{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.9867197875166,
  "global_step": 5640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 9.411764705882353e-05,
      "loss": 1.501,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00018823529411764707,
      "loss": 1.4128,
      "step": 40
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002823529411764706,
      "loss": 1.325,
      "step": 60
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00037647058823529414,
      "loss": 1.2828,
      "step": 80
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00047058823529411766,
      "loss": 1.2758,
      "step": 100
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0005647058823529412,
      "loss": 1.2667,
      "step": 120
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0006588235294117648,
      "loss": 1.2504,
      "step": 140
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0007529411764705883,
      "loss": 1.2394,
      "step": 160
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0007999934028874321,
      "loss": 1.228,
      "step": 180
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0007999406272925394,
      "loss": 1.2138,
      "step": 200
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0007998350830660272,
      "loss": 1.2122,
      "step": 220
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0007996767841335234,
      "loss": 1.219,
      "step": 240
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0007994657513811737,
      "loss": 1.1998,
      "step": 260
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0007992020126528848,
      "loss": 1.188,
      "step": 280
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0007988856027466511,
      "loss": 1.1931,
      "step": 300
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.000798516563409964,
      "loss": 1.1753,
      "step": 320
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0007980949433343026,
      "loss": 1.173,
      "step": 340
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0007976207981487104,
      "loss": 1.1637,
      "step": 360
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0007970941904124546,
      "loss": 1.1651,
      "step": 380
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0007965151896067728,
      "loss": 1.1559,
      "step": 400
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0007958838721257046,
      "loss": 1.1658,
      "step": 420
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0007952003212660127,
      "loss": 1.1496,
      "step": 440
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0007944646272161933,
      "loss": 1.1471,
      "step": 460
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0007936768870445747,
      "loss": 1.1311,
      "step": 480
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0007928372046865116,
      "loss": 1.1301,
      "step": 500
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0007919456909306711,
      "loss": 1.134,
      "step": 520
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0007910024634044154,
      "loss": 1.1235,
      "step": 540
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0007900076465582816,
      "loss": 1.1239,
      "step": 560
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0007889613716495616,
      "loss": 1.0878,
      "step": 580
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0007878637767249839,
      "loss": 1.0879,
      "step": 600
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0007867150066024996,
      "loss": 1.0671,
      "step": 620
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0007855152128521754,
      "loss": 1.0689,
      "step": 640
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0007842645537761941,
      "loss": 1.0794,
      "step": 660
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0007829631943879694,
      "loss": 1.0653,
      "step": 680
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0007816113063903726,
      "loss": 1.066,
      "step": 700
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0007802090681530788,
      "loss": 1.0675,
      "step": 720
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0007787566646890325,
      "loss": 1.0598,
      "step": 740
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0007772542876300359,
      "loss": 1.0669,
      "step": 760
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0007757021352014663,
      "loss": 1.0558,
      "step": 780
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0007741004121961207,
      "loss": 1.0578,
      "step": 800
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0007724493299471956,
      "loss": 1.056,
      "step": 820
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0007707491063004035,
      "loss": 1.0491,
      "step": 840
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0007689999655852306,
      "loss": 1.0497,
      "step": 860
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0007672021385853376,
      "loss": 1.0393,
      "step": 880
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0007653558625081099,
      "loss": 1.0379,
      "step": 900
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0007634613809533613,
      "loss": 1.049,
      "step": 920
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0007615189438811918,
      "loss": 1.0594,
      "step": 940
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0007595288075790085,
      "loss": 1.0375,
      "step": 960
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0007574912346277103,
      "loss": 1.0453,
      "step": 980
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0007554064938670426,
      "loss": 1.0348,
      "step": 1000
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0007532748603601265,
      "loss": 1.0441,
      "step": 1020
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0007510966153571667,
      "loss": 1.0237,
      "step": 1040
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.000748872046258343,
      "loss": 1.0478,
      "step": 1060
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0007466014465758899,
      "loss": 1.0289,
      "step": 1080
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0007442851158953712,
      "loss": 1.026,
      "step": 1100
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007419233598361512,
      "loss": 1.0244,
      "step": 1120
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0007395164900110721,
      "loss": 0.9968,
      "step": 1140
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0007370648239853385,
      "loss": 0.9798,
      "step": 1160
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0007345686852346176,
      "loss": 0.9529,
      "step": 1180
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0007320284031023603,
      "loss": 0.9666,
      "step": 1200
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.000729444312756346,
      "loss": 0.9797,
      "step": 1220
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0007268167551444611,
      "loss": 0.9674,
      "step": 1240
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0007241460769497138,
      "loss": 0.9666,
      "step": 1260
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0007214326305444917,
      "loss": 0.9785,
      "step": 1280
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0007186767739440701,
      "loss": 0.9629,
      "step": 1300
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0007158788707593748,
      "loss": 0.973,
      "step": 1320
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0007130392901490069,
      "loss": 0.9649,
      "step": 1340
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0007101584067705355,
      "loss": 0.9766,
      "step": 1360
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0007072366007310646,
      "loss": 0.954,
      "step": 1380
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0007042742575370822,
      "loss": 0.9576,
      "step": 1400
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0007012717680435956,
      "loss": 0.9783,
      "step": 1420
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0006982295284025612,
      "loss": 0.9553,
      "step": 1440
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0006951479400106161,
      "loss": 0.951,
      "step": 1460
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000692027409456118,
      "loss": 0.9647,
      "step": 1480
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0006888683484654981,
      "loss": 0.9656,
      "step": 1500
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0006856711738489386,
      "loss": 0.9507,
      "step": 1520
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0006824363074453778,
      "loss": 0.9496,
      "step": 1540
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0006791641760668519,
      "loss": 0.9571,
      "step": 1560
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0006758552114421815,
      "loss": 0.9494,
      "step": 1580
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0006725098501600088,
      "loss": 0.9556,
      "step": 1600
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0006691285336111928,
      "loss": 0.9411,
      "step": 1620
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0006657117079305725,
      "loss": 0.9501,
      "step": 1640
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0006622598239381033,
      "loss": 0.9598,
      "step": 1660
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0006587733370793743,
      "loss": 0.9599,
      "step": 1680
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0006552527073655178,
      "loss": 0.9306,
      "step": 1700
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0006516983993125138,
      "loss": 0.9013,
      "step": 1720
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0006481108818799015,
      "loss": 0.8798,
      "step": 1740
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0006444906284089044,
      "loss": 0.903,
      "step": 1760
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.000640838116559977,
      "loss": 0.9017,
      "step": 1780
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.0006371538282497815,
      "loss": 0.9051,
      "step": 1800
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.0006334382495876036,
      "loss": 0.8966,
      "step": 1820
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.0006296918708112143,
      "loss": 0.8863,
      "step": 1840
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0006259151862221875,
      "loss": 0.8926,
      "step": 1860
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0006221086941206817,
      "loss": 0.908,
      "step": 1880
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0006182728967396925,
      "loss": 0.9007,
      "step": 1900
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0006144083001787886,
      "loss": 0.8872,
      "step": 1920
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.0006105154143373362,
      "loss": 0.8984,
      "step": 1940
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0006065947528472215,
      "loss": 0.9123,
      "step": 1960
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.0006026468330050827,
      "loss": 0.8929,
      "step": 1980
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0005986721757040564,
      "loss": 0.9145,
      "step": 2000
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0005946713053650507,
      "loss": 0.8867,
      "step": 2020
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0005906447498675521,
      "loss": 0.8914,
      "step": 2040
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0005865930404799774,
      "loss": 0.8946,
      "step": 2060
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.0005825167117895765,
      "loss": 0.892,
      "step": 2080
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.0005784163016318987,
      "loss": 0.8875,
      "step": 2100
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0005742923510198303,
      "loss": 0.888,
      "step": 2120
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.0005701454040722124,
      "loss": 0.9078,
      "step": 2140
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.0005659760079420498,
      "loss": 0.9027,
      "step": 2160
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.000561784712744318,
      "loss": 0.8997,
      "step": 2180
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.0005575720714833808,
      "loss": 0.9053,
      "step": 2200
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.0005533386399800275,
      "loss": 0.9054,
      "step": 2220
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.0005490849767981348,
      "loss": 0.8988,
      "step": 2240
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0005448116431709716,
      "loss": 0.8903,
      "step": 2260
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0005405192029271477,
      "loss": 0.8373,
      "step": 2280
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0005362082224162223,
      "loss": 0.8336,
      "step": 2300
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.0005318792704339792,
      "loss": 0.8483,
      "step": 2320
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.0005275329181473787,
      "loss": 0.8453,
      "step": 2340
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.0005231697390191976,
      "loss": 0.8351,
      "step": 2360
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.000518790308732366,
      "loss": 0.8329,
      "step": 2380
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.0005143952051140103,
      "loss": 0.8394,
      "step": 2400
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.000509985008059215,
      "loss": 0.8526,
      "step": 2420
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.0005055602994545098,
      "loss": 0.826,
      "step": 2440
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.0005011216631010953,
      "loss": 0.849,
      "step": 2460
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.0004966696846378156,
      "loss": 0.8507,
      "step": 2480
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.000492204951463888,
      "loss": 0.8461,
      "step": 2500
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00048772805266140154,
      "loss": 0.8533,
      "step": 2520
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00048323957891759203,
      "loss": 0.8384,
      "step": 2540
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00047874012244690696,
      "loss": 0.842,
      "step": 2560
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.000474230276912867,
      "loss": 0.8608,
      "step": 2580
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.00046971063734973833,
      "loss": 0.8562,
      "step": 2600
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.0004651818000840229,
      "loss": 0.8594,
      "step": 2620
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.0004606443626557778,
      "loss": 0.8608,
      "step": 2640
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.0004560989237397758,
      "loss": 0.8486,
      "step": 2660
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.00045154608306651514,
      "loss": 0.869,
      "step": 2680
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.0004469864413430907,
      "loss": 0.8482,
      "step": 2700
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.00044242060017393573,
      "loss": 0.8583,
      "step": 2720
    },
    {
      "epoch": 4.85,
      "learning_rate": 0.00043784916198144543,
      "loss": 0.8582,
      "step": 2740
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.00043327272992649317,
      "loss": 0.8504,
      "step": 2760
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.00042869190782884794,
      "loss": 0.8592,
      "step": 2780
    },
    {
      "epoch": 4.96,
      "learning_rate": 0.00042410730008750623,
      "loss": 0.8545,
      "step": 2800
    },
    {
      "epoch": 4.99,
      "learning_rate": 0.00041951951160094664,
      "loss": 0.855,
      "step": 2820
    },
    {
      "epoch": 5.03,
      "learning_rate": 0.00041492914768731927,
      "loss": 0.7869,
      "step": 2840
    },
    {
      "epoch": 5.06,
      "learning_rate": 0.0004103368140045789,
      "loss": 0.8083,
      "step": 2860
    },
    {
      "epoch": 5.1,
      "learning_rate": 0.00040574311647057366,
      "loss": 0.8108,
      "step": 2880
    },
    {
      "epoch": 5.14,
      "learning_rate": 0.00040114866118310045,
      "loss": 0.7968,
      "step": 2900
    },
    {
      "epoch": 5.17,
      "learning_rate": 0.0003965540543399344,
      "loss": 0.8016,
      "step": 2920
    },
    {
      "epoch": 5.21,
      "learning_rate": 0.00039195990215884756,
      "loss": 0.7967,
      "step": 2940
    },
    {
      "epoch": 5.24,
      "learning_rate": 0.00038736681079762293,
      "loss": 0.8096,
      "step": 2960
    },
    {
      "epoch": 5.28,
      "learning_rate": 0.0003827753862740779,
      "loss": 0.8073,
      "step": 2980
    },
    {
      "epoch": 5.31,
      "learning_rate": 0.0003781862343861055,
      "loss": 0.804,
      "step": 3000
    },
    {
      "epoch": 5.35,
      "learning_rate": 0.00037359996063174425,
      "loss": 0.8119,
      "step": 3020
    },
    {
      "epoch": 5.38,
      "learning_rate": 0.0003690171701292887,
      "loss": 0.7997,
      "step": 3040
    },
    {
      "epoch": 5.42,
      "learning_rate": 0.0003644384675374489,
      "loss": 0.8202,
      "step": 3060
    },
    {
      "epoch": 5.45,
      "learning_rate": 0.0003598644569755713,
      "loss": 0.815,
      "step": 3080
    },
    {
      "epoch": 5.49,
      "learning_rate": 0.00035529574194393033,
      "loss": 0.825,
      "step": 3100
    },
    {
      "epoch": 5.52,
      "learning_rate": 0.00035073292524410207,
      "loss": 0.8171,
      "step": 3120
    },
    {
      "epoch": 5.56,
      "learning_rate": 0.00034617660889943,
      "loss": 0.7921,
      "step": 3140
    },
    {
      "epoch": 5.6,
      "learning_rate": 0.00034162739407559285,
      "loss": 0.8299,
      "step": 3160
    },
    {
      "epoch": 5.63,
      "learning_rate": 0.0003370858810012869,
      "loss": 0.811,
      "step": 3180
    },
    {
      "epoch": 5.67,
      "learning_rate": 0.00033255266888903006,
      "loss": 0.8093,
      "step": 3200
    },
    {
      "epoch": 5.7,
      "learning_rate": 0.00032802835585610225,
      "loss": 0.8106,
      "step": 3220
    },
    {
      "epoch": 5.74,
      "learning_rate": 0.00032351353884562783,
      "loss": 0.8053,
      "step": 3240
    },
    {
      "epoch": 5.77,
      "learning_rate": 0.00031900881354781556,
      "loss": 0.8161,
      "step": 3260
    },
    {
      "epoch": 5.81,
      "learning_rate": 0.00031451477432136154,
      "loss": 0.8186,
      "step": 3280
    },
    {
      "epoch": 5.84,
      "learning_rate": 0.0003100320141150293,
      "loss": 0.8046,
      "step": 3300
    },
    {
      "epoch": 5.88,
      "learning_rate": 0.00030556112438941526,
      "loss": 0.8236,
      "step": 3320
    },
    {
      "epoch": 5.91,
      "learning_rate": 0.00030110269503891084,
      "loss": 0.8057,
      "step": 3340
    },
    {
      "epoch": 5.95,
      "learning_rate": 0.0002966573143138713,
      "loss": 0.8109,
      "step": 3360
    },
    {
      "epoch": 5.98,
      "learning_rate": 0.00029222556874300036,
      "loss": 0.8163,
      "step": 3380
    },
    {
      "epoch": 6.02,
      "learning_rate": 0.0002878080430559646,
      "loss": 0.7901,
      "step": 3400
    },
    {
      "epoch": 6.06,
      "learning_rate": 0.0002834053201062417,
      "loss": 0.7749,
      "step": 3420
    },
    {
      "epoch": 6.09,
      "learning_rate": 0.00027901798079421977,
      "loss": 0.7775,
      "step": 3440
    },
    {
      "epoch": 6.13,
      "learning_rate": 0.0002746466039905513,
      "loss": 0.7589,
      "step": 3460
    },
    {
      "epoch": 6.16,
      "learning_rate": 0.000270291766459777,
      "loss": 0.7742,
      "step": 3480
    },
    {
      "epoch": 6.2,
      "learning_rate": 0.00026595404278422684,
      "loss": 0.7914,
      "step": 3500
    },
    {
      "epoch": 6.23,
      "learning_rate": 0.00026163400528820836,
      "loss": 0.7909,
      "step": 3520
    },
    {
      "epoch": 6.27,
      "learning_rate": 0.0002573322239624947,
      "loss": 0.7653,
      "step": 3540
    },
    {
      "epoch": 6.3,
      "learning_rate": 0.000253049266389118,
      "loss": 0.7779,
      "step": 3560
    },
    {
      "epoch": 6.34,
      "learning_rate": 0.0002487856976664831,
      "loss": 0.7771,
      "step": 3580
    },
    {
      "epoch": 6.37,
      "learning_rate": 0.00024454208033480683,
      "loss": 0.771,
      "step": 3600
    },
    {
      "epoch": 6.41,
      "learning_rate": 0.00024031897430189695,
      "loss": 0.7788,
      "step": 3620
    },
    {
      "epoch": 6.45,
      "learning_rate": 0.00023611693676927606,
      "loss": 0.7798,
      "step": 3640
    },
    {
      "epoch": 6.48,
      "learning_rate": 0.00023193652215866429,
      "loss": 0.7779,
      "step": 3660
    },
    {
      "epoch": 6.52,
      "learning_rate": 0.00022777828203882875,
      "loss": 0.7784,
      "step": 3680
    },
    {
      "epoch": 6.55,
      "learning_rate": 0.00022364276505280794,
      "loss": 0.7946,
      "step": 3700
    },
    {
      "epoch": 6.59,
      "learning_rate": 0.0002195305168455239,
      "loss": 0.774,
      "step": 3720
    },
    {
      "epoch": 6.62,
      "learning_rate": 0.00021544207999178917,
      "loss": 0.7823,
      "step": 3740
    },
    {
      "epoch": 6.66,
      "learning_rate": 0.00021137799392471814,
      "loss": 0.7779,
      "step": 3760
    },
    {
      "epoch": 6.69,
      "learning_rate": 0.00020733879486455433,
      "loss": 0.7867,
      "step": 3780
    },
    {
      "epoch": 6.73,
      "learning_rate": 0.0002033250157479206,
      "loss": 0.7738,
      "step": 3800
    },
    {
      "epoch": 6.76,
      "learning_rate": 0.0001993371861575028,
      "loss": 0.7814,
      "step": 3820
    },
    {
      "epoch": 6.8,
      "learning_rate": 0.00019537583225217605,
      "loss": 0.7695,
      "step": 3840
    },
    {
      "epoch": 6.83,
      "learning_rate": 0.00019144147669758322,
      "loss": 0.7846,
      "step": 3860
    },
    {
      "epoch": 6.87,
      "learning_rate": 0.00018753463859717283,
      "loss": 0.7779,
      "step": 3880
    },
    {
      "epoch": 6.91,
      "learning_rate": 0.0001836558334237088,
      "loss": 0.7769,
      "step": 3900
    },
    {
      "epoch": 6.94,
      "learning_rate": 0.0001798055729512579,
      "loss": 0.782,
      "step": 3920
    },
    {
      "epoch": 6.98,
      "learning_rate": 0.00017598436518766596,
      "loss": 0.7683,
      "step": 3940
    },
    {
      "epoch": 7.01,
      "learning_rate": 0.0001721927143075305,
      "loss": 0.7674,
      "step": 3960
    },
    {
      "epoch": 7.05,
      "learning_rate": 0.00016843112058567935,
      "loss": 0.7427,
      "step": 3980
    },
    {
      "epoch": 7.08,
      "learning_rate": 0.00016470008033116443,
      "loss": 0.7627,
      "step": 4000
    },
    {
      "epoch": 7.12,
      "learning_rate": 0.00016100008582177705,
      "loss": 0.7541,
      "step": 4020
    },
    {
      "epoch": 7.15,
      "learning_rate": 0.00015733162523909707,
      "loss": 0.7654,
      "step": 4040
    },
    {
      "epoch": 7.19,
      "learning_rate": 0.0001536951826040813,
      "loss": 0.7382,
      "step": 4060
    },
    {
      "epoch": 7.22,
      "learning_rate": 0.0001500912377132013,
      "loss": 0.7555,
      "step": 4080
    },
    {
      "epoch": 7.26,
      "learning_rate": 0.00014652026607513848,
      "loss": 0.748,
      "step": 4100
    },
    {
      "epoch": 7.3,
      "learning_rate": 0.00014298273884804478,
      "loss": 0.753,
      "step": 4120
    },
    {
      "epoch": 7.33,
      "learning_rate": 0.00013947912277737808,
      "loss": 0.7594,
      "step": 4140
    },
    {
      "epoch": 7.37,
      "learning_rate": 0.00013600988013431832,
      "loss": 0.7387,
      "step": 4160
    },
    {
      "epoch": 7.4,
      "learning_rate": 0.00013257546865477572,
      "loss": 0.7551,
      "step": 4180
    },
    {
      "epoch": 7.44,
      "learning_rate": 0.00012917634147899607,
      "loss": 0.7609,
      "step": 4200
    },
    {
      "epoch": 7.47,
      "learning_rate": 0.00012581294709177327,
      "loss": 0.762,
      "step": 4220
    },
    {
      "epoch": 7.51,
      "learning_rate": 0.00012248572926327537,
      "loss": 0.755,
      "step": 4240
    },
    {
      "epoch": 7.54,
      "learning_rate": 0.00011919512699049314,
      "loss": 0.753,
      "step": 4260
    },
    {
      "epoch": 7.58,
      "learning_rate": 0.00011594157443931872,
      "loss": 0.7603,
      "step": 4280
    },
    {
      "epoch": 7.61,
      "learning_rate": 0.0001127255008872604,
      "loss": 0.7565,
      "step": 4300
    },
    {
      "epoch": 7.65,
      "learning_rate": 0.00010954733066680401,
      "loss": 0.7542,
      "step": 4320
    },
    {
      "epoch": 7.68,
      "learning_rate": 0.00010640748310942559,
      "loss": 0.7674,
      "step": 4340
    },
    {
      "epoch": 7.72,
      "learning_rate": 0.00010330637249026445,
      "loss": 0.7447,
      "step": 4360
    },
    {
      "epoch": 7.76,
      "learning_rate": 0.00010024440797346324,
      "loss": 0.7582,
      "step": 4380
    },
    {
      "epoch": 7.79,
      "learning_rate": 9.722199355818227e-05,
      "loss": 0.7598,
      "step": 4400
    },
    {
      "epoch": 7.83,
      "learning_rate": 9.423952802529564e-05,
      "loss": 0.7389,
      "step": 4420
    },
    {
      "epoch": 7.86,
      "learning_rate": 9.129740488477518e-05,
      "loss": 0.759,
      "step": 4440
    },
    {
      "epoch": 7.9,
      "learning_rate": 8.83960123237706e-05,
      "loss": 0.7442,
      "step": 4460
    },
    {
      "epoch": 7.93,
      "learning_rate": 8.553573315539188e-05,
      "loss": 0.7627,
      "step": 4480
    },
    {
      "epoch": 7.97,
      "learning_rate": 8.271694476819956e-05,
      "loss": 0.7531,
      "step": 4500
    },
    {
      "epoch": 8.0,
      "learning_rate": 7.994001907641262e-05,
      "loss": 0.7567,
      "step": 4520
    },
    {
      "epoch": 8.04,
      "learning_rate": 7.720532247083743e-05,
      "loss": 0.7274,
      "step": 4540
    },
    {
      "epoch": 8.07,
      "learning_rate": 7.451321577052533e-05,
      "loss": 0.7347,
      "step": 4560
    },
    {
      "epoch": 8.11,
      "learning_rate": 7.18640541751661e-05,
      "loss": 0.752,
      "step": 4580
    },
    {
      "epoch": 8.15,
      "learning_rate": 6.925818721822239e-05,
      "loss": 0.7293,
      "step": 4600
    },
    {
      "epoch": 8.18,
      "learning_rate": 6.669595872081211e-05,
      "loss": 0.7213,
      "step": 4620
    },
    {
      "epoch": 8.22,
      "learning_rate": 6.417770674634365e-05,
      "loss": 0.7246,
      "step": 4640
    },
    {
      "epoch": 8.25,
      "learning_rate": 6.170376355591204e-05,
      "loss": 0.7353,
      "step": 4660
    },
    {
      "epoch": 8.29,
      "learning_rate": 5.9274455564459896e-05,
      "loss": 0.7396,
      "step": 4680
    },
    {
      "epoch": 8.32,
      "learning_rate": 5.689010329770965e-05,
      "loss": 0.7401,
      "step": 4700
    },
    {
      "epoch": 8.36,
      "learning_rate": 5.455102134987304e-05,
      "loss": 0.7399,
      "step": 4720
    },
    {
      "epoch": 8.39,
      "learning_rate": 5.225751834214339e-05,
      "loss": 0.7427,
      "step": 4740
    },
    {
      "epoch": 8.43,
      "learning_rate": 5.000989688197555e-05,
      "loss": 0.7289,
      "step": 4760
    },
    {
      "epoch": 8.46,
      "learning_rate": 4.780845352315968e-05,
      "loss": 0.7398,
      "step": 4780
    },
    {
      "epoch": 8.5,
      "learning_rate": 4.565347872669339e-05,
      "loss": 0.7332,
      "step": 4800
    },
    {
      "epoch": 8.53,
      "learning_rate": 4.3545256822458445e-05,
      "loss": 0.7365,
      "step": 4820
    },
    {
      "epoch": 8.57,
      "learning_rate": 4.148406597170529e-05,
      "loss": 0.7347,
      "step": 4840
    },
    {
      "epoch": 8.61,
      "learning_rate": 3.947017813035254e-05,
      "loss": 0.7515,
      "step": 4860
    },
    {
      "epoch": 8.64,
      "learning_rate": 3.7503859013104806e-05,
      "loss": 0.7397,
      "step": 4880
    },
    {
      "epoch": 8.68,
      "learning_rate": 3.5585368058393834e-05,
      "loss": 0.7597,
      "step": 4900
    },
    {
      "epoch": 8.71,
      "learning_rate": 3.3714958394147975e-05,
      "loss": 0.7391,
      "step": 4920
    },
    {
      "epoch": 8.75,
      "learning_rate": 3.1892876804394144e-05,
      "loss": 0.7359,
      "step": 4940
    },
    {
      "epoch": 8.78,
      "learning_rate": 3.0119363696697078e-05,
      "loss": 0.7412,
      "step": 4960
    },
    {
      "epoch": 8.82,
      "learning_rate": 2.839465307043927e-05,
      "loss": 0.7485,
      "step": 4980
    },
    {
      "epoch": 8.85,
      "learning_rate": 2.6718972485947037e-05,
      "loss": 0.7534,
      "step": 5000
    },
    {
      "epoch": 8.89,
      "learning_rate": 2.5092543034466264e-05,
      "loss": 0.7405,
      "step": 5020
    },
    {
      "epoch": 8.92,
      "learning_rate": 2.3515579308990597e-05,
      "loss": 0.7415,
      "step": 5040
    },
    {
      "epoch": 8.96,
      "learning_rate": 2.1988289375948524e-05,
      "loss": 0.7309,
      "step": 5060
    },
    {
      "epoch": 9.0,
      "learning_rate": 2.0510874747750575e-05,
      "loss": 0.7418,
      "step": 5080
    },
    {
      "epoch": 9.03,
      "learning_rate": 1.9083530356201407e-05,
      "loss": 0.7469,
      "step": 5100
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.7706444526780585e-05,
      "loss": 0.7385,
      "step": 5120
    },
    {
      "epoch": 9.1,
      "learning_rate": 1.637979895379429e-05,
      "loss": 0.7333,
      "step": 5140
    },
    {
      "epoch": 9.14,
      "learning_rate": 1.5103768676402885e-05,
      "loss": 0.7314,
      "step": 5160
    },
    {
      "epoch": 9.17,
      "learning_rate": 1.38785220555254e-05,
      "loss": 0.74,
      "step": 5180
    },
    {
      "epoch": 9.21,
      "learning_rate": 1.270422075162645e-05,
      "loss": 0.7481,
      "step": 5200
    },
    {
      "epoch": 9.24,
      "learning_rate": 1.1581019703386143e-05,
      "loss": 0.7222,
      "step": 5220
    },
    {
      "epoch": 9.28,
      "learning_rate": 1.0509067107257365e-05,
      "loss": 0.7259,
      "step": 5240
    },
    {
      "epoch": 9.31,
      "learning_rate": 9.488504397912712e-06,
      "loss": 0.7309,
      "step": 5260
    },
    {
      "epoch": 9.35,
      "learning_rate": 8.51946622958324e-06,
      "loss": 0.7212,
      "step": 5280
    },
    {
      "epoch": 9.38,
      "learning_rate": 7.602080458292227e-06,
      "loss": 0.7254,
      "step": 5300
    },
    {
      "epoch": 9.42,
      "learning_rate": 6.7364681249854735e-06,
      "loss": 0.728,
      "step": 5320
    },
    {
      "epoch": 9.46,
      "learning_rate": 5.922743439561229e-06,
      "loss": 0.7341,
      "step": 5340
    },
    {
      "epoch": 9.49,
      "learning_rate": 5.161013765801137e-06,
      "loss": 0.721,
      "step": 5360
    },
    {
      "epoch": 9.53,
      "learning_rate": 4.451379607204453e-06,
      "loss": 0.725,
      "step": 5380
    },
    {
      "epoch": 9.56,
      "learning_rate": 3.7939345937275884e-06,
      "loss": 0.7399,
      "step": 5400
    },
    {
      "epoch": 9.6,
      "learning_rate": 3.1887654694303883e-06,
      "loss": 0.7301,
      "step": 5420
    },
    {
      "epoch": 9.63,
      "learning_rate": 2.635952081031201e-06,
      "loss": 0.7222,
      "step": 5440
    },
    {
      "epoch": 9.67,
      "learning_rate": 2.1355673673715716e-06,
      "loss": 0.7308,
      "step": 5460
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.6876773497926046e-06,
      "loss": 0.7417,
      "step": 5480
    },
    {
      "epoch": 9.74,
      "learning_rate": 1.292341123424423e-06,
      "loss": 0.7348,
      "step": 5500
    },
    {
      "epoch": 9.77,
      "learning_rate": 9.496108493884936e-07,
      "loss": 0.7278,
      "step": 5520
    },
    {
      "epoch": 9.81,
      "learning_rate": 6.595317479159313e-07,
      "loss": 0.7257,
      "step": 5540
    },
    {
      "epoch": 9.85,
      "learning_rate": 4.2214209238085054e-07,
      "loss": 0.7296,
      "step": 5560
    },
    {
      "epoch": 9.88,
      "learning_rate": 2.3747320425053786e-07,
      "loss": 0.7301,
      "step": 5580
    },
    {
      "epoch": 9.92,
      "learning_rate": 1.0554944895293517e-07,
      "loss": 0.739,
      "step": 5600
    },
    {
      "epoch": 9.95,
      "learning_rate": 2.638823266174484e-08,
      "loss": 0.7284,
      "step": 5620
    },
    {
      "epoch": 9.99,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 5640
    },
    {
      "epoch": 9.99,
      "step": 5640,
      "total_flos": 1.4686171215861645e+19,
      "train_loss": 0.8784972819876163,
      "train_runtime": 13393.5014,
      "train_samples_per_second": 53.949,
      "train_steps_per_second": 0.421
    }
  ],
  "max_steps": 5640,
  "num_train_epochs": 10,
  "total_flos": 1.4686171215861645e+19,
  "trial_name": null,
  "trial_params": null
}