Commit
•
69fe7d1
1
Parent(s):
b7c3311
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +1825 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 500.0,
|
| 3 |
+
"eval_loss": 1.7564287185668945,
|
| 4 |
+
"eval_runtime": 6.7205,
|
| 5 |
+
"eval_samples": 142,
|
| 6 |
+
"eval_samples_per_second": 21.129,
|
| 7 |
+
"eval_steps_per_second": 2.678,
|
| 8 |
+
"eval_wer": 0.7201394943330427,
|
| 9 |
+
"train_loss": 0.7251964689095814,
|
| 10 |
+
"train_runtime": 16734.3269,
|
| 11 |
+
"train_samples": 378,
|
| 12 |
+
"train_samples_per_second": 11.294,
|
| 13 |
+
"train_steps_per_second": 0.717
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 500.0,
|
| 3 |
+
"eval_loss": 1.7564287185668945,
|
| 4 |
+
"eval_runtime": 6.7205,
|
| 5 |
+
"eval_samples": 142,
|
| 6 |
+
"eval_samples_per_second": 21.129,
|
| 7 |
+
"eval_steps_per_second": 2.678,
|
| 8 |
+
"eval_wer": 0.7201394943330427
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 500.0,
|
| 3 |
+
"train_loss": 0.7251964689095814,
|
| 4 |
+
"train_runtime": 16734.3269,
|
| 5 |
+
"train_samples": 378,
|
| 6 |
+
"train_samples_per_second": 11.294,
|
| 7 |
+
"train_steps_per_second": 0.717
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1825 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 500.0,
|
| 5 |
+
"global_step": 12000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 4.17,
|
| 12 |
+
"learning_rate": 7.35e-06,
|
| 13 |
+
"loss": 18.5133,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 4.17,
|
| 18 |
+
"eval_loss": 12.037576675415039,
|
| 19 |
+
"eval_runtime": 6.8573,
|
| 20 |
+
"eval_samples_per_second": 20.708,
|
| 21 |
+
"eval_steps_per_second": 2.625,
|
| 22 |
+
"eval_wer": 1.0,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 8.33,
|
| 27 |
+
"learning_rate": 1.485e-05,
|
| 28 |
+
"loss": 7.9152,
|
| 29 |
+
"step": 200
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 8.33,
|
| 33 |
+
"eval_loss": 6.240577220916748,
|
| 34 |
+
"eval_runtime": 6.2762,
|
| 35 |
+
"eval_samples_per_second": 22.625,
|
| 36 |
+
"eval_steps_per_second": 2.868,
|
| 37 |
+
"eval_wer": 1.0,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 12.5,
|
| 42 |
+
"learning_rate": 2.2349999999999998e-05,
|
| 43 |
+
"loss": 5.363,
|
| 44 |
+
"step": 300
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 12.5,
|
| 48 |
+
"eval_loss": 4.664888858795166,
|
| 49 |
+
"eval_runtime": 7.3621,
|
| 50 |
+
"eval_samples_per_second": 19.288,
|
| 51 |
+
"eval_steps_per_second": 2.445,
|
| 52 |
+
"eval_wer": 1.0,
|
| 53 |
+
"step": 300
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 16.67,
|
| 57 |
+
"learning_rate": 2.985e-05,
|
| 58 |
+
"loss": 4.05,
|
| 59 |
+
"step": 400
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 16.67,
|
| 63 |
+
"eval_loss": 3.5798044204711914,
|
| 64 |
+
"eval_runtime": 6.7794,
|
| 65 |
+
"eval_samples_per_second": 20.946,
|
| 66 |
+
"eval_steps_per_second": 2.655,
|
| 67 |
+
"eval_wer": 1.0,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 20.83,
|
| 72 |
+
"learning_rate": 3.735e-05,
|
| 73 |
+
"loss": 3.3467,
|
| 74 |
+
"step": 500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 20.83,
|
| 78 |
+
"eval_loss": 3.1930062770843506,
|
| 79 |
+
"eval_runtime": 7.0411,
|
| 80 |
+
"eval_samples_per_second": 20.167,
|
| 81 |
+
"eval_steps_per_second": 2.556,
|
| 82 |
+
"eval_wer": 1.0,
|
| 83 |
+
"step": 500
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 25.0,
|
| 87 |
+
"learning_rate": 4.484999999999999e-05,
|
| 88 |
+
"loss": 3.1638,
|
| 89 |
+
"step": 600
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 25.0,
|
| 93 |
+
"eval_loss": 3.098419666290283,
|
| 94 |
+
"eval_runtime": 6.9505,
|
| 95 |
+
"eval_samples_per_second": 20.43,
|
| 96 |
+
"eval_steps_per_second": 2.59,
|
| 97 |
+
"eval_wer": 1.0,
|
| 98 |
+
"step": 600
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 29.17,
|
| 102 |
+
"learning_rate": 5.234999999999999e-05,
|
| 103 |
+
"loss": 3.043,
|
| 104 |
+
"step": 700
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 29.17,
|
| 108 |
+
"eval_loss": 2.970651865005493,
|
| 109 |
+
"eval_runtime": 6.7343,
|
| 110 |
+
"eval_samples_per_second": 21.086,
|
| 111 |
+
"eval_steps_per_second": 2.673,
|
| 112 |
+
"eval_wer": 0.998256320836966,
|
| 113 |
+
"step": 700
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 33.33,
|
| 117 |
+
"learning_rate": 5.985e-05,
|
| 118 |
+
"loss": 2.9566,
|
| 119 |
+
"step": 800
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 33.33,
|
| 123 |
+
"eval_loss": 2.9295754432678223,
|
| 124 |
+
"eval_runtime": 6.7892,
|
| 125 |
+
"eval_samples_per_second": 20.916,
|
| 126 |
+
"eval_steps_per_second": 2.651,
|
| 127 |
+
"eval_wer": 1.0,
|
| 128 |
+
"step": 800
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 37.5,
|
| 132 |
+
"learning_rate": 6.735e-05,
|
| 133 |
+
"loss": 2.8994,
|
| 134 |
+
"step": 900
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 37.5,
|
| 138 |
+
"eval_loss": 2.8289785385131836,
|
| 139 |
+
"eval_runtime": 6.3572,
|
| 140 |
+
"eval_samples_per_second": 22.337,
|
| 141 |
+
"eval_steps_per_second": 2.831,
|
| 142 |
+
"eval_wer": 0.998256320836966,
|
| 143 |
+
"step": 900
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 41.67,
|
| 147 |
+
"learning_rate": 7.484999999999999e-05,
|
| 148 |
+
"loss": 2.6469,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 41.67,
|
| 153 |
+
"eval_loss": 2.244255781173706,
|
| 154 |
+
"eval_runtime": 7.0587,
|
| 155 |
+
"eval_samples_per_second": 20.117,
|
| 156 |
+
"eval_steps_per_second": 2.55,
|
| 157 |
+
"eval_wer": 0.976460331299041,
|
| 158 |
+
"step": 1000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 45.83,
|
| 162 |
+
"learning_rate": 7.433181818181818e-05,
|
| 163 |
+
"loss": 2.1557,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 45.83,
|
| 168 |
+
"eval_loss": 1.5079460144042969,
|
| 169 |
+
"eval_runtime": 6.593,
|
| 170 |
+
"eval_samples_per_second": 21.538,
|
| 171 |
+
"eval_steps_per_second": 2.73,
|
| 172 |
+
"eval_wer": 0.8204010462074979,
|
| 173 |
+
"step": 1100
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 50.0,
|
| 177 |
+
"learning_rate": 7.364999999999999e-05,
|
| 178 |
+
"loss": 1.7524,
|
| 179 |
+
"step": 1200
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 50.0,
|
| 183 |
+
"eval_loss": 1.2472423315048218,
|
| 184 |
+
"eval_runtime": 6.884,
|
| 185 |
+
"eval_samples_per_second": 20.628,
|
| 186 |
+
"eval_steps_per_second": 2.615,
|
| 187 |
+
"eval_wer": 0.7593722755013078,
|
| 188 |
+
"step": 1200
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 54.17,
|
| 192 |
+
"learning_rate": 7.296818181818182e-05,
|
| 193 |
+
"loss": 1.54,
|
| 194 |
+
"step": 1300
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 54.17,
|
| 198 |
+
"eval_loss": 1.1611692905426025,
|
| 199 |
+
"eval_runtime": 6.7498,
|
| 200 |
+
"eval_samples_per_second": 21.038,
|
| 201 |
+
"eval_steps_per_second": 2.667,
|
| 202 |
+
"eval_wer": 0.7096774193548387,
|
| 203 |
+
"step": 1300
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 58.33,
|
| 207 |
+
"learning_rate": 7.228636363636363e-05,
|
| 208 |
+
"loss": 1.3985,
|
| 209 |
+
"step": 1400
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 58.33,
|
| 213 |
+
"eval_loss": 1.1158100366592407,
|
| 214 |
+
"eval_runtime": 6.9751,
|
| 215 |
+
"eval_samples_per_second": 20.358,
|
| 216 |
+
"eval_steps_per_second": 2.581,
|
| 217 |
+
"eval_wer": 0.7358326068003488,
|
| 218 |
+
"step": 1400
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 62.5,
|
| 222 |
+
"learning_rate": 7.160454545454545e-05,
|
| 223 |
+
"loss": 1.2869,
|
| 224 |
+
"step": 1500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 62.5,
|
| 228 |
+
"eval_loss": 1.0453699827194214,
|
| 229 |
+
"eval_runtime": 6.542,
|
| 230 |
+
"eval_samples_per_second": 21.706,
|
| 231 |
+
"eval_steps_per_second": 2.751,
|
| 232 |
+
"eval_wer": 0.6931124673060157,
|
| 233 |
+
"step": 1500
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 66.67,
|
| 237 |
+
"learning_rate": 7.092272727272727e-05,
|
| 238 |
+
"loss": 1.1952,
|
| 239 |
+
"step": 1600
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 66.67,
|
| 243 |
+
"eval_loss": 1.0130492448806763,
|
| 244 |
+
"eval_runtime": 7.0228,
|
| 245 |
+
"eval_samples_per_second": 20.22,
|
| 246 |
+
"eval_steps_per_second": 2.563,
|
| 247 |
+
"eval_wer": 0.6852659110723627,
|
| 248 |
+
"step": 1600
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 70.83,
|
| 252 |
+
"learning_rate": 7.024090909090908e-05,
|
| 253 |
+
"loss": 1.1022,
|
| 254 |
+
"step": 1700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 70.83,
|
| 258 |
+
"eval_loss": 1.017554521560669,
|
| 259 |
+
"eval_runtime": 6.8272,
|
| 260 |
+
"eval_samples_per_second": 20.799,
|
| 261 |
+
"eval_steps_per_second": 2.637,
|
| 262 |
+
"eval_wer": 0.6965998256320837,
|
| 263 |
+
"step": 1700
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 75.0,
|
| 267 |
+
"learning_rate": 6.95590909090909e-05,
|
| 268 |
+
"loss": 1.0346,
|
| 269 |
+
"step": 1800
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 75.0,
|
| 273 |
+
"eval_loss": 1.0053126811981201,
|
| 274 |
+
"eval_runtime": 6.9741,
|
| 275 |
+
"eval_samples_per_second": 20.361,
|
| 276 |
+
"eval_steps_per_second": 2.581,
|
| 277 |
+
"eval_wer": 0.6817785527462947,
|
| 278 |
+
"step": 1800
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 79.17,
|
| 282 |
+
"learning_rate": 6.887727272727272e-05,
|
| 283 |
+
"loss": 0.9707,
|
| 284 |
+
"step": 1900
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 79.17,
|
| 288 |
+
"eval_loss": 1.022377610206604,
|
| 289 |
+
"eval_runtime": 7.0064,
|
| 290 |
+
"eval_samples_per_second": 20.267,
|
| 291 |
+
"eval_steps_per_second": 2.569,
|
| 292 |
+
"eval_wer": 0.6713164777680907,
|
| 293 |
+
"step": 1900
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 83.33,
|
| 297 |
+
"learning_rate": 6.819545454545453e-05,
|
| 298 |
+
"loss": 0.917,
|
| 299 |
+
"step": 2000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 83.33,
|
| 303 |
+
"eval_loss": 1.0382641553878784,
|
| 304 |
+
"eval_runtime": 6.7309,
|
| 305 |
+
"eval_samples_per_second": 21.097,
|
| 306 |
+
"eval_steps_per_second": 2.674,
|
| 307 |
+
"eval_wer": 0.6530078465562337,
|
| 308 |
+
"step": 2000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 87.5,
|
| 312 |
+
"learning_rate": 6.751363636363636e-05,
|
| 313 |
+
"loss": 0.8574,
|
| 314 |
+
"step": 2100
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 87.5,
|
| 318 |
+
"eval_loss": 1.0632084608078003,
|
| 319 |
+
"eval_runtime": 7.3312,
|
| 320 |
+
"eval_samples_per_second": 19.369,
|
| 321 |
+
"eval_steps_per_second": 2.455,
|
| 322 |
+
"eval_wer": 0.6756756756756757,
|
| 323 |
+
"step": 2100
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 91.67,
|
| 327 |
+
"learning_rate": 6.683181818181818e-05,
|
| 328 |
+
"loss": 0.8021,
|
| 329 |
+
"step": 2200
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 91.67,
|
| 333 |
+
"eval_loss": 1.012629747390747,
|
| 334 |
+
"eval_runtime": 6.9334,
|
| 335 |
+
"eval_samples_per_second": 20.481,
|
| 336 |
+
"eval_steps_per_second": 2.596,
|
| 337 |
+
"eval_wer": 0.6442894507410637,
|
| 338 |
+
"step": 2200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 95.83,
|
| 342 |
+
"learning_rate": 6.615e-05,
|
| 343 |
+
"loss": 0.7563,
|
| 344 |
+
"step": 2300
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 95.83,
|
| 348 |
+
"eval_loss": 1.0677976608276367,
|
| 349 |
+
"eval_runtime": 7.0411,
|
| 350 |
+
"eval_samples_per_second": 20.167,
|
| 351 |
+
"eval_steps_per_second": 2.556,
|
| 352 |
+
"eval_wer": 0.6713164777680907,
|
| 353 |
+
"step": 2300
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 100.0,
|
| 357 |
+
"learning_rate": 6.546818181818181e-05,
|
| 358 |
+
"loss": 0.709,
|
| 359 |
+
"step": 2400
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 100.0,
|
| 363 |
+
"eval_loss": 1.0755900144577026,
|
| 364 |
+
"eval_runtime": 6.7348,
|
| 365 |
+
"eval_samples_per_second": 21.085,
|
| 366 |
+
"eval_steps_per_second": 2.673,
|
| 367 |
+
"eval_wer": 0.6756756756756757,
|
| 368 |
+
"step": 2400
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 104.17,
|
| 372 |
+
"learning_rate": 6.478636363636363e-05,
|
| 373 |
+
"loss": 0.6775,
|
| 374 |
+
"step": 2500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 104.17,
|
| 378 |
+
"eval_loss": 1.03966224193573,
|
| 379 |
+
"eval_runtime": 6.7793,
|
| 380 |
+
"eval_samples_per_second": 20.946,
|
| 381 |
+
"eval_steps_per_second": 2.655,
|
| 382 |
+
"eval_wer": 0.6913687881429816,
|
| 383 |
+
"step": 2500
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 108.33,
|
| 387 |
+
"learning_rate": 6.410454545454546e-05,
|
| 388 |
+
"loss": 0.6325,
|
| 389 |
+
"step": 2600
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 108.33,
|
| 393 |
+
"eval_loss": 1.055609107017517,
|
| 394 |
+
"eval_runtime": 6.9878,
|
| 395 |
+
"eval_samples_per_second": 20.321,
|
| 396 |
+
"eval_steps_per_second": 2.576,
|
| 397 |
+
"eval_wer": 0.6512641673931997,
|
| 398 |
+
"step": 2600
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 112.5,
|
| 402 |
+
"learning_rate": 6.342272727272726e-05,
|
| 403 |
+
"loss": 0.617,
|
| 404 |
+
"step": 2700
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 112.5,
|
| 408 |
+
"eval_loss": 1.1218976974487305,
|
| 409 |
+
"eval_runtime": 6.8023,
|
| 410 |
+
"eval_samples_per_second": 20.875,
|
| 411 |
+
"eval_steps_per_second": 2.646,
|
| 412 |
+
"eval_wer": 0.6922406277244987,
|
| 413 |
+
"step": 2700
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 116.67,
|
| 417 |
+
"learning_rate": 6.274090909090909e-05,
|
| 418 |
+
"loss": 0.5801,
|
| 419 |
+
"step": 2800
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 116.67,
|
| 423 |
+
"eval_loss": 1.112548589706421,
|
| 424 |
+
"eval_runtime": 6.7506,
|
| 425 |
+
"eval_samples_per_second": 21.035,
|
| 426 |
+
"eval_steps_per_second": 2.666,
|
| 427 |
+
"eval_wer": 0.6870095902353966,
|
| 428 |
+
"step": 2800
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 120.83,
|
| 432 |
+
"learning_rate": 6.205909090909091e-05,
|
| 433 |
+
"loss": 0.5367,
|
| 434 |
+
"step": 2900
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 120.83,
|
| 438 |
+
"eval_loss": 1.1397372484207153,
|
| 439 |
+
"eval_runtime": 6.876,
|
| 440 |
+
"eval_samples_per_second": 20.652,
|
| 441 |
+
"eval_steps_per_second": 2.618,
|
| 442 |
+
"eval_wer": 0.6564952048823016,
|
| 443 |
+
"step": 2900
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 125.0,
|
| 447 |
+
"learning_rate": 6.137727272727272e-05,
|
| 448 |
+
"loss": 0.5132,
|
| 449 |
+
"step": 3000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 125.0,
|
| 453 |
+
"eval_loss": 1.1678012609481812,
|
| 454 |
+
"eval_runtime": 6.8221,
|
| 455 |
+
"eval_samples_per_second": 20.815,
|
| 456 |
+
"eval_steps_per_second": 2.638,
|
| 457 |
+
"eval_wer": 0.7000871839581517,
|
| 458 |
+
"step": 3000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 129.17,
|
| 462 |
+
"learning_rate": 6.069545454545454e-05,
|
| 463 |
+
"loss": 0.4948,
|
| 464 |
+
"step": 3100
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 129.17,
|
| 468 |
+
"eval_loss": 1.1275851726531982,
|
| 469 |
+
"eval_runtime": 6.9051,
|
| 470 |
+
"eval_samples_per_second": 20.564,
|
| 471 |
+
"eval_steps_per_second": 2.607,
|
| 472 |
+
"eval_wer": 0.6643417611159547,
|
| 473 |
+
"step": 3100
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 133.33,
|
| 477 |
+
"learning_rate": 6.001363636363636e-05,
|
| 478 |
+
"loss": 0.457,
|
| 479 |
+
"step": 3200
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 133.33,
|
| 483 |
+
"eval_loss": 1.1464685201644897,
|
| 484 |
+
"eval_runtime": 6.6513,
|
| 485 |
+
"eval_samples_per_second": 21.349,
|
| 486 |
+
"eval_steps_per_second": 2.706,
|
| 487 |
+
"eval_wer": 0.6791630340017437,
|
| 488 |
+
"step": 3200
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 137.5,
|
| 492 |
+
"learning_rate": 5.933181818181817e-05,
|
| 493 |
+
"loss": 0.4538,
|
| 494 |
+
"step": 3300
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 137.5,
|
| 498 |
+
"eval_loss": 1.143470287322998,
|
| 499 |
+
"eval_runtime": 5.4611,
|
| 500 |
+
"eval_samples_per_second": 26.002,
|
| 501 |
+
"eval_steps_per_second": 3.296,
|
| 502 |
+
"eval_wer": 0.6809067131647777,
|
| 503 |
+
"step": 3300
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 141.67,
|
| 507 |
+
"learning_rate": 5.8649999999999996e-05,
|
| 508 |
+
"loss": 0.4227,
|
| 509 |
+
"step": 3400
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 141.67,
|
| 513 |
+
"eval_loss": 1.1456971168518066,
|
| 514 |
+
"eval_runtime": 6.7115,
|
| 515 |
+
"eval_samples_per_second": 21.158,
|
| 516 |
+
"eval_steps_per_second": 2.682,
|
| 517 |
+
"eval_wer": 0.6599825632083697,
|
| 518 |
+
"step": 3400
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 145.83,
|
| 522 |
+
"learning_rate": 5.796818181818181e-05,
|
| 523 |
+
"loss": 0.4083,
|
| 524 |
+
"step": 3500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 145.83,
|
| 528 |
+
"eval_loss": 1.179282307624817,
|
| 529 |
+
"eval_runtime": 7.0949,
|
| 530 |
+
"eval_samples_per_second": 20.014,
|
| 531 |
+
"eval_steps_per_second": 2.537,
|
| 532 |
+
"eval_wer": 0.6652136006974717,
|
| 533 |
+
"step": 3500
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 150.0,
|
| 537 |
+
"learning_rate": 5.7286363636363635e-05,
|
| 538 |
+
"loss": 0.3965,
|
| 539 |
+
"step": 3600
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 150.0,
|
| 543 |
+
"eval_loss": 1.2435046434402466,
|
| 544 |
+
"eval_runtime": 7.0689,
|
| 545 |
+
"eval_samples_per_second": 20.088,
|
| 546 |
+
"eval_steps_per_second": 2.546,
|
| 547 |
+
"eval_wer": 0.6878814298169137,
|
| 548 |
+
"step": 3600
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 154.17,
|
| 552 |
+
"learning_rate": 5.6604545454545445e-05,
|
| 553 |
+
"loss": 0.382,
|
| 554 |
+
"step": 3700
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 154.17,
|
| 558 |
+
"eval_loss": 1.2228599786758423,
|
| 559 |
+
"eval_runtime": 7.0115,
|
| 560 |
+
"eval_samples_per_second": 20.252,
|
| 561 |
+
"eval_steps_per_second": 2.567,
|
| 562 |
+
"eval_wer": 0.6826503923278117,
|
| 563 |
+
"step": 3700
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 158.33,
|
| 567 |
+
"learning_rate": 5.592272727272727e-05,
|
| 568 |
+
"loss": 0.3452,
|
| 569 |
+
"step": 3800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 158.33,
|
| 573 |
+
"eval_loss": 1.229727029800415,
|
| 574 |
+
"eval_runtime": 6.7478,
|
| 575 |
+
"eval_samples_per_second": 21.044,
|
| 576 |
+
"eval_steps_per_second": 2.668,
|
| 577 |
+
"eval_wer": 0.6878814298169137,
|
| 578 |
+
"step": 3800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 162.5,
|
| 582 |
+
"learning_rate": 5.5240909090909085e-05,
|
| 583 |
+
"loss": 0.3434,
|
| 584 |
+
"step": 3900
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 162.5,
|
| 588 |
+
"eval_loss": 1.2349668741226196,
|
| 589 |
+
"eval_runtime": 7.0134,
|
| 590 |
+
"eval_samples_per_second": 20.247,
|
| 591 |
+
"eval_steps_per_second": 2.567,
|
| 592 |
+
"eval_wer": 0.6887532693984307,
|
| 593 |
+
"step": 3900
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 166.67,
|
| 597 |
+
"learning_rate": 5.455909090909091e-05,
|
| 598 |
+
"loss": 0.3276,
|
| 599 |
+
"step": 4000
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 166.67,
|
| 603 |
+
"eval_loss": 1.2186285257339478,
|
| 604 |
+
"eval_runtime": 6.7562,
|
| 605 |
+
"eval_samples_per_second": 21.018,
|
| 606 |
+
"eval_steps_per_second": 2.664,
|
| 607 |
+
"eval_wer": 0.6922406277244987,
|
| 608 |
+
"step": 4000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 170.83,
|
| 612 |
+
"learning_rate": 5.387727272727272e-05,
|
| 613 |
+
"loss": 0.3052,
|
| 614 |
+
"step": 4100
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 170.83,
|
| 618 |
+
"eval_loss": 1.233783483505249,
|
| 619 |
+
"eval_runtime": 6.5925,
|
| 620 |
+
"eval_samples_per_second": 21.54,
|
| 621 |
+
"eval_steps_per_second": 2.73,
|
| 622 |
+
"eval_wer": 0.6870095902353966,
|
| 623 |
+
"step": 4100
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 175.0,
|
| 627 |
+
"learning_rate": 5.319545454545454e-05,
|
| 628 |
+
"loss": 0.3025,
|
| 629 |
+
"step": 4200
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 175.0,
|
| 633 |
+
"eval_loss": 1.272440791130066,
|
| 634 |
+
"eval_runtime": 6.8253,
|
| 635 |
+
"eval_samples_per_second": 20.805,
|
| 636 |
+
"eval_steps_per_second": 2.637,
|
| 637 |
+
"eval_wer": 0.7079337401918047,
|
| 638 |
+
"step": 4200
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 179.17,
|
| 642 |
+
"learning_rate": 5.251363636363636e-05,
|
| 643 |
+
"loss": 0.2916,
|
| 644 |
+
"step": 4300
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 179.17,
|
| 648 |
+
"eval_loss": 1.2757943868637085,
|
| 649 |
+
"eval_runtime": 6.6107,
|
| 650 |
+
"eval_samples_per_second": 21.48,
|
| 651 |
+
"eval_steps_per_second": 2.723,
|
| 652 |
+
"eval_wer": 0.6974716652136007,
|
| 653 |
+
"step": 4300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 183.33,
|
| 657 |
+
"learning_rate": 5.183181818181818e-05,
|
| 658 |
+
"loss": 0.2709,
|
| 659 |
+
"step": 4400
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 183.33,
|
| 663 |
+
"eval_loss": 1.2726093530654907,
|
| 664 |
+
"eval_runtime": 6.7657,
|
| 665 |
+
"eval_samples_per_second": 20.988,
|
| 666 |
+
"eval_steps_per_second": 2.66,
|
| 667 |
+
"eval_wer": 0.6748038360941587,
|
| 668 |
+
"step": 4400
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 187.5,
|
| 672 |
+
"learning_rate": 5.1149999999999996e-05,
|
| 673 |
+
"loss": 0.2707,
|
| 674 |
+
"step": 4500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 187.5,
|
| 678 |
+
"eval_loss": 1.264315128326416,
|
| 679 |
+
"eval_runtime": 6.6322,
|
| 680 |
+
"eval_samples_per_second": 21.411,
|
| 681 |
+
"eval_steps_per_second": 2.714,
|
| 682 |
+
"eval_wer": 0.6957279860505667,
|
| 683 |
+
"step": 4500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 191.67,
|
| 687 |
+
"learning_rate": 5.046818181818181e-05,
|
| 688 |
+
"loss": 0.262,
|
| 689 |
+
"step": 4600
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 191.67,
|
| 693 |
+
"eval_loss": 1.321448564529419,
|
| 694 |
+
"eval_runtime": 6.7016,
|
| 695 |
+
"eval_samples_per_second": 21.189,
|
| 696 |
+
"eval_steps_per_second": 2.686,
|
| 697 |
+
"eval_wer": 0.7131647776809067,
|
| 698 |
+
"step": 4600
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 195.83,
|
| 702 |
+
"learning_rate": 4.9786363636363636e-05,
|
| 703 |
+
"loss": 0.2453,
|
| 704 |
+
"step": 4700
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 195.83,
|
| 708 |
+
"eval_loss": 1.2952940464019775,
|
| 709 |
+
"eval_runtime": 6.8362,
|
| 710 |
+
"eval_samples_per_second": 20.772,
|
| 711 |
+
"eval_steps_per_second": 2.633,
|
| 712 |
+
"eval_wer": 0.6861377506538797,
|
| 713 |
+
"step": 4700
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 200.0,
|
| 717 |
+
"learning_rate": 4.910454545454545e-05,
|
| 718 |
+
"loss": 0.248,
|
| 719 |
+
"step": 4800
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 200.0,
|
| 723 |
+
"eval_loss": 1.362170696258545,
|
| 724 |
+
"eval_runtime": 6.5267,
|
| 725 |
+
"eval_samples_per_second": 21.757,
|
| 726 |
+
"eval_steps_per_second": 2.758,
|
| 727 |
+
"eval_wer": 0.6774193548387096,
|
| 728 |
+
"step": 4800
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 204.17,
|
| 732 |
+
"learning_rate": 4.842272727272727e-05,
|
| 733 |
+
"loss": 0.2325,
|
| 734 |
+
"step": 4900
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 204.17,
|
| 738 |
+
"eval_loss": 1.3594108819961548,
|
| 739 |
+
"eval_runtime": 6.7956,
|
| 740 |
+
"eval_samples_per_second": 20.896,
|
| 741 |
+
"eval_steps_per_second": 2.649,
|
| 742 |
+
"eval_wer": 0.6835222319093287,
|
| 743 |
+
"step": 4900
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 208.33,
|
| 747 |
+
"learning_rate": 4.7740909090909085e-05,
|
| 748 |
+
"loss": 0.2124,
|
| 749 |
+
"step": 5000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 208.33,
|
| 753 |
+
"eval_loss": 1.3367135524749756,
|
| 754 |
+
"eval_runtime": 6.8257,
|
| 755 |
+
"eval_samples_per_second": 20.804,
|
| 756 |
+
"eval_steps_per_second": 2.637,
|
| 757 |
+
"eval_wer": 0.6652136006974717,
|
| 758 |
+
"step": 5000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 212.5,
|
| 762 |
+
"learning_rate": 4.706590909090908e-05,
|
| 763 |
+
"loss": 0.2253,
|
| 764 |
+
"step": 5100
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 212.5,
|
| 768 |
+
"eval_loss": 1.415678858757019,
|
| 769 |
+
"eval_runtime": 6.8381,
|
| 770 |
+
"eval_samples_per_second": 20.766,
|
| 771 |
+
"eval_steps_per_second": 2.632,
|
| 772 |
+
"eval_wer": 0.6878814298169137,
|
| 773 |
+
"step": 5100
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 216.67,
|
| 777 |
+
"learning_rate": 4.6384090909090906e-05,
|
| 778 |
+
"loss": 0.2059,
|
| 779 |
+
"step": 5200
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 216.67,
|
| 783 |
+
"eval_loss": 1.4359543323516846,
|
| 784 |
+
"eval_runtime": 6.8901,
|
| 785 |
+
"eval_samples_per_second": 20.609,
|
| 786 |
+
"eval_steps_per_second": 2.612,
|
| 787 |
+
"eval_wer": 0.7131647776809067,
|
| 788 |
+
"step": 5200
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 220.83,
|
| 792 |
+
"learning_rate": 4.570227272727272e-05,
|
| 793 |
+
"loss": 0.1951,
|
| 794 |
+
"step": 5300
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 220.83,
|
| 798 |
+
"eval_loss": 1.4606153964996338,
|
| 799 |
+
"eval_runtime": 6.7349,
|
| 800 |
+
"eval_samples_per_second": 21.084,
|
| 801 |
+
"eval_steps_per_second": 2.673,
|
| 802 |
+
"eval_wer": 0.7157802964254577,
|
| 803 |
+
"step": 5300
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 225.0,
|
| 807 |
+
"learning_rate": 4.5020454545454545e-05,
|
| 808 |
+
"loss": 0.1861,
|
| 809 |
+
"step": 5400
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 225.0,
|
| 813 |
+
"eval_loss": 1.469545841217041,
|
| 814 |
+
"eval_runtime": 6.7814,
|
| 815 |
+
"eval_samples_per_second": 20.94,
|
| 816 |
+
"eval_steps_per_second": 2.654,
|
| 817 |
+
"eval_wer": 0.7018308631211857,
|
| 818 |
+
"step": 5400
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 229.17,
|
| 822 |
+
"learning_rate": 4.4338636363636355e-05,
|
| 823 |
+
"loss": 0.1916,
|
| 824 |
+
"step": 5500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 229.17,
|
| 828 |
+
"eval_loss": 1.4031471014022827,
|
| 829 |
+
"eval_runtime": 6.9155,
|
| 830 |
+
"eval_samples_per_second": 20.534,
|
| 831 |
+
"eval_steps_per_second": 2.603,
|
| 832 |
+
"eval_wer": 0.6739319965126417,
|
| 833 |
+
"step": 5500
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 233.33,
|
| 837 |
+
"learning_rate": 4.365681818181818e-05,
|
| 838 |
+
"loss": 0.1822,
|
| 839 |
+
"step": 5600
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 233.33,
|
| 843 |
+
"eval_loss": 1.4425878524780273,
|
| 844 |
+
"eval_runtime": 6.658,
|
| 845 |
+
"eval_samples_per_second": 21.328,
|
| 846 |
+
"eval_steps_per_second": 2.704,
|
| 847 |
+
"eval_wer": 0.6870095902353966,
|
| 848 |
+
"step": 5600
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 237.5,
|
| 852 |
+
"learning_rate": 4.2974999999999994e-05,
|
| 853 |
+
"loss": 0.1684,
|
| 854 |
+
"step": 5700
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 237.5,
|
| 858 |
+
"eval_loss": 1.4068546295166016,
|
| 859 |
+
"eval_runtime": 5.8333,
|
| 860 |
+
"eval_samples_per_second": 24.343,
|
| 861 |
+
"eval_steps_per_second": 3.086,
|
| 862 |
+
"eval_wer": 0.7053182214472538,
|
| 863 |
+
"step": 5700
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 241.67,
|
| 867 |
+
"learning_rate": 4.229318181818182e-05,
|
| 868 |
+
"loss": 0.1719,
|
| 869 |
+
"step": 5800
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 241.67,
|
| 873 |
+
"eval_loss": 1.4765515327453613,
|
| 874 |
+
"eval_runtime": 6.853,
|
| 875 |
+
"eval_samples_per_second": 20.721,
|
| 876 |
+
"eval_steps_per_second": 2.627,
|
| 877 |
+
"eval_wer": 0.6965998256320837,
|
| 878 |
+
"step": 5800
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 245.83,
|
| 882 |
+
"learning_rate": 4.161136363636363e-05,
|
| 883 |
+
"loss": 0.1569,
|
| 884 |
+
"step": 5900
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 245.83,
|
| 888 |
+
"eval_loss": 1.4509494304656982,
|
| 889 |
+
"eval_runtime": 7.6508,
|
| 890 |
+
"eval_samples_per_second": 18.56,
|
| 891 |
+
"eval_steps_per_second": 2.353,
|
| 892 |
+
"eval_wer": 0.6931124673060157,
|
| 893 |
+
"step": 5900
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 250.0,
|
| 897 |
+
"learning_rate": 4.092954545454545e-05,
|
| 898 |
+
"loss": 0.159,
|
| 899 |
+
"step": 6000
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 250.0,
|
| 903 |
+
"eval_loss": 1.4466707706451416,
|
| 904 |
+
"eval_runtime": 6.8247,
|
| 905 |
+
"eval_samples_per_second": 20.807,
|
| 906 |
+
"eval_steps_per_second": 2.637,
|
| 907 |
+
"eval_wer": 0.7096774193548387,
|
| 908 |
+
"step": 6000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 254.17,
|
| 912 |
+
"learning_rate": 4.0247727272727273e-05,
|
| 913 |
+
"loss": 0.1476,
|
| 914 |
+
"step": 6100
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 254.17,
|
| 918 |
+
"eval_loss": 1.4616789817810059,
|
| 919 |
+
"eval_runtime": 6.6822,
|
| 920 |
+
"eval_samples_per_second": 21.25,
|
| 921 |
+
"eval_steps_per_second": 2.694,
|
| 922 |
+
"eval_wer": 0.6870095902353966,
|
| 923 |
+
"step": 6100
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 258.33,
|
| 927 |
+
"learning_rate": 3.956590909090909e-05,
|
| 928 |
+
"loss": 0.1497,
|
| 929 |
+
"step": 6200
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 258.33,
|
| 933 |
+
"eval_loss": 1.4459782838821411,
|
| 934 |
+
"eval_runtime": 7.0114,
|
| 935 |
+
"eval_samples_per_second": 20.253,
|
| 936 |
+
"eval_steps_per_second": 2.567,
|
| 937 |
+
"eval_wer": 0.6843940714908456,
|
| 938 |
+
"step": 6200
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 262.5,
|
| 942 |
+
"learning_rate": 3.8884090909090906e-05,
|
| 943 |
+
"loss": 0.1446,
|
| 944 |
+
"step": 6300
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 262.5,
|
| 948 |
+
"eval_loss": 1.5556844472885132,
|
| 949 |
+
"eval_runtime": 6.9912,
|
| 950 |
+
"eval_samples_per_second": 20.311,
|
| 951 |
+
"eval_steps_per_second": 2.575,
|
| 952 |
+
"eval_wer": 0.7088055797733217,
|
| 953 |
+
"step": 6300
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 266.67,
|
| 957 |
+
"learning_rate": 3.820227272727272e-05,
|
| 958 |
+
"loss": 0.1389,
|
| 959 |
+
"step": 6400
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 266.67,
|
| 963 |
+
"eval_loss": 1.4885756969451904,
|
| 964 |
+
"eval_runtime": 6.5474,
|
| 965 |
+
"eval_samples_per_second": 21.688,
|
| 966 |
+
"eval_steps_per_second": 2.749,
|
| 967 |
+
"eval_wer": 0.7140366172624237,
|
| 968 |
+
"step": 6400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 270.83,
|
| 972 |
+
"learning_rate": 3.7520454545454546e-05,
|
| 973 |
+
"loss": 0.1331,
|
| 974 |
+
"step": 6500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 270.83,
|
| 978 |
+
"eval_loss": 1.5526471138000488,
|
| 979 |
+
"eval_runtime": 6.5806,
|
| 980 |
+
"eval_samples_per_second": 21.579,
|
| 981 |
+
"eval_steps_per_second": 2.735,
|
| 982 |
+
"eval_wer": 0.7061900610287707,
|
| 983 |
+
"step": 6500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 275.0,
|
| 987 |
+
"learning_rate": 3.683863636363636e-05,
|
| 988 |
+
"loss": 0.1344,
|
| 989 |
+
"step": 6600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 275.0,
|
| 993 |
+
"eval_loss": 1.5419210195541382,
|
| 994 |
+
"eval_runtime": 6.8756,
|
| 995 |
+
"eval_samples_per_second": 20.653,
|
| 996 |
+
"eval_steps_per_second": 2.618,
|
| 997 |
+
"eval_wer": 0.7027027027027027,
|
| 998 |
+
"step": 6600
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 279.17,
|
| 1002 |
+
"learning_rate": 3.615681818181818e-05,
|
| 1003 |
+
"loss": 0.1198,
|
| 1004 |
+
"step": 6700
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 279.17,
|
| 1008 |
+
"eval_loss": 1.564064383506775,
|
| 1009 |
+
"eval_runtime": 6.7671,
|
| 1010 |
+
"eval_samples_per_second": 20.984,
|
| 1011 |
+
"eval_steps_per_second": 2.66,
|
| 1012 |
+
"eval_wer": 0.7000871839581517,
|
| 1013 |
+
"step": 6700
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 283.33,
|
| 1017 |
+
"learning_rate": 3.5474999999999995e-05,
|
| 1018 |
+
"loss": 0.1242,
|
| 1019 |
+
"step": 6800
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 283.33,
|
| 1023 |
+
"eval_loss": 1.5390304327011108,
|
| 1024 |
+
"eval_runtime": 6.7114,
|
| 1025 |
+
"eval_samples_per_second": 21.158,
|
| 1026 |
+
"eval_steps_per_second": 2.682,
|
| 1027 |
+
"eval_wer": 0.7061900610287707,
|
| 1028 |
+
"step": 6800
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 287.5,
|
| 1032 |
+
"learning_rate": 3.479318181818181e-05,
|
| 1033 |
+
"loss": 0.12,
|
| 1034 |
+
"step": 6900
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 287.5,
|
| 1038 |
+
"eval_loss": 1.540635585784912,
|
| 1039 |
+
"eval_runtime": 6.7141,
|
| 1040 |
+
"eval_samples_per_second": 21.149,
|
| 1041 |
+
"eval_steps_per_second": 2.681,
|
| 1042 |
+
"eval_wer": 0.7105492589363557,
|
| 1043 |
+
"step": 6900
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 291.67,
|
| 1047 |
+
"learning_rate": 3.4111363636363634e-05,
|
| 1048 |
+
"loss": 0.1096,
|
| 1049 |
+
"step": 7000
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 291.67,
|
| 1053 |
+
"eval_loss": 1.5737296342849731,
|
| 1054 |
+
"eval_runtime": 6.6657,
|
| 1055 |
+
"eval_samples_per_second": 21.303,
|
| 1056 |
+
"eval_steps_per_second": 2.7,
|
| 1057 |
+
"eval_wer": 0.6974716652136007,
|
| 1058 |
+
"step": 7000
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 295.83,
|
| 1062 |
+
"learning_rate": 3.342954545454545e-05,
|
| 1063 |
+
"loss": 0.113,
|
| 1064 |
+
"step": 7100
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 295.83,
|
| 1068 |
+
"eval_loss": 1.5494580268859863,
|
| 1069 |
+
"eval_runtime": 6.8629,
|
| 1070 |
+
"eval_samples_per_second": 20.691,
|
| 1071 |
+
"eval_steps_per_second": 2.623,
|
| 1072 |
+
"eval_wer": 0.7210113339145597,
|
| 1073 |
+
"step": 7100
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 300.0,
|
| 1077 |
+
"learning_rate": 3.2747727272727274e-05,
|
| 1078 |
+
"loss": 0.108,
|
| 1079 |
+
"step": 7200
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 300.0,
|
| 1083 |
+
"eval_loss": 1.5374633073806763,
|
| 1084 |
+
"eval_runtime": 7.1279,
|
| 1085 |
+
"eval_samples_per_second": 19.922,
|
| 1086 |
+
"eval_steps_per_second": 2.525,
|
| 1087 |
+
"eval_wer": 0.6948561464690497,
|
| 1088 |
+
"step": 7200
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 304.17,
|
| 1092 |
+
"learning_rate": 3.207272727272727e-05,
|
| 1093 |
+
"loss": 0.1072,
|
| 1094 |
+
"step": 7300
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 304.17,
|
| 1098 |
+
"eval_loss": 1.5337363481521606,
|
| 1099 |
+
"eval_runtime": 7.1049,
|
| 1100 |
+
"eval_samples_per_second": 19.986,
|
| 1101 |
+
"eval_steps_per_second": 2.533,
|
| 1102 |
+
"eval_wer": 0.7009590235396687,
|
| 1103 |
+
"step": 7300
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 308.33,
|
| 1107 |
+
"learning_rate": 3.139090909090909e-05,
|
| 1108 |
+
"loss": 0.0979,
|
| 1109 |
+
"step": 7400
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 308.33,
|
| 1113 |
+
"eval_loss": 1.5927180051803589,
|
| 1114 |
+
"eval_runtime": 6.6123,
|
| 1115 |
+
"eval_samples_per_second": 21.475,
|
| 1116 |
+
"eval_steps_per_second": 2.722,
|
| 1117 |
+
"eval_wer": 0.7061900610287707,
|
| 1118 |
+
"step": 7400
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 312.5,
|
| 1122 |
+
"learning_rate": 3.0709090909090904e-05,
|
| 1123 |
+
"loss": 0.0983,
|
| 1124 |
+
"step": 7500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 312.5,
|
| 1128 |
+
"eval_loss": 1.5882079601287842,
|
| 1129 |
+
"eval_runtime": 6.7832,
|
| 1130 |
+
"eval_samples_per_second": 20.934,
|
| 1131 |
+
"eval_steps_per_second": 2.654,
|
| 1132 |
+
"eval_wer": 0.6843940714908456,
|
| 1133 |
+
"step": 7500
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 316.67,
|
| 1137 |
+
"learning_rate": 3.0027272727272724e-05,
|
| 1138 |
+
"loss": 0.0977,
|
| 1139 |
+
"step": 7600
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 316.67,
|
| 1143 |
+
"eval_loss": 1.6189255714416504,
|
| 1144 |
+
"eval_runtime": 6.9471,
|
| 1145 |
+
"eval_samples_per_second": 20.44,
|
| 1146 |
+
"eval_steps_per_second": 2.591,
|
| 1147 |
+
"eval_wer": 0.6957279860505667,
|
| 1148 |
+
"step": 7600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 320.83,
|
| 1152 |
+
"learning_rate": 2.9352272727272724e-05,
|
| 1153 |
+
"loss": 0.0947,
|
| 1154 |
+
"step": 7700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 320.83,
|
| 1158 |
+
"eval_loss": 1.5098397731781006,
|
| 1159 |
+
"eval_runtime": 6.6241,
|
| 1160 |
+
"eval_samples_per_second": 21.437,
|
| 1161 |
+
"eval_steps_per_second": 2.717,
|
| 1162 |
+
"eval_wer": 0.6817785527462947,
|
| 1163 |
+
"step": 7700
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 325.0,
|
| 1167 |
+
"learning_rate": 2.867045454545454e-05,
|
| 1168 |
+
"loss": 0.0996,
|
| 1169 |
+
"step": 7800
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 325.0,
|
| 1173 |
+
"eval_loss": 1.6268917322158813,
|
| 1174 |
+
"eval_runtime": 6.6505,
|
| 1175 |
+
"eval_samples_per_second": 21.352,
|
| 1176 |
+
"eval_steps_per_second": 2.707,
|
| 1177 |
+
"eval_wer": 0.7253705318221447,
|
| 1178 |
+
"step": 7800
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 329.17,
|
| 1182 |
+
"learning_rate": 2.798863636363636e-05,
|
| 1183 |
+
"loss": 0.0846,
|
| 1184 |
+
"step": 7900
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 329.17,
|
| 1188 |
+
"eval_loss": 1.6366547346115112,
|
| 1189 |
+
"eval_runtime": 6.9003,
|
| 1190 |
+
"eval_samples_per_second": 20.579,
|
| 1191 |
+
"eval_steps_per_second": 2.609,
|
| 1192 |
+
"eval_wer": 0.7088055797733217,
|
| 1193 |
+
"step": 7900
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 333.33,
|
| 1197 |
+
"learning_rate": 2.7306818181818177e-05,
|
| 1198 |
+
"loss": 0.0953,
|
| 1199 |
+
"step": 8000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 333.33,
|
| 1203 |
+
"eval_loss": 1.5965033769607544,
|
| 1204 |
+
"eval_runtime": 7.0137,
|
| 1205 |
+
"eval_samples_per_second": 20.246,
|
| 1206 |
+
"eval_steps_per_second": 2.566,
|
| 1207 |
+
"eval_wer": 0.7122929380993898,
|
| 1208 |
+
"step": 8000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 337.5,
|
| 1212 |
+
"learning_rate": 2.6624999999999997e-05,
|
| 1213 |
+
"loss": 0.0906,
|
| 1214 |
+
"step": 8100
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 337.5,
|
| 1218 |
+
"eval_loss": 1.6095737218856812,
|
| 1219 |
+
"eval_runtime": 6.5521,
|
| 1220 |
+
"eval_samples_per_second": 21.672,
|
| 1221 |
+
"eval_steps_per_second": 2.747,
|
| 1222 |
+
"eval_wer": 0.7122929380993898,
|
| 1223 |
+
"step": 8100
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 341.67,
|
| 1227 |
+
"learning_rate": 2.5943181818181813e-05,
|
| 1228 |
+
"loss": 0.093,
|
| 1229 |
+
"step": 8200
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 341.67,
|
| 1233 |
+
"eval_loss": 1.5953401327133179,
|
| 1234 |
+
"eval_runtime": 6.9512,
|
| 1235 |
+
"eval_samples_per_second": 20.428,
|
| 1236 |
+
"eval_steps_per_second": 2.589,
|
| 1237 |
+
"eval_wer": 0.6983435047951178,
|
| 1238 |
+
"step": 8200
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 345.83,
|
| 1242 |
+
"learning_rate": 2.5261363636363633e-05,
|
| 1243 |
+
"loss": 0.0784,
|
| 1244 |
+
"step": 8300
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 345.83,
|
| 1248 |
+
"eval_loss": 1.5884095430374146,
|
| 1249 |
+
"eval_runtime": 6.8928,
|
| 1250 |
+
"eval_samples_per_second": 20.601,
|
| 1251 |
+
"eval_steps_per_second": 2.611,
|
| 1252 |
+
"eval_wer": 0.6913687881429816,
|
| 1253 |
+
"step": 8300
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 350.0,
|
| 1257 |
+
"learning_rate": 2.457954545454545e-05,
|
| 1258 |
+
"loss": 0.0769,
|
| 1259 |
+
"step": 8400
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 350.0,
|
| 1263 |
+
"eval_loss": 1.5793629884719849,
|
| 1264 |
+
"eval_runtime": 6.8179,
|
| 1265 |
+
"eval_samples_per_second": 20.828,
|
| 1266 |
+
"eval_steps_per_second": 2.64,
|
| 1267 |
+
"eval_wer": 0.6870095902353966,
|
| 1268 |
+
"step": 8400
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 354.17,
|
| 1272 |
+
"learning_rate": 2.3897727272727272e-05,
|
| 1273 |
+
"loss": 0.0782,
|
| 1274 |
+
"step": 8500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 354.17,
|
| 1278 |
+
"eval_loss": 1.6580848693847656,
|
| 1279 |
+
"eval_runtime": 6.9462,
|
| 1280 |
+
"eval_samples_per_second": 20.443,
|
| 1281 |
+
"eval_steps_per_second": 2.591,
|
| 1282 |
+
"eval_wer": 0.6817785527462947,
|
| 1283 |
+
"step": 8500
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 358.33,
|
| 1287 |
+
"learning_rate": 2.3215909090909092e-05,
|
| 1288 |
+
"loss": 0.0764,
|
| 1289 |
+
"step": 8600
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 358.33,
|
| 1293 |
+
"eval_loss": 1.6554986238479614,
|
| 1294 |
+
"eval_runtime": 7.0359,
|
| 1295 |
+
"eval_samples_per_second": 20.182,
|
| 1296 |
+
"eval_steps_per_second": 2.558,
|
| 1297 |
+
"eval_wer": 0.7088055797733217,
|
| 1298 |
+
"step": 8600
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 362.5,
|
| 1302 |
+
"learning_rate": 2.253409090909091e-05,
|
| 1303 |
+
"loss": 0.073,
|
| 1304 |
+
"step": 8700
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 362.5,
|
| 1308 |
+
"eval_loss": 1.6465544700622559,
|
| 1309 |
+
"eval_runtime": 6.8477,
|
| 1310 |
+
"eval_samples_per_second": 20.737,
|
| 1311 |
+
"eval_steps_per_second": 2.629,
|
| 1312 |
+
"eval_wer": 0.6931124673060157,
|
| 1313 |
+
"step": 8700
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 366.67,
|
| 1317 |
+
"learning_rate": 2.1852272727272725e-05,
|
| 1318 |
+
"loss": 0.0703,
|
| 1319 |
+
"step": 8800
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 366.67,
|
| 1323 |
+
"eval_loss": 1.6614816188812256,
|
| 1324 |
+
"eval_runtime": 7.0115,
|
| 1325 |
+
"eval_samples_per_second": 20.252,
|
| 1326 |
+
"eval_steps_per_second": 2.567,
|
| 1327 |
+
"eval_wer": 0.7114210985178727,
|
| 1328 |
+
"step": 8800
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 370.83,
|
| 1332 |
+
"learning_rate": 2.1170454545454545e-05,
|
| 1333 |
+
"loss": 0.0707,
|
| 1334 |
+
"step": 8900
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 370.83,
|
| 1338 |
+
"eval_loss": 1.6742826700210571,
|
| 1339 |
+
"eval_runtime": 6.7568,
|
| 1340 |
+
"eval_samples_per_second": 21.016,
|
| 1341 |
+
"eval_steps_per_second": 2.664,
|
| 1342 |
+
"eval_wer": 0.7079337401918047,
|
| 1343 |
+
"step": 8900
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 375.0,
|
| 1347 |
+
"learning_rate": 2.048863636363636e-05,
|
| 1348 |
+
"loss": 0.0647,
|
| 1349 |
+
"step": 9000
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 375.0,
|
| 1353 |
+
"eval_loss": 1.6451914310455322,
|
| 1354 |
+
"eval_runtime": 6.781,
|
| 1355 |
+
"eval_samples_per_second": 20.941,
|
| 1356 |
+
"eval_steps_per_second": 2.654,
|
| 1357 |
+
"eval_wer": 0.7166521360069747,
|
| 1358 |
+
"step": 9000
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 379.17,
|
| 1362 |
+
"learning_rate": 1.980681818181818e-05,
|
| 1363 |
+
"loss": 0.0614,
|
| 1364 |
+
"step": 9100
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 379.17,
|
| 1368 |
+
"eval_loss": 1.7081646919250488,
|
| 1369 |
+
"eval_runtime": 6.6825,
|
| 1370 |
+
"eval_samples_per_second": 21.25,
|
| 1371 |
+
"eval_steps_per_second": 2.694,
|
| 1372 |
+
"eval_wer": 0.7122929380993898,
|
| 1373 |
+
"step": 9100
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 383.33,
|
| 1377 |
+
"learning_rate": 1.9124999999999997e-05,
|
| 1378 |
+
"loss": 0.0646,
|
| 1379 |
+
"step": 9200
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 383.33,
|
| 1383 |
+
"eval_loss": 1.684810757637024,
|
| 1384 |
+
"eval_runtime": 6.7457,
|
| 1385 |
+
"eval_samples_per_second": 21.05,
|
| 1386 |
+
"eval_steps_per_second": 2.668,
|
| 1387 |
+
"eval_wer": 0.7183958151700087,
|
| 1388 |
+
"step": 9200
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 387.5,
|
| 1392 |
+
"learning_rate": 1.8443181818181817e-05,
|
| 1393 |
+
"loss": 0.0648,
|
| 1394 |
+
"step": 9300
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 387.5,
|
| 1398 |
+
"eval_loss": 1.65809166431427,
|
| 1399 |
+
"eval_runtime": 6.5015,
|
| 1400 |
+
"eval_samples_per_second": 21.841,
|
| 1401 |
+
"eval_steps_per_second": 2.769,
|
| 1402 |
+
"eval_wer": 0.7088055797733217,
|
| 1403 |
+
"step": 9300
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 391.67,
|
| 1407 |
+
"learning_rate": 1.7761363636363633e-05,
|
| 1408 |
+
"loss": 0.0625,
|
| 1409 |
+
"step": 9400
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 391.67,
|
| 1413 |
+
"eval_loss": 1.7315229177474976,
|
| 1414 |
+
"eval_runtime": 7.2222,
|
| 1415 |
+
"eval_samples_per_second": 19.662,
|
| 1416 |
+
"eval_steps_per_second": 2.492,
|
| 1417 |
+
"eval_wer": 0.7340889276373147,
|
| 1418 |
+
"step": 9400
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 395.83,
|
| 1422 |
+
"learning_rate": 1.7079545454545453e-05,
|
| 1423 |
+
"loss": 0.0637,
|
| 1424 |
+
"step": 9500
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 395.83,
|
| 1428 |
+
"eval_loss": 1.683098316192627,
|
| 1429 |
+
"eval_runtime": 6.8952,
|
| 1430 |
+
"eval_samples_per_second": 20.594,
|
| 1431 |
+
"eval_steps_per_second": 2.611,
|
| 1432 |
+
"eval_wer": 0.7027027027027027,
|
| 1433 |
+
"step": 9500
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 400.0,
|
| 1437 |
+
"learning_rate": 1.639772727272727e-05,
|
| 1438 |
+
"loss": 0.0558,
|
| 1439 |
+
"step": 9600
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 400.0,
|
| 1443 |
+
"eval_loss": 1.7159340381622314,
|
| 1444 |
+
"eval_runtime": 6.6968,
|
| 1445 |
+
"eval_samples_per_second": 21.204,
|
| 1446 |
+
"eval_steps_per_second": 2.688,
|
| 1447 |
+
"eval_wer": 0.7279860505666957,
|
| 1448 |
+
"step": 9600
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 404.17,
|
| 1452 |
+
"learning_rate": 1.571590909090909e-05,
|
| 1453 |
+
"loss": 0.0563,
|
| 1454 |
+
"step": 9700
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 404.17,
|
| 1458 |
+
"eval_loss": 1.7474530935287476,
|
| 1459 |
+
"eval_runtime": 6.7706,
|
| 1460 |
+
"eval_samples_per_second": 20.973,
|
| 1461 |
+
"eval_steps_per_second": 2.659,
|
| 1462 |
+
"eval_wer": 0.7157802964254577,
|
| 1463 |
+
"step": 9700
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 408.33,
|
| 1467 |
+
"learning_rate": 1.5034090909090908e-05,
|
| 1468 |
+
"loss": 0.0568,
|
| 1469 |
+
"step": 9800
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 408.33,
|
| 1473 |
+
"eval_loss": 1.6776412725448608,
|
| 1474 |
+
"eval_runtime": 6.9524,
|
| 1475 |
+
"eval_samples_per_second": 20.425,
|
| 1476 |
+
"eval_steps_per_second": 2.589,
|
| 1477 |
+
"eval_wer": 0.6992153443766347,
|
| 1478 |
+
"step": 9800
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 412.5,
|
| 1482 |
+
"learning_rate": 1.4352272727272727e-05,
|
| 1483 |
+
"loss": 0.0574,
|
| 1484 |
+
"step": 9900
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 412.5,
|
| 1488 |
+
"eval_loss": 1.7150009870529175,
|
| 1489 |
+
"eval_runtime": 6.8865,
|
| 1490 |
+
"eval_samples_per_second": 20.62,
|
| 1491 |
+
"eval_steps_per_second": 2.614,
|
| 1492 |
+
"eval_wer": 0.6983435047951178,
|
| 1493 |
+
"step": 9900
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 416.67,
|
| 1497 |
+
"learning_rate": 1.3670454545454545e-05,
|
| 1498 |
+
"loss": 0.0561,
|
| 1499 |
+
"step": 10000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 416.67,
|
| 1503 |
+
"eval_loss": 1.7315118312835693,
|
| 1504 |
+
"eval_runtime": 6.8566,
|
| 1505 |
+
"eval_samples_per_second": 20.71,
|
| 1506 |
+
"eval_steps_per_second": 2.625,
|
| 1507 |
+
"eval_wer": 0.7140366172624237,
|
| 1508 |
+
"step": 10000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 420.83,
|
| 1512 |
+
"learning_rate": 1.2988636363636363e-05,
|
| 1513 |
+
"loss": 0.0494,
|
| 1514 |
+
"step": 10100
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 420.83,
|
| 1518 |
+
"eval_loss": 1.6868910789489746,
|
| 1519 |
+
"eval_runtime": 6.9481,
|
| 1520 |
+
"eval_samples_per_second": 20.437,
|
| 1521 |
+
"eval_steps_per_second": 2.591,
|
| 1522 |
+
"eval_wer": 0.7218831734960767,
|
| 1523 |
+
"step": 10100
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 425.0,
|
| 1527 |
+
"learning_rate": 1.2306818181818182e-05,
|
| 1528 |
+
"loss": 0.0495,
|
| 1529 |
+
"step": 10200
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 425.0,
|
| 1533 |
+
"eval_loss": 1.749950885772705,
|
| 1534 |
+
"eval_runtime": 6.8281,
|
| 1535 |
+
"eval_samples_per_second": 20.796,
|
| 1536 |
+
"eval_steps_per_second": 2.636,
|
| 1537 |
+
"eval_wer": 0.7262423714036618,
|
| 1538 |
+
"step": 10200
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 429.17,
|
| 1542 |
+
"learning_rate": 1.1625e-05,
|
| 1543 |
+
"loss": 0.0542,
|
| 1544 |
+
"step": 10300
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 429.17,
|
| 1548 |
+
"eval_loss": 1.7298214435577393,
|
| 1549 |
+
"eval_runtime": 6.7073,
|
| 1550 |
+
"eval_samples_per_second": 21.171,
|
| 1551 |
+
"eval_steps_per_second": 2.684,
|
| 1552 |
+
"eval_wer": 0.7271142109851787,
|
| 1553 |
+
"step": 10300
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 433.33,
|
| 1557 |
+
"learning_rate": 1.0943181818181818e-05,
|
| 1558 |
+
"loss": 0.0509,
|
| 1559 |
+
"step": 10400
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 433.33,
|
| 1563 |
+
"eval_loss": 1.7334100008010864,
|
| 1564 |
+
"eval_runtime": 6.7752,
|
| 1565 |
+
"eval_samples_per_second": 20.959,
|
| 1566 |
+
"eval_steps_per_second": 2.657,
|
| 1567 |
+
"eval_wer": 0.7262423714036618,
|
| 1568 |
+
"step": 10400
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 437.5,
|
| 1572 |
+
"learning_rate": 1.0268181818181817e-05,
|
| 1573 |
+
"loss": 0.046,
|
| 1574 |
+
"step": 10500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 437.5,
|
| 1578 |
+
"eval_loss": 1.7047654390335083,
|
| 1579 |
+
"eval_runtime": 6.3102,
|
| 1580 |
+
"eval_samples_per_second": 22.503,
|
| 1581 |
+
"eval_steps_per_second": 2.853,
|
| 1582 |
+
"eval_wer": 0.7192676547515258,
|
| 1583 |
+
"step": 10500
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 441.67,
|
| 1587 |
+
"learning_rate": 9.586363636363636e-06,
|
| 1588 |
+
"loss": 0.0423,
|
| 1589 |
+
"step": 10600
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 441.67,
|
| 1593 |
+
"eval_loss": 1.716819167137146,
|
| 1594 |
+
"eval_runtime": 7.0132,
|
| 1595 |
+
"eval_samples_per_second": 20.247,
|
| 1596 |
+
"eval_steps_per_second": 2.567,
|
| 1597 |
+
"eval_wer": 0.7192676547515258,
|
| 1598 |
+
"step": 10600
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 445.83,
|
| 1602 |
+
"learning_rate": 8.904545454545453e-06,
|
| 1603 |
+
"loss": 0.0477,
|
| 1604 |
+
"step": 10700
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 445.83,
|
| 1608 |
+
"eval_loss": 1.7387615442276,
|
| 1609 |
+
"eval_runtime": 6.7039,
|
| 1610 |
+
"eval_samples_per_second": 21.182,
|
| 1611 |
+
"eval_steps_per_second": 2.685,
|
| 1612 |
+
"eval_wer": 0.7210113339145597,
|
| 1613 |
+
"step": 10700
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 450.0,
|
| 1617 |
+
"learning_rate": 8.222727272727273e-06,
|
| 1618 |
+
"loss": 0.0436,
|
| 1619 |
+
"step": 10800
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 450.0,
|
| 1623 |
+
"eval_loss": 1.7278592586517334,
|
| 1624 |
+
"eval_runtime": 6.9747,
|
| 1625 |
+
"eval_samples_per_second": 20.359,
|
| 1626 |
+
"eval_steps_per_second": 2.581,
|
| 1627 |
+
"eval_wer": 0.7166521360069747,
|
| 1628 |
+
"step": 10800
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 454.17,
|
| 1632 |
+
"learning_rate": 7.540909090909091e-06,
|
| 1633 |
+
"loss": 0.0466,
|
| 1634 |
+
"step": 10900
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 454.17,
|
| 1638 |
+
"eval_loss": 1.696805715560913,
|
| 1639 |
+
"eval_runtime": 6.8023,
|
| 1640 |
+
"eval_samples_per_second": 20.875,
|
| 1641 |
+
"eval_steps_per_second": 2.646,
|
| 1642 |
+
"eval_wer": 0.7053182214472538,
|
| 1643 |
+
"step": 10900
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 458.33,
|
| 1647 |
+
"learning_rate": 6.859090909090909e-06,
|
| 1648 |
+
"loss": 0.0424,
|
| 1649 |
+
"step": 11000
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 458.33,
|
| 1653 |
+
"eval_loss": 1.7237237691879272,
|
| 1654 |
+
"eval_runtime": 7.027,
|
| 1655 |
+
"eval_samples_per_second": 20.208,
|
| 1656 |
+
"eval_steps_per_second": 2.562,
|
| 1657 |
+
"eval_wer": 0.7183958151700087,
|
| 1658 |
+
"step": 11000
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 462.5,
|
| 1662 |
+
"learning_rate": 6.177272727272727e-06,
|
| 1663 |
+
"loss": 0.0447,
|
| 1664 |
+
"step": 11100
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 462.5,
|
| 1668 |
+
"eval_loss": 1.721848726272583,
|
| 1669 |
+
"eval_runtime": 6.7393,
|
| 1670 |
+
"eval_samples_per_second": 21.07,
|
| 1671 |
+
"eval_steps_per_second": 2.671,
|
| 1672 |
+
"eval_wer": 0.7183958151700087,
|
| 1673 |
+
"step": 11100
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 466.67,
|
| 1677 |
+
"learning_rate": 5.495454545454545e-06,
|
| 1678 |
+
"loss": 0.0455,
|
| 1679 |
+
"step": 11200
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 466.67,
|
| 1683 |
+
"eval_loss": 1.7505738735198975,
|
| 1684 |
+
"eval_runtime": 6.8936,
|
| 1685 |
+
"eval_samples_per_second": 20.599,
|
| 1686 |
+
"eval_steps_per_second": 2.611,
|
| 1687 |
+
"eval_wer": 0.7218831734960767,
|
| 1688 |
+
"step": 11200
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 470.83,
|
| 1692 |
+
"learning_rate": 4.813636363636364e-06,
|
| 1693 |
+
"loss": 0.0446,
|
| 1694 |
+
"step": 11300
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 470.83,
|
| 1698 |
+
"eval_loss": 1.7541626691818237,
|
| 1699 |
+
"eval_runtime": 6.9153,
|
| 1700 |
+
"eval_samples_per_second": 20.534,
|
| 1701 |
+
"eval_steps_per_second": 2.603,
|
| 1702 |
+
"eval_wer": 0.7279860505666957,
|
| 1703 |
+
"step": 11300
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 475.0,
|
| 1707 |
+
"learning_rate": 4.131818181818182e-06,
|
| 1708 |
+
"loss": 0.043,
|
| 1709 |
+
"step": 11400
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 475.0,
|
| 1713 |
+
"eval_loss": 1.750115156173706,
|
| 1714 |
+
"eval_runtime": 6.8671,
|
| 1715 |
+
"eval_samples_per_second": 20.678,
|
| 1716 |
+
"eval_steps_per_second": 2.621,
|
| 1717 |
+
"eval_wer": 0.7201394943330427,
|
| 1718 |
+
"step": 11400
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 479.17,
|
| 1722 |
+
"learning_rate": 3.4499999999999996e-06,
|
| 1723 |
+
"loss": 0.0397,
|
| 1724 |
+
"step": 11500
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 479.17,
|
| 1728 |
+
"eval_loss": 1.783726692199707,
|
| 1729 |
+
"eval_runtime": 6.759,
|
| 1730 |
+
"eval_samples_per_second": 21.009,
|
| 1731 |
+
"eval_steps_per_second": 2.663,
|
| 1732 |
+
"eval_wer": 0.7244986922406277,
|
| 1733 |
+
"step": 11500
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 483.33,
|
| 1737 |
+
"learning_rate": 2.768181818181818e-06,
|
| 1738 |
+
"loss": 0.0402,
|
| 1739 |
+
"step": 11600
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 483.33,
|
| 1743 |
+
"eval_loss": 1.7762095928192139,
|
| 1744 |
+
"eval_runtime": 6.8013,
|
| 1745 |
+
"eval_samples_per_second": 20.878,
|
| 1746 |
+
"eval_steps_per_second": 2.647,
|
| 1747 |
+
"eval_wer": 0.7175239755884917,
|
| 1748 |
+
"step": 11600
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 487.5,
|
| 1752 |
+
"learning_rate": 2.086363636363636e-06,
|
| 1753 |
+
"loss": 0.039,
|
| 1754 |
+
"step": 11700
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 487.5,
|
| 1758 |
+
"eval_loss": 1.7771002054214478,
|
| 1759 |
+
"eval_runtime": 6.9246,
|
| 1760 |
+
"eval_samples_per_second": 20.507,
|
| 1761 |
+
"eval_steps_per_second": 2.599,
|
| 1762 |
+
"eval_wer": 0.7262423714036618,
|
| 1763 |
+
"step": 11700
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 491.67,
|
| 1767 |
+
"learning_rate": 1.4045454545454545e-06,
|
| 1768 |
+
"loss": 0.0402,
|
| 1769 |
+
"step": 11800
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 491.67,
|
| 1773 |
+
"eval_loss": 1.7563738822937012,
|
| 1774 |
+
"eval_runtime": 7.1476,
|
| 1775 |
+
"eval_samples_per_second": 19.867,
|
| 1776 |
+
"eval_steps_per_second": 2.518,
|
| 1777 |
+
"eval_wer": 0.7218831734960767,
|
| 1778 |
+
"step": 11800
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 495.83,
|
| 1782 |
+
"learning_rate": 7.227272727272726e-07,
|
| 1783 |
+
"loss": 0.0368,
|
| 1784 |
+
"step": 11900
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 495.83,
|
| 1788 |
+
"eval_loss": 1.7552615404129028,
|
| 1789 |
+
"eval_runtime": 6.927,
|
| 1790 |
+
"eval_samples_per_second": 20.499,
|
| 1791 |
+
"eval_steps_per_second": 2.599,
|
| 1792 |
+
"eval_wer": 0.7192676547515258,
|
| 1793 |
+
"step": 11900
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 500.0,
|
| 1797 |
+
"learning_rate": 4.090909090909091e-08,
|
| 1798 |
+
"loss": 0.0395,
|
| 1799 |
+
"step": 12000
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 500.0,
|
| 1803 |
+
"eval_loss": 1.7564287185668945,
|
| 1804 |
+
"eval_runtime": 6.7877,
|
| 1805 |
+
"eval_samples_per_second": 20.92,
|
| 1806 |
+
"eval_steps_per_second": 2.652,
|
| 1807 |
+
"eval_wer": 0.7201394943330427,
|
| 1808 |
+
"step": 12000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 500.0,
|
| 1812 |
+
"step": 12000,
|
| 1813 |
+
"total_flos": 2.5005546169759453e+19,
|
| 1814 |
+
"train_loss": 0.7251964689095814,
|
| 1815 |
+
"train_runtime": 16734.3269,
|
| 1816 |
+
"train_samples_per_second": 11.294,
|
| 1817 |
+
"train_steps_per_second": 0.717
|
| 1818 |
+
}
|
| 1819 |
+
],
|
| 1820 |
+
"max_steps": 12000,
|
| 1821 |
+
"num_train_epochs": 500,
|
| 1822 |
+
"total_flos": 2.5005546169759453e+19,
|
| 1823 |
+
"trial_name": null,
|
| 1824 |
+
"trial_params": null
|
| 1825 |
+
}
|