{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.24560541735377706,
  "global_step": 3500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.3156,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 1.2717,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.428571428571429e-05,
      "loss": 1.1819,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 1.038,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.9192,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.857142857142858e-05,
      "loss": 0.8204,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 8e-05,
      "loss": 0.7139,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.142857142857143e-05,
      "loss": 0.6191,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00010285714285714286,
      "loss": 0.559,
      "step": 90
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00011428571428571428,
      "loss": 0.5244,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00012571428571428572,
      "loss": 0.5149,
      "step": 110
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00013714285714285716,
      "loss": 0.4988,
      "step": 120
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00014857142857142857,
      "loss": 0.475,
      "step": 130
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016,
      "loss": 0.4527,
      "step": 140
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017142857142857143,
      "loss": 0.4563,
      "step": 150
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018285714285714286,
      "loss": 0.4443,
      "step": 160
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001942857142857143,
      "loss": 0.4247,
      "step": 170
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00020571428571428572,
      "loss": 0.4369,
      "step": 180
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00021714285714285713,
      "loss": 0.4371,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00022857142857142857,
      "loss": 0.42,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00024,
      "loss": 0.4221,
      "step": 210
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00025142857142857145,
      "loss": 0.4162,
      "step": 220
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00026285714285714286,
      "loss": 0.4065,
      "step": 230
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0002742857142857143,
      "loss": 0.4094,
      "step": 240
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00028571428571428574,
      "loss": 0.4017,
      "step": 250
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00029714285714285715,
      "loss": 0.4291,
      "step": 260
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003085714285714286,
      "loss": 0.4198,
      "step": 270
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00032,
      "loss": 0.4069,
      "step": 280
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003314285714285715,
      "loss": 0.4118,
      "step": 290
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00034285714285714285,
      "loss": 0.3921,
      "step": 300
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00035428571428571426,
      "loss": 0.3911,
      "step": 310
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00036571428571428573,
      "loss": 0.4013,
      "step": 320
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00037714285714285714,
      "loss": 0.4114,
      "step": 330
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0003885714285714286,
      "loss": 0.4098,
      "step": 340
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0004,
      "loss": 0.4114,
      "step": 350
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039999005339118863,
      "loss": 0.3901,
      "step": 360
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003999602145541047,
      "loss": 0.405,
      "step": 370
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039991048645670067,
      "loss": 0.3946,
      "step": 380
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039984087404523587,
      "loss": 0.398,
      "step": 390
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039975138424378447,
      "loss": 0.3979,
      "step": 400
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000399642025953547,
      "loss": 0.3996,
      "step": 410
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003995128100519649,
      "loss": 0.3935,
      "step": 420
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003993637493916382,
      "loss": 0.4008,
      "step": 430
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039919485879904784,
      "loss": 0.3875,
      "step": 440
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039900615507308033,
      "loss": 0.3918,
      "step": 450
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039879765698335704,
      "loss": 0.3907,
      "step": 460
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003985693852683675,
      "loss": 0.3899,
      "step": 470
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00039832136263340603,
      "loss": 0.3916,
      "step": 480
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0003980536137483141,
      "loss": 0.3971,
      "step": 490
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003977661652450257,
      "loss": 0.3915,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039745904571491917,
      "loss": 0.3838,
      "step": 510
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003971322857059726,
      "loss": 0.378,
      "step": 520
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003967859177197259,
      "loss": 0.3795,
      "step": 530
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003964199762080478,
      "loss": 0.3697,
      "step": 540
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039603449756970877,
      "loss": 0.3785,
      "step": 550
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039562952014676116,
      "loss": 0.3868,
      "step": 560
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003952050842207249,
      "loss": 0.3938,
      "step": 570
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003947612320085813,
      "loss": 0.385,
      "step": 580
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003942980076585735,
      "loss": 0.3884,
      "step": 590
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003938154572458156,
      "loss": 0.3835,
      "step": 600
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039331362876770945,
      "loss": 0.3884,
      "step": 610
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039279257213917066,
      "loss": 0.3873,
      "step": 620
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003922523391876638,
      "loss": 0.3824,
      "step": 630
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00039169298364804715,
      "loss": 0.3853,
      "step": 640
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003911145611572282,
      "loss": 0.3783,
      "step": 650
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003905171292486292,
      "loss": 0.3806,
      "step": 660
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038990074734646525,
      "loss": 0.3794,
      "step": 670
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038926547675983286,
      "loss": 0.3883,
      "step": 680
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003886113806766121,
      "loss": 0.3764,
      "step": 690
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003879385241571817,
      "loss": 0.3872,
      "step": 700
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038724697412794747,
      "loss": 0.381,
      "step": 710
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038653679937468557,
      "loss": 0.3777,
      "step": 720
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003858080705357005,
      "loss": 0.378,
      "step": 730
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038506086009479937,
      "loss": 0.3838,
      "step": 740
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003842952423740815,
      "loss": 0.3748,
      "step": 750
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038351129352654677,
      "loss": 0.3732,
      "step": 760
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003827090915285202,
      "loss": 0.3677,
      "step": 770
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00038188871617189646,
      "loss": 0.3635,
      "step": 780
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000381050249056203,
      "loss": 0.369,
      "step": 790
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003801937735804838,
      "loss": 0.3823,
      "step": 800
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003793193749350042,
      "loss": 0.385,
      "step": 810
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003784271400927767,
      "loss": 0.3785,
      "step": 820
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037751715780091086,
      "loss": 0.3779,
      "step": 830
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037658951857178543,
      "loss": 0.3604,
      "step": 840
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037564431467404573,
      "loss": 0.3742,
      "step": 850
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037468164012342597,
      "loss": 0.365,
      "step": 860
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003737015906733978,
      "loss": 0.382,
      "step": 870
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003727042638056462,
      "loss": 0.3804,
      "step": 880
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037168975872037323,
      "loss": 0.3794,
      "step": 890
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00037065817632643116,
      "loss": 0.3794,
      "step": 900
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003696096192312852,
      "loss": 0.3725,
      "step": 910
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003685441917308078,
      "loss": 0.3698,
      "step": 920
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003674619997989047,
      "loss": 0.3679,
      "step": 930
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003663631510769738,
      "loss": 0.3671,
      "step": 940
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.000365247754863199,
      "loss": 0.3719,
      "step": 950
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003641159221016783,
      "loss": 0.3685,
      "step": 960
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00036296776537138905,
      "loss": 0.3703,
      "step": 970
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003618033988749895,
      "loss": 0.3708,
      "step": 980
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00036062293842746036,
      "loss": 0.3642,
      "step": 990
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003594265014445845,
      "loss": 0.3692,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003582142069312683,
      "loss": 0.3753,
      "step": 1010
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003569861754697045,
      "loss": 0.3673,
      "step": 1020
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003557425292073788,
      "loss": 0.3728,
      "step": 1030
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003544833918449199,
      "loss": 0.3711,
      "step": 1040
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003532088886237956,
      "loss": 0.3657,
      "step": 1050
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00035191914631385563,
      "loss": 0.3714,
      "step": 1060
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00035061429320072223,
      "loss": 0.3705,
      "step": 1070
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00034929445907303007,
      "loss": 0.3672,
      "step": 1080
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003479597752095168,
      "loss": 0.3628,
      "step": 1090
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003466103743659653,
      "loss": 0.3647,
      "step": 1100
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003452463907619986,
      "loss": 0.3699,
      "step": 1110
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003438679600677302,
      "loss": 0.3699,
      "step": 1120
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000342475219390269,
      "loss": 0.3659,
      "step": 1130
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003410683072600818,
      "loss": 0.3708,
      "step": 1140
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00033964736361721457,
      "loss": 0.3667,
      "step": 1150
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00033821252979737297,
      "loss": 0.3628,
      "step": 1160
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00033676394851786404,
      "loss": 0.3618,
      "step": 1170
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00033530176386340106,
      "loss": 0.372,
      "step": 1180
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00033382612127177166,
      "loss": 0.3637,
      "step": 1190
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003323371675193719,
      "loss": 0.3623,
      "step": 1200
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0003308350507066069,
      "loss": 0.3684,
      "step": 1210
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003293199202431599,
      "loss": 0.3648,
      "step": 1220
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003277919268331312,
      "loss": 0.3555,
      "step": 1230
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003262512224600478,
      "loss": 0.3592,
      "step": 1240
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00032469796037174674,
      "loss": 0.3587,
      "step": 1250
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003231322950651316,
      "loss": 0.3645,
      "step": 1260
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00032155438227080604,
      "loss": 0.3647,
      "step": 1270
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00031996437893758276,
      "loss": 0.3563,
      "step": 1280
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00031836244321687356,
      "loss": 0.3613,
      "step": 1290
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000316748734446958,
      "loss": 0.3726,
      "step": 1300
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00031512341313713474,
      "loss": 0.377,
      "step": 1310
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003134866409517564,
      "loss": 0.3629,
      "step": 1320
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003118385806941494,
      "loss": 0.363,
      "step": 1330
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003101793962904205,
      "loss": 0.353,
      "step": 1340
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0003085092527731519,
      "loss": 0.3537,
      "step": 1350
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00030682831626498575,
      "loss": 0.3606,
      "step": 1360
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00030513675396210094,
      "loss": 0.357,
      "step": 1370
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00030343473411758253,
      "loss": 0.3603,
      "step": 1380
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0003017224260246863,
      "loss": 0.3618,
      "step": 1390
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00030000000000000003,
      "loss": 0.3674,
      "step": 1400
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002982676273665023,
      "loss": 0.3711,
      "step": 1410
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002965254804365222,
      "loss": 0.366,
      "step": 1420
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002947737324945997,
      "loss": 0.3727,
      "step": 1430
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00029318909447116496,
      "loss": 0.372,
      "step": 1440
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002914195854117389,
      "loss": 0.3683,
      "step": 1450
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002896409832037748,
      "loss": 0.356,
      "step": 1460
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002878534647578768,
      "loss": 0.3599,
      "step": 1470
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002860572078715121,
      "loss": 0.3605,
      "step": 1480
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0002842523912113264,
      "loss": 0.3595,
      "step": 1490
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00028243919429537265,
      "loss": 0.3592,
      "step": 1500
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00028061779747525504,
      "loss": 0.3582,
      "step": 1510
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00027878838191819024,
      "loss": 0.3473,
      "step": 1520
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002769511295889872,
      "loss": 0.3577,
      "step": 1530
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000275106223231948,
      "loss": 0.3465,
      "step": 1540
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00027325384635269094,
      "loss": 0.3547,
      "step": 1550
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00027139418319989785,
      "loss": 0.3583,
      "step": 1560
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002695274187469878,
      "loss": 0.3558,
      "step": 1570
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002676537386737183,
      "loss": 0.3547,
      "step": 1580
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00026577332934771665,
      "loss": 0.3591,
      "step": 1590
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002638863778059426,
      "loss": 0.3443,
      "step": 1600
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00026199307173608433,
      "loss": 0.3481,
      "step": 1610
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002600935994578904,
      "loss": 0.349,
      "step": 1620
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0002581881499044377,
      "loss": 0.3565,
      "step": 1630
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002562769126033394,
      "loss": 0.3539,
      "step": 1640
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00025436007765789327,
      "loss": 0.353,
      "step": 1650
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00025243783572817297,
      "loss": 0.3612,
      "step": 1660
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002505103780120636,
      "loss": 0.3573,
      "step": 1670
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002485778962262443,
      "loss": 0.3536,
      "step": 1680
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00024664058258711853,
      "loss": 0.3632,
      "step": 1690
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002446986297916954,
      "loss": 0.3522,
      "step": 1700
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00024275223099842291,
      "loss": 0.3509,
      "step": 1710
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00024080157980797484,
      "loss": 0.348,
      "step": 1720
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002388468702439944,
      "loss": 0.3556,
      "step": 1730
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00023688829673379534,
      "loss": 0.3561,
      "step": 1740
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00023492605408902297,
      "loss": 0.3532,
      "step": 1750
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00023296033748627712,
      "loss": 0.346,
      "step": 1760
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0002309913424476986,
      "loss": 0.3521,
      "step": 1770
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00022901926482152138,
      "loss": 0.3505,
      "step": 1780
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00022704430076259246,
      "loss": 0.3444,
      "step": 1790
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00022506664671286087,
      "loss": 0.3467,
      "step": 1800
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00022308649938183864,
      "loss": 0.3561,
      "step": 1810
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00022110405572703466,
      "loss": 0.3529,
      "step": 1820
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00021911951293436416,
      "loss": 0.348,
      "step": 1830
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00021713306839853545,
      "loss": 0.3449,
      "step": 1840
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002151449197034157,
      "loss": 0.3525,
      "step": 1850
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002131552646023783,
      "loss": 0.3431,
      "step": 1860
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00021116430099863277,
      "loss": 0.35,
      "step": 1870
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002091722269255404,
      "loss": 0.3397,
      "step": 1880
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0002071792405269165,
      "loss": 0.3532,
      "step": 1890
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00020518554003732167,
      "loss": 0.3487,
      "step": 1900
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00020319132376234462,
      "loss": 0.3508,
      "step": 1910
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00020119679005887702,
      "loss": 0.35,
      "step": 1920
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019920213731538394,
      "loss": 0.3479,
      "step": 1930
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019720756393217098,
      "loss": 0.3439,
      "step": 1940
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019521326830164998,
      "loss": 0.337,
      "step": 1950
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019321944878860587,
      "loss": 0.3419,
      "step": 1960
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000191226303710466,
      "loss": 0.3508,
      "step": 1970
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00018923403131757439,
      "loss": 0.3474,
      "step": 1980
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00018724282977347235,
      "loss": 0.3451,
      "step": 1990
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00018525289713518817,
      "loss": 0.3449,
      "step": 2000
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00018326443133353693,
      "loss": 0.3437,
      "step": 2010
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00018127763015343332,
      "loss": 0.3425,
      "step": 2020
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017929269121421857,
      "loss": 0.3478,
      "step": 2030
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017730981195000406,
      "loss": 0.3408,
      "step": 2040
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017532918959003353,
      "loss": 0.3452,
      "step": 2050
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00017335102113906505,
      "loss": 0.3395,
      "step": 2060
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00017137550335777612,
      "loss": 0.3462,
      "step": 2070
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001694028327431924,
      "loss": 0.3397,
      "step": 2080
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001674332055091431,
      "loss": 0.3434,
      "step": 2090
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001654668175667442,
      "loss": 0.3431,
      "step": 2100
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00016350386450491208,
      "loss": 0.3326,
      "step": 2110
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00016154454157090884,
      "loss": 0.3392,
      "step": 2120
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00015958904365092225,
      "loss": 0.339,
      "step": 2130
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00015763756525068065,
      "loss": 0.336,
      "step": 2140
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00015569030047610656,
      "loss": 0.3363,
      "step": 2150
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001537474430140096,
      "loss": 0.3373,
      "step": 2160
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001518091861128213,
      "loss": 0.3419,
      "step": 2170
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00014987572256337336,
      "loss": 0.3493,
      "step": 2180
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001479472446797216,
      "loss": 0.3334,
      "step": 2190
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00014602394428001712,
      "loss": 0.3384,
      "step": 2200
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00014410601266742691,
      "loss": 0.3435,
      "step": 2210
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00014219364061110565,
      "loss": 0.341,
      "step": 2220
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00014028701832722104,
      "loss": 0.3317,
      "step": 2230
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00013838633546003302,
      "loss": 0.3422,
      "step": 2240
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00013649178106303115,
      "loss": 0.3457,
      "step": 2250
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00013460354358013,
      "loss": 0.3356,
      "step": 2260
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001327218108269255,
      "loss": 0.3377,
      "step": 2270
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00013084676997201342,
      "loss": 0.3304,
      "step": 2280
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00012897860751837263,
      "loss": 0.3413,
      "step": 2290
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00012711750928481443,
      "loss": 0.3433,
      "step": 2300
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00012526366038749956,
      "loss": 0.3419,
      "step": 2310
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000123417245221526,
      "loss": 0.3439,
      "step": 2320
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00012157844744258722,
      "loss": 0.3439,
      "step": 2330
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00011974744994870517,
      "loss": 0.3406,
      "step": 2340
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00011792443486203788,
      "loss": 0.3424,
      "step": 2350
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011610958351076458,
      "loss": 0.34,
      "step": 2360
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011430307641104971,
      "loss": 0.3345,
      "step": 2370
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011250509324908767,
      "loss": 0.3411,
      "step": 2380
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011071581286323,
      "loss": 0.3393,
      "step": 2390
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00010893541322619732,
      "loss": 0.3337,
      "step": 2400
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00010716407142737659,
      "loss": 0.3428,
      "step": 2410
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00010540196365520754,
      "loss": 0.3406,
      "step": 2420
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00010364926517965692,
      "loss": 0.3411,
      "step": 2430
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001019061503347858,
      "loss": 0.3356,
      "step": 2440
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00010017279250140891,
      "loss": 0.3342,
      "step": 2450
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.844936408984924e-05,
      "loss": 0.3381,
      "step": 2460
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.673603652278904e-05,
      "loss": 0.3324,
      "step": 2470
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.503298021821905e-05,
      "loss": 0.3255,
      "step": 2480
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.334036457248774e-05,
      "loss": 0.3333,
      "step": 2490
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.165835794345205e-05,
      "loss": 0.33,
      "step": 2500
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.99871276337315e-05,
      "loss": 0.3396,
      "step": 2510
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.832683987406746e-05,
      "loss": 0.3321,
      "step": 2520
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.667765980678851e-05,
      "loss": 0.3279,
      "step": 2530
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.503975146938444e-05,
      "loss": 0.3337,
      "step": 2540
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.341327777819035e-05,
      "loss": 0.3369,
      "step": 2550
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.179840051218167e-05,
      "loss": 0.3326,
      "step": 2560
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.019528029688286e-05,
      "loss": 0.3292,
      "step": 2570
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.860407658839049e-05,
      "loss": 0.3377,
      "step": 2580
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.702494765751285e-05,
      "loss": 0.333,
      "step": 2590
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.545805057402733e-05,
      "loss": 0.3248,
      "step": 2600
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.390354119105722e-05,
      "loss": 0.3357,
      "step": 2610
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.236157412956994e-05,
      "loss": 0.3326,
      "step": 2620
    },
    {
      "epoch": 0.18,
      "learning_rate": 7.0832302762997e-05,
      "loss": 0.328,
      "step": 2630
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.93158792019789e-05,
      "loss": 0.3235,
      "step": 2640
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.781245427923522e-05,
      "loss": 0.3293,
      "step": 2650
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.632217753456174e-05,
      "loss": 0.3238,
      "step": 2660
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.484519719995647e-05,
      "loss": 0.3301,
      "step": 2670
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.338166018487555e-05,
      "loss": 0.3273,
      "step": 2680
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.193171206162065e-05,
      "loss": 0.3305,
      "step": 2690
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.0495497050859574e-05,
      "loss": 0.3335,
      "step": 2700
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.907315800728106e-05,
      "loss": 0.3396,
      "step": 2710
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.766483640538587e-05,
      "loss": 0.3281,
      "step": 2720
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.62706723254145e-05,
      "loss": 0.3255,
      "step": 2730
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.489080443941415e-05,
      "loss": 0.3336,
      "step": 2740
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.352536999744557e-05,
      "loss": 0.3233,
      "step": 2750
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.217450481393129e-05,
      "loss": 0.3273,
      "step": 2760
    },
    {
      "epoch": 0.19,
      "learning_rate": 5.083834325414667e-05,
      "loss": 0.3329,
      "step": 2770
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.951701822085515e-05,
      "loss": 0.3241,
      "step": 2780
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.821066114108892e-05,
      "loss": 0.3346,
      "step": 2790
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.69194019530764e-05,
      "loss": 0.3358,
      "step": 2800
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.564336909331768e-05,
      "loss": 0.3275,
      "step": 2810
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.438268948380972e-05,
      "loss": 0.3245,
      "step": 2820
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.3137488519421656e-05,
      "loss": 0.33,
      "step": 2830
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.1907890055422286e-05,
      "loss": 0.326,
      "step": 2840
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.069401639516075e-05,
      "loss": 0.3262,
      "step": 2850
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.949598827790155e-05,
      "loss": 0.3266,
      "step": 2860
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.831392486681495e-05,
      "loss": 0.3228,
      "step": 2870
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.714794373712431e-05,
      "loss": 0.3327,
      "step": 2880
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.5998160864411476e-05,
      "loss": 0.3171,
      "step": 2890
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.486469061308093e-05,
      "loss": 0.3197,
      "step": 2900
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.3747645724984544e-05,
      "loss": 0.3268,
      "step": 2910
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.264713730820768e-05,
      "loss": 0.3254,
      "step": 2920
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.156327482601742e-05,
      "loss": 0.3251,
      "step": 2930
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.0496166085974943e-05,
      "loss": 0.3302,
      "step": 2940
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.9445917229212193e-05,
      "loss": 0.3176,
      "step": 2950
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.8412632719874532e-05,
      "loss": 0.3203,
      "step": 2960
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.7396415334729964e-05,
      "loss": 0.3183,
      "step": 2970
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.6397366152946523e-05,
      "loss": 0.3285,
      "step": 2980
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.5415584546038096e-05,
      "loss": 0.3279,
      "step": 2990
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.4451168167980497e-05,
      "loss": 0.3186,
      "step": 3000
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.350421294549825e-05,
      "loss": 0.3222,
      "step": 3010
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.2574813068522894e-05,
      "loss": 0.3271,
      "step": 3020
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.166306098082451e-05,
      "loss": 0.3253,
      "step": 3030
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.0769047370816553e-05,
      "loss": 0.3234,
      "step": 3040
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.989286116253557e-05,
      "loss": 0.3256,
      "step": 3050
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.903458950679613e-05,
      "loss": 0.3234,
      "step": 3060
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8194317772522362e-05,
      "loss": 0.3228,
      "step": 3070
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.7372129538256667e-05,
      "loss": 0.32,
      "step": 3080
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.6568106583846378e-05,
      "loss": 0.3198,
      "step": 3090
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5782328882309484e-05,
      "loss": 0.331,
      "step": 3100
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5014874591880157e-05,
      "loss": 0.3264,
      "step": 3110
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.4265820048234447e-05,
      "loss": 0.3253,
      "step": 3120
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.3535239756897566e-05,
      "loss": 0.3187,
      "step": 3130
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.2823206385833187e-05,
      "loss": 0.3199,
      "step": 3140
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.212979075821532e-05,
      "loss": 0.3225,
      "step": 3150
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.1455061845383852e-05,
      "loss": 0.3195,
      "step": 3160
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.0799086759984333e-05,
      "loss": 0.3163,
      "step": 3170
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.016193074929237e-05,
      "loss": 0.3217,
      "step": 3180
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.543657188723876e-06,
      "loss": 0.3204,
      "step": 3190
    },
    {
      "epoch": 0.22,
      "learning_rate": 8.944327575531275e-06,
      "loss": 0.3196,
      "step": 3200
    },
    {
      "epoch": 0.23,
      "learning_rate": 8.364001522686726e-06,
      "loss": 0.3326,
      "step": 3210
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.802736752952533e-06,
      "loss": 0.3153,
      "step": 3220
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.260589093139736e-06,
      "loss": 0.323,
      "step": 3230
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.737612468555221e-06,
      "loss": 0.3192,
      "step": 3240
    },
    {
      "epoch": 0.23,
      "learning_rate": 6.2338588976380115e-06,
      "loss": 0.3166,
      "step": 3250
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.749378486785162e-06,
      "loss": 0.3134,
      "step": 3260
    },
    {
      "epoch": 0.23,
      "learning_rate": 5.2842194253679424e-06,
      "loss": 0.3225,
      "step": 3270
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8384279809385426e-06,
      "loss": 0.3251,
      "step": 3280
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.41204849462804e-06,
      "loss": 0.3228,
      "step": 3290
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.005123376735997e-06,
      "loss": 0.3181,
      "step": 3300
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.617693102512032e-06,
      "loss": 0.3135,
      "step": 3310
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.2497962081299514e-06,
      "loss": 0.3196,
      "step": 3320
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.9014692868546633e-06,
      "loss": 0.3143,
      "step": 3330
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.572746985402419e-06,
      "loss": 0.3124,
      "step": 3340
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.2636620004946154e-06,
      "loss": 0.3157,
      "step": 3350
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.974245075605574e-06,
      "loss": 0.3225,
      "step": 3360
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7045249979046995e-06,
      "loss": 0.3165,
      "step": 3370
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.4545285953929677e-06,
      "loss": 0.3178,
      "step": 3380
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.224280734234573e-06,
      "loss": 0.3129,
      "step": 3390
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.013804316283573e-06,
      "loss": 0.3187,
      "step": 3400
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.231202768059332e-07,
      "loss": 0.3178,
      "step": 3410
    },
    {
      "epoch": 0.24,
      "learning_rate": 6.522475823970808e-07,
      "loss": 0.3252,
      "step": 3420
    },
    {
      "epoch": 0.24,
      "learning_rate": 5.012032290955037e-07,
      "loss": 0.311,
      "step": 3430
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.7000224069216883e-07,
      "loss": 0.3253,
      "step": 3440
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.586576672361396e-07,
      "loss": 0.3205,
      "step": 3450
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.671805837365703e-07,
      "loss": 0.3196,
      "step": 3460
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.558008906112026e-08,
      "loss": 0.3166,
      "step": 3470
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.386330503090008e-08,
      "loss": 0.3255,
      "step": 3480
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.2035375712105001e-08,
      "loss": 0.3181,
      "step": 3490
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.946690433526584e-11,
      "loss": 0.3212,
      "step": 3500
    }
  ],
  "max_steps": 3500,
  "num_train_epochs": 1,
  "total_flos": 2.21324249137152e+18,
  "trial_name": null,
  "trial_params": null
}