Commit
•
4029416
1
Parent(s):
23df469
End of training
Browse files- all_results.json +13 -0
- test_results.json +13 -0
- trainer_state.json +1092 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.092857142857143,
|
| 3 |
+
"eval_accuracy": 0.946058091286307,
|
| 4 |
+
"eval_accuracy_hold": 1.0,
|
| 5 |
+
"eval_accuracy_preparation": 0.9685534591194969,
|
| 6 |
+
"eval_accuracy_recovery": 0.8947368421052632,
|
| 7 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 8 |
+
"eval_accuracy_unknown": 0.9285714285714286,
|
| 9 |
+
"eval_loss": 0.19672317802906036,
|
| 10 |
+
"eval_runtime": 30.5504,
|
| 11 |
+
"eval_samples_per_second": 7.889,
|
| 12 |
+
"eval_steps_per_second": 1.015
|
| 13 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.092857142857143,
|
| 3 |
+
"eval_accuracy": 0.946058091286307,
|
| 4 |
+
"eval_accuracy_hold": 1.0,
|
| 5 |
+
"eval_accuracy_preparation": 0.9685534591194969,
|
| 6 |
+
"eval_accuracy_recovery": 0.8947368421052632,
|
| 7 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 8 |
+
"eval_accuracy_unknown": 0.9285714285714286,
|
| 9 |
+
"eval_loss": 0.19672317802906036,
|
| 10 |
+
"eval_runtime": 30.5504,
|
| 11 |
+
"eval_samples_per_second": 7.889,
|
| 12 |
+
"eval_steps_per_second": 1.015
|
| 13 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1092 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9737991266375546,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-good-gesturePhaseV11/checkpoint-1260",
|
| 4 |
+
"epoch": 9.092857142857143,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1260,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.007936507936507936,
|
| 13 |
+
"grad_norm": 10.67099666595459,
|
| 14 |
+
"learning_rate": 7.936507936507937e-07,
|
| 15 |
+
"loss": 1.6937,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.015873015873015872,
|
| 20 |
+
"grad_norm": 10.839055061340332,
|
| 21 |
+
"learning_rate": 1.5873015873015873e-06,
|
| 22 |
+
"loss": 1.6052,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.023809523809523808,
|
| 27 |
+
"grad_norm": 10.671605110168457,
|
| 28 |
+
"learning_rate": 2.380952380952381e-06,
|
| 29 |
+
"loss": 1.5128,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.031746031746031744,
|
| 34 |
+
"grad_norm": 10.129815101623535,
|
| 35 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 36 |
+
"loss": 1.3375,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03968253968253968,
|
| 41 |
+
"grad_norm": 7.233930587768555,
|
| 42 |
+
"learning_rate": 3.968253968253968e-06,
|
| 43 |
+
"loss": 1.1729,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.047619047619047616,
|
| 48 |
+
"grad_norm": 7.941068172454834,
|
| 49 |
+
"learning_rate": 4.761904761904762e-06,
|
| 50 |
+
"loss": 1.0557,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05555555555555555,
|
| 55 |
+
"grad_norm": 5.641613483428955,
|
| 56 |
+
"learning_rate": 5.555555555555557e-06,
|
| 57 |
+
"loss": 0.8538,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.06349206349206349,
|
| 62 |
+
"grad_norm": 11.44336986541748,
|
| 63 |
+
"learning_rate": 6.349206349206349e-06,
|
| 64 |
+
"loss": 0.9715,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.07142857142857142,
|
| 69 |
+
"grad_norm": 6.217037200927734,
|
| 70 |
+
"learning_rate": 7.1428571428571436e-06,
|
| 71 |
+
"loss": 0.9786,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.07936507936507936,
|
| 76 |
+
"grad_norm": 8.418871879577637,
|
| 77 |
+
"learning_rate": 7.936507936507936e-06,
|
| 78 |
+
"loss": 0.8612,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.0873015873015873,
|
| 83 |
+
"grad_norm": 8.026611328125,
|
| 84 |
+
"learning_rate": 8.730158730158731e-06,
|
| 85 |
+
"loss": 0.9033,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.09523809523809523,
|
| 90 |
+
"grad_norm": 5.777014255523682,
|
| 91 |
+
"learning_rate": 9.523809523809525e-06,
|
| 92 |
+
"loss": 1.2097,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1007936507936508,
|
| 97 |
+
"eval_accuracy": 0.6899563318777293,
|
| 98 |
+
"eval_accuracy_hold": 0.0,
|
| 99 |
+
"eval_accuracy_preparation": 1.0,
|
| 100 |
+
"eval_accuracy_recovery": 0.0,
|
| 101 |
+
"eval_accuracy_stroke": 0.0,
|
| 102 |
+
"eval_accuracy_unknown": 0.0,
|
| 103 |
+
"eval_loss": 1.0243256092071533,
|
| 104 |
+
"eval_runtime": 22.3266,
|
| 105 |
+
"eval_samples_per_second": 10.257,
|
| 106 |
+
"eval_steps_per_second": 1.299,
|
| 107 |
+
"step": 127
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.0023809523809524,
|
| 111 |
+
"grad_norm": 4.667547225952148,
|
| 112 |
+
"learning_rate": 9.9647266313933e-06,
|
| 113 |
+
"loss": 0.9759,
|
| 114 |
+
"step": 130
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.0103174603174603,
|
| 118 |
+
"grad_norm": 6.570101261138916,
|
| 119 |
+
"learning_rate": 9.876543209876543e-06,
|
| 120 |
+
"loss": 0.8687,
|
| 121 |
+
"step": 140
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0182539682539682,
|
| 125 |
+
"grad_norm": 3.454606533050537,
|
| 126 |
+
"learning_rate": 9.788359788359789e-06,
|
| 127 |
+
"loss": 0.9626,
|
| 128 |
+
"step": 150
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.026190476190476,
|
| 132 |
+
"grad_norm": 5.504855632781982,
|
| 133 |
+
"learning_rate": 9.700176366843034e-06,
|
| 134 |
+
"loss": 0.8062,
|
| 135 |
+
"step": 160
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0341269841269842,
|
| 139 |
+
"grad_norm": 11.000463485717773,
|
| 140 |
+
"learning_rate": 9.61199294532628e-06,
|
| 141 |
+
"loss": 0.8868,
|
| 142 |
+
"step": 170
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.0420634920634921,
|
| 146 |
+
"grad_norm": 9.436511039733887,
|
| 147 |
+
"learning_rate": 9.523809523809525e-06,
|
| 148 |
+
"loss": 0.7972,
|
| 149 |
+
"step": 180
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.05,
|
| 153 |
+
"grad_norm": 4.840676784515381,
|
| 154 |
+
"learning_rate": 9.43562610229277e-06,
|
| 155 |
+
"loss": 0.9835,
|
| 156 |
+
"step": 190
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.057936507936508,
|
| 160 |
+
"grad_norm": 5.491481781005859,
|
| 161 |
+
"learning_rate": 9.347442680776014e-06,
|
| 162 |
+
"loss": 0.8156,
|
| 163 |
+
"step": 200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.0658730158730159,
|
| 167 |
+
"grad_norm": 9.045355796813965,
|
| 168 |
+
"learning_rate": 9.25925925925926e-06,
|
| 169 |
+
"loss": 0.9068,
|
| 170 |
+
"step": 210
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.0738095238095238,
|
| 174 |
+
"grad_norm": 6.567234516143799,
|
| 175 |
+
"learning_rate": 9.171075837742504e-06,
|
| 176 |
+
"loss": 0.9518,
|
| 177 |
+
"step": 220
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.0817460317460317,
|
| 181 |
+
"grad_norm": 4.041014194488525,
|
| 182 |
+
"learning_rate": 9.08289241622575e-06,
|
| 183 |
+
"loss": 0.8425,
|
| 184 |
+
"step": 230
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 1.0896825396825396,
|
| 188 |
+
"grad_norm": 5.872761249542236,
|
| 189 |
+
"learning_rate": 8.994708994708995e-06,
|
| 190 |
+
"loss": 0.7499,
|
| 191 |
+
"step": 240
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.0976190476190477,
|
| 195 |
+
"grad_norm": 5.4285888671875,
|
| 196 |
+
"learning_rate": 8.90652557319224e-06,
|
| 197 |
+
"loss": 0.8,
|
| 198 |
+
"step": 250
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.1007936507936509,
|
| 202 |
+
"eval_accuracy": 0.6899563318777293,
|
| 203 |
+
"eval_accuracy_hold": 0.0,
|
| 204 |
+
"eval_accuracy_preparation": 1.0,
|
| 205 |
+
"eval_accuracy_recovery": 0.0,
|
| 206 |
+
"eval_accuracy_stroke": 0.0,
|
| 207 |
+
"eval_accuracy_unknown": 0.0,
|
| 208 |
+
"eval_loss": 0.8716774582862854,
|
| 209 |
+
"eval_runtime": 18.6999,
|
| 210 |
+
"eval_samples_per_second": 12.246,
|
| 211 |
+
"eval_steps_per_second": 1.551,
|
| 212 |
+
"step": 254
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.0047619047619047,
|
| 216 |
+
"grad_norm": 7.916933536529541,
|
| 217 |
+
"learning_rate": 8.818342151675486e-06,
|
| 218 |
+
"loss": 0.8939,
|
| 219 |
+
"step": 260
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.0126984126984127,
|
| 223 |
+
"grad_norm": 10.032241821289062,
|
| 224 |
+
"learning_rate": 8.730158730158731e-06,
|
| 225 |
+
"loss": 0.9507,
|
| 226 |
+
"step": 270
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 2.0206349206349206,
|
| 230 |
+
"grad_norm": 5.500053405761719,
|
| 231 |
+
"learning_rate": 8.641975308641975e-06,
|
| 232 |
+
"loss": 0.7524,
|
| 233 |
+
"step": 280
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 2.0285714285714285,
|
| 237 |
+
"grad_norm": 2.9224343299865723,
|
| 238 |
+
"learning_rate": 8.553791887125222e-06,
|
| 239 |
+
"loss": 0.8033,
|
| 240 |
+
"step": 290
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 2.0365079365079364,
|
| 244 |
+
"grad_norm": 9.924591064453125,
|
| 245 |
+
"learning_rate": 8.465608465608466e-06,
|
| 246 |
+
"loss": 0.7143,
|
| 247 |
+
"step": 300
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 2.0444444444444443,
|
| 251 |
+
"grad_norm": 6.131772518157959,
|
| 252 |
+
"learning_rate": 8.377425044091711e-06,
|
| 253 |
+
"loss": 0.8056,
|
| 254 |
+
"step": 310
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.052380952380952,
|
| 258 |
+
"grad_norm": 5.074908256530762,
|
| 259 |
+
"learning_rate": 8.289241622574956e-06,
|
| 260 |
+
"loss": 0.7329,
|
| 261 |
+
"step": 320
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 2.06031746031746,
|
| 265 |
+
"grad_norm": 6.0842061042785645,
|
| 266 |
+
"learning_rate": 8.201058201058202e-06,
|
| 267 |
+
"loss": 0.5516,
|
| 268 |
+
"step": 330
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 2.0682539682539685,
|
| 272 |
+
"grad_norm": 6.810462951660156,
|
| 273 |
+
"learning_rate": 8.112874779541447e-06,
|
| 274 |
+
"loss": 0.6635,
|
| 275 |
+
"step": 340
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.0761904761904764,
|
| 279 |
+
"grad_norm": 8.190900802612305,
|
| 280 |
+
"learning_rate": 8.024691358024692e-06,
|
| 281 |
+
"loss": 0.707,
|
| 282 |
+
"step": 350
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 2.0841269841269843,
|
| 286 |
+
"grad_norm": 6.851507663726807,
|
| 287 |
+
"learning_rate": 7.936507936507936e-06,
|
| 288 |
+
"loss": 0.8107,
|
| 289 |
+
"step": 360
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 2.092063492063492,
|
| 293 |
+
"grad_norm": 10.986032485961914,
|
| 294 |
+
"learning_rate": 7.848324514991183e-06,
|
| 295 |
+
"loss": 0.5181,
|
| 296 |
+
"step": 370
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.1,
|
| 300 |
+
"grad_norm": 7.873786926269531,
|
| 301 |
+
"learning_rate": 7.760141093474427e-06,
|
| 302 |
+
"loss": 0.5199,
|
| 303 |
+
"step": 380
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 2.100793650793651,
|
| 307 |
+
"eval_accuracy": 0.7729257641921398,
|
| 308 |
+
"eval_accuracy_hold": 0.0,
|
| 309 |
+
"eval_accuracy_preparation": 0.9810126582278481,
|
| 310 |
+
"eval_accuracy_recovery": 0.17647058823529413,
|
| 311 |
+
"eval_accuracy_stroke": 0.0,
|
| 312 |
+
"eval_accuracy_unknown": 0.8636363636363636,
|
| 313 |
+
"eval_loss": 0.6724938750267029,
|
| 314 |
+
"eval_runtime": 17.1616,
|
| 315 |
+
"eval_samples_per_second": 13.344,
|
| 316 |
+
"eval_steps_per_second": 1.69,
|
| 317 |
+
"step": 381
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.007142857142857,
|
| 321 |
+
"grad_norm": 9.23725700378418,
|
| 322 |
+
"learning_rate": 7.671957671957672e-06,
|
| 323 |
+
"loss": 0.5796,
|
| 324 |
+
"step": 390
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 3.015079365079365,
|
| 328 |
+
"grad_norm": 14.487571716308594,
|
| 329 |
+
"learning_rate": 7.583774250440918e-06,
|
| 330 |
+
"loss": 0.6356,
|
| 331 |
+
"step": 400
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 3.023015873015873,
|
| 335 |
+
"grad_norm": 8.111766815185547,
|
| 336 |
+
"learning_rate": 7.495590828924163e-06,
|
| 337 |
+
"loss": 0.609,
|
| 338 |
+
"step": 410
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 3.030952380952381,
|
| 342 |
+
"grad_norm": 10.31395149230957,
|
| 343 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 344 |
+
"loss": 0.4853,
|
| 345 |
+
"step": 420
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 3.0388888888888888,
|
| 349 |
+
"grad_norm": 9.713062286376953,
|
| 350 |
+
"learning_rate": 7.319223985890654e-06,
|
| 351 |
+
"loss": 0.4254,
|
| 352 |
+
"step": 430
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 3.0468253968253967,
|
| 356 |
+
"grad_norm": 24.236248016357422,
|
| 357 |
+
"learning_rate": 7.231040564373898e-06,
|
| 358 |
+
"loss": 0.6144,
|
| 359 |
+
"step": 440
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.0547619047619046,
|
| 363 |
+
"grad_norm": 9.259011268615723,
|
| 364 |
+
"learning_rate": 7.1428571428571436e-06,
|
| 365 |
+
"loss": 0.5036,
|
| 366 |
+
"step": 450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 3.0626984126984125,
|
| 370 |
+
"grad_norm": 9.141166687011719,
|
| 371 |
+
"learning_rate": 7.054673721340388e-06,
|
| 372 |
+
"loss": 0.5061,
|
| 373 |
+
"step": 460
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 3.070634920634921,
|
| 377 |
+
"grad_norm": 1.9605706930160522,
|
| 378 |
+
"learning_rate": 6.966490299823634e-06,
|
| 379 |
+
"loss": 0.3654,
|
| 380 |
+
"step": 470
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 3.0785714285714287,
|
| 384 |
+
"grad_norm": 23.66200828552246,
|
| 385 |
+
"learning_rate": 6.878306878306879e-06,
|
| 386 |
+
"loss": 0.5331,
|
| 387 |
+
"step": 480
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 3.0865079365079366,
|
| 391 |
+
"grad_norm": 17.01007843017578,
|
| 392 |
+
"learning_rate": 6.790123456790124e-06,
|
| 393 |
+
"loss": 0.491,
|
| 394 |
+
"step": 490
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 3.0944444444444446,
|
| 398 |
+
"grad_norm": 18.177013397216797,
|
| 399 |
+
"learning_rate": 6.701940035273369e-06,
|
| 400 |
+
"loss": 0.3134,
|
| 401 |
+
"step": 500
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.100793650793651,
|
| 405 |
+
"eval_accuracy": 0.8427947598253275,
|
| 406 |
+
"eval_accuracy_hold": 0.3076923076923077,
|
| 407 |
+
"eval_accuracy_preparation": 1.0,
|
| 408 |
+
"eval_accuracy_recovery": 0.4117647058823529,
|
| 409 |
+
"eval_accuracy_stroke": 0.0,
|
| 410 |
+
"eval_accuracy_unknown": 0.9090909090909091,
|
| 411 |
+
"eval_loss": 0.47145798802375793,
|
| 412 |
+
"eval_runtime": 20.9475,
|
| 413 |
+
"eval_samples_per_second": 10.932,
|
| 414 |
+
"eval_steps_per_second": 1.384,
|
| 415 |
+
"step": 508
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 4.001587301587302,
|
| 419 |
+
"grad_norm": 3.7459118366241455,
|
| 420 |
+
"learning_rate": 6.613756613756615e-06,
|
| 421 |
+
"loss": 0.2951,
|
| 422 |
+
"step": 510
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 4.0095238095238095,
|
| 426 |
+
"grad_norm": 1.2948265075683594,
|
| 427 |
+
"learning_rate": 6.525573192239859e-06,
|
| 428 |
+
"loss": 0.3015,
|
| 429 |
+
"step": 520
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 4.017460317460317,
|
| 433 |
+
"grad_norm": 28.894683837890625,
|
| 434 |
+
"learning_rate": 6.437389770723105e-06,
|
| 435 |
+
"loss": 0.5519,
|
| 436 |
+
"step": 530
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 4.025396825396825,
|
| 440 |
+
"grad_norm": 12.930984497070312,
|
| 441 |
+
"learning_rate": 6.349206349206349e-06,
|
| 442 |
+
"loss": 0.3167,
|
| 443 |
+
"step": 540
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 4.033333333333333,
|
| 447 |
+
"grad_norm": 14.674068450927734,
|
| 448 |
+
"learning_rate": 6.2610229276895955e-06,
|
| 449 |
+
"loss": 0.3852,
|
| 450 |
+
"step": 550
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 4.041269841269841,
|
| 454 |
+
"grad_norm": 5.561566352844238,
|
| 455 |
+
"learning_rate": 6.17283950617284e-06,
|
| 456 |
+
"loss": 0.2827,
|
| 457 |
+
"step": 560
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 4.049206349206349,
|
| 461 |
+
"grad_norm": 2.773526906967163,
|
| 462 |
+
"learning_rate": 6.084656084656085e-06,
|
| 463 |
+
"loss": 0.3025,
|
| 464 |
+
"step": 570
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.057142857142857,
|
| 468 |
+
"grad_norm": 2.659675121307373,
|
| 469 |
+
"learning_rate": 5.99647266313933e-06,
|
| 470 |
+
"loss": 0.2043,
|
| 471 |
+
"step": 580
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 4.065079365079365,
|
| 475 |
+
"grad_norm": 3.98606538772583,
|
| 476 |
+
"learning_rate": 5.908289241622576e-06,
|
| 477 |
+
"loss": 0.3236,
|
| 478 |
+
"step": 590
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 4.073015873015873,
|
| 482 |
+
"grad_norm": 27.466781616210938,
|
| 483 |
+
"learning_rate": 5.820105820105821e-06,
|
| 484 |
+
"loss": 0.1979,
|
| 485 |
+
"step": 600
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 4.080952380952381,
|
| 489 |
+
"grad_norm": 5.185789108276367,
|
| 490 |
+
"learning_rate": 5.731922398589066e-06,
|
| 491 |
+
"loss": 0.2881,
|
| 492 |
+
"step": 610
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 4.088888888888889,
|
| 496 |
+
"grad_norm": 11.606480598449707,
|
| 497 |
+
"learning_rate": 5.6437389770723105e-06,
|
| 498 |
+
"loss": 0.1399,
|
| 499 |
+
"step": 620
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 4.0968253968253965,
|
| 503 |
+
"grad_norm": 31.712146759033203,
|
| 504 |
+
"learning_rate": 5.555555555555557e-06,
|
| 505 |
+
"loss": 0.1561,
|
| 506 |
+
"step": 630
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.10079365079365,
|
| 510 |
+
"eval_accuracy": 0.8951965065502183,
|
| 511 |
+
"eval_accuracy_hold": 0.7692307692307693,
|
| 512 |
+
"eval_accuracy_preparation": 1.0,
|
| 513 |
+
"eval_accuracy_recovery": 0.7058823529411765,
|
| 514 |
+
"eval_accuracy_stroke": 0.0,
|
| 515 |
+
"eval_accuracy_unknown": 0.6818181818181818,
|
| 516 |
+
"eval_loss": 0.43630751967430115,
|
| 517 |
+
"eval_runtime": 18.0386,
|
| 518 |
+
"eval_samples_per_second": 12.695,
|
| 519 |
+
"eval_steps_per_second": 1.608,
|
| 520 |
+
"step": 635
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 5.003968253968254,
|
| 524 |
+
"grad_norm": 0.8358302116394043,
|
| 525 |
+
"learning_rate": 5.467372134038801e-06,
|
| 526 |
+
"loss": 0.4914,
|
| 527 |
+
"step": 640
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 5.011904761904762,
|
| 531 |
+
"grad_norm": 26.23114585876465,
|
| 532 |
+
"learning_rate": 5.3791887125220465e-06,
|
| 533 |
+
"loss": 0.1426,
|
| 534 |
+
"step": 650
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 5.01984126984127,
|
| 538 |
+
"grad_norm": 21.137874603271484,
|
| 539 |
+
"learning_rate": 5.291005291005291e-06,
|
| 540 |
+
"loss": 0.3177,
|
| 541 |
+
"step": 660
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 5.027777777777778,
|
| 545 |
+
"grad_norm": 13.185637474060059,
|
| 546 |
+
"learning_rate": 5.202821869488537e-06,
|
| 547 |
+
"loss": 0.2641,
|
| 548 |
+
"step": 670
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 5.035714285714286,
|
| 552 |
+
"grad_norm": 22.404172897338867,
|
| 553 |
+
"learning_rate": 5.114638447971782e-06,
|
| 554 |
+
"loss": 0.122,
|
| 555 |
+
"step": 680
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 5.0436507936507935,
|
| 559 |
+
"grad_norm": 1.1821787357330322,
|
| 560 |
+
"learning_rate": 5.026455026455027e-06,
|
| 561 |
+
"loss": 0.214,
|
| 562 |
+
"step": 690
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 5.051587301587301,
|
| 566 |
+
"grad_norm": 0.8503882884979248,
|
| 567 |
+
"learning_rate": 4.938271604938272e-06,
|
| 568 |
+
"loss": 0.3925,
|
| 569 |
+
"step": 700
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 5.059523809523809,
|
| 573 |
+
"grad_norm": 26.318893432617188,
|
| 574 |
+
"learning_rate": 4.850088183421517e-06,
|
| 575 |
+
"loss": 0.1443,
|
| 576 |
+
"step": 710
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 5.067460317460317,
|
| 580 |
+
"grad_norm": 111.941162109375,
|
| 581 |
+
"learning_rate": 4.761904761904762e-06,
|
| 582 |
+
"loss": 0.1638,
|
| 583 |
+
"step": 720
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 5.075396825396825,
|
| 587 |
+
"grad_norm": 27.32931137084961,
|
| 588 |
+
"learning_rate": 4.673721340388007e-06,
|
| 589 |
+
"loss": 0.3608,
|
| 590 |
+
"step": 730
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 5.083333333333333,
|
| 594 |
+
"grad_norm": 117.25592041015625,
|
| 595 |
+
"learning_rate": 4.585537918871252e-06,
|
| 596 |
+
"loss": 0.0645,
|
| 597 |
+
"step": 740
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 5.091269841269841,
|
| 601 |
+
"grad_norm": 12.846317291259766,
|
| 602 |
+
"learning_rate": 4.497354497354498e-06,
|
| 603 |
+
"loss": 0.2111,
|
| 604 |
+
"step": 750
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 5.099206349206349,
|
| 608 |
+
"grad_norm": 45.810054779052734,
|
| 609 |
+
"learning_rate": 4.409171075837743e-06,
|
| 610 |
+
"loss": 0.0429,
|
| 611 |
+
"step": 760
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 5.10079365079365,
|
| 615 |
+
"eval_accuracy": 0.9432314410480349,
|
| 616 |
+
"eval_accuracy_hold": 0.8846153846153846,
|
| 617 |
+
"eval_accuracy_preparation": 0.9936708860759493,
|
| 618 |
+
"eval_accuracy_recovery": 0.7058823529411765,
|
| 619 |
+
"eval_accuracy_stroke": 0.5,
|
| 620 |
+
"eval_accuracy_unknown": 0.9545454545454546,
|
| 621 |
+
"eval_loss": 0.2210746556520462,
|
| 622 |
+
"eval_runtime": 17.8804,
|
| 623 |
+
"eval_samples_per_second": 12.807,
|
| 624 |
+
"eval_steps_per_second": 1.622,
|
| 625 |
+
"step": 762
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 6.006349206349206,
|
| 629 |
+
"grad_norm": 35.88644027709961,
|
| 630 |
+
"learning_rate": 4.3209876543209875e-06,
|
| 631 |
+
"loss": 0.1086,
|
| 632 |
+
"step": 770
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.014285714285714,
|
| 636 |
+
"grad_norm": 0.47389915585517883,
|
| 637 |
+
"learning_rate": 4.232804232804233e-06,
|
| 638 |
+
"loss": 0.2244,
|
| 639 |
+
"step": 780
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 6.022222222222222,
|
| 643 |
+
"grad_norm": 0.22194349765777588,
|
| 644 |
+
"learning_rate": 4.144620811287478e-06,
|
| 645 |
+
"loss": 0.1538,
|
| 646 |
+
"step": 790
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 6.03015873015873,
|
| 650 |
+
"grad_norm": 50.78633499145508,
|
| 651 |
+
"learning_rate": 4.0564373897707236e-06,
|
| 652 |
+
"loss": 0.1374,
|
| 653 |
+
"step": 800
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 6.038095238095238,
|
| 657 |
+
"grad_norm": 0.39046162366867065,
|
| 658 |
+
"learning_rate": 3.968253968253968e-06,
|
| 659 |
+
"loss": 0.3255,
|
| 660 |
+
"step": 810
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 6.046031746031746,
|
| 664 |
+
"grad_norm": 56.010379791259766,
|
| 665 |
+
"learning_rate": 3.8800705467372134e-06,
|
| 666 |
+
"loss": 0.1618,
|
| 667 |
+
"step": 820
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 6.053968253968254,
|
| 671 |
+
"grad_norm": 0.16263973712921143,
|
| 672 |
+
"learning_rate": 3.791887125220459e-06,
|
| 673 |
+
"loss": 0.116,
|
| 674 |
+
"step": 830
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 6.061904761904762,
|
| 678 |
+
"grad_norm": 0.23908407986164093,
|
| 679 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 680 |
+
"loss": 0.1012,
|
| 681 |
+
"step": 840
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 6.06984126984127,
|
| 685 |
+
"grad_norm": 53.480140686035156,
|
| 686 |
+
"learning_rate": 3.615520282186949e-06,
|
| 687 |
+
"loss": 0.1767,
|
| 688 |
+
"step": 850
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 6.0777777777777775,
|
| 692 |
+
"grad_norm": 0.681148886680603,
|
| 693 |
+
"learning_rate": 3.527336860670194e-06,
|
| 694 |
+
"loss": 0.1364,
|
| 695 |
+
"step": 860
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 6.085714285714285,
|
| 699 |
+
"grad_norm": 6.9544267654418945,
|
| 700 |
+
"learning_rate": 3.4391534391534394e-06,
|
| 701 |
+
"loss": 0.0884,
|
| 702 |
+
"step": 870
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 6.093650793650793,
|
| 706 |
+
"grad_norm": 0.3266388177871704,
|
| 707 |
+
"learning_rate": 3.3509700176366843e-06,
|
| 708 |
+
"loss": 0.2294,
|
| 709 |
+
"step": 880
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 6.10079365079365,
|
| 713 |
+
"eval_accuracy": 0.9475982532751092,
|
| 714 |
+
"eval_accuracy_hold": 0.8846153846153846,
|
| 715 |
+
"eval_accuracy_preparation": 1.0,
|
| 716 |
+
"eval_accuracy_recovery": 0.8823529411764706,
|
| 717 |
+
"eval_accuracy_stroke": 0.16666666666666666,
|
| 718 |
+
"eval_accuracy_unknown": 0.9090909090909091,
|
| 719 |
+
"eval_loss": 0.20944689214229584,
|
| 720 |
+
"eval_runtime": 16.986,
|
| 721 |
+
"eval_samples_per_second": 13.482,
|
| 722 |
+
"eval_steps_per_second": 1.707,
|
| 723 |
+
"step": 889
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 7.000793650793651,
|
| 727 |
+
"grad_norm": 4.641857147216797,
|
| 728 |
+
"learning_rate": 3.2627865961199297e-06,
|
| 729 |
+
"loss": 0.1946,
|
| 730 |
+
"step": 890
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 7.008730158730159,
|
| 734 |
+
"grad_norm": 33.782859802246094,
|
| 735 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 736 |
+
"loss": 0.0758,
|
| 737 |
+
"step": 900
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 7.016666666666667,
|
| 741 |
+
"grad_norm": 0.22343800961971283,
|
| 742 |
+
"learning_rate": 3.08641975308642e-06,
|
| 743 |
+
"loss": 0.0885,
|
| 744 |
+
"step": 910
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 7.0246031746031745,
|
| 748 |
+
"grad_norm": 4.360742092132568,
|
| 749 |
+
"learning_rate": 2.998236331569665e-06,
|
| 750 |
+
"loss": 0.016,
|
| 751 |
+
"step": 920
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 7.032539682539682,
|
| 755 |
+
"grad_norm": 0.2488667219877243,
|
| 756 |
+
"learning_rate": 2.9100529100529103e-06,
|
| 757 |
+
"loss": 0.1495,
|
| 758 |
+
"step": 930
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 7.04047619047619,
|
| 762 |
+
"grad_norm": 28.20427703857422,
|
| 763 |
+
"learning_rate": 2.8218694885361552e-06,
|
| 764 |
+
"loss": 0.1959,
|
| 765 |
+
"step": 940
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 7.048412698412698,
|
| 769 |
+
"grad_norm": 8.172640800476074,
|
| 770 |
+
"learning_rate": 2.7336860670194006e-06,
|
| 771 |
+
"loss": 0.0934,
|
| 772 |
+
"step": 950
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 7.056349206349206,
|
| 776 |
+
"grad_norm": 9.601517677307129,
|
| 777 |
+
"learning_rate": 2.6455026455026455e-06,
|
| 778 |
+
"loss": 0.2069,
|
| 779 |
+
"step": 960
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 7.064285714285714,
|
| 783 |
+
"grad_norm": 17.030954360961914,
|
| 784 |
+
"learning_rate": 2.557319223985891e-06,
|
| 785 |
+
"loss": 0.0529,
|
| 786 |
+
"step": 970
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 7.072222222222222,
|
| 790 |
+
"grad_norm": 0.2699105143547058,
|
| 791 |
+
"learning_rate": 2.469135802469136e-06,
|
| 792 |
+
"loss": 0.236,
|
| 793 |
+
"step": 980
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 7.08015873015873,
|
| 797 |
+
"grad_norm": 0.7129911184310913,
|
| 798 |
+
"learning_rate": 2.380952380952381e-06,
|
| 799 |
+
"loss": 0.015,
|
| 800 |
+
"step": 990
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 7.088095238095238,
|
| 804 |
+
"grad_norm": 20.803991317749023,
|
| 805 |
+
"learning_rate": 2.292768959435626e-06,
|
| 806 |
+
"loss": 0.07,
|
| 807 |
+
"step": 1000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 7.096031746031746,
|
| 811 |
+
"grad_norm": 31.60879135131836,
|
| 812 |
+
"learning_rate": 2.2045855379188715e-06,
|
| 813 |
+
"loss": 0.1214,
|
| 814 |
+
"step": 1010
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 7.10079365079365,
|
| 818 |
+
"eval_accuracy": 0.9606986899563319,
|
| 819 |
+
"eval_accuracy_hold": 0.8846153846153846,
|
| 820 |
+
"eval_accuracy_preparation": 0.9936708860759493,
|
| 821 |
+
"eval_accuracy_recovery": 0.8823529411764706,
|
| 822 |
+
"eval_accuracy_stroke": 0.6666666666666666,
|
| 823 |
+
"eval_accuracy_unknown": 0.9545454545454546,
|
| 824 |
+
"eval_loss": 0.15864819288253784,
|
| 825 |
+
"eval_runtime": 198.6859,
|
| 826 |
+
"eval_samples_per_second": 1.153,
|
| 827 |
+
"eval_steps_per_second": 0.146,
|
| 828 |
+
"step": 1016
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 8.003174603174603,
|
| 832 |
+
"grad_norm": 0.32036104798316956,
|
| 833 |
+
"learning_rate": 2.1164021164021164e-06,
|
| 834 |
+
"loss": 0.0981,
|
| 835 |
+
"step": 1020
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 8.011111111111111,
|
| 839 |
+
"grad_norm": 0.1587836891412735,
|
| 840 |
+
"learning_rate": 2.0282186948853618e-06,
|
| 841 |
+
"loss": 0.1222,
|
| 842 |
+
"step": 1030
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 8.019047619047619,
|
| 846 |
+
"grad_norm": 9.007894515991211,
|
| 847 |
+
"learning_rate": 1.9400352733686067e-06,
|
| 848 |
+
"loss": 0.2571,
|
| 849 |
+
"step": 1040
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 8.026984126984127,
|
| 853 |
+
"grad_norm": 0.25755736231803894,
|
| 854 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 855 |
+
"loss": 0.0791,
|
| 856 |
+
"step": 1050
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 8.034920634920635,
|
| 860 |
+
"grad_norm": 6.88518762588501,
|
| 861 |
+
"learning_rate": 1.763668430335097e-06,
|
| 862 |
+
"loss": 0.0597,
|
| 863 |
+
"step": 1060
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 8.042857142857143,
|
| 867 |
+
"grad_norm": 28.681108474731445,
|
| 868 |
+
"learning_rate": 1.6754850088183422e-06,
|
| 869 |
+
"loss": 0.1355,
|
| 870 |
+
"step": 1070
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 8.05079365079365,
|
| 874 |
+
"grad_norm": 0.7820481061935425,
|
| 875 |
+
"learning_rate": 1.5873015873015873e-06,
|
| 876 |
+
"loss": 0.0102,
|
| 877 |
+
"step": 1080
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 8.058730158730159,
|
| 881 |
+
"grad_norm": 0.11372612416744232,
|
| 882 |
+
"learning_rate": 1.4991181657848325e-06,
|
| 883 |
+
"loss": 0.0969,
|
| 884 |
+
"step": 1090
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 8.066666666666666,
|
| 888 |
+
"grad_norm": 0.962800145149231,
|
| 889 |
+
"learning_rate": 1.4109347442680776e-06,
|
| 890 |
+
"loss": 0.0633,
|
| 891 |
+
"step": 1100
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 8.074603174603174,
|
| 895 |
+
"grad_norm": 0.29850736260414124,
|
| 896 |
+
"learning_rate": 1.3227513227513228e-06,
|
| 897 |
+
"loss": 0.0743,
|
| 898 |
+
"step": 1110
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 8.082539682539682,
|
| 902 |
+
"grad_norm": 0.2638128101825714,
|
| 903 |
+
"learning_rate": 1.234567901234568e-06,
|
| 904 |
+
"loss": 0.0603,
|
| 905 |
+
"step": 1120
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 8.09047619047619,
|
| 909 |
+
"grad_norm": 4.3052215576171875,
|
| 910 |
+
"learning_rate": 1.146384479717813e-06,
|
| 911 |
+
"loss": 0.1776,
|
| 912 |
+
"step": 1130
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 8.098412698412698,
|
| 916 |
+
"grad_norm": 7.065666675567627,
|
| 917 |
+
"learning_rate": 1.0582010582010582e-06,
|
| 918 |
+
"loss": 0.1478,
|
| 919 |
+
"step": 1140
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 8.100793650793651,
|
| 923 |
+
"eval_accuracy": 0.9432314410480349,
|
| 924 |
+
"eval_accuracy_hold": 0.8846153846153846,
|
| 925 |
+
"eval_accuracy_preparation": 0.9683544303797469,
|
| 926 |
+
"eval_accuracy_recovery": 0.8823529411764706,
|
| 927 |
+
"eval_accuracy_stroke": 0.6666666666666666,
|
| 928 |
+
"eval_accuracy_unknown": 0.9545454545454546,
|
| 929 |
+
"eval_loss": 0.16072985529899597,
|
| 930 |
+
"eval_runtime": 38.5322,
|
| 931 |
+
"eval_samples_per_second": 5.943,
|
| 932 |
+
"eval_steps_per_second": 0.753,
|
| 933 |
+
"step": 1143
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 9.005555555555556,
|
| 937 |
+
"grad_norm": 1.259048342704773,
|
| 938 |
+
"learning_rate": 9.700176366843034e-07,
|
| 939 |
+
"loss": 0.0626,
|
| 940 |
+
"step": 1150
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 9.013492063492064,
|
| 944 |
+
"grad_norm": 13.264341354370117,
|
| 945 |
+
"learning_rate": 8.818342151675485e-07,
|
| 946 |
+
"loss": 0.1718,
|
| 947 |
+
"step": 1160
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 9.021428571428572,
|
| 951 |
+
"grad_norm": 0.916714608669281,
|
| 952 |
+
"learning_rate": 7.936507936507937e-07,
|
| 953 |
+
"loss": 0.0135,
|
| 954 |
+
"step": 1170
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 9.02936507936508,
|
| 958 |
+
"grad_norm": 0.167047917842865,
|
| 959 |
+
"learning_rate": 7.054673721340388e-07,
|
| 960 |
+
"loss": 0.087,
|
| 961 |
+
"step": 1180
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 9.037301587301588,
|
| 965 |
+
"grad_norm": 0.09918953478336334,
|
| 966 |
+
"learning_rate": 6.17283950617284e-07,
|
| 967 |
+
"loss": 0.0936,
|
| 968 |
+
"step": 1190
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 9.045238095238096,
|
| 972 |
+
"grad_norm": 9.908056259155273,
|
| 973 |
+
"learning_rate": 5.291005291005291e-07,
|
| 974 |
+
"loss": 0.1706,
|
| 975 |
+
"step": 1200
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 9.053174603174604,
|
| 979 |
+
"grad_norm": 0.15841275453567505,
|
| 980 |
+
"learning_rate": 4.4091710758377425e-07,
|
| 981 |
+
"loss": 0.0411,
|
| 982 |
+
"step": 1210
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 9.061111111111112,
|
| 986 |
+
"grad_norm": 9.066829681396484,
|
| 987 |
+
"learning_rate": 3.527336860670194e-07,
|
| 988 |
+
"loss": 0.0798,
|
| 989 |
+
"step": 1220
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 9.06904761904762,
|
| 993 |
+
"grad_norm": 15.856799125671387,
|
| 994 |
+
"learning_rate": 2.6455026455026455e-07,
|
| 995 |
+
"loss": 0.0698,
|
| 996 |
+
"step": 1230
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 9.076984126984128,
|
| 1000 |
+
"grad_norm": 57.29991149902344,
|
| 1001 |
+
"learning_rate": 1.763668430335097e-07,
|
| 1002 |
+
"loss": 0.0994,
|
| 1003 |
+
"step": 1240
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 9.084920634920636,
|
| 1007 |
+
"grad_norm": 0.18446692824363708,
|
| 1008 |
+
"learning_rate": 8.818342151675485e-08,
|
| 1009 |
+
"loss": 0.1576,
|
| 1010 |
+
"step": 1250
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 9.092857142857143,
|
| 1014 |
+
"grad_norm": 0.1289648562669754,
|
| 1015 |
+
"learning_rate": 0.0,
|
| 1016 |
+
"loss": 0.1156,
|
| 1017 |
+
"step": 1260
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 9.092857142857143,
|
| 1021 |
+
"eval_accuracy": 0.9737991266375546,
|
| 1022 |
+
"eval_accuracy_hold": 1.0,
|
| 1023 |
+
"eval_accuracy_preparation": 0.9936708860759493,
|
| 1024 |
+
"eval_accuracy_recovery": 0.8823529411764706,
|
| 1025 |
+
"eval_accuracy_stroke": 0.6666666666666666,
|
| 1026 |
+
"eval_accuracy_unknown": 0.9545454545454546,
|
| 1027 |
+
"eval_loss": 0.11774494498968124,
|
| 1028 |
+
"eval_runtime": 38.5752,
|
| 1029 |
+
"eval_samples_per_second": 5.936,
|
| 1030 |
+
"eval_steps_per_second": 0.752,
|
| 1031 |
+
"step": 1260
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 9.092857142857143,
|
| 1035 |
+
"step": 1260,
|
| 1036 |
+
"total_flos": 1.2549452224921436e+19,
|
| 1037 |
+
"train_loss": 0.42608252665589724,
|
| 1038 |
+
"train_runtime": 2594.6501,
|
| 1039 |
+
"train_samples_per_second": 3.885,
|
| 1040 |
+
"train_steps_per_second": 0.486
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 9.092857142857143,
|
| 1044 |
+
"eval_accuracy": 0.946058091286307,
|
| 1045 |
+
"eval_accuracy_hold": 1.0,
|
| 1046 |
+
"eval_accuracy_preparation": 0.9685534591194969,
|
| 1047 |
+
"eval_accuracy_recovery": 0.8947368421052632,
|
| 1048 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 1049 |
+
"eval_accuracy_unknown": 0.9285714285714286,
|
| 1050 |
+
"eval_loss": 0.19672314822673798,
|
| 1051 |
+
"eval_runtime": 33.9773,
|
| 1052 |
+
"eval_samples_per_second": 7.093,
|
| 1053 |
+
"eval_steps_per_second": 0.912,
|
| 1054 |
+
"step": 1260
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 9.092857142857143,
|
| 1058 |
+
"eval_accuracy": 0.946058091286307,
|
| 1059 |
+
"eval_accuracy_hold": 1.0,
|
| 1060 |
+
"eval_accuracy_preparation": 0.9685534591194969,
|
| 1061 |
+
"eval_accuracy_recovery": 0.8947368421052632,
|
| 1062 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 1063 |
+
"eval_accuracy_unknown": 0.9285714285714286,
|
| 1064 |
+
"eval_loss": 0.19672317802906036,
|
| 1065 |
+
"eval_runtime": 30.5504,
|
| 1066 |
+
"eval_samples_per_second": 7.889,
|
| 1067 |
+
"eval_steps_per_second": 1.015,
|
| 1068 |
+
"step": 1260
|
| 1069 |
+
}
|
| 1070 |
+
],
|
| 1071 |
+
"logging_steps": 10,
|
| 1072 |
+
"max_steps": 1260,
|
| 1073 |
+
"num_input_tokens_seen": 0,
|
| 1074 |
+
"num_train_epochs": 9223372036854775807,
|
| 1075 |
+
"save_steps": 500,
|
| 1076 |
+
"stateful_callbacks": {
|
| 1077 |
+
"TrainerControl": {
|
| 1078 |
+
"args": {
|
| 1079 |
+
"should_epoch_stop": false,
|
| 1080 |
+
"should_evaluate": false,
|
| 1081 |
+
"should_log": false,
|
| 1082 |
+
"should_save": true,
|
| 1083 |
+
"should_training_stop": true
|
| 1084 |
+
},
|
| 1085 |
+
"attributes": {}
|
| 1086 |
+
}
|
| 1087 |
+
},
|
| 1088 |
+
"total_flos": 1.2549452224921436e+19,
|
| 1089 |
+
"train_batch_size": 8,
|
| 1090 |
+
"trial_name": null,
|
| 1091 |
+
"trial_params": null
|
| 1092 |
+
}
|