Upload trainer_state.json
Browse files- trainer_state.json +73 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 4481,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.11,
|
| 12 |
+
"learning_rate": 2.665253291675965e-05,
|
| 13 |
+
"loss": 2.8266,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.22,
|
| 18 |
+
"learning_rate": 2.3305065833519305e-05,
|
| 19 |
+
"loss": 1.8389,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.33,
|
| 24 |
+
"learning_rate": 1.9957598750278953e-05,
|
| 25 |
+
"loss": 1.6042,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.45,
|
| 30 |
+
"learning_rate": 1.661013166703861e-05,
|
| 31 |
+
"loss": 1.5406,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.56,
|
| 36 |
+
"learning_rate": 1.326266458379826e-05,
|
| 37 |
+
"loss": 1.4462,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.67,
|
| 42 |
+
"learning_rate": 9.915197500557913e-06,
|
| 43 |
+
"loss": 1.4261,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.78,
|
| 48 |
+
"learning_rate": 6.567730417317563e-06,
|
| 49 |
+
"loss": 1.3561,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.89,
|
| 54 |
+
"learning_rate": 3.220263334077215e-06,
|
| 55 |
+
"loss": 1.3676,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.0,
|
| 60 |
+
"step": 4481,
|
| 61 |
+
"total_flos": 1.0536269121865728e+16,
|
| 62 |
+
"train_loss": 1.6402719195893076,
|
| 63 |
+
"train_runtime": 2258.7241,
|
| 64 |
+
"train_samples_per_second": 23.803,
|
| 65 |
+
"train_steps_per_second": 1.984
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
"max_steps": 4481,
|
| 69 |
+
"num_train_epochs": 1,
|
| 70 |
+
"total_flos": 1.0536269121865728e+16,
|
| 71 |
+
"trial_name": null,
|
| 72 |
+
"trial_params": null
|
| 73 |
+
}
|