Commit
•
2ac0073
1
Parent(s):
7348b28
End of training
Browse files- all_results.json +12 -6
- eval_results.json +7 -6
- train_results.json +8 -0
- trainer_state.json +1225 -0
all_results.json
CHANGED
|
@@ -1,9 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
|
|
|
| 5 |
"eval_samples": 100,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"perplexity":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.03550813008130081,
|
| 4 |
+
"eval_loss": 3.548828125,
|
| 5 |
+
"eval_runtime": 2.5024,
|
| 6 |
"eval_samples": 100,
|
| 7 |
+
"eval_samples_per_second": 39.962,
|
| 8 |
+
"eval_steps_per_second": 0.799,
|
| 9 |
+
"perplexity": 34.77254452624438,
|
| 10 |
+
"train_loss": 1.6890087127685547,
|
| 11 |
+
"train_runtime": 406.6939,
|
| 12 |
+
"train_samples": 1000,
|
| 13 |
+
"train_samples_per_second": 12.294,
|
| 14 |
+
"train_steps_per_second": 0.197
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,9 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"
|
| 3 |
-
"
|
| 4 |
-
"
|
|
|
|
| 5 |
"eval_samples": 100,
|
| 6 |
-
"eval_samples_per_second":
|
| 7 |
-
"eval_steps_per_second": 0.
|
| 8 |
-
"perplexity":
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.03550813008130081,
|
| 4 |
+
"eval_loss": 3.548828125,
|
| 5 |
+
"eval_runtime": 2.5024,
|
| 6 |
"eval_samples": 100,
|
| 7 |
+
"eval_samples_per_second": 39.962,
|
| 8 |
+
"eval_steps_per_second": 0.799,
|
| 9 |
+
"perplexity": 34.77254452624438
|
| 10 |
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 1.6890087127685547,
|
| 4 |
+
"train_runtime": 406.6939,
|
| 5 |
+
"train_samples": 1000,
|
| 6 |
+
"train_samples_per_second": 12.294,
|
| 7 |
+
"train_steps_per_second": 0.197
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1225 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 80,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 0,
|
| 13 |
+
"loss": 3.2024,
|
| 14 |
+
"step": 1
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"eval_accuracy": 0.038495934959349595,
|
| 19 |
+
"eval_loss": 2.912109375,
|
| 20 |
+
"eval_runtime": 2.0553,
|
| 21 |
+
"eval_samples_per_second": 48.656,
|
| 22 |
+
"eval_steps_per_second": 0.973,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.12,
|
| 27 |
+
"learning_rate": 0,
|
| 28 |
+
"loss": 3.1226,
|
| 29 |
+
"step": 2
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.12,
|
| 33 |
+
"eval_accuracy": 0.038495934959349595,
|
| 34 |
+
"eval_loss": 2.912109375,
|
| 35 |
+
"eval_runtime": 2.5287,
|
| 36 |
+
"eval_samples_per_second": 39.546,
|
| 37 |
+
"eval_steps_per_second": 0.791,
|
| 38 |
+
"step": 2
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.19,
|
| 42 |
+
"learning_rate": 0.0,
|
| 43 |
+
"loss": 3.1321,
|
| 44 |
+
"step": 3
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.19,
|
| 48 |
+
"eval_accuracy": 0.03936991869918699,
|
| 49 |
+
"eval_loss": 2.84765625,
|
| 50 |
+
"eval_runtime": 2.4969,
|
| 51 |
+
"eval_samples_per_second": 40.05,
|
| 52 |
+
"eval_steps_per_second": 0.801,
|
| 53 |
+
"step": 3
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.25,
|
| 57 |
+
"learning_rate": 5e-05,
|
| 58 |
+
"loss": 2.9875,
|
| 59 |
+
"step": 4
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.25,
|
| 63 |
+
"eval_accuracy": 0.03936991869918699,
|
| 64 |
+
"eval_loss": 2.84765625,
|
| 65 |
+
"eval_runtime": 2.5942,
|
| 66 |
+
"eval_samples_per_second": 38.548,
|
| 67 |
+
"eval_steps_per_second": 0.771,
|
| 68 |
+
"step": 4
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.31,
|
| 72 |
+
"learning_rate": 5e-05,
|
| 73 |
+
"loss": 2.9717,
|
| 74 |
+
"step": 5
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.31,
|
| 78 |
+
"eval_accuracy": 0.03910569105691057,
|
| 79 |
+
"eval_loss": 2.85546875,
|
| 80 |
+
"eval_runtime": 2.4958,
|
| 81 |
+
"eval_samples_per_second": 40.068,
|
| 82 |
+
"eval_steps_per_second": 0.801,
|
| 83 |
+
"step": 5
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.38,
|
| 87 |
+
"learning_rate": 5e-05,
|
| 88 |
+
"loss": 2.9341,
|
| 89 |
+
"step": 6
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.38,
|
| 93 |
+
"eval_accuracy": 0.03920731707317073,
|
| 94 |
+
"eval_loss": 2.84375,
|
| 95 |
+
"eval_runtime": 2.6044,
|
| 96 |
+
"eval_samples_per_second": 38.396,
|
| 97 |
+
"eval_steps_per_second": 0.768,
|
| 98 |
+
"step": 6
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.44,
|
| 102 |
+
"learning_rate": 5e-05,
|
| 103 |
+
"loss": 3.0376,
|
| 104 |
+
"step": 7
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.44,
|
| 108 |
+
"eval_accuracy": 0.03957317073170732,
|
| 109 |
+
"eval_loss": 2.818359375,
|
| 110 |
+
"eval_runtime": 2.2607,
|
| 111 |
+
"eval_samples_per_second": 44.235,
|
| 112 |
+
"eval_steps_per_second": 0.885,
|
| 113 |
+
"step": 7
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.5,
|
| 117 |
+
"learning_rate": 5e-05,
|
| 118 |
+
"loss": 2.8164,
|
| 119 |
+
"step": 8
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.5,
|
| 123 |
+
"eval_accuracy": 0.039532520325203255,
|
| 124 |
+
"eval_loss": 2.798828125,
|
| 125 |
+
"eval_runtime": 2.0579,
|
| 126 |
+
"eval_samples_per_second": 48.594,
|
| 127 |
+
"eval_steps_per_second": 0.972,
|
| 128 |
+
"step": 8
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.56,
|
| 132 |
+
"learning_rate": 5e-05,
|
| 133 |
+
"loss": 3.0857,
|
| 134 |
+
"step": 9
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.56,
|
| 138 |
+
"eval_accuracy": 0.03936991869918699,
|
| 139 |
+
"eval_loss": 2.798828125,
|
| 140 |
+
"eval_runtime": 2.0436,
|
| 141 |
+
"eval_samples_per_second": 48.933,
|
| 142 |
+
"eval_steps_per_second": 0.979,
|
| 143 |
+
"step": 9
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.62,
|
| 147 |
+
"learning_rate": 5e-05,
|
| 148 |
+
"loss": 2.9492,
|
| 149 |
+
"step": 10
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.62,
|
| 153 |
+
"eval_accuracy": 0.03947154471544716,
|
| 154 |
+
"eval_loss": 2.796875,
|
| 155 |
+
"eval_runtime": 2.6006,
|
| 156 |
+
"eval_samples_per_second": 38.452,
|
| 157 |
+
"eval_steps_per_second": 0.769,
|
| 158 |
+
"step": 10
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.69,
|
| 162 |
+
"learning_rate": 5e-05,
|
| 163 |
+
"loss": 2.8633,
|
| 164 |
+
"step": 11
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.69,
|
| 168 |
+
"eval_accuracy": 0.03947154471544716,
|
| 169 |
+
"eval_loss": 2.796875,
|
| 170 |
+
"eval_runtime": 2.0146,
|
| 171 |
+
"eval_samples_per_second": 49.639,
|
| 172 |
+
"eval_steps_per_second": 0.993,
|
| 173 |
+
"step": 11
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.75,
|
| 177 |
+
"learning_rate": 5e-05,
|
| 178 |
+
"loss": 2.8994,
|
| 179 |
+
"step": 12
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.75,
|
| 183 |
+
"eval_accuracy": 0.03979674796747967,
|
| 184 |
+
"eval_loss": 2.791015625,
|
| 185 |
+
"eval_runtime": 1.8459,
|
| 186 |
+
"eval_samples_per_second": 54.174,
|
| 187 |
+
"eval_steps_per_second": 1.083,
|
| 188 |
+
"step": 12
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.81,
|
| 192 |
+
"learning_rate": 5e-05,
|
| 193 |
+
"loss": 3.0024,
|
| 194 |
+
"step": 13
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.81,
|
| 198 |
+
"eval_accuracy": 0.040121951219512196,
|
| 199 |
+
"eval_loss": 2.78125,
|
| 200 |
+
"eval_runtime": 2.5067,
|
| 201 |
+
"eval_samples_per_second": 39.892,
|
| 202 |
+
"eval_steps_per_second": 0.798,
|
| 203 |
+
"step": 13
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.88,
|
| 207 |
+
"learning_rate": 5e-05,
|
| 208 |
+
"loss": 2.937,
|
| 209 |
+
"step": 14
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.88,
|
| 213 |
+
"eval_accuracy": 0.03991869918699187,
|
| 214 |
+
"eval_loss": 2.78125,
|
| 215 |
+
"eval_runtime": 2.3271,
|
| 216 |
+
"eval_samples_per_second": 42.972,
|
| 217 |
+
"eval_steps_per_second": 0.859,
|
| 218 |
+
"step": 14
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.94,
|
| 222 |
+
"learning_rate": 5e-05,
|
| 223 |
+
"loss": 2.9963,
|
| 224 |
+
"step": 15
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.94,
|
| 228 |
+
"eval_accuracy": 0.039878048780487806,
|
| 229 |
+
"eval_loss": 2.78125,
|
| 230 |
+
"eval_runtime": 1.8105,
|
| 231 |
+
"eval_samples_per_second": 55.232,
|
| 232 |
+
"eval_steps_per_second": 1.105,
|
| 233 |
+
"step": 15
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.0,
|
| 237 |
+
"learning_rate": 5e-05,
|
| 238 |
+
"loss": 3.0168,
|
| 239 |
+
"step": 16
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.0,
|
| 243 |
+
"eval_accuracy": 0.04,
|
| 244 |
+
"eval_loss": 2.775390625,
|
| 245 |
+
"eval_runtime": 2.6001,
|
| 246 |
+
"eval_samples_per_second": 38.46,
|
| 247 |
+
"eval_steps_per_second": 0.769,
|
| 248 |
+
"step": 16
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.06,
|
| 252 |
+
"learning_rate": 5e-05,
|
| 253 |
+
"loss": 2.2589,
|
| 254 |
+
"step": 17
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.06,
|
| 258 |
+
"eval_accuracy": 0.03971544715447155,
|
| 259 |
+
"eval_loss": 2.771484375,
|
| 260 |
+
"eval_runtime": 2.0378,
|
| 261 |
+
"eval_samples_per_second": 49.072,
|
| 262 |
+
"eval_steps_per_second": 0.981,
|
| 263 |
+
"step": 17
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.12,
|
| 267 |
+
"learning_rate": 5e-05,
|
| 268 |
+
"loss": 2.2568,
|
| 269 |
+
"step": 18
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.12,
|
| 273 |
+
"eval_accuracy": 0.03951219512195122,
|
| 274 |
+
"eval_loss": 2.779296875,
|
| 275 |
+
"eval_runtime": 2.5986,
|
| 276 |
+
"eval_samples_per_second": 38.482,
|
| 277 |
+
"eval_steps_per_second": 0.77,
|
| 278 |
+
"step": 18
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.19,
|
| 282 |
+
"learning_rate": 5e-05,
|
| 283 |
+
"loss": 2.3138,
|
| 284 |
+
"step": 19
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.19,
|
| 288 |
+
"eval_accuracy": 0.03930894308943089,
|
| 289 |
+
"eval_loss": 2.802734375,
|
| 290 |
+
"eval_runtime": 2.2441,
|
| 291 |
+
"eval_samples_per_second": 44.561,
|
| 292 |
+
"eval_steps_per_second": 0.891,
|
| 293 |
+
"step": 19
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.25,
|
| 297 |
+
"learning_rate": 5e-05,
|
| 298 |
+
"loss": 2.2759,
|
| 299 |
+
"step": 20
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.25,
|
| 303 |
+
"eval_accuracy": 0.039288617886178864,
|
| 304 |
+
"eval_loss": 2.818359375,
|
| 305 |
+
"eval_runtime": 2.3006,
|
| 306 |
+
"eval_samples_per_second": 43.467,
|
| 307 |
+
"eval_steps_per_second": 0.869,
|
| 308 |
+
"step": 20
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.31,
|
| 312 |
+
"learning_rate": 5e-05,
|
| 313 |
+
"loss": 2.5137,
|
| 314 |
+
"step": 21
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.31,
|
| 318 |
+
"eval_accuracy": 0.039044715447154474,
|
| 319 |
+
"eval_loss": 2.826171875,
|
| 320 |
+
"eval_runtime": 2.6018,
|
| 321 |
+
"eval_samples_per_second": 38.434,
|
| 322 |
+
"eval_steps_per_second": 0.769,
|
| 323 |
+
"step": 21
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.38,
|
| 327 |
+
"learning_rate": 5e-05,
|
| 328 |
+
"loss": 2.2997,
|
| 329 |
+
"step": 22
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.38,
|
| 333 |
+
"eval_accuracy": 0.03878048780487805,
|
| 334 |
+
"eval_loss": 2.83203125,
|
| 335 |
+
"eval_runtime": 2.6159,
|
| 336 |
+
"eval_samples_per_second": 38.228,
|
| 337 |
+
"eval_steps_per_second": 0.765,
|
| 338 |
+
"step": 22
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.44,
|
| 342 |
+
"learning_rate": 5e-05,
|
| 343 |
+
"loss": 2.2693,
|
| 344 |
+
"step": 23
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.44,
|
| 348 |
+
"eval_accuracy": 0.03916666666666667,
|
| 349 |
+
"eval_loss": 2.8359375,
|
| 350 |
+
"eval_runtime": 2.0432,
|
| 351 |
+
"eval_samples_per_second": 48.943,
|
| 352 |
+
"eval_steps_per_second": 0.979,
|
| 353 |
+
"step": 23
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.5,
|
| 357 |
+
"learning_rate": 5e-05,
|
| 358 |
+
"loss": 2.204,
|
| 359 |
+
"step": 24
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.5,
|
| 363 |
+
"eval_accuracy": 0.038739837398373986,
|
| 364 |
+
"eval_loss": 2.837890625,
|
| 365 |
+
"eval_runtime": 2.2494,
|
| 366 |
+
"eval_samples_per_second": 44.456,
|
| 367 |
+
"eval_steps_per_second": 0.889,
|
| 368 |
+
"step": 24
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.56,
|
| 372 |
+
"learning_rate": 5e-05,
|
| 373 |
+
"loss": 2.3713,
|
| 374 |
+
"step": 25
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.56,
|
| 378 |
+
"eval_accuracy": 0.039146341463414634,
|
| 379 |
+
"eval_loss": 2.8359375,
|
| 380 |
+
"eval_runtime": 2.0551,
|
| 381 |
+
"eval_samples_per_second": 48.659,
|
| 382 |
+
"eval_steps_per_second": 0.973,
|
| 383 |
+
"step": 25
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.62,
|
| 387 |
+
"learning_rate": 5e-05,
|
| 388 |
+
"loss": 2.3448,
|
| 389 |
+
"step": 26
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.62,
|
| 393 |
+
"eval_accuracy": 0.03910569105691057,
|
| 394 |
+
"eval_loss": 2.833984375,
|
| 395 |
+
"eval_runtime": 2.6076,
|
| 396 |
+
"eval_samples_per_second": 38.349,
|
| 397 |
+
"eval_steps_per_second": 0.767,
|
| 398 |
+
"step": 26
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.69,
|
| 402 |
+
"learning_rate": 5e-05,
|
| 403 |
+
"loss": 2.217,
|
| 404 |
+
"step": 27
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.69,
|
| 408 |
+
"eval_accuracy": 0.03910569105691057,
|
| 409 |
+
"eval_loss": 2.8359375,
|
| 410 |
+
"eval_runtime": 1.8526,
|
| 411 |
+
"eval_samples_per_second": 53.979,
|
| 412 |
+
"eval_steps_per_second": 1.08,
|
| 413 |
+
"step": 27
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.75,
|
| 417 |
+
"learning_rate": 5e-05,
|
| 418 |
+
"loss": 2.3082,
|
| 419 |
+
"step": 28
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 1.75,
|
| 423 |
+
"eval_accuracy": 0.038455284552845526,
|
| 424 |
+
"eval_loss": 2.837890625,
|
| 425 |
+
"eval_runtime": 2.2452,
|
| 426 |
+
"eval_samples_per_second": 44.54,
|
| 427 |
+
"eval_steps_per_second": 0.891,
|
| 428 |
+
"step": 28
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.81,
|
| 432 |
+
"learning_rate": 5e-05,
|
| 433 |
+
"loss": 2.2878,
|
| 434 |
+
"step": 29
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.81,
|
| 438 |
+
"eval_accuracy": 0.03855691056910569,
|
| 439 |
+
"eval_loss": 2.837890625,
|
| 440 |
+
"eval_runtime": 2.2035,
|
| 441 |
+
"eval_samples_per_second": 45.381,
|
| 442 |
+
"eval_steps_per_second": 0.908,
|
| 443 |
+
"step": 29
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.88,
|
| 447 |
+
"learning_rate": 5e-05,
|
| 448 |
+
"loss": 2.2429,
|
| 449 |
+
"step": 30
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.88,
|
| 453 |
+
"eval_accuracy": 0.03847560975609756,
|
| 454 |
+
"eval_loss": 2.837890625,
|
| 455 |
+
"eval_runtime": 2.0495,
|
| 456 |
+
"eval_samples_per_second": 48.792,
|
| 457 |
+
"eval_steps_per_second": 0.976,
|
| 458 |
+
"step": 30
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.94,
|
| 462 |
+
"learning_rate": 5e-05,
|
| 463 |
+
"loss": 2.2838,
|
| 464 |
+
"step": 31
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.94,
|
| 468 |
+
"eval_accuracy": 0.03851626016260162,
|
| 469 |
+
"eval_loss": 2.8359375,
|
| 470 |
+
"eval_runtime": 2.5995,
|
| 471 |
+
"eval_samples_per_second": 38.469,
|
| 472 |
+
"eval_steps_per_second": 0.769,
|
| 473 |
+
"step": 31
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 2.0,
|
| 477 |
+
"learning_rate": 5e-05,
|
| 478 |
+
"loss": 2.4038,
|
| 479 |
+
"step": 32
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 2.0,
|
| 483 |
+
"eval_accuracy": 0.038739837398373986,
|
| 484 |
+
"eval_loss": 2.837890625,
|
| 485 |
+
"eval_runtime": 2.6048,
|
| 486 |
+
"eval_samples_per_second": 38.391,
|
| 487 |
+
"eval_steps_per_second": 0.768,
|
| 488 |
+
"step": 32
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.06,
|
| 492 |
+
"learning_rate": 5e-05,
|
| 493 |
+
"loss": 1.8481,
|
| 494 |
+
"step": 33
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.06,
|
| 498 |
+
"eval_accuracy": 0.03841463414634146,
|
| 499 |
+
"eval_loss": 2.85546875,
|
| 500 |
+
"eval_runtime": 2.6153,
|
| 501 |
+
"eval_samples_per_second": 38.237,
|
| 502 |
+
"eval_steps_per_second": 0.765,
|
| 503 |
+
"step": 33
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 2.12,
|
| 507 |
+
"learning_rate": 5e-05,
|
| 508 |
+
"loss": 1.657,
|
| 509 |
+
"step": 34
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 2.12,
|
| 513 |
+
"eval_accuracy": 0.038211382113821135,
|
| 514 |
+
"eval_loss": 2.896484375,
|
| 515 |
+
"eval_runtime": 2.1039,
|
| 516 |
+
"eval_samples_per_second": 47.53,
|
| 517 |
+
"eval_steps_per_second": 0.951,
|
| 518 |
+
"step": 34
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.19,
|
| 522 |
+
"learning_rate": 5e-05,
|
| 523 |
+
"loss": 1.6996,
|
| 524 |
+
"step": 35
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.19,
|
| 528 |
+
"eval_accuracy": 0.03804878048780488,
|
| 529 |
+
"eval_loss": 2.958984375,
|
| 530 |
+
"eval_runtime": 2.051,
|
| 531 |
+
"eval_samples_per_second": 48.757,
|
| 532 |
+
"eval_steps_per_second": 0.975,
|
| 533 |
+
"step": 35
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 2.25,
|
| 537 |
+
"learning_rate": 5e-05,
|
| 538 |
+
"loss": 1.6741,
|
| 539 |
+
"step": 36
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 2.25,
|
| 543 |
+
"eval_accuracy": 0.037865853658536584,
|
| 544 |
+
"eval_loss": 3.03125,
|
| 545 |
+
"eval_runtime": 2.6053,
|
| 546 |
+
"eval_samples_per_second": 38.383,
|
| 547 |
+
"eval_steps_per_second": 0.768,
|
| 548 |
+
"step": 36
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.31,
|
| 552 |
+
"learning_rate": 5e-05,
|
| 553 |
+
"loss": 1.594,
|
| 554 |
+
"step": 37
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.31,
|
| 558 |
+
"eval_accuracy": 0.037967479674796745,
|
| 559 |
+
"eval_loss": 3.041015625,
|
| 560 |
+
"eval_runtime": 2.5975,
|
| 561 |
+
"eval_samples_per_second": 38.498,
|
| 562 |
+
"eval_steps_per_second": 0.77,
|
| 563 |
+
"step": 37
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 2.38,
|
| 567 |
+
"learning_rate": 5e-05,
|
| 568 |
+
"loss": 1.5201,
|
| 569 |
+
"step": 38
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 2.38,
|
| 573 |
+
"eval_accuracy": 0.038109756097560975,
|
| 574 |
+
"eval_loss": 3.015625,
|
| 575 |
+
"eval_runtime": 2.6045,
|
| 576 |
+
"eval_samples_per_second": 38.396,
|
| 577 |
+
"eval_steps_per_second": 0.768,
|
| 578 |
+
"step": 38
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.44,
|
| 582 |
+
"learning_rate": 5e-05,
|
| 583 |
+
"loss": 1.5149,
|
| 584 |
+
"step": 39
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.44,
|
| 588 |
+
"eval_accuracy": 0.03798780487804878,
|
| 589 |
+
"eval_loss": 3.013671875,
|
| 590 |
+
"eval_runtime": 2.3613,
|
| 591 |
+
"eval_samples_per_second": 42.349,
|
| 592 |
+
"eval_steps_per_second": 0.847,
|
| 593 |
+
"step": 39
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 2.5,
|
| 597 |
+
"learning_rate": 5e-05,
|
| 598 |
+
"loss": 1.5521,
|
| 599 |
+
"step": 40
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 2.5,
|
| 603 |
+
"eval_accuracy": 0.03788617886178862,
|
| 604 |
+
"eval_loss": 3.017578125,
|
| 605 |
+
"eval_runtime": 2.601,
|
| 606 |
+
"eval_samples_per_second": 38.446,
|
| 607 |
+
"eval_steps_per_second": 0.769,
|
| 608 |
+
"step": 40
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.56,
|
| 612 |
+
"learning_rate": 5e-05,
|
| 613 |
+
"loss": 1.5364,
|
| 614 |
+
"step": 41
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.56,
|
| 618 |
+
"eval_accuracy": 0.03780487804878049,
|
| 619 |
+
"eval_loss": 3.02734375,
|
| 620 |
+
"eval_runtime": 2.6044,
|
| 621 |
+
"eval_samples_per_second": 38.397,
|
| 622 |
+
"eval_steps_per_second": 0.768,
|
| 623 |
+
"step": 41
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 2.62,
|
| 627 |
+
"learning_rate": 5e-05,
|
| 628 |
+
"loss": 1.5385,
|
| 629 |
+
"step": 42
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 2.62,
|
| 633 |
+
"eval_accuracy": 0.037967479674796745,
|
| 634 |
+
"eval_loss": 3.0390625,
|
| 635 |
+
"eval_runtime": 2.5064,
|
| 636 |
+
"eval_samples_per_second": 39.899,
|
| 637 |
+
"eval_steps_per_second": 0.798,
|
| 638 |
+
"step": 42
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 2.69,
|
| 642 |
+
"learning_rate": 5e-05,
|
| 643 |
+
"loss": 1.4794,
|
| 644 |
+
"step": 43
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 2.69,
|
| 648 |
+
"eval_accuracy": 0.038008130081300814,
|
| 649 |
+
"eval_loss": 3.048828125,
|
| 650 |
+
"eval_runtime": 2.1015,
|
| 651 |
+
"eval_samples_per_second": 47.584,
|
| 652 |
+
"eval_steps_per_second": 0.952,
|
| 653 |
+
"step": 43
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 2.75,
|
| 657 |
+
"learning_rate": 5e-05,
|
| 658 |
+
"loss": 1.4313,
|
| 659 |
+
"step": 44
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 2.75,
|
| 663 |
+
"eval_accuracy": 0.03782520325203252,
|
| 664 |
+
"eval_loss": 3.052734375,
|
| 665 |
+
"eval_runtime": 2.4538,
|
| 666 |
+
"eval_samples_per_second": 40.753,
|
| 667 |
+
"eval_steps_per_second": 0.815,
|
| 668 |
+
"step": 44
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 2.81,
|
| 672 |
+
"learning_rate": 5e-05,
|
| 673 |
+
"loss": 1.5071,
|
| 674 |
+
"step": 45
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 2.81,
|
| 678 |
+
"eval_accuracy": 0.03784552845528455,
|
| 679 |
+
"eval_loss": 3.046875,
|
| 680 |
+
"eval_runtime": 2.1003,
|
| 681 |
+
"eval_samples_per_second": 47.611,
|
| 682 |
+
"eval_steps_per_second": 0.952,
|
| 683 |
+
"step": 45
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.88,
|
| 687 |
+
"learning_rate": 5e-05,
|
| 688 |
+
"loss": 1.4799,
|
| 689 |
+
"step": 46
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 2.88,
|
| 693 |
+
"eval_accuracy": 0.037764227642276424,
|
| 694 |
+
"eval_loss": 3.044921875,
|
| 695 |
+
"eval_runtime": 2.6039,
|
| 696 |
+
"eval_samples_per_second": 38.404,
|
| 697 |
+
"eval_steps_per_second": 0.768,
|
| 698 |
+
"step": 46
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 2.94,
|
| 702 |
+
"learning_rate": 5e-05,
|
| 703 |
+
"loss": 1.521,
|
| 704 |
+
"step": 47
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.94,
|
| 708 |
+
"eval_accuracy": 0.037967479674796745,
|
| 709 |
+
"eval_loss": 3.037109375,
|
| 710 |
+
"eval_runtime": 2.6013,
|
| 711 |
+
"eval_samples_per_second": 38.443,
|
| 712 |
+
"eval_steps_per_second": 0.769,
|
| 713 |
+
"step": 47
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 3.0,
|
| 717 |
+
"learning_rate": 5e-05,
|
| 718 |
+
"loss": 1.4603,
|
| 719 |
+
"step": 48
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 3.0,
|
| 723 |
+
"eval_accuracy": 0.03788617886178862,
|
| 724 |
+
"eval_loss": 3.041015625,
|
| 725 |
+
"eval_runtime": 2.0826,
|
| 726 |
+
"eval_samples_per_second": 48.016,
|
| 727 |
+
"eval_steps_per_second": 0.96,
|
| 728 |
+
"step": 48
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.06,
|
| 732 |
+
"learning_rate": 5e-05,
|
| 733 |
+
"loss": 1.25,
|
| 734 |
+
"step": 49
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.06,
|
| 738 |
+
"eval_accuracy": 0.03813008130081301,
|
| 739 |
+
"eval_loss": 3.0859375,
|
| 740 |
+
"eval_runtime": 2.5454,
|
| 741 |
+
"eval_samples_per_second": 39.287,
|
| 742 |
+
"eval_steps_per_second": 0.786,
|
| 743 |
+
"step": 49
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 3.12,
|
| 747 |
+
"learning_rate": 5e-05,
|
| 748 |
+
"loss": 1.0411,
|
| 749 |
+
"step": 50
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.12,
|
| 753 |
+
"eval_accuracy": 0.03754065040650407,
|
| 754 |
+
"eval_loss": 3.1796875,
|
| 755 |
+
"eval_runtime": 2.6016,
|
| 756 |
+
"eval_samples_per_second": 38.438,
|
| 757 |
+
"eval_steps_per_second": 0.769,
|
| 758 |
+
"step": 50
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 3.19,
|
| 762 |
+
"learning_rate": 5e-05,
|
| 763 |
+
"loss": 1.0385,
|
| 764 |
+
"step": 51
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 3.19,
|
| 768 |
+
"eval_accuracy": 0.037113821138211385,
|
| 769 |
+
"eval_loss": 3.296875,
|
| 770 |
+
"eval_runtime": 2.6072,
|
| 771 |
+
"eval_samples_per_second": 38.355,
|
| 772 |
+
"eval_steps_per_second": 0.767,
|
| 773 |
+
"step": 51
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 3.25,
|
| 777 |
+
"learning_rate": 5e-05,
|
| 778 |
+
"loss": 1.0254,
|
| 779 |
+
"step": 52
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.25,
|
| 783 |
+
"eval_accuracy": 0.03670731707317073,
|
| 784 |
+
"eval_loss": 3.361328125,
|
| 785 |
+
"eval_runtime": 2.2651,
|
| 786 |
+
"eval_samples_per_second": 44.148,
|
| 787 |
+
"eval_steps_per_second": 0.883,
|
| 788 |
+
"step": 52
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 3.31,
|
| 792 |
+
"learning_rate": 5e-05,
|
| 793 |
+
"loss": 0.9656,
|
| 794 |
+
"step": 53
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 3.31,
|
| 798 |
+
"eval_accuracy": 0.036829268292682925,
|
| 799 |
+
"eval_loss": 3.36328125,
|
| 800 |
+
"eval_runtime": 2.6237,
|
| 801 |
+
"eval_samples_per_second": 38.115,
|
| 802 |
+
"eval_steps_per_second": 0.762,
|
| 803 |
+
"step": 53
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 3.38,
|
| 807 |
+
"learning_rate": 5e-05,
|
| 808 |
+
"loss": 1.036,
|
| 809 |
+
"step": 54
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.38,
|
| 813 |
+
"eval_accuracy": 0.03664634146341463,
|
| 814 |
+
"eval_loss": 3.3359375,
|
| 815 |
+
"eval_runtime": 2.5895,
|
| 816 |
+
"eval_samples_per_second": 38.618,
|
| 817 |
+
"eval_steps_per_second": 0.772,
|
| 818 |
+
"step": 54
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 3.44,
|
| 822 |
+
"learning_rate": 5e-05,
|
| 823 |
+
"loss": 0.9366,
|
| 824 |
+
"step": 55
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 3.44,
|
| 828 |
+
"eval_accuracy": 0.036565040650406506,
|
| 829 |
+
"eval_loss": 3.294921875,
|
| 830 |
+
"eval_runtime": 2.2574,
|
| 831 |
+
"eval_samples_per_second": 44.298,
|
| 832 |
+
"eval_steps_per_second": 0.886,
|
| 833 |
+
"step": 55
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 3.5,
|
| 837 |
+
"learning_rate": 5e-05,
|
| 838 |
+
"loss": 0.9712,
|
| 839 |
+
"step": 56
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.5,
|
| 843 |
+
"eval_accuracy": 0.03666666666666667,
|
| 844 |
+
"eval_loss": 3.26953125,
|
| 845 |
+
"eval_runtime": 2.6146,
|
| 846 |
+
"eval_samples_per_second": 38.247,
|
| 847 |
+
"eval_steps_per_second": 0.765,
|
| 848 |
+
"step": 56
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 3.56,
|
| 852 |
+
"learning_rate": 5e-05,
|
| 853 |
+
"loss": 1.0066,
|
| 854 |
+
"step": 57
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 3.56,
|
| 858 |
+
"eval_accuracy": 0.036565040650406506,
|
| 859 |
+
"eval_loss": 3.267578125,
|
| 860 |
+
"eval_runtime": 2.6062,
|
| 861 |
+
"eval_samples_per_second": 38.37,
|
| 862 |
+
"eval_steps_per_second": 0.767,
|
| 863 |
+
"step": 57
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 3.62,
|
| 867 |
+
"learning_rate": 5e-05,
|
| 868 |
+
"loss": 0.9952,
|
| 869 |
+
"step": 58
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 3.62,
|
| 873 |
+
"eval_accuracy": 0.03676829268292683,
|
| 874 |
+
"eval_loss": 3.27734375,
|
| 875 |
+
"eval_runtime": 2.5965,
|
| 876 |
+
"eval_samples_per_second": 38.513,
|
| 877 |
+
"eval_steps_per_second": 0.77,
|
| 878 |
+
"step": 58
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 3.69,
|
| 882 |
+
"learning_rate": 5e-05,
|
| 883 |
+
"loss": 1.0352,
|
| 884 |
+
"step": 59
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 3.69,
|
| 888 |
+
"eval_accuracy": 0.0366869918699187,
|
| 889 |
+
"eval_loss": 3.2890625,
|
| 890 |
+
"eval_runtime": 2.3047,
|
| 891 |
+
"eval_samples_per_second": 43.39,
|
| 892 |
+
"eval_steps_per_second": 0.868,
|
| 893 |
+
"step": 59
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.75,
|
| 897 |
+
"learning_rate": 5e-05,
|
| 898 |
+
"loss": 1.0212,
|
| 899 |
+
"step": 60
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 3.75,
|
| 903 |
+
"eval_accuracy": 0.036178861788617886,
|
| 904 |
+
"eval_loss": 3.31640625,
|
| 905 |
+
"eval_runtime": 2.6015,
|
| 906 |
+
"eval_samples_per_second": 38.439,
|
| 907 |
+
"eval_steps_per_second": 0.769,
|
| 908 |
+
"step": 60
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 3.81,
|
| 912 |
+
"learning_rate": 5e-05,
|
| 913 |
+
"loss": 0.9468,
|
| 914 |
+
"step": 61
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 3.81,
|
| 918 |
+
"eval_accuracy": 0.036036585365853656,
|
| 919 |
+
"eval_loss": 3.3203125,
|
| 920 |
+
"eval_runtime": 2.2607,
|
| 921 |
+
"eval_samples_per_second": 44.233,
|
| 922 |
+
"eval_steps_per_second": 0.885,
|
| 923 |
+
"step": 61
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 3.88,
|
| 927 |
+
"learning_rate": 5e-05,
|
| 928 |
+
"loss": 0.9155,
|
| 929 |
+
"step": 62
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 3.88,
|
| 933 |
+
"eval_accuracy": 0.03664634146341463,
|
| 934 |
+
"eval_loss": 3.322265625,
|
| 935 |
+
"eval_runtime": 2.6074,
|
| 936 |
+
"eval_samples_per_second": 38.352,
|
| 937 |
+
"eval_steps_per_second": 0.767,
|
| 938 |
+
"step": 62
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 3.94,
|
| 942 |
+
"learning_rate": 5e-05,
|
| 943 |
+
"loss": 0.8552,
|
| 944 |
+
"step": 63
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 3.94,
|
| 948 |
+
"eval_accuracy": 0.03701219512195122,
|
| 949 |
+
"eval_loss": 3.326171875,
|
| 950 |
+
"eval_runtime": 2.2425,
|
| 951 |
+
"eval_samples_per_second": 44.594,
|
| 952 |
+
"eval_steps_per_second": 0.892,
|
| 953 |
+
"step": 63
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 4.0,
|
| 957 |
+
"learning_rate": 5e-05,
|
| 958 |
+
"loss": 0.9575,
|
| 959 |
+
"step": 64
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 4.0,
|
| 963 |
+
"eval_accuracy": 0.03699186991869919,
|
| 964 |
+
"eval_loss": 3.333984375,
|
| 965 |
+
"eval_runtime": 2.0082,
|
| 966 |
+
"eval_samples_per_second": 49.796,
|
| 967 |
+
"eval_steps_per_second": 0.996,
|
| 968 |
+
"step": 64
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 4.06,
|
| 972 |
+
"learning_rate": 5e-05,
|
| 973 |
+
"loss": 0.6384,
|
| 974 |
+
"step": 65
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 4.06,
|
| 978 |
+
"eval_accuracy": 0.036971544715447155,
|
| 979 |
+
"eval_loss": 3.375,
|
| 980 |
+
"eval_runtime": 2.5098,
|
| 981 |
+
"eval_samples_per_second": 39.844,
|
| 982 |
+
"eval_steps_per_second": 0.797,
|
| 983 |
+
"step": 65
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 4.12,
|
| 987 |
+
"learning_rate": 5e-05,
|
| 988 |
+
"loss": 0.6436,
|
| 989 |
+
"step": 66
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 4.12,
|
| 993 |
+
"eval_accuracy": 0.03636178861788618,
|
| 994 |
+
"eval_loss": 3.4453125,
|
| 995 |
+
"eval_runtime": 2.6062,
|
| 996 |
+
"eval_samples_per_second": 38.37,
|
| 997 |
+
"eval_steps_per_second": 0.767,
|
| 998 |
+
"step": 66
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 4.19,
|
| 1002 |
+
"learning_rate": 5e-05,
|
| 1003 |
+
"loss": 0.5752,
|
| 1004 |
+
"step": 67
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 4.19,
|
| 1008 |
+
"eval_accuracy": 0.035792682926829265,
|
| 1009 |
+
"eval_loss": 3.5390625,
|
| 1010 |
+
"eval_runtime": 2.6001,
|
| 1011 |
+
"eval_samples_per_second": 38.46,
|
| 1012 |
+
"eval_steps_per_second": 0.769,
|
| 1013 |
+
"step": 67
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 4.25,
|
| 1017 |
+
"learning_rate": 5e-05,
|
| 1018 |
+
"loss": 0.6542,
|
| 1019 |
+
"step": 68
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 4.25,
|
| 1023 |
+
"eval_accuracy": 0.03540650406504065,
|
| 1024 |
+
"eval_loss": 3.6015625,
|
| 1025 |
+
"eval_runtime": 2.4568,
|
| 1026 |
+
"eval_samples_per_second": 40.704,
|
| 1027 |
+
"eval_steps_per_second": 0.814,
|
| 1028 |
+
"step": 68
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 4.31,
|
| 1032 |
+
"learning_rate": 5e-05,
|
| 1033 |
+
"loss": 0.6724,
|
| 1034 |
+
"step": 69
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 4.31,
|
| 1038 |
+
"eval_accuracy": 0.03540650406504065,
|
| 1039 |
+
"eval_loss": 3.6015625,
|
| 1040 |
+
"eval_runtime": 2.0755,
|
| 1041 |
+
"eval_samples_per_second": 48.182,
|
| 1042 |
+
"eval_steps_per_second": 0.964,
|
| 1043 |
+
"step": 69
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 4.38,
|
| 1047 |
+
"learning_rate": 5e-05,
|
| 1048 |
+
"loss": 0.591,
|
| 1049 |
+
"step": 70
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 4.38,
|
| 1053 |
+
"eval_accuracy": 0.03591463414634146,
|
| 1054 |
+
"eval_loss": 3.59375,
|
| 1055 |
+
"eval_runtime": 2.3021,
|
| 1056 |
+
"eval_samples_per_second": 43.439,
|
| 1057 |
+
"eval_steps_per_second": 0.869,
|
| 1058 |
+
"step": 70
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 4.44,
|
| 1062 |
+
"learning_rate": 5e-05,
|
| 1063 |
+
"loss": 0.5346,
|
| 1064 |
+
"step": 71
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 4.44,
|
| 1068 |
+
"eval_accuracy": 0.03613821138211382,
|
| 1069 |
+
"eval_loss": 3.580078125,
|
| 1070 |
+
"eval_runtime": 2.0073,
|
| 1071 |
+
"eval_samples_per_second": 49.819,
|
| 1072 |
+
"eval_steps_per_second": 0.996,
|
| 1073 |
+
"step": 71
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 4.5,
|
| 1077 |
+
"learning_rate": 5e-05,
|
| 1078 |
+
"loss": 0.5112,
|
| 1079 |
+
"step": 72
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 4.5,
|
| 1083 |
+
"eval_accuracy": 0.036077235772357726,
|
| 1084 |
+
"eval_loss": 3.576171875,
|
| 1085 |
+
"eval_runtime": 2.6065,
|
| 1086 |
+
"eval_samples_per_second": 38.365,
|
| 1087 |
+
"eval_steps_per_second": 0.767,
|
| 1088 |
+
"step": 72
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 4.56,
|
| 1092 |
+
"learning_rate": 5e-05,
|
| 1093 |
+
"loss": 0.5443,
|
| 1094 |
+
"step": 73
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 4.56,
|
| 1098 |
+
"eval_accuracy": 0.03619918699186992,
|
| 1099 |
+
"eval_loss": 3.583984375,
|
| 1100 |
+
"eval_runtime": 2.6395,
|
| 1101 |
+
"eval_samples_per_second": 37.886,
|
| 1102 |
+
"eval_steps_per_second": 0.758,
|
| 1103 |
+
"step": 73
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 4.62,
|
| 1107 |
+
"learning_rate": 5e-05,
|
| 1108 |
+
"loss": 0.5689,
|
| 1109 |
+
"step": 74
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 4.62,
|
| 1113 |
+
"eval_accuracy": 0.035833333333333335,
|
| 1114 |
+
"eval_loss": 3.615234375,
|
| 1115 |
+
"eval_runtime": 2.4972,
|
| 1116 |
+
"eval_samples_per_second": 40.046,
|
| 1117 |
+
"eval_steps_per_second": 0.801,
|
| 1118 |
+
"step": 74
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 4.69,
|
| 1122 |
+
"learning_rate": 5e-05,
|
| 1123 |
+
"loss": 0.5667,
|
| 1124 |
+
"step": 75
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 4.69,
|
| 1128 |
+
"eval_accuracy": 0.0357520325203252,
|
| 1129 |
+
"eval_loss": 3.6328125,
|
| 1130 |
+
"eval_runtime": 2.4508,
|
| 1131 |
+
"eval_samples_per_second": 40.803,
|
| 1132 |
+
"eval_steps_per_second": 0.816,
|
| 1133 |
+
"step": 75
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 4.75,
|
| 1137 |
+
"learning_rate": 5e-05,
|
| 1138 |
+
"loss": 0.554,
|
| 1139 |
+
"step": 76
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 4.75,
|
| 1143 |
+
"eval_accuracy": 0.03573170731707317,
|
| 1144 |
+
"eval_loss": 3.634765625,
|
| 1145 |
+
"eval_runtime": 2.0574,
|
| 1146 |
+
"eval_samples_per_second": 48.604,
|
| 1147 |
+
"eval_steps_per_second": 0.972,
|
| 1148 |
+
"step": 76
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 4.81,
|
| 1152 |
+
"learning_rate": 5e-05,
|
| 1153 |
+
"loss": 0.6087,
|
| 1154 |
+
"step": 77
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 4.81,
|
| 1158 |
+
"eval_accuracy": 0.03552845528455285,
|
| 1159 |
+
"eval_loss": 3.625,
|
| 1160 |
+
"eval_runtime": 2.6082,
|
| 1161 |
+
"eval_samples_per_second": 38.34,
|
| 1162 |
+
"eval_steps_per_second": 0.767,
|
| 1163 |
+
"step": 77
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 4.88,
|
| 1167 |
+
"learning_rate": 5e-05,
|
| 1168 |
+
"loss": 0.5236,
|
| 1169 |
+
"step": 78
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 4.88,
|
| 1173 |
+
"eval_accuracy": 0.03552845528455285,
|
| 1174 |
+
"eval_loss": 3.615234375,
|
| 1175 |
+
"eval_runtime": 1.8631,
|
| 1176 |
+
"eval_samples_per_second": 53.674,
|
| 1177 |
+
"eval_steps_per_second": 1.073,
|
| 1178 |
+
"step": 78
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 4.94,
|
| 1182 |
+
"learning_rate": 5e-05,
|
| 1183 |
+
"loss": 0.5458,
|
| 1184 |
+
"step": 79
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 4.94,
|
| 1188 |
+
"eval_accuracy": 0.03550813008130081,
|
| 1189 |
+
"eval_loss": 3.578125,
|
| 1190 |
+
"eval_runtime": 2.6009,
|
| 1191 |
+
"eval_samples_per_second": 38.449,
|
| 1192 |
+
"eval_steps_per_second": 0.769,
|
| 1193 |
+
"step": 79
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 5.0,
|
| 1197 |
+
"learning_rate": 5e-05,
|
| 1198 |
+
"loss": 0.5702,
|
| 1199 |
+
"step": 80
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 5.0,
|
| 1203 |
+
"eval_accuracy": 0.03550813008130081,
|
| 1204 |
+
"eval_loss": 3.548828125,
|
| 1205 |
+
"eval_runtime": 2.5127,
|
| 1206 |
+
"eval_samples_per_second": 39.798,
|
| 1207 |
+
"eval_steps_per_second": 0.796,
|
| 1208 |
+
"step": 80
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 5.0,
|
| 1212 |
+
"step": 80,
|
| 1213 |
+
"total_flos": 9903514583040.0,
|
| 1214 |
+
"train_loss": 1.6890087127685547,
|
| 1215 |
+
"train_runtime": 406.6939,
|
| 1216 |
+
"train_samples_per_second": 12.294,
|
| 1217 |
+
"train_steps_per_second": 0.197
|
| 1218 |
+
}
|
| 1219 |
+
],
|
| 1220 |
+
"max_steps": 80,
|
| 1221 |
+
"num_train_epochs": 5,
|
| 1222 |
+
"total_flos": 9903514583040.0,
|
| 1223 |
+
"trial_name": null,
|
| 1224 |
+
"trial_params": null
|
| 1225 |
+
}
|