Asimok's picture
Upload 36 files
89c2adb
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 522,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.11,
"global_step": 20,
"learning_rate": 3.7735849056603776e-05,
"loss": 0.3277,
"step": 20
},
{
"epoch": 0.23,
"global_step": 40,
"learning_rate": 7.358490566037736e-05,
"loss": 0.2854,
"step": 40
},
{
"epoch": 0.34,
"global_step": 60,
"learning_rate": 0.0001,
"loss": 0.2546,
"step": 60
},
{
"epoch": 0.46,
"global_step": 80,
"learning_rate": 0.0001,
"loss": 0.2746,
"step": 80
},
{
"epoch": 0.57,
"global_step": 100,
"learning_rate": 0.0001,
"loss": 0.2706,
"step": 100
},
{
"epoch": 0.69,
"global_step": 120,
"learning_rate": 0.0001,
"loss": 0.2405,
"step": 120
},
{
"epoch": 0.8,
"global_step": 140,
"learning_rate": 0.0001,
"loss": 0.31,
"step": 140
},
{
"epoch": 0.92,
"global_step": 160,
"learning_rate": 0.0001,
"loss": 0.3137,
"step": 160
},
{
"epoch": 1.03,
"global_step": 180,
"learning_rate": 0.0001,
"loss": 0.2409,
"step": 180
},
{
"epoch": 1.15,
"global_step": 200,
"learning_rate": 0.0001,
"loss": 0.1111,
"step": 200
},
{
"epoch": 1.26,
"global_step": 220,
"learning_rate": 0.0001,
"loss": 0.1792,
"step": 220
},
{
"epoch": 1.38,
"global_step": 240,
"learning_rate": 0.0001,
"loss": 0.1353,
"step": 240
},
{
"epoch": 1.49,
"global_step": 260,
"learning_rate": 0.0001,
"loss": 0.1418,
"step": 260
},
{
"epoch": 1.61,
"global_step": 280,
"learning_rate": 0.0001,
"loss": 0.1249,
"step": 280
},
{
"epoch": 1.72,
"global_step": 300,
"learning_rate": 0.0001,
"loss": 0.1365,
"step": 300
},
{
"epoch": 1.84,
"global_step": 320,
"learning_rate": 0.0001,
"loss": 0.1525,
"step": 320
},
{
"epoch": 1.95,
"global_step": 340,
"learning_rate": 0.0001,
"loss": 0.1233,
"step": 340
},
{
"epoch": 2.07,
"global_step": 360,
"learning_rate": 0.0001,
"loss": 0.0656,
"step": 360
},
{
"epoch": 2.18,
"global_step": 380,
"learning_rate": 0.0001,
"loss": 0.0608,
"step": 380
},
{
"epoch": 2.3,
"global_step": 400,
"learning_rate": 0.0001,
"loss": 0.0805,
"step": 400
},
{
"epoch": 2.41,
"global_step": 420,
"learning_rate": 0.0001,
"loss": 0.0562,
"step": 420
},
{
"epoch": 2.53,
"global_step": 440,
"learning_rate": 0.0001,
"loss": 0.0932,
"step": 440
},
{
"epoch": 2.64,
"global_step": 460,
"learning_rate": 0.0001,
"loss": 0.0551,
"step": 460
},
{
"epoch": 2.76,
"global_step": 480,
"learning_rate": 0.0001,
"loss": 0.0659,
"step": 480
},
{
"epoch": 2.87,
"global_step": 500,
"learning_rate": 0.0001,
"loss": 0.0959,
"step": 500
},
{
"epoch": 2.99,
"global_step": 520,
"learning_rate": 0.0001,
"loss": 0.0682,
"step": 520
},
{
"epoch": 3.0,
"step": 522,
"total_flos": 2.6798272947001754e+17,
"train_loss": 0.16342519847783207,
"train_runtime": 12079.5406,
"train_samples_per_second": 0.863,
"train_steps_per_second": 0.043
}
],
"logging_steps": 20,
"max_steps": 522,
"num_train_epochs": 3,
"save_steps": 300,
"total_flos": 2.6798272947001754e+17,
"trial_name": null,
"trial_params": null
}