{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.336,
  "eval_steps": 500,
  "global_step": 2100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999999122701883e-05,
      "loss": 1.1656,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999964908081455e-05,
      "loss": 1.0536,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999992104320636e-05,
      "loss": 1.007,
      "step": 15
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999985963242432e-05,
      "loss": 0.9506,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999978067577844e-05,
      "loss": 1.0037,
      "step": 25
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999968417332415e-05,
      "loss": 1.0402,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999957012512916e-05,
      "loss": 1.0713,
      "step": 35
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999943853127351e-05,
      "loss": 0.991,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999928939184958e-05,
      "loss": 1.0264,
      "step": 45
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999912270696202e-05,
      "loss": 0.9843,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998938476727826e-05,
      "loss": 0.9999,
      "step": 55
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998736701276295e-05,
      "loss": 0.9625,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999851738074904e-05,
      "loss": 0.9601,
      "step": 65
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99982805153e-05,
      "loss": 0.9828,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998026105095405e-05,
      "loss": 0.9715,
      "step": 75
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997754150313815e-05,
      "loss": 0.9534,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999746465114609e-05,
      "loss": 1.0176,
      "step": 85
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999715760779541e-05,
      "loss": 0.9303,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9996833020477285e-05,
      "loss": 0.8311,
      "step": 95
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9996490889419514e-05,
      "loss": 0.9668,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999613121486222e-05,
      "loss": 0.9011,
      "step": 105
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999575399705783e-05,
      "loss": 0.9496,
      "step": 110
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999535923627109e-05,
      "loss": 0.9257,
      "step": 115
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999494693277907e-05,
      "loss": 0.9277,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999451708687114e-05,
      "loss": 1.0239,
      "step": 125
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999406969884897e-05,
      "loss": 0.9119,
      "step": 130
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999360476902656e-05,
      "loss": 0.9641,
      "step": 135
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999312229773022e-05,
      "loss": 0.872,
      "step": 140
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999262228529855e-05,
      "loss": 0.9925,
      "step": 145
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99921047320825e-05,
      "loss": 0.8343,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99915696384453e-05,
      "loss": 1.0187,
      "step": 155
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9991017004762496e-05,
      "loss": 0.9613,
      "step": 160
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9990446831421955e-05,
      "loss": 1.0168,
      "step": 165
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998985911882384e-05,
      "loss": 1.0068,
      "step": 170
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998925386738063e-05,
      "loss": 0.8927,
      "step": 175
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998863107751711e-05,
      "loss": 0.8887,
      "step": 180
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9987990749670395e-05,
      "loss": 0.9027,
      "step": 185
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998733288428987e-05,
      "loss": 0.985,
      "step": 190
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9986657481837277e-05,
      "loss": 0.9419,
      "step": 195
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9985964542786614e-05,
      "loss": 1.0437,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9985254067624215e-05,
      "loss": 0.9169,
      "step": 205
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998452605684874e-05,
      "loss": 0.891,
      "step": 210
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998378051097111e-05,
      "loss": 0.8631,
      "step": 215
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998301743051459e-05,
      "loss": 0.8966,
      "step": 220
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998223681601473e-05,
      "loss": 0.9779,
      "step": 225
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998143866801942e-05,
      "loss": 0.9671,
      "step": 230
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9980622987088795e-05,
      "loss": 0.9224,
      "step": 235
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997978977379536e-05,
      "loss": 0.9042,
      "step": 240
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9978939028723894e-05,
      "loss": 0.9182,
      "step": 245
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997807075247146e-05,
      "loss": 0.9076,
      "step": 250
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9977184945647473e-05,
      "loss": 1.0458,
      "step": 255
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997628160887361e-05,
      "loss": 0.8966,
      "step": 260
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997536074278387e-05,
      "loss": 0.8368,
      "step": 265
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997442234802456e-05,
      "loss": 0.8587,
      "step": 270
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9973466425254286e-05,
      "loss": 0.8835,
      "step": 275
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997249297514394e-05,
      "loss": 0.9406,
      "step": 280
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.997150199837671e-05,
      "loss": 0.904,
      "step": 285
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.997049349564814e-05,
      "loss": 0.9765,
      "step": 290
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996946746766601e-05,
      "loss": 0.8974,
      "step": 295
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996842391515044e-05,
      "loss": 0.868,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996736283883382e-05,
      "loss": 0.9357,
      "step": 305
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996628423946087e-05,
      "loss": 0.8766,
      "step": 310
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996518811778858e-05,
      "loss": 0.8624,
      "step": 315
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996407447458626e-05,
      "loss": 0.8213,
      "step": 320
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.99629433106355e-05,
      "loss": 0.9129,
      "step": 325
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.99617946267302e-05,
      "loss": 0.889,
      "step": 330
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996062842367654e-05,
      "loss": 0.9491,
      "step": 335
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.995944470229302e-05,
      "loss": 0.8673,
      "step": 340
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9958243463410414e-05,
      "loss": 0.9436,
      "step": 345
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9957024707871806e-05,
      "loss": 0.8837,
      "step": 350
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9955788436532545e-05,
      "loss": 0.8275,
      "step": 355
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995453465026032e-05,
      "loss": 0.8712,
      "step": 360
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9953263349935074e-05,
      "loss": 0.8837,
      "step": 365
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995197453644905e-05,
      "loss": 0.8224,
      "step": 370
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995066821070679e-05,
      "loss": 0.7904,
      "step": 375
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.994934437362513e-05,
      "loss": 0.8966,
      "step": 380
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.994800302613318e-05,
      "loss": 0.8933,
      "step": 385
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9946644169172355e-05,
      "loss": 0.8844,
      "step": 390
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9945267803696364e-05,
      "loss": 0.9564,
      "step": 395
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.994387393067117e-05,
      "loss": 0.8834,
      "step": 400
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9942462551075056e-05,
      "loss": 0.9638,
      "step": 405
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.994103366589859e-05,
      "loss": 0.8059,
      "step": 410
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9939587276144616e-05,
      "loss": 0.8947,
      "step": 415
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993812338282826e-05,
      "loss": 0.9133,
      "step": 420
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993664198697694e-05,
      "loss": 0.8608,
      "step": 425
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993514308963036e-05,
      "loss": 0.8155,
      "step": 430
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993362669184051e-05,
      "loss": 0.966,
      "step": 435
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993209279467164e-05,
      "loss": 0.941,
      "step": 440
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993054139920032e-05,
      "loss": 0.8789,
      "step": 445
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.992897250651535e-05,
      "loss": 0.8314,
      "step": 450
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.992738611771787e-05,
      "loss": 0.8749,
      "step": 455
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.992578223392124e-05,
      "loss": 0.9051,
      "step": 460
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.992416085625115e-05,
      "loss": 0.9447,
      "step": 465
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.992252198584554e-05,
      "loss": 0.8177,
      "step": 470
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9920865623854615e-05,
      "loss": 0.9483,
      "step": 475
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9919191771440905e-05,
      "loss": 0.9583,
      "step": 480
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991750042977916e-05,
      "loss": 0.9164,
      "step": 485
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991579160005644e-05,
      "loss": 0.9171,
      "step": 490
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991406528347206e-05,
      "loss": 0.9225,
      "step": 495
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991232148123761e-05,
      "loss": 0.8825,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991056019457697e-05,
      "loss": 0.8752,
      "step": 505
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.990878142472628e-05,
      "loss": 0.8622,
      "step": 510
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.990698517293395e-05,
      "loss": 0.9081,
      "step": 515
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.990517144046064e-05,
      "loss": 0.9334,
      "step": 520
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.990334022857932e-05,
      "loss": 0.7778,
      "step": 525
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9901491538575185e-05,
      "loss": 0.9381,
      "step": 530
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9899625371745726e-05,
      "loss": 0.8496,
      "step": 535
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9897741729400705e-05,
      "loss": 0.8012,
      "step": 540
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.989584061286211e-05,
      "loss": 1.0092,
      "step": 545
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9893922023464236e-05,
      "loss": 0.9182,
      "step": 550
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9891985962553606e-05,
      "loss": 0.9456,
      "step": 555
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.989003243148904e-05,
      "loss": 0.8976,
      "step": 560
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988806143164159e-05,
      "loss": 0.8505,
      "step": 565
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988607296439458e-05,
      "loss": 0.8364,
      "step": 570
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.98840670311436e-05,
      "loss": 0.8363,
      "step": 575
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988204363329648e-05,
      "loss": 0.8958,
      "step": 580
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988000277227334e-05,
      "loss": 0.8503,
      "step": 585
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.987794444950651e-05,
      "loss": 0.8074,
      "step": 590
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9875868666440604e-05,
      "loss": 0.8401,
      "step": 595
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.987377542453251e-05,
      "loss": 0.8383,
      "step": 600
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9871664725251314e-05,
      "loss": 0.7884,
      "step": 605
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.986953657007841e-05,
      "loss": 0.8736,
      "step": 610
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.98673909605074e-05,
      "loss": 0.9165,
      "step": 615
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.986522789804417e-05,
      "loss": 0.8373,
      "step": 620
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9863047384206835e-05,
      "loss": 0.8809,
      "step": 625
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9860849420525766e-05,
      "loss": 0.7754,
      "step": 630
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.985863400854358e-05,
      "loss": 0.7965,
      "step": 635
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9856401149815126e-05,
      "loss": 0.7485,
      "step": 640
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.985415084590752e-05,
      "loss": 0.9094,
      "step": 645
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.985188309840012e-05,
      "loss": 0.8977,
      "step": 650
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.98495979088845e-05,
      "loss": 0.8415,
      "step": 655
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9847295278964514e-05,
      "loss": 0.8873,
      "step": 660
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9844975210256217e-05,
      "loss": 0.976,
      "step": 665
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.984263770438793e-05,
      "loss": 0.9412,
      "step": 670
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.984028276300021e-05,
      "loss": 0.7609,
      "step": 675
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9837910387745845e-05,
      "loss": 0.8231,
      "step": 680
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9835520580289854e-05,
      "loss": 0.8444,
      "step": 685
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.98331133423095e-05,
      "loss": 0.8402,
      "step": 690
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9830688675494265e-05,
      "loss": 0.9041,
      "step": 695
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.982824658154589e-05,
      "loss": 0.8746,
      "step": 700
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9825787062178315e-05,
      "loss": 0.8755,
      "step": 705
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.982331011911774e-05,
      "loss": 0.9024,
      "step": 710
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.982081575410256e-05,
      "loss": 0.8178,
      "step": 715
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981830396888344e-05,
      "loss": 0.8501,
      "step": 720
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9815774765223226e-05,
      "loss": 0.8927,
      "step": 725
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981322814489703e-05,
      "loss": 0.8197,
      "step": 730
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981066410969215e-05,
      "loss": 0.8939,
      "step": 735
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980808266140813e-05,
      "loss": 0.8914,
      "step": 740
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980548380185674e-05,
      "loss": 0.8876,
      "step": 745
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980286753286195e-05,
      "loss": 0.9568,
      "step": 750
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980023385625996e-05,
      "loss": 0.8387,
      "step": 755
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.979758277389919e-05,
      "loss": 0.7872,
      "step": 760
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.979491428764026e-05,
      "loss": 0.8559,
      "step": 765
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.979222839935602e-05,
      "loss": 0.9081,
      "step": 770
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9789525110931545e-05,
      "loss": 0.8869,
      "step": 775
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9786804424264085e-05,
      "loss": 0.8222,
      "step": 780
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.978406634126315e-05,
      "loss": 0.9353,
      "step": 785
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9781310863850405e-05,
      "loss": 0.8829,
      "step": 790
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.977853799395976e-05,
      "loss": 0.8707,
      "step": 795
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.977574773353732e-05,
      "loss": 0.9178,
      "step": 800
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9772940084541405e-05,
      "loss": 0.8582,
      "step": 805
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.977011504894252e-05,
      "loss": 0.8964,
      "step": 810
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9767272628723396e-05,
      "loss": 0.6945,
      "step": 815
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9764412825878943e-05,
      "loss": 0.838,
      "step": 820
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.976153564241628e-05,
      "loss": 0.7924,
      "step": 825
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.975864108035474e-05,
      "loss": 0.75,
      "step": 830
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.975572914172582e-05,
      "loss": 0.8376,
      "step": 835
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.975279982857324e-05,
      "loss": 0.8017,
      "step": 840
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.97498531429529e-05,
      "loss": 0.7804,
      "step": 845
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9746889086932895e-05,
      "loss": 0.8152,
      "step": 850
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9743907662593524e-05,
      "loss": 0.8042,
      "step": 855
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.974090887202726e-05,
      "loss": 0.8054,
      "step": 860
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9737892717338774e-05,
      "loss": 0.8108,
      "step": 865
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9734859200644905e-05,
      "loss": 0.8086,
      "step": 870
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9731808324074717e-05,
      "loss": 1.0072,
      "step": 875
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.97287400897694e-05,
      "loss": 0.8551,
      "step": 880
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.972565449988239e-05,
      "loss": 0.6961,
      "step": 885
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.972255155657925e-05,
      "loss": 0.793,
      "step": 890
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9719431262037755e-05,
      "loss": 0.7999,
      "step": 895
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.971629361844785e-05,
      "loss": 0.8461,
      "step": 900
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9713138628011654e-05,
      "loss": 0.8966,
      "step": 905
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9709966292943455e-05,
      "loss": 0.9542,
      "step": 910
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9706776615469716e-05,
      "loss": 0.7947,
      "step": 915
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.970356959782909e-05,
      "loss": 0.914,
      "step": 920
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.970034524227238e-05,
      "loss": 0.8311,
      "step": 925
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9697103551062556e-05,
      "loss": 0.8571,
      "step": 930
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.969384452647477e-05,
      "loss": 0.9157,
      "step": 935
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.969056817079633e-05,
      "loss": 0.756,
      "step": 940
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.968727448632669e-05,
      "loss": 0.803,
      "step": 945
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.968396347537751e-05,
      "loss": 0.7612,
      "step": 950
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9680635140272575e-05,
      "loss": 0.759,
      "step": 955
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.967728948334784e-05,
      "loss": 0.7543,
      "step": 960
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9673926506951404e-05,
      "loss": 0.8701,
      "step": 965
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.967054621344356e-05,
      "loss": 0.8874,
      "step": 970
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.96671486051967e-05,
      "loss": 0.8365,
      "step": 975
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.966373368459541e-05,
      "loss": 0.7625,
      "step": 980
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.966030145403642e-05,
      "loss": 0.874,
      "step": 985
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.965685191592859e-05,
      "loss": 0.8794,
      "step": 990
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.965338507269294e-05,
      "loss": 0.8623,
      "step": 995
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.964990092676263e-05,
      "loss": 0.806,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.964639948058297e-05,
      "loss": 0.8127,
      "step": 1005
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.964288073661142e-05,
      "loss": 0.8409,
      "step": 1010
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.963934469731756e-05,
      "loss": 0.8788,
      "step": 1015
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.963579136518312e-05,
      "loss": 0.8135,
      "step": 1020
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9632220742701965e-05,
      "loss": 0.8258,
      "step": 1025
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.96286328323801e-05,
      "loss": 0.8422,
      "step": 1030
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.962502763673565e-05,
      "loss": 0.8156,
      "step": 1035
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.96214051582989e-05,
      "loss": 0.7606,
      "step": 1040
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.961776539961222e-05,
      "loss": 0.8642,
      "step": 1045
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9614108363230135e-05,
      "loss": 0.8211,
      "step": 1050
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.961043405171931e-05,
      "loss": 0.8341,
      "step": 1055
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.960674246765851e-05,
      "loss": 0.9787,
      "step": 1060
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9603033613638626e-05,
      "loss": 0.8983,
      "step": 1065
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.959930749226269e-05,
      "loss": 0.7955,
      "step": 1070
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.959556410614582e-05,
      "loss": 0.8516,
      "step": 1075
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.959180345791528e-05,
      "loss": 0.8472,
      "step": 1080
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.958802555021042e-05,
      "loss": 0.7969,
      "step": 1085
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.958423038568274e-05,
      "loss": 0.8582,
      "step": 1090
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.958041796699583e-05,
      "loss": 0.8102,
      "step": 1095
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9576588296825386e-05,
      "loss": 0.9105,
      "step": 1100
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.957274137785922e-05,
      "loss": 0.7878,
      "step": 1105
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.956887721279726e-05,
      "loss": 0.7164,
      "step": 1110
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.95649958043515e-05,
      "loss": 0.7911,
      "step": 1115
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.956109715524608e-05,
      "loss": 0.8541,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9557181268217227e-05,
      "loss": 0.7509,
      "step": 1125
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.955324814601324e-05,
      "loss": 0.8026,
      "step": 1130
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.954929779139455e-05,
      "loss": 0.9155,
      "step": 1135
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9545330207133664e-05,
      "loss": 0.8988,
      "step": 1140
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9541345396015193e-05,
      "loss": 0.7643,
      "step": 1145
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.953734336083583e-05,
      "loss": 0.8627,
      "step": 1150
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.953332410440435e-05,
      "loss": 0.8157,
      "step": 1155
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.952928762954161e-05,
      "loss": 0.8046,
      "step": 1160
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.952523393908059e-05,
      "loss": 0.8714,
      "step": 1165
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.952116303586631e-05,
      "loss": 0.795,
      "step": 1170
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.951707492275589e-05,
      "loss": 0.8813,
      "step": 1175
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.951296960261853e-05,
      "loss": 0.7807,
      "step": 1180
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9508847078335495e-05,
      "loss": 0.8237,
      "step": 1185
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9504707352800125e-05,
      "loss": 0.8191,
      "step": 1190
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.950055042891786e-05,
      "loss": 0.8957,
      "step": 1195
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.949637630960617e-05,
      "loss": 0.8884,
      "step": 1200
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9492184997794624e-05,
      "loss": 0.8054,
      "step": 1205
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.948797649642484e-05,
      "loss": 0.8055,
      "step": 1210
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.94837508084505e-05,
      "loss": 0.8127,
      "step": 1215
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9479507936837364e-05,
      "loss": 0.8751,
      "step": 1220
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.947524788456325e-05,
      "loss": 0.7474,
      "step": 1225
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.947097065461801e-05,
      "loss": 0.791,
      "step": 1230
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9466676250003576e-05,
      "loss": 0.8068,
      "step": 1235
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.946236467373392e-05,
      "loss": 0.8031,
      "step": 1240
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.945803592883509e-05,
      "loss": 0.852,
      "step": 1245
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 0.8317,
      "step": 1250
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.944932694531422e-05,
      "loss": 0.8149,
      "step": 1255
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9444946712804494e-05,
      "loss": 0.8335,
      "step": 1260
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9440549323890176e-05,
      "loss": 0.8224,
      "step": 1265
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.943613478165753e-05,
      "loss": 0.8027,
      "step": 1270
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.943170308920484e-05,
      "loss": 0.7803,
      "step": 1275
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9427254249642444e-05,
      "loss": 0.8416,
      "step": 1280
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9422788266092715e-05,
      "loss": 0.8652,
      "step": 1285
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.941830514169004e-05,
      "loss": 0.768,
      "step": 1290
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.941380487958086e-05,
      "loss": 0.7578,
      "step": 1295
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.940928748292363e-05,
      "loss": 0.78,
      "step": 1300
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9404752954888824e-05,
      "loss": 0.8722,
      "step": 1305
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.940020129865895e-05,
      "loss": 0.7997,
      "step": 1310
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.939563251742855e-05,
      "loss": 0.8842,
      "step": 1315
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.939104661440415e-05,
      "loss": 0.758,
      "step": 1320
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.938644359280433e-05,
      "loss": 0.8025,
      "step": 1325
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.938182345585966e-05,
      "loss": 0.8497,
      "step": 1330
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.937718620681273e-05,
      "loss": 0.744,
      "step": 1335
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9372531848918145e-05,
      "loss": 0.9146,
      "step": 1340
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.936786038544251e-05,
      "loss": 0.8052,
      "step": 1345
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9363171819664434e-05,
      "loss": 0.7883,
      "step": 1350
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.935846615487453e-05,
      "loss": 0.7909,
      "step": 1355
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.935374339437543e-05,
      "loss": 0.825,
      "step": 1360
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.934900354148173e-05,
      "loss": 0.8479,
      "step": 1365
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.934424659952006e-05,
      "loss": 0.7657,
      "step": 1370
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.933947257182901e-05,
      "loss": 0.7758,
      "step": 1375
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.933468146175918e-05,
      "loss": 0.8303,
      "step": 1380
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.932987327267316e-05,
      "loss": 0.8576,
      "step": 1385
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9325048007945526e-05,
      "loss": 0.8819,
      "step": 1390
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9320205670962814e-05,
      "loss": 0.7849,
      "step": 1395
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9315346265123594e-05,
      "loss": 0.8793,
      "step": 1400
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.931046979383835e-05,
      "loss": 0.8143,
      "step": 1405
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9305576260529607e-05,
      "loss": 0.8128,
      "step": 1410
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.930066566863182e-05,
      "loss": 0.8271,
      "step": 1415
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.929573802159143e-05,
      "loss": 0.8397,
      "step": 1420
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.929079332286685e-05,
      "loss": 0.8498,
      "step": 1425
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9285831575928465e-05,
      "loss": 0.7958,
      "step": 1430
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9280852784258624e-05,
      "loss": 0.7522,
      "step": 1435
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.927585695135162e-05,
      "loss": 0.8168,
      "step": 1440
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.927084408071373e-05,
      "loss": 0.7932,
      "step": 1445
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9265814175863186e-05,
      "loss": 0.8002,
      "step": 1450
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.926076724033016e-05,
      "loss": 0.7272,
      "step": 1455
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.925570327765678e-05,
      "loss": 0.8137,
      "step": 1460
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.925062229139714e-05,
      "loss": 0.7828,
      "step": 1465
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9245524285117274e-05,
      "loss": 0.8363,
      "step": 1470
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.924040926239515e-05,
      "loss": 0.7636,
      "step": 1475
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9235277226820695e-05,
      "loss": 0.8556,
      "step": 1480
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.923012818199576e-05,
      "loss": 0.8344,
      "step": 1485
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.922496213153416e-05,
      "loss": 0.8093,
      "step": 1490
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.92197790790616e-05,
      "loss": 0.7333,
      "step": 1495
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9214579028215776e-05,
      "loss": 0.765,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.920936198264627e-05,
      "loss": 0.8155,
      "step": 1505
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.920412794601461e-05,
      "loss": 0.7858,
      "step": 1510
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.919887692199423e-05,
      "loss": 0.8073,
      "step": 1515
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9193608914270515e-05,
      "loss": 0.76,
      "step": 1520
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9188323926540746e-05,
      "loss": 0.7419,
      "step": 1525
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.918302196251415e-05,
      "loss": 0.7942,
      "step": 1530
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9177703025911825e-05,
      "loss": 0.8357,
      "step": 1535
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.917236712046682e-05,
      "loss": 0.8111,
      "step": 1540
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9167014249924075e-05,
      "loss": 0.7631,
      "step": 1545
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.916164441804044e-05,
      "loss": 0.8014,
      "step": 1550
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.915625762858467e-05,
      "loss": 0.8524,
      "step": 1555
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9150853885337426e-05,
      "loss": 0.9321,
      "step": 1560
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.914543319209126e-05,
      "loss": 0.7353,
      "step": 1565
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.913999555265062e-05,
      "loss": 0.8118,
      "step": 1570
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.913454097083185e-05,
      "loss": 0.7877,
      "step": 1575
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9129069450463186e-05,
      "loss": 0.8804,
      "step": 1580
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.912358099538476e-05,
      "loss": 0.7856,
      "step": 1585
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.911807560944858e-05,
      "loss": 0.8434,
      "step": 1590
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.911255329651851e-05,
      "loss": 0.7606,
      "step": 1595
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.910701406047037e-05,
      "loss": 0.8149,
      "step": 1600
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9101457905191774e-05,
      "loss": 0.7455,
      "step": 1605
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.909588483458225e-05,
      "loss": 0.7615,
      "step": 1610
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.909029485255321e-05,
      "loss": 0.7266,
      "step": 1615
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9084687963027894e-05,
      "loss": 0.7535,
      "step": 1620
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.907906416994146e-05,
      "loss": 0.844,
      "step": 1625
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.907342347724087e-05,
      "loss": 0.6995,
      "step": 1630
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.906776588888502e-05,
      "loss": 0.7789,
      "step": 1635
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.906209140884459e-05,
      "loss": 0.8476,
      "step": 1640
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.905640004110216e-05,
      "loss": 0.7863,
      "step": 1645
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.905069178965215e-05,
      "loss": 0.7288,
      "step": 1650
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.904496665850084e-05,
      "loss": 0.8017,
      "step": 1655
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9039224651666325e-05,
      "loss": 0.8351,
      "step": 1660
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.903346577317859e-05,
      "loss": 0.7895,
      "step": 1665
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.902769002707942e-05,
      "loss": 0.7417,
      "step": 1670
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.902189741742247e-05,
      "loss": 0.743,
      "step": 1675
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.90160879482732e-05,
      "loss": 0.8576,
      "step": 1680
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9010261623708944e-05,
      "loss": 0.6991,
      "step": 1685
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9004418447818815e-05,
      "loss": 0.8136,
      "step": 1690
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.89985584247038e-05,
      "loss": 0.852,
      "step": 1695
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.899268155847667e-05,
      "loss": 0.8095,
      "step": 1700
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.898678785326205e-05,
      "loss": 0.7535,
      "step": 1705
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.898087731319636e-05,
      "loss": 0.8795,
      "step": 1710
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.897494994242785e-05,
      "loss": 0.7611,
      "step": 1715
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.896900574511657e-05,
      "loss": 0.6977,
      "step": 1720
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.89630447254344e-05,
      "loss": 0.7481,
      "step": 1725
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8957066887565e-05,
      "loss": 0.7695,
      "step": 1730
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8951072235703855e-05,
      "loss": 0.7652,
      "step": 1735
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.894506077405824e-05,
      "loss": 0.8048,
      "step": 1740
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8939032506847224e-05,
      "loss": 0.7717,
      "step": 1745
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.893298743830168e-05,
      "loss": 0.8183,
      "step": 1750
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.892692557266429e-05,
      "loss": 0.848,
      "step": 1755
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.892084691418947e-05,
      "loss": 0.8795,
      "step": 1760
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.891475146714347e-05,
      "loss": 0.8638,
      "step": 1765
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8908639235804324e-05,
      "loss": 0.7328,
      "step": 1770
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.890251022446181e-05,
      "loss": 0.8872,
      "step": 1775
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.889636443741752e-05,
      "loss": 0.8033,
      "step": 1780
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8890201878984796e-05,
      "loss": 0.8466,
      "step": 1785
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.888402255348876e-05,
      "loss": 0.7987,
      "step": 1790
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.887782646526631e-05,
      "loss": 0.8626,
      "step": 1795
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.887161361866608e-05,
      "loss": 0.8178,
      "step": 1800
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8865384018048494e-05,
      "loss": 0.7933,
      "step": 1805
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8859137667785735e-05,
      "loss": 0.9044,
      "step": 1810
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.885287457226172e-05,
      "loss": 0.7814,
      "step": 1815
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.884659473587213e-05,
      "loss": 0.7293,
      "step": 1820
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.88402981630244e-05,
      "loss": 0.9027,
      "step": 1825
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8833984858137715e-05,
      "loss": 0.839,
      "step": 1830
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.882765482564298e-05,
      "loss": 0.788,
      "step": 1835
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8821308069982867e-05,
      "loss": 0.7907,
      "step": 1840
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8814944595611776e-05,
      "loss": 0.8303,
      "step": 1845
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.880856440699582e-05,
      "loss": 0.8853,
      "step": 1850
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.880216750861288e-05,
      "loss": 0.7154,
      "step": 1855
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8795753904952534e-05,
      "loss": 0.7562,
      "step": 1860
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8789323600516104e-05,
      "loss": 0.8036,
      "step": 1865
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.878287659981662e-05,
      "loss": 0.6877,
      "step": 1870
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.877641290737884e-05,
      "loss": 0.8234,
      "step": 1875
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8769932527739225e-05,
      "loss": 0.9163,
      "step": 1880
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8763435465445964e-05,
      "loss": 0.918,
      "step": 1885
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8756921725058934e-05,
      "loss": 0.8422,
      "step": 1890
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.875039131114975e-05,
      "loss": 0.8704,
      "step": 1895
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.874384422830167e-05,
      "loss": 0.772,
      "step": 1900
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8737280481109724e-05,
      "loss": 0.7513,
      "step": 1905
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.873070007418059e-05,
      "loss": 0.7595,
      "step": 1910
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.872410301213265e-05,
      "loss": 0.7778,
      "step": 1915
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.871748929959598e-05,
      "loss": 0.8576,
      "step": 1920
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.871085894121233e-05,
      "loss": 0.843,
      "step": 1925
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.870421194163515e-05,
      "loss": 0.8354,
      "step": 1930
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.869754830552956e-05,
      "loss": 0.7699,
      "step": 1935
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8690868037572346e-05,
      "loss": 0.7549,
      "step": 1940
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8684171142451986e-05,
      "loss": 0.8241,
      "step": 1945
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.867745762486861e-05,
      "loss": 0.8673,
      "step": 1950
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8670727489534034e-05,
      "loss": 0.8047,
      "step": 1955
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8663980741171724e-05,
      "loss": 0.811,
      "step": 1960
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.86572173845168e-05,
      "loss": 0.6631,
      "step": 1965
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.865043742431605e-05,
      "loss": 0.787,
      "step": 1970
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.864364086532792e-05,
      "loss": 0.8221,
      "step": 1975
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.863682771232248e-05,
      "loss": 0.7942,
      "step": 1980
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.862999797008149e-05,
      "loss": 0.788,
      "step": 1985
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.862315164339829e-05,
      "loss": 0.8077,
      "step": 1990
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.861628873707792e-05,
      "loss": 0.8043,
      "step": 1995
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.860940925593703e-05,
      "loss": 0.804,
      "step": 2000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8602513204803896e-05,
      "loss": 0.855,
      "step": 2005
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.859560058851844e-05,
      "loss": 0.798,
      "step": 2010
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.858867141193219e-05,
      "loss": 0.88,
      "step": 2015
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8581725679908317e-05,
      "loss": 0.6838,
      "step": 2020
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8574763397321614e-05,
      "loss": 0.8288,
      "step": 2025
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.856778456905846e-05,
      "loss": 0.8132,
      "step": 2030
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8560789200016884e-05,
      "loss": 1.0326,
      "step": 2035
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.855377729510648e-05,
      "loss": 0.7732,
      "step": 2040
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.85467488592485e-05,
      "loss": 0.7843,
      "step": 2045
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8539703897375755e-05,
      "loss": 0.8764,
      "step": 2050
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8532642414432674e-05,
      "loss": 0.8422,
      "step": 2055
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.852556441537528e-05,
      "loss": 0.8688,
      "step": 2060
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.851846990517118e-05,
      "loss": 0.8208,
      "step": 2065
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.851135888879958e-05,
      "loss": 0.8681,
      "step": 2070
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8504231371251255e-05,
      "loss": 0.7863,
      "step": 2075
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.849708735752859e-05,
      "loss": 0.7919,
      "step": 2080
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8489926852645505e-05,
      "loss": 0.7518,
      "step": 2085
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.848274986162754e-05,
      "loss": 0.776,
      "step": 2090
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.847555638951177e-05,
      "loss": 0.7668,
      "step": 2095
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.846834644134686e-05,
      "loss": 0.7175,
      "step": 2100
    }
  ],
  "logging_steps": 5,
  "max_steps": 18750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 1.2438877592027136e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}