{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6666370185893444,
  "eval_steps": 500,
  "global_step": 14990,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 9.986880725784932e-07,
      "loss": 5.3038,
      "step": 64
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.972649648670284e-07,
      "loss": 2.7293,
      "step": 128
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.959085653295383e-07,
      "loss": 1.6865,
      "step": 192
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.944854576180734e-07,
      "loss": 0.3066,
      "step": 256
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.930623499066086e-07,
      "loss": 0.2533,
      "step": 320
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.916392421951437e-07,
      "loss": 0.3489,
      "step": 384
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.902161344836786e-07,
      "loss": 0.2908,
      "step": 448
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.887930267722138e-07,
      "loss": 0.3666,
      "step": 512
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.873699190607489e-07,
      "loss": 0.3543,
      "step": 576
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.85946811349284e-07,
      "loss": 0.2938,
      "step": 640
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.84523703637819e-07,
      "loss": 0.2674,
      "step": 704
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.83100595926354e-07,
      "loss": 0.2186,
      "step": 768
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.816774882148892e-07,
      "loss": 0.3472,
      "step": 832
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.802543805034244e-07,
      "loss": 0.3157,
      "step": 896
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.788312727919595e-07,
      "loss": 0.432,
      "step": 960
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.774081650804944e-07,
      "loss": 0.2131,
      "step": 1024
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.759850573690296e-07,
      "loss": 0.249,
      "step": 1088
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.745619496575647e-07,
      "loss": 0.239,
      "step": 1152
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.731388419460998e-07,
      "loss": 0.3577,
      "step": 1216
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.717157342346347e-07,
      "loss": 0.1945,
      "step": 1280
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.702926265231699e-07,
      "loss": 0.3107,
      "step": 1344
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.68869518811705e-07,
      "loss": 0.1983,
      "step": 1408
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.674464111002402e-07,
      "loss": 0.188,
      "step": 1472
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.660233033887753e-07,
      "loss": 0.2659,
      "step": 1536
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.646001956773104e-07,
      "loss": 0.442,
      "step": 1600
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.631770879658454e-07,
      "loss": 0.1774,
      "step": 1664
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.617539802543805e-07,
      "loss": 0.2944,
      "step": 1728
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.603308725429156e-07,
      "loss": 0.3228,
      "step": 1792
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.589077648314505e-07,
      "loss": 0.2818,
      "step": 1856
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.574846571199857e-07,
      "loss": 0.2816,
      "step": 1920
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.560615494085208e-07,
      "loss": 0.3663,
      "step": 1984
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.54638441697056e-07,
      "loss": 0.2614,
      "step": 2048
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.53215333985591e-07,
      "loss": 0.2934,
      "step": 2112
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.51792226274126e-07,
      "loss": 0.4422,
      "step": 2176
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.503691185626613e-07,
      "loss": 0.2562,
      "step": 2240
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.489460108511963e-07,
      "loss": 0.3432,
      "step": 2304
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.475229031397314e-07,
      "loss": 0.3384,
      "step": 2368
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.460997954282665e-07,
      "loss": 0.1679,
      "step": 2432
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.446766877168015e-07,
      "loss": 0.1767,
      "step": 2496
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.432535800053366e-07,
      "loss": 0.2161,
      "step": 2560
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.418304722938717e-07,
      "loss": 0.2729,
      "step": 2624
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.404073645824068e-07,
      "loss": 0.3314,
      "step": 2688
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.389842568709418e-07,
      "loss": 0.242,
      "step": 2752
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.37561149159477e-07,
      "loss": 0.3446,
      "step": 2816
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.36138041448012e-07,
      "loss": 0.1292,
      "step": 2880
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.347149337365472e-07,
      "loss": 0.1854,
      "step": 2944
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.332918260250823e-07,
      "loss": 0.2536,
      "step": 3008
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.318687183136174e-07,
      "loss": 0.2224,
      "step": 3072
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.304456106021524e-07,
      "loss": 0.237,
      "step": 3136
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.290225028906876e-07,
      "loss": 0.3345,
      "step": 3200
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.275993951792226e-07,
      "loss": 0.175,
      "step": 3264
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.261762874677576e-07,
      "loss": 0.2693,
      "step": 3328
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.247531797562928e-07,
      "loss": 0.254,
      "step": 3392
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.233300720448278e-07,
      "loss": 0.2855,
      "step": 3456
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.219069643333629e-07,
      "loss": 0.1507,
      "step": 3520
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.20483856621898e-07,
      "loss": 0.2464,
      "step": 3584
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.190607489104332e-07,
      "loss": 0.2522,
      "step": 3648
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.176376411989682e-07,
      "loss": 0.2378,
      "step": 3712
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.162145334875034e-07,
      "loss": 0.2433,
      "step": 3776
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.147914257760384e-07,
      "loss": 0.2932,
      "step": 3840
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.133683180645735e-07,
      "loss": 0.2602,
      "step": 3904
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.119452103531086e-07,
      "loss": 0.3365,
      "step": 3968
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.105221026416437e-07,
      "loss": 0.1738,
      "step": 4032
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.090989949301787e-07,
      "loss": 0.3132,
      "step": 4096
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.076758872187137e-07,
      "loss": 0.2353,
      "step": 4160
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.062527795072489e-07,
      "loss": 0.2868,
      "step": 4224
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.048296717957839e-07,
      "loss": 0.1807,
      "step": 4288
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.034065640843192e-07,
      "loss": 0.3052,
      "step": 4352
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.019834563728542e-07,
      "loss": 0.2848,
      "step": 4416
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.005603486613893e-07,
      "loss": 0.2031,
      "step": 4480
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.991372409499244e-07,
      "loss": 0.2588,
      "step": 4544
    },
    {
      "epoch": 0.2,
      "learning_rate": 8.977141332384595e-07,
      "loss": 0.2356,
      "step": 4608
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.962910255269945e-07,
      "loss": 0.2445,
      "step": 4672
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.948679178155297e-07,
      "loss": 0.2537,
      "step": 4736
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.934448101040647e-07,
      "loss": 0.2345,
      "step": 4800
    },
    {
      "epoch": 0.22,
      "learning_rate": 8.920217023925998e-07,
      "loss": 0.2098,
      "step": 4864
    },
    {
      "epoch": 0.22,
      "learning_rate": 8.905985946811349e-07,
      "loss": 0.3487,
      "step": 4928
    },
    {
      "epoch": 0.22,
      "learning_rate": 8.891754869696699e-07,
      "loss": 0.1884,
      "step": 4992
    },
    {
      "epoch": 0.22,
      "learning_rate": 8.877523792582051e-07,
      "loss": 0.273,
      "step": 5056
    },
    {
      "epoch": 0.23,
      "learning_rate": 8.863292715467402e-07,
      "loss": 0.27,
      "step": 5120
    },
    {
      "epoch": 0.23,
      "learning_rate": 8.849061638352753e-07,
      "loss": 0.1494,
      "step": 5184
    },
    {
      "epoch": 0.23,
      "learning_rate": 8.834830561238103e-07,
      "loss": 0.1452,
      "step": 5248
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.820599484123455e-07,
      "loss": 0.177,
      "step": 5312
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.806368407008805e-07,
      "loss": 0.3317,
      "step": 5376
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.792137329894156e-07,
      "loss": 0.1638,
      "step": 5440
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.777906252779507e-07,
      "loss": 0.2928,
      "step": 5504
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.763675175664858e-07,
      "loss": 0.1851,
      "step": 5568
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.749444098550208e-07,
      "loss": 0.3108,
      "step": 5632
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.73521302143556e-07,
      "loss": 0.3002,
      "step": 5696
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.720981944320911e-07,
      "loss": 0.2726,
      "step": 5760
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.706750867206262e-07,
      "loss": 0.1775,
      "step": 5824
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.692519790091613e-07,
      "loss": 0.2632,
      "step": 5888
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.678288712976963e-07,
      "loss": 0.1625,
      "step": 5952
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.664057635862314e-07,
      "loss": 0.235,
      "step": 6016
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.649826558747665e-07,
      "loss": 0.2707,
      "step": 6080
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.635595481633016e-07,
      "loss": 0.1704,
      "step": 6144
    },
    {
      "epoch": 0.28,
      "learning_rate": 8.621364404518366e-07,
      "loss": 0.1925,
      "step": 6208
    },
    {
      "epoch": 0.28,
      "learning_rate": 8.607133327403718e-07,
      "loss": 0.2159,
      "step": 6272
    },
    {
      "epoch": 0.28,
      "learning_rate": 8.592902250289068e-07,
      "loss": 0.2228,
      "step": 6336
    },
    {
      "epoch": 0.28,
      "learning_rate": 8.578671173174419e-07,
      "loss": 0.2184,
      "step": 6400
    },
    {
      "epoch": 0.29,
      "learning_rate": 8.56444009605977e-07,
      "loss": 0.2803,
      "step": 6464
    },
    {
      "epoch": 0.29,
      "learning_rate": 8.550209018945122e-07,
      "loss": 0.1729,
      "step": 6528
    },
    {
      "epoch": 0.29,
      "learning_rate": 8.535977941830472e-07,
      "loss": 0.2702,
      "step": 6592
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.521746864715824e-07,
      "loss": 0.2938,
      "step": 6656
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.507515787601174e-07,
      "loss": 0.2692,
      "step": 6720
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.493284710486524e-07,
      "loss": 0.3317,
      "step": 6784
    },
    {
      "epoch": 0.3,
      "learning_rate": 8.479053633371876e-07,
      "loss": 0.2815,
      "step": 6848
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.464822556257226e-07,
      "loss": 0.2568,
      "step": 6912
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.450591479142577e-07,
      "loss": 0.2247,
      "step": 6976
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.436360402027927e-07,
      "loss": 0.2673,
      "step": 7040
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.422129324913279e-07,
      "loss": 0.1592,
      "step": 7104
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.407898247798629e-07,
      "loss": 0.2298,
      "step": 7168
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.393667170683982e-07,
      "loss": 0.2149,
      "step": 7232
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.379436093569332e-07,
      "loss": 0.3177,
      "step": 7296
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.365205016454683e-07,
      "loss": 0.2083,
      "step": 7360
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.350973939340034e-07,
      "loss": 0.2936,
      "step": 7424
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.336742862225385e-07,
      "loss": 0.2263,
      "step": 7488
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.322511785110735e-07,
      "loss": 0.2245,
      "step": 7552
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.308280707996085e-07,
      "loss": 0.1807,
      "step": 7616
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.294049630881437e-07,
      "loss": 0.2844,
      "step": 7680
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.279818553766787e-07,
      "loss": 0.1844,
      "step": 7744
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.265587476652138e-07,
      "loss": 0.1911,
      "step": 7808
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.251578760117405e-07,
      "loss": 0.2085,
      "step": 7872
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.237347683002757e-07,
      "loss": 0.1718,
      "step": 7936
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.223116605888107e-07,
      "loss": 0.2387,
      "step": 8000
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.208885528773458e-07,
      "loss": 0.1725,
      "step": 8064
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.19465445165881e-07,
      "loss": 0.1448,
      "step": 8128
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.180423374544161e-07,
      "loss": 0.154,
      "step": 8192
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.166192297429512e-07,
      "loss": 0.3209,
      "step": 8256
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.151961220314863e-07,
      "loss": 0.2067,
      "step": 8320
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.137730143200213e-07,
      "loss": 0.2147,
      "step": 8384
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.123499066085565e-07,
      "loss": 0.1771,
      "step": 8448
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.109267988970915e-07,
      "loss": 0.1212,
      "step": 8512
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.095036911856265e-07,
      "loss": 0.1944,
      "step": 8576
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.080805834741616e-07,
      "loss": 0.2416,
      "step": 8640
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.066574757626967e-07,
      "loss": 0.2004,
      "step": 8704
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.052343680512318e-07,
      "loss": 0.1667,
      "step": 8768
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.03811260339767e-07,
      "loss": 0.203,
      "step": 8832
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.023881526283021e-07,
      "loss": 0.2355,
      "step": 8896
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.009650449168371e-07,
      "loss": 0.2247,
      "step": 8960
    },
    {
      "epoch": 0.4,
      "learning_rate": 7.995419372053723e-07,
      "loss": 0.1768,
      "step": 9024
    },
    {
      "epoch": 0.4,
      "learning_rate": 7.981188294939073e-07,
      "loss": 0.2669,
      "step": 9088
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.966957217824424e-07,
      "loss": 0.2288,
      "step": 9152
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.952726140709774e-07,
      "loss": 0.2397,
      "step": 9216
    },
    {
      "epoch": 0.41,
      "learning_rate": 7.938495063595126e-07,
      "loss": 0.2276,
      "step": 9280
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.924263986480476e-07,
      "loss": 0.1367,
      "step": 9344
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.910032909365826e-07,
      "loss": 0.222,
      "step": 9408
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.895801832251178e-07,
      "loss": 0.2611,
      "step": 9472
    },
    {
      "epoch": 0.42,
      "learning_rate": 7.881570755136529e-07,
      "loss": 0.2373,
      "step": 9536
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.867339678021881e-07,
      "loss": 0.2052,
      "step": 9600
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.853108600907231e-07,
      "loss": 0.1998,
      "step": 9664
    },
    {
      "epoch": 0.43,
      "learning_rate": 7.838877523792582e-07,
      "loss": 0.2254,
      "step": 9728
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.824646446677932e-07,
      "loss": 0.1955,
      "step": 9792
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.810415369563284e-07,
      "loss": 0.1238,
      "step": 9856
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.796184292448634e-07,
      "loss": 0.1806,
      "step": 9920
    },
    {
      "epoch": 0.44,
      "learning_rate": 7.781953215333986e-07,
      "loss": 0.31,
      "step": 9984
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.767722138219336e-07,
      "loss": 0.2349,
      "step": 10048
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.753491061104687e-07,
      "loss": 0.2277,
      "step": 10112
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.739259983990037e-07,
      "loss": 0.2239,
      "step": 10176
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.725028906875389e-07,
      "loss": 0.2707,
      "step": 10240
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.71079782976074e-07,
      "loss": 0.1957,
      "step": 10304
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.69656675264609e-07,
      "loss": 0.1649,
      "step": 10368
    },
    {
      "epoch": 0.46,
      "learning_rate": 7.682335675531442e-07,
      "loss": 0.2773,
      "step": 10432
    },
    {
      "epoch": 0.47,
      "learning_rate": 7.668104598416792e-07,
      "loss": 0.2146,
      "step": 10496
    },
    {
      "epoch": 0.47,
      "learning_rate": 7.653873521302144e-07,
      "loss": 0.2134,
      "step": 10560
    },
    {
      "epoch": 0.47,
      "learning_rate": 7.639642444187494e-07,
      "loss": 0.2168,
      "step": 10624
    },
    {
      "epoch": 0.48,
      "learning_rate": 7.625411367072845e-07,
      "loss": 0.1122,
      "step": 10688
    },
    {
      "epoch": 0.48,
      "learning_rate": 7.611180289958195e-07,
      "loss": 0.1628,
      "step": 10752
    },
    {
      "epoch": 0.48,
      "learning_rate": 7.596949212843547e-07,
      "loss": 0.1566,
      "step": 10816
    },
    {
      "epoch": 0.48,
      "learning_rate": 7.582718135728897e-07,
      "loss": 0.1824,
      "step": 10880
    },
    {
      "epoch": 0.49,
      "learning_rate": 7.56848705861425e-07,
      "loss": 0.292,
      "step": 10944
    },
    {
      "epoch": 0.49,
      "learning_rate": 7.5542559814996e-07,
      "loss": 0.1534,
      "step": 11008
    },
    {
      "epoch": 0.49,
      "learning_rate": 7.54002490438495e-07,
      "loss": 0.2107,
      "step": 11072
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.525793827270302e-07,
      "loss": 0.1689,
      "step": 11136
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.511562750155652e-07,
      "loss": 0.3489,
      "step": 11200
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.497331673041003e-07,
      "loss": 0.1535,
      "step": 11264
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.483100595926353e-07,
      "loss": 0.1439,
      "step": 11328
    },
    {
      "epoch": 0.51,
      "learning_rate": 7.468869518811705e-07,
      "loss": 0.2032,
      "step": 11392
    },
    {
      "epoch": 0.51,
      "learning_rate": 7.454638441697055e-07,
      "loss": 0.4177,
      "step": 11456
    },
    {
      "epoch": 0.51,
      "learning_rate": 7.440407364582406e-07,
      "loss": 0.2552,
      "step": 11520
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.426176287467757e-07,
      "loss": 0.181,
      "step": 11584
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.411945210353108e-07,
      "loss": 0.1128,
      "step": 11648
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.39771413323846e-07,
      "loss": 0.2184,
      "step": 11712
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.383483056123811e-07,
      "loss": 0.3284,
      "step": 11776
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.369251979009161e-07,
      "loss": 0.1976,
      "step": 11840
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.355020901894511e-07,
      "loss": 0.1274,
      "step": 11904
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.340789824779863e-07,
      "loss": 0.1578,
      "step": 11968
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.326558747665213e-07,
      "loss": 0.2599,
      "step": 12032
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.312327670550564e-07,
      "loss": 0.1812,
      "step": 12096
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.298096593435915e-07,
      "loss": 0.2296,
      "step": 12160
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.283865516321266e-07,
      "loss": 0.2716,
      "step": 12224
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.269634439206616e-07,
      "loss": 0.3587,
      "step": 12288
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.255403362091968e-07,
      "loss": 0.1644,
      "step": 12352
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.241394645557235e-07,
      "loss": 0.2707,
      "step": 12416
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.227163568442586e-07,
      "loss": 0.1747,
      "step": 12480
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.212932491327936e-07,
      "loss": 0.2109,
      "step": 12544
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.198701414213289e-07,
      "loss": 0.2764,
      "step": 12608
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.184470337098639e-07,
      "loss": 0.1857,
      "step": 12672
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.170239259983991e-07,
      "loss": 0.2993,
      "step": 12736
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.156008182869341e-07,
      "loss": 0.3019,
      "step": 12800
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.141777105754691e-07,
      "loss": 0.2171,
      "step": 12864
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.127546028640042e-07,
      "loss": 0.1616,
      "step": 12928
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.113314951525393e-07,
      "loss": 0.2056,
      "step": 12992
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.099083874410744e-07,
      "loss": 0.1741,
      "step": 13056
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.084852797296094e-07,
      "loss": 0.1881,
      "step": 13120
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.070621720181446e-07,
      "loss": 0.3325,
      "step": 13184
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.056390643066796e-07,
      "loss": 0.1797,
      "step": 13248
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.042159565952149e-07,
      "loss": 0.1904,
      "step": 13312
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.027928488837499e-07,
      "loss": 0.182,
      "step": 13376
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.01369741172285e-07,
      "loss": 0.1893,
      "step": 13440
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.9994663346082e-07,
      "loss": 0.2757,
      "step": 13504
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.985235257493552e-07,
      "loss": 0.1684,
      "step": 13568
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.971004180378902e-07,
      "loss": 0.1678,
      "step": 13632
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.956773103264252e-07,
      "loss": 0.2164,
      "step": 13696
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.942542026149604e-07,
      "loss": 0.3885,
      "step": 13760
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.928310949034954e-07,
      "loss": 0.153,
      "step": 13824
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.914079871920305e-07,
      "loss": 0.211,
      "step": 13888
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.899848794805656e-07,
      "loss": 0.2952,
      "step": 13952
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.885617717691008e-07,
      "loss": 0.227,
      "step": 14016
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.871386640576358e-07,
      "loss": 0.1933,
      "step": 14080
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.85715556346171e-07,
      "loss": 0.2457,
      "step": 14144
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.84292448634706e-07,
      "loss": 0.1388,
      "step": 14208
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.828693409232412e-07,
      "loss": 0.175,
      "step": 14272
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.814462332117762e-07,
      "loss": 0.1735,
      "step": 14336
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.800231255003113e-07,
      "loss": 0.2027,
      "step": 14400
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.786000177888463e-07,
      "loss": 0.3326,
      "step": 14464
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.771769100773814e-07,
      "loss": 0.2391,
      "step": 14528
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.757538023659165e-07,
      "loss": 0.1766,
      "step": 14592
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.743306946544515e-07,
      "loss": 0.195,
      "step": 14656
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.729075869429868e-07,
      "loss": 0.2176,
      "step": 14720
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.714844792315218e-07,
      "loss": 0.2863,
      "step": 14784
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.700836075780485e-07,
      "loss": 0.2664,
      "step": 14848
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.686604998665835e-07,
      "loss": 0.1482,
      "step": 14912
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.672373921551188e-07,
      "loss": 0.3435,
      "step": 14976
    }
  ],
  "logging_steps": 64,
  "max_steps": 44972,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 14990,
  "total_flos": 3.1826333601792e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}