M2M1.2b_NL_subtitles / trainer_state.json
wangpichao's picture
Upload 13 files
6a84e87 verified
raw
history blame contribute delete
No virus
82.1 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 383970,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 0.6562,
"step": 500
},
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 0.5864,
"step": 1000
},
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 0.5613,
"step": 1500
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.5443,
"step": 2000
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.5272,
"step": 2500
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.5199,
"step": 3000
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.5136,
"step": 3500
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.502,
"step": 4000
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.4965,
"step": 4500
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.4904,
"step": 5000
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.4871,
"step": 5500
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.481,
"step": 6000
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.4765,
"step": 6500
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.473,
"step": 7000
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.4725,
"step": 7500
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.4677,
"step": 8000
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.463,
"step": 8500
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.4633,
"step": 9000
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.4615,
"step": 9500
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4586,
"step": 10000
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4561,
"step": 10500
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.453,
"step": 11000
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4528,
"step": 11500
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4497,
"step": 12000
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4478,
"step": 12500
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.4475,
"step": 13000
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4457,
"step": 13500
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4432,
"step": 14000
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4412,
"step": 14500
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4383,
"step": 15000
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.439,
"step": 15500
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4391,
"step": 16000
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4359,
"step": 16500
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.4354,
"step": 17000
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4328,
"step": 17500
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4316,
"step": 18000
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4311,
"step": 18500
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4288,
"step": 19000
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4259,
"step": 19500
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4266,
"step": 20000
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.426,
"step": 20500
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.4233,
"step": 21000
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4231,
"step": 21500
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4206,
"step": 22000
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4219,
"step": 22500
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4199,
"step": 23000
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4191,
"step": 23500
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4181,
"step": 24000
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.4172,
"step": 24500
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4166,
"step": 25000
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4142,
"step": 25500
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4145,
"step": 26000
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4148,
"step": 26500
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4133,
"step": 27000
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4128,
"step": 27500
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.412,
"step": 28000
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.4108,
"step": 28500
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4101,
"step": 29000
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4091,
"step": 29500
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4107,
"step": 30000
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4097,
"step": 30500
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4056,
"step": 31000
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4075,
"step": 31500
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4075,
"step": 32000
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.4048,
"step": 32500
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4042,
"step": 33000
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4041,
"step": 33500
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4051,
"step": 34000
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4038,
"step": 34500
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4044,
"step": 35000
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4048,
"step": 35500
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.4026,
"step": 36000
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.402,
"step": 36500
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.4018,
"step": 37000
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.4018,
"step": 37500
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.3993,
"step": 38000
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.4021,
"step": 38500
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.4001,
"step": 39000
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.3987,
"step": 39500
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.3994,
"step": 40000
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.4001,
"step": 40500
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3975,
"step": 41000
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3954,
"step": 41500
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3958,
"step": 42000
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3963,
"step": 42500
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3949,
"step": 43000
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3946,
"step": 43500
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.3957,
"step": 44000
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3932,
"step": 44500
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3942,
"step": 45000
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3919,
"step": 45500
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3935,
"step": 46000
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3919,
"step": 46500
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.3924,
"step": 47000
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.39,
"step": 47500
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3896,
"step": 48000
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3901,
"step": 48500
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3916,
"step": 49000
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3903,
"step": 49500
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3885,
"step": 50000
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3897,
"step": 50500
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3883,
"step": 51000
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.3874,
"step": 51500
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3892,
"step": 52000
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.388,
"step": 52500
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3873,
"step": 53000
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3879,
"step": 53500
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3867,
"step": 54000
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3873,
"step": 54500
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3869,
"step": 55000
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.3872,
"step": 55500
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3864,
"step": 56000
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3869,
"step": 56500
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3838,
"step": 57000
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3856,
"step": 57500
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3848,
"step": 58000
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3861,
"step": 58500
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3854,
"step": 59000
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.3843,
"step": 59500
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.386,
"step": 60000
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3816,
"step": 60500
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3814,
"step": 61000
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3835,
"step": 61500
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3818,
"step": 62000
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3823,
"step": 62500
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.3825,
"step": 63000
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3823,
"step": 63500
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3806,
"step": 64000
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3803,
"step": 64500
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3797,
"step": 65000
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3797,
"step": 65500
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3793,
"step": 66000
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.3794,
"step": 66500
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.379,
"step": 67000
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3783,
"step": 67500
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3784,
"step": 68000
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3778,
"step": 68500
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3773,
"step": 69000
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.378,
"step": 69500
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3759,
"step": 70000
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.377,
"step": 70500
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.3763,
"step": 71000
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.376,
"step": 71500
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3767,
"step": 72000
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3775,
"step": 72500
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3794,
"step": 73000
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3774,
"step": 73500
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3762,
"step": 74000
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.3765,
"step": 74500
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3776,
"step": 75000
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3762,
"step": 75500
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3761,
"step": 76000
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3767,
"step": 76500
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3756,
"step": 77000
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3744,
"step": 77500
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3739,
"step": 78000
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.3746,
"step": 78500
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3736,
"step": 79000
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3723,
"step": 79500
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3739,
"step": 80000
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3734,
"step": 80500
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3726,
"step": 81000
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3737,
"step": 81500
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3726,
"step": 82000
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.3712,
"step": 82500
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.3719,
"step": 83000
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.3714,
"step": 83500
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.3711,
"step": 84000
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.3707,
"step": 84500
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.369,
"step": 85000
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.3706,
"step": 85500
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.371,
"step": 86000
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3706,
"step": 86500
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3708,
"step": 87000
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3687,
"step": 87500
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3705,
"step": 88000
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3699,
"step": 88500
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3695,
"step": 89000
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.37,
"step": 89500
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.3707,
"step": 90000
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3698,
"step": 90500
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3692,
"step": 91000
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.37,
"step": 91500
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3674,
"step": 92000
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3694,
"step": 92500
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3696,
"step": 93000
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3677,
"step": 93500
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.3685,
"step": 94000
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3671,
"step": 94500
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3684,
"step": 95000
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3664,
"step": 95500
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3678,
"step": 96000
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3677,
"step": 96500
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3667,
"step": 97000
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.3676,
"step": 97500
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3677,
"step": 98000
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3651,
"step": 98500
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3658,
"step": 99000
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3651,
"step": 99500
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3663,
"step": 100000
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3657,
"step": 100500
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3655,
"step": 101000
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.3637,
"step": 101500
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3662,
"step": 102000
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3638,
"step": 102500
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3638,
"step": 103000
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3663,
"step": 103500
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3645,
"step": 104000
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3661,
"step": 104500
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3659,
"step": 105000
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.3651,
"step": 105500
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3646,
"step": 106000
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3645,
"step": 106500
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3637,
"step": 107000
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3634,
"step": 107500
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3626,
"step": 108000
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3636,
"step": 108500
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.3629,
"step": 109000
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3625,
"step": 109500
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3627,
"step": 110000
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3628,
"step": 110500
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3619,
"step": 111000
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3627,
"step": 111500
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.3612,
"step": 112000
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.362,
"step": 112500
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.363,
"step": 113000
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3607,
"step": 113500
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3617,
"step": 114000
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3605,
"step": 114500
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3626,
"step": 115000
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3619,
"step": 115500
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3616,
"step": 116000
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.361,
"step": 116500
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.3609,
"step": 117000
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3625,
"step": 117500
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3619,
"step": 118000
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3611,
"step": 118500
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3603,
"step": 119000
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3609,
"step": 119500
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.36,
"step": 120000
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.3612,
"step": 120500
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3595,
"step": 121000
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3602,
"step": 121500
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3601,
"step": 122000
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3593,
"step": 122500
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3595,
"step": 123000
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3586,
"step": 123500
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3591,
"step": 124000
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.3582,
"step": 124500
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3585,
"step": 125000
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3589,
"step": 125500
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3575,
"step": 126000
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3599,
"step": 126500
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3582,
"step": 127000
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3566,
"step": 127500
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3559,
"step": 128000
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.3598,
"step": 128500
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3579,
"step": 129000
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3581,
"step": 129500
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3592,
"step": 130000
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3585,
"step": 130500
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3577,
"step": 131000
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.357,
"step": 131500
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.3588,
"step": 132000
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.358,
"step": 132500
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.3566,
"step": 133000
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.3574,
"step": 133500
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.357,
"step": 134000
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.3553,
"step": 134500
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.357,
"step": 135000
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.3578,
"step": 135500
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.3559,
"step": 136000
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3559,
"step": 136500
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.356,
"step": 137000
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3553,
"step": 137500
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3565,
"step": 138000
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3557,
"step": 138500
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3541,
"step": 139000
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3552,
"step": 139500
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.3571,
"step": 140000
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3565,
"step": 140500
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3573,
"step": 141000
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3561,
"step": 141500
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3569,
"step": 142000
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3558,
"step": 142500
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3544,
"step": 143000
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.3559,
"step": 143500
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3558,
"step": 144000
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3545,
"step": 144500
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3542,
"step": 145000
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3543,
"step": 145500
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3541,
"step": 146000
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3545,
"step": 146500
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3542,
"step": 147000
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.3538,
"step": 147500
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3544,
"step": 148000
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3538,
"step": 148500
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3525,
"step": 149000
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3535,
"step": 149500
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3527,
"step": 150000
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.353,
"step": 150500
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.356,
"step": 151000
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.3541,
"step": 151500
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3542,
"step": 152000
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3541,
"step": 152500
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3541,
"step": 153000
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3537,
"step": 153500
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3543,
"step": 154000
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3547,
"step": 154500
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.352,
"step": 155000
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.3523,
"step": 155500
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3515,
"step": 156000
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3508,
"step": 156500
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3524,
"step": 157000
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3519,
"step": 157500
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3525,
"step": 158000
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3524,
"step": 158500
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.3508,
"step": 159000
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3529,
"step": 159500
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3529,
"step": 160000
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3525,
"step": 160500
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3535,
"step": 161000
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3515,
"step": 161500
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3526,
"step": 162000
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3507,
"step": 162500
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.3521,
"step": 163000
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3512,
"step": 163500
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.351,
"step": 164000
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3507,
"step": 164500
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3514,
"step": 165000
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3505,
"step": 165500
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3492,
"step": 166000
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3507,
"step": 166500
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.3511,
"step": 167000
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3497,
"step": 167500
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3506,
"step": 168000
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3507,
"step": 168500
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3501,
"step": 169000
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3491,
"step": 169500
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.3509,
"step": 170000
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.35,
"step": 170500
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3505,
"step": 171000
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3507,
"step": 171500
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3496,
"step": 172000
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3503,
"step": 172500
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3502,
"step": 173000
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3498,
"step": 173500
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3496,
"step": 174000
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.3498,
"step": 174500
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3498,
"step": 175000
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3473,
"step": 175500
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3482,
"step": 176000
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3482,
"step": 176500
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3486,
"step": 177000
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3504,
"step": 177500
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3492,
"step": 178000
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.3502,
"step": 178500
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3475,
"step": 179000
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3487,
"step": 179500
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3501,
"step": 180000
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3478,
"step": 180500
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.348,
"step": 181000
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3479,
"step": 181500
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.3467,
"step": 182000
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3474,
"step": 182500
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3454,
"step": 183000
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.347,
"step": 183500
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3491,
"step": 184000
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3487,
"step": 184500
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3489,
"step": 185000
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3481,
"step": 185500
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.3462,
"step": 186000
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.346,
"step": 186500
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3458,
"step": 187000
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3467,
"step": 187500
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3465,
"step": 188000
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3462,
"step": 188500
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3474,
"step": 189000
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.3471,
"step": 189500
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.348,
"step": 190000
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3466,
"step": 190500
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3465,
"step": 191000
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3472,
"step": 191500
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3463,
"step": 192000
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3478,
"step": 192500
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3479,
"step": 193000
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.3468,
"step": 193500
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3463,
"step": 194000
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3449,
"step": 194500
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3465,
"step": 195000
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3452,
"step": 195500
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3451,
"step": 196000
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3463,
"step": 196500
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.345,
"step": 197000
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.3458,
"step": 197500
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3465,
"step": 198000
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3445,
"step": 198500
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3443,
"step": 199000
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3457,
"step": 199500
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3459,
"step": 200000
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3454,
"step": 200500
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3439,
"step": 201000
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.3456,
"step": 201500
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3453,
"step": 202000
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3459,
"step": 202500
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3446,
"step": 203000
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3441,
"step": 203500
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3469,
"step": 204000
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3457,
"step": 204500
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.3446,
"step": 205000
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3433,
"step": 205500
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3436,
"step": 206000
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.345,
"step": 206500
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3436,
"step": 207000
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3442,
"step": 207500
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3444,
"step": 208000
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3436,
"step": 208500
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.3439,
"step": 209000
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3452,
"step": 209500
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3448,
"step": 210000
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3447,
"step": 210500
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3447,
"step": 211000
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3433,
"step": 211500
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3434,
"step": 212000
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3444,
"step": 212500
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.3443,
"step": 213000
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3441,
"step": 213500
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3434,
"step": 214000
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3443,
"step": 214500
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3425,
"step": 215000
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3437,
"step": 215500
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3432,
"step": 216000
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.3436,
"step": 216500
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3426,
"step": 217000
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3419,
"step": 217500
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3435,
"step": 218000
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3422,
"step": 218500
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3439,
"step": 219000
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3431,
"step": 219500
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3431,
"step": 220000
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.3428,
"step": 220500
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3426,
"step": 221000
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3416,
"step": 221500
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3422,
"step": 222000
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3413,
"step": 222500
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3406,
"step": 223000
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3418,
"step": 223500
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3415,
"step": 224000
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.3418,
"step": 224500
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3415,
"step": 225000
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3422,
"step": 225500
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3414,
"step": 226000
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3415,
"step": 226500
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3425,
"step": 227000
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3426,
"step": 227500
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.3431,
"step": 228000
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3411,
"step": 228500
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3416,
"step": 229000
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3422,
"step": 229500
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3414,
"step": 230000
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3426,
"step": 230500
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3412,
"step": 231000
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3414,
"step": 231500
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.3402,
"step": 232000
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3412,
"step": 232500
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3405,
"step": 233000
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3409,
"step": 233500
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3409,
"step": 234000
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3421,
"step": 234500
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3409,
"step": 235000
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3413,
"step": 235500
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.3397,
"step": 236000
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3401,
"step": 236500
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.342,
"step": 237000
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3402,
"step": 237500
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3399,
"step": 238000
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3413,
"step": 238500
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3397,
"step": 239000
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.3395,
"step": 239500
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3411,
"step": 240000
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3409,
"step": 240500
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3411,
"step": 241000
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3395,
"step": 241500
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3418,
"step": 242000
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3383,
"step": 242500
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3402,
"step": 243000
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.3393,
"step": 243500
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3407,
"step": 244000
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3413,
"step": 244500
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.34,
"step": 245000
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3401,
"step": 245500
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3391,
"step": 246000
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3395,
"step": 246500
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3394,
"step": 247000
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.3403,
"step": 247500
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3395,
"step": 248000
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3403,
"step": 248500
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3394,
"step": 249000
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3394,
"step": 249500
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3394,
"step": 250000
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3388,
"step": 250500
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3395,
"step": 251000
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.3396,
"step": 251500
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3401,
"step": 252000
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3374,
"step": 252500
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3398,
"step": 253000
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3376,
"step": 253500
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3394,
"step": 254000
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3393,
"step": 254500
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.3385,
"step": 255000
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3397,
"step": 255500
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3374,
"step": 256000
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3374,
"step": 256500
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3377,
"step": 257000
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3388,
"step": 257500
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3381,
"step": 258000
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3386,
"step": 258500
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.3379,
"step": 259000
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3387,
"step": 259500
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3369,
"step": 260000
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3388,
"step": 260500
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3373,
"step": 261000
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3399,
"step": 261500
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3366,
"step": 262000
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3377,
"step": 262500
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.3385,
"step": 263000
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3377,
"step": 263500
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3382,
"step": 264000
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3369,
"step": 264500
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3374,
"step": 265000
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3373,
"step": 265500
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3361,
"step": 266000
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.3362,
"step": 266500
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3376,
"step": 267000
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3372,
"step": 267500
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3371,
"step": 268000
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3359,
"step": 268500
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3354,
"step": 269000
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3367,
"step": 269500
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.337,
"step": 270000
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.3365,
"step": 270500
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3374,
"step": 271000
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3365,
"step": 271500
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3359,
"step": 272000
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3363,
"step": 272500
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3352,
"step": 273000
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3369,
"step": 273500
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3354,
"step": 274000
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.3373,
"step": 274500
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3366,
"step": 275000
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3365,
"step": 275500
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3373,
"step": 276000
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3368,
"step": 276500
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3375,
"step": 277000
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3356,
"step": 277500
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.3356,
"step": 278000
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3373,
"step": 278500
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3362,
"step": 279000
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3361,
"step": 279500
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3357,
"step": 280000
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3367,
"step": 280500
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3371,
"step": 281000
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3358,
"step": 281500
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.3354,
"step": 282000
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3363,
"step": 282500
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3366,
"step": 283000
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3353,
"step": 283500
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3367,
"step": 284000
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3357,
"step": 284500
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3364,
"step": 285000
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3338,
"step": 285500
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.3346,
"step": 286000
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3341,
"step": 286500
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3368,
"step": 287000
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3357,
"step": 287500
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3345,
"step": 288000
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.335,
"step": 288500
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3352,
"step": 289000
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.3338,
"step": 289500
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3356,
"step": 290000
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3355,
"step": 290500
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3354,
"step": 291000
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3358,
"step": 291500
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3343,
"step": 292000
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.335,
"step": 292500
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.335,
"step": 293000
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.3365,
"step": 293500
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3355,
"step": 294000
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3347,
"step": 294500
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3349,
"step": 295000
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3344,
"step": 295500
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3345,
"step": 296000
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3341,
"step": 296500
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3353,
"step": 297000
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.3358,
"step": 297500
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3353,
"step": 298000
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3337,
"step": 298500
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3341,
"step": 299000
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3351,
"step": 299500
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3347,
"step": 300000
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3334,
"step": 300500
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.3348,
"step": 301000
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3352,
"step": 301500
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3343,
"step": 302000
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.334,
"step": 302500
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.335,
"step": 303000
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3337,
"step": 303500
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3338,
"step": 304000
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3354,
"step": 304500
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.3339,
"step": 305000
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3332,
"step": 305500
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3336,
"step": 306000
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3344,
"step": 306500
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3333,
"step": 307000
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3347,
"step": 307500
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3337,
"step": 308000
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3327,
"step": 308500
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.3335,
"step": 309000
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.333,
"step": 309500
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.3341,
"step": 310000
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.334,
"step": 310500
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.3336,
"step": 311000
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.3328,
"step": 311500
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.3341,
"step": 312000
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.3333,
"step": 312500
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3335,
"step": 313000
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3312,
"step": 313500
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3332,
"step": 314000
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3334,
"step": 314500
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3322,
"step": 315000
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3324,
"step": 315500
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3327,
"step": 316000
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.3331,
"step": 316500
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3331,
"step": 317000
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3319,
"step": 317500
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3322,
"step": 318000
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3329,
"step": 318500
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3335,
"step": 319000
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.332,
"step": 319500
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3322,
"step": 320000
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.3327,
"step": 320500
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.331,
"step": 321000
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.333,
"step": 321500
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.3342,
"step": 322000
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.3321,
"step": 322500
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.3324,
"step": 323000
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.3318,
"step": 323500
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.3324,
"step": 324000
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.3324,
"step": 324500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.3318,
"step": 325000
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.332,
"step": 325500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.3319,
"step": 326000
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.332,
"step": 326500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.3325,
"step": 327000
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.331,
"step": 327500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.3314,
"step": 328000
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3323,
"step": 328500
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3316,
"step": 329000
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3331,
"step": 329500
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3308,
"step": 330000
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3309,
"step": 330500
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3312,
"step": 331000
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3321,
"step": 331500
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.3318,
"step": 332000
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.331,
"step": 332500
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3301,
"step": 333000
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3315,
"step": 333500
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3314,
"step": 334000
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3319,
"step": 334500
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3315,
"step": 335000
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.3316,
"step": 335500
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3311,
"step": 336000
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3302,
"step": 336500
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3306,
"step": 337000
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3297,
"step": 337500
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3319,
"step": 338000
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3297,
"step": 338500
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 339000
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.3317,
"step": 339500
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 340000
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.3309,
"step": 340500
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.332,
"step": 341000
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.331,
"step": 341500
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.3313,
"step": 342000
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.33,
"step": 342500
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.3313,
"step": 343000
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 343500
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3322,
"step": 344000
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.331,
"step": 344500
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3304,
"step": 345000
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3312,
"step": 345500
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3317,
"step": 346000
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3306,
"step": 346500
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.3288,
"step": 347000
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3304,
"step": 347500
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3307,
"step": 348000
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3305,
"step": 348500
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3294,
"step": 349000
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3309,
"step": 349500
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3293,
"step": 350000
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 350500
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.3303,
"step": 351000
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3312,
"step": 351500
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3308,
"step": 352000
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3291,
"step": 352500
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3289,
"step": 353000
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3302,
"step": 353500
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 354000
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3302,
"step": 354500
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.3309,
"step": 355000
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3287,
"step": 355500
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3297,
"step": 356000
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3304,
"step": 356500
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3295,
"step": 357000
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 357500
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3305,
"step": 358000
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3297,
"step": 358500
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.3298,
"step": 359000
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3282,
"step": 359500
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3301,
"step": 360000
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3304,
"step": 360500
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3281,
"step": 361000
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.329,
"step": 361500
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3293,
"step": 362000
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.3293,
"step": 362500
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.329,
"step": 363000
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3303,
"step": 363500
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 364000
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.329,
"step": 364500
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3298,
"step": 365000
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3279,
"step": 365500
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3298,
"step": 366000
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.3291,
"step": 366500
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3298,
"step": 367000
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3283,
"step": 367500
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3274,
"step": 368000
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3287,
"step": 368500
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3299,
"step": 369000
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3289,
"step": 369500
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3282,
"step": 370000
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.3293,
"step": 370500
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3291,
"step": 371000
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3279,
"step": 371500
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3283,
"step": 372000
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3304,
"step": 372500
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3291,
"step": 373000
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.3284,
"step": 373500
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.328,
"step": 374000
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3289,
"step": 374500
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3281,
"step": 375000
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3295,
"step": 375500
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3287,
"step": 376000
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3281,
"step": 376500
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3265,
"step": 377000
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3283,
"step": 377500
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.3283,
"step": 378000
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.328,
"step": 378500
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.3268,
"step": 379000
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.3273,
"step": 379500
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.3274,
"step": 380000
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.326,
"step": 380500
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.3273,
"step": 381000
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.328,
"step": 381500
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.3273,
"step": 382000
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.3282,
"step": 382500
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.3281,
"step": 383000
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.3272,
"step": 383500
}
],
"logging_steps": 500,
"max_steps": 383970,
"num_train_epochs": 1,
"save_steps": 383970,
"total_flos": 1.3356439865463197e+20,
"trial_name": null,
"trial_params": null
}