Asimok's picture
Upload 36 files
89c2adb
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 1149,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.05,
"global_step": 20,
"learning_rate": 1.652173913043478e-05,
"loss": 5.7171,
"step": 20
},
{
"epoch": 0.1,
"global_step": 40,
"learning_rate": 3.304347826086956e-05,
"loss": 3.3801,
"step": 40
},
{
"epoch": 0.16,
"global_step": 60,
"learning_rate": 4.956521739130435e-05,
"loss": 0.7521,
"step": 60
},
{
"epoch": 0.21,
"global_step": 80,
"learning_rate": 6.695652173913044e-05,
"loss": 0.7016,
"step": 80
},
{
"epoch": 0.26,
"global_step": 100,
"learning_rate": 8.434782608695653e-05,
"loss": 0.7053,
"step": 100
},
{
"epoch": 0.31,
"global_step": 120,
"learning_rate": 0.0001,
"loss": 0.7202,
"step": 120
},
{
"epoch": 0.37,
"global_step": 140,
"learning_rate": 0.0001,
"loss": 0.7062,
"step": 140
},
{
"epoch": 0.42,
"global_step": 160,
"learning_rate": 0.0001,
"loss": 0.709,
"step": 160
},
{
"epoch": 0.47,
"global_step": 180,
"learning_rate": 0.0001,
"loss": 0.6773,
"step": 180
},
{
"epoch": 0.52,
"global_step": 200,
"learning_rate": 0.0001,
"loss": 0.6675,
"step": 200
},
{
"epoch": 0.57,
"global_step": 220,
"learning_rate": 0.0001,
"loss": 0.6582,
"step": 220
},
{
"epoch": 0.63,
"global_step": 240,
"learning_rate": 0.0001,
"loss": 0.7473,
"step": 240
},
{
"epoch": 0.68,
"global_step": 260,
"learning_rate": 0.0001,
"loss": 0.6401,
"step": 260
},
{
"epoch": 0.73,
"global_step": 280,
"learning_rate": 0.0001,
"loss": 0.6312,
"step": 280
},
{
"epoch": 0.78,
"global_step": 300,
"learning_rate": 0.0001,
"loss": 0.6399,
"step": 300
},
{
"epoch": 0.84,
"global_step": 320,
"learning_rate": 0.0001,
"loss": 0.6484,
"step": 320
},
{
"epoch": 0.89,
"global_step": 340,
"learning_rate": 0.0001,
"loss": 0.6193,
"step": 340
},
{
"epoch": 0.94,
"global_step": 360,
"learning_rate": 0.0001,
"loss": 0.5908,
"step": 360
},
{
"epoch": 0.99,
"global_step": 380,
"learning_rate": 0.0001,
"loss": 0.6047,
"step": 380
},
{
"epoch": 1.04,
"global_step": 400,
"learning_rate": 0.0001,
"loss": 0.5931,
"step": 400
},
{
"epoch": 1.1,
"global_step": 420,
"learning_rate": 0.0001,
"loss": 0.6038,
"step": 420
},
{
"epoch": 1.15,
"global_step": 440,
"learning_rate": 0.0001,
"loss": 0.5726,
"step": 440
},
{
"epoch": 1.2,
"global_step": 460,
"learning_rate": 0.0001,
"loss": 0.5738,
"step": 460
},
{
"epoch": 1.25,
"global_step": 480,
"learning_rate": 0.0001,
"loss": 0.5536,
"step": 480
},
{
"epoch": 1.31,
"global_step": 500,
"learning_rate": 0.0001,
"loss": 0.576,
"step": 500
},
{
"epoch": 1.36,
"global_step": 520,
"learning_rate": 0.0001,
"loss": 0.5598,
"step": 520
},
{
"epoch": 1.41,
"global_step": 540,
"learning_rate": 0.0001,
"loss": 0.547,
"step": 540
},
{
"epoch": 1.46,
"global_step": 560,
"learning_rate": 0.0001,
"loss": 0.5269,
"step": 560
},
{
"epoch": 1.51,
"global_step": 580,
"learning_rate": 0.0001,
"loss": 0.5244,
"step": 580
},
{
"epoch": 1.57,
"global_step": 600,
"learning_rate": 0.0001,
"loss": 0.5017,
"step": 600
},
{
"epoch": 1.62,
"global_step": 620,
"learning_rate": 0.0001,
"loss": 0.557,
"step": 620
},
{
"epoch": 1.67,
"global_step": 640,
"learning_rate": 0.0001,
"loss": 0.4926,
"step": 640
},
{
"epoch": 1.72,
"global_step": 660,
"learning_rate": 0.0001,
"loss": 0.4458,
"step": 660
},
{
"epoch": 1.78,
"global_step": 680,
"learning_rate": 0.0001,
"loss": 0.4503,
"step": 680
},
{
"epoch": 1.83,
"global_step": 700,
"learning_rate": 0.0001,
"loss": 0.4413,
"step": 700
},
{
"epoch": 1.88,
"global_step": 720,
"learning_rate": 0.0001,
"loss": 0.4396,
"step": 720
},
{
"epoch": 1.93,
"global_step": 740,
"learning_rate": 0.0001,
"loss": 0.3788,
"step": 740
},
{
"epoch": 1.98,
"global_step": 760,
"learning_rate": 0.0001,
"loss": 0.3731,
"step": 760
},
{
"epoch": 2.04,
"global_step": 780,
"learning_rate": 0.0001,
"loss": 0.3797,
"step": 780
},
{
"epoch": 2.09,
"global_step": 800,
"learning_rate": 0.0001,
"loss": 0.4084,
"step": 800
},
{
"epoch": 2.14,
"global_step": 820,
"learning_rate": 0.0001,
"loss": 0.3583,
"step": 820
},
{
"epoch": 2.19,
"global_step": 840,
"learning_rate": 0.0001,
"loss": 0.3262,
"step": 840
},
{
"epoch": 2.25,
"global_step": 860,
"learning_rate": 0.0001,
"loss": 0.3407,
"step": 860
},
{
"epoch": 2.3,
"global_step": 880,
"learning_rate": 0.0001,
"loss": 0.3476,
"step": 880
},
{
"epoch": 2.35,
"global_step": 900,
"learning_rate": 0.0001,
"loss": 0.3041,
"step": 900
},
{
"epoch": 2.4,
"global_step": 920,
"learning_rate": 0.0001,
"loss": 0.3037,
"step": 920
},
{
"epoch": 2.45,
"global_step": 940,
"learning_rate": 0.0001,
"loss": 0.2546,
"step": 940
},
{
"epoch": 2.51,
"global_step": 960,
"learning_rate": 0.0001,
"loss": 0.2807,
"step": 960
},
{
"epoch": 2.56,
"global_step": 980,
"learning_rate": 0.0001,
"loss": 0.2338,
"step": 980
},
{
"epoch": 2.61,
"global_step": 1000,
"learning_rate": 0.0001,
"loss": 0.3036,
"step": 1000
},
{
"epoch": 2.66,
"global_step": 1020,
"learning_rate": 0.0001,
"loss": 0.2492,
"step": 1020
},
{
"epoch": 2.72,
"global_step": 1040,
"learning_rate": 0.0001,
"loss": 0.2118,
"step": 1040
},
{
"epoch": 2.77,
"global_step": 1060,
"learning_rate": 0.0001,
"loss": 0.2155,
"step": 1060
},
{
"epoch": 2.82,
"global_step": 1080,
"learning_rate": 0.0001,
"loss": 0.2032,
"step": 1080
},
{
"epoch": 2.87,
"global_step": 1100,
"learning_rate": 0.0001,
"loss": 0.1983,
"step": 1100
},
{
"epoch": 2.92,
"global_step": 1120,
"learning_rate": 0.0001,
"loss": 0.1879,
"step": 1120
},
{
"epoch": 2.98,
"global_step": 1140,
"learning_rate": 0.0001,
"loss": 0.1938,
"step": 1140
},
{
"epoch": 3.0,
"step": 1149,
"total_flos": 1.3282309617703977e+18,
"train_loss": 0.6198576812229538,
"train_runtime": 37640.6404,
"train_samples_per_second": 1.221,
"train_steps_per_second": 0.031
}
],
"logging_steps": 20,
"max_steps": 1149,
"num_train_epochs": 3,
"save_steps": 300,
"total_flos": 1.3282309617703977e+18,
"trial_name": null,
"trial_params": null
}