061263e-06,
      "loss": 0.4223,
      "step": 1222
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.925235298049906e-06,
      "loss": 0.4069,
      "step": 1223
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.3334,
      "step": 1224
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.894432918061579e-06,
      "loss": 0.404,
      "step": 1225
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.8790439238602576e-06,
      "loss": 0.403,
      "step": 1226
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.863663114002411e-06,
      "loss": 0.3739,
      "step": 1227
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.848290528822417e-06,
      "loss": 0.3936,
      "step": 1228
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.8329262086330864e-06,
      "loss": 0.4142,
      "step": 1229
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.8175701937255645e-06,
      "loss": 0.4611,
      "step": 1230
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.802222524369202e-06,
      "loss": 0.4569,
      "step": 1231
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.786883240811479e-06,
      "loss": 0.44,
      "step": 1232
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.771552383277875e-06,
      "loss": 0.3902,
      "step": 1233
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.756229991971779e-06,
      "loss": 0.467,
      "step": 1234
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.740916107074372e-06,
      "loss": 0.389,
      "step": 1235
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.725610768744535e-06,
      "loss": 0.4507,
      "step": 1236
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.710314017118734e-06,
      "loss": 0.4448,
      "step": 1237
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.695025892310913e-06,
      "loss": 0.4476,
      "step": 1238
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.6797464344124045e-06,
      "loss": 0.4785,
      "step": 1239
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.664475683491797e-06,
      "loss": 0.4106,
      "step": 1240
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.649213679594859e-06,
      "loss": 0.4091,
      "step": 1241
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.633960462744415e-06,
      "loss": 0.4373,
      "step": 1242
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.618716072940248e-06,
      "loss": 0.4227,
      "step": 1243
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.603480550158995e-06,
      "loss": 0.4241,
      "step": 1244
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.588253934354039e-06,
      "loss": 0.3771,
      "step": 1245
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.5730362654554015e-06,
      "loss": 0.4489,
      "step": 1246
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.5578275833696485e-06,
      "loss": 0.4633,
      "step": 1247
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.542627927979772e-06,
      "loss": 0.3883,
      "step": 1248
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.527437339145097e-06,
      "loss": 0.4436,
      "step": 1249
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.5122558567011775e-06,
      "loss": 0.4504,
      "step": 1250
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.497083520459674e-06,
      "loss": 0.3626,
      "step": 1251
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.481920370208274e-06,
      "loss": 0.4091,
      "step": 1252
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.466766445710568e-06,
      "loss": 0.3863,
      "step": 1253
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.4516217867059615e-06,
      "loss": 0.4081,
      "step": 1254
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.43648643290955e-06,
      "loss": 0.3932,
      "step": 1255
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.421360424012039e-06,
      "loss": 0.4043,
      "step": 1256
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.406243799679625e-06,
      "loss": 0.4142,
      "step": 1257
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.39113659955389e-06,
      "loss": 0.3747,
      "step": 1258
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.376038863251706e-06,
      "loss": 0.3748,
      "step": 1259
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.360950630365126e-06,
      "loss": 0.4162,
      "step": 1260
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.345871940461282e-06,
      "loss": 0.4242,
      "step": 1261
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.33080283308228e-06,
      "loss": 0.3819,
      "step": 1262
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.315743347745098e-06,
      "loss": 0.4009,
      "step": 1263
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.300693523941481e-06,
      "loss": 0.3622,
      "step": 1264
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.2856534011378365e-06,
      "loss": 0.4283,
      "step": 1265
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.270623018775135e-06,
      "loss": 0.4575,
      "step": 1266
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.255602416268799e-06,
      "loss": 0.4358,
      "step": 1267
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.2405916330086106e-06,
      "loss": 0.3554,
      "step": 1268
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.225590708358596e-06,
      "loss": 0.4037,
      "step": 1269
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.210599681656933e-06,
      "loss": 0.3967,
      "step": 1270
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.1956185922158445e-06,
      "loss": 0.4144,
      "step": 1271
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.180647479321484e-06,
      "loss": 0.4534,
      "step": 1272
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.165686382233856e-06,
      "loss": 0.4557,
      "step": 1273
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.150735340186689e-06,
      "loss": 0.3985,
      "step": 1274
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.135794392387353e-06,
      "loss": 0.4285,
      "step": 1275
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.120863578016736e-06,
      "loss": 0.3867,
      "step": 1276
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.1059429362291615e-06,
      "loss": 0.4173,
      "step": 1277
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.091032506152274e-06,
      "loss": 0.3591,
      "step": 1278
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.076132326886934e-06,
      "loss": 0.4235,
      "step": 1279
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.061242437507131e-06,
      "loss": 0.3599,
      "step": 1280
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.0463628770598574e-06,
      "loss": 0.3687,
      "step": 1281
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.0314936845650296e-06,
      "loss": 0.4174,
      "step": 1282
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.016634899015369e-06,
      "loss": 0.3781,
      "step": 1283
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.00178655937631e-06,
      "loss": 0.3717,
      "step": 1284
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.986948704585895e-06,
      "loss": 0.3987,
      "step": 1285
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.972121373554665e-06,
      "loss": 0.3913,
      "step": 1286
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.957304605165567e-06,
      "loss": 0.4595,
      "step": 1287
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.942498438273849e-06,
      "loss": 0.375,
      "step": 1288
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.927702911706961e-06,
      "loss": 0.4359,
      "step": 1289
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.912918064264441e-06,
      "loss": 0.3799,
      "step": 1290
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.898143934717831e-06,
      "loss": 0.4056,
      "step": 1291
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.8833805618105635e-06,
      "loss": 0.4361,
      "step": 1292
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.868627984257862e-06,
      "loss": 0.4293,
      "step": 1293
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.853886240746643e-06,
      "loss": 0.3594,
      "step": 1294
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.839155369935407e-06,
      "loss": 0.4376,
      "step": 1295
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.82443541045415e-06,
      "loss": 0.3927,
      "step": 1296
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.809726400904242e-06,
      "loss": 0.4547,
      "step": 1297
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.795028379858355e-06,
      "loss": 0.3932,
      "step": 1298
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.780341385860333e-06,
      "loss": 0.4543,
      "step": 1299
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.765665457425102e-06,
      "loss": 0.409,
      "step": 1300
    },
    {
      "epoch": 0.65,
      "eval_code_gate_load": [
        206.65,
        175.25,
        174.65,
        157.15,
        173.1,
        187.8,
        194.85,
        170.55
      ],
      "eval_code_loss": 0.2615722715854645,
      "eval_code_runtime": 1.7786,
      "eval_code_samples_per_second": 562.233,
      "eval_code_steps_per_second": 35.421,
      "step": 1300
    },
    {
      "epoch": 0.65,
      "eval_orca_gate_load": [
        499.85,
        346.7,
        403.75,
        402.4,
        350.05,
        418.15,
        370.1,
        351.3
      ],
      "eval_orca_loss": 0.4473632872104645,
      "eval_orca_runtime": 1.9968,
      "eval_orca_samples_per_second": 500.804,
      "eval_orca_steps_per_second": 31.551,
      "step": 1300
    },
    {
      "epoch": 0.65,
      "eval_math_gate_load": [
        312.9,
        224.8,
        227.45,
        234.4,
        250.9,
        253.95,
        276.6,
        251.7
      ],
      "eval_math_loss": 0.3106445372104645,
      "eval_math_runtime": 1.8419,
      "eval_math_samples_per_second": 542.921,
      "eval_math_steps_per_second": 34.204,
      "step": 1300
    },
    {
      "epoch": 0.65,
      "eval_sharegpt_gate_load": [
        1519.25,
        1096.15,
        1298.05,
        1219.5,
        1133.1,
        1354.6,
        1183.8,
        1107.15
      ],
      "eval_sharegpt_loss": 0.533984363079071,
      "eval_sharegpt_runtime": 2.9971,
      "eval_sharegpt_samples_per_second": 333.658,
      "eval_sharegpt_steps_per_second": 21.02,
      "step": 1300
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.751000633038573e-06,
      "loss": 0.4107,
      "step": 1301
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.736346951157544e-06,
      "loss": 0.3936,
      "step": 1302
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.721704450209581e-06,
      "loss": 0.3587,
      "step": 1303
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.707073168592943e-06,
      "loss": 0.3612,
      "step": 1304
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.692453144676451e-06,
      "loss": 0.3779,
      "step": 1305
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.677844416799424e-06,
      "loss": 0.4237,
      "step": 1306
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.663247023271543e-06,
      "loss": 0.4095,
      "step": 1307
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.648661002372769e-06,
      "loss": 0.3707,
      "step": 1308
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.63408639235324e-06,
      "loss": 0.4192,
      "step": 1309
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.619523231433177e-06,
      "loss": 0.4076,
      "step": 1310
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.604971557802769e-06,
      "loss": 0.3707,
      "step": 1311
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.590431409622081e-06,
      "loss": 0.3618,
      "step": 1312
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.575902825020962e-06,
      "loss": 0.4583,
      "step": 1313
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.56138584209893e-06,
      "loss": 0.3753,
      "step": 1314
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.546880498925079e-06,
      "loss": 0.4015,
      "step": 1315
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.5323868335379775e-06,
      "loss": 0.4733,
      "step": 1316
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.517904883945577e-06,
      "loss": 0.4065,
      "step": 1317
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.503434688125104e-06,
      "loss": 0.391,
      "step": 1318
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.488976284022953e-06,
      "loss": 0.413,
      "step": 1319
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.4745297095546125e-06,
      "loss": 0.4226,
      "step": 1320
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.460095002604533e-06,
      "loss": 0.4065,
      "step": 1321
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.445672201026054e-06,
      "loss": 0.3748,
      "step": 1322
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.431261342641287e-06,
      "loss": 0.3484,
      "step": 1323
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.416862465241033e-06,
      "loss": 0.3729,
      "step": 1324
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.40247560658467e-06,
      "loss": 0.4519,
      "step": 1325
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.3881008044000495e-06,
      "loss": 0.3754,
      "step": 1326
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.373738096383423e-06,
      "loss": 0.399,
      "step": 1327
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.359387520199317e-06,
      "loss": 0.3878,
      "step": 1328
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.3450491134804416e-06,
      "loss": 0.3375,
      "step": 1329
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.330722913827594e-06,
      "loss": 0.3478,
      "step": 1330
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.3164089588095705e-06,
      "loss": 0.4146,
      "step": 1331
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.302107285963045e-06,
      "loss": 0.4672,
      "step": 1332
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.287817932792485e-06,
      "loss": 0.389,
      "step": 1333
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.273540936770059e-06,
      "loss": 0.3968,
      "step": 1334
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.259276335335522e-06,
      "loss": 0.4544,
      "step": 1335
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.245024165896126e-06,
      "loss": 0.4088,
      "step": 1336
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.2307844658265236e-06,
      "loss": 0.3966,
      "step": 1337
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.216557272468675e-06,
      "loss": 0.3967,
      "step": 1338
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.202342623131731e-06,
      "loss": 0.4211,
      "step": 1339
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.18814055509195e-06,
      "loss": 0.3969,
      "step": 1340
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.173951105592605e-06,
      "loss": 0.3714,
      "step": 1341
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.1597743118438725e-06,
      "loss": 0.4332,
      "step": 1342
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.145610211022738e-06,
      "loss": 0.4368,
      "step": 1343
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.131458840272905e-06,
      "loss": 0.3699,
      "step": 1344
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.117320236704697e-06,
      "loss": 0.3549,
      "step": 1345
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.103194437394952e-06,
      "loss": 0.3848,
      "step": 1346
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.089081479386928e-06,
      "loss": 0.4111,
      "step": 1347
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.074981399690219e-06,
      "loss": 0.4432,
      "step": 1348
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.060894235280637e-06,
      "loss": 0.4486,
      "step": 1349
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.046820023100129e-06,
      "loss": 0.3844,
      "step": 1350
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.03275880005667e-06,
      "loss": 0.393,
      "step": 1351
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.018710603024187e-06,
      "loss": 0.3565,
      "step": 1352
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.004675468842436e-06,
      "loss": 0.3473,
      "step": 1353
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.990653434316915e-06,
      "loss": 0.4119,
      "step": 1354
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.976644536218783e-06,
      "loss": 0.3434,
      "step": 1355
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.9626488112847384e-06,
      "loss": 0.3702,
      "step": 1356
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.948666296216938e-06,
      "loss": 0.4028,
      "step": 1357
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.934697027682894e-06,
      "loss": 0.3527,
      "step": 1358
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.9207410423153925e-06,
      "loss": 0.3867,
      "step": 1359
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.9067983767123736e-06,
      "loss": 0.4232,
      "step": 1360
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.8928690674368495e-06,
      "loss": 0.4539,
      "step": 1361
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.878953151016816e-06,
      "loss": 0.4188,
      "step": 1362
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.8650506639451385e-06,
      "loss": 0.4357,
      "step": 1363
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.851161642679466e-06,
      "loss": 0.3801,
      "step": 1364
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.837286123642141e-06,
      "loss": 0.4286,
      "step": 1365
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.823424143220097e-06,
      "loss": 0.411,
      "step": 1366
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.809575737764759e-06,
      "loss": 0.4003,
      "step": 1367
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.795740943591955e-06,
      "loss": 0.4079,
      "step": 1368
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.781919796981818e-06,
      "loss": 0.4187,
      "step": 1369
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.7681123341787e-06,
      "loss": 0.4027,
      "step": 1370
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.754318591391057e-06,
      "loss": 0.3967,
      "step": 1371
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.740538604791371e-06,
      "loss": 0.3976,
      "step": 1372
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.726772410516055e-06,
      "loss": 0.4034,
      "step": 1373
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.713020044665348e-06,
      "loss": 0.4319,
      "step": 1374
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.699281543303222e-06,
      "loss": 0.4228,
      "step": 1375
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.685556942457296e-06,
      "loss": 0.3096,
      "step": 1376
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.67184627811874e-06,
      "loss": 0.4083,
      "step": 1377
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.65814958624217e-06,
      "loss": 0.3911,
      "step": 1378
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.6444669027455615e-06,
      "loss": 0.4363,
      "step": 1379
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.630798263510162e-06,
      "loss": 0.3926,
      "step": 1380
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.617143704380382e-06,
      "loss": 0.3413,
      "step": 1381
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.60350326116371e-06,
      "loss": 0.414,
      "step": 1382
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.589876969630616e-06,
      "loss": 0.3942,
      "step": 1383
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.576264865514467e-06,
      "loss": 0.4239,
      "step": 1384
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.562666984511416e-06,
      "loss": 0.4214,
      "step": 1385
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.549083362280318e-06,
      "loss": 0.3739,
      "step": 1386
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.535514034442644e-06,
      "loss": 0.372,
      "step": 1387
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.521959036582372e-06,
      "loss": 0.3932,
      "step": 1388
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.508418404245903e-06,
      "loss": 0.3606,
      "step": 1389
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.494892172941965e-06,
      "loss": 0.4025,
      "step": 1390
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.481380378141528e-06,
      "loss": 0.3945,
      "step": 1391
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.467883055277696e-06,
      "loss": 0.4146,
      "step": 1392
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.454400239745619e-06,
      "loss": 0.4055,
      "step": 1393
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.440931966902419e-06,
      "loss": 0.3951,
      "step": 1394
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.427478272067066e-06,
      "loss": 0.4238,
      "step": 1395
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.414039190520308e-06,
      "loss": 0.4598,
      "step": 1396
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.400614757504565e-06,
      "loss": 0.4238,
      "step": 1397
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.3872050082238535e-06,
      "loss": 0.3732,
      "step": 1398
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.373809977843676e-06,
      "loss": 0.4321,
      "step": 1399
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.360429701490935e-06,
      "loss": 0.4115,
      "step": 1400
    },
    {
      "epoch": 0.7,
      "eval_code_gate_load": [
        205.5,
        171.95,
        176.4,
        155.0,
        176.15,
        185.45,
        196.15,
        173.4
      ],
      "eval_code_loss": 0.2625488340854645,
      "eval_code_runtime": 1.7811,
      "eval_code_samples_per_second": 561.449,
      "eval_code_steps_per_second": 35.371,
      "step": 1400
    },
    {
      "epoch": 0.7,
      "eval_orca_gate_load": [
        499.7,
        344.5,
        406.5,
        399.4,
        350.35,
        416.7,
        369.6,
        355.55
      ],
      "eval_orca_loss": 0.39360350370407104,
      "eval_orca_runtime": 1.9989,
      "eval_orca_samples_per_second": 500.287,
      "eval_orca_steps_per_second": 31.518,
      "step": 1400
    },
    {
      "epoch": 0.7,
      "eval_math_gate_load": [
        315.6,
        219.15,
        229.9,
        225.75,
        253.55,
        256.25,
        275.3,
        257.2
      ],
      "eval_math_loss": 0.30537110567092896,
      "eval_math_runtime": 1.8428,
      "eval_math_samples_per_second": 542.647,
      "eval_math_steps_per_second": 34.187,
      "step": 1400
    },
    {
      "epoch": 0.7,
      "eval_sharegpt_gate_load": [
        1527.85,
        1080.7,
        1307.7,
        1213.2,
        1131.55,
        1358.05,
        1182.35,
        1110.2
      ],
      "eval_sharegpt_loss": 0.53271484375,
      "eval_sharegpt_runtime": 3.0058,
      "eval_sharegpt_samples_per_second": 332.695,
      "eval_sharegpt_steps_per_second": 20.96,
      "step": 1400
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.34706421425385e-06,
      "loss": 0.3626,
      "step": 1401
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.3337135511818514e-06,
      "loss": 0.3482,
      "step": 1402
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.320377747285497e-06,
      "loss": 0.3597,
      "step": 1403
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.307056837536373e-06,
      "loss": 0.4391,
      "step": 1404
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.2937508568670194e-06,
      "loss": 0.4752,
      "step": 1405
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.280459840170818e-06,
      "loss": 0.4315,
      "step": 1406
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.267183822301903e-06,
      "loss": 0.4123,
      "step": 1407
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.2539228380750955e-06,
      "loss": 0.3806,
      "step": 1408
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.240676922265774e-06,
      "loss": 0.3826,
      "step": 1409
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.2274461096098085e-06,
      "loss": 0.4211,
      "step": 1410
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.21423043480346e-06,
      "loss": 0.382,
      "step": 1411
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.201029932503303e-06,
      "loss": 0.415,
      "step": 1412
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.18784463732611e-06,
      "loss": 0.4246,
      "step": 1413
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.17467458384878e-06,
      "loss": 0.4311,
      "step": 1414
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.1615198066082475e-06,
      "loss": 0.4051,
      "step": 1415
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.14838034010138e-06,
      "loss": 0.3679,
      "step": 1416
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.135256218784896e-06,
      "loss": 0.3997,
      "step": 1417
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.12214747707527e-06,
      "loss": 0.4115,
      "step": 1418
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.1090541493486555e-06,
      "loss": 0.413,
      "step": 1419
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.095976269940777e-06,
      "loss": 0.3356,
      "step": 1420
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.082913873146842e-06,
      "loss": 0.3641,
      "step": 1421
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.069866993221473e-06,
      "loss": 0.3875,
      "step": 1422
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.056835664378585e-06,
      "loss": 0.3374,
      "step": 1423
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.043819920791322e-06,
      "loss": 0.3604,
      "step": 1424
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.03081979659195e-06,
      "loss": 0.3292,
      "step": 1425
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.017835325871781e-06,
      "loss": 0.3759,
      "step": 1426
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.004866542681079e-06,
      "loss": 0.4933,
      "step": 1427
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.991913481028965e-06,
      "loss": 0.343,
      "step": 1428
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.978976174883329e-06,
      "loss": 0.3699,
      "step": 1429
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.966054658170754e-06,
      "loss": 0.3808,
      "step": 1430
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.953148964776408e-06,
      "loss": 0.3943,
      "step": 1431
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.940259128543967e-06,
      "loss": 0.4061,
      "step": 1432
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.927385183275522e-06,
      "loss": 0.3791,
      "step": 1433
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.914527162731498e-06,
      "loss": 0.3934,
      "step": 1434
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.901685100630554e-06,
      "loss": 0.3913,
      "step": 1435
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.888859030649498e-06,
      "loss": 0.3934,
      "step": 1436
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.876048986423207e-06,
      "loss": 0.4283,
      "step": 1437
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.863255001544526e-06,
      "loss": 0.3086,
      "step": 1438
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.8504771095641905e-06,
      "loss": 0.358,
      "step": 1439
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.837715343990727e-06,
      "loss": 0.3836,
      "step": 1440
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.824969738290386e-06,
      "loss": 0.351,
      "step": 1441
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.81224032588703e-06,
      "loss": 0.3598,
      "step": 1442
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7995271401620548e-06,
      "loss": 0.4081,
      "step": 1443
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7868302144543146e-06,
      "loss": 0.3872,
      "step": 1444
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7741495820600128e-06,
      "loss": 0.3478,
      "step": 1445
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7614852762326303e-06,
      "loss": 0.4235,
      "step": 1446
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7488373301828296e-06,
      "loss": 0.3883,
      "step": 1447
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.736205777078381e-06,
      "loss": 0.3884,
      "step": 1448
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7235906500440576e-06,
      "loss": 0.4015,
      "step": 1449
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7109919821615546e-06,
      "loss": 0.3689,
      "step": 1450
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6984098064694174e-06,
      "loss": 0.3799,
      "step": 1451
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.685844155962931e-06,
      "loss": 0.3901,
      "step": 1452
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.673295063594049e-06,
      "loss": 0.4099,
      "step": 1453
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6607625622713005e-06,
      "loss": 0.3802,
      "step": 1454
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6482466848597164e-06,
      "loss": 0.3853,
      "step": 1455
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.63574746418072e-06,
      "loss": 0.3828,
      "step": 1456
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6232649330120608e-06,
      "loss": 0.3936,
      "step": 1457
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.610799124087725e-06,
      "loss": 0.3601,
      "step": 1458
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5983500700978425e-06,
      "loss": 0.3573,
      "step": 1459
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.585917803688603e-06,
      "loss": 0.4001,
      "step": 1460
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5735023574621765e-06,
      "loss": 0.3791,
      "step": 1461
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5611037639766267e-06,
      "loss": 0.3672,
      "step": 1462
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.548722055745818e-06,
      "loss": 0.4283,
      "step": 1463
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.536357265239333e-06,
      "loss": 0.3697,
      "step": 1464
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5240094248824e-06,
      "loss": 0.3795,
      "step": 1465
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.511678567055786e-06,
      "loss": 0.3783,
      "step": 1466
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4993647240957307e-06,
      "loss": 0.3738,
      "step": 1467
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.487067928293848e-06,
      "loss": 0.3223,
      "step": 1468
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4747882118970565e-06,
      "loss": 0.4047,
      "step": 1469
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4625256071074776e-06,
      "loss": 0.3838,
      "step": 1470
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.450280146082361e-06,
      "loss": 0.3525,
      "step": 1471
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4380518609340076e-06,
      "loss": 0.4043,
      "step": 1472
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4258407837296635e-06,
      "loss": 0.3922,
      "step": 1473
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.413646946491458e-06,
      "loss": 0.4094,
      "step": 1474
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4014703811963024e-06,
      "loss": 0.4139,
      "step": 1475
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3893111197758276e-06,
      "loss": 0.3647,
      "step": 1476
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3771691941162755e-06,
      "loss": 0.3618,
      "step": 1477
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3650446360584276e-06,
      "loss": 0.4106,
      "step": 1478
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.35293747739753e-06,
      "loss": 0.3881,
      "step": 1479
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3408477498831917e-06,
      "loss": 0.4383,
      "step": 1480
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3287754852193143e-06,
      "loss": 0.3771,
      "step": 1481
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.3167207150640003e-06,
      "loss": 0.3858,
      "step": 1482
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.304683471029485e-06,
      "loss": 0.3392,
      "step": 1483
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2926637846820366e-06,
      "loss": 0.4236,
      "step": 1484
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.280661687541876e-06,
      "loss": 0.3373,
      "step": 1485
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.268677211083109e-06,
      "loss": 0.424,
      "step": 1486
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.256710386733629e-06,
      "loss": 0.4352,
      "step": 1487
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2447612458750365e-06,
      "loss": 0.3318,
      "step": 1488
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2328298198425556e-06,
      "loss": 0.3844,
      "step": 1489
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2209161399249677e-06,
      "loss": 0.3987,
      "step": 1490
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.209020237364505e-06,
      "loss": 0.3508,
      "step": 1491
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.197142143356787e-06,
      "loss": 0.3412,
      "step": 1492
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1852818890507255e-06,
      "loss": 0.3601,
      "step": 1493
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1734395055484623e-06,
      "loss": 0.3728,
      "step": 1494
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1616150239052647e-06,
      "loss": 0.4014,
      "step": 1495
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1498084751294523e-06,
      "loss": 0.3344,
      "step": 1496
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1380198901823313e-06,
      "loss": 0.3488,
      "step": 1497
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.126249299978086e-06,
      "loss": 0.3459,
      "step": 1498
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1144967353837196e-06,
      "loss": 0.4079,
      "step": 1499
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1027622272189572e-06,
      "loss": 0.3557,
      "step": 1500
    },
    {
      "epoch": 0.75,
      "eval_code_gate_load": [
        207.05,
        170.35,
        182.3,
        156.9,
        179.05,
        184.35,
        193.1,
        166.9
      ],
      "eval_code_loss": 0.2623046934604645,
      "eval_code_runtime": 1.7777,
      "eval_code_samples_per_second": 562.517,
      "eval_code_steps_per_second": 35.439,
      "step": 1500
    },
    {
      "epoch": 0.75,
      "eval_orca_gate_load": [
        501.2,
        343.6,
        407.1,
        397.6,
        349.35,
        418.55,
        371.3,
        353.6
      ],
      "eval_orca_loss": 0.35224610567092896,
      "eval_orca_runtime": 2.0086,
      "eval_orca_samples_per_second": 497.848,
      "eval_orca_steps_per_second": 31.364,
      "step": 1500
    },
    {
      "epoch": 0.75,
      "eval_math_gate_load": [
        318.45,
        216.6,
        235.85,
        226.0,
        251.4,
        258.6,
        275.95,
        249.85
      ],
      "eval_math_loss": 0.3041015565395355,
      "eval_math_runtime": 1.8624,
      "eval_math_samples_per_second": 536.934,
      "eval_math_steps_per_second": 33.827,
      "step": 1500
    },
    {
      "epoch": 0.75,
      "eval_sharegpt_gate_load": [
        1534.6,
        1081.35,
        1305.0,
        1210.7,
        1133.0,
        1357.35,
        1191.1,
        1098.5
      ],
      "eval_sharegpt_loss": 0.530078113079071,
      "eval_sharegpt_runtime": 2.9848,
      "eval_sharegpt_samples_per_second": 335.033,
      "eval_sharegpt_steps_per_second": 21.107,
      "step": 1500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0910458062561865e-06,
      "loss": 0.3699,
      "step": 1501
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0793475032203513e-06,
      "loss": 0.4025,
      "step": 1502
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0676673487888854e-06,
      "loss": 0.3846,
      "step": 1503
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0560053735916372e-06,
      "loss": 0.3701,
      "step": 1504
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0443616082107753e-06,
      "loss": 0.4251,
      "step": 1505
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.032736083180716e-06,
      "loss": 0.3944,
      "step": 1506
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0211288289880404e-06,
      "loss": 0.3492,
      "step": 1507
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.009539876071427e-06,
      "loss": 0.3408,
      "step": 1508
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.997969254821548e-06,
      "loss": 0.3257,
      "step": 1509
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9864169955810085e-06,
      "loss": 0.3895,
      "step": 1510
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.974883128644266e-06,
      "loss": 0.367,
      "step": 1511
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9633676842575386e-06,
      "loss": 0.3883,
      "step": 1512
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.951870692618739e-06,
      "loss": 0.3662,
      "step": 1513
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.940392183877382e-06,
      "loss": 0.338,
      "step": 1514
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.3624,
      "step": 1515
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9174907354426696e-06,
      "loss": 0.3262,
      "step": 1516
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9060678558056876e-06,
      "loss": 0.3512,
      "step": 1517
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8946635791787546e-06,
      "loss": 0.4026,
      "step": 1518
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.883277935468254e-06,
      "loss": 0.3844,
      "step": 1519
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8719109545317102e-06,
      "loss": 0.3645,
      "step": 1520
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8605626661776995e-06,
      "loss": 0.3448,
      "step": 1521
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.849233100165795e-06,
      "loss": 0.405,
      "step": 1522
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.837922286206457e-06,
      "loss": 0.3771,
      "step": 1523
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8266302539609747e-06,
      "loss": 0.354,
      "step": 1524
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8153570330413925e-06,
      "loss": 0.3807,
      "step": 1525
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8041026530104144e-06,
      "loss": 0.3702,
      "step": 1526
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7928671433813392e-06,
      "loss": 0.3836,
      "step": 1527
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.78165053361798e-06,
      "loss": 0.3682,
      "step": 1528
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.770452853134593e-06,
      "loss": 0.3843,
      "step": 1529
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.759274131295787e-06,
      "loss": 0.4317,
      "step": 1530
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.7481143974164548e-06,
      "loss": 0.3398,
      "step": 1531
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.736973680761702e-06,
      "loss": 0.3638,
      "step": 1532
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.7258520105467566e-06,
      "loss": 0.3795,
      "step": 1533
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.714749415936904e-06,
      "loss": 0.3695,
      "step": 1534
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.7036659260473973e-06,
      "loss": 0.3323,
      "step": 1535
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.692601569943407e-06,
      "loss": 0.3591,
      "step": 1536
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6815563766399122e-06,
      "loss": 0.3609,
      "step": 1537
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.670530375101641e-06,
      "loss": 0.3286,
      "step": 1538
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6595235942430044e-06,
      "loss": 0.347,
      "step": 1539
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.648536062927999e-06,
      "loss": 0.3776,
      "step": 1540
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.637567809970143e-06,
      "loss": 0.469,
      "step": 1541
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6266188641324e-06,
      "loss": 0.3424,
      "step": 1542
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6156892541271083e-06,
      "loss": 0.384,
      "step": 1543
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.604779008615895e-06,
      "loss": 0.3755,
      "step": 1544
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.593888156209603e-06,
      "loss": 0.37,
      "step": 1545
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.583016725468226e-06,
      "loss": 0.3906,
      "step": 1546
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.572164744900827e-06,
      "loss": 0.3445,
      "step": 1547
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5613322429654573e-06,
      "loss": 0.3819,
      "step": 1548
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5505192480690865e-06,
      "loss": 0.305,
      "step": 1549
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.5397257885675396e-06,
      "loss": 0.2964,
      "step": 1550
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.528951892765402e-06,
      "loss": 0.3686,
      "step": 1551
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.5181975889159615e-06,
      "loss": 0.3582,
      "step": 1552
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.507462905221122e-06,
      "loss": 0.3419,
      "step": 1553
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.496747869831345e-06,
      "loss": 0.3447,
      "step": 1554
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.48605251084556e-06,
      "loss": 0.3513,
      "step": 1555
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.475376856311097e-06,
      "loss": 0.3515,
      "step": 1556
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.464720934223619e-06,
      "loss": 0.347,
      "step": 1557
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4540847725270376e-06,
      "loss": 0.3591,
      "step": 1558
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4434683991134476e-06,
      "loss": 0.3858,
      "step": 1559
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.432871841823047e-06,
      "loss": 0.3432,
      "step": 1560
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4222951284440776e-06,
      "loss": 0.3379,
      "step": 1561
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.411738286712735e-06,
      "loss": 0.3651,
      "step": 1562
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.401201344313102e-06,
      "loss": 0.411,
      "step": 1563
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.390684328877089e-06,
      "loss": 0.3929,
      "step": 1564
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3801872679843384e-06,
      "loss": 0.335,
      "step": 1565
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.36971018916217e-06,
      "loss": 0.347,
      "step": 1566
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3592531198854974e-06,
      "loss": 0.36,
      "step": 1567
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3488160875767717e-06,
      "loss": 0.3374,
      "step": 1568
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3383991196058918e-06,
      "loss": 0.287,
      "step": 1569
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.328002243290138e-06,
      "loss": 0.3322,
      "step": 1570
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.317625485894113e-06,
      "loss": 0.324,
      "step": 1571
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.307268874629649e-06,
      "loss": 0.3286,
      "step": 1572
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.296932436655752e-06,
      "loss": 0.3224,
      "step": 1573
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2866161990785228e-06,
      "loss": 0.3541,
      "step": 1574
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2763201889510987e-06,
      "loss": 0.3138,
      "step": 1575
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.266044433273562e-06,
      "loss": 0.3317,
      "step": 1576
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2557889589928815e-06,
      "loss": 0.3571,
      "step": 1577
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.245553793002849e-06,
      "loss": 0.3136,
      "step": 1578
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.23533896214399e-06,
      "loss": 0.3203,
      "step": 1579
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2251444932035094e-06,
      "loss": 0.3453,
      "step": 1580
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2149704129152083e-06,
      "loss": 0.4132,
      "step": 1581
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.204816747959434e-06,
      "loss": 0.368,
      "step": 1582
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.194683524962986e-06,
      "loss": 0.3031,
      "step": 1583
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.184570770499056e-06,
      "loss": 0.3408,
      "step": 1584
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1744785110871713e-06,
      "loss": 0.3161,
      "step": 1585
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1644067731931005e-06,
      "loss": 0.2998,
      "step": 1586
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1543555832288056e-06,
      "loss": 0.3601,
      "step": 1587
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1443249675523536e-06,
      "loss": 0.343,
      "step": 1588
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.134314952467873e-06,
      "loss": 0.306,
      "step": 1589
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.124325564225458e-06,
      "loss": 0.3214,
      "step": 1590
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.1143568290211115e-06,
      "loss": 0.3413,
      "step": 1591
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.1044087729966856e-06,
      "loss": 0.3369,
      "step": 1592
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0944814222397948e-06,
      "loss": 0.3431,
      "step": 1593
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0845748027837585e-06,
      "loss": 0.2723,
      "step": 1594
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.074688940607529e-06,
      "loss": 0.326,
      "step": 1595
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.064823861635633e-06,
      "loss": 0.2964,
      "step": 1596
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0549795917380867e-06,
      "loss": 0.3481,
      "step": 1597
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0451561567303378e-06,
      "loss": 0.3613,
      "step": 1598
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0353535823732053e-06,
      "loss": 0.3455,
      "step": 1599
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.025571894372794e-06,
      "loss": 0.2833,
      "step": 1600
    },
    {
      "epoch": 0.8,
      "eval_code_gate_load": [
        206.95,
        170.4,
        179.9,
        153.6,
        179.0,
        186.85,
        195.1,
        168.2
      ],
      "eval_code_loss": 0.261962890625,
      "eval_code_runtime": 1.7896,
      "eval_code_samples_per_second": 558.77,
      "eval_code_steps_per_second": 35.203,
      "step": 1600
    },
    {
      "epoch": 0.8,
      "eval_orca_gate_load": [
        498.0,
        341.05,
        410.9,
        396.4,
        351.45,
        419.25,
        369.95,
        355.3
      ],
      "eval_orca_loss": 0.3511718809604645,
      "eval_orca_runtime": 2.0018,
      "eval_orca_samples_per_second": 499.561,
      "eval_orca_steps_per_second": 31.472,
      "step": 1600
    },
    {
      "epoch": 0.8,
      "eval_math_gate_load": [
        314.8,
        213.8,
        239.55,
        221.0,
        254.25,
        260.1,
        278.1,
        251.1
      ],
      "eval_math_loss": 0.24335937201976776,
      "eval_math_runtime": 1.8471,
      "eval_math_samples_per_second": 541.403,
      "eval_math_steps_per_second": 34.108,
      "step": 1600
    },
    {
      "epoch": 0.8,
      "eval_sharegpt_gate_load": [
        1525.0,
        1074.95,
        1316.25,
        1205.0,
        1138.7,
        1360.4,
        1181.75,
        1109.55
      ],
      "eval_sharegpt_loss": 0.48750001192092896,
      "eval_sharegpt_runtime": 2.9974,
      "eval_sharegpt_samples_per_second": 333.624,
      "eval_sharegpt_steps_per_second": 21.018,
      "step": 1600
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0158111183804407e-06,
      "loss": 0.3204,
      "step": 1601
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0060712799926407e-06,
      "loss": 0.2845,
      "step": 1602
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9963524047509898e-06,
      "loss": 0.3192,
      "step": 1603
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9866545181421016e-06,
      "loss": 0.321,
      "step": 1604
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.976977645597552e-06,
      "loss": 0.3466,
      "step": 1605
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.967321812493813e-06,
      "loss": 0.341,
      "step": 1606
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9576870441521834e-06,
      "loss": 0.2962,
      "step": 1607
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9480733658387175e-06,
      "loss": 0.3237,
      "step": 1608
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9384808027641666e-06,
      "loss": 0.3164,
      "step": 1609
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9289093800839067e-06,
      "loss": 0.3583,
      "step": 1610
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9193591228978815e-06,
      "loss": 0.2978,
      "step": 1611
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.2947,
      "step": 1612
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9003222051307046e-06,
      "loss": 0.2788,
      "step": 1613
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8908355944716516e-06,
      "loss": 0.3126,
      "step": 1614
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8813702491508956e-06,
      "loss": 0.3048,
      "step": 1615
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8719261939902023e-06,
      "loss": 0.2466,
      "step": 1616
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.862503453755502e-06,
      "loss": 0.3255,
      "step": 1617
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8531020531568377e-06,
      "loss": 0.3581,
      "step": 1618
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8437220168482839e-06,
      "loss": 0.3437,
      "step": 1619
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8343633694278895e-06,
      "loss": 0.3263,
      "step": 1620
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.825026135437622e-06,
      "loss": 0.323,
      "step": 1621
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8157103393632869e-06,
      "loss": 0.2819,
      "step": 1622
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8064160056344714e-06,
      "loss": 0.2967,
      "step": 1623
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7971431586244814e-06,
      "loss": 0.3171,
      "step": 1624
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7878918226502816e-06,
      "loss": 0.3275,
      "step": 1625
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7786620219724205e-06,
      "loss": 0.3351,
      "step": 1626
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7694537807949707e-06,
      "loss": 0.2992,
      "step": 1627
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7602671232654755e-06,
      "loss": 0.3353,
      "step": 1628
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.751102073474873e-06,
      "loss": 0.3463,
      "step": 1629
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7419586554574364e-06,
      "loss": 0.3037,
      "step": 1630
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7328368931907114e-06,
      "loss": 0.2644,
      "step": 1631
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.723736810595461e-06,
      "loss": 0.2871,
      "step": 1632
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7146584315355886e-06,
      "loss": 0.3261,
      "step": 1633
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7056017798180824e-06,
      "loss": 0.3192,
      "step": 1634
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.69656687919296e-06,
      "loss": 0.3082,
      "step": 1635
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.687553753353195e-06,
      "loss": 0.3397,
      "step": 1636
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6785624259346556e-06,
      "loss": 0.3451,
      "step": 1637
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.669592920516049e-06,
      "loss": 0.3263,
      "step": 1638
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.660645260618864e-06,
      "loss": 0.2625,
      "step": 1639
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6517194697072903e-06,
      "loss": 0.2828,
      "step": 1640
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6428155711881722e-06,
      "loss": 0.3153,
      "step": 1641
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.633933588410952e-06,
      "loss": 0.3264,
      "step": 1642
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6250735446675914e-06,
      "loss": 0.2978,
      "step": 1643
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6162354631925203e-06,
      "loss": 0.3242,
      "step": 1644
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.607419367162577e-06,
      "loss": 0.3276,
      "step": 1645
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5986252796969482e-06,
      "loss": 0.3074,
      "step": 1646
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.589853223857103e-06,
      "loss": 0.3643,
      "step": 1647
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5811032226467304e-06,
      "loss": 0.3557,
      "step": 1648
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5723752990116948e-06,
      "loss": 0.3167,
      "step": 1649
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5636694758399563e-06,
      "loss": 0.2939,
      "step": 1650
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5549857759615195e-06,
      "loss": 0.2991,
      "step": 1651
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5463242221483742e-06,
      "loss": 0.3062,
      "step": 1652
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5376848371144404e-06,
      "loss": 0.2834,
      "step": 1653
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5290676435154949e-06,
      "loss": 0.3087,
      "step": 1654
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.520472663949122e-06,
      "loss": 0.3508,
      "step": 1655
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.511899920954656e-06,
      "loss": 0.2915,
      "step": 1656
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5033494370131162e-06,
      "loss": 0.2971,
      "step": 1657
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4948212345471492e-06,
      "loss": 0.3288,
      "step": 1658
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4863153359209693e-06,
      "loss": 0.3216,
      "step": 1659
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4778317634403082e-06,
      "loss": 0.3173,
      "step": 1660
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.469370539352345e-06,
      "loss": 0.3572,
      "step": 1661
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.460931685845649e-06,
      "loss": 0.3071,
      "step": 1662
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4525152250501362e-06,
      "loss": 0.3556,
      "step": 1663
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4441211790369892e-06,
      "loss": 0.294,
      "step": 1664
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4357495698186186e-06,
      "loss": 0.2905,
      "step": 1665
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.427400419348588e-06,
      "loss": 0.3211,
      "step": 1666
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4190737495215746e-06,
      "loss": 0.2948,
      "step": 1667
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4107695821733026e-06,
      "loss": 0.2951,
      "step": 1668
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.402487939080479e-06,
      "loss": 0.2877,
      "step": 1669
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3942288419607476e-06,
      "loss": 0.3135,
      "step": 1670
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3859923124726283e-06,
      "loss": 0.2892,
      "step": 1671
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3777783722154603e-06,
      "loss": 0.286,
      "step": 1672
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.369587042729341e-06,
      "loss": 0.2768,
      "step": 1673
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3614183454950824e-06,
      "loss": 0.3297,
      "step": 1674
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3532723019341376e-06,
      "loss": 0.2959,
      "step": 1675
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3451489334085555e-06,
      "loss": 0.3296,
      "step": 1676
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3370482612209224e-06,
      "loss": 0.3014,
      "step": 1677
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3289703066143112e-06,
      "loss": 0.3268,
      "step": 1678
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3209150907722124e-06,
      "loss": 0.2981,
      "step": 1679
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3128826348184886e-06,
      "loss": 0.2967,
      "step": 1680
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3048729598173248e-06,
      "loss": 0.2178,
      "step": 1681
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.296886086773157e-06,
      "loss": 0.2973,
      "step": 1682
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2889220366306276e-06,
      "loss": 0.3261,
      "step": 1683
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2809808302745298e-06,
      "loss": 0.2954,
      "step": 1684
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2730624885297537e-06,
      "loss": 0.3454,
      "step": 1685
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2651670321612264e-06,
      "loss": 0.3073,
      "step": 1686
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2572944818738587e-06,
      "loss": 0.2906,
      "step": 1687
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.249444858312502e-06,
      "loss": 0.3021,
      "step": 1688
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2416181820618745e-06,
      "loss": 0.2942,
      "step": 1689
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.233814473646524e-06,
      "loss": 0.3381,
      "step": 1690
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.226033753530763e-06,
      "loss": 0.3046,
      "step": 1691
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.218276042118629e-06,
      "loss": 0.3362,
      "step": 1692
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2105413597538107e-06,
      "loss": 0.3149,
      "step": 1693
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.202829726719611e-06,
      "loss": 0.2529,
      "step": 1694
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.195141163238892e-06,
      "loss": 0.3058,
      "step": 1695
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1874756894740137e-06,
      "loss": 0.3242,
      "step": 1696
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1798333255267857e-06,
      "loss": 0.2882,
      "step": 1697
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1722140914384162e-06,
      "loss": 0.2735,
      "step": 1698
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1646180071894608e-06,
      "loss": 0.2994,
      "step": 1699
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1570450926997657e-06,
      "loss": 0.2764,
      "step": 1700
    },
    {
      "epoch": 0.85,
      "eval_code_gate_load": [
        207.4,
        170.6,
        179.4,
        154.8,
        177.85,
        186.5,
        195.0,
        168.45
      ],
      "eval_code_loss": 0.26079100370407104,
      "eval_code_runtime": 1.7851,
      "eval_code_samples_per_second": 560.183,
      "eval_code_steps_per_second": 35.292,
      "step": 1700
    },
    {
      "epoch": 0.85,
      "eval_orca_gate_load": [
        499.15,
        341.7,
        406.8,
        396.0,
        351.7,
        421.4,
        368.95,
        356.6
      ],
      "eval_orca_loss": 0.35004884004592896,
      "eval_orca_runtime": 2.0049,
      "eval_orca_samples_per_second": 498.779,
      "eval_orca_steps_per_second": 31.423,
      "step": 1700
    },
    {
      "epoch": 0.85,
      "eval_math_gate_load": [
        314.35,
        215.9,
        240.2,
        220.65,
        253.5,
        262.05,
        277.05,
        249.0
      ],
      "eval_math_loss": 0.2432861328125,
      "eval_math_runtime": 1.8844,
      "eval_math_samples_per_second": 530.665,
      "eval_math_steps_per_second": 33.432,
      "step": 1700
    },
    {
      "epoch": 0.85,
      "eval_sharegpt_gate_load": [
        1523.75,
        1077.65,
        1308.6,
        1203.3,
        1140.45,
        1368.2,
        1179.25,
        1110.4
      ],
      "eval_sharegpt_loss": 0.4883789122104645,
      "eval_sharegpt_runtime": 3.003,
      "eval_sharegpt_samples_per_second": 333.001,
      "eval_sharegpt_steps_per_second": 20.979,
      "step": 1700
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1494953678284105e-06,
      "loss": 0.2901,
      "step": 1701
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1419688523736761e-06,
      "loss": 0.3298,
      "step": 1702
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1344655660729676e-06,
      "loss": 0.3355,
      "step": 1703
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1269855286027798e-06,
      "loss": 0.323,
      "step": 1704
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1195287595786352e-06,
      "loss": 0.3004,
      "step": 1705
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1120952785550477e-06,
      "loss": 0.258,
      "step": 1706
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1046851050254504e-06,
      "loss": 0.2928,
      "step": 1707
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0972982584221592e-06,
      "loss": 0.294,
      "step": 1708
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0899347581163222e-06,
      "loss": 0.3543,
      "step": 1709
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0825946234178575e-06,
      "loss": 0.3076,
      "step": 1710
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0752778735754121e-06,
      "loss": 0.2811,
      "step": 1711
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.067984527776309e-06,
      "loss": 0.2797,
      "step": 1712
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0607146051465011e-06,
      "loss": 0.2882,
      "step": 1713
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0534681247505107e-06,
      "loss": 0.2896,
      "step": 1714
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0462451055913847e-06,
      "loss": 0.2878,
      "step": 1715
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0390455666106547e-06,
      "loss": 0.3366,
      "step": 1716
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0318695266882696e-06,
      "loss": 0.2932,
      "step": 1717
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.024717004642557e-06,
      "loss": 0.3224,
      "step": 1718
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0175880192301713e-06,
      "loss": 0.3163,
      "step": 1719
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.010482589146048e-06,
      "loss": 0.3059,
      "step": 1720
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0034007330233487e-06,
      "loss": 0.3024,
      "step": 1721
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.963424694334122e-07,
      "loss": 0.31,
      "step": 1722
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.893078168857173e-07,
      "loss": 0.2936,
      "step": 1723
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.822967938278172e-07,
      "loss": 0.244,
      "step": 1724
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.753094186453028e-07,
      "loss": 0.362,
      "step": 1725
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.683457096617487e-07,
      "loss": 0.3021,
      "step": 1726
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.614056851386743e-07,
      "loss": 0.2698,
      "step": 1727
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.544893632754816e-07,
      "loss": 0.3254,
      "step": 1728
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.475967622094207e-07,
      "loss": 0.3472,
      "step": 1729
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.407279000155311e-07,
      "loss": 0.3558,
      "step": 1730
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.338827947066076e-07,
      "loss": 0.2826,
      "step": 1731
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.270614642331377e-07,
      "loss": 0.287,
      "step": 1732
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.202639264832669e-07,
      "loss": 0.3271,
      "step": 1733
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.134901992827427e-07,
      "loss": 0.3017,
      "step": 1734
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.067403003948783e-07,
      "loss": 0.2957,
      "step": 1735
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.000142475204965e-07,
      "loss": 0.2881,
      "step": 1736
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.933120582978827e-07,
      "loss": 0.3096,
      "step": 1737
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.866337503027523e-07,
      "loss": 0.3048,
      "step": 1738
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.799793410481871e-07,
      "loss": 0.2805,
      "step": 1739
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.733488479845997e-07,
      "loss": 0.3241,
      "step": 1740
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.667422884996823e-07,
      "loss": 0.2841,
      "step": 1741
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.60159679918372e-07,
      "loss": 0.3075,
      "step": 1742
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.536010395027905e-07,
      "loss": 0.3022,
      "step": 1743
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.470663844522053e-07,
      "loss": 0.2859,
      "step": 1744
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.405557319029911e-07,
      "loss": 0.2964,
      "step": 1745
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.340690989285727e-07,
      "loss": 0.2967,
      "step": 1746
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.276065025393909e-07,
      "loss": 0.3191,
      "step": 1747
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.211679596828481e-07,
      "loss": 0.27,
      "step": 1748
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.147534872432761e-07,
      "loss": 0.3067,
      "step": 1749
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.083631020418792e-07,
      "loss": 0.3003,
      "step": 1750
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.019968208366958e-07,
      "loss": 0.3156,
      "step": 1751
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.956546603225601e-07,
      "loss": 0.2852,
      "step": 1752
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.893366371310463e-07,
      "loss": 0.3485,
      "step": 1753
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.830427678304353e-07,
      "loss": 0.3002,
      "step": 1754
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.767730689256614e-07,
      "loss": 0.3188,
      "step": 1755
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.705275568582848e-07,
      "loss": 0.3151,
      "step": 1756
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.643062480064301e-07,
      "loss": 0.2763,
      "step": 1757
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.581091586847522e-07,
      "loss": 0.3074,
      "step": 1758
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.519363051443996e-07,
      "loss": 0.2607,
      "step": 1759
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.457877035729588e-07,
      "loss": 0.333,
      "step": 1760
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.3966337009442e-07,
      "loss": 0.3025,
      "step": 1761
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.335633207691362e-07,
      "loss": 0.2867,
      "step": 1762
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.274875715937746e-07,
      "loss": 0.2828,
      "step": 1763
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.21436138501278e-07,
      "loss": 0.2814,
      "step": 1764
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.154090373608236e-07,
      "loss": 0.2994,
      "step": 1765
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.094062839777838e-07,
      "loss": 0.2821,
      "step": 1766
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.03427894093679e-07,
      "loss": 0.2824,
      "step": 1767
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.974738833861383e-07,
      "loss": 0.2706,
      "step": 1768
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.915442674688633e-07,
      "loss": 0.2929,
      "step": 1769
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.856390618915775e-07,
      "loss": 0.3311,
      "step": 1770
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.797582821399973e-07,
      "loss": 0.31,
      "step": 1771
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.739019436357774e-07,
      "loss": 0.3079,
      "step": 1772
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.680700617364877e-07,
      "loss": 0.307,
      "step": 1773
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.622626517355557e-07,
      "loss": 0.3126,
      "step": 1774
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.564797288622371e-07,
      "loss": 0.3571,
      "step": 1775
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.507213082815745e-07,
      "loss": 0.277,
      "step": 1776
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.449874050943549e-07,
      "loss": 0.3365,
      "step": 1777
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.392780343370686e-07,
      "loss": 0.2917,
      "step": 1778
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.335932109818754e-07,
      "loss": 0.2717,
      "step": 1779
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.279329499365649e-07,
      "loss": 0.2477,
      "step": 1780
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.222972660445082e-07,
      "loss": 0.3076,
      "step": 1781
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.166861740846297e-07,
      "loss": 0.2606,
      "step": 1782
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.11099688771366e-07,
      "loss": 0.2695,
      "step": 1783
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.055378247546217e-07,
      "loss": 0.3152,
      "step": 1784
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.000005966197387e-07,
      "loss": 0.268,
      "step": 1785
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.94488018887448e-07,
      "loss": 0.3017,
      "step": 1786
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.890001060138484e-07,
      "loss": 0.3279,
      "step": 1787
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.835368723903456e-07,
      "loss": 0.2949,
      "step": 1788
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.780983323436374e-07,
      "loss": 0.3345,
      "step": 1789
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.726845001356573e-07,
      "loss": 0.3232,
      "step": 1790
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.672953899635524e-07,
      "loss": 0.2993,
      "step": 1791
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.619310159596358e-07,
      "loss": 0.3065,
      "step": 1792
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.565913921913513e-07,
      "loss": 0.2778,
      "step": 1793
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.51276532661238e-07,
      "loss": 0.2822,
      "step": 1794
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.459864513068991e-07,
      "loss": 0.3144,
      "step": 1795
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.407211620009545e-07,
      "loss": 0.3007,
      "step": 1796
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.354806785510113e-07,
      "loss": 0.2625,
      "step": 1797
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.30265014699628e-07,
      "loss": 0.293,
      "step": 1798
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.250741841242735e-07,
      "loss": 0.2848,
      "step": 1799
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.199082004372958e-07,
      "loss": 0.2789,
      "step": 1800
    },
    {
      "epoch": 0.9,
      "eval_code_gate_load": [
        206.3,
        171.0,
        178.4,
        153.6,
        178.55,
        188.05,
        195.05,
        169.05
      ],
      "eval_code_loss": 0.24262695014476776,
      "eval_code_runtime": 1.786,
      "eval_code_samples_per_second": 559.908,
      "eval_code_steps_per_second": 35.274,
      "step": 1800
    },
    {
      "epoch": 0.9,
      "eval_orca_gate_load": [
        498.15,
        341.45,
        405.0,
        396.0,
        352.45,
        421.95,
        368.95,
        358.35
      ],
      "eval_orca_loss": 0.3505859375,
      "eval_orca_runtime": 2.0051,
      "eval_orca_samples_per_second": 498.718,
      "eval_orca_steps_per_second": 31.419,
      "step": 1800
    },
    {
      "epoch": 0.9,
      "eval_math_gate_load": [
        314.35,
        216.7,
        239.55,
        220.1,
        253.35,
        261.95,
        275.8,
        250.9
      ],
      "eval_math_loss": 0.24282225966453552,
      "eval_math_runtime": 1.865,
      "eval_math_samples_per_second": 536.202,
      "eval_math_steps_per_second": 33.781,
      "step": 1800
    },
    {
      "epoch": 0.9,
      "eval_sharegpt_gate_load": [
        1525.35,
        1079.5,
        1303.65,
        1203.5,
        1140.4,
        1370.1,
        1178.8,
        1110.3
      ],
      "eval_sharegpt_loss": 0.48857420682907104,
      "eval_sharegpt_runtime": 2.9886,
      "eval_sharegpt_samples_per_second": 334.604,
      "eval_sharegpt_steps_per_second": 21.08,
      "step": 1800
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.147670771858848e-07,
      "loss": 0.3078,
      "step": 1801
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.096508278520385e-07,
      "loss": 0.3373,
      "step": 1802
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.045594658525232e-07,
      "loss": 0.3121,
      "step": 1803
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.994930045388414e-07,
      "loss": 0.2671,
      "step": 1804
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.944514571971981e-07,
      "loss": 0.3293,
      "step": 1805
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.894348370484648e-07,
      "loss": 0.3154,
      "step": 1806
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.844431572481412e-07,
      "loss": 0.2988,
      "step": 1807
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.794764308863242e-07,
      "loss": 0.2988,
      "step": 1808
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.745346709876786e-07,
      "loss": 0.292,
      "step": 1809
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.696178905113913e-07,
      "loss": 0.2773,
      "step": 1810
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.6472610235114513e-07,
      "loss": 0.2962,
      "step": 1811
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.5985931933508757e-07,
      "loss": 0.2758,
      "step": 1812
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.550175542257862e-07,
      "loss": 0.3309,
      "step": 1813
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.502008197202068e-07,
      "loss": 0.3072,
      "step": 1814
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.454091284496731e-07,
      "loss": 0.3043,
      "step": 1815
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.406424929798403e-07,
      "loss": 0.2964,
      "step": 1816
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3590092581065055e-07,
      "loss": 0.2578,
      "step": 1817
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3118443937631094e-07,
      "loss": 0.2903,
      "step": 1818
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.26493046045261e-07,
      "loss": 0.2921,
      "step": 1819
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.218267581201296e-07,
      "loss": 0.3116,
      "step": 1820
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.17185587837714e-07,
      "loss": 0.2662,
      "step": 1821
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.125695473689406e-07,
      "loss": 0.2796,
      "step": 1822
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.0797864881883977e-07,
      "loss": 0.3705,
      "step": 1823
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.034129042265067e-07,
      "loss": 0.265,
      "step": 1824
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.988723255650728e-07,
      "loss": 0.3237,
      "step": 1825
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.943569247416801e-07,
      "loss": 0.2654,
      "step": 1826
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.8986671359743767e-07,
      "loss": 0.3055,
      "step": 1827
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.8540170390740097e-07,
      "loss": 0.2912,
      "step": 1828
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.8096190738053815e-07,
      "loss": 0.2577,
      "step": 1829
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.7654733565969826e-07,
      "loss": 0.3198,
      "step": 1830
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.721580003215808e-07,
      "loss": 0.3015,
      "step": 1831
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.67793912876705e-07,
      "loss": 0.2554,
      "step": 1832
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.6345508476938296e-07,
      "loss": 0.2986,
      "step": 1833
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.591415273776855e-07,
      "loss": 0.2911,
      "step": 1834
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.548532520134129e-07,
      "loss": 0.2769,
      "step": 1835
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.5059026992206645e-07,
      "loss": 0.3182,
      "step": 1836
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4635259228282256e-07,
      "loss": 0.2591,
      "step": 1837
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.421402302084953e-07,
      "loss": 0.3009,
      "step": 1838
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.379531947455128e-07,
      "loss": 0.2904,
      "step": 1839
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.3379149687388866e-07,
      "loss": 0.3221,
      "step": 1840
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2965514750718964e-07,
      "loss": 0.3366,
      "step": 1841
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.255441574925089e-07,
      "loss": 0.2937,
      "step": 1842
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2145853761043844e-07,
      "loss": 0.3,
      "step": 1843
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.1739829857504235e-07,
      "loss": 0.2915,
      "step": 1844
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.133634510338235e-07,
      "loss": 0.2622,
      "step": 1845
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.093540055676958e-07,
      "loss": 0.2865,
      "step": 1846
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.053699726909676e-07,
      "loss": 0.2946,
      "step": 1847
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.0141136285129825e-07,
      "loss": 0.3349,
      "step": 1848
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.974781864296783e-07,
      "loss": 0.2425,
      "step": 1849
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.935704537404083e-07,
      "loss": 0.2605,
      "step": 1850
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.8968817503105984e-07,
      "loss": 0.2733,
      "step": 1851
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.8583136048245697e-07,
      "loss": 0.2423,
      "step": 1852
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.820000202086459e-07,
      "loss": 0.2926,
      "step": 1853
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.781941642568686e-07,
      "loss": 0.242,
      "step": 1854
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.744138026075405e-07,
      "loss": 0.3168,
      "step": 1855
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.706589451742181e-07,
      "loss": 0.3208,
      "step": 1856
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.669296018035772e-07,
      "loss": 0.2808,
      "step": 1857
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.632257822753881e-07,
      "loss": 0.2936,
      "step": 1858
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5954749630248355e-07,
      "loss": 0.2861,
      "step": 1859
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5589475353073987e-07,
      "loss": 0.2806,
      "step": 1860
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5226756353904925e-07,
      "loss": 0.3313,
      "step": 1861
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.486659358392951e-07,
      "loss": 0.288,
      "step": 1862
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.450898798763268e-07,
      "loss": 0.2649,
      "step": 1863
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.4153940502793185e-07,
      "loss": 0.2993,
      "step": 1864
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.380145206048201e-07,
      "loss": 0.2831,
      "step": 1865
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.3451523585058756e-07,
      "loss": 0.269,
      "step": 1866
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.3104155994170042e-07,
      "loss": 0.2863,
      "step": 1867
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.2759350198746978e-07,
      "loss": 0.2895,
      "step": 1868
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.24171071030026e-07,
      "loss": 0.3511,
      "step": 1869
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.2077427604429435e-07,
      "loss": 0.3163,
      "step": 1870
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.1740312593797274e-07,
      "loss": 0.2954,
      "step": 1871
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.1405762955151178e-07,
      "loss": 0.3207,
      "step": 1872
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.1073779565808471e-07,
      "loss": 0.2869,
      "step": 1873
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.0744363296356872e-07,
      "loss": 0.296,
      "step": 1874
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.0417515010652032e-07,
      "loss": 0.3008,
      "step": 1875
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.009323556581566e-07,
      "loss": 0.2822,
      "step": 1876
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.977152581223274e-07,
      "loss": 0.2678,
      "step": 1877
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9452386593549534e-07,
      "loss": 0.3009,
      "step": 1878
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9135818746671587e-07,
      "loss": 0.2525,
      "step": 1879
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8821823101760949e-07,
      "loss": 0.342,
      "step": 1880
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8510400482234848e-07,
      "loss": 0.3283,
      "step": 1881
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8201551704762453e-07,
      "loss": 0.3051,
      "step": 1882
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7895277579264015e-07,
      "loss": 0.2982,
      "step": 1883
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7591578908907724e-07,
      "loss": 0.2743,
      "step": 1884
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7290456490107522e-07,
      "loss": 0.2717,
      "step": 1885
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.699191111252241e-07,
      "loss": 0.3157,
      "step": 1886
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6695943559052463e-07,
      "loss": 0.3009,
      "step": 1887
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6402554605838173e-07,
      "loss": 0.2954,
      "step": 1888
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6111745022257873e-07,
      "loss": 0.2454,
      "step": 1889
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5823515570925763e-07,
      "loss": 0.325,
      "step": 1890
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5537867007690111e-07,
      "loss": 0.3027,
      "step": 1891
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5254800081630828e-07,
      "loss": 0.2612,
      "step": 1892
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4974315535058016e-07,
      "loss": 0.3325,
      "step": 1893
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.469641410350964e-07,
      "loss": 0.2978,
      "step": 1894
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4421096515749855e-07,
      "loss": 0.284,
      "step": 1895
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.4148363493766803e-07,
      "loss": 0.3069,
      "step": 1896
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3878215752771264e-07,
      "loss": 0.3225,
      "step": 1897
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.361065400119399e-07,
      "loss": 0.3019,
      "step": 1898
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3345678940684615e-07,
      "loss": 0.3103,
      "step": 1899
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.30832912661093e-07,
      "loss": 0.2691,
      "step": 1900
    },
    {
      "epoch": 0.95,
      "eval_code_gate_load": [
        205.55,
        171.4,
        178.65,
        153.4,
        178.2,
        188.1,
        195.95,
        168.75
      ],
      "eval_code_loss": 0.24282225966453552,
      "eval_code_runtime": 1.7813,
      "eval_code_samples_per_second": 561.397,
      "eval_code_steps_per_second": 35.368,
      "step": 1900
    },
    {
      "epoch": 0.95,
      "eval_orca_gate_load": [
        497.35,
        342.7,
        406.6,
        396.1,
        351.0,
        421.1,
        369.95,
        357.5
      ],
      "eval_orca_loss": 0.34912109375,
      "eval_orca_runtime": 2.0183,
      "eval_orca_samples_per_second": 495.476,
      "eval_orca_steps_per_second": 31.215,
      "step": 1900
    },
    {
      "epoch": 0.95,
      "eval_math_gate_load": [
        314.85,
        216.8,
        238.95,
        220.8,
        253.1,
        261.85,
        276.2,
        250.15
      ],
      "eval_math_loss": 0.24270018935203552,
      "eval_math_runtime": 1.8503,
      "eval_math_samples_per_second": 540.439,
      "eval_math_steps_per_second": 34.048,
      "step": 1900
    },
    {
      "epoch": 0.95,
      "eval_sharegpt_gate_load": [
        1522.5,
        1078.8,
        1304.65,
        1205.55,
        1138.55,
        1369.8,
        1181.35,
        1110.4
      ],
      "eval_sharegpt_loss": 0.4888671934604645,
      "eval_sharegpt_runtime": 3.0127,
      "eval_sharegpt_samples_per_second": 331.925,
      "eval_sharegpt_steps_per_second": 20.911,
      "step": 1900
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2823491665549193e-07,
      "loss": 0.3058,
      "step": 1901
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2566280820298427e-07,
      "loss": 0.3011,
      "step": 1902
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.231165940486234e-07,
      "loss": 0.2873,
      "step": 1903
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2059628086956044e-07,
      "loss": 0.3044,
      "step": 1904
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1810187527502182e-07,
      "loss": 0.2932,
      "step": 1905
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1563338380629618e-07,
      "loss": 0.3145,
      "step": 1906
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1319081293671541e-07,
      "loss": 0.3078,
      "step": 1907
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1077416907163573e-07,
      "loss": 0.2436,
      "step": 1908
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0838345854842447e-07,
      "loss": 0.2853,
      "step": 1909
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0601868763643997e-07,
      "loss": 0.3035,
      "step": 1910
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.0367986253701945e-07,
      "loss": 0.3109,
      "step": 1911
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.0136698938346012e-07,
      "loss": 0.2506,
      "step": 1912
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.90800742410003e-08,
      "loss": 0.3005,
      "step": 1913
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.68191231068083e-08,
      "loss": 0.284,
      "step": 1914
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.45841419099669e-08,
      "loss": 0.2738,
      "step": 1915
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.237513651145224e-08,
      "loss": 0.3183,
      "step": 1916
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.019211270412275e-08,
      "loss": 0.3705,
      "step": 1917
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.80350762127058e-08,
      "loss": 0.2912,
      "step": 1918
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.590403269377656e-08,
      "loss": 0.334,
      "step": 1919
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.379898773574924e-08,
      "loss": 0.3033,
      "step": 1920
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.171994685885698e-08,
      "loss": 0.2956,
      "step": 1921
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.966691551514527e-08,
      "loss": 0.3074,
      "step": 1922
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.763989908844749e-08,
      "loss": 0.3667,
      "step": 1923
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.563890289437825e-08,
      "loss": 0.328,
      "step": 1924
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.366393218031564e-08,
      "loss": 0.317,
      "step": 1925
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.171499212539124e-08,
      "loss": 0.263,
      "step": 1926
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.979208784047454e-08,
      "loss": 0.2675,
      "step": 1927
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.78952243681541e-08,
      "loss": 0.3353,
      "step": 1928
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.602440668273758e-08,
      "loss": 0.2805,
      "step": 1929
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.417963969022389e-08,
      "loss": 0.2993,
      "step": 1930
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.236092822829887e-08,
      "loss": 0.2835,
      "step": 1931
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.056827706632185e-08,
      "loss": 0.2756,
      "step": 1932
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.880169090531351e-08,
      "loss": 0.2952,
      "step": 1933
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.7061174377937015e-08,
      "loss": 0.3181,
      "step": 1934
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.534673204849572e-08,
      "loss": 0.2758,
      "step": 1935
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.365836841291439e-08,
      "loss": 0.3501,
      "step": 1936
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.199608789873134e-08,
      "loss": 0.285,
      "step": 1937
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.035989486508075e-08,
      "loss": 0.3217,
      "step": 1938
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.874979360268928e-08,
      "loss": 0.3108,
      "step": 1939
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.716578833386054e-08,
      "loss": 0.2918,
      "step": 1940
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.56078832124629e-08,
      "loss": 0.2769,
      "step": 1941
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.4076082323920576e-08,
      "loss": 0.2785,
      "step": 1942
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.257038968520366e-08,
      "loss": 0.3826,
      "step": 1943
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.109080924481479e-08,
      "loss": 0.3202,
      "step": 1944
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.963734488278248e-08,
      "loss": 0.3273,
      "step": 1945
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.82100004106456e-08,
      "loss": 0.2841,
      "step": 1946
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.680877957145112e-08,
      "loss": 0.2521,
      "step": 1947
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.543368603973529e-08,
      "loss": 0.2639,
      "step": 1948
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.408472342152136e-08,
      "loss": 0.2813,
      "step": 1949
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.2761895254306285e-08,
      "loss": 0.2856,
      "step": 1950
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.1465205007052965e-08,
      "loss": 0.2849,
      "step": 1951
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.019465608018024e-08,
      "loss": 0.313,
      "step": 1952
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.8950251805553997e-08,
      "loss": 0.2965,
      "step": 1953
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.773199544648164e-08,
      "loss": 0.2648,
      "step": 1954
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.6539890197695428e-08,
      "loss": 0.3016,
      "step": 1955
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.537393918535358e-08,
      "loss": 0.2957,
      "step": 1956
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.423414546702807e-08,
      "loss": 0.2827,
      "step": 1957
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.312051203169352e-08,
      "loss": 0.3606,
      "step": 1958
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.2033041799723877e-08,
      "loss": 0.3286,
      "step": 1959
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.0971737622883515e-08,
      "loss": 0.2846,
      "step": 1960
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9936602284318375e-08,
      "loss": 0.3424,
      "step": 1961
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.8927638498551502e-08,
      "loss": 0.3063,
      "step": 1962
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.7944848911470857e-08,
      "loss": 0.2941,
      "step": 1963
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.698823610032929e-08,
      "loss": 0.3948,
      "step": 1964
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.605780257373124e-08,
      "loss": 0.3256,
      "step": 1965
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5153550771630498e-08,
      "loss": 0.2806,
      "step": 1966
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.4275483065321338e-08,
      "loss": 0.2948,
      "step": 1967
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3423601757436289e-08,
      "loss": 0.2675,
      "step": 1968
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2597909081931702e-08,
      "loss": 0.2999,
      "step": 1969
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.179840720409331e-08,
      "loss": 0.2986,
      "step": 1970
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.102509822051845e-08,
      "loss": 0.3122,
      "step": 1971
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0277984159122734e-08,
      "loss": 0.3172,
      "step": 1972
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.557066979123398e-09,
      "loss": 0.315,
      "step": 1973
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.862348571043733e-09,
      "loss": 0.3141,
      "step": 1974
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.193830756699773e-09,
      "loss": 0.3203,
      "step": 1975
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.551515289203615e-09,
      "loss": 0.3238,
      "step": 1976
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.935403852950107e-09,
      "loss": 0.2921,
      "step": 1977
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.345498063622391e-09,
      "loss": 0.3407,
      "step": 1978
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.781799468177473e-09,
      "loss": 0.2725,
      "step": 1979
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.2443095448506674e-09,
      "loss": 0.3079,
      "step": 1980
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.733029703146708e-09,
      "loss": 0.3236,
      "step": 1981
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.247961283835311e-09,
      "loss": 0.3236,
      "step": 1982
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.789105558954509e-09,
      "loss": 0.3095,
      "step": 1983
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3564637317984318e-09,
      "loss": 0.3494,
      "step": 1984
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.9500369369195313e-09,
      "loss": 0.3102,
      "step": 1985
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.5698262401263607e-09,
      "loss": 0.2826,
      "step": 1986
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.215832638474691e-09,
      "loss": 0.2811,
      "step": 1987
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.888057060274173e-09,
      "loss": 0.2664,
      "step": 1988
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5865003650761268e-09,
      "loss": 0.3725,
      "step": 1989
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3111633436779792e-09,
      "loss": 0.3254,
      "step": 1990
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.062046718121046e-09,
      "loss": 0.3177,
      "step": 1991
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.391511416816489e-10,
      "loss": 0.2967,
      "step": 1992
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.424771988788881e-10,
      "loss": 0.3657,
      "step": 1993
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.720254054679796e-10,
      "loss": 0.2909,
      "step": 1994
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.277962084369257e-10,
      "loss": 0.3087,
      "step": 1995
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.0978998601206558e-10,
      "loss": 0.3516,
      "step": 1996
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.1800704765030367e-10,
      "loss": 0.3215,
      "step": 1997
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.244763404133046e-11,
      "loss": 0.3044,
      "step": 1998
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.311191710651194e-11,
      "loss": 0.3368,
      "step": 1999
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 0.2367,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_code_gate_load": [
        205.2,
        169.7,
        178.05,
        154.0,
        178.95,
        188.05,
        195.85,
        170.2
      ],
      "eval_code_loss": 0.2437744140625,
      "eval_code_runtime": 1.7981,
      "eval_code_samples_per_second": 556.137,
      "eval_code_steps_per_second": 35.037,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_orca_gate_load": [
        497.05,
        341.55,
        406.6,
        396.8,
        351.5,
        422.2,
        370.25,
        356.35
      ],
      "eval_orca_loss": 0.34916990995407104,
      "eval_orca_runtime": 2.0032,
      "eval_orca_samples_per_second": 499.21,
      "eval_orca_steps_per_second": 31.45,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_math_gate_load": [
        314.55,
        216.95,
        239.45,
        221.2,
        253.25,
        261.05,
        276.5,
        249.75
      ],
      "eval_math_loss": 0.24287109076976776,
      "eval_math_runtime": 1.8382,
      "eval_math_samples_per_second": 544.019,
      "eval_math_steps_per_second": 34.273,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_sharegpt_gate_load": [
        1524.05,
        1079.85,
        1307.8,
        1203.45,
        1136.45,
        1369.95,
        1180.15,
        1109.9
      ],
      "eval_sharegpt_loss": 0.4881835877895355,
      "eval_sharegpt_runtime": 2.9976,
      "eval_sharegpt_samples_per_second": 333.601,
      "eval_sharegpt_steps_per_second": 21.017,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "step": 2000,
      "total_flos": 2.079150933069005e+19,
      "train_loss": 0.5031886091232299,
      "train_runtime": 25719.4214,
      "train_samples_per_second": 9.954,
      "train_steps_per_second": 0.078
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 9999999999999,
  "total_flos": 2.079150933069005e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}