End of training
Browse files- all_results.json +13 -0
- test_results.json +13 -0
- trainer_state.json +1316 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.036231884057973,
|
| 3 |
+
"eval_accuracy": 0.6870748299319728,
|
| 4 |
+
"eval_accuracy_hold": 0.0,
|
| 5 |
+
"eval_accuracy_preparation": 0.922077922077922,
|
| 6 |
+
"eval_accuracy_recovery": 0.0,
|
| 7 |
+
"eval_accuracy_stroke": 0.0,
|
| 8 |
+
"eval_accuracy_unknown": 0.8571428571428571,
|
| 9 |
+
"eval_loss": 0.9822250604629517,
|
| 10 |
+
"eval_runtime": 12.2253,
|
| 11 |
+
"eval_samples_per_second": 12.024,
|
| 12 |
+
"eval_steps_per_second": 1.554
|
| 13 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.036231884057973,
|
| 3 |
+
"eval_accuracy": 0.6870748299319728,
|
| 4 |
+
"eval_accuracy_hold": 0.0,
|
| 5 |
+
"eval_accuracy_preparation": 0.922077922077922,
|
| 6 |
+
"eval_accuracy_recovery": 0.0,
|
| 7 |
+
"eval_accuracy_stroke": 0.0,
|
| 8 |
+
"eval_accuracy_unknown": 0.8571428571428571,
|
| 9 |
+
"eval_loss": 0.9822250604629517,
|
| 10 |
+
"eval_runtime": 12.2253,
|
| 11 |
+
"eval_samples_per_second": 12.024,
|
| 12 |
+
"eval_steps_per_second": 1.554
|
| 13 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1316 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7164179104477612,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-good-gesturePhaseV5/checkpoint-490",
|
| 4 |
+
"epoch": 19.036231884057973,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1380,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.007246376811594203,
|
| 13 |
+
"grad_norm": 10.268256187438965,
|
| 14 |
+
"learning_rate": 7.246376811594204e-07,
|
| 15 |
+
"loss": 1.7861,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.014492753623188406,
|
| 20 |
+
"grad_norm": 9.569662094116211,
|
| 21 |
+
"learning_rate": 1.4492753623188408e-06,
|
| 22 |
+
"loss": 1.7539,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.021739130434782608,
|
| 27 |
+
"grad_norm": 11.259232521057129,
|
| 28 |
+
"learning_rate": 2.173913043478261e-06,
|
| 29 |
+
"loss": 1.7245,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.028985507246376812,
|
| 34 |
+
"grad_norm": 7.182712554931641,
|
| 35 |
+
"learning_rate": 2.8985507246376816e-06,
|
| 36 |
+
"loss": 1.5819,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.036231884057971016,
|
| 41 |
+
"grad_norm": 7.182448387145996,
|
| 42 |
+
"learning_rate": 3.6231884057971017e-06,
|
| 43 |
+
"loss": 1.4508,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.043478260869565216,
|
| 48 |
+
"grad_norm": 6.293506622314453,
|
| 49 |
+
"learning_rate": 4.347826086956522e-06,
|
| 50 |
+
"loss": 1.2736,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.050724637681159424,
|
| 55 |
+
"grad_norm": 6.1872944831848145,
|
| 56 |
+
"learning_rate": 5.072463768115943e-06,
|
| 57 |
+
"loss": 1.1475,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.050724637681159424,
|
| 62 |
+
"eval_accuracy": 0.5597014925373134,
|
| 63 |
+
"eval_accuracy_hold": 0.0,
|
| 64 |
+
"eval_accuracy_preparation": 1.0,
|
| 65 |
+
"eval_accuracy_recovery": 0.0,
|
| 66 |
+
"eval_accuracy_stroke": 0.0,
|
| 67 |
+
"eval_accuracy_unknown": 0.0,
|
| 68 |
+
"eval_loss": 1.2558053731918335,
|
| 69 |
+
"eval_runtime": 12.049,
|
| 70 |
+
"eval_samples_per_second": 11.121,
|
| 71 |
+
"eval_steps_per_second": 1.411,
|
| 72 |
+
"step": 70
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0072463768115942,
|
| 76 |
+
"grad_norm": 6.2471137046813965,
|
| 77 |
+
"learning_rate": 5.797101449275363e-06,
|
| 78 |
+
"loss": 1.1874,
|
| 79 |
+
"step": 80
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.0144927536231885,
|
| 83 |
+
"grad_norm": 5.99643611907959,
|
| 84 |
+
"learning_rate": 6.521739130434783e-06,
|
| 85 |
+
"loss": 1.1562,
|
| 86 |
+
"step": 90
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.0217391304347827,
|
| 90 |
+
"grad_norm": 5.748469829559326,
|
| 91 |
+
"learning_rate": 7.246376811594203e-06,
|
| 92 |
+
"loss": 1.1299,
|
| 93 |
+
"step": 100
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.0289855072463767,
|
| 97 |
+
"grad_norm": 4.690842628479004,
|
| 98 |
+
"learning_rate": 7.971014492753623e-06,
|
| 99 |
+
"loss": 1.0645,
|
| 100 |
+
"step": 110
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.036231884057971,
|
| 104 |
+
"grad_norm": 4.702245712280273,
|
| 105 |
+
"learning_rate": 8.695652173913044e-06,
|
| 106 |
+
"loss": 1.3726,
|
| 107 |
+
"step": 120
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.0434782608695652,
|
| 111 |
+
"grad_norm": 4.664677143096924,
|
| 112 |
+
"learning_rate": 9.420289855072464e-06,
|
| 113 |
+
"loss": 0.9989,
|
| 114 |
+
"step": 130
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.0507246376811594,
|
| 118 |
+
"grad_norm": 6.247674465179443,
|
| 119 |
+
"learning_rate": 9.98389694041868e-06,
|
| 120 |
+
"loss": 1.2103,
|
| 121 |
+
"step": 140
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0507246376811594,
|
| 125 |
+
"eval_accuracy": 0.5597014925373134,
|
| 126 |
+
"eval_accuracy_hold": 0.0,
|
| 127 |
+
"eval_accuracy_preparation": 1.0,
|
| 128 |
+
"eval_accuracy_recovery": 0.0,
|
| 129 |
+
"eval_accuracy_stroke": 0.0,
|
| 130 |
+
"eval_accuracy_unknown": 0.0,
|
| 131 |
+
"eval_loss": 1.2703653573989868,
|
| 132 |
+
"eval_runtime": 10.5504,
|
| 133 |
+
"eval_samples_per_second": 12.701,
|
| 134 |
+
"eval_steps_per_second": 1.611,
|
| 135 |
+
"step": 140
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 2.0072463768115942,
|
| 139 |
+
"grad_norm": 6.113107681274414,
|
| 140 |
+
"learning_rate": 9.903381642512077e-06,
|
| 141 |
+
"loss": 0.9008,
|
| 142 |
+
"step": 150
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.0144927536231885,
|
| 146 |
+
"grad_norm": 6.213434219360352,
|
| 147 |
+
"learning_rate": 9.822866344605476e-06,
|
| 148 |
+
"loss": 1.3296,
|
| 149 |
+
"step": 160
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.0217391304347827,
|
| 153 |
+
"grad_norm": 5.969342231750488,
|
| 154 |
+
"learning_rate": 9.742351046698874e-06,
|
| 155 |
+
"loss": 1.2484,
|
| 156 |
+
"step": 170
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 2.028985507246377,
|
| 160 |
+
"grad_norm": 7.0032196044921875,
|
| 161 |
+
"learning_rate": 9.66183574879227e-06,
|
| 162 |
+
"loss": 1.1239,
|
| 163 |
+
"step": 180
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.036231884057971,
|
| 167 |
+
"grad_norm": 5.675483703613281,
|
| 168 |
+
"learning_rate": 9.581320450885669e-06,
|
| 169 |
+
"loss": 1.1508,
|
| 170 |
+
"step": 190
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.0434782608695654,
|
| 174 |
+
"grad_norm": 5.311893939971924,
|
| 175 |
+
"learning_rate": 9.500805152979067e-06,
|
| 176 |
+
"loss": 1.1376,
|
| 177 |
+
"step": 200
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.050724637681159,
|
| 181 |
+
"grad_norm": 6.503353118896484,
|
| 182 |
+
"learning_rate": 9.420289855072464e-06,
|
| 183 |
+
"loss": 0.9964,
|
| 184 |
+
"step": 210
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.050724637681159,
|
| 188 |
+
"eval_accuracy": 0.5597014925373134,
|
| 189 |
+
"eval_accuracy_hold": 0.0,
|
| 190 |
+
"eval_accuracy_preparation": 1.0,
|
| 191 |
+
"eval_accuracy_recovery": 0.0,
|
| 192 |
+
"eval_accuracy_stroke": 0.0,
|
| 193 |
+
"eval_accuracy_unknown": 0.0,
|
| 194 |
+
"eval_loss": 1.2141785621643066,
|
| 195 |
+
"eval_runtime": 10.6183,
|
| 196 |
+
"eval_samples_per_second": 12.62,
|
| 197 |
+
"eval_steps_per_second": 1.601,
|
| 198 |
+
"step": 210
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 3.0072463768115942,
|
| 202 |
+
"grad_norm": 4.335814476013184,
|
| 203 |
+
"learning_rate": 9.339774557165862e-06,
|
| 204 |
+
"loss": 1.059,
|
| 205 |
+
"step": 220
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 3.0144927536231885,
|
| 209 |
+
"grad_norm": 6.591854095458984,
|
| 210 |
+
"learning_rate": 9.25925925925926e-06,
|
| 211 |
+
"loss": 1.149,
|
| 212 |
+
"step": 230
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.0217391304347827,
|
| 216 |
+
"grad_norm": 6.057372570037842,
|
| 217 |
+
"learning_rate": 9.178743961352658e-06,
|
| 218 |
+
"loss": 0.9137,
|
| 219 |
+
"step": 240
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.028985507246377,
|
| 223 |
+
"grad_norm": 18.623048782348633,
|
| 224 |
+
"learning_rate": 9.098228663446056e-06,
|
| 225 |
+
"loss": 1.1594,
|
| 226 |
+
"step": 250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 3.036231884057971,
|
| 230 |
+
"grad_norm": 8.108168601989746,
|
| 231 |
+
"learning_rate": 9.017713365539453e-06,
|
| 232 |
+
"loss": 1.1682,
|
| 233 |
+
"step": 260
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.0434782608695654,
|
| 237 |
+
"grad_norm": 8.030190467834473,
|
| 238 |
+
"learning_rate": 8.937198067632851e-06,
|
| 239 |
+
"loss": 1.2166,
|
| 240 |
+
"step": 270
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 3.050724637681159,
|
| 244 |
+
"grad_norm": 9.22143268585205,
|
| 245 |
+
"learning_rate": 8.85668276972625e-06,
|
| 246 |
+
"loss": 0.9975,
|
| 247 |
+
"step": 280
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 3.050724637681159,
|
| 251 |
+
"eval_accuracy": 0.5970149253731343,
|
| 252 |
+
"eval_accuracy_hold": 0.0,
|
| 253 |
+
"eval_accuracy_preparation": 0.9733333333333334,
|
| 254 |
+
"eval_accuracy_recovery": 0.0,
|
| 255 |
+
"eval_accuracy_stroke": 0.0,
|
| 256 |
+
"eval_accuracy_unknown": 0.2692307692307692,
|
| 257 |
+
"eval_loss": 1.0746699571609497,
|
| 258 |
+
"eval_runtime": 10.8059,
|
| 259 |
+
"eval_samples_per_second": 12.401,
|
| 260 |
+
"eval_steps_per_second": 1.573,
|
| 261 |
+
"step": 280
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 4.007246376811594,
|
| 265 |
+
"grad_norm": 9.18785285949707,
|
| 266 |
+
"learning_rate": 8.776167471819646e-06,
|
| 267 |
+
"loss": 1.1711,
|
| 268 |
+
"step": 290
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 4.0144927536231885,
|
| 272 |
+
"grad_norm": 13.558119773864746,
|
| 273 |
+
"learning_rate": 8.695652173913044e-06,
|
| 274 |
+
"loss": 0.8541,
|
| 275 |
+
"step": 300
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 4.021739130434782,
|
| 279 |
+
"grad_norm": 25.042736053466797,
|
| 280 |
+
"learning_rate": 8.615136876006443e-06,
|
| 281 |
+
"loss": 0.9376,
|
| 282 |
+
"step": 310
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 4.028985507246377,
|
| 286 |
+
"grad_norm": 11.34363079071045,
|
| 287 |
+
"learning_rate": 8.53462157809984e-06,
|
| 288 |
+
"loss": 1.0186,
|
| 289 |
+
"step": 320
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 4.036231884057971,
|
| 293 |
+
"grad_norm": 10.07709789276123,
|
| 294 |
+
"learning_rate": 8.454106280193238e-06,
|
| 295 |
+
"loss": 1.0577,
|
| 296 |
+
"step": 330
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 4.043478260869565,
|
| 300 |
+
"grad_norm": 10.198676109313965,
|
| 301 |
+
"learning_rate": 8.373590982286636e-06,
|
| 302 |
+
"loss": 1.0924,
|
| 303 |
+
"step": 340
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 4.050724637681159,
|
| 307 |
+
"grad_norm": 19.099363327026367,
|
| 308 |
+
"learning_rate": 8.293075684380033e-06,
|
| 309 |
+
"loss": 1.0538,
|
| 310 |
+
"step": 350
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 4.050724637681159,
|
| 314 |
+
"eval_accuracy": 0.664179104477612,
|
| 315 |
+
"eval_accuracy_hold": 0.0,
|
| 316 |
+
"eval_accuracy_preparation": 0.88,
|
| 317 |
+
"eval_accuracy_recovery": 0.0,
|
| 318 |
+
"eval_accuracy_stroke": 0.0,
|
| 319 |
+
"eval_accuracy_unknown": 0.8846153846153846,
|
| 320 |
+
"eval_loss": 0.9621986150741577,
|
| 321 |
+
"eval_runtime": 10.8298,
|
| 322 |
+
"eval_samples_per_second": 12.373,
|
| 323 |
+
"eval_steps_per_second": 1.57,
|
| 324 |
+
"step": 350
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 5.007246376811594,
|
| 328 |
+
"grad_norm": 4.384646892547607,
|
| 329 |
+
"learning_rate": 8.212560386473431e-06,
|
| 330 |
+
"loss": 0.9923,
|
| 331 |
+
"step": 360
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 5.0144927536231885,
|
| 335 |
+
"grad_norm": 16.15943717956543,
|
| 336 |
+
"learning_rate": 8.132045088566828e-06,
|
| 337 |
+
"loss": 0.937,
|
| 338 |
+
"step": 370
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.021739130434782,
|
| 342 |
+
"grad_norm": 23.731212615966797,
|
| 343 |
+
"learning_rate": 8.051529790660226e-06,
|
| 344 |
+
"loss": 1.0034,
|
| 345 |
+
"step": 380
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 5.028985507246377,
|
| 349 |
+
"grad_norm": 9.929147720336914,
|
| 350 |
+
"learning_rate": 7.971014492753623e-06,
|
| 351 |
+
"loss": 0.9414,
|
| 352 |
+
"step": 390
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 5.036231884057971,
|
| 356 |
+
"grad_norm": 7.188724517822266,
|
| 357 |
+
"learning_rate": 7.890499194847021e-06,
|
| 358 |
+
"loss": 0.9633,
|
| 359 |
+
"step": 400
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.043478260869565,
|
| 363 |
+
"grad_norm": 7.971716403961182,
|
| 364 |
+
"learning_rate": 7.80998389694042e-06,
|
| 365 |
+
"loss": 0.9313,
|
| 366 |
+
"step": 410
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 5.050724637681159,
|
| 370 |
+
"grad_norm": 23.818294525146484,
|
| 371 |
+
"learning_rate": 7.729468599033817e-06,
|
| 372 |
+
"loss": 1.0321,
|
| 373 |
+
"step": 420
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 5.050724637681159,
|
| 377 |
+
"eval_accuracy": 0.6567164179104478,
|
| 378 |
+
"eval_accuracy_hold": 0.0,
|
| 379 |
+
"eval_accuracy_preparation": 0.8533333333333334,
|
| 380 |
+
"eval_accuracy_recovery": 0.0,
|
| 381 |
+
"eval_accuracy_stroke": 0.0,
|
| 382 |
+
"eval_accuracy_unknown": 0.9230769230769231,
|
| 383 |
+
"eval_loss": 0.9450957775115967,
|
| 384 |
+
"eval_runtime": 10.5545,
|
| 385 |
+
"eval_samples_per_second": 12.696,
|
| 386 |
+
"eval_steps_per_second": 1.611,
|
| 387 |
+
"step": 420
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 6.007246376811594,
|
| 391 |
+
"grad_norm": 6.358923435211182,
|
| 392 |
+
"learning_rate": 7.648953301127215e-06,
|
| 393 |
+
"loss": 0.7638,
|
| 394 |
+
"step": 430
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 6.0144927536231885,
|
| 398 |
+
"grad_norm": 7.487329483032227,
|
| 399 |
+
"learning_rate": 7.568438003220613e-06,
|
| 400 |
+
"loss": 0.8216,
|
| 401 |
+
"step": 440
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 6.021739130434782,
|
| 405 |
+
"grad_norm": 8.15117359161377,
|
| 406 |
+
"learning_rate": 7.48792270531401e-06,
|
| 407 |
+
"loss": 1.042,
|
| 408 |
+
"step": 450
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 6.028985507246377,
|
| 412 |
+
"grad_norm": 8.114563941955566,
|
| 413 |
+
"learning_rate": 7.4074074074074075e-06,
|
| 414 |
+
"loss": 0.8528,
|
| 415 |
+
"step": 460
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 6.036231884057971,
|
| 419 |
+
"grad_norm": 10.497995376586914,
|
| 420 |
+
"learning_rate": 7.326892109500806e-06,
|
| 421 |
+
"loss": 1.0731,
|
| 422 |
+
"step": 470
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 6.043478260869565,
|
| 426 |
+
"grad_norm": 7.0596818923950195,
|
| 427 |
+
"learning_rate": 7.246376811594203e-06,
|
| 428 |
+
"loss": 0.89,
|
| 429 |
+
"step": 480
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 6.050724637681159,
|
| 433 |
+
"grad_norm": 7.714406490325928,
|
| 434 |
+
"learning_rate": 7.165861513687601e-06,
|
| 435 |
+
"loss": 0.7822,
|
| 436 |
+
"step": 490
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 6.050724637681159,
|
| 440 |
+
"eval_accuracy": 0.7164179104477612,
|
| 441 |
+
"eval_accuracy_hold": 0.0,
|
| 442 |
+
"eval_accuracy_preparation": 0.96,
|
| 443 |
+
"eval_accuracy_recovery": 0.08333333333333333,
|
| 444 |
+
"eval_accuracy_stroke": 0.0,
|
| 445 |
+
"eval_accuracy_unknown": 0.8846153846153846,
|
| 446 |
+
"eval_loss": 0.8797218203544617,
|
| 447 |
+
"eval_runtime": 10.7094,
|
| 448 |
+
"eval_samples_per_second": 12.512,
|
| 449 |
+
"eval_steps_per_second": 1.587,
|
| 450 |
+
"step": 490
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 7.007246376811594,
|
| 454 |
+
"grad_norm": 6.505290508270264,
|
| 455 |
+
"learning_rate": 7.085346215780999e-06,
|
| 456 |
+
"loss": 0.7396,
|
| 457 |
+
"step": 500
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 7.0144927536231885,
|
| 461 |
+
"grad_norm": 7.676311492919922,
|
| 462 |
+
"learning_rate": 7.004830917874397e-06,
|
| 463 |
+
"loss": 1.0383,
|
| 464 |
+
"step": 510
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 7.021739130434782,
|
| 468 |
+
"grad_norm": 8.093998908996582,
|
| 469 |
+
"learning_rate": 6.924315619967794e-06,
|
| 470 |
+
"loss": 0.841,
|
| 471 |
+
"step": 520
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 7.028985507246377,
|
| 475 |
+
"grad_norm": 4.555324077606201,
|
| 476 |
+
"learning_rate": 6.843800322061193e-06,
|
| 477 |
+
"loss": 0.6253,
|
| 478 |
+
"step": 530
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 7.036231884057971,
|
| 482 |
+
"grad_norm": 6.43816614151001,
|
| 483 |
+
"learning_rate": 6.76328502415459e-06,
|
| 484 |
+
"loss": 0.9515,
|
| 485 |
+
"step": 540
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 7.043478260869565,
|
| 489 |
+
"grad_norm": 10.472518920898438,
|
| 490 |
+
"learning_rate": 6.682769726247987e-06,
|
| 491 |
+
"loss": 1.1703,
|
| 492 |
+
"step": 550
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 7.050724637681159,
|
| 496 |
+
"grad_norm": 12.832620620727539,
|
| 497 |
+
"learning_rate": 6.602254428341386e-06,
|
| 498 |
+
"loss": 0.8743,
|
| 499 |
+
"step": 560
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 7.050724637681159,
|
| 503 |
+
"eval_accuracy": 0.6791044776119403,
|
| 504 |
+
"eval_accuracy_hold": 0.0,
|
| 505 |
+
"eval_accuracy_preparation": 0.8533333333333334,
|
| 506 |
+
"eval_accuracy_recovery": 0.08333333333333333,
|
| 507 |
+
"eval_accuracy_stroke": 0.0,
|
| 508 |
+
"eval_accuracy_unknown": 1.0,
|
| 509 |
+
"eval_loss": 0.9399013519287109,
|
| 510 |
+
"eval_runtime": 10.6086,
|
| 511 |
+
"eval_samples_per_second": 12.631,
|
| 512 |
+
"eval_steps_per_second": 1.602,
|
| 513 |
+
"step": 560
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 8.007246376811594,
|
| 517 |
+
"grad_norm": 6.019155502319336,
|
| 518 |
+
"learning_rate": 6.521739130434783e-06,
|
| 519 |
+
"loss": 0.8912,
|
| 520 |
+
"step": 570
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 8.014492753623188,
|
| 524 |
+
"grad_norm": 5.795660018920898,
|
| 525 |
+
"learning_rate": 6.44122383252818e-06,
|
| 526 |
+
"loss": 0.8103,
|
| 527 |
+
"step": 580
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 8.021739130434783,
|
| 531 |
+
"grad_norm": 22.207584381103516,
|
| 532 |
+
"learning_rate": 6.360708534621579e-06,
|
| 533 |
+
"loss": 0.9291,
|
| 534 |
+
"step": 590
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 8.028985507246377,
|
| 538 |
+
"grad_norm": 8.41496467590332,
|
| 539 |
+
"learning_rate": 6.280193236714976e-06,
|
| 540 |
+
"loss": 0.8249,
|
| 541 |
+
"step": 600
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 8.03623188405797,
|
| 545 |
+
"grad_norm": 19.649620056152344,
|
| 546 |
+
"learning_rate": 6.199677938808374e-06,
|
| 547 |
+
"loss": 0.9209,
|
| 548 |
+
"step": 610
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 8.043478260869565,
|
| 552 |
+
"grad_norm": 33.21004867553711,
|
| 553 |
+
"learning_rate": 6.119162640901772e-06,
|
| 554 |
+
"loss": 0.7371,
|
| 555 |
+
"step": 620
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 8.05072463768116,
|
| 559 |
+
"grad_norm": 10.654520988464355,
|
| 560 |
+
"learning_rate": 6.03864734299517e-06,
|
| 561 |
+
"loss": 0.7515,
|
| 562 |
+
"step": 630
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 8.05072463768116,
|
| 566 |
+
"eval_accuracy": 0.6791044776119403,
|
| 567 |
+
"eval_accuracy_hold": 0.0,
|
| 568 |
+
"eval_accuracy_preparation": 0.8666666666666667,
|
| 569 |
+
"eval_accuracy_recovery": 0.0,
|
| 570 |
+
"eval_accuracy_stroke": 0.0,
|
| 571 |
+
"eval_accuracy_unknown": 1.0,
|
| 572 |
+
"eval_loss": 0.928985059261322,
|
| 573 |
+
"eval_runtime": 11.5822,
|
| 574 |
+
"eval_samples_per_second": 11.57,
|
| 575 |
+
"eval_steps_per_second": 1.468,
|
| 576 |
+
"step": 630
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 9.007246376811594,
|
| 580 |
+
"grad_norm": 9.228910446166992,
|
| 581 |
+
"learning_rate": 5.958132045088567e-06,
|
| 582 |
+
"loss": 0.8266,
|
| 583 |
+
"step": 640
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 9.014492753623188,
|
| 587 |
+
"grad_norm": 6.276782035827637,
|
| 588 |
+
"learning_rate": 5.877616747181965e-06,
|
| 589 |
+
"loss": 0.611,
|
| 590 |
+
"step": 650
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 9.021739130434783,
|
| 594 |
+
"grad_norm": 27.36185646057129,
|
| 595 |
+
"learning_rate": 5.797101449275363e-06,
|
| 596 |
+
"loss": 0.8595,
|
| 597 |
+
"step": 660
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 9.028985507246377,
|
| 601 |
+
"grad_norm": 13.051177024841309,
|
| 602 |
+
"learning_rate": 5.716586151368761e-06,
|
| 603 |
+
"loss": 1.0073,
|
| 604 |
+
"step": 670
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 9.03623188405797,
|
| 608 |
+
"grad_norm": 12.05443000793457,
|
| 609 |
+
"learning_rate": 5.6360708534621574e-06,
|
| 610 |
+
"loss": 0.6386,
|
| 611 |
+
"step": 680
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 9.043478260869565,
|
| 615 |
+
"grad_norm": 9.477375984191895,
|
| 616 |
+
"learning_rate": 5.555555555555557e-06,
|
| 617 |
+
"loss": 1.1109,
|
| 618 |
+
"step": 690
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 9.05072463768116,
|
| 622 |
+
"grad_norm": 16.6785945892334,
|
| 623 |
+
"learning_rate": 5.475040257648953e-06,
|
| 624 |
+
"loss": 0.8525,
|
| 625 |
+
"step": 700
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 9.05072463768116,
|
| 629 |
+
"eval_accuracy": 0.7089552238805971,
|
| 630 |
+
"eval_accuracy_hold": 0.0,
|
| 631 |
+
"eval_accuracy_preparation": 0.9466666666666667,
|
| 632 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 633 |
+
"eval_accuracy_stroke": 0.0,
|
| 634 |
+
"eval_accuracy_unknown": 0.8461538461538461,
|
| 635 |
+
"eval_loss": 0.8446803092956543,
|
| 636 |
+
"eval_runtime": 15.7965,
|
| 637 |
+
"eval_samples_per_second": 8.483,
|
| 638 |
+
"eval_steps_per_second": 1.076,
|
| 639 |
+
"step": 700
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 10.007246376811594,
|
| 643 |
+
"grad_norm": 12.451231002807617,
|
| 644 |
+
"learning_rate": 5.394524959742351e-06,
|
| 645 |
+
"loss": 0.7396,
|
| 646 |
+
"step": 710
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 10.014492753623188,
|
| 650 |
+
"grad_norm": 6.226474761962891,
|
| 651 |
+
"learning_rate": 5.314009661835749e-06,
|
| 652 |
+
"loss": 0.7467,
|
| 653 |
+
"step": 720
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 10.021739130434783,
|
| 657 |
+
"grad_norm": 9.991294860839844,
|
| 658 |
+
"learning_rate": 5.233494363929147e-06,
|
| 659 |
+
"loss": 0.8203,
|
| 660 |
+
"step": 730
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 10.028985507246377,
|
| 664 |
+
"grad_norm": 18.472450256347656,
|
| 665 |
+
"learning_rate": 5.152979066022544e-06,
|
| 666 |
+
"loss": 0.766,
|
| 667 |
+
"step": 740
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 10.03623188405797,
|
| 671 |
+
"grad_norm": 8.452324867248535,
|
| 672 |
+
"learning_rate": 5.072463768115943e-06,
|
| 673 |
+
"loss": 0.8539,
|
| 674 |
+
"step": 750
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 10.043478260869565,
|
| 678 |
+
"grad_norm": 18.313405990600586,
|
| 679 |
+
"learning_rate": 4.99194847020934e-06,
|
| 680 |
+
"loss": 0.7443,
|
| 681 |
+
"step": 760
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 10.05072463768116,
|
| 685 |
+
"grad_norm": 18.442798614501953,
|
| 686 |
+
"learning_rate": 4.911433172302738e-06,
|
| 687 |
+
"loss": 0.7661,
|
| 688 |
+
"step": 770
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 10.05072463768116,
|
| 692 |
+
"eval_accuracy": 0.7089552238805971,
|
| 693 |
+
"eval_accuracy_hold": 0.0,
|
| 694 |
+
"eval_accuracy_preparation": 0.9066666666666666,
|
| 695 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 696 |
+
"eval_accuracy_stroke": 0.0,
|
| 697 |
+
"eval_accuracy_unknown": 0.9615384615384616,
|
| 698 |
+
"eval_loss": 0.7856659293174744,
|
| 699 |
+
"eval_runtime": 10.3807,
|
| 700 |
+
"eval_samples_per_second": 12.909,
|
| 701 |
+
"eval_steps_per_second": 1.638,
|
| 702 |
+
"step": 770
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 11.007246376811594,
|
| 706 |
+
"grad_norm": 8.811756134033203,
|
| 707 |
+
"learning_rate": 4.830917874396135e-06,
|
| 708 |
+
"loss": 0.7784,
|
| 709 |
+
"step": 780
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 11.014492753623188,
|
| 713 |
+
"grad_norm": 8.056985855102539,
|
| 714 |
+
"learning_rate": 4.750402576489534e-06,
|
| 715 |
+
"loss": 0.7112,
|
| 716 |
+
"step": 790
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 11.021739130434783,
|
| 720 |
+
"grad_norm": 5.6117939949035645,
|
| 721 |
+
"learning_rate": 4.669887278582931e-06,
|
| 722 |
+
"loss": 0.6901,
|
| 723 |
+
"step": 800
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 11.028985507246377,
|
| 727 |
+
"grad_norm": 8.331842422485352,
|
| 728 |
+
"learning_rate": 4.589371980676329e-06,
|
| 729 |
+
"loss": 0.5963,
|
| 730 |
+
"step": 810
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 11.03623188405797,
|
| 734 |
+
"grad_norm": 7.48464822769165,
|
| 735 |
+
"learning_rate": 4.508856682769726e-06,
|
| 736 |
+
"loss": 0.9118,
|
| 737 |
+
"step": 820
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 11.043478260869565,
|
| 741 |
+
"grad_norm": 14.966755867004395,
|
| 742 |
+
"learning_rate": 4.428341384863125e-06,
|
| 743 |
+
"loss": 0.8462,
|
| 744 |
+
"step": 830
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 11.05072463768116,
|
| 748 |
+
"grad_norm": 13.086145401000977,
|
| 749 |
+
"learning_rate": 4.347826086956522e-06,
|
| 750 |
+
"loss": 0.8363,
|
| 751 |
+
"step": 840
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 11.05072463768116,
|
| 755 |
+
"eval_accuracy": 0.6865671641791045,
|
| 756 |
+
"eval_accuracy_hold": 0.0,
|
| 757 |
+
"eval_accuracy_preparation": 0.92,
|
| 758 |
+
"eval_accuracy_recovery": 0.08333333333333333,
|
| 759 |
+
"eval_accuracy_stroke": 0.0,
|
| 760 |
+
"eval_accuracy_unknown": 0.8461538461538461,
|
| 761 |
+
"eval_loss": 0.8165404200553894,
|
| 762 |
+
"eval_runtime": 10.6177,
|
| 763 |
+
"eval_samples_per_second": 12.62,
|
| 764 |
+
"eval_steps_per_second": 1.601,
|
| 765 |
+
"step": 840
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 12.007246376811594,
|
| 769 |
+
"grad_norm": 13.03532886505127,
|
| 770 |
+
"learning_rate": 4.26731078904992e-06,
|
| 771 |
+
"loss": 0.8312,
|
| 772 |
+
"step": 850
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 12.014492753623188,
|
| 776 |
+
"grad_norm": 10.994771957397461,
|
| 777 |
+
"learning_rate": 4.186795491143318e-06,
|
| 778 |
+
"loss": 0.7242,
|
| 779 |
+
"step": 860
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 12.021739130434783,
|
| 783 |
+
"grad_norm": 37.28443908691406,
|
| 784 |
+
"learning_rate": 4.106280193236716e-06,
|
| 785 |
+
"loss": 0.8542,
|
| 786 |
+
"step": 870
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 12.028985507246377,
|
| 790 |
+
"grad_norm": 15.367025375366211,
|
| 791 |
+
"learning_rate": 4.025764895330113e-06,
|
| 792 |
+
"loss": 0.7253,
|
| 793 |
+
"step": 880
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 12.03623188405797,
|
| 797 |
+
"grad_norm": 18.963815689086914,
|
| 798 |
+
"learning_rate": 3.945249597423511e-06,
|
| 799 |
+
"loss": 0.6786,
|
| 800 |
+
"step": 890
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 12.043478260869565,
|
| 804 |
+
"grad_norm": 6.912456512451172,
|
| 805 |
+
"learning_rate": 3.864734299516908e-06,
|
| 806 |
+
"loss": 0.6771,
|
| 807 |
+
"step": 900
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 12.05072463768116,
|
| 811 |
+
"grad_norm": 11.315908432006836,
|
| 812 |
+
"learning_rate": 3.7842190016103066e-06,
|
| 813 |
+
"loss": 0.659,
|
| 814 |
+
"step": 910
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 12.05072463768116,
|
| 818 |
+
"eval_accuracy": 0.7164179104477612,
|
| 819 |
+
"eval_accuracy_hold": 0.0,
|
| 820 |
+
"eval_accuracy_preparation": 0.9066666666666666,
|
| 821 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 822 |
+
"eval_accuracy_stroke": 0.0,
|
| 823 |
+
"eval_accuracy_unknown": 1.0,
|
| 824 |
+
"eval_loss": 0.7950567007064819,
|
| 825 |
+
"eval_runtime": 10.5,
|
| 826 |
+
"eval_samples_per_second": 12.762,
|
| 827 |
+
"eval_steps_per_second": 1.619,
|
| 828 |
+
"step": 910
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 13.007246376811594,
|
| 832 |
+
"grad_norm": 14.643826484680176,
|
| 833 |
+
"learning_rate": 3.7037037037037037e-06,
|
| 834 |
+
"loss": 0.7824,
|
| 835 |
+
"step": 920
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 13.014492753623188,
|
| 839 |
+
"grad_norm": 10.328850746154785,
|
| 840 |
+
"learning_rate": 3.6231884057971017e-06,
|
| 841 |
+
"loss": 0.7485,
|
| 842 |
+
"step": 930
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 13.021739130434783,
|
| 846 |
+
"grad_norm": 13.339781761169434,
|
| 847 |
+
"learning_rate": 3.5426731078904997e-06,
|
| 848 |
+
"loss": 0.6207,
|
| 849 |
+
"step": 940
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 13.028985507246377,
|
| 853 |
+
"grad_norm": 15.779186248779297,
|
| 854 |
+
"learning_rate": 3.462157809983897e-06,
|
| 855 |
+
"loss": 0.7358,
|
| 856 |
+
"step": 950
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 13.03623188405797,
|
| 860 |
+
"grad_norm": 9.903864860534668,
|
| 861 |
+
"learning_rate": 3.381642512077295e-06,
|
| 862 |
+
"loss": 0.7412,
|
| 863 |
+
"step": 960
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 13.043478260869565,
|
| 867 |
+
"grad_norm": 19.546676635742188,
|
| 868 |
+
"learning_rate": 3.301127214170693e-06,
|
| 869 |
+
"loss": 0.6234,
|
| 870 |
+
"step": 970
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 13.05072463768116,
|
| 874 |
+
"grad_norm": 11.30717945098877,
|
| 875 |
+
"learning_rate": 3.22061191626409e-06,
|
| 876 |
+
"loss": 0.6274,
|
| 877 |
+
"step": 980
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 13.05072463768116,
|
| 881 |
+
"eval_accuracy": 0.7014925373134329,
|
| 882 |
+
"eval_accuracy_hold": 0.0,
|
| 883 |
+
"eval_accuracy_preparation": 0.8933333333333333,
|
| 884 |
+
"eval_accuracy_recovery": 0.08333333333333333,
|
| 885 |
+
"eval_accuracy_stroke": 0.0,
|
| 886 |
+
"eval_accuracy_unknown": 1.0,
|
| 887 |
+
"eval_loss": 0.7753632068634033,
|
| 888 |
+
"eval_runtime": 10.5138,
|
| 889 |
+
"eval_samples_per_second": 12.745,
|
| 890 |
+
"eval_steps_per_second": 1.617,
|
| 891 |
+
"step": 980
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 14.007246376811594,
|
| 895 |
+
"grad_norm": 14.222391128540039,
|
| 896 |
+
"learning_rate": 3.140096618357488e-06,
|
| 897 |
+
"loss": 0.6564,
|
| 898 |
+
"step": 990
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 14.014492753623188,
|
| 902 |
+
"grad_norm": 9.9340238571167,
|
| 903 |
+
"learning_rate": 3.059581320450886e-06,
|
| 904 |
+
"loss": 0.6258,
|
| 905 |
+
"step": 1000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 14.021739130434783,
|
| 909 |
+
"grad_norm": 21.6451416015625,
|
| 910 |
+
"learning_rate": 2.9790660225442837e-06,
|
| 911 |
+
"loss": 0.7534,
|
| 912 |
+
"step": 1010
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 14.028985507246377,
|
| 916 |
+
"grad_norm": 5.2904462814331055,
|
| 917 |
+
"learning_rate": 2.8985507246376816e-06,
|
| 918 |
+
"loss": 0.6742,
|
| 919 |
+
"step": 1020
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 14.03623188405797,
|
| 923 |
+
"grad_norm": 12.198628425598145,
|
| 924 |
+
"learning_rate": 2.8180354267310787e-06,
|
| 925 |
+
"loss": 0.6938,
|
| 926 |
+
"step": 1030
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 14.043478260869565,
|
| 930 |
+
"grad_norm": 16.807910919189453,
|
| 931 |
+
"learning_rate": 2.7375201288244767e-06,
|
| 932 |
+
"loss": 0.6308,
|
| 933 |
+
"step": 1040
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 14.05072463768116,
|
| 937 |
+
"grad_norm": 29.11658477783203,
|
| 938 |
+
"learning_rate": 2.6570048309178746e-06,
|
| 939 |
+
"loss": 0.7292,
|
| 940 |
+
"step": 1050
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 14.05072463768116,
|
| 944 |
+
"eval_accuracy": 0.6791044776119403,
|
| 945 |
+
"eval_accuracy_hold": 0.0,
|
| 946 |
+
"eval_accuracy_preparation": 0.8266666666666667,
|
| 947 |
+
"eval_accuracy_recovery": 0.25,
|
| 948 |
+
"eval_accuracy_stroke": 0.0,
|
| 949 |
+
"eval_accuracy_unknown": 1.0,
|
| 950 |
+
"eval_loss": 0.8128386735916138,
|
| 951 |
+
"eval_runtime": 10.6524,
|
| 952 |
+
"eval_samples_per_second": 12.579,
|
| 953 |
+
"eval_steps_per_second": 1.596,
|
| 954 |
+
"step": 1050
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 15.007246376811594,
|
| 958 |
+
"grad_norm": 19.81916046142578,
|
| 959 |
+
"learning_rate": 2.576489533011272e-06,
|
| 960 |
+
"loss": 0.6481,
|
| 961 |
+
"step": 1060
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 15.014492753623188,
|
| 965 |
+
"grad_norm": 23.879323959350586,
|
| 966 |
+
"learning_rate": 2.49597423510467e-06,
|
| 967 |
+
"loss": 0.6273,
|
| 968 |
+
"step": 1070
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 15.021739130434783,
|
| 972 |
+
"grad_norm": 26.635896682739258,
|
| 973 |
+
"learning_rate": 2.4154589371980677e-06,
|
| 974 |
+
"loss": 0.629,
|
| 975 |
+
"step": 1080
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 15.028985507246377,
|
| 979 |
+
"grad_norm": 15.82916259765625,
|
| 980 |
+
"learning_rate": 2.3349436392914656e-06,
|
| 981 |
+
"loss": 0.6142,
|
| 982 |
+
"step": 1090
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 15.03623188405797,
|
| 986 |
+
"grad_norm": 29.63594627380371,
|
| 987 |
+
"learning_rate": 2.254428341384863e-06,
|
| 988 |
+
"loss": 0.6727,
|
| 989 |
+
"step": 1100
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 15.043478260869565,
|
| 993 |
+
"grad_norm": 15.376564979553223,
|
| 994 |
+
"learning_rate": 2.173913043478261e-06,
|
| 995 |
+
"loss": 0.6719,
|
| 996 |
+
"step": 1110
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 15.05072463768116,
|
| 1000 |
+
"grad_norm": 19.15326499938965,
|
| 1001 |
+
"learning_rate": 2.093397745571659e-06,
|
| 1002 |
+
"loss": 0.7447,
|
| 1003 |
+
"step": 1120
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 15.05072463768116,
|
| 1007 |
+
"eval_accuracy": 0.6865671641791045,
|
| 1008 |
+
"eval_accuracy_hold": 0.0,
|
| 1009 |
+
"eval_accuracy_preparation": 0.84,
|
| 1010 |
+
"eval_accuracy_recovery": 0.25,
|
| 1011 |
+
"eval_accuracy_stroke": 0.0,
|
| 1012 |
+
"eval_accuracy_unknown": 1.0,
|
| 1013 |
+
"eval_loss": 0.7860263586044312,
|
| 1014 |
+
"eval_runtime": 10.5021,
|
| 1015 |
+
"eval_samples_per_second": 12.759,
|
| 1016 |
+
"eval_steps_per_second": 1.619,
|
| 1017 |
+
"step": 1120
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 16.007246376811594,
|
| 1021 |
+
"grad_norm": 12.135031700134277,
|
| 1022 |
+
"learning_rate": 2.0128824476650566e-06,
|
| 1023 |
+
"loss": 0.6189,
|
| 1024 |
+
"step": 1130
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 16.014492753623188,
|
| 1028 |
+
"grad_norm": 13.124465942382812,
|
| 1029 |
+
"learning_rate": 1.932367149758454e-06,
|
| 1030 |
+
"loss": 0.5972,
|
| 1031 |
+
"step": 1140
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 16.02173913043478,
|
| 1035 |
+
"grad_norm": 15.266533851623535,
|
| 1036 |
+
"learning_rate": 1.8518518518518519e-06,
|
| 1037 |
+
"loss": 0.8582,
|
| 1038 |
+
"step": 1150
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 16.028985507246375,
|
| 1042 |
+
"grad_norm": 15.358236312866211,
|
| 1043 |
+
"learning_rate": 1.7713365539452498e-06,
|
| 1044 |
+
"loss": 0.5844,
|
| 1045 |
+
"step": 1160
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 16.036231884057973,
|
| 1049 |
+
"grad_norm": 28.783262252807617,
|
| 1050 |
+
"learning_rate": 1.6908212560386476e-06,
|
| 1051 |
+
"loss": 0.6861,
|
| 1052 |
+
"step": 1170
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 16.043478260869566,
|
| 1056 |
+
"grad_norm": 26.525894165039062,
|
| 1057 |
+
"learning_rate": 1.610305958132045e-06,
|
| 1058 |
+
"loss": 0.5341,
|
| 1059 |
+
"step": 1180
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 16.05072463768116,
|
| 1063 |
+
"grad_norm": 20.598716735839844,
|
| 1064 |
+
"learning_rate": 1.529790660225443e-06,
|
| 1065 |
+
"loss": 0.5512,
|
| 1066 |
+
"step": 1190
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 16.05072463768116,
|
| 1070 |
+
"eval_accuracy": 0.7014925373134329,
|
| 1071 |
+
"eval_accuracy_hold": 0.0625,
|
| 1072 |
+
"eval_accuracy_preparation": 0.8666666666666667,
|
| 1073 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 1074 |
+
"eval_accuracy_stroke": 0.0,
|
| 1075 |
+
"eval_accuracy_unknown": 1.0,
|
| 1076 |
+
"eval_loss": 0.7839486598968506,
|
| 1077 |
+
"eval_runtime": 10.6658,
|
| 1078 |
+
"eval_samples_per_second": 12.564,
|
| 1079 |
+
"eval_steps_per_second": 1.594,
|
| 1080 |
+
"step": 1190
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 17.007246376811594,
|
| 1084 |
+
"grad_norm": 15.926338195800781,
|
| 1085 |
+
"learning_rate": 1.4492753623188408e-06,
|
| 1086 |
+
"loss": 0.6122,
|
| 1087 |
+
"step": 1200
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 17.014492753623188,
|
| 1091 |
+
"grad_norm": 9.888009071350098,
|
| 1092 |
+
"learning_rate": 1.3687600644122383e-06,
|
| 1093 |
+
"loss": 0.6741,
|
| 1094 |
+
"step": 1210
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 17.02173913043478,
|
| 1098 |
+
"grad_norm": 7.94158935546875,
|
| 1099 |
+
"learning_rate": 1.288244766505636e-06,
|
| 1100 |
+
"loss": 0.5769,
|
| 1101 |
+
"step": 1220
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 17.028985507246375,
|
| 1105 |
+
"grad_norm": 12.030184745788574,
|
| 1106 |
+
"learning_rate": 1.2077294685990338e-06,
|
| 1107 |
+
"loss": 0.7706,
|
| 1108 |
+
"step": 1230
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 17.036231884057973,
|
| 1112 |
+
"grad_norm": 12.405618667602539,
|
| 1113 |
+
"learning_rate": 1.1272141706924316e-06,
|
| 1114 |
+
"loss": 0.7073,
|
| 1115 |
+
"step": 1240
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 17.043478260869566,
|
| 1119 |
+
"grad_norm": 18.144861221313477,
|
| 1120 |
+
"learning_rate": 1.0466988727858295e-06,
|
| 1121 |
+
"loss": 0.5714,
|
| 1122 |
+
"step": 1250
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 17.05072463768116,
|
| 1126 |
+
"grad_norm": 51.77016067504883,
|
| 1127 |
+
"learning_rate": 9.66183574879227e-07,
|
| 1128 |
+
"loss": 0.3404,
|
| 1129 |
+
"step": 1260
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 17.05072463768116,
|
| 1133 |
+
"eval_accuracy": 0.7014925373134329,
|
| 1134 |
+
"eval_accuracy_hold": 0.0,
|
| 1135 |
+
"eval_accuracy_preparation": 0.8533333333333334,
|
| 1136 |
+
"eval_accuracy_recovery": 0.3333333333333333,
|
| 1137 |
+
"eval_accuracy_stroke": 0.0,
|
| 1138 |
+
"eval_accuracy_unknown": 1.0,
|
| 1139 |
+
"eval_loss": 0.8054906725883484,
|
| 1140 |
+
"eval_runtime": 10.759,
|
| 1141 |
+
"eval_samples_per_second": 12.455,
|
| 1142 |
+
"eval_steps_per_second": 1.58,
|
| 1143 |
+
"step": 1260
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 18.007246376811594,
|
| 1147 |
+
"grad_norm": 22.187854766845703,
|
| 1148 |
+
"learning_rate": 8.856682769726249e-07,
|
| 1149 |
+
"loss": 0.6113,
|
| 1150 |
+
"step": 1270
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 18.014492753623188,
|
| 1154 |
+
"grad_norm": 16.749889373779297,
|
| 1155 |
+
"learning_rate": 8.051529790660226e-07,
|
| 1156 |
+
"loss": 0.6058,
|
| 1157 |
+
"step": 1280
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 18.02173913043478,
|
| 1161 |
+
"grad_norm": 16.847583770751953,
|
| 1162 |
+
"learning_rate": 7.246376811594204e-07,
|
| 1163 |
+
"loss": 0.7215,
|
| 1164 |
+
"step": 1290
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 18.028985507246375,
|
| 1168 |
+
"grad_norm": 71.80619049072266,
|
| 1169 |
+
"learning_rate": 6.44122383252818e-07,
|
| 1170 |
+
"loss": 0.483,
|
| 1171 |
+
"step": 1300
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 18.036231884057973,
|
| 1175 |
+
"grad_norm": 13.316573143005371,
|
| 1176 |
+
"learning_rate": 5.636070853462158e-07,
|
| 1177 |
+
"loss": 0.585,
|
| 1178 |
+
"step": 1310
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 18.043478260869566,
|
| 1182 |
+
"grad_norm": 22.565717697143555,
|
| 1183 |
+
"learning_rate": 4.830917874396135e-07,
|
| 1184 |
+
"loss": 0.5584,
|
| 1185 |
+
"step": 1320
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 18.05072463768116,
|
| 1189 |
+
"grad_norm": 9.562509536743164,
|
| 1190 |
+
"learning_rate": 4.025764895330113e-07,
|
| 1191 |
+
"loss": 0.4406,
|
| 1192 |
+
"step": 1330
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 18.05072463768116,
|
| 1196 |
+
"eval_accuracy": 0.6865671641791045,
|
| 1197 |
+
"eval_accuracy_hold": 0.0,
|
| 1198 |
+
"eval_accuracy_preparation": 0.8533333333333334,
|
| 1199 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 1200 |
+
"eval_accuracy_stroke": 0.0,
|
| 1201 |
+
"eval_accuracy_unknown": 1.0,
|
| 1202 |
+
"eval_loss": 0.7800428867340088,
|
| 1203 |
+
"eval_runtime": 10.5812,
|
| 1204 |
+
"eval_samples_per_second": 12.664,
|
| 1205 |
+
"eval_steps_per_second": 1.607,
|
| 1206 |
+
"step": 1330
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 19.007246376811594,
|
| 1210 |
+
"grad_norm": 14.316792488098145,
|
| 1211 |
+
"learning_rate": 3.22061191626409e-07,
|
| 1212 |
+
"loss": 0.5768,
|
| 1213 |
+
"step": 1340
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 19.014492753623188,
|
| 1217 |
+
"grad_norm": 12.971110343933105,
|
| 1218 |
+
"learning_rate": 2.4154589371980677e-07,
|
| 1219 |
+
"loss": 0.5324,
|
| 1220 |
+
"step": 1350
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 19.02173913043478,
|
| 1224 |
+
"grad_norm": 18.0727596282959,
|
| 1225 |
+
"learning_rate": 1.610305958132045e-07,
|
| 1226 |
+
"loss": 0.4406,
|
| 1227 |
+
"step": 1360
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 19.028985507246375,
|
| 1231 |
+
"grad_norm": 20.428890228271484,
|
| 1232 |
+
"learning_rate": 8.051529790660226e-08,
|
| 1233 |
+
"loss": 0.5222,
|
| 1234 |
+
"step": 1370
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 19.036231884057973,
|
| 1238 |
+
"grad_norm": 28.665119171142578,
|
| 1239 |
+
"learning_rate": 0.0,
|
| 1240 |
+
"loss": 0.6358,
|
| 1241 |
+
"step": 1380
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 19.036231884057973,
|
| 1245 |
+
"eval_accuracy": 0.7014925373134329,
|
| 1246 |
+
"eval_accuracy_hold": 0.0,
|
| 1247 |
+
"eval_accuracy_preparation": 0.88,
|
| 1248 |
+
"eval_accuracy_recovery": 0.16666666666666666,
|
| 1249 |
+
"eval_accuracy_stroke": 0.0,
|
| 1250 |
+
"eval_accuracy_unknown": 1.0,
|
| 1251 |
+
"eval_loss": 0.7816389203071594,
|
| 1252 |
+
"eval_runtime": 10.5317,
|
| 1253 |
+
"eval_samples_per_second": 12.724,
|
| 1254 |
+
"eval_steps_per_second": 1.614,
|
| 1255 |
+
"step": 1380
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 19.036231884057973,
|
| 1259 |
+
"step": 1380,
|
| 1260 |
+
"total_flos": 1.3709569395152978e+19,
|
| 1261 |
+
"train_loss": 0.8566418366155762,
|
| 1262 |
+
"train_runtime": 2187.3823,
|
| 1263 |
+
"train_samples_per_second": 5.047,
|
| 1264 |
+
"train_steps_per_second": 0.631
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 19.036231884057973,
|
| 1268 |
+
"eval_accuracy": 0.6870748299319728,
|
| 1269 |
+
"eval_accuracy_hold": 0.0,
|
| 1270 |
+
"eval_accuracy_preparation": 0.922077922077922,
|
| 1271 |
+
"eval_accuracy_recovery": 0.0,
|
| 1272 |
+
"eval_accuracy_stroke": 0.0,
|
| 1273 |
+
"eval_accuracy_unknown": 0.8571428571428571,
|
| 1274 |
+
"eval_loss": 0.9822249412536621,
|
| 1275 |
+
"eval_runtime": 13.5675,
|
| 1276 |
+
"eval_samples_per_second": 10.835,
|
| 1277 |
+
"eval_steps_per_second": 1.4,
|
| 1278 |
+
"step": 1380
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 19.036231884057973,
|
| 1282 |
+
"eval_accuracy": 0.6870748299319728,
|
| 1283 |
+
"eval_accuracy_hold": 0.0,
|
| 1284 |
+
"eval_accuracy_preparation": 0.922077922077922,
|
| 1285 |
+
"eval_accuracy_recovery": 0.0,
|
| 1286 |
+
"eval_accuracy_stroke": 0.0,
|
| 1287 |
+
"eval_accuracy_unknown": 0.8571428571428571,
|
| 1288 |
+
"eval_loss": 0.9822250604629517,
|
| 1289 |
+
"eval_runtime": 12.2253,
|
| 1290 |
+
"eval_samples_per_second": 12.024,
|
| 1291 |
+
"eval_steps_per_second": 1.554,
|
| 1292 |
+
"step": 1380
|
| 1293 |
+
}
|
| 1294 |
+
],
|
| 1295 |
+
"logging_steps": 10,
|
| 1296 |
+
"max_steps": 1380,
|
| 1297 |
+
"num_input_tokens_seen": 0,
|
| 1298 |
+
"num_train_epochs": 9223372036854775807,
|
| 1299 |
+
"save_steps": 500,
|
| 1300 |
+
"stateful_callbacks": {
|
| 1301 |
+
"TrainerControl": {
|
| 1302 |
+
"args": {
|
| 1303 |
+
"should_epoch_stop": false,
|
| 1304 |
+
"should_evaluate": false,
|
| 1305 |
+
"should_log": false,
|
| 1306 |
+
"should_save": true,
|
| 1307 |
+
"should_training_stop": true
|
| 1308 |
+
},
|
| 1309 |
+
"attributes": {}
|
| 1310 |
+
}
|
| 1311 |
+
},
|
| 1312 |
+
"total_flos": 1.3709569395152978e+19,
|
| 1313 |
+
"train_batch_size": 8,
|
| 1314 |
+
"trial_name": null,
|
| 1315 |
+
"trial_params": null
|
| 1316 |
+
}
|