End of training
Browse files- all_results.json +11 -0
- test_results.json +11 -0
- trainer_state.json +1574 -0
all_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 79.01,
|
| 3 |
+
"eval_accuracy": 0.6697312588401697,
|
| 4 |
+
"eval_f1": 0.635434606603408,
|
| 5 |
+
"eval_loss": 1.1357100009918213,
|
| 6 |
+
"eval_precision": 0.642912037609069,
|
| 7 |
+
"eval_recall": 0.6697312588401697,
|
| 8 |
+
"eval_runtime": 860.1584,
|
| 9 |
+
"eval_samples_per_second": 4.932,
|
| 10 |
+
"eval_steps_per_second": 0.617
|
| 11 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 79.01,
|
| 3 |
+
"eval_accuracy": 0.6697312588401697,
|
| 4 |
+
"eval_f1": 0.635434606603408,
|
| 5 |
+
"eval_loss": 1.1357100009918213,
|
| 6 |
+
"eval_precision": 0.642912037609069,
|
| 7 |
+
"eval_recall": 0.6697312588401697,
|
| 8 |
+
"eval_runtime": 860.1584,
|
| 9 |
+
"eval_samples_per_second": 4.932,
|
| 10 |
+
"eval_steps_per_second": 0.617
|
| 11 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1574 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6680810938236681,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-chickenbehaviour-2/checkpoint-23865",
|
| 4 |
+
"epoch": 79.01218956303049,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 127240,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"grad_norm": 34.09493637084961,
|
| 14 |
+
"learning_rate": 6.251964790946243e-06,
|
| 15 |
+
"loss": 2.035,
|
| 16 |
+
"step": 1591
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01,
|
| 20 |
+
"eval_accuracy": 0.5115511551155115,
|
| 21 |
+
"eval_f1": 0.4139359299241216,
|
| 22 |
+
"eval_loss": 1.6961450576782227,
|
| 23 |
+
"eval_precision": 0.426389188260245,
|
| 24 |
+
"eval_recall": 0.5115511551155115,
|
| 25 |
+
"eval_runtime": 443.2234,
|
| 26 |
+
"eval_samples_per_second": 9.571,
|
| 27 |
+
"eval_steps_per_second": 1.198,
|
| 28 |
+
"step": 1591
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 1.01,
|
| 32 |
+
"grad_norm": 25.231983184814453,
|
| 33 |
+
"learning_rate": 1.2503929581892487e-05,
|
| 34 |
+
"loss": 1.5431,
|
| 35 |
+
"step": 3182
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.01,
|
| 39 |
+
"eval_accuracy": 0.5898161244695899,
|
| 40 |
+
"eval_f1": 0.5166622256451852,
|
| 41 |
+
"eval_loss": 1.439492106437683,
|
| 42 |
+
"eval_precision": 0.5240058255527255,
|
| 43 |
+
"eval_recall": 0.5898161244695899,
|
| 44 |
+
"eval_runtime": 437.564,
|
| 45 |
+
"eval_samples_per_second": 9.695,
|
| 46 |
+
"eval_steps_per_second": 1.214,
|
| 47 |
+
"step": 3182
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 2.01,
|
| 51 |
+
"grad_norm": 24.721105575561523,
|
| 52 |
+
"learning_rate": 1.875589437283873e-05,
|
| 53 |
+
"loss": 1.4118,
|
| 54 |
+
"step": 4773
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 2.01,
|
| 58 |
+
"eval_accuracy": 0.6051390853371051,
|
| 59 |
+
"eval_f1": 0.5535236961754626,
|
| 60 |
+
"eval_loss": 1.363150715827942,
|
| 61 |
+
"eval_precision": 0.5552721379615939,
|
| 62 |
+
"eval_recall": 0.6051390853371051,
|
| 63 |
+
"eval_runtime": 683.9412,
|
| 64 |
+
"eval_samples_per_second": 6.202,
|
| 65 |
+
"eval_steps_per_second": 0.776,
|
| 66 |
+
"step": 4773
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 3.01,
|
| 70 |
+
"grad_norm": 7.513462066650391,
|
| 71 |
+
"learning_rate": 2.5007859163784973e-05,
|
| 72 |
+
"loss": 1.3413,
|
| 73 |
+
"step": 6364
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 3.01,
|
| 77 |
+
"eval_accuracy": 0.6020744931636021,
|
| 78 |
+
"eval_f1": 0.5384080508073906,
|
| 79 |
+
"eval_loss": 1.331170678138733,
|
| 80 |
+
"eval_precision": 0.551611875314686,
|
| 81 |
+
"eval_recall": 0.6020744931636021,
|
| 82 |
+
"eval_runtime": 680.73,
|
| 83 |
+
"eval_samples_per_second": 6.232,
|
| 84 |
+
"eval_steps_per_second": 0.78,
|
| 85 |
+
"step": 6364
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 4.01,
|
| 89 |
+
"grad_norm": 23.583574295043945,
|
| 90 |
+
"learning_rate": 3.125982395473121e-05,
|
| 91 |
+
"loss": 1.2969,
|
| 92 |
+
"step": 7955
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 4.01,
|
| 96 |
+
"eval_accuracy": 0.6211692597831212,
|
| 97 |
+
"eval_f1": 0.5663241746963936,
|
| 98 |
+
"eval_loss": 1.273905873298645,
|
| 99 |
+
"eval_precision": 0.6121530259266542,
|
| 100 |
+
"eval_recall": 0.6211692597831212,
|
| 101 |
+
"eval_runtime": 663.3464,
|
| 102 |
+
"eval_samples_per_second": 6.395,
|
| 103 |
+
"eval_steps_per_second": 0.8,
|
| 104 |
+
"step": 7955
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 5.01,
|
| 108 |
+
"grad_norm": 12.484142303466797,
|
| 109 |
+
"learning_rate": 3.751178874567746e-05,
|
| 110 |
+
"loss": 1.2636,
|
| 111 |
+
"step": 9546
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 5.01,
|
| 115 |
+
"eval_accuracy": 0.6058462989156058,
|
| 116 |
+
"eval_f1": 0.5429975108084805,
|
| 117 |
+
"eval_loss": 1.3211930990219116,
|
| 118 |
+
"eval_precision": 0.6186956973637323,
|
| 119 |
+
"eval_recall": 0.6058462989156058,
|
| 120 |
+
"eval_runtime": 661.8451,
|
| 121 |
+
"eval_samples_per_second": 6.409,
|
| 122 |
+
"eval_steps_per_second": 0.802,
|
| 123 |
+
"step": 9546
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 6.01,
|
| 127 |
+
"grad_norm": 6.579521179199219,
|
| 128 |
+
"learning_rate": 4.3763753536623704e-05,
|
| 129 |
+
"loss": 1.2231,
|
| 130 |
+
"step": 11137
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 6.01,
|
| 134 |
+
"eval_accuracy": 0.6242338519566243,
|
| 135 |
+
"eval_f1": 0.5747264230461347,
|
| 136 |
+
"eval_loss": 1.2543139457702637,
|
| 137 |
+
"eval_precision": 0.6460983768240532,
|
| 138 |
+
"eval_recall": 0.6242338519566243,
|
| 139 |
+
"eval_runtime": 669.9717,
|
| 140 |
+
"eval_samples_per_second": 6.332,
|
| 141 |
+
"eval_steps_per_second": 0.793,
|
| 142 |
+
"step": 11137
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 7.01,
|
| 146 |
+
"grad_norm": 12.255209922790527,
|
| 147 |
+
"learning_rate": 4.99982535191589e-05,
|
| 148 |
+
"loss": 1.1989,
|
| 149 |
+
"step": 12728
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 7.01,
|
| 153 |
+
"eval_accuracy": 0.6404997642621405,
|
| 154 |
+
"eval_f1": 0.5869125503826516,
|
| 155 |
+
"eval_loss": 1.2377874851226807,
|
| 156 |
+
"eval_precision": 0.6356351683189131,
|
| 157 |
+
"eval_recall": 0.6404997642621405,
|
| 158 |
+
"eval_runtime": 678.5987,
|
| 159 |
+
"eval_samples_per_second": 6.251,
|
| 160 |
+
"eval_steps_per_second": 0.782,
|
| 161 |
+
"step": 12728
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 8.01,
|
| 165 |
+
"grad_norm": 9.07247257232666,
|
| 166 |
+
"learning_rate": 4.9303590764609316e-05,
|
| 167 |
+
"loss": 1.1566,
|
| 168 |
+
"step": 14319
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 8.01,
|
| 172 |
+
"eval_accuracy": 0.6527581329561527,
|
| 173 |
+
"eval_f1": 0.594207871765373,
|
| 174 |
+
"eval_loss": 1.2123682498931885,
|
| 175 |
+
"eval_precision": 0.6198764093734253,
|
| 176 |
+
"eval_recall": 0.6527581329561527,
|
| 177 |
+
"eval_runtime": 667.9014,
|
| 178 |
+
"eval_samples_per_second": 6.351,
|
| 179 |
+
"eval_steps_per_second": 0.795,
|
| 180 |
+
"step": 14319
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 9.01,
|
| 184 |
+
"grad_norm": 13.849576950073242,
|
| 185 |
+
"learning_rate": 4.8608928010059734e-05,
|
| 186 |
+
"loss": 1.1145,
|
| 187 |
+
"step": 15910
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 9.01,
|
| 191 |
+
"eval_accuracy": 0.6475719000471476,
|
| 192 |
+
"eval_f1": 0.605227949854272,
|
| 193 |
+
"eval_loss": 1.1802877187728882,
|
| 194 |
+
"eval_precision": 0.6340925434803425,
|
| 195 |
+
"eval_recall": 0.6475719000471476,
|
| 196 |
+
"eval_runtime": 672.4937,
|
| 197 |
+
"eval_samples_per_second": 6.308,
|
| 198 |
+
"eval_steps_per_second": 0.79,
|
| 199 |
+
"step": 15910
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 10.01,
|
| 203 |
+
"grad_norm": 26.01058006286621,
|
| 204 |
+
"learning_rate": 4.791426525551015e-05,
|
| 205 |
+
"loss": 1.0567,
|
| 206 |
+
"step": 17501
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 10.01,
|
| 210 |
+
"eval_accuracy": 0.6265912305516266,
|
| 211 |
+
"eval_f1": 0.5969458291906395,
|
| 212 |
+
"eval_loss": 1.2576700448989868,
|
| 213 |
+
"eval_precision": 0.6278990782548687,
|
| 214 |
+
"eval_recall": 0.6265912305516266,
|
| 215 |
+
"eval_runtime": 662.684,
|
| 216 |
+
"eval_samples_per_second": 6.401,
|
| 217 |
+
"eval_steps_per_second": 0.801,
|
| 218 |
+
"step": 17501
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 11.01,
|
| 222 |
+
"grad_norm": 4.436341285705566,
|
| 223 |
+
"learning_rate": 4.721960250096057e-05,
|
| 224 |
+
"loss": 1.0172,
|
| 225 |
+
"step": 19092
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 11.01,
|
| 229 |
+
"eval_accuracy": 0.657001414427157,
|
| 230 |
+
"eval_f1": 0.6082570131553916,
|
| 231 |
+
"eval_loss": 1.1961308717727661,
|
| 232 |
+
"eval_precision": 0.6369039125738234,
|
| 233 |
+
"eval_recall": 0.657001414427157,
|
| 234 |
+
"eval_runtime": 753.1481,
|
| 235 |
+
"eval_samples_per_second": 5.632,
|
| 236 |
+
"eval_steps_per_second": 0.705,
|
| 237 |
+
"step": 19092
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 12.01,
|
| 241 |
+
"grad_norm": 12.711715698242188,
|
| 242 |
+
"learning_rate": 4.652493974641098e-05,
|
| 243 |
+
"loss": 0.9817,
|
| 244 |
+
"step": 20683
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 12.01,
|
| 248 |
+
"eval_accuracy": 0.6619519094766619,
|
| 249 |
+
"eval_f1": 0.6048958777519221,
|
| 250 |
+
"eval_loss": 1.2287497520446777,
|
| 251 |
+
"eval_precision": 0.6498679093091749,
|
| 252 |
+
"eval_recall": 0.6619519094766619,
|
| 253 |
+
"eval_runtime": 765.9815,
|
| 254 |
+
"eval_samples_per_second": 5.538,
|
| 255 |
+
"eval_steps_per_second": 0.693,
|
| 256 |
+
"step": 20683
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 13.01,
|
| 260 |
+
"grad_norm": 6.408486366271973,
|
| 261 |
+
"learning_rate": 4.583027699186141e-05,
|
| 262 |
+
"loss": 0.9279,
|
| 263 |
+
"step": 22274
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 13.01,
|
| 267 |
+
"eval_accuracy": 0.6548797736916548,
|
| 268 |
+
"eval_f1": 0.6213080237690763,
|
| 269 |
+
"eval_loss": 1.2358391284942627,
|
| 270 |
+
"eval_precision": 0.6503596413080538,
|
| 271 |
+
"eval_recall": 0.6548797736916548,
|
| 272 |
+
"eval_runtime": 754.2864,
|
| 273 |
+
"eval_samples_per_second": 5.624,
|
| 274 |
+
"eval_steps_per_second": 0.704,
|
| 275 |
+
"step": 22274
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 14.01,
|
| 279 |
+
"grad_norm": 14.318334579467773,
|
| 280 |
+
"learning_rate": 4.513561423731182e-05,
|
| 281 |
+
"loss": 0.8913,
|
| 282 |
+
"step": 23865
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 14.01,
|
| 286 |
+
"eval_accuracy": 0.6680810938236681,
|
| 287 |
+
"eval_f1": 0.6307923814434653,
|
| 288 |
+
"eval_loss": 1.181541919708252,
|
| 289 |
+
"eval_precision": 0.6325361512013947,
|
| 290 |
+
"eval_recall": 0.6680810938236681,
|
| 291 |
+
"eval_runtime": 766.2903,
|
| 292 |
+
"eval_samples_per_second": 5.536,
|
| 293 |
+
"eval_steps_per_second": 0.693,
|
| 294 |
+
"step": 23865
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 15.01,
|
| 298 |
+
"grad_norm": 14.112875938415527,
|
| 299 |
+
"learning_rate": 4.444095148276224e-05,
|
| 300 |
+
"loss": 0.8559,
|
| 301 |
+
"step": 25456
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 15.01,
|
| 305 |
+
"eval_accuracy": 0.639085337105139,
|
| 306 |
+
"eval_f1": 0.6036680470772597,
|
| 307 |
+
"eval_loss": 1.3211859464645386,
|
| 308 |
+
"eval_precision": 0.639173575618385,
|
| 309 |
+
"eval_recall": 0.639085337105139,
|
| 310 |
+
"eval_runtime": 741.2024,
|
| 311 |
+
"eval_samples_per_second": 5.723,
|
| 312 |
+
"eval_steps_per_second": 0.716,
|
| 313 |
+
"step": 25456
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 16.01,
|
| 317 |
+
"grad_norm": 19.837888717651367,
|
| 318 |
+
"learning_rate": 4.3746288728212655e-05,
|
| 319 |
+
"loss": 0.8083,
|
| 320 |
+
"step": 27047
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 16.01,
|
| 324 |
+
"eval_accuracy": 0.6230551626591231,
|
| 325 |
+
"eval_f1": 0.6006019601279614,
|
| 326 |
+
"eval_loss": 1.307276725769043,
|
| 327 |
+
"eval_precision": 0.6250962347990193,
|
| 328 |
+
"eval_recall": 0.6230551626591231,
|
| 329 |
+
"eval_runtime": 755.5211,
|
| 330 |
+
"eval_samples_per_second": 5.615,
|
| 331 |
+
"eval_steps_per_second": 0.703,
|
| 332 |
+
"step": 27047
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 17.01,
|
| 336 |
+
"grad_norm": 7.695098876953125,
|
| 337 |
+
"learning_rate": 4.305162597366307e-05,
|
| 338 |
+
"loss": 0.7662,
|
| 339 |
+
"step": 28638
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 17.01,
|
| 343 |
+
"eval_accuracy": 0.6461574728901461,
|
| 344 |
+
"eval_f1": 0.6214078287802024,
|
| 345 |
+
"eval_loss": 1.2981834411621094,
|
| 346 |
+
"eval_precision": 0.625200272023183,
|
| 347 |
+
"eval_recall": 0.6461574728901461,
|
| 348 |
+
"eval_runtime": 760.7104,
|
| 349 |
+
"eval_samples_per_second": 5.576,
|
| 350 |
+
"eval_steps_per_second": 0.698,
|
| 351 |
+
"step": 28638
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 18.01,
|
| 355 |
+
"grad_norm": 17.791141510009766,
|
| 356 |
+
"learning_rate": 4.2356963219113485e-05,
|
| 357 |
+
"loss": 0.7363,
|
| 358 |
+
"step": 30229
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 18.01,
|
| 362 |
+
"eval_accuracy": 0.6574728901461575,
|
| 363 |
+
"eval_f1": 0.6264016299127368,
|
| 364 |
+
"eval_loss": 1.3019486665725708,
|
| 365 |
+
"eval_precision": 0.6427931632515602,
|
| 366 |
+
"eval_recall": 0.6574728901461575,
|
| 367 |
+
"eval_runtime": 732.8997,
|
| 368 |
+
"eval_samples_per_second": 5.788,
|
| 369 |
+
"eval_steps_per_second": 0.725,
|
| 370 |
+
"step": 30229
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 19.01,
|
| 374 |
+
"grad_norm": 12.370431900024414,
|
| 375 |
+
"learning_rate": 4.166230046456391e-05,
|
| 376 |
+
"loss": 0.6787,
|
| 377 |
+
"step": 31820
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 19.01,
|
| 381 |
+
"eval_accuracy": 0.6511079679396511,
|
| 382 |
+
"eval_f1": 0.6229627569285623,
|
| 383 |
+
"eval_loss": 1.3867465257644653,
|
| 384 |
+
"eval_precision": 0.6368072307821887,
|
| 385 |
+
"eval_recall": 0.6511079679396511,
|
| 386 |
+
"eval_runtime": 662.5188,
|
| 387 |
+
"eval_samples_per_second": 6.403,
|
| 388 |
+
"eval_steps_per_second": 0.801,
|
| 389 |
+
"step": 31820
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 20.01,
|
| 393 |
+
"grad_norm": 10.867647171020508,
|
| 394 |
+
"learning_rate": 4.096763771001432e-05,
|
| 395 |
+
"loss": 0.6433,
|
| 396 |
+
"step": 33411
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 20.01,
|
| 400 |
+
"eval_accuracy": 0.6364922206506365,
|
| 401 |
+
"eval_f1": 0.6138991311356594,
|
| 402 |
+
"eval_loss": 1.4018532037734985,
|
| 403 |
+
"eval_precision": 0.6374756941286471,
|
| 404 |
+
"eval_recall": 0.6364922206506365,
|
| 405 |
+
"eval_runtime": 667.8552,
|
| 406 |
+
"eval_samples_per_second": 6.352,
|
| 407 |
+
"eval_steps_per_second": 0.795,
|
| 408 |
+
"step": 33411
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 21.01,
|
| 412 |
+
"grad_norm": 23.982513427734375,
|
| 413 |
+
"learning_rate": 4.0272974955464746e-05,
|
| 414 |
+
"loss": 0.5969,
|
| 415 |
+
"step": 35002
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 21.01,
|
| 419 |
+
"eval_accuracy": 0.6341348420556341,
|
| 420 |
+
"eval_f1": 0.6104193509325564,
|
| 421 |
+
"eval_loss": 1.4419147968292236,
|
| 422 |
+
"eval_precision": 0.6211641893240438,
|
| 423 |
+
"eval_recall": 0.6341348420556341,
|
| 424 |
+
"eval_runtime": 657.2327,
|
| 425 |
+
"eval_samples_per_second": 6.454,
|
| 426 |
+
"eval_steps_per_second": 0.808,
|
| 427 |
+
"step": 35002
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 22.01,
|
| 431 |
+
"grad_norm": 4.0233635902404785,
|
| 432 |
+
"learning_rate": 3.957831220091516e-05,
|
| 433 |
+
"loss": 0.563,
|
| 434 |
+
"step": 36593
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 22.01,
|
| 438 |
+
"eval_accuracy": 0.6508722300801508,
|
| 439 |
+
"eval_f1": 0.6170283959147944,
|
| 440 |
+
"eval_loss": 1.4777988195419312,
|
| 441 |
+
"eval_precision": 0.6292668904937919,
|
| 442 |
+
"eval_recall": 0.6508722300801508,
|
| 443 |
+
"eval_runtime": 657.9566,
|
| 444 |
+
"eval_samples_per_second": 6.447,
|
| 445 |
+
"eval_steps_per_second": 0.807,
|
| 446 |
+
"step": 36593
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 23.01,
|
| 450 |
+
"grad_norm": 23.503877639770508,
|
| 451 |
+
"learning_rate": 3.8883649446365576e-05,
|
| 452 |
+
"loss": 0.5252,
|
| 453 |
+
"step": 38184
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 23.01,
|
| 457 |
+
"eval_accuracy": 0.6433286185761433,
|
| 458 |
+
"eval_f1": 0.6213976507987007,
|
| 459 |
+
"eval_loss": 1.486406683921814,
|
| 460 |
+
"eval_precision": 0.6316274985409115,
|
| 461 |
+
"eval_recall": 0.6433286185761433,
|
| 462 |
+
"eval_runtime": 652.8988,
|
| 463 |
+
"eval_samples_per_second": 6.497,
|
| 464 |
+
"eval_steps_per_second": 0.813,
|
| 465 |
+
"step": 38184
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 24.01,
|
| 469 |
+
"grad_norm": 29.642290115356445,
|
| 470 |
+
"learning_rate": 3.8188986691815994e-05,
|
| 471 |
+
"loss": 0.5,
|
| 472 |
+
"step": 39775
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 24.01,
|
| 476 |
+
"eval_accuracy": 0.6232909005186232,
|
| 477 |
+
"eval_f1": 0.6023496407577721,
|
| 478 |
+
"eval_loss": 1.6704081296920776,
|
| 479 |
+
"eval_precision": 0.6273005697774543,
|
| 480 |
+
"eval_recall": 0.6232909005186232,
|
| 481 |
+
"eval_runtime": 651.691,
|
| 482 |
+
"eval_samples_per_second": 6.509,
|
| 483 |
+
"eval_steps_per_second": 0.815,
|
| 484 |
+
"step": 39775
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 25.01,
|
| 488 |
+
"grad_norm": 0.25407856702804565,
|
| 489 |
+
"learning_rate": 3.749432393726641e-05,
|
| 490 |
+
"loss": 0.4622,
|
| 491 |
+
"step": 41366
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 25.01,
|
| 495 |
+
"eval_accuracy": 0.6487505893446488,
|
| 496 |
+
"eval_f1": 0.6119499708020253,
|
| 497 |
+
"eval_loss": 1.665787935256958,
|
| 498 |
+
"eval_precision": 0.6259817520615986,
|
| 499 |
+
"eval_recall": 0.6487505893446488,
|
| 500 |
+
"eval_runtime": 664.0356,
|
| 501 |
+
"eval_samples_per_second": 6.388,
|
| 502 |
+
"eval_steps_per_second": 0.8,
|
| 503 |
+
"step": 41366
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 26.01,
|
| 507 |
+
"grad_norm": 12.53039836883545,
|
| 508 |
+
"learning_rate": 3.6799661182716824e-05,
|
| 509 |
+
"loss": 0.4292,
|
| 510 |
+
"step": 42957
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 26.01,
|
| 514 |
+
"eval_accuracy": 0.6494578029231495,
|
| 515 |
+
"eval_f1": 0.6242771027666587,
|
| 516 |
+
"eval_loss": 1.6428192853927612,
|
| 517 |
+
"eval_precision": 0.6286829818492489,
|
| 518 |
+
"eval_recall": 0.6494578029231495,
|
| 519 |
+
"eval_runtime": 673.3255,
|
| 520 |
+
"eval_samples_per_second": 6.3,
|
| 521 |
+
"eval_steps_per_second": 0.789,
|
| 522 |
+
"step": 42957
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 27.01,
|
| 526 |
+
"grad_norm": 2.96919846534729,
|
| 527 |
+
"learning_rate": 3.610499842816725e-05,
|
| 528 |
+
"loss": 0.4044,
|
| 529 |
+
"step": 44548
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 27.01,
|
| 533 |
+
"eval_accuracy": 0.6586515794436586,
|
| 534 |
+
"eval_f1": 0.6387100670675616,
|
| 535 |
+
"eval_loss": 1.6702899932861328,
|
| 536 |
+
"eval_precision": 0.6310773161399348,
|
| 537 |
+
"eval_recall": 0.6586515794436586,
|
| 538 |
+
"eval_runtime": 657.3579,
|
| 539 |
+
"eval_samples_per_second": 6.453,
|
| 540 |
+
"eval_steps_per_second": 0.808,
|
| 541 |
+
"step": 44548
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 28.01,
|
| 545 |
+
"grad_norm": 2.961900472640991,
|
| 546 |
+
"learning_rate": 3.541033567361766e-05,
|
| 547 |
+
"loss": 0.3952,
|
| 548 |
+
"step": 46139
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 28.01,
|
| 552 |
+
"eval_accuracy": 0.632956152758133,
|
| 553 |
+
"eval_f1": 0.6123356581191824,
|
| 554 |
+
"eval_loss": 1.7576137781143188,
|
| 555 |
+
"eval_precision": 0.6170827944289541,
|
| 556 |
+
"eval_recall": 0.632956152758133,
|
| 557 |
+
"eval_runtime": 659.0162,
|
| 558 |
+
"eval_samples_per_second": 6.437,
|
| 559 |
+
"eval_steps_per_second": 0.806,
|
| 560 |
+
"step": 46139
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 29.01,
|
| 564 |
+
"grad_norm": 17.30776596069336,
|
| 565 |
+
"learning_rate": 3.471567291906808e-05,
|
| 566 |
+
"loss": 0.3681,
|
| 567 |
+
"step": 47730
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 29.01,
|
| 571 |
+
"eval_accuracy": 0.6553512494106554,
|
| 572 |
+
"eval_f1": 0.6231268362993411,
|
| 573 |
+
"eval_loss": 1.9031915664672852,
|
| 574 |
+
"eval_precision": 0.6349163976683712,
|
| 575 |
+
"eval_recall": 0.6553512494106554,
|
| 576 |
+
"eval_runtime": 657.4951,
|
| 577 |
+
"eval_samples_per_second": 6.452,
|
| 578 |
+
"eval_steps_per_second": 0.808,
|
| 579 |
+
"step": 47730
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 30.01,
|
| 583 |
+
"grad_norm": 15.424605369567871,
|
| 584 |
+
"learning_rate": 3.40210101645185e-05,
|
| 585 |
+
"loss": 0.3541,
|
| 586 |
+
"step": 49321
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 30.01,
|
| 590 |
+
"eval_accuracy": 0.6445073078736445,
|
| 591 |
+
"eval_f1": 0.6207211544517277,
|
| 592 |
+
"eval_loss": 1.9507993459701538,
|
| 593 |
+
"eval_precision": 0.632026480012122,
|
| 594 |
+
"eval_recall": 0.6445073078736445,
|
| 595 |
+
"eval_runtime": 655.6116,
|
| 596 |
+
"eval_samples_per_second": 6.47,
|
| 597 |
+
"eval_steps_per_second": 0.81,
|
| 598 |
+
"step": 49321
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 31.01,
|
| 602 |
+
"grad_norm": 0.31878066062927246,
|
| 603 |
+
"learning_rate": 3.3326347409968915e-05,
|
| 604 |
+
"loss": 0.322,
|
| 605 |
+
"step": 50912
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 31.01,
|
| 609 |
+
"eval_accuracy": 0.6225836869401226,
|
| 610 |
+
"eval_f1": 0.6098707323593034,
|
| 611 |
+
"eval_loss": 2.1316964626312256,
|
| 612 |
+
"eval_precision": 0.6276797078299916,
|
| 613 |
+
"eval_recall": 0.6225836869401226,
|
| 614 |
+
"eval_runtime": 655.8263,
|
| 615 |
+
"eval_samples_per_second": 6.468,
|
| 616 |
+
"eval_steps_per_second": 0.81,
|
| 617 |
+
"step": 50912
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 32.01,
|
| 621 |
+
"grad_norm": 1.1452162265777588,
|
| 622 |
+
"learning_rate": 3.263168465541933e-05,
|
| 623 |
+
"loss": 0.3239,
|
| 624 |
+
"step": 52503
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 32.01,
|
| 628 |
+
"eval_accuracy": 0.6508722300801508,
|
| 629 |
+
"eval_f1": 0.6327759879930807,
|
| 630 |
+
"eval_loss": 1.9785257577896118,
|
| 631 |
+
"eval_precision": 0.6320825441180629,
|
| 632 |
+
"eval_recall": 0.6508722300801508,
|
| 633 |
+
"eval_runtime": 656.8912,
|
| 634 |
+
"eval_samples_per_second": 6.458,
|
| 635 |
+
"eval_steps_per_second": 0.808,
|
| 636 |
+
"step": 52503
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 33.01,
|
| 640 |
+
"grad_norm": 18.911890029907227,
|
| 641 |
+
"learning_rate": 3.193702190086975e-05,
|
| 642 |
+
"loss": 0.301,
|
| 643 |
+
"step": 54094
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 33.01,
|
| 647 |
+
"eval_accuracy": 0.6435643564356436,
|
| 648 |
+
"eval_f1": 0.6097216315179437,
|
| 649 |
+
"eval_loss": 2.2050163745880127,
|
| 650 |
+
"eval_precision": 0.6258806132730886,
|
| 651 |
+
"eval_recall": 0.6435643564356436,
|
| 652 |
+
"eval_runtime": 659.2017,
|
| 653 |
+
"eval_samples_per_second": 6.435,
|
| 654 |
+
"eval_steps_per_second": 0.806,
|
| 655 |
+
"step": 54094
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 34.01,
|
| 659 |
+
"grad_norm": 0.37402623891830444,
|
| 660 |
+
"learning_rate": 3.124235914632016e-05,
|
| 661 |
+
"loss": 0.28,
|
| 662 |
+
"step": 55685
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 34.01,
|
| 666 |
+
"eval_accuracy": 0.6320132013201321,
|
| 667 |
+
"eval_f1": 0.6173670056289928,
|
| 668 |
+
"eval_loss": 2.2267725467681885,
|
| 669 |
+
"eval_precision": 0.6318691154473476,
|
| 670 |
+
"eval_recall": 0.6320132013201321,
|
| 671 |
+
"eval_runtime": 655.5706,
|
| 672 |
+
"eval_samples_per_second": 6.471,
|
| 673 |
+
"eval_steps_per_second": 0.81,
|
| 674 |
+
"step": 55685
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 35.01,
|
| 678 |
+
"grad_norm": 0.24996362626552582,
|
| 679 |
+
"learning_rate": 3.054769639177059e-05,
|
| 680 |
+
"loss": 0.2742,
|
| 681 |
+
"step": 57276
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 35.01,
|
| 685 |
+
"eval_accuracy": 0.641914191419142,
|
| 686 |
+
"eval_f1": 0.6158490856542709,
|
| 687 |
+
"eval_loss": 2.3538448810577393,
|
| 688 |
+
"eval_precision": 0.6239469548156946,
|
| 689 |
+
"eval_recall": 0.641914191419142,
|
| 690 |
+
"eval_runtime": 658.9065,
|
| 691 |
+
"eval_samples_per_second": 6.438,
|
| 692 |
+
"eval_steps_per_second": 0.806,
|
| 693 |
+
"step": 57276
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 36.01,
|
| 697 |
+
"grad_norm": 40.534271240234375,
|
| 698 |
+
"learning_rate": 2.9853033637221e-05,
|
| 699 |
+
"loss": 0.2433,
|
| 700 |
+
"step": 58867
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 36.01,
|
| 704 |
+
"eval_accuracy": 0.6478076379066479,
|
| 705 |
+
"eval_f1": 0.6184054664095173,
|
| 706 |
+
"eval_loss": 2.3947157859802246,
|
| 707 |
+
"eval_precision": 0.6237019229335129,
|
| 708 |
+
"eval_recall": 0.6478076379066479,
|
| 709 |
+
"eval_runtime": 658.8157,
|
| 710 |
+
"eval_samples_per_second": 6.439,
|
| 711 |
+
"eval_steps_per_second": 0.806,
|
| 712 |
+
"step": 58867
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 37.01,
|
| 716 |
+
"grad_norm": 0.18938298523426056,
|
| 717 |
+
"learning_rate": 2.915837088267142e-05,
|
| 718 |
+
"loss": 0.2677,
|
| 719 |
+
"step": 60458
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 37.01,
|
| 723 |
+
"eval_accuracy": 0.6454502593116455,
|
| 724 |
+
"eval_f1": 0.6234092398599342,
|
| 725 |
+
"eval_loss": 2.400697708129883,
|
| 726 |
+
"eval_precision": 0.6284538940001326,
|
| 727 |
+
"eval_recall": 0.6454502593116455,
|
| 728 |
+
"eval_runtime": 672.9538,
|
| 729 |
+
"eval_samples_per_second": 6.304,
|
| 730 |
+
"eval_steps_per_second": 0.789,
|
| 731 |
+
"step": 60458
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 38.01,
|
| 735 |
+
"grad_norm": 0.05054619163274765,
|
| 736 |
+
"learning_rate": 2.8463708128121836e-05,
|
| 737 |
+
"loss": 0.2316,
|
| 738 |
+
"step": 62049
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 38.01,
|
| 742 |
+
"eval_accuracy": 0.6296558227251297,
|
| 743 |
+
"eval_f1": 0.6119646490374726,
|
| 744 |
+
"eval_loss": 2.5197205543518066,
|
| 745 |
+
"eval_precision": 0.6245868321675843,
|
| 746 |
+
"eval_recall": 0.6296558227251297,
|
| 747 |
+
"eval_runtime": 677.5713,
|
| 748 |
+
"eval_samples_per_second": 6.261,
|
| 749 |
+
"eval_steps_per_second": 0.784,
|
| 750 |
+
"step": 62049
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 39.01,
|
| 754 |
+
"grad_norm": 0.9518815875053406,
|
| 755 |
+
"learning_rate": 2.7769045373572254e-05,
|
| 756 |
+
"loss": 0.2229,
|
| 757 |
+
"step": 63640
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 39.01,
|
| 761 |
+
"eval_accuracy": 0.6506364922206507,
|
| 762 |
+
"eval_f1": 0.6235476374240025,
|
| 763 |
+
"eval_loss": 2.547842264175415,
|
| 764 |
+
"eval_precision": 0.6321911444043747,
|
| 765 |
+
"eval_recall": 0.6506364922206507,
|
| 766 |
+
"eval_runtime": 680.5865,
|
| 767 |
+
"eval_samples_per_second": 6.233,
|
| 768 |
+
"eval_steps_per_second": 0.78,
|
| 769 |
+
"step": 63640
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 40.01,
|
| 773 |
+
"grad_norm": 0.030849022790789604,
|
| 774 |
+
"learning_rate": 2.707438261902267e-05,
|
| 775 |
+
"loss": 0.215,
|
| 776 |
+
"step": 65231
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 40.01,
|
| 780 |
+
"eval_accuracy": 0.6445073078736445,
|
| 781 |
+
"eval_f1": 0.6209448574869749,
|
| 782 |
+
"eval_loss": 2.516798734664917,
|
| 783 |
+
"eval_precision": 0.6454990917874304,
|
| 784 |
+
"eval_recall": 0.6445073078736445,
|
| 785 |
+
"eval_runtime": 663.9051,
|
| 786 |
+
"eval_samples_per_second": 6.389,
|
| 787 |
+
"eval_steps_per_second": 0.8,
|
| 788 |
+
"step": 65231
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 41.01,
|
| 792 |
+
"grad_norm": 0.38587260246276855,
|
| 793 |
+
"learning_rate": 2.637971986447309e-05,
|
| 794 |
+
"loss": 0.2032,
|
| 795 |
+
"step": 66822
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 41.01,
|
| 799 |
+
"eval_accuracy": 0.6442715700141443,
|
| 800 |
+
"eval_f1": 0.6161042310233847,
|
| 801 |
+
"eval_loss": 2.6606993675231934,
|
| 802 |
+
"eval_precision": 0.6304245945338769,
|
| 803 |
+
"eval_recall": 0.6442715700141443,
|
| 804 |
+
"eval_runtime": 670.0548,
|
| 805 |
+
"eval_samples_per_second": 6.331,
|
| 806 |
+
"eval_steps_per_second": 0.792,
|
| 807 |
+
"step": 66822
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 42.01,
|
| 811 |
+
"grad_norm": 52.90748977661133,
|
| 812 |
+
"learning_rate": 2.5685057109923506e-05,
|
| 813 |
+
"loss": 0.1957,
|
| 814 |
+
"step": 68413
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 42.01,
|
| 818 |
+
"eval_accuracy": 0.6218764733616219,
|
| 819 |
+
"eval_f1": 0.6058915424282973,
|
| 820 |
+
"eval_loss": 2.6433801651000977,
|
| 821 |
+
"eval_precision": 0.6206059185377255,
|
| 822 |
+
"eval_recall": 0.6218764733616219,
|
| 823 |
+
"eval_runtime": 666.3612,
|
| 824 |
+
"eval_samples_per_second": 6.366,
|
| 825 |
+
"eval_steps_per_second": 0.797,
|
| 826 |
+
"step": 68413
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 43.01,
|
| 830 |
+
"grad_norm": 0.5584044456481934,
|
| 831 |
+
"learning_rate": 2.499039435537392e-05,
|
| 832 |
+
"loss": 0.1839,
|
| 833 |
+
"step": 70004
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 43.01,
|
| 837 |
+
"eval_accuracy": 0.648043375766148,
|
| 838 |
+
"eval_f1": 0.6202014579788635,
|
| 839 |
+
"eval_loss": 2.637795925140381,
|
| 840 |
+
"eval_precision": 0.6181521426299653,
|
| 841 |
+
"eval_recall": 0.648043375766148,
|
| 842 |
+
"eval_runtime": 669.3865,
|
| 843 |
+
"eval_samples_per_second": 6.337,
|
| 844 |
+
"eval_steps_per_second": 0.793,
|
| 845 |
+
"step": 70004
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 44.01,
|
| 849 |
+
"grad_norm": 0.02387963980436325,
|
| 850 |
+
"learning_rate": 2.429573160082434e-05,
|
| 851 |
+
"loss": 0.1672,
|
| 852 |
+
"step": 71595
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 44.01,
|
| 856 |
+
"eval_accuracy": 0.632956152758133,
|
| 857 |
+
"eval_f1": 0.6095258779451574,
|
| 858 |
+
"eval_loss": 2.8354904651641846,
|
| 859 |
+
"eval_precision": 0.6175432549732301,
|
| 860 |
+
"eval_recall": 0.632956152758133,
|
| 861 |
+
"eval_runtime": 658.5104,
|
| 862 |
+
"eval_samples_per_second": 6.442,
|
| 863 |
+
"eval_steps_per_second": 0.806,
|
| 864 |
+
"step": 71595
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 45.01,
|
| 868 |
+
"grad_norm": 0.04022861644625664,
|
| 869 |
+
"learning_rate": 2.3601068846274757e-05,
|
| 870 |
+
"loss": 0.1554,
|
| 871 |
+
"step": 73186
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 45.01,
|
| 875 |
+
"eval_accuracy": 0.6296558227251297,
|
| 876 |
+
"eval_f1": 0.609015505263741,
|
| 877 |
+
"eval_loss": 2.8833281993865967,
|
| 878 |
+
"eval_precision": 0.6179849508221367,
|
| 879 |
+
"eval_recall": 0.6296558227251297,
|
| 880 |
+
"eval_runtime": 666.4351,
|
| 881 |
+
"eval_samples_per_second": 6.365,
|
| 882 |
+
"eval_steps_per_second": 0.797,
|
| 883 |
+
"step": 73186
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 46.01,
|
| 887 |
+
"grad_norm": 3.8721201419830322,
|
| 888 |
+
"learning_rate": 2.2906406091725175e-05,
|
| 889 |
+
"loss": 0.1525,
|
| 890 |
+
"step": 74777
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 46.01,
|
| 894 |
+
"eval_accuracy": 0.6499292786421499,
|
| 895 |
+
"eval_f1": 0.6246638391623945,
|
| 896 |
+
"eval_loss": 2.8732240200042725,
|
| 897 |
+
"eval_precision": 0.6212359780695843,
|
| 898 |
+
"eval_recall": 0.6499292786421499,
|
| 899 |
+
"eval_runtime": 672.5613,
|
| 900 |
+
"eval_samples_per_second": 6.307,
|
| 901 |
+
"eval_steps_per_second": 0.79,
|
| 902 |
+
"step": 74777
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 47.01,
|
| 906 |
+
"grad_norm": 0.009451803751289845,
|
| 907 |
+
"learning_rate": 2.221174333717559e-05,
|
| 908 |
+
"loss": 0.1443,
|
| 909 |
+
"step": 76368
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 47.01,
|
| 913 |
+
"eval_accuracy": 0.6513437057991514,
|
| 914 |
+
"eval_f1": 0.6297491645477588,
|
| 915 |
+
"eval_loss": 2.7935521602630615,
|
| 916 |
+
"eval_precision": 0.6239725322581114,
|
| 917 |
+
"eval_recall": 0.6513437057991514,
|
| 918 |
+
"eval_runtime": 673.447,
|
| 919 |
+
"eval_samples_per_second": 6.299,
|
| 920 |
+
"eval_steps_per_second": 0.788,
|
| 921 |
+
"step": 76368
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 48.01,
|
| 925 |
+
"grad_norm": 0.0011284707579761744,
|
| 926 |
+
"learning_rate": 2.151708058262601e-05,
|
| 927 |
+
"loss": 0.1361,
|
| 928 |
+
"step": 77959
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 48.01,
|
| 932 |
+
"eval_accuracy": 0.6442715700141443,
|
| 933 |
+
"eval_f1": 0.6229754386330917,
|
| 934 |
+
"eval_loss": 2.8814539909362793,
|
| 935 |
+
"eval_precision": 0.6187358231984663,
|
| 936 |
+
"eval_recall": 0.6442715700141443,
|
| 937 |
+
"eval_runtime": 674.3209,
|
| 938 |
+
"eval_samples_per_second": 6.291,
|
| 939 |
+
"eval_steps_per_second": 0.787,
|
| 940 |
+
"step": 77959
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 49.01,
|
| 944 |
+
"grad_norm": 102.50770568847656,
|
| 945 |
+
"learning_rate": 2.0822417828076427e-05,
|
| 946 |
+
"loss": 0.1351,
|
| 947 |
+
"step": 79550
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 49.01,
|
| 951 |
+
"eval_accuracy": 0.6428571428571429,
|
| 952 |
+
"eval_f1": 0.6174587269654099,
|
| 953 |
+
"eval_loss": 3.070270299911499,
|
| 954 |
+
"eval_precision": 0.6243702935522684,
|
| 955 |
+
"eval_recall": 0.6428571428571429,
|
| 956 |
+
"eval_runtime": 669.5589,
|
| 957 |
+
"eval_samples_per_second": 6.336,
|
| 958 |
+
"eval_steps_per_second": 0.793,
|
| 959 |
+
"step": 79550
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 50.01,
|
| 963 |
+
"grad_norm": 8.653002738952637,
|
| 964 |
+
"learning_rate": 2.0127755073526845e-05,
|
| 965 |
+
"loss": 0.1196,
|
| 966 |
+
"step": 81141
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 50.01,
|
| 970 |
+
"eval_accuracy": 0.6423856671381424,
|
| 971 |
+
"eval_f1": 0.6190424793456692,
|
| 972 |
+
"eval_loss": 3.027528762817383,
|
| 973 |
+
"eval_precision": 0.625005878732158,
|
| 974 |
+
"eval_recall": 0.6423856671381424,
|
| 975 |
+
"eval_runtime": 677.2082,
|
| 976 |
+
"eval_samples_per_second": 6.264,
|
| 977 |
+
"eval_steps_per_second": 0.784,
|
| 978 |
+
"step": 81141
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 51.01,
|
| 982 |
+
"grad_norm": 62.25484848022461,
|
| 983 |
+
"learning_rate": 1.943309231897726e-05,
|
| 984 |
+
"loss": 0.111,
|
| 985 |
+
"step": 82732
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 51.01,
|
| 989 |
+
"eval_accuracy": 0.641914191419142,
|
| 990 |
+
"eval_f1": 0.618920231303522,
|
| 991 |
+
"eval_loss": 3.1254563331604004,
|
| 992 |
+
"eval_precision": 0.6280670220001587,
|
| 993 |
+
"eval_recall": 0.641914191419142,
|
| 994 |
+
"eval_runtime": 651.0502,
|
| 995 |
+
"eval_samples_per_second": 6.516,
|
| 996 |
+
"eval_steps_per_second": 0.816,
|
| 997 |
+
"step": 82732
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 52.01,
|
| 1001 |
+
"grad_norm": 18.133270263671875,
|
| 1002 |
+
"learning_rate": 1.8738429564427678e-05,
|
| 1003 |
+
"loss": 0.1119,
|
| 1004 |
+
"step": 84323
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 52.01,
|
| 1008 |
+
"eval_accuracy": 0.6471004243281471,
|
| 1009 |
+
"eval_f1": 0.6214547873156199,
|
| 1010 |
+
"eval_loss": 3.1854350566864014,
|
| 1011 |
+
"eval_precision": 0.629875200916356,
|
| 1012 |
+
"eval_recall": 0.6471004243281471,
|
| 1013 |
+
"eval_runtime": 653.5,
|
| 1014 |
+
"eval_samples_per_second": 6.491,
|
| 1015 |
+
"eval_steps_per_second": 0.813,
|
| 1016 |
+
"step": 84323
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 53.01,
|
| 1020 |
+
"grad_norm": 19.144214630126953,
|
| 1021 |
+
"learning_rate": 1.8043766809878096e-05,
|
| 1022 |
+
"loss": 0.1069,
|
| 1023 |
+
"step": 85914
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 53.01,
|
| 1027 |
+
"eval_accuracy": 0.6383781235266384,
|
| 1028 |
+
"eval_f1": 0.6195011414926834,
|
| 1029 |
+
"eval_loss": 3.2136049270629883,
|
| 1030 |
+
"eval_precision": 0.6251413165179668,
|
| 1031 |
+
"eval_recall": 0.6383781235266384,
|
| 1032 |
+
"eval_runtime": 653.7257,
|
| 1033 |
+
"eval_samples_per_second": 6.489,
|
| 1034 |
+
"eval_steps_per_second": 0.812,
|
| 1035 |
+
"step": 85914
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 54.01,
|
| 1039 |
+
"grad_norm": 0.010962074622511864,
|
| 1040 |
+
"learning_rate": 1.7349104055328515e-05,
|
| 1041 |
+
"loss": 0.093,
|
| 1042 |
+
"step": 87505
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 54.01,
|
| 1046 |
+
"eval_accuracy": 0.6506364922206507,
|
| 1047 |
+
"eval_f1": 0.6154588049677499,
|
| 1048 |
+
"eval_loss": 3.3124778270721436,
|
| 1049 |
+
"eval_precision": 0.614483037031078,
|
| 1050 |
+
"eval_recall": 0.6506364922206507,
|
| 1051 |
+
"eval_runtime": 653.1839,
|
| 1052 |
+
"eval_samples_per_second": 6.494,
|
| 1053 |
+
"eval_steps_per_second": 0.813,
|
| 1054 |
+
"step": 87505
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 55.01,
|
| 1058 |
+
"grad_norm": 4.032207489013672,
|
| 1059 |
+
"learning_rate": 1.665444130077893e-05,
|
| 1060 |
+
"loss": 0.0901,
|
| 1061 |
+
"step": 89096
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 55.01,
|
| 1065 |
+
"eval_accuracy": 0.6383781235266384,
|
| 1066 |
+
"eval_f1": 0.6217233442126047,
|
| 1067 |
+
"eval_loss": 3.3028151988983154,
|
| 1068 |
+
"eval_precision": 0.6277366002975752,
|
| 1069 |
+
"eval_recall": 0.6383781235266384,
|
| 1070 |
+
"eval_runtime": 651.3473,
|
| 1071 |
+
"eval_samples_per_second": 6.513,
|
| 1072 |
+
"eval_steps_per_second": 0.815,
|
| 1073 |
+
"step": 89096
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 56.01,
|
| 1077 |
+
"grad_norm": 0.039573218673467636,
|
| 1078 |
+
"learning_rate": 1.5959778546229348e-05,
|
| 1079 |
+
"loss": 0.0776,
|
| 1080 |
+
"step": 90687
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 56.01,
|
| 1084 |
+
"eval_accuracy": 0.6487505893446488,
|
| 1085 |
+
"eval_f1": 0.6297653212708532,
|
| 1086 |
+
"eval_loss": 3.3314833641052246,
|
| 1087 |
+
"eval_precision": 0.6272429266150353,
|
| 1088 |
+
"eval_recall": 0.6487505893446488,
|
| 1089 |
+
"eval_runtime": 649.3376,
|
| 1090 |
+
"eval_samples_per_second": 6.533,
|
| 1091 |
+
"eval_steps_per_second": 0.818,
|
| 1092 |
+
"step": 90687
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 57.01,
|
| 1096 |
+
"grad_norm": 0.0005812808522023261,
|
| 1097 |
+
"learning_rate": 1.5265115791679766e-05,
|
| 1098 |
+
"loss": 0.0837,
|
| 1099 |
+
"step": 92278
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 57.01,
|
| 1103 |
+
"eval_accuracy": 0.6558227251296558,
|
| 1104 |
+
"eval_f1": 0.6242411319775614,
|
| 1105 |
+
"eval_loss": 3.438481092453003,
|
| 1106 |
+
"eval_precision": 0.6374388645077231,
|
| 1107 |
+
"eval_recall": 0.6558227251296558,
|
| 1108 |
+
"eval_runtime": 658.6971,
|
| 1109 |
+
"eval_samples_per_second": 6.44,
|
| 1110 |
+
"eval_steps_per_second": 0.806,
|
| 1111 |
+
"step": 92278
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 58.01,
|
| 1115 |
+
"grad_norm": 0.04262514412403107,
|
| 1116 |
+
"learning_rate": 1.4570453037130182e-05,
|
| 1117 |
+
"loss": 0.0701,
|
| 1118 |
+
"step": 93869
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 58.01,
|
| 1122 |
+
"eval_accuracy": 0.644035832154644,
|
| 1123 |
+
"eval_f1": 0.6285778416417378,
|
| 1124 |
+
"eval_loss": 3.3799736499786377,
|
| 1125 |
+
"eval_precision": 0.6320916006510833,
|
| 1126 |
+
"eval_recall": 0.644035832154644,
|
| 1127 |
+
"eval_runtime": 669.2439,
|
| 1128 |
+
"eval_samples_per_second": 6.338,
|
| 1129 |
+
"eval_steps_per_second": 0.793,
|
| 1130 |
+
"step": 93869
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 59.01,
|
| 1134 |
+
"grad_norm": 0.001757206628099084,
|
| 1135 |
+
"learning_rate": 1.38757902825806e-05,
|
| 1136 |
+
"loss": 0.0682,
|
| 1137 |
+
"step": 95460
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 59.01,
|
| 1141 |
+
"eval_accuracy": 0.6541725601131542,
|
| 1142 |
+
"eval_f1": 0.6262011153983429,
|
| 1143 |
+
"eval_loss": 3.4473154544830322,
|
| 1144 |
+
"eval_precision": 0.6343707774148433,
|
| 1145 |
+
"eval_recall": 0.6541725601131542,
|
| 1146 |
+
"eval_runtime": 660.2778,
|
| 1147 |
+
"eval_samples_per_second": 6.425,
|
| 1148 |
+
"eval_steps_per_second": 0.804,
|
| 1149 |
+
"step": 95460
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 60.01,
|
| 1153 |
+
"grad_norm": 0.08907134085893631,
|
| 1154 |
+
"learning_rate": 1.3181127528031017e-05,
|
| 1155 |
+
"loss": 0.0763,
|
| 1156 |
+
"step": 97051
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 60.01,
|
| 1160 |
+
"eval_accuracy": 0.6315417256011315,
|
| 1161 |
+
"eval_f1": 0.6148352414634933,
|
| 1162 |
+
"eval_loss": 3.450514316558838,
|
| 1163 |
+
"eval_precision": 0.6148635373339948,
|
| 1164 |
+
"eval_recall": 0.6315417256011315,
|
| 1165 |
+
"eval_runtime": 666.8615,
|
| 1166 |
+
"eval_samples_per_second": 6.361,
|
| 1167 |
+
"eval_steps_per_second": 0.796,
|
| 1168 |
+
"step": 97051
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 61.01,
|
| 1172 |
+
"grad_norm": 0.0005003924597986042,
|
| 1173 |
+
"learning_rate": 1.2486464773481436e-05,
|
| 1174 |
+
"loss": 0.0629,
|
| 1175 |
+
"step": 98642
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 61.01,
|
| 1179 |
+
"eval_accuracy": 0.6504007543611504,
|
| 1180 |
+
"eval_f1": 0.6253235814801552,
|
| 1181 |
+
"eval_loss": 3.440239906311035,
|
| 1182 |
+
"eval_precision": 0.6233462363989224,
|
| 1183 |
+
"eval_recall": 0.6504007543611504,
|
| 1184 |
+
"eval_runtime": 653.614,
|
| 1185 |
+
"eval_samples_per_second": 6.49,
|
| 1186 |
+
"eval_steps_per_second": 0.812,
|
| 1187 |
+
"step": 98642
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 62.01,
|
| 1191 |
+
"grad_norm": 0.06604283303022385,
|
| 1192 |
+
"learning_rate": 1.1791802018931852e-05,
|
| 1193 |
+
"loss": 0.0552,
|
| 1194 |
+
"step": 100233
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 62.01,
|
| 1198 |
+
"eval_accuracy": 0.6537010843941538,
|
| 1199 |
+
"eval_f1": 0.6314537569564033,
|
| 1200 |
+
"eval_loss": 3.4401602745056152,
|
| 1201 |
+
"eval_precision": 0.6324239134320033,
|
| 1202 |
+
"eval_recall": 0.6537010843941538,
|
| 1203 |
+
"eval_runtime": 663.5087,
|
| 1204 |
+
"eval_samples_per_second": 6.393,
|
| 1205 |
+
"eval_steps_per_second": 0.8,
|
| 1206 |
+
"step": 100233
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 63.01,
|
| 1210 |
+
"grad_norm": 0.00693962536752224,
|
| 1211 |
+
"learning_rate": 1.109713926438227e-05,
|
| 1212 |
+
"loss": 0.0463,
|
| 1213 |
+
"step": 101824
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 63.01,
|
| 1217 |
+
"eval_accuracy": 0.6466289486091467,
|
| 1218 |
+
"eval_f1": 0.6216674773439607,
|
| 1219 |
+
"eval_loss": 3.529994010925293,
|
| 1220 |
+
"eval_precision": 0.6217012879573426,
|
| 1221 |
+
"eval_recall": 0.6466289486091467,
|
| 1222 |
+
"eval_runtime": 686.7487,
|
| 1223 |
+
"eval_samples_per_second": 6.177,
|
| 1224 |
+
"eval_steps_per_second": 0.773,
|
| 1225 |
+
"step": 101824
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 64.01,
|
| 1229 |
+
"grad_norm": 0.0005720761837437749,
|
| 1230 |
+
"learning_rate": 1.0402476509832687e-05,
|
| 1231 |
+
"loss": 0.0471,
|
| 1232 |
+
"step": 103415
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 64.01,
|
| 1236 |
+
"eval_accuracy": 0.6511079679396511,
|
| 1237 |
+
"eval_f1": 0.6222590555700158,
|
| 1238 |
+
"eval_loss": 3.6793229579925537,
|
| 1239 |
+
"eval_precision": 0.6346230225230813,
|
| 1240 |
+
"eval_recall": 0.6511079679396511,
|
| 1241 |
+
"eval_runtime": 669.336,
|
| 1242 |
+
"eval_samples_per_second": 6.338,
|
| 1243 |
+
"eval_steps_per_second": 0.793,
|
| 1244 |
+
"step": 103415
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 65.01,
|
| 1248 |
+
"grad_norm": 0.0010839367751032114,
|
| 1249 |
+
"learning_rate": 9.707813755283105e-06,
|
| 1250 |
+
"loss": 0.0448,
|
| 1251 |
+
"step": 105006
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 65.01,
|
| 1255 |
+
"eval_accuracy": 0.6449787835926449,
|
| 1256 |
+
"eval_f1": 0.6169893577308092,
|
| 1257 |
+
"eval_loss": 3.685042142868042,
|
| 1258 |
+
"eval_precision": 0.6265011843748146,
|
| 1259 |
+
"eval_recall": 0.6449787835926449,
|
| 1260 |
+
"eval_runtime": 670.2367,
|
| 1261 |
+
"eval_samples_per_second": 6.329,
|
| 1262 |
+
"eval_steps_per_second": 0.792,
|
| 1263 |
+
"step": 105006
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 66.01,
|
| 1267 |
+
"grad_norm": 0.03215405344963074,
|
| 1268 |
+
"learning_rate": 9.013151000733522e-06,
|
| 1269 |
+
"loss": 0.0362,
|
| 1270 |
+
"step": 106597
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 66.01,
|
| 1274 |
+
"eval_accuracy": 0.6482791136256483,
|
| 1275 |
+
"eval_f1": 0.624213528202086,
|
| 1276 |
+
"eval_loss": 3.658543348312378,
|
| 1277 |
+
"eval_precision": 0.6265115566919579,
|
| 1278 |
+
"eval_recall": 0.6482791136256483,
|
| 1279 |
+
"eval_runtime": 655.7488,
|
| 1280 |
+
"eval_samples_per_second": 6.469,
|
| 1281 |
+
"eval_steps_per_second": 0.81,
|
| 1282 |
+
"step": 106597
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 67.01,
|
| 1286 |
+
"grad_norm": 0.00014659887528978288,
|
| 1287 |
+
"learning_rate": 8.31848824618394e-06,
|
| 1288 |
+
"loss": 0.0419,
|
| 1289 |
+
"step": 108188
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 67.01,
|
| 1293 |
+
"eval_accuracy": 0.6343705799151343,
|
| 1294 |
+
"eval_f1": 0.6168968855551049,
|
| 1295 |
+
"eval_loss": 3.6284878253936768,
|
| 1296 |
+
"eval_precision": 0.6192216504465796,
|
| 1297 |
+
"eval_recall": 0.6343705799151343,
|
| 1298 |
+
"eval_runtime": 667.7711,
|
| 1299 |
+
"eval_samples_per_second": 6.352,
|
| 1300 |
+
"eval_steps_per_second": 0.795,
|
| 1301 |
+
"step": 108188
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 68.01,
|
| 1305 |
+
"grad_norm": 0.0006045685149729252,
|
| 1306 |
+
"learning_rate": 7.6238254916343565e-06,
|
| 1307 |
+
"loss": 0.0309,
|
| 1308 |
+
"step": 109779
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 68.01,
|
| 1312 |
+
"eval_accuracy": 0.648986327204149,
|
| 1313 |
+
"eval_f1": 0.6269316812634602,
|
| 1314 |
+
"eval_loss": 3.665743112564087,
|
| 1315 |
+
"eval_precision": 0.6264133629902356,
|
| 1316 |
+
"eval_recall": 0.648986327204149,
|
| 1317 |
+
"eval_runtime": 670.5591,
|
| 1318 |
+
"eval_samples_per_second": 6.326,
|
| 1319 |
+
"eval_steps_per_second": 0.792,
|
| 1320 |
+
"step": 109779
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 69.01,
|
| 1324 |
+
"grad_norm": 0.0056765577755868435,
|
| 1325 |
+
"learning_rate": 6.929162737084774e-06,
|
| 1326 |
+
"loss": 0.0312,
|
| 1327 |
+
"step": 111370
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 69.01,
|
| 1331 |
+
"eval_accuracy": 0.6416784535596417,
|
| 1332 |
+
"eval_f1": 0.6205254067422245,
|
| 1333 |
+
"eval_loss": 3.7122817039489746,
|
| 1334 |
+
"eval_precision": 0.6239256088005212,
|
| 1335 |
+
"eval_recall": 0.6416784535596417,
|
| 1336 |
+
"eval_runtime": 661.8739,
|
| 1337 |
+
"eval_samples_per_second": 6.409,
|
| 1338 |
+
"eval_steps_per_second": 0.802,
|
| 1339 |
+
"step": 111370
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 70.01,
|
| 1343 |
+
"grad_norm": 0.0024060504510998726,
|
| 1344 |
+
"learning_rate": 6.234499982535192e-06,
|
| 1345 |
+
"loss": 0.0315,
|
| 1346 |
+
"step": 112961
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 70.01,
|
| 1350 |
+
"eval_accuracy": 0.648986327204149,
|
| 1351 |
+
"eval_f1": 0.6189187635444573,
|
| 1352 |
+
"eval_loss": 3.753802537918091,
|
| 1353 |
+
"eval_precision": 0.6224224422132874,
|
| 1354 |
+
"eval_recall": 0.648986327204149,
|
| 1355 |
+
"eval_runtime": 664.7996,
|
| 1356 |
+
"eval_samples_per_second": 6.381,
|
| 1357 |
+
"eval_steps_per_second": 0.799,
|
| 1358 |
+
"step": 112961
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 71.01,
|
| 1362 |
+
"grad_norm": 0.047281160950660706,
|
| 1363 |
+
"learning_rate": 5.53983722798561e-06,
|
| 1364 |
+
"loss": 0.0294,
|
| 1365 |
+
"step": 114552
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 71.01,
|
| 1369 |
+
"eval_accuracy": 0.6482791136256483,
|
| 1370 |
+
"eval_f1": 0.6236630466746416,
|
| 1371 |
+
"eval_loss": 3.706387519836426,
|
| 1372 |
+
"eval_precision": 0.6234001348683471,
|
| 1373 |
+
"eval_recall": 0.6482791136256483,
|
| 1374 |
+
"eval_runtime": 677.1081,
|
| 1375 |
+
"eval_samples_per_second": 6.265,
|
| 1376 |
+
"eval_steps_per_second": 0.784,
|
| 1377 |
+
"step": 114552
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 72.01,
|
| 1381 |
+
"grad_norm": 0.00014872274186927825,
|
| 1382 |
+
"learning_rate": 4.845174473436027e-06,
|
| 1383 |
+
"loss": 0.0282,
|
| 1384 |
+
"step": 116143
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 72.01,
|
| 1388 |
+
"eval_accuracy": 0.6428571428571429,
|
| 1389 |
+
"eval_f1": 0.6192426747671007,
|
| 1390 |
+
"eval_loss": 3.7945356369018555,
|
| 1391 |
+
"eval_precision": 0.624713412639683,
|
| 1392 |
+
"eval_recall": 0.6428571428571429,
|
| 1393 |
+
"eval_runtime": 677.029,
|
| 1394 |
+
"eval_samples_per_second": 6.266,
|
| 1395 |
+
"eval_steps_per_second": 0.784,
|
| 1396 |
+
"step": 116143
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 73.01,
|
| 1400 |
+
"grad_norm": 0.0006783913122490048,
|
| 1401 |
+
"learning_rate": 4.150511718886444e-06,
|
| 1402 |
+
"loss": 0.0275,
|
| 1403 |
+
"step": 117734
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 73.01,
|
| 1407 |
+
"eval_accuracy": 0.6527581329561527,
|
| 1408 |
+
"eval_f1": 0.6271859964586763,
|
| 1409 |
+
"eval_loss": 3.7549855709075928,
|
| 1410 |
+
"eval_precision": 0.6297463780952456,
|
| 1411 |
+
"eval_recall": 0.6527581329561527,
|
| 1412 |
+
"eval_runtime": 694.5725,
|
| 1413 |
+
"eval_samples_per_second": 6.107,
|
| 1414 |
+
"eval_steps_per_second": 0.764,
|
| 1415 |
+
"step": 117734
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 74.01,
|
| 1419 |
+
"grad_norm": 0.10241026431322098,
|
| 1420 |
+
"learning_rate": 3.4558489643368614e-06,
|
| 1421 |
+
"loss": 0.0319,
|
| 1422 |
+
"step": 119325
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 74.01,
|
| 1426 |
+
"eval_accuracy": 0.6508722300801508,
|
| 1427 |
+
"eval_f1": 0.6233733427088337,
|
| 1428 |
+
"eval_loss": 3.740715980529785,
|
| 1429 |
+
"eval_precision": 0.628877991091339,
|
| 1430 |
+
"eval_recall": 0.6508722300801508,
|
| 1431 |
+
"eval_runtime": 654.6009,
|
| 1432 |
+
"eval_samples_per_second": 6.48,
|
| 1433 |
+
"eval_steps_per_second": 0.811,
|
| 1434 |
+
"step": 119325
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 75.01,
|
| 1438 |
+
"grad_norm": 0.00016077565669547766,
|
| 1439 |
+
"learning_rate": 2.7611862097872788e-06,
|
| 1440 |
+
"loss": 0.021,
|
| 1441 |
+
"step": 120916
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 75.01,
|
| 1445 |
+
"eval_accuracy": 0.6532296086751532,
|
| 1446 |
+
"eval_f1": 0.6269729664014984,
|
| 1447 |
+
"eval_loss": 3.75272536277771,
|
| 1448 |
+
"eval_precision": 0.6290218177795819,
|
| 1449 |
+
"eval_recall": 0.6532296086751532,
|
| 1450 |
+
"eval_runtime": 658.8706,
|
| 1451 |
+
"eval_samples_per_second": 6.438,
|
| 1452 |
+
"eval_steps_per_second": 0.806,
|
| 1453 |
+
"step": 120916
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 76.01,
|
| 1457 |
+
"grad_norm": 0.0008236940484493971,
|
| 1458 |
+
"learning_rate": 2.066523455237696e-06,
|
| 1459 |
+
"loss": 0.0159,
|
| 1460 |
+
"step": 122507
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 76.01,
|
| 1464 |
+
"eval_accuracy": 0.6515794436586516,
|
| 1465 |
+
"eval_f1": 0.6242553810002985,
|
| 1466 |
+
"eval_loss": 3.7779977321624756,
|
| 1467 |
+
"eval_precision": 0.6240515257248215,
|
| 1468 |
+
"eval_recall": 0.6515794436586516,
|
| 1469 |
+
"eval_runtime": 660.4492,
|
| 1470 |
+
"eval_samples_per_second": 6.423,
|
| 1471 |
+
"eval_steps_per_second": 0.804,
|
| 1472 |
+
"step": 122507
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 77.01,
|
| 1476 |
+
"grad_norm": 0.004408856853842735,
|
| 1477 |
+
"learning_rate": 1.3718607006881136e-06,
|
| 1478 |
+
"loss": 0.0133,
|
| 1479 |
+
"step": 124098
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 77.01,
|
| 1483 |
+
"eval_accuracy": 0.6499292786421499,
|
| 1484 |
+
"eval_f1": 0.624040880732074,
|
| 1485 |
+
"eval_loss": 3.7923333644866943,
|
| 1486 |
+
"eval_precision": 0.6271592552161068,
|
| 1487 |
+
"eval_recall": 0.6499292786421499,
|
| 1488 |
+
"eval_runtime": 666.8007,
|
| 1489 |
+
"eval_samples_per_second": 6.362,
|
| 1490 |
+
"eval_steps_per_second": 0.796,
|
| 1491 |
+
"step": 124098
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 78.01,
|
| 1495 |
+
"grad_norm": 0.0014601564034819603,
|
| 1496 |
+
"learning_rate": 6.771979461385309e-07,
|
| 1497 |
+
"loss": 0.0125,
|
| 1498 |
+
"step": 125689
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 78.01,
|
| 1502 |
+
"eval_accuracy": 0.6504007543611504,
|
| 1503 |
+
"eval_f1": 0.6216504935099332,
|
| 1504 |
+
"eval_loss": 3.8070006370544434,
|
| 1505 |
+
"eval_precision": 0.6262856082095383,
|
| 1506 |
+
"eval_recall": 0.6504007543611504,
|
| 1507 |
+
"eval_runtime": 667.8654,
|
| 1508 |
+
"eval_samples_per_second": 6.352,
|
| 1509 |
+
"eval_steps_per_second": 0.795,
|
| 1510 |
+
"step": 125689
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 79.01,
|
| 1514 |
+
"grad_norm": 0.0005495115183293819,
|
| 1515 |
+
"learning_rate": 0.0,
|
| 1516 |
+
"loss": 0.0132,
|
| 1517 |
+
"step": 127240
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 79.01,
|
| 1521 |
+
"eval_accuracy": 0.6506364922206507,
|
| 1522 |
+
"eval_f1": 0.6225468142229464,
|
| 1523 |
+
"eval_loss": 3.796358108520508,
|
| 1524 |
+
"eval_precision": 0.6264178866401664,
|
| 1525 |
+
"eval_recall": 0.6506364922206507,
|
| 1526 |
+
"eval_runtime": 698.0179,
|
| 1527 |
+
"eval_samples_per_second": 6.077,
|
| 1528 |
+
"eval_steps_per_second": 0.761,
|
| 1529 |
+
"step": 127240
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 79.01,
|
| 1533 |
+
"step": 127240,
|
| 1534 |
+
"total_flos": 1.2682038581932563e+21,
|
| 1535 |
+
"train_loss": 0.41676221998635044,
|
| 1536 |
+
"train_runtime": 266875.9954,
|
| 1537 |
+
"train_samples_per_second": 3.814,
|
| 1538 |
+
"train_steps_per_second": 0.477
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 79.01,
|
| 1542 |
+
"eval_accuracy": 0.6697312588401697,
|
| 1543 |
+
"eval_f1": 0.635434606603408,
|
| 1544 |
+
"eval_loss": 1.1357100009918213,
|
| 1545 |
+
"eval_precision": 0.642912037609069,
|
| 1546 |
+
"eval_recall": 0.6697312588401697,
|
| 1547 |
+
"eval_runtime": 859.5426,
|
| 1548 |
+
"eval_samples_per_second": 4.935,
|
| 1549 |
+
"eval_steps_per_second": 0.618,
|
| 1550 |
+
"step": 127240
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 79.01,
|
| 1554 |
+
"eval_accuracy": 0.6697312588401697,
|
| 1555 |
+
"eval_f1": 0.635434606603408,
|
| 1556 |
+
"eval_loss": 1.1357100009918213,
|
| 1557 |
+
"eval_precision": 0.642912037609069,
|
| 1558 |
+
"eval_recall": 0.6697312588401697,
|
| 1559 |
+
"eval_runtime": 860.1584,
|
| 1560 |
+
"eval_samples_per_second": 4.932,
|
| 1561 |
+
"eval_steps_per_second": 0.617,
|
| 1562 |
+
"step": 127240
|
| 1563 |
+
}
|
| 1564 |
+
],
|
| 1565 |
+
"logging_steps": 10,
|
| 1566 |
+
"max_steps": 127240,
|
| 1567 |
+
"num_input_tokens_seen": 0,
|
| 1568 |
+
"num_train_epochs": 9223372036854775807,
|
| 1569 |
+
"save_steps": 500,
|
| 1570 |
+
"total_flos": 1.2682038581932563e+21,
|
| 1571 |
+
"train_batch_size": 8,
|
| 1572 |
+
"trial_name": null,
|
| 1573 |
+
"trial_params": null
|
| 1574 |
+
}
|