{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.4997766860205448,
  "eval_steps": 500,
  "global_step": 10074,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3.722084367245657e-08,
      "loss": 0.2836,
      "step": 32
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.692307692307692e-08,
      "loss": 0.2111,
      "step": 64
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1662531017369727e-07,
      "loss": 0.2145,
      "step": 96
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.563275434243176e-07,
      "loss": 0.2375,
      "step": 128
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9602977667493795e-07,
      "loss": 0.1839,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.3573200992555832e-07,
      "loss": 0.2864,
      "step": 192
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.7543424317617863e-07,
      "loss": 0.1799,
      "step": 224
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.1513647642679897e-07,
      "loss": 0.1609,
      "step": 256
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.5483870967741936e-07,
      "loss": 0.1434,
      "step": 288
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.945409429280397e-07,
      "loss": 0.1699,
      "step": 320
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.3424317617866004e-07,
      "loss": 0.1898,
      "step": 352
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.739454094292804e-07,
      "loss": 0.1665,
      "step": 384
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.136476426799007e-07,
      "loss": 0.1358,
      "step": 416
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.533498759305211e-07,
      "loss": 0.1713,
      "step": 448
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.930521091811415e-07,
      "loss": 0.1817,
      "step": 480
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.327543424317618e-07,
      "loss": 0.149,
      "step": 512
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.724565756823821e-07,
      "loss": 0.165,
      "step": 544
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.121588089330024e-07,
      "loss": 0.2282,
      "step": 576
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.518610421836227e-07,
      "loss": 0.1666,
      "step": 608
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.915632754342431e-07,
      "loss": 0.1229,
      "step": 640
    },
    {
      "epoch": 0.1,
      "learning_rate": 8.312655086848634e-07,
      "loss": 0.2118,
      "step": 672
    },
    {
      "epoch": 0.1,
      "learning_rate": 8.709677419354838e-07,
      "loss": 0.1869,
      "step": 704
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.106699751861042e-07,
      "loss": 0.2196,
      "step": 736
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.503722084367245e-07,
      "loss": 0.0936,
      "step": 768
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.90074441687345e-07,
      "loss": 0.1839,
      "step": 800
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.980994615140957e-07,
      "loss": 0.1675,
      "step": 832
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.955654101995564e-07,
      "loss": 0.1595,
      "step": 864
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.930313588850174e-07,
      "loss": 0.1556,
      "step": 896
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.905764966740576e-07,
      "loss": 0.1711,
      "step": 928
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.880424453595185e-07,
      "loss": 0.1777,
      "step": 960
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.855083940449792e-07,
      "loss": 0.2031,
      "step": 992
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.829743427304402e-07,
      "loss": 0.1529,
      "step": 1024
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.804402914159011e-07,
      "loss": 0.1365,
      "step": 1056
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.77906240101362e-07,
      "loss": 0.1922,
      "step": 1088
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.75372188786823e-07,
      "loss": 0.133,
      "step": 1120
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.728381374722837e-07,
      "loss": 0.1692,
      "step": 1152
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.703040861577447e-07,
      "loss": 0.1022,
      "step": 1184
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.677700348432054e-07,
      "loss": 0.2052,
      "step": 1216
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.652359835286664e-07,
      "loss": 0.1546,
      "step": 1248
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.627019322141273e-07,
      "loss": 0.149,
      "step": 1280
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.601678808995883e-07,
      "loss": 0.1281,
      "step": 1312
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.57633829585049e-07,
      "loss": 0.1437,
      "step": 1344
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.5509977827051e-07,
      "loss": 0.2097,
      "step": 1376
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.525657269559708e-07,
      "loss": 0.1308,
      "step": 1408
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.500316756414317e-07,
      "loss": 0.1691,
      "step": 1440
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.474976243268927e-07,
      "loss": 0.2319,
      "step": 1472
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.449635730123534e-07,
      "loss": 0.2226,
      "step": 1504
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.424295216978143e-07,
      "loss": 0.1789,
      "step": 1536
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.398954703832752e-07,
      "loss": 0.1932,
      "step": 1568
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.373614190687361e-07,
      "loss": 0.1718,
      "step": 1600
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.34827367754197e-07,
      "loss": 0.156,
      "step": 1632
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.322933164396578e-07,
      "loss": 0.1512,
      "step": 1664
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.297592651251187e-07,
      "loss": 0.0968,
      "step": 1696
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.272252138105796e-07,
      "loss": 0.0932,
      "step": 1728
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.246911624960405e-07,
      "loss": 0.2464,
      "step": 1760
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.221571111815014e-07,
      "loss": 0.2036,
      "step": 1792
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.196230598669623e-07,
      "loss": 0.1245,
      "step": 1824
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.170890085524232e-07,
      "loss": 0.1097,
      "step": 1856
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.14554957237884e-07,
      "loss": 0.1844,
      "step": 1888
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.120209059233449e-07,
      "loss": 0.1114,
      "step": 1920
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.094868546088058e-07,
      "loss": 0.1992,
      "step": 1952
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.069528032942667e-07,
      "loss": 0.1721,
      "step": 1984
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.044187519797275e-07,
      "loss": 0.1473,
      "step": 2016
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.018847006651884e-07,
      "loss": 0.1865,
      "step": 2048
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.993506493506493e-07,
      "loss": 0.1583,
      "step": 2080
    },
    {
      "epoch": 0.31,
      "learning_rate": 8.968165980361102e-07,
      "loss": 0.1866,
      "step": 2112
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.942825467215711e-07,
      "loss": 0.1617,
      "step": 2144
    },
    {
      "epoch": 0.32,
      "learning_rate": 8.917484954070319e-07,
      "loss": 0.1189,
      "step": 2176
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.892144440924928e-07,
      "loss": 0.148,
      "step": 2208
    },
    {
      "epoch": 0.33,
      "learning_rate": 8.866803927779537e-07,
      "loss": 0.131,
      "step": 2240
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.841463414634146e-07,
      "loss": 0.2261,
      "step": 2272
    },
    {
      "epoch": 0.34,
      "learning_rate": 8.816122901488755e-07,
      "loss": 0.1742,
      "step": 2304
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.790782388343364e-07,
      "loss": 0.164,
      "step": 2336
    },
    {
      "epoch": 0.35,
      "learning_rate": 8.765441875197972e-07,
      "loss": 0.1161,
      "step": 2368
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.74010136205258e-07,
      "loss": 0.1636,
      "step": 2400
    },
    {
      "epoch": 0.36,
      "learning_rate": 8.71476084890719e-07,
      "loss": 0.2416,
      "step": 2432
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.689420335761799e-07,
      "loss": 0.1632,
      "step": 2464
    },
    {
      "epoch": 0.37,
      "learning_rate": 8.664079822616408e-07,
      "loss": 0.1477,
      "step": 2496
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.638739309471016e-07,
      "loss": 0.2083,
      "step": 2528
    },
    {
      "epoch": 0.38,
      "learning_rate": 8.613398796325625e-07,
      "loss": 0.1599,
      "step": 2560
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.588058283180234e-07,
      "loss": 0.1817,
      "step": 2592
    },
    {
      "epoch": 0.39,
      "learning_rate": 8.562717770034843e-07,
      "loss": 0.1005,
      "step": 2624
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.537377256889452e-07,
      "loss": 0.168,
      "step": 2656
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.51203674374406e-07,
      "loss": 0.2418,
      "step": 2688
    },
    {
      "epoch": 0.4,
      "learning_rate": 8.486696230598669e-07,
      "loss": 0.1881,
      "step": 2720
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.461355717453278e-07,
      "loss": 0.1829,
      "step": 2752
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.436015204307887e-07,
      "loss": 0.1073,
      "step": 2784
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.410674691162496e-07,
      "loss": 0.1324,
      "step": 2816
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.385334178017105e-07,
      "loss": 0.2077,
      "step": 2848
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.359993664871713e-07,
      "loss": 0.2248,
      "step": 2880
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.334653151726322e-07,
      "loss": 0.1337,
      "step": 2912
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.30931263858093e-07,
      "loss": 0.1906,
      "step": 2944
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.28397212543554e-07,
      "loss": 0.1893,
      "step": 2976
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.259423503325942e-07,
      "loss": 0.2029,
      "step": 3008
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.234082990180551e-07,
      "loss": 0.157,
      "step": 3040
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.208742477035159e-07,
      "loss": 0.1433,
      "step": 3072
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.183401963889769e-07,
      "loss": 0.1689,
      "step": 3104
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.158061450744377e-07,
      "loss": 0.2012,
      "step": 3136
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.132720937598986e-07,
      "loss": 0.175,
      "step": 3168
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.107380424453595e-07,
      "loss": 0.1961,
      "step": 3200
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.082039911308203e-07,
      "loss": 0.2547,
      "step": 3232
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.056699398162813e-07,
      "loss": 0.1935,
      "step": 3264
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.031358885017421e-07,
      "loss": 0.2149,
      "step": 3296
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.00601837187203e-07,
      "loss": 0.1809,
      "step": 3328
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.980677858726639e-07,
      "loss": 0.2072,
      "step": 3360
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.955337345581247e-07,
      "loss": 0.2116,
      "step": 3392
    },
    {
      "epoch": 0.51,
      "learning_rate": 7.929996832435857e-07,
      "loss": 0.1737,
      "step": 3424
    },
    {
      "epoch": 0.51,
      "learning_rate": 7.904656319290464e-07,
      "loss": 0.2219,
      "step": 3456
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.879315806145074e-07,
      "loss": 0.1849,
      "step": 3488
    },
    {
      "epoch": 0.52,
      "learning_rate": 7.853975292999683e-07,
      "loss": 0.1884,
      "step": 3520
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.828634779854292e-07,
      "loss": 0.2192,
      "step": 3552
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.803294266708901e-07,
      "loss": 0.1958,
      "step": 3584
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.777953753563509e-07,
      "loss": 0.1433,
      "step": 3616
    },
    {
      "epoch": 0.54,
      "learning_rate": 7.752613240418118e-07,
      "loss": 0.2151,
      "step": 3648
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.727272727272727e-07,
      "loss": 0.1675,
      "step": 3680
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.701932214127336e-07,
      "loss": 0.1586,
      "step": 3712
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.676591700981945e-07,
      "loss": 0.2881,
      "step": 3744
    },
    {
      "epoch": 0.56,
      "learning_rate": 7.651251187836553e-07,
      "loss": 0.196,
      "step": 3776
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.625910674691162e-07,
      "loss": 0.1285,
      "step": 3808
    },
    {
      "epoch": 0.57,
      "learning_rate": 7.60057016154577e-07,
      "loss": 0.2262,
      "step": 3840
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.57522964840038e-07,
      "loss": 0.2309,
      "step": 3872
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.549889135254989e-07,
      "loss": 0.1533,
      "step": 3904
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.524548622109597e-07,
      "loss": 0.1297,
      "step": 3936
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.499208108964206e-07,
      "loss": 0.1808,
      "step": 3968
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.473867595818814e-07,
      "loss": 0.2401,
      "step": 4000
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.448527082673424e-07,
      "loss": 0.2507,
      "step": 4032
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.423186569528033e-07,
      "loss": 0.1562,
      "step": 4064
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.397846056382642e-07,
      "loss": 0.1912,
      "step": 4096
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.373297434273043e-07,
      "loss": 0.1703,
      "step": 4128
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.347956921127653e-07,
      "loss": 0.1471,
      "step": 4160
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.322616407982262e-07,
      "loss": 0.1539,
      "step": 4192
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.297275894836869e-07,
      "loss": 0.1521,
      "step": 4224
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.271935381691479e-07,
      "loss": 0.2623,
      "step": 4256
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.246594868546087e-07,
      "loss": 0.1753,
      "step": 4288
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.221254355400697e-07,
      "loss": 0.1945,
      "step": 4320
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.195913842255306e-07,
      "loss": 0.2153,
      "step": 4352
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.170573329109915e-07,
      "loss": 0.2841,
      "step": 4384
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.145232815964523e-07,
      "loss": 0.1759,
      "step": 4416
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.119892302819131e-07,
      "loss": 0.2214,
      "step": 4448
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.094551789673741e-07,
      "loss": 0.188,
      "step": 4480
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.069211276528349e-07,
      "loss": 0.1579,
      "step": 4512
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.043870763382959e-07,
      "loss": 0.2213,
      "step": 4544
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.018530250237567e-07,
      "loss": 0.2042,
      "step": 4576
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.993189737092175e-07,
      "loss": 0.1852,
      "step": 4608
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.967849223946785e-07,
      "loss": 0.1716,
      "step": 4640
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.942508710801393e-07,
      "loss": 0.1645,
      "step": 4672
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.917168197656003e-07,
      "loss": 0.1986,
      "step": 4704
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.89182768451061e-07,
      "loss": 0.2531,
      "step": 4736
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.866487171365219e-07,
      "loss": 0.1792,
      "step": 4768
    },
    {
      "epoch": 0.71,
      "learning_rate": 6.841146658219829e-07,
      "loss": 0.1843,
      "step": 4800
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.815806145074437e-07,
      "loss": 0.2175,
      "step": 4832
    },
    {
      "epoch": 0.72,
      "learning_rate": 6.790465631929047e-07,
      "loss": 0.2083,
      "step": 4864
    },
    {
      "epoch": 0.73,
      "learning_rate": 6.765125118783655e-07,
      "loss": 0.1729,
      "step": 4896
    },
    {
      "epoch": 0.73,
      "learning_rate": 6.739784605638263e-07,
      "loss": 0.1849,
      "step": 4928
    },
    {
      "epoch": 0.74,
      "learning_rate": 6.714444092492873e-07,
      "loss": 0.2374,
      "step": 4960
    },
    {
      "epoch": 0.74,
      "learning_rate": 6.689103579347481e-07,
      "loss": 0.241,
      "step": 4992
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.663763066202091e-07,
      "loss": 0.1853,
      "step": 5024
    },
    {
      "epoch": 0.75,
      "learning_rate": 6.638422553056699e-07,
      "loss": 0.1957,
      "step": 5056
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.613082039911308e-07,
      "loss": 0.2052,
      "step": 5088
    },
    {
      "epoch": 0.76,
      "learning_rate": 6.587741526765917e-07,
      "loss": 0.2321,
      "step": 5120
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.562401013620525e-07,
      "loss": 0.1804,
      "step": 5152
    },
    {
      "epoch": 0.77,
      "learning_rate": 6.537060500475135e-07,
      "loss": 0.1842,
      "step": 5184
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.511719987329743e-07,
      "loss": 0.2388,
      "step": 5216
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.486379474184352e-07,
      "loss": 0.2417,
      "step": 5248
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.46103896103896e-07,
      "loss": 0.2224,
      "step": 5280
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.435698447893569e-07,
      "loss": 0.2029,
      "step": 5312
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.410357934748179e-07,
      "loss": 0.2807,
      "step": 5344
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.385017421602787e-07,
      "loss": 0.192,
      "step": 5376
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.359676908457397e-07,
      "loss": 0.1848,
      "step": 5408
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.334336395312004e-07,
      "loss": 0.2143,
      "step": 5440
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.308995882166613e-07,
      "loss": 0.2421,
      "step": 5472
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.283655369021223e-07,
      "loss": 0.1724,
      "step": 5504
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.258314855875831e-07,
      "loss": 0.1207,
      "step": 5536
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.232974342730441e-07,
      "loss": 0.2259,
      "step": 5568
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.207633829585048e-07,
      "loss": 0.2504,
      "step": 5600
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.182293316439658e-07,
      "loss": 0.188,
      "step": 5632
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.156952803294266e-07,
      "loss": 0.1893,
      "step": 5664
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.131612290148875e-07,
      "loss": 0.1905,
      "step": 5696
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.106271777003485e-07,
      "loss": 0.2594,
      "step": 5728
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.080931263858092e-07,
      "loss": 0.3084,
      "step": 5760
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.055590750712702e-07,
      "loss": 0.1925,
      "step": 5792
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.03025023756731e-07,
      "loss": 0.186,
      "step": 5824
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.004909724421919e-07,
      "loss": 0.2302,
      "step": 5856
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.979569211276529e-07,
      "loss": 0.1371,
      "step": 5888
    },
    {
      "epoch": 0.88,
      "learning_rate": 5.954228698131137e-07,
      "loss": 0.231,
      "step": 5920
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.928888184985746e-07,
      "loss": 0.2012,
      "step": 5952
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.903547671840354e-07,
      "loss": 0.2006,
      "step": 5984
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.878207158694963e-07,
      "loss": 0.215,
      "step": 6016
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.852866645549572e-07,
      "loss": 0.1471,
      "step": 6048
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.827526132404181e-07,
      "loss": 0.2364,
      "step": 6080
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.80218561925879e-07,
      "loss": 0.2881,
      "step": 6112
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.776845106113398e-07,
      "loss": 0.1536,
      "step": 6144
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.751504592968008e-07,
      "loss": 0.2317,
      "step": 6176
    },
    {
      "epoch": 0.92,
      "learning_rate": 5.726164079822616e-07,
      "loss": 0.1952,
      "step": 6208
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.700823566677225e-07,
      "loss": 0.1602,
      "step": 6240
    },
    {
      "epoch": 0.93,
      "learning_rate": 5.675483053531834e-07,
      "loss": 0.212,
      "step": 6272
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.650142540386442e-07,
      "loss": 0.2401,
      "step": 6304
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.624802027241052e-07,
      "loss": 0.1992,
      "step": 6336
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.59946151409566e-07,
      "loss": 0.2616,
      "step": 6368
    },
    {
      "epoch": 0.95,
      "learning_rate": 5.574121000950269e-07,
      "loss": 0.146,
      "step": 6400
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.548780487804878e-07,
      "loss": 0.2081,
      "step": 6432
    },
    {
      "epoch": 0.96,
      "learning_rate": 5.523439974659486e-07,
      "loss": 0.207,
      "step": 6464
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.498099461514096e-07,
      "loss": 0.2631,
      "step": 6496
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.472758948368704e-07,
      "loss": 0.1721,
      "step": 6528
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.447418435223313e-07,
      "loss": 0.1908,
      "step": 6560
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.422869813113715e-07,
      "loss": 0.2238,
      "step": 6592
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.397529299968325e-07,
      "loss": 0.2524,
      "step": 6624
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.372188786822933e-07,
      "loss": 0.1968,
      "step": 6656
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.346848273677542e-07,
      "loss": 0.2379,
      "step": 6688
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.32150776053215e-07,
      "loss": 0.246,
      "step": 6720
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.296167247386759e-07,
      "loss": 0.104,
      "step": 6752
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.270826734241369e-07,
      "loss": 0.102,
      "step": 6784
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.245486221095977e-07,
      "loss": 0.1432,
      "step": 6816
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.220145707950586e-07,
      "loss": 0.1262,
      "step": 6848
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.194805194805194e-07,
      "loss": 0.1299,
      "step": 6880
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.169464681659803e-07,
      "loss": 0.1319,
      "step": 6912
    },
    {
      "epoch": 1.03,
      "learning_rate": 5.144124168514412e-07,
      "loss": 0.105,
      "step": 6944
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.118783655369021e-07,
      "loss": 0.1233,
      "step": 6976
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.09344314222363e-07,
      "loss": 0.0922,
      "step": 7008
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.068102629078239e-07,
      "loss": 0.144,
      "step": 7040
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.042762115932847e-07,
      "loss": 0.1828,
      "step": 7072
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.017421602787456e-07,
      "loss": 0.1097,
      "step": 7104
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.992081089642065e-07,
      "loss": 0.1867,
      "step": 7136
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.966740576496675e-07,
      "loss": 0.1338,
      "step": 7168
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.941400063351283e-07,
      "loss": 0.1535,
      "step": 7200
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.916059550205891e-07,
      "loss": 0.1861,
      "step": 7232
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.8907190370605e-07,
      "loss": 0.1284,
      "step": 7264
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.865378523915109e-07,
      "loss": 0.1037,
      "step": 7296
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.840038010769719e-07,
      "loss": 0.1217,
      "step": 7328
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.814697497624327e-07,
      "loss": 0.1469,
      "step": 7360
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.789356984478935e-07,
      "loss": 0.1218,
      "step": 7392
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.764016471333545e-07,
      "loss": 0.1486,
      "step": 7424
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.738675958188153e-07,
      "loss": 0.0796,
      "step": 7456
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.7141273360785554e-07,
      "loss": 0.1163,
      "step": 7488
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.688786822933164e-07,
      "loss": 0.0821,
      "step": 7520
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.663446309787773e-07,
      "loss": 0.1701,
      "step": 7552
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.638105796642382e-07,
      "loss": 0.1002,
      "step": 7584
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.613557174532784e-07,
      "loss": 0.0914,
      "step": 7616
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.5882166613873927e-07,
      "loss": 0.1374,
      "step": 7648
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.562876148242002e-07,
      "loss": 0.1142,
      "step": 7680
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.5375356350966105e-07,
      "loss": 0.1351,
      "step": 7712
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.5121951219512194e-07,
      "loss": 0.1513,
      "step": 7744
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.486854608805828e-07,
      "loss": 0.0998,
      "step": 7776
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.461514095660437e-07,
      "loss": 0.1115,
      "step": 7808
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.4361735825150457e-07,
      "loss": 0.1373,
      "step": 7840
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.4108330693696546e-07,
      "loss": 0.1614,
      "step": 7872
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.3854925562242635e-07,
      "loss": 0.0826,
      "step": 7904
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.3601520430788724e-07,
      "loss": 0.1129,
      "step": 7936
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.334811529933481e-07,
      "loss": 0.0825,
      "step": 7968
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.3094710167880897e-07,
      "loss": 0.1497,
      "step": 8000
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.284130503642698e-07,
      "loss": 0.1575,
      "step": 8032
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.2587899904973075e-07,
      "loss": 0.1377,
      "step": 8064
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.2334494773519165e-07,
      "loss": 0.152,
      "step": 8096
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.208108964206525e-07,
      "loss": 0.1142,
      "step": 8128
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.182768451061134e-07,
      "loss": 0.1458,
      "step": 8160
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.1574279379157427e-07,
      "loss": 0.1216,
      "step": 8192
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.132087424770351e-07,
      "loss": 0.112,
      "step": 8224
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.1067469116249605e-07,
      "loss": 0.1517,
      "step": 8256
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.081406398479569e-07,
      "loss": 0.1046,
      "step": 8288
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.056065885334178e-07,
      "loss": 0.1538,
      "step": 8320
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.030725372188787e-07,
      "loss": 0.1734,
      "step": 8352
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.005384859043395e-07,
      "loss": 0.1236,
      "step": 8384
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.980044345898004e-07,
      "loss": 0.1129,
      "step": 8416
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.9547038327526135e-07,
      "loss": 0.1193,
      "step": 8448
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.929363319607222e-07,
      "loss": 0.1124,
      "step": 8480
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.904022806461831e-07,
      "loss": 0.103,
      "step": 8512
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.878682293316439e-07,
      "loss": 0.1526,
      "step": 8544
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.853341780171048e-07,
      "loss": 0.1747,
      "step": 8576
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.8280012670256575e-07,
      "loss": 0.0711,
      "step": 8608
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.802660753880266e-07,
      "loss": 0.131,
      "step": 8640
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.777320240734875e-07,
      "loss": 0.0695,
      "step": 8672
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.751979727589484e-07,
      "loss": 0.1176,
      "step": 8704
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.726639214444092e-07,
      "loss": 0.1141,
      "step": 8736
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.701298701298701e-07,
      "loss": 0.1437,
      "step": 8768
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.6759581881533095e-07,
      "loss": 0.1273,
      "step": 8800
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.650617675007919e-07,
      "loss": 0.0765,
      "step": 8832
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.625277161862528e-07,
      "loss": 0.1322,
      "step": 8864
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.599936648717136e-07,
      "loss": 0.1643,
      "step": 8896
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.574596135571745e-07,
      "loss": 0.1005,
      "step": 8928
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.549255622426354e-07,
      "loss": 0.1545,
      "step": 8960
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.523915109280963e-07,
      "loss": 0.0755,
      "step": 8992
    },
    {
      "epoch": 1.34,
      "learning_rate": 3.498574596135572e-07,
      "loss": 0.0917,
      "step": 9024
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.47323408299018e-07,
      "loss": 0.0893,
      "step": 9056
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.447893569844789e-07,
      "loss": 0.152,
      "step": 9088
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.422553056699398e-07,
      "loss": 0.1154,
      "step": 9120
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.3972125435540065e-07,
      "loss": 0.0817,
      "step": 9152
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.371872030408616e-07,
      "loss": 0.083,
      "step": 9184
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.346531517263225e-07,
      "loss": 0.1084,
      "step": 9216
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.321191004117833e-07,
      "loss": 0.1514,
      "step": 9248
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.295850490972442e-07,
      "loss": 0.1195,
      "step": 9280
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2705099778270505e-07,
      "loss": 0.0939,
      "step": 9312
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2451694646816595e-07,
      "loss": 0.1227,
      "step": 9344
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.219828951536269e-07,
      "loss": 0.1342,
      "step": 9376
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.1944884383908773e-07,
      "loss": 0.1214,
      "step": 9408
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.169147925245486e-07,
      "loss": 0.1624,
      "step": 9440
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.143807412100095e-07,
      "loss": 0.1261,
      "step": 9472
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.1184668989547035e-07,
      "loss": 0.1674,
      "step": 9504
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.0931263858093124e-07,
      "loss": 0.1613,
      "step": 9536
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.0677858726639213e-07,
      "loss": 0.1519,
      "step": 9568
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.04244535951853e-07,
      "loss": 0.1107,
      "step": 9600
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.017104846373139e-07,
      "loss": 0.0992,
      "step": 9632
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.9917643332277476e-07,
      "loss": 0.0921,
      "step": 9664
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.9664238200823565e-07,
      "loss": 0.1478,
      "step": 9696
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.9410833069369654e-07,
      "loss": 0.1268,
      "step": 9728
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.9157427937915743e-07,
      "loss": 0.1086,
      "step": 9760
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.890402280646183e-07,
      "loss": 0.1913,
      "step": 9792
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.8650617675007916e-07,
      "loss": 0.1626,
      "step": 9824
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.8397212543554005e-07,
      "loss": 0.1647,
      "step": 9856
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.8143807412100095e-07,
      "loss": 0.1245,
      "step": 9888
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.789040228064618e-07,
      "loss": 0.1326,
      "step": 9920
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.7636997149192273e-07,
      "loss": 0.1095,
      "step": 9952
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.738359201773836e-07,
      "loss": 0.1089,
      "step": 9984
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.7130186886284446e-07,
      "loss": 0.1074,
      "step": 10016
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.6876781754830535e-07,
      "loss": 0.1084,
      "step": 10048
    }
  ],
  "logging_steps": 32,
  "max_steps": 13434,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 3358,
  "total_flos": 4.27776497270784e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}