End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +3046 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.15054723620414734,
|
| 4 |
+
"eval_runtime": 688.169,
|
| 5 |
+
"eval_samples": 8301,
|
| 6 |
+
"eval_samples_per_second": 12.062,
|
| 7 |
+
"eval_steps_per_second": 1.508,
|
| 8 |
+
"eval_wer": 0.13517827708857466,
|
| 9 |
+
"train_loss": 0.9556344533304143,
|
| 10 |
+
"train_runtime": 205033.5418,
|
| 11 |
+
"train_samples": 24877,
|
| 12 |
+
"train_samples_per_second": 6.067,
|
| 13 |
+
"train_steps_per_second": 0.189
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.15054723620414734,
|
| 4 |
+
"eval_runtime": 688.169,
|
| 5 |
+
"eval_samples": 8301,
|
| 6 |
+
"eval_samples_per_second": 12.062,
|
| 7 |
+
"eval_steps_per_second": 1.508,
|
| 8 |
+
"eval_wer": 0.13517827708857466
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"train_loss": 0.9556344533304143,
|
| 4 |
+
"train_runtime": 205033.5418,
|
| 5 |
+
"train_samples": 24877,
|
| 6 |
+
"train_samples_per_second": 6.067,
|
| 7 |
+
"train_steps_per_second": 0.189
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3046 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 49.99935691318328,
|
| 5 |
+
"global_step": 38850,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.13,
|
| 12 |
+
"learning_rate": 3.675e-06,
|
| 13 |
+
"loss": 21.0125,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.26,
|
| 18 |
+
"learning_rate": 7.425e-06,
|
| 19 |
+
"loss": 10.9603,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.39,
|
| 24 |
+
"learning_rate": 1.1174999999999999e-05,
|
| 25 |
+
"loss": 6.2714,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.51,
|
| 30 |
+
"learning_rate": 1.4925e-05,
|
| 31 |
+
"loss": 5.0154,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.64,
|
| 36 |
+
"learning_rate": 1.8675e-05,
|
| 37 |
+
"loss": 4.1367,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.64,
|
| 42 |
+
"eval_loss": 3.88252592086792,
|
| 43 |
+
"eval_runtime": 666.8464,
|
| 44 |
+
"eval_samples_per_second": 12.448,
|
| 45 |
+
"eval_steps_per_second": 1.557,
|
| 46 |
+
"eval_wer": 1.0,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.77,
|
| 51 |
+
"learning_rate": 2.2424999999999996e-05,
|
| 52 |
+
"loss": 3.5318,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.9,
|
| 57 |
+
"learning_rate": 2.6174999999999996e-05,
|
| 58 |
+
"loss": 3.2175,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.03,
|
| 63 |
+
"learning_rate": 2.9925e-05,
|
| 64 |
+
"loss": 3.1431,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.16,
|
| 69 |
+
"learning_rate": 3.3675e-05,
|
| 70 |
+
"loss": 3.1247,
|
| 71 |
+
"step": 900
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.29,
|
| 75 |
+
"learning_rate": 3.7424999999999995e-05,
|
| 76 |
+
"loss": 2.9677,
|
| 77 |
+
"step": 1000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.29,
|
| 81 |
+
"eval_loss": 2.949765205383301,
|
| 82 |
+
"eval_runtime": 679.306,
|
| 83 |
+
"eval_samples_per_second": 12.22,
|
| 84 |
+
"eval_steps_per_second": 1.528,
|
| 85 |
+
"eval_wer": 1.0,
|
| 86 |
+
"step": 1000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.42,
|
| 90 |
+
"learning_rate": 4.1175e-05,
|
| 91 |
+
"loss": 2.9224,
|
| 92 |
+
"step": 1100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.54,
|
| 96 |
+
"learning_rate": 4.4924999999999994e-05,
|
| 97 |
+
"loss": 2.8662,
|
| 98 |
+
"step": 1200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.67,
|
| 102 |
+
"learning_rate": 4.8675e-05,
|
| 103 |
+
"loss": 2.591,
|
| 104 |
+
"step": 1300
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.8,
|
| 108 |
+
"learning_rate": 5.2424999999999994e-05,
|
| 109 |
+
"loss": 1.959,
|
| 110 |
+
"step": 1400
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.93,
|
| 114 |
+
"learning_rate": 5.6175e-05,
|
| 115 |
+
"loss": 1.5884,
|
| 116 |
+
"step": 1500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.93,
|
| 120 |
+
"eval_loss": 0.6722400188446045,
|
| 121 |
+
"eval_runtime": 671.1611,
|
| 122 |
+
"eval_samples_per_second": 12.368,
|
| 123 |
+
"eval_steps_per_second": 1.547,
|
| 124 |
+
"eval_wer": 0.6492582632081201,
|
| 125 |
+
"step": 1500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 2.06,
|
| 129 |
+
"learning_rate": 5.9925e-05,
|
| 130 |
+
"loss": 1.4568,
|
| 131 |
+
"step": 1600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 2.19,
|
| 135 |
+
"learning_rate": 6.367499999999999e-05,
|
| 136 |
+
"loss": 1.3268,
|
| 137 |
+
"step": 1700
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.32,
|
| 141 |
+
"learning_rate": 6.7425e-05,
|
| 142 |
+
"loss": 1.29,
|
| 143 |
+
"step": 1800
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.45,
|
| 147 |
+
"learning_rate": 7.1175e-05,
|
| 148 |
+
"loss": 1.2493,
|
| 149 |
+
"step": 1900
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.57,
|
| 153 |
+
"learning_rate": 7.492499999999999e-05,
|
| 154 |
+
"loss": 1.2292,
|
| 155 |
+
"step": 2000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 2.57,
|
| 159 |
+
"eval_loss": 0.3634721636772156,
|
| 160 |
+
"eval_runtime": 665.3111,
|
| 161 |
+
"eval_samples_per_second": 12.477,
|
| 162 |
+
"eval_steps_per_second": 1.56,
|
| 163 |
+
"eval_wer": 0.3202047367051271,
|
| 164 |
+
"step": 2000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.7,
|
| 168 |
+
"learning_rate": 7.480054274084124e-05,
|
| 169 |
+
"loss": 1.1888,
|
| 170 |
+
"step": 2100
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.83,
|
| 174 |
+
"learning_rate": 7.459905020352781e-05,
|
| 175 |
+
"loss": 1.168,
|
| 176 |
+
"step": 2200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.96,
|
| 180 |
+
"learning_rate": 7.43955223880597e-05,
|
| 181 |
+
"loss": 1.1837,
|
| 182 |
+
"step": 2300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.09,
|
| 186 |
+
"learning_rate": 7.419199457259158e-05,
|
| 187 |
+
"loss": 1.1336,
|
| 188 |
+
"step": 2400
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 3.22,
|
| 192 |
+
"learning_rate": 7.399050203527815e-05,
|
| 193 |
+
"loss": 1.1314,
|
| 194 |
+
"step": 2500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.22,
|
| 198 |
+
"eval_loss": 0.2969689965248108,
|
| 199 |
+
"eval_runtime": 670.1792,
|
| 200 |
+
"eval_samples_per_second": 12.386,
|
| 201 |
+
"eval_steps_per_second": 1.549,
|
| 202 |
+
"eval_wer": 0.2680142274659495,
|
| 203 |
+
"step": 2500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 3.35,
|
| 207 |
+
"learning_rate": 7.378697421981004e-05,
|
| 208 |
+
"loss": 1.1159,
|
| 209 |
+
"step": 2600
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 3.47,
|
| 213 |
+
"learning_rate": 7.358344640434192e-05,
|
| 214 |
+
"loss": 1.0976,
|
| 215 |
+
"step": 2700
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 3.6,
|
| 219 |
+
"learning_rate": 7.33799185888738e-05,
|
| 220 |
+
"loss": 1.102,
|
| 221 |
+
"step": 2800
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 3.73,
|
| 225 |
+
"learning_rate": 7.317639077340569e-05,
|
| 226 |
+
"loss": 1.0915,
|
| 227 |
+
"step": 2900
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 3.86,
|
| 231 |
+
"learning_rate": 7.297286295793758e-05,
|
| 232 |
+
"loss": 1.0879,
|
| 233 |
+
"step": 3000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.86,
|
| 237 |
+
"eval_loss": 0.2671239972114563,
|
| 238 |
+
"eval_runtime": 666.0789,
|
| 239 |
+
"eval_samples_per_second": 12.462,
|
| 240 |
+
"eval_steps_per_second": 1.558,
|
| 241 |
+
"eval_wer": 0.24856424047887568,
|
| 242 |
+
"step": 3000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 3.99,
|
| 246 |
+
"learning_rate": 7.276933514246946e-05,
|
| 247 |
+
"loss": 1.0838,
|
| 248 |
+
"step": 3100
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 4.12,
|
| 252 |
+
"learning_rate": 7.256580732700136e-05,
|
| 253 |
+
"loss": 1.0662,
|
| 254 |
+
"step": 3200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 4.25,
|
| 258 |
+
"learning_rate": 7.236227951153323e-05,
|
| 259 |
+
"loss": 1.0418,
|
| 260 |
+
"step": 3300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 4.38,
|
| 264 |
+
"learning_rate": 7.215875169606512e-05,
|
| 265 |
+
"loss": 1.0358,
|
| 266 |
+
"step": 3400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 4.5,
|
| 270 |
+
"learning_rate": 7.19572591587517e-05,
|
| 271 |
+
"loss": 1.0344,
|
| 272 |
+
"step": 3500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 4.5,
|
| 276 |
+
"eval_loss": 0.26251843571662903,
|
| 277 |
+
"eval_runtime": 663.0388,
|
| 278 |
+
"eval_samples_per_second": 12.52,
|
| 279 |
+
"eval_steps_per_second": 1.566,
|
| 280 |
+
"eval_wer": 0.2239264335907001,
|
| 281 |
+
"step": 3500
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 4.63,
|
| 285 |
+
"learning_rate": 7.175373134328358e-05,
|
| 286 |
+
"loss": 1.0395,
|
| 287 |
+
"step": 3600
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 4.76,
|
| 291 |
+
"learning_rate": 7.155020352781546e-05,
|
| 292 |
+
"loss": 1.031,
|
| 293 |
+
"step": 3700
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 4.89,
|
| 297 |
+
"learning_rate": 7.134667571234734e-05,
|
| 298 |
+
"loss": 1.016,
|
| 299 |
+
"step": 3800
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 5.02,
|
| 303 |
+
"learning_rate": 7.114314789687924e-05,
|
| 304 |
+
"loss": 1.0319,
|
| 305 |
+
"step": 3900
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 5.15,
|
| 309 |
+
"learning_rate": 7.093962008141112e-05,
|
| 310 |
+
"loss": 1.0109,
|
| 311 |
+
"step": 4000
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 5.15,
|
| 315 |
+
"eval_loss": 0.2520427405834198,
|
| 316 |
+
"eval_runtime": 660.1809,
|
| 317 |
+
"eval_samples_per_second": 12.574,
|
| 318 |
+
"eval_steps_per_second": 1.572,
|
| 319 |
+
"eval_wer": 0.22297215233798906,
|
| 320 |
+
"step": 4000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 5.28,
|
| 324 |
+
"learning_rate": 7.073609226594301e-05,
|
| 325 |
+
"loss": 1.0109,
|
| 326 |
+
"step": 4100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 5.41,
|
| 330 |
+
"learning_rate": 7.05325644504749e-05,
|
| 331 |
+
"loss": 1.0064,
|
| 332 |
+
"step": 4200
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 5.53,
|
| 336 |
+
"learning_rate": 7.032903663500678e-05,
|
| 337 |
+
"loss": 0.9985,
|
| 338 |
+
"step": 4300
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 5.66,
|
| 342 |
+
"learning_rate": 7.012550881953866e-05,
|
| 343 |
+
"loss": 1.0407,
|
| 344 |
+
"step": 4400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 5.79,
|
| 348 |
+
"learning_rate": 6.992198100407055e-05,
|
| 349 |
+
"loss": 0.9966,
|
| 350 |
+
"step": 4500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 5.79,
|
| 354 |
+
"eval_loss": 0.22795377671718597,
|
| 355 |
+
"eval_runtime": 658.6571,
|
| 356 |
+
"eval_samples_per_second": 12.603,
|
| 357 |
+
"eval_steps_per_second": 1.576,
|
| 358 |
+
"eval_wer": 0.2105144443480524,
|
| 359 |
+
"step": 4500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 5.92,
|
| 363 |
+
"learning_rate": 6.971845318860243e-05,
|
| 364 |
+
"loss": 0.9938,
|
| 365 |
+
"step": 4600
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 6.05,
|
| 369 |
+
"learning_rate": 6.951492537313432e-05,
|
| 370 |
+
"loss": 1.007,
|
| 371 |
+
"step": 4700
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 6.18,
|
| 375 |
+
"learning_rate": 6.93113975576662e-05,
|
| 376 |
+
"loss": 0.9835,
|
| 377 |
+
"step": 4800
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 6.31,
|
| 381 |
+
"learning_rate": 6.910786974219809e-05,
|
| 382 |
+
"loss": 0.9927,
|
| 383 |
+
"step": 4900
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 6.43,
|
| 387 |
+
"learning_rate": 6.890434192672997e-05,
|
| 388 |
+
"loss": 0.9815,
|
| 389 |
+
"step": 5000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 6.43,
|
| 393 |
+
"eval_loss": 0.225350484251976,
|
| 394 |
+
"eval_runtime": 716.6616,
|
| 395 |
+
"eval_samples_per_second": 11.583,
|
| 396 |
+
"eval_steps_per_second": 1.448,
|
| 397 |
+
"eval_wer": 0.2179057864145051,
|
| 398 |
+
"step": 5000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 6.56,
|
| 402 |
+
"learning_rate": 6.870081411126187e-05,
|
| 403 |
+
"loss": 0.9874,
|
| 404 |
+
"step": 5100
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 6.69,
|
| 408 |
+
"learning_rate": 6.849728629579376e-05,
|
| 409 |
+
"loss": 0.9856,
|
| 410 |
+
"step": 5200
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 6.82,
|
| 414 |
+
"learning_rate": 6.829375848032564e-05,
|
| 415 |
+
"loss": 0.9915,
|
| 416 |
+
"step": 5300
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 6.95,
|
| 420 |
+
"learning_rate": 6.809023066485753e-05,
|
| 421 |
+
"loss": 1.0046,
|
| 422 |
+
"step": 5400
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 7.08,
|
| 426 |
+
"learning_rate": 6.78887381275441e-05,
|
| 427 |
+
"loss": 0.9744,
|
| 428 |
+
"step": 5500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 7.08,
|
| 432 |
+
"eval_loss": 0.23012416064739227,
|
| 433 |
+
"eval_runtime": 684.7726,
|
| 434 |
+
"eval_samples_per_second": 12.122,
|
| 435 |
+
"eval_steps_per_second": 1.516,
|
| 436 |
+
"eval_wer": 0.21365489719788322,
|
| 437 |
+
"step": 5500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 7.21,
|
| 441 |
+
"learning_rate": 6.768724559023065e-05,
|
| 442 |
+
"loss": 0.9836,
|
| 443 |
+
"step": 5600
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 7.34,
|
| 447 |
+
"learning_rate": 6.748371777476255e-05,
|
| 448 |
+
"loss": 0.9752,
|
| 449 |
+
"step": 5700
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 7.46,
|
| 453 |
+
"learning_rate": 6.728018995929444e-05,
|
| 454 |
+
"loss": 0.9677,
|
| 455 |
+
"step": 5800
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 7.59,
|
| 459 |
+
"learning_rate": 6.707666214382632e-05,
|
| 460 |
+
"loss": 0.9678,
|
| 461 |
+
"step": 5900
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 7.72,
|
| 465 |
+
"learning_rate": 6.68731343283582e-05,
|
| 466 |
+
"loss": 0.9487,
|
| 467 |
+
"step": 6000
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 7.72,
|
| 471 |
+
"eval_loss": 0.22240765392780304,
|
| 472 |
+
"eval_runtime": 695.5316,
|
| 473 |
+
"eval_samples_per_second": 11.935,
|
| 474 |
+
"eval_steps_per_second": 1.492,
|
| 475 |
+
"eval_wer": 0.20506636592348398,
|
| 476 |
+
"step": 6000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 7.85,
|
| 480 |
+
"learning_rate": 6.666960651289009e-05,
|
| 481 |
+
"loss": 0.9644,
|
| 482 |
+
"step": 6100
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 7.98,
|
| 486 |
+
"learning_rate": 6.646607869742198e-05,
|
| 487 |
+
"loss": 0.9797,
|
| 488 |
+
"step": 6200
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 8.11,
|
| 492 |
+
"learning_rate": 6.626255088195386e-05,
|
| 493 |
+
"loss": 0.9697,
|
| 494 |
+
"step": 6300
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 8.24,
|
| 498 |
+
"learning_rate": 6.605902306648575e-05,
|
| 499 |
+
"loss": 0.9502,
|
| 500 |
+
"step": 6400
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 8.37,
|
| 504 |
+
"learning_rate": 6.585549525101763e-05,
|
| 505 |
+
"loss": 0.9431,
|
| 506 |
+
"step": 6500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 8.37,
|
| 510 |
+
"eval_loss": 0.21053080260753632,
|
| 511 |
+
"eval_runtime": 691.0445,
|
| 512 |
+
"eval_samples_per_second": 12.012,
|
| 513 |
+
"eval_steps_per_second": 1.502,
|
| 514 |
+
"eval_wer": 0.19918452329313785,
|
| 515 |
+
"step": 6500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 8.49,
|
| 519 |
+
"learning_rate": 6.565196743554952e-05,
|
| 520 |
+
"loss": 0.9414,
|
| 521 |
+
"step": 6600
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 8.62,
|
| 525 |
+
"learning_rate": 6.54484396200814e-05,
|
| 526 |
+
"loss": 0.9491,
|
| 527 |
+
"step": 6700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 8.75,
|
| 531 |
+
"learning_rate": 6.524491180461329e-05,
|
| 532 |
+
"loss": 0.9548,
|
| 533 |
+
"step": 6800
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 8.88,
|
| 537 |
+
"learning_rate": 6.504138398914518e-05,
|
| 538 |
+
"loss": 0.9409,
|
| 539 |
+
"step": 6900
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 9.01,
|
| 543 |
+
"learning_rate": 6.483785617367707e-05,
|
| 544 |
+
"loss": 0.9365,
|
| 545 |
+
"step": 7000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 9.01,
|
| 549 |
+
"eval_loss": 0.21139565110206604,
|
| 550 |
+
"eval_runtime": 668.2218,
|
| 551 |
+
"eval_samples_per_second": 12.423,
|
| 552 |
+
"eval_steps_per_second": 1.553,
|
| 553 |
+
"eval_wer": 0.20185651080072872,
|
| 554 |
+
"step": 7000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 9.14,
|
| 558 |
+
"learning_rate": 6.463432835820895e-05,
|
| 559 |
+
"loss": 0.9264,
|
| 560 |
+
"step": 7100
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 9.27,
|
| 564 |
+
"learning_rate": 6.443080054274083e-05,
|
| 565 |
+
"loss": 0.9129,
|
| 566 |
+
"step": 7200
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 9.39,
|
| 570 |
+
"learning_rate": 6.422727272727272e-05,
|
| 571 |
+
"loss": 0.9267,
|
| 572 |
+
"step": 7300
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 9.52,
|
| 576 |
+
"learning_rate": 6.402374491180461e-05,
|
| 577 |
+
"loss": 0.9213,
|
| 578 |
+
"step": 7400
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 9.65,
|
| 582 |
+
"learning_rate": 6.38202170963365e-05,
|
| 583 |
+
"loss": 0.9268,
|
| 584 |
+
"step": 7500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 9.65,
|
| 588 |
+
"eval_loss": 0.20965604484081268,
|
| 589 |
+
"eval_runtime": 672.8441,
|
| 590 |
+
"eval_samples_per_second": 12.337,
|
| 591 |
+
"eval_steps_per_second": 1.543,
|
| 592 |
+
"eval_wer": 0.19880281079205345,
|
| 593 |
+
"step": 7500
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 9.78,
|
| 597 |
+
"learning_rate": 6.361668928086838e-05,
|
| 598 |
+
"loss": 0.9087,
|
| 599 |
+
"step": 7600
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 9.91,
|
| 603 |
+
"learning_rate": 6.341316146540027e-05,
|
| 604 |
+
"loss": 0.9142,
|
| 605 |
+
"step": 7700
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 10.04,
|
| 609 |
+
"learning_rate": 6.320963364993215e-05,
|
| 610 |
+
"loss": 0.9274,
|
| 611 |
+
"step": 7800
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 10.17,
|
| 615 |
+
"learning_rate": 6.300610583446404e-05,
|
| 616 |
+
"loss": 0.9223,
|
| 617 |
+
"step": 7900
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 10.3,
|
| 621 |
+
"learning_rate": 6.280257801899593e-05,
|
| 622 |
+
"loss": 0.9292,
|
| 623 |
+
"step": 8000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 10.3,
|
| 627 |
+
"eval_loss": 0.21196416020393372,
|
| 628 |
+
"eval_runtime": 663.5835,
|
| 629 |
+
"eval_samples_per_second": 12.509,
|
| 630 |
+
"eval_steps_per_second": 1.564,
|
| 631 |
+
"eval_wer": 0.1985599028368179,
|
| 632 |
+
"step": 8000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 10.42,
|
| 636 |
+
"learning_rate": 6.25990502035278e-05,
|
| 637 |
+
"loss": 0.9355,
|
| 638 |
+
"step": 8100
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 10.55,
|
| 642 |
+
"learning_rate": 6.239552238805969e-05,
|
| 643 |
+
"loss": 0.9095,
|
| 644 |
+
"step": 8200
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 10.68,
|
| 648 |
+
"learning_rate": 6.219199457259158e-05,
|
| 649 |
+
"loss": 0.8979,
|
| 650 |
+
"step": 8300
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 10.81,
|
| 654 |
+
"learning_rate": 6.198846675712346e-05,
|
| 655 |
+
"loss": 0.9174,
|
| 656 |
+
"step": 8400
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 10.94,
|
| 660 |
+
"learning_rate": 6.178493894165536e-05,
|
| 661 |
+
"loss": 0.929,
|
| 662 |
+
"step": 8500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 10.94,
|
| 666 |
+
"eval_loss": 0.2047712206840515,
|
| 667 |
+
"eval_runtime": 660.2531,
|
| 668 |
+
"eval_samples_per_second": 12.572,
|
| 669 |
+
"eval_steps_per_second": 1.572,
|
| 670 |
+
"eval_wer": 0.1998264943176889,
|
| 671 |
+
"step": 8500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 11.07,
|
| 675 |
+
"learning_rate": 6.158141112618724e-05,
|
| 676 |
+
"loss": 0.921,
|
| 677 |
+
"step": 8600
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 11.2,
|
| 681 |
+
"learning_rate": 6.137788331071913e-05,
|
| 682 |
+
"loss": 0.9058,
|
| 683 |
+
"step": 8700
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 11.33,
|
| 687 |
+
"learning_rate": 6.117435549525101e-05,
|
| 688 |
+
"loss": 0.887,
|
| 689 |
+
"step": 8800
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 11.45,
|
| 693 |
+
"learning_rate": 6.09708276797829e-05,
|
| 694 |
+
"loss": 0.9142,
|
| 695 |
+
"step": 8900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 11.58,
|
| 699 |
+
"learning_rate": 6.076729986431478e-05,
|
| 700 |
+
"loss": 0.9017,
|
| 701 |
+
"step": 9000
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 11.58,
|
| 705 |
+
"eval_loss": 0.20354785025119781,
|
| 706 |
+
"eval_runtime": 663.071,
|
| 707 |
+
"eval_samples_per_second": 12.519,
|
| 708 |
+
"eval_steps_per_second": 1.565,
|
| 709 |
+
"eval_wer": 0.19993059772707555,
|
| 710 |
+
"step": 9000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 11.71,
|
| 714 |
+
"learning_rate": 6.056377204884667e-05,
|
| 715 |
+
"loss": 0.9031,
|
| 716 |
+
"step": 9100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 11.84,
|
| 720 |
+
"learning_rate": 6.0360244233378555e-05,
|
| 721 |
+
"loss": 0.8996,
|
| 722 |
+
"step": 9200
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 11.97,
|
| 726 |
+
"learning_rate": 6.015671641791044e-05,
|
| 727 |
+
"loss": 0.9138,
|
| 728 |
+
"step": 9300
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 12.1,
|
| 732 |
+
"learning_rate": 5.9953188602442325e-05,
|
| 733 |
+
"loss": 0.8957,
|
| 734 |
+
"step": 9400
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 12.23,
|
| 738 |
+
"learning_rate": 5.974966078697422e-05,
|
| 739 |
+
"loss": 0.8898,
|
| 740 |
+
"step": 9500
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 12.23,
|
| 744 |
+
"eval_loss": 0.1960887461900711,
|
| 745 |
+
"eval_runtime": 671.0322,
|
| 746 |
+
"eval_samples_per_second": 12.37,
|
| 747 |
+
"eval_steps_per_second": 1.547,
|
| 748 |
+
"eval_wer": 0.1907521471328186,
|
| 749 |
+
"step": 9500
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 12.35,
|
| 753 |
+
"learning_rate": 5.95461329715061e-05,
|
| 754 |
+
"loss": 0.884,
|
| 755 |
+
"step": 9600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 12.48,
|
| 759 |
+
"learning_rate": 5.934260515603799e-05,
|
| 760 |
+
"loss": 0.8872,
|
| 761 |
+
"step": 9700
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 12.61,
|
| 765 |
+
"learning_rate": 5.913907734056988e-05,
|
| 766 |
+
"loss": 0.8993,
|
| 767 |
+
"step": 9800
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 12.74,
|
| 771 |
+
"learning_rate": 5.893554952510176e-05,
|
| 772 |
+
"loss": 0.8845,
|
| 773 |
+
"step": 9900
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 12.87,
|
| 777 |
+
"learning_rate": 5.873202170963364e-05,
|
| 778 |
+
"loss": 0.8799,
|
| 779 |
+
"step": 10000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 12.87,
|
| 783 |
+
"eval_loss": 0.19447088241577148,
|
| 784 |
+
"eval_runtime": 677.1371,
|
| 785 |
+
"eval_samples_per_second": 12.259,
|
| 786 |
+
"eval_steps_per_second": 1.533,
|
| 787 |
+
"eval_wer": 0.1816951505161794,
|
| 788 |
+
"step": 10000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 13.0,
|
| 792 |
+
"learning_rate": 5.852849389416553e-05,
|
| 793 |
+
"loss": 0.8997,
|
| 794 |
+
"step": 10100
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 13.13,
|
| 798 |
+
"learning_rate": 5.832496607869741e-05,
|
| 799 |
+
"loss": 0.8774,
|
| 800 |
+
"step": 10200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 13.26,
|
| 804 |
+
"learning_rate": 5.8121438263229305e-05,
|
| 805 |
+
"loss": 0.8761,
|
| 806 |
+
"step": 10300
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 13.38,
|
| 810 |
+
"learning_rate": 5.791791044776119e-05,
|
| 811 |
+
"loss": 0.8731,
|
| 812 |
+
"step": 10400
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 13.51,
|
| 816 |
+
"learning_rate": 5.7714382632293075e-05,
|
| 817 |
+
"loss": 0.869,
|
| 818 |
+
"step": 10500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 13.51,
|
| 822 |
+
"eval_loss": 0.19293242692947388,
|
| 823 |
+
"eval_runtime": 658.1115,
|
| 824 |
+
"eval_samples_per_second": 12.613,
|
| 825 |
+
"eval_steps_per_second": 1.577,
|
| 826 |
+
"eval_wer": 0.18436713802377028,
|
| 827 |
+
"step": 10500
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 13.64,
|
| 831 |
+
"learning_rate": 5.751085481682497e-05,
|
| 832 |
+
"loss": 0.8771,
|
| 833 |
+
"step": 10600
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 13.77,
|
| 837 |
+
"learning_rate": 5.730936227951153e-05,
|
| 838 |
+
"loss": 0.8718,
|
| 839 |
+
"step": 10700
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 13.9,
|
| 843 |
+
"learning_rate": 5.710583446404341e-05,
|
| 844 |
+
"loss": 0.8873,
|
| 845 |
+
"step": 10800
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 14.03,
|
| 849 |
+
"learning_rate": 5.6902306648575305e-05,
|
| 850 |
+
"loss": 0.878,
|
| 851 |
+
"step": 10900
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 14.16,
|
| 855 |
+
"learning_rate": 5.669877883310719e-05,
|
| 856 |
+
"loss": 0.8572,
|
| 857 |
+
"step": 11000
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 14.16,
|
| 861 |
+
"eval_loss": 0.19409486651420593,
|
| 862 |
+
"eval_runtime": 670.5041,
|
| 863 |
+
"eval_samples_per_second": 12.38,
|
| 864 |
+
"eval_steps_per_second": 1.548,
|
| 865 |
+
"eval_wer": 0.18882623405916543,
|
| 866 |
+
"step": 11000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 14.29,
|
| 870 |
+
"learning_rate": 5.6495251017639075e-05,
|
| 871 |
+
"loss": 0.8561,
|
| 872 |
+
"step": 11100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 14.41,
|
| 876 |
+
"learning_rate": 5.6291723202170953e-05,
|
| 877 |
+
"loss": 0.8739,
|
| 878 |
+
"step": 11200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 14.54,
|
| 882 |
+
"learning_rate": 5.608819538670284e-05,
|
| 883 |
+
"loss": 0.8706,
|
| 884 |
+
"step": 11300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 14.67,
|
| 888 |
+
"learning_rate": 5.588466757123473e-05,
|
| 889 |
+
"loss": 0.8693,
|
| 890 |
+
"step": 11400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 14.8,
|
| 894 |
+
"learning_rate": 5.5681139755766616e-05,
|
| 895 |
+
"loss": 0.8691,
|
| 896 |
+
"step": 11500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 14.8,
|
| 900 |
+
"eval_loss": 0.1912171095609665,
|
| 901 |
+
"eval_runtime": 691.3958,
|
| 902 |
+
"eval_samples_per_second": 12.006,
|
| 903 |
+
"eval_steps_per_second": 1.501,
|
| 904 |
+
"eval_wer": 0.18037650733061508,
|
| 905 |
+
"step": 11500
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 14.93,
|
| 909 |
+
"learning_rate": 5.54776119402985e-05,
|
| 910 |
+
"loss": 0.8562,
|
| 911 |
+
"step": 11600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 15.06,
|
| 915 |
+
"learning_rate": 5.527408412483039e-05,
|
| 916 |
+
"loss": 0.8874,
|
| 917 |
+
"step": 11700
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 15.19,
|
| 921 |
+
"learning_rate": 5.507055630936228e-05,
|
| 922 |
+
"loss": 0.8503,
|
| 923 |
+
"step": 11800
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 15.32,
|
| 927 |
+
"learning_rate": 5.486702849389416e-05,
|
| 928 |
+
"loss": 0.8558,
|
| 929 |
+
"step": 11900
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 15.44,
|
| 933 |
+
"learning_rate": 5.466350067842605e-05,
|
| 934 |
+
"loss": 0.8645,
|
| 935 |
+
"step": 12000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 15.44,
|
| 939 |
+
"eval_loss": 0.1950196474790573,
|
| 940 |
+
"eval_runtime": 670.3998,
|
| 941 |
+
"eval_samples_per_second": 12.382,
|
| 942 |
+
"eval_steps_per_second": 1.548,
|
| 943 |
+
"eval_wer": 0.18506116075301465,
|
| 944 |
+
"step": 12000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 15.57,
|
| 948 |
+
"learning_rate": 5.4459972862957926e-05,
|
| 949 |
+
"loss": 0.8668,
|
| 950 |
+
"step": 12100
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 15.7,
|
| 954 |
+
"learning_rate": 5.425644504748982e-05,
|
| 955 |
+
"loss": 0.8461,
|
| 956 |
+
"step": 12200
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 15.83,
|
| 960 |
+
"learning_rate": 5.40529172320217e-05,
|
| 961 |
+
"loss": 0.8508,
|
| 962 |
+
"step": 12300
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 15.96,
|
| 966 |
+
"learning_rate": 5.384938941655359e-05,
|
| 967 |
+
"loss": 0.8529,
|
| 968 |
+
"step": 12400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 16.09,
|
| 972 |
+
"learning_rate": 5.364586160108548e-05,
|
| 973 |
+
"loss": 0.8468,
|
| 974 |
+
"step": 12500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 16.09,
|
| 978 |
+
"eval_loss": 0.1879136562347412,
|
| 979 |
+
"eval_runtime": 672.4666,
|
| 980 |
+
"eval_samples_per_second": 12.344,
|
| 981 |
+
"eval_steps_per_second": 1.544,
|
| 982 |
+
"eval_wer": 0.17701049709377983,
|
| 983 |
+
"step": 12500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 16.22,
|
| 987 |
+
"learning_rate": 5.3442333785617365e-05,
|
| 988 |
+
"loss": 0.8342,
|
| 989 |
+
"step": 12600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 16.34,
|
| 993 |
+
"learning_rate": 5.323880597014925e-05,
|
| 994 |
+
"loss": 0.8465,
|
| 995 |
+
"step": 12700
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 16.47,
|
| 999 |
+
"learning_rate": 5.303731343283582e-05,
|
| 1000 |
+
"loss": 0.8206,
|
| 1001 |
+
"step": 12800
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 16.6,
|
| 1005 |
+
"learning_rate": 5.28337856173677e-05,
|
| 1006 |
+
"loss": 0.8453,
|
| 1007 |
+
"step": 12900
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 16.73,
|
| 1011 |
+
"learning_rate": 5.263025780189959e-05,
|
| 1012 |
+
"loss": 0.8405,
|
| 1013 |
+
"step": 13000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 16.73,
|
| 1017 |
+
"eval_loss": 0.18807044625282288,
|
| 1018 |
+
"eval_runtime": 680.7317,
|
| 1019 |
+
"eval_samples_per_second": 12.194,
|
| 1020 |
+
"eval_steps_per_second": 1.525,
|
| 1021 |
+
"eval_wer": 0.1758827101587577,
|
| 1022 |
+
"step": 13000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 16.86,
|
| 1026 |
+
"learning_rate": 5.2426729986431473e-05,
|
| 1027 |
+
"loss": 0.8571,
|
| 1028 |
+
"step": 13100
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 16.99,
|
| 1032 |
+
"learning_rate": 5.2223202170963365e-05,
|
| 1033 |
+
"loss": 0.854,
|
| 1034 |
+
"step": 13200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 17.12,
|
| 1038 |
+
"learning_rate": 5.201967435549525e-05,
|
| 1039 |
+
"loss": 0.8578,
|
| 1040 |
+
"step": 13300
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 17.25,
|
| 1044 |
+
"learning_rate": 5.181614654002713e-05,
|
| 1045 |
+
"loss": 0.8382,
|
| 1046 |
+
"step": 13400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 17.37,
|
| 1050 |
+
"learning_rate": 5.1612618724559014e-05,
|
| 1051 |
+
"loss": 0.8647,
|
| 1052 |
+
"step": 13500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 17.37,
|
| 1056 |
+
"eval_loss": 0.1861347109079361,
|
| 1057 |
+
"eval_runtime": 674.9378,
|
| 1058 |
+
"eval_samples_per_second": 12.299,
|
| 1059 |
+
"eval_steps_per_second": 1.538,
|
| 1060 |
+
"eval_wer": 0.17395679708510453,
|
| 1061 |
+
"step": 13500
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 17.5,
|
| 1065 |
+
"learning_rate": 5.1409090909090906e-05,
|
| 1066 |
+
"loss": 0.8372,
|
| 1067 |
+
"step": 13600
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 17.63,
|
| 1071 |
+
"learning_rate": 5.120556309362279e-05,
|
| 1072 |
+
"loss": 0.8253,
|
| 1073 |
+
"step": 13700
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 17.76,
|
| 1077 |
+
"learning_rate": 5.1002035278154676e-05,
|
| 1078 |
+
"loss": 0.8353,
|
| 1079 |
+
"step": 13800
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 17.89,
|
| 1083 |
+
"learning_rate": 5.079850746268656e-05,
|
| 1084 |
+
"loss": 0.8548,
|
| 1085 |
+
"step": 13900
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 18.02,
|
| 1089 |
+
"learning_rate": 5.059497964721845e-05,
|
| 1090 |
+
"loss": 0.8477,
|
| 1091 |
+
"step": 14000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 18.02,
|
| 1095 |
+
"eval_loss": 0.1782081127166748,
|
| 1096 |
+
"eval_runtime": 701.0008,
|
| 1097 |
+
"eval_samples_per_second": 11.842,
|
| 1098 |
+
"eval_steps_per_second": 1.481,
|
| 1099 |
+
"eval_wer": 0.17017437321072265,
|
| 1100 |
+
"step": 14000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 18.15,
|
| 1104 |
+
"learning_rate": 5.039145183175034e-05,
|
| 1105 |
+
"loss": 0.8121,
|
| 1106 |
+
"step": 14100
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 18.28,
|
| 1110 |
+
"learning_rate": 5.018792401628222e-05,
|
| 1111 |
+
"loss": 0.8243,
|
| 1112 |
+
"step": 14200
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 18.4,
|
| 1116 |
+
"learning_rate": 4.99843962008141e-05,
|
| 1117 |
+
"loss": 0.8312,
|
| 1118 |
+
"step": 14300
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 18.53,
|
| 1122 |
+
"learning_rate": 4.978086838534599e-05,
|
| 1123 |
+
"loss": 0.8288,
|
| 1124 |
+
"step": 14400
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 18.66,
|
| 1128 |
+
"learning_rate": 4.957734056987788e-05,
|
| 1129 |
+
"loss": 0.811,
|
| 1130 |
+
"step": 14500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 18.66,
|
| 1134 |
+
"eval_loss": 0.19152408838272095,
|
| 1135 |
+
"eval_runtime": 690.1271,
|
| 1136 |
+
"eval_samples_per_second": 12.028,
|
| 1137 |
+
"eval_steps_per_second": 1.504,
|
| 1138 |
+
"eval_wer": 0.1757265550446777,
|
| 1139 |
+
"step": 14500
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 18.79,
|
| 1143 |
+
"learning_rate": 4.937381275440976e-05,
|
| 1144 |
+
"loss": 0.8276,
|
| 1145 |
+
"step": 14600
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 18.92,
|
| 1149 |
+
"learning_rate": 4.917028493894165e-05,
|
| 1150 |
+
"loss": 0.8337,
|
| 1151 |
+
"step": 14700
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 19.05,
|
| 1155 |
+
"learning_rate": 4.896675712347354e-05,
|
| 1156 |
+
"loss": 0.8343,
|
| 1157 |
+
"step": 14800
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 19.18,
|
| 1161 |
+
"learning_rate": 4.8763229308005425e-05,
|
| 1162 |
+
"loss": 0.8046,
|
| 1163 |
+
"step": 14900
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 19.3,
|
| 1167 |
+
"learning_rate": 4.855970149253731e-05,
|
| 1168 |
+
"loss": 0.8165,
|
| 1169 |
+
"step": 15000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 19.3,
|
| 1173 |
+
"eval_loss": 0.18196436762809753,
|
| 1174 |
+
"eval_runtime": 694.7382,
|
| 1175 |
+
"eval_samples_per_second": 11.948,
|
| 1176 |
+
"eval_steps_per_second": 1.494,
|
| 1177 |
+
"eval_wer": 0.172447297648998,
|
| 1178 |
+
"step": 15000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 19.43,
|
| 1182 |
+
"learning_rate": 4.8356173677069196e-05,
|
| 1183 |
+
"loss": 0.8109,
|
| 1184 |
+
"step": 15100
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 19.56,
|
| 1188 |
+
"learning_rate": 4.815264586160108e-05,
|
| 1189 |
+
"loss": 0.824,
|
| 1190 |
+
"step": 15200
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 19.69,
|
| 1194 |
+
"learning_rate": 4.7949118046132966e-05,
|
| 1195 |
+
"loss": 0.8213,
|
| 1196 |
+
"step": 15300
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 19.82,
|
| 1200 |
+
"learning_rate": 4.774559023066485e-05,
|
| 1201 |
+
"loss": 0.8197,
|
| 1202 |
+
"step": 15400
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 19.95,
|
| 1206 |
+
"learning_rate": 4.7542062415196736e-05,
|
| 1207 |
+
"loss": 0.8166,
|
| 1208 |
+
"step": 15500
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 19.95,
|
| 1212 |
+
"eval_loss": 0.17982707917690277,
|
| 1213 |
+
"eval_runtime": 678.5205,
|
| 1214 |
+
"eval_samples_per_second": 12.234,
|
| 1215 |
+
"eval_steps_per_second": 1.53,
|
| 1216 |
+
"eval_wer": 0.16974060900494492,
|
| 1217 |
+
"step": 15500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 20.08,
|
| 1221 |
+
"learning_rate": 4.733853459972863e-05,
|
| 1222 |
+
"loss": 0.8024,
|
| 1223 |
+
"step": 15600
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 20.21,
|
| 1227 |
+
"learning_rate": 4.713500678426051e-05,
|
| 1228 |
+
"loss": 0.8093,
|
| 1229 |
+
"step": 15700
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 20.33,
|
| 1233 |
+
"learning_rate": 4.69314789687924e-05,
|
| 1234 |
+
"loss": 0.8215,
|
| 1235 |
+
"step": 15800
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 20.46,
|
| 1239 |
+
"learning_rate": 4.672795115332428e-05,
|
| 1240 |
+
"loss": 0.8061,
|
| 1241 |
+
"step": 15900
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 20.59,
|
| 1245 |
+
"learning_rate": 4.652645861601085e-05,
|
| 1246 |
+
"loss": 0.8167,
|
| 1247 |
+
"step": 16000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 20.59,
|
| 1251 |
+
"eval_loss": 0.18053852021694183,
|
| 1252 |
+
"eval_runtime": 661.7443,
|
| 1253 |
+
"eval_samples_per_second": 12.544,
|
| 1254 |
+
"eval_steps_per_second": 1.569,
|
| 1255 |
+
"eval_wer": 0.1751539862930511,
|
| 1256 |
+
"step": 16000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 20.72,
|
| 1260 |
+
"learning_rate": 4.6322930800542736e-05,
|
| 1261 |
+
"loss": 0.8003,
|
| 1262 |
+
"step": 16100
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 20.85,
|
| 1266 |
+
"learning_rate": 4.6121438263229304e-05,
|
| 1267 |
+
"loss": 0.8008,
|
| 1268 |
+
"step": 16200
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 20.98,
|
| 1272 |
+
"learning_rate": 4.591791044776119e-05,
|
| 1273 |
+
"loss": 0.8064,
|
| 1274 |
+
"step": 16300
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 21.11,
|
| 1278 |
+
"learning_rate": 4.5714382632293075e-05,
|
| 1279 |
+
"loss": 0.8184,
|
| 1280 |
+
"step": 16400
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 21.24,
|
| 1284 |
+
"learning_rate": 4.551085481682496e-05,
|
| 1285 |
+
"loss": 0.7908,
|
| 1286 |
+
"step": 16500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 21.24,
|
| 1290 |
+
"eval_loss": 0.17608480155467987,
|
| 1291 |
+
"eval_runtime": 665.8329,
|
| 1292 |
+
"eval_samples_per_second": 12.467,
|
| 1293 |
+
"eval_steps_per_second": 1.559,
|
| 1294 |
+
"eval_wer": 0.169914114687256,
|
| 1295 |
+
"step": 16500
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 21.36,
|
| 1299 |
+
"learning_rate": 4.530732700135685e-05,
|
| 1300 |
+
"loss": 0.7861,
|
| 1301 |
+
"step": 16600
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 21.49,
|
| 1305 |
+
"learning_rate": 4.5103799185888737e-05,
|
| 1306 |
+
"loss": 0.8065,
|
| 1307 |
+
"step": 16700
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 21.62,
|
| 1311 |
+
"learning_rate": 4.490027137042062e-05,
|
| 1312 |
+
"loss": 0.7982,
|
| 1313 |
+
"step": 16800
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 21.75,
|
| 1317 |
+
"learning_rate": 4.46967435549525e-05,
|
| 1318 |
+
"loss": 0.803,
|
| 1319 |
+
"step": 16900
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 21.88,
|
| 1323 |
+
"learning_rate": 4.449321573948439e-05,
|
| 1324 |
+
"loss": 0.7925,
|
| 1325 |
+
"step": 17000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 21.88,
|
| 1329 |
+
"eval_loss": 0.1740112453699112,
|
| 1330 |
+
"eval_runtime": 660.0442,
|
| 1331 |
+
"eval_samples_per_second": 12.576,
|
| 1332 |
+
"eval_steps_per_second": 1.573,
|
| 1333 |
+
"eval_wer": 0.17093779821289148,
|
| 1334 |
+
"step": 17000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 22.01,
|
| 1338 |
+
"learning_rate": 4.428968792401628e-05,
|
| 1339 |
+
"loss": 0.7822,
|
| 1340 |
+
"step": 17100
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 22.14,
|
| 1344 |
+
"learning_rate": 4.408616010854816e-05,
|
| 1345 |
+
"loss": 0.7851,
|
| 1346 |
+
"step": 17200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 22.26,
|
| 1350 |
+
"learning_rate": 4.388263229308005e-05,
|
| 1351 |
+
"loss": 0.7871,
|
| 1352 |
+
"step": 17300
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 22.39,
|
| 1356 |
+
"learning_rate": 4.367910447761194e-05,
|
| 1357 |
+
"loss": 0.7964,
|
| 1358 |
+
"step": 17400
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 22.52,
|
| 1362 |
+
"learning_rate": 4.3475576662143824e-05,
|
| 1363 |
+
"loss": 0.7803,
|
| 1364 |
+
"step": 17500
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 22.52,
|
| 1368 |
+
"eval_loss": 0.1815224587917328,
|
| 1369 |
+
"eval_runtime": 660.2602,
|
| 1370 |
+
"eval_samples_per_second": 12.572,
|
| 1371 |
+
"eval_steps_per_second": 1.572,
|
| 1372 |
+
"eval_wer": 0.17269020560423354,
|
| 1373 |
+
"step": 17500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 22.65,
|
| 1377 |
+
"learning_rate": 4.327204884667571e-05,
|
| 1378 |
+
"loss": 0.7793,
|
| 1379 |
+
"step": 17600
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 22.78,
|
| 1383 |
+
"learning_rate": 4.30685210312076e-05,
|
| 1384 |
+
"loss": 0.7872,
|
| 1385 |
+
"step": 17700
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 22.91,
|
| 1389 |
+
"learning_rate": 4.286499321573948e-05,
|
| 1390 |
+
"loss": 0.7895,
|
| 1391 |
+
"step": 17800
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 23.04,
|
| 1395 |
+
"learning_rate": 4.2661465400271364e-05,
|
| 1396 |
+
"loss": 0.7925,
|
| 1397 |
+
"step": 17900
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 23.17,
|
| 1401 |
+
"learning_rate": 4.245793758480325e-05,
|
| 1402 |
+
"loss": 0.7839,
|
| 1403 |
+
"step": 18000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 23.17,
|
| 1407 |
+
"eval_loss": 0.17369158565998077,
|
| 1408 |
+
"eval_runtime": 664.004,
|
| 1409 |
+
"eval_samples_per_second": 12.501,
|
| 1410 |
+
"eval_steps_per_second": 1.563,
|
| 1411 |
+
"eval_wer": 0.1693588965038605,
|
| 1412 |
+
"step": 18000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 23.29,
|
| 1416 |
+
"learning_rate": 4.2254409769335135e-05,
|
| 1417 |
+
"loss": 0.7662,
|
| 1418 |
+
"step": 18100
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 23.42,
|
| 1422 |
+
"learning_rate": 4.2050881953867026e-05,
|
| 1423 |
+
"loss": 0.7779,
|
| 1424 |
+
"step": 18200
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 23.55,
|
| 1428 |
+
"learning_rate": 4.184735413839891e-05,
|
| 1429 |
+
"loss": 0.7867,
|
| 1430 |
+
"step": 18300
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 23.68,
|
| 1434 |
+
"learning_rate": 4.1643826322930797e-05,
|
| 1435 |
+
"loss": 0.7645,
|
| 1436 |
+
"step": 18400
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 23.81,
|
| 1440 |
+
"learning_rate": 4.144029850746269e-05,
|
| 1441 |
+
"loss": 0.7815,
|
| 1442 |
+
"step": 18500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 23.81,
|
| 1446 |
+
"eval_loss": 0.17323149740695953,
|
| 1447 |
+
"eval_runtime": 656.7033,
|
| 1448 |
+
"eval_samples_per_second": 12.64,
|
| 1449 |
+
"eval_steps_per_second": 1.581,
|
| 1450 |
+
"eval_wer": 0.16304328966773662,
|
| 1451 |
+
"step": 18500
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 23.94,
|
| 1455 |
+
"learning_rate": 4.1236770691994574e-05,
|
| 1456 |
+
"loss": 0.7783,
|
| 1457 |
+
"step": 18600
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 24.07,
|
| 1461 |
+
"learning_rate": 4.103324287652645e-05,
|
| 1462 |
+
"loss": 0.79,
|
| 1463 |
+
"step": 18700
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 24.2,
|
| 1467 |
+
"learning_rate": 4.082971506105834e-05,
|
| 1468 |
+
"loss": 0.7503,
|
| 1469 |
+
"step": 18800
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 24.32,
|
| 1473 |
+
"learning_rate": 4.062618724559022e-05,
|
| 1474 |
+
"loss": 0.7941,
|
| 1475 |
+
"step": 18900
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 24.45,
|
| 1479 |
+
"learning_rate": 4.0422659430122114e-05,
|
| 1480 |
+
"loss": 0.767,
|
| 1481 |
+
"step": 19000
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 24.45,
|
| 1485 |
+
"eval_loss": 0.17244413495063782,
|
| 1486 |
+
"eval_runtime": 658.5684,
|
| 1487 |
+
"eval_samples_per_second": 12.605,
|
| 1488 |
+
"eval_steps_per_second": 1.576,
|
| 1489 |
+
"eval_wer": 0.16479569705907868,
|
| 1490 |
+
"step": 19000
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 24.58,
|
| 1494 |
+
"learning_rate": 4.0219131614654e-05,
|
| 1495 |
+
"loss": 0.7828,
|
| 1496 |
+
"step": 19100
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 24.71,
|
| 1500 |
+
"learning_rate": 4.0015603799185884e-05,
|
| 1501 |
+
"loss": 0.7594,
|
| 1502 |
+
"step": 19200
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 24.84,
|
| 1506 |
+
"learning_rate": 3.9812075983717776e-05,
|
| 1507 |
+
"loss": 0.7649,
|
| 1508 |
+
"step": 19300
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 24.97,
|
| 1512 |
+
"learning_rate": 3.960854816824966e-05,
|
| 1513 |
+
"loss": 0.7619,
|
| 1514 |
+
"step": 19400
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 25.1,
|
| 1518 |
+
"learning_rate": 3.9405020352781546e-05,
|
| 1519 |
+
"loss": 0.7672,
|
| 1520 |
+
"step": 19500
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 25.1,
|
| 1524 |
+
"eval_loss": 0.1706162840127945,
|
| 1525 |
+
"eval_runtime": 662.0343,
|
| 1526 |
+
"eval_samples_per_second": 12.539,
|
| 1527 |
+
"eval_steps_per_second": 1.568,
|
| 1528 |
+
"eval_wer": 0.15959052658974582,
|
| 1529 |
+
"step": 19500
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 25.23,
|
| 1533 |
+
"learning_rate": 3.9201492537313424e-05,
|
| 1534 |
+
"loss": 0.7685,
|
| 1535 |
+
"step": 19600
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 25.35,
|
| 1539 |
+
"learning_rate": 3.899796472184531e-05,
|
| 1540 |
+
"loss": 0.7739,
|
| 1541 |
+
"step": 19700
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 25.48,
|
| 1545 |
+
"learning_rate": 3.87944369063772e-05,
|
| 1546 |
+
"loss": 0.7729,
|
| 1547 |
+
"step": 19800
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 25.61,
|
| 1551 |
+
"learning_rate": 3.8590909090909086e-05,
|
| 1552 |
+
"loss": 0.7573,
|
| 1553 |
+
"step": 19900
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 25.74,
|
| 1557 |
+
"learning_rate": 3.838738127544097e-05,
|
| 1558 |
+
"loss": 0.7691,
|
| 1559 |
+
"step": 20000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 25.74,
|
| 1563 |
+
"eval_loss": 0.17184105515480042,
|
| 1564 |
+
"eval_runtime": 658.3755,
|
| 1565 |
+
"eval_samples_per_second": 12.608,
|
| 1566 |
+
"eval_steps_per_second": 1.577,
|
| 1567 |
+
"eval_wer": 0.16177669818686563,
|
| 1568 |
+
"step": 20000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 25.87,
|
| 1572 |
+
"learning_rate": 3.818385345997286e-05,
|
| 1573 |
+
"loss": 0.7646,
|
| 1574 |
+
"step": 20100
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 26.0,
|
| 1578 |
+
"learning_rate": 3.798032564450475e-05,
|
| 1579 |
+
"loss": 0.754,
|
| 1580 |
+
"step": 20200
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 26.13,
|
| 1584 |
+
"learning_rate": 3.7776797829036634e-05,
|
| 1585 |
+
"loss": 0.7629,
|
| 1586 |
+
"step": 20300
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 26.25,
|
| 1590 |
+
"learning_rate": 3.757327001356852e-05,
|
| 1591 |
+
"loss": 0.7601,
|
| 1592 |
+
"step": 20400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 26.38,
|
| 1596 |
+
"learning_rate": 3.7369742198100404e-05,
|
| 1597 |
+
"loss": 0.7547,
|
| 1598 |
+
"step": 20500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 26.38,
|
| 1602 |
+
"eval_loss": 0.1694486141204834,
|
| 1603 |
+
"eval_runtime": 662.0708,
|
| 1604 |
+
"eval_samples_per_second": 12.538,
|
| 1605 |
+
"eval_steps_per_second": 1.568,
|
| 1606 |
+
"eval_wer": 0.15648477487637721,
|
| 1607 |
+
"step": 20500
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 26.51,
|
| 1611 |
+
"learning_rate": 3.716621438263229e-05,
|
| 1612 |
+
"loss": 0.7612,
|
| 1613 |
+
"step": 20600
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 26.64,
|
| 1617 |
+
"learning_rate": 3.696268656716418e-05,
|
| 1618 |
+
"loss": 0.7699,
|
| 1619 |
+
"step": 20700
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 26.77,
|
| 1623 |
+
"learning_rate": 3.675915875169606e-05,
|
| 1624 |
+
"loss": 0.7609,
|
| 1625 |
+
"step": 20800
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 26.9,
|
| 1629 |
+
"learning_rate": 3.655563093622795e-05,
|
| 1630 |
+
"loss": 0.739,
|
| 1631 |
+
"step": 20900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 27.03,
|
| 1635 |
+
"learning_rate": 3.6352103120759836e-05,
|
| 1636 |
+
"loss": 0.7498,
|
| 1637 |
+
"step": 21000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 27.03,
|
| 1641 |
+
"eval_loss": 0.17055003345012665,
|
| 1642 |
+
"eval_runtime": 660.5207,
|
| 1643 |
+
"eval_samples_per_second": 12.567,
|
| 1644 |
+
"eval_steps_per_second": 1.571,
|
| 1645 |
+
"eval_wer": 0.15821983169948817,
|
| 1646 |
+
"step": 21000
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 27.16,
|
| 1650 |
+
"learning_rate": 3.614857530529172e-05,
|
| 1651 |
+
"loss": 0.7583,
|
| 1652 |
+
"step": 21100
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 27.28,
|
| 1656 |
+
"learning_rate": 3.5945047489823606e-05,
|
| 1657 |
+
"loss": 0.7591,
|
| 1658 |
+
"step": 21200
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 27.41,
|
| 1662 |
+
"learning_rate": 3.574151967435549e-05,
|
| 1663 |
+
"loss": 0.7415,
|
| 1664 |
+
"step": 21300
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 27.54,
|
| 1668 |
+
"learning_rate": 3.5537991858887376e-05,
|
| 1669 |
+
"loss": 0.734,
|
| 1670 |
+
"step": 21400
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 27.67,
|
| 1674 |
+
"learning_rate": 3.533446404341927e-05,
|
| 1675 |
+
"loss": 0.7459,
|
| 1676 |
+
"step": 21500
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 27.67,
|
| 1680 |
+
"eval_loss": 0.1662723422050476,
|
| 1681 |
+
"eval_runtime": 658.6276,
|
| 1682 |
+
"eval_samples_per_second": 12.603,
|
| 1683 |
+
"eval_steps_per_second": 1.576,
|
| 1684 |
+
"eval_wer": 0.1586362453370348,
|
| 1685 |
+
"step": 21500
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 27.8,
|
| 1689 |
+
"learning_rate": 3.513093622795115e-05,
|
| 1690 |
+
"loss": 0.7579,
|
| 1691 |
+
"step": 21600
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 27.93,
|
| 1695 |
+
"learning_rate": 3.492740841248304e-05,
|
| 1696 |
+
"loss": 0.7687,
|
| 1697 |
+
"step": 21700
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 28.06,
|
| 1701 |
+
"learning_rate": 3.472388059701492e-05,
|
| 1702 |
+
"loss": 0.7493,
|
| 1703 |
+
"step": 21800
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 28.19,
|
| 1707 |
+
"learning_rate": 3.452035278154681e-05,
|
| 1708 |
+
"loss": 0.7279,
|
| 1709 |
+
"step": 21900
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 28.31,
|
| 1713 |
+
"learning_rate": 3.4316824966078694e-05,
|
| 1714 |
+
"loss": 0.7374,
|
| 1715 |
+
"step": 22000
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 28.31,
|
| 1719 |
+
"eval_loss": 0.16514363884925842,
|
| 1720 |
+
"eval_runtime": 662.3435,
|
| 1721 |
+
"eval_samples_per_second": 12.533,
|
| 1722 |
+
"eval_steps_per_second": 1.567,
|
| 1723 |
+
"eval_wer": 0.15674503339984386,
|
| 1724 |
+
"step": 22000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 28.44,
|
| 1728 |
+
"learning_rate": 3.411329715061058e-05,
|
| 1729 |
+
"loss": 0.7525,
|
| 1730 |
+
"step": 22100
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 28.57,
|
| 1734 |
+
"learning_rate": 3.3909769335142464e-05,
|
| 1735 |
+
"loss": 0.7336,
|
| 1736 |
+
"step": 22200
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 28.7,
|
| 1740 |
+
"learning_rate": 3.370827679782903e-05,
|
| 1741 |
+
"loss": 0.7426,
|
| 1742 |
+
"step": 22300
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 28.83,
|
| 1746 |
+
"learning_rate": 3.3504748982360924e-05,
|
| 1747 |
+
"loss": 0.7502,
|
| 1748 |
+
"step": 22400
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 28.96,
|
| 1752 |
+
"learning_rate": 3.3303256445047485e-05,
|
| 1753 |
+
"loss": 0.7499,
|
| 1754 |
+
"step": 22500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 28.96,
|
| 1758 |
+
"eval_loss": 0.1668335199356079,
|
| 1759 |
+
"eval_runtime": 660.1378,
|
| 1760 |
+
"eval_samples_per_second": 12.575,
|
| 1761 |
+
"eval_steps_per_second": 1.572,
|
| 1762 |
+
"eval_wer": 0.1548538214626529,
|
| 1763 |
+
"step": 22500
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 29.09,
|
| 1767 |
+
"learning_rate": 3.309972862957937e-05,
|
| 1768 |
+
"loss": 0.7375,
|
| 1769 |
+
"step": 22600
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 29.21,
|
| 1773 |
+
"learning_rate": 3.289620081411126e-05,
|
| 1774 |
+
"loss": 0.7397,
|
| 1775 |
+
"step": 22700
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 29.34,
|
| 1779 |
+
"learning_rate": 3.269267299864315e-05,
|
| 1780 |
+
"loss": 0.7362,
|
| 1781 |
+
"step": 22800
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 29.47,
|
| 1785 |
+
"learning_rate": 3.248914518317503e-05,
|
| 1786 |
+
"loss": 0.7417,
|
| 1787 |
+
"step": 22900
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 29.6,
|
| 1791 |
+
"learning_rate": 3.228561736770692e-05,
|
| 1792 |
+
"loss": 0.7471,
|
| 1793 |
+
"step": 23000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 29.6,
|
| 1797 |
+
"eval_loss": 0.16665612161159515,
|
| 1798 |
+
"eval_runtime": 665.4206,
|
| 1799 |
+
"eval_samples_per_second": 12.475,
|
| 1800 |
+
"eval_steps_per_second": 1.56,
|
| 1801 |
+
"eval_wer": 0.15528758566843065,
|
| 1802 |
+
"step": 23000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 29.73,
|
| 1806 |
+
"learning_rate": 3.20820895522388e-05,
|
| 1807 |
+
"loss": 0.7211,
|
| 1808 |
+
"step": 23100
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 29.86,
|
| 1812 |
+
"learning_rate": 3.187856173677069e-05,
|
| 1813 |
+
"loss": 0.7264,
|
| 1814 |
+
"step": 23200
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 29.99,
|
| 1818 |
+
"learning_rate": 3.167503392130258e-05,
|
| 1819 |
+
"loss": 0.7621,
|
| 1820 |
+
"step": 23300
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 30.12,
|
| 1824 |
+
"learning_rate": 3.147150610583446e-05,
|
| 1825 |
+
"loss": 0.7484,
|
| 1826 |
+
"step": 23400
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 30.24,
|
| 1830 |
+
"learning_rate": 3.126797829036635e-05,
|
| 1831 |
+
"loss": 0.7369,
|
| 1832 |
+
"step": 23500
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 30.24,
|
| 1836 |
+
"eval_loss": 0.16586658358573914,
|
| 1837 |
+
"eval_runtime": 659.5988,
|
| 1838 |
+
"eval_samples_per_second": 12.585,
|
| 1839 |
+
"eval_steps_per_second": 1.574,
|
| 1840 |
+
"eval_wer": 0.1555651947601284,
|
| 1841 |
+
"step": 23500
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 30.37,
|
| 1845 |
+
"learning_rate": 3.1064450474898235e-05,
|
| 1846 |
+
"loss": 0.7319,
|
| 1847 |
+
"step": 23600
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 30.5,
|
| 1851 |
+
"learning_rate": 3.086092265943012e-05,
|
| 1852 |
+
"loss": 0.7514,
|
| 1853 |
+
"step": 23700
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 30.63,
|
| 1857 |
+
"learning_rate": 3.0657394843962005e-05,
|
| 1858 |
+
"loss": 0.7341,
|
| 1859 |
+
"step": 23800
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 30.76,
|
| 1863 |
+
"learning_rate": 3.045386702849389e-05,
|
| 1864 |
+
"loss": 0.716,
|
| 1865 |
+
"step": 23900
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 30.89,
|
| 1869 |
+
"learning_rate": 3.025033921302578e-05,
|
| 1870 |
+
"loss": 0.7389,
|
| 1871 |
+
"step": 24000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 30.89,
|
| 1875 |
+
"eval_loss": 0.1668270081281662,
|
| 1876 |
+
"eval_runtime": 659.9227,
|
| 1877 |
+
"eval_samples_per_second": 12.579,
|
| 1878 |
+
"eval_steps_per_second": 1.573,
|
| 1879 |
+
"eval_wer": 0.1537954368005552,
|
| 1880 |
+
"step": 24000
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 31.02,
|
| 1884 |
+
"learning_rate": 3.0046811397557663e-05,
|
| 1885 |
+
"loss": 0.7348,
|
| 1886 |
+
"step": 24100
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 31.15,
|
| 1890 |
+
"learning_rate": 2.9843283582089552e-05,
|
| 1891 |
+
"loss": 0.7297,
|
| 1892 |
+
"step": 24200
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 31.27,
|
| 1896 |
+
"learning_rate": 2.9639755766621434e-05,
|
| 1897 |
+
"loss": 0.7189,
|
| 1898 |
+
"step": 24300
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 31.4,
|
| 1902 |
+
"learning_rate": 2.9436227951153322e-05,
|
| 1903 |
+
"loss": 0.7254,
|
| 1904 |
+
"step": 24400
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 31.53,
|
| 1908 |
+
"learning_rate": 2.9232700135685207e-05,
|
| 1909 |
+
"loss": 0.7197,
|
| 1910 |
+
"step": 24500
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 31.53,
|
| 1914 |
+
"eval_loss": 0.1687065213918686,
|
| 1915 |
+
"eval_runtime": 661.8428,
|
| 1916 |
+
"eval_samples_per_second": 12.542,
|
| 1917 |
+
"eval_steps_per_second": 1.568,
|
| 1918 |
+
"eval_wer": 0.1561204129435239,
|
| 1919 |
+
"step": 24500
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 31.66,
|
| 1923 |
+
"learning_rate": 2.9029172320217096e-05,
|
| 1924 |
+
"loss": 0.7213,
|
| 1925 |
+
"step": 24600
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 31.79,
|
| 1929 |
+
"learning_rate": 2.8825644504748977e-05,
|
| 1930 |
+
"loss": 0.7266,
|
| 1931 |
+
"step": 24700
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 31.92,
|
| 1935 |
+
"learning_rate": 2.8622116689280866e-05,
|
| 1936 |
+
"loss": 0.7088,
|
| 1937 |
+
"step": 24800
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 32.05,
|
| 1941 |
+
"learning_rate": 2.841858887381275e-05,
|
| 1942 |
+
"loss": 0.7324,
|
| 1943 |
+
"step": 24900
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 32.17,
|
| 1947 |
+
"learning_rate": 2.821506105834464e-05,
|
| 1948 |
+
"loss": 0.71,
|
| 1949 |
+
"step": 25000
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 32.17,
|
| 1953 |
+
"eval_loss": 0.16655509173870087,
|
| 1954 |
+
"eval_runtime": 666.1737,
|
| 1955 |
+
"eval_samples_per_second": 12.461,
|
| 1956 |
+
"eval_steps_per_second": 1.558,
|
| 1957 |
+
"eval_wer": 0.1516092652034354,
|
| 1958 |
+
"step": 25000
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 32.3,
|
| 1962 |
+
"learning_rate": 2.8011533242876524e-05,
|
| 1963 |
+
"loss": 0.7224,
|
| 1964 |
+
"step": 25100
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 32.43,
|
| 1968 |
+
"learning_rate": 2.780800542740841e-05,
|
| 1969 |
+
"loss": 0.7158,
|
| 1970 |
+
"step": 25200
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 32.56,
|
| 1974 |
+
"learning_rate": 2.7604477611940295e-05,
|
| 1975 |
+
"loss": 0.7154,
|
| 1976 |
+
"step": 25300
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 32.69,
|
| 1980 |
+
"learning_rate": 2.7402985074626863e-05,
|
| 1981 |
+
"loss": 0.6996,
|
| 1982 |
+
"step": 25400
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 32.82,
|
| 1986 |
+
"learning_rate": 2.719945725915875e-05,
|
| 1987 |
+
"loss": 0.7199,
|
| 1988 |
+
"step": 25500
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 32.82,
|
| 1992 |
+
"eval_loss": 0.16397617757320404,
|
| 1993 |
+
"eval_runtime": 665.0744,
|
| 1994 |
+
"eval_samples_per_second": 12.481,
|
| 1995 |
+
"eval_steps_per_second": 1.561,
|
| 1996 |
+
"eval_wer": 0.152337989069142,
|
| 1997 |
+
"step": 25500
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 32.95,
|
| 2001 |
+
"learning_rate": 2.6995929443690633e-05,
|
| 2002 |
+
"loss": 0.7132,
|
| 2003 |
+
"step": 25600
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 33.08,
|
| 2007 |
+
"learning_rate": 2.679240162822252e-05,
|
| 2008 |
+
"loss": 0.7134,
|
| 2009 |
+
"step": 25700
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 33.2,
|
| 2013 |
+
"learning_rate": 2.6588873812754406e-05,
|
| 2014 |
+
"loss": 0.6905,
|
| 2015 |
+
"step": 25800
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 33.33,
|
| 2019 |
+
"learning_rate": 2.6385345997286295e-05,
|
| 2020 |
+
"loss": 0.7102,
|
| 2021 |
+
"step": 25900
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 33.46,
|
| 2025 |
+
"learning_rate": 2.618181818181818e-05,
|
| 2026 |
+
"loss": 0.7194,
|
| 2027 |
+
"step": 26000
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 33.46,
|
| 2031 |
+
"eval_loss": 0.1659289002418518,
|
| 2032 |
+
"eval_runtime": 660.5596,
|
| 2033 |
+
"eval_samples_per_second": 12.567,
|
| 2034 |
+
"eval_steps_per_second": 1.571,
|
| 2035 |
+
"eval_wer": 0.15280645441138196,
|
| 2036 |
+
"step": 26000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 33.59,
|
| 2040 |
+
"learning_rate": 2.5978290366350065e-05,
|
| 2041 |
+
"loss": 0.7092,
|
| 2042 |
+
"step": 26100
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 33.72,
|
| 2046 |
+
"learning_rate": 2.5776797829036633e-05,
|
| 2047 |
+
"loss": 0.7152,
|
| 2048 |
+
"step": 26200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 33.85,
|
| 2052 |
+
"learning_rate": 2.557327001356852e-05,
|
| 2053 |
+
"loss": 0.7224,
|
| 2054 |
+
"step": 26300
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 33.98,
|
| 2058 |
+
"learning_rate": 2.5369742198100407e-05,
|
| 2059 |
+
"loss": 0.7078,
|
| 2060 |
+
"step": 26400
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 34.11,
|
| 2064 |
+
"learning_rate": 2.516621438263229e-05,
|
| 2065 |
+
"loss": 0.6923,
|
| 2066 |
+
"step": 26500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 34.11,
|
| 2070 |
+
"eval_loss": 0.1662396788597107,
|
| 2071 |
+
"eval_runtime": 674.0602,
|
| 2072 |
+
"eval_samples_per_second": 12.315,
|
| 2073 |
+
"eval_steps_per_second": 1.54,
|
| 2074 |
+
"eval_wer": 0.1507070356554177,
|
| 2075 |
+
"step": 26500
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 34.23,
|
| 2079 |
+
"learning_rate": 2.4962686567164177e-05,
|
| 2080 |
+
"loss": 0.7214,
|
| 2081 |
+
"step": 26600
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 34.36,
|
| 2085 |
+
"learning_rate": 2.4759158751696062e-05,
|
| 2086 |
+
"loss": 0.7136,
|
| 2087 |
+
"step": 26700
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 34.49,
|
| 2091 |
+
"learning_rate": 2.455563093622795e-05,
|
| 2092 |
+
"loss": 0.6958,
|
| 2093 |
+
"step": 26800
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 34.62,
|
| 2097 |
+
"learning_rate": 2.4352103120759832e-05,
|
| 2098 |
+
"loss": 0.7271,
|
| 2099 |
+
"step": 26900
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 34.75,
|
| 2103 |
+
"learning_rate": 2.414857530529172e-05,
|
| 2104 |
+
"loss": 0.7054,
|
| 2105 |
+
"step": 27000
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 34.75,
|
| 2109 |
+
"eval_loss": 0.16411349177360535,
|
| 2110 |
+
"eval_runtime": 660.8038,
|
| 2111 |
+
"eval_samples_per_second": 12.562,
|
| 2112 |
+
"eval_steps_per_second": 1.571,
|
| 2113 |
+
"eval_wer": 0.14857291576299123,
|
| 2114 |
+
"step": 27000
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 34.88,
|
| 2118 |
+
"learning_rate": 2.3945047489823606e-05,
|
| 2119 |
+
"loss": 0.6913,
|
| 2120 |
+
"step": 27100
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 35.01,
|
| 2124 |
+
"learning_rate": 2.3741519674355494e-05,
|
| 2125 |
+
"loss": 0.697,
|
| 2126 |
+
"step": 27200
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 35.14,
|
| 2130 |
+
"learning_rate": 2.3537991858887383e-05,
|
| 2131 |
+
"loss": 0.6981,
|
| 2132 |
+
"step": 27300
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 35.26,
|
| 2136 |
+
"learning_rate": 2.3334464043419264e-05,
|
| 2137 |
+
"loss": 0.6934,
|
| 2138 |
+
"step": 27400
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 35.39,
|
| 2142 |
+
"learning_rate": 2.313093622795115e-05,
|
| 2143 |
+
"loss": 0.6955,
|
| 2144 |
+
"step": 27500
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 35.39,
|
| 2148 |
+
"eval_loss": 0.16339896619319916,
|
| 2149 |
+
"eval_runtime": 679.5402,
|
| 2150 |
+
"eval_samples_per_second": 12.216,
|
| 2151 |
+
"eval_steps_per_second": 1.528,
|
| 2152 |
+
"eval_wer": 0.14966600156155113,
|
| 2153 |
+
"step": 27500
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 35.52,
|
| 2157 |
+
"learning_rate": 2.2927408412483038e-05,
|
| 2158 |
+
"loss": 0.7035,
|
| 2159 |
+
"step": 27600
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 35.65,
|
| 2163 |
+
"learning_rate": 2.2723880597014926e-05,
|
| 2164 |
+
"loss": 0.6886,
|
| 2165 |
+
"step": 27700
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 35.78,
|
| 2169 |
+
"learning_rate": 2.2522388059701488e-05,
|
| 2170 |
+
"loss": 0.7097,
|
| 2171 |
+
"step": 27800
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 35.91,
|
| 2175 |
+
"learning_rate": 2.2318860244233376e-05,
|
| 2176 |
+
"loss": 0.6856,
|
| 2177 |
+
"step": 27900
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 36.04,
|
| 2181 |
+
"learning_rate": 2.211533242876526e-05,
|
| 2182 |
+
"loss": 0.7084,
|
| 2183 |
+
"step": 28000
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 36.04,
|
| 2187 |
+
"eval_loss": 0.16183346509933472,
|
| 2188 |
+
"eval_runtime": 676.475,
|
| 2189 |
+
"eval_samples_per_second": 12.271,
|
| 2190 |
+
"eval_steps_per_second": 1.534,
|
| 2191 |
+
"eval_wer": 0.1477574390561291,
|
| 2192 |
+
"step": 28000
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 36.16,
|
| 2196 |
+
"learning_rate": 2.191180461329715e-05,
|
| 2197 |
+
"loss": 0.6912,
|
| 2198 |
+
"step": 28100
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 36.29,
|
| 2202 |
+
"learning_rate": 2.170827679782903e-05,
|
| 2203 |
+
"loss": 0.6962,
|
| 2204 |
+
"step": 28200
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 36.42,
|
| 2208 |
+
"learning_rate": 2.150474898236092e-05,
|
| 2209 |
+
"loss": 0.6828,
|
| 2210 |
+
"step": 28300
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 36.55,
|
| 2214 |
+
"learning_rate": 2.1301221166892805e-05,
|
| 2215 |
+
"loss": 0.6844,
|
| 2216 |
+
"step": 28400
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 36.68,
|
| 2220 |
+
"learning_rate": 2.1097693351424694e-05,
|
| 2221 |
+
"loss": 0.6917,
|
| 2222 |
+
"step": 28500
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 36.68,
|
| 2226 |
+
"eval_loss": 0.1589200794696808,
|
| 2227 |
+
"eval_runtime": 681.9046,
|
| 2228 |
+
"eval_samples_per_second": 12.173,
|
| 2229 |
+
"eval_steps_per_second": 1.522,
|
| 2230 |
+
"eval_wer": 0.14708076689511582,
|
| 2231 |
+
"step": 28500
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 36.81,
|
| 2235 |
+
"learning_rate": 2.0894165535956582e-05,
|
| 2236 |
+
"loss": 0.6813,
|
| 2237 |
+
"step": 28600
|
| 2238 |
+
},
|
| 2239 |
+
{
|
| 2240 |
+
"epoch": 36.94,
|
| 2241 |
+
"learning_rate": 2.0690637720488464e-05,
|
| 2242 |
+
"loss": 0.6917,
|
| 2243 |
+
"step": 28700
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 37.07,
|
| 2247 |
+
"learning_rate": 2.048710990502035e-05,
|
| 2248 |
+
"loss": 0.6864,
|
| 2249 |
+
"step": 28800
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 37.19,
|
| 2253 |
+
"learning_rate": 2.0283582089552237e-05,
|
| 2254 |
+
"loss": 0.6789,
|
| 2255 |
+
"step": 28900
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 37.32,
|
| 2259 |
+
"learning_rate": 2.0080054274084126e-05,
|
| 2260 |
+
"loss": 0.687,
|
| 2261 |
+
"step": 29000
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 37.32,
|
| 2265 |
+
"eval_loss": 0.15887242555618286,
|
| 2266 |
+
"eval_runtime": 671.8399,
|
| 2267 |
+
"eval_samples_per_second": 12.356,
|
| 2268 |
+
"eval_steps_per_second": 1.545,
|
| 2269 |
+
"eval_wer": 0.1450160492756138,
|
| 2270 |
+
"step": 29000
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 37.45,
|
| 2274 |
+
"learning_rate": 1.9876526458616007e-05,
|
| 2275 |
+
"loss": 0.6909,
|
| 2276 |
+
"step": 29100
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 37.58,
|
| 2280 |
+
"learning_rate": 1.9672998643147893e-05,
|
| 2281 |
+
"loss": 0.6757,
|
| 2282 |
+
"step": 29200
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 37.71,
|
| 2286 |
+
"learning_rate": 1.946947082767978e-05,
|
| 2287 |
+
"loss": 0.6963,
|
| 2288 |
+
"step": 29300
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 37.84,
|
| 2292 |
+
"learning_rate": 1.926594301221167e-05,
|
| 2293 |
+
"loss": 0.693,
|
| 2294 |
+
"step": 29400
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 37.97,
|
| 2298 |
+
"learning_rate": 1.9062415196743555e-05,
|
| 2299 |
+
"loss": 0.6914,
|
| 2300 |
+
"step": 29500
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 37.97,
|
| 2304 |
+
"eval_loss": 0.15876132249832153,
|
| 2305 |
+
"eval_runtime": 679.0742,
|
| 2306 |
+
"eval_samples_per_second": 12.224,
|
| 2307 |
+
"eval_steps_per_second": 1.529,
|
| 2308 |
+
"eval_wer": 0.14645614643879587,
|
| 2309 |
+
"step": 29500
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 38.1,
|
| 2313 |
+
"learning_rate": 1.8858887381275436e-05,
|
| 2314 |
+
"loss": 0.6769,
|
| 2315 |
+
"step": 29600
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 38.22,
|
| 2319 |
+
"learning_rate": 1.8655359565807325e-05,
|
| 2320 |
+
"loss": 0.675,
|
| 2321 |
+
"step": 29700
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 38.35,
|
| 2325 |
+
"learning_rate": 1.8451831750339213e-05,
|
| 2326 |
+
"loss": 0.6785,
|
| 2327 |
+
"step": 29800
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 38.48,
|
| 2331 |
+
"learning_rate": 1.82483039348711e-05,
|
| 2332 |
+
"loss": 0.6719,
|
| 2333 |
+
"step": 29900
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 38.61,
|
| 2337 |
+
"learning_rate": 1.8044776119402983e-05,
|
| 2338 |
+
"loss": 0.6646,
|
| 2339 |
+
"step": 30000
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 38.61,
|
| 2343 |
+
"eval_loss": 0.16017936170101166,
|
| 2344 |
+
"eval_runtime": 683.2054,
|
| 2345 |
+
"eval_samples_per_second": 12.15,
|
| 2346 |
+
"eval_steps_per_second": 1.519,
|
| 2347 |
+
"eval_wer": 0.14682050837164917,
|
| 2348 |
+
"step": 30000
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 38.74,
|
| 2352 |
+
"learning_rate": 1.784124830393487e-05,
|
| 2353 |
+
"loss": 0.6905,
|
| 2354 |
+
"step": 30100
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 38.87,
|
| 2358 |
+
"learning_rate": 1.7637720488466757e-05,
|
| 2359 |
+
"loss": 0.6724,
|
| 2360 |
+
"step": 30200
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 39.0,
|
| 2364 |
+
"learning_rate": 1.7434192672998642e-05,
|
| 2365 |
+
"loss": 0.6501,
|
| 2366 |
+
"step": 30300
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 39.12,
|
| 2370 |
+
"learning_rate": 1.7230664857530527e-05,
|
| 2371 |
+
"loss": 0.6705,
|
| 2372 |
+
"step": 30400
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 39.25,
|
| 2376 |
+
"learning_rate": 1.7027137042062416e-05,
|
| 2377 |
+
"loss": 0.6667,
|
| 2378 |
+
"step": 30500
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 39.25,
|
| 2382 |
+
"eval_loss": 0.1588352620601654,
|
| 2383 |
+
"eval_runtime": 689.4941,
|
| 2384 |
+
"eval_samples_per_second": 12.039,
|
| 2385 |
+
"eval_steps_per_second": 1.505,
|
| 2386 |
+
"eval_wer": 0.14435672768283161,
|
| 2387 |
+
"step": 30500
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 39.38,
|
| 2391 |
+
"learning_rate": 1.68236092265943e-05,
|
| 2392 |
+
"loss": 0.6622,
|
| 2393 |
+
"step": 30600
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 39.51,
|
| 2397 |
+
"learning_rate": 1.6620081411126186e-05,
|
| 2398 |
+
"loss": 0.6746,
|
| 2399 |
+
"step": 30700
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 39.64,
|
| 2403 |
+
"learning_rate": 1.641655359565807e-05,
|
| 2404 |
+
"loss": 0.6753,
|
| 2405 |
+
"step": 30800
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 39.77,
|
| 2409 |
+
"learning_rate": 1.621302578018996e-05,
|
| 2410 |
+
"loss": 0.6749,
|
| 2411 |
+
"step": 30900
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 39.9,
|
| 2415 |
+
"learning_rate": 1.6011533242876524e-05,
|
| 2416 |
+
"loss": 0.6754,
|
| 2417 |
+
"step": 31000
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 39.9,
|
| 2421 |
+
"eval_loss": 0.15873819589614868,
|
| 2422 |
+
"eval_runtime": 677.5943,
|
| 2423 |
+
"eval_samples_per_second": 12.251,
|
| 2424 |
+
"eval_steps_per_second": 1.532,
|
| 2425 |
+
"eval_wer": 0.14550186518608485,
|
| 2426 |
+
"step": 31000
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 40.03,
|
| 2430 |
+
"learning_rate": 1.5808005427408413e-05,
|
| 2431 |
+
"loss": 0.6753,
|
| 2432 |
+
"step": 31100
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 40.15,
|
| 2436 |
+
"learning_rate": 1.5604477611940298e-05,
|
| 2437 |
+
"loss": 0.6644,
|
| 2438 |
+
"step": 31200
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 40.28,
|
| 2442 |
+
"learning_rate": 1.5400949796472183e-05,
|
| 2443 |
+
"loss": 0.6619,
|
| 2444 |
+
"step": 31300
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 40.41,
|
| 2448 |
+
"learning_rate": 1.519742198100407e-05,
|
| 2449 |
+
"loss": 0.6597,
|
| 2450 |
+
"step": 31400
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 40.54,
|
| 2454 |
+
"learning_rate": 1.4993894165535955e-05,
|
| 2455 |
+
"loss": 0.6632,
|
| 2456 |
+
"step": 31500
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 40.54,
|
| 2460 |
+
"eval_loss": 0.15860143303871155,
|
| 2461 |
+
"eval_runtime": 682.8945,
|
| 2462 |
+
"eval_samples_per_second": 12.156,
|
| 2463 |
+
"eval_steps_per_second": 1.52,
|
| 2464 |
+
"eval_wer": 0.14609178450594257,
|
| 2465 |
+
"step": 31500
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 40.67,
|
| 2469 |
+
"learning_rate": 1.4790366350067841e-05,
|
| 2470 |
+
"loss": 0.6584,
|
| 2471 |
+
"step": 31600
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 40.8,
|
| 2475 |
+
"learning_rate": 1.4586838534599726e-05,
|
| 2476 |
+
"loss": 0.6524,
|
| 2477 |
+
"step": 31700
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 40.93,
|
| 2481 |
+
"learning_rate": 1.4383310719131613e-05,
|
| 2482 |
+
"loss": 0.6502,
|
| 2483 |
+
"step": 31800
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 41.06,
|
| 2487 |
+
"learning_rate": 1.4179782903663498e-05,
|
| 2488 |
+
"loss": 0.6773,
|
| 2489 |
+
"step": 31900
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 41.18,
|
| 2493 |
+
"learning_rate": 1.3976255088195385e-05,
|
| 2494 |
+
"loss": 0.6619,
|
| 2495 |
+
"step": 32000
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 41.18,
|
| 2499 |
+
"eval_loss": 0.15714022517204285,
|
| 2500 |
+
"eval_runtime": 688.8586,
|
| 2501 |
+
"eval_samples_per_second": 12.05,
|
| 2502 |
+
"eval_steps_per_second": 1.507,
|
| 2503 |
+
"eval_wer": 0.14406176802290274,
|
| 2504 |
+
"step": 32000
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 41.31,
|
| 2508 |
+
"learning_rate": 1.377272727272727e-05,
|
| 2509 |
+
"loss": 0.645,
|
| 2510 |
+
"step": 32100
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 41.44,
|
| 2514 |
+
"learning_rate": 1.3569199457259157e-05,
|
| 2515 |
+
"loss": 0.6433,
|
| 2516 |
+
"step": 32200
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 41.57,
|
| 2520 |
+
"learning_rate": 1.3365671641791044e-05,
|
| 2521 |
+
"loss": 0.6376,
|
| 2522 |
+
"step": 32300
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 41.7,
|
| 2526 |
+
"learning_rate": 1.3162143826322929e-05,
|
| 2527 |
+
"loss": 0.6542,
|
| 2528 |
+
"step": 32400
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 41.83,
|
| 2532 |
+
"learning_rate": 1.2958616010854816e-05,
|
| 2533 |
+
"loss": 0.6561,
|
| 2534 |
+
"step": 32500
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 41.83,
|
| 2538 |
+
"eval_loss": 0.15637381374835968,
|
| 2539 |
+
"eval_runtime": 690.1652,
|
| 2540 |
+
"eval_samples_per_second": 12.028,
|
| 2541 |
+
"eval_steps_per_second": 1.504,
|
| 2542 |
+
"eval_wer": 0.14204910210809404,
|
| 2543 |
+
"step": 32500
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 41.96,
|
| 2547 |
+
"learning_rate": 1.27550881953867e-05,
|
| 2548 |
+
"loss": 0.6565,
|
| 2549 |
+
"step": 32600
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 42.08,
|
| 2553 |
+
"learning_rate": 1.2551560379918588e-05,
|
| 2554 |
+
"loss": 0.6591,
|
| 2555 |
+
"step": 32700
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 42.21,
|
| 2559 |
+
"learning_rate": 1.2348032564450473e-05,
|
| 2560 |
+
"loss": 0.6511,
|
| 2561 |
+
"step": 32800
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 42.34,
|
| 2565 |
+
"learning_rate": 1.214450474898236e-05,
|
| 2566 |
+
"loss": 0.6532,
|
| 2567 |
+
"step": 32900
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 42.47,
|
| 2571 |
+
"learning_rate": 1.1940976933514244e-05,
|
| 2572 |
+
"loss": 0.6492,
|
| 2573 |
+
"step": 33000
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 42.47,
|
| 2577 |
+
"eval_loss": 0.15392744541168213,
|
| 2578 |
+
"eval_runtime": 693.1544,
|
| 2579 |
+
"eval_samples_per_second": 11.976,
|
| 2580 |
+
"eval_steps_per_second": 1.498,
|
| 2581 |
+
"eval_wer": 0.14373210722651167,
|
| 2582 |
+
"step": 33000
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 42.6,
|
| 2586 |
+
"learning_rate": 1.1739484396200814e-05,
|
| 2587 |
+
"loss": 0.6624,
|
| 2588 |
+
"step": 33100
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 42.73,
|
| 2592 |
+
"learning_rate": 1.1535956580732698e-05,
|
| 2593 |
+
"loss": 0.6608,
|
| 2594 |
+
"step": 33200
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 42.86,
|
| 2598 |
+
"learning_rate": 1.1332428765264586e-05,
|
| 2599 |
+
"loss": 0.6556,
|
| 2600 |
+
"step": 33300
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 42.99,
|
| 2604 |
+
"learning_rate": 1.112890094979647e-05,
|
| 2605 |
+
"loss": 0.6382,
|
| 2606 |
+
"step": 33400
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 43.11,
|
| 2610 |
+
"learning_rate": 1.0925373134328358e-05,
|
| 2611 |
+
"loss": 0.6649,
|
| 2612 |
+
"step": 33500
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 43.11,
|
| 2616 |
+
"eval_loss": 0.15117992460727692,
|
| 2617 |
+
"eval_runtime": 672.9964,
|
| 2618 |
+
"eval_samples_per_second": 12.334,
|
| 2619 |
+
"eval_steps_per_second": 1.542,
|
| 2620 |
+
"eval_wer": 0.14059165437668084,
|
| 2621 |
+
"step": 33500
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 43.24,
|
| 2625 |
+
"learning_rate": 1.0721845318860245e-05,
|
| 2626 |
+
"loss": 0.631,
|
| 2627 |
+
"step": 33600
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 43.37,
|
| 2631 |
+
"learning_rate": 1.051831750339213e-05,
|
| 2632 |
+
"loss": 0.6476,
|
| 2633 |
+
"step": 33700
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 43.5,
|
| 2637 |
+
"learning_rate": 1.0314789687924017e-05,
|
| 2638 |
+
"loss": 0.6498,
|
| 2639 |
+
"step": 33800
|
| 2640 |
+
},
|
| 2641 |
+
{
|
| 2642 |
+
"epoch": 43.63,
|
| 2643 |
+
"learning_rate": 1.0111261872455902e-05,
|
| 2644 |
+
"loss": 0.6426,
|
| 2645 |
+
"step": 33900
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 43.76,
|
| 2649 |
+
"learning_rate": 9.907734056987789e-06,
|
| 2650 |
+
"loss": 0.6511,
|
| 2651 |
+
"step": 34000
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 43.76,
|
| 2655 |
+
"eval_loss": 0.1538574993610382,
|
| 2656 |
+
"eval_runtime": 664.8185,
|
| 2657 |
+
"eval_samples_per_second": 12.486,
|
| 2658 |
+
"eval_steps_per_second": 1.561,
|
| 2659 |
+
"eval_wer": 0.1383534310748677,
|
| 2660 |
+
"step": 34000
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 43.89,
|
| 2664 |
+
"learning_rate": 9.704206241519674e-06,
|
| 2665 |
+
"loss": 0.6355,
|
| 2666 |
+
"step": 34100
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 44.02,
|
| 2670 |
+
"learning_rate": 9.50067842605156e-06,
|
| 2671 |
+
"loss": 0.6512,
|
| 2672 |
+
"step": 34200
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 44.14,
|
| 2676 |
+
"learning_rate": 9.297150610583445e-06,
|
| 2677 |
+
"loss": 0.6472,
|
| 2678 |
+
"step": 34300
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 44.27,
|
| 2682 |
+
"learning_rate": 9.093622795115332e-06,
|
| 2683 |
+
"loss": 0.6421,
|
| 2684 |
+
"step": 34400
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 44.4,
|
| 2688 |
+
"learning_rate": 8.890094979647217e-06,
|
| 2689 |
+
"loss": 0.6551,
|
| 2690 |
+
"step": 34500
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 44.4,
|
| 2694 |
+
"eval_loss": 0.152027428150177,
|
| 2695 |
+
"eval_runtime": 659.3672,
|
| 2696 |
+
"eval_samples_per_second": 12.589,
|
| 2697 |
+
"eval_steps_per_second": 1.574,
|
| 2698 |
+
"eval_wer": 0.13844018391602325,
|
| 2699 |
+
"step": 34500
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 44.53,
|
| 2703 |
+
"learning_rate": 8.686567164179104e-06,
|
| 2704 |
+
"loss": 0.6483,
|
| 2705 |
+
"step": 34600
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 44.66,
|
| 2709 |
+
"learning_rate": 8.48303934871099e-06,
|
| 2710 |
+
"loss": 0.6577,
|
| 2711 |
+
"step": 34700
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 44.79,
|
| 2715 |
+
"learning_rate": 8.279511533242876e-06,
|
| 2716 |
+
"loss": 0.6509,
|
| 2717 |
+
"step": 34800
|
| 2718 |
+
},
|
| 2719 |
+
{
|
| 2720 |
+
"epoch": 44.92,
|
| 2721 |
+
"learning_rate": 8.075983717774763e-06,
|
| 2722 |
+
"loss": 0.634,
|
| 2723 |
+
"step": 34900
|
| 2724 |
+
},
|
| 2725 |
+
{
|
| 2726 |
+
"epoch": 45.05,
|
| 2727 |
+
"learning_rate": 7.872455902306648e-06,
|
| 2728 |
+
"loss": 0.6452,
|
| 2729 |
+
"step": 35000
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 45.05,
|
| 2733 |
+
"eval_loss": 0.1509818583726883,
|
| 2734 |
+
"eval_runtime": 663.5705,
|
| 2735 |
+
"eval_samples_per_second": 12.51,
|
| 2736 |
+
"eval_steps_per_second": 1.564,
|
| 2737 |
+
"eval_wer": 0.13679187993406783,
|
| 2738 |
+
"step": 35000
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 45.17,
|
| 2742 |
+
"learning_rate": 7.668928086838535e-06,
|
| 2743 |
+
"loss": 0.6415,
|
| 2744 |
+
"step": 35100
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 45.3,
|
| 2748 |
+
"learning_rate": 7.46540027137042e-06,
|
| 2749 |
+
"loss": 0.6385,
|
| 2750 |
+
"step": 35200
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 45.43,
|
| 2754 |
+
"learning_rate": 7.261872455902306e-06,
|
| 2755 |
+
"loss": 0.6495,
|
| 2756 |
+
"step": 35300
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 45.56,
|
| 2760 |
+
"learning_rate": 7.058344640434192e-06,
|
| 2761 |
+
"loss": 0.6348,
|
| 2762 |
+
"step": 35400
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 45.69,
|
| 2766 |
+
"learning_rate": 6.8548168249660775e-06,
|
| 2767 |
+
"loss": 0.6155,
|
| 2768 |
+
"step": 35500
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 45.69,
|
| 2772 |
+
"eval_loss": 0.15219098329544067,
|
| 2773 |
+
"eval_runtime": 665.3938,
|
| 2774 |
+
"eval_samples_per_second": 12.475,
|
| 2775 |
+
"eval_steps_per_second": 1.56,
|
| 2776 |
+
"eval_wer": 0.13750325323154333,
|
| 2777 |
+
"step": 35500
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 45.82,
|
| 2781 |
+
"learning_rate": 6.6512890094979635e-06,
|
| 2782 |
+
"loss": 0.6502,
|
| 2783 |
+
"step": 35600
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 45.95,
|
| 2787 |
+
"learning_rate": 6.447761194029849e-06,
|
| 2788 |
+
"loss": 0.6522,
|
| 2789 |
+
"step": 35700
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 46.07,
|
| 2793 |
+
"learning_rate": 6.244233378561737e-06,
|
| 2794 |
+
"loss": 0.6241,
|
| 2795 |
+
"step": 35800
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 46.2,
|
| 2799 |
+
"learning_rate": 6.040705563093623e-06,
|
| 2800 |
+
"loss": 0.6403,
|
| 2801 |
+
"step": 35900
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 46.33,
|
| 2805 |
+
"learning_rate": 5.837177747625509e-06,
|
| 2806 |
+
"loss": 0.628,
|
| 2807 |
+
"step": 36000
|
| 2808 |
+
},
|
| 2809 |
+
{
|
| 2810 |
+
"epoch": 46.33,
|
| 2811 |
+
"eval_loss": 0.15221403539180756,
|
| 2812 |
+
"eval_runtime": 658.0569,
|
| 2813 |
+
"eval_samples_per_second": 12.614,
|
| 2814 |
+
"eval_steps_per_second": 1.577,
|
| 2815 |
+
"eval_wer": 0.13660102368352564,
|
| 2816 |
+
"step": 36000
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 46.46,
|
| 2820 |
+
"learning_rate": 5.633649932157395e-06,
|
| 2821 |
+
"loss": 0.6394,
|
| 2822 |
+
"step": 36100
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 46.59,
|
| 2826 |
+
"learning_rate": 5.430122116689281e-06,
|
| 2827 |
+
"loss": 0.6202,
|
| 2828 |
+
"step": 36200
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 46.72,
|
| 2832 |
+
"learning_rate": 5.226594301221167e-06,
|
| 2833 |
+
"loss": 0.6261,
|
| 2834 |
+
"step": 36300
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 46.85,
|
| 2838 |
+
"learning_rate": 5.023066485753053e-06,
|
| 2839 |
+
"loss": 0.6158,
|
| 2840 |
+
"step": 36400
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 46.97,
|
| 2844 |
+
"learning_rate": 4.8195386702849386e-06,
|
| 2845 |
+
"loss": 0.6389,
|
| 2846 |
+
"step": 36500
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 46.97,
|
| 2850 |
+
"eval_loss": 0.15134426951408386,
|
| 2851 |
+
"eval_runtime": 663.8499,
|
| 2852 |
+
"eval_samples_per_second": 12.504,
|
| 2853 |
+
"eval_steps_per_second": 1.564,
|
| 2854 |
+
"eval_wer": 0.13774616118677888,
|
| 2855 |
+
"step": 36500
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 47.1,
|
| 2859 |
+
"learning_rate": 4.6160108548168245e-06,
|
| 2860 |
+
"loss": 0.6383,
|
| 2861 |
+
"step": 36600
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 47.23,
|
| 2865 |
+
"learning_rate": 4.4124830393487104e-06,
|
| 2866 |
+
"loss": 0.638,
|
| 2867 |
+
"step": 36700
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 47.36,
|
| 2871 |
+
"learning_rate": 4.208955223880596e-06,
|
| 2872 |
+
"loss": 0.6193,
|
| 2873 |
+
"step": 36800
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 47.49,
|
| 2877 |
+
"learning_rate": 4.005427408412483e-06,
|
| 2878 |
+
"loss": 0.6313,
|
| 2879 |
+
"step": 36900
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 47.62,
|
| 2883 |
+
"learning_rate": 3.801899592944369e-06,
|
| 2884 |
+
"loss": 0.6265,
|
| 2885 |
+
"step": 37000
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 47.62,
|
| 2889 |
+
"eval_loss": 0.15123163163661957,
|
| 2890 |
+
"eval_runtime": 661.0461,
|
| 2891 |
+
"eval_samples_per_second": 12.557,
|
| 2892 |
+
"eval_steps_per_second": 1.57,
|
| 2893 |
+
"eval_wer": 0.13691333391168561,
|
| 2894 |
+
"step": 37000
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 47.75,
|
| 2898 |
+
"learning_rate": 3.5983717774762546e-06,
|
| 2899 |
+
"loss": 0.6362,
|
| 2900 |
+
"step": 37100
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 47.88,
|
| 2904 |
+
"learning_rate": 3.3948439620081405e-06,
|
| 2905 |
+
"loss": 0.6336,
|
| 2906 |
+
"step": 37200
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 48.01,
|
| 2910 |
+
"learning_rate": 3.1913161465400265e-06,
|
| 2911 |
+
"loss": 0.6276,
|
| 2912 |
+
"step": 37300
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 48.13,
|
| 2916 |
+
"learning_rate": 2.9877883310719132e-06,
|
| 2917 |
+
"loss": 0.623,
|
| 2918 |
+
"step": 37400
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 48.26,
|
| 2922 |
+
"learning_rate": 2.784260515603799e-06,
|
| 2923 |
+
"loss": 0.6197,
|
| 2924 |
+
"step": 37500
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 48.26,
|
| 2928 |
+
"eval_loss": 0.1511179655790329,
|
| 2929 |
+
"eval_runtime": 661.0804,
|
| 2930 |
+
"eval_samples_per_second": 12.557,
|
| 2931 |
+
"eval_steps_per_second": 1.57,
|
| 2932 |
+
"eval_wer": 0.136184610045979,
|
| 2933 |
+
"step": 37500
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 48.39,
|
| 2937 |
+
"learning_rate": 2.580732700135685e-06,
|
| 2938 |
+
"loss": 0.6209,
|
| 2939 |
+
"step": 37600
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 48.52,
|
| 2943 |
+
"learning_rate": 2.377204884667571e-06,
|
| 2944 |
+
"loss": 0.6361,
|
| 2945 |
+
"step": 37700
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 48.65,
|
| 2949 |
+
"learning_rate": 2.173677069199457e-06,
|
| 2950 |
+
"loss": 0.6226,
|
| 2951 |
+
"step": 37800
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 48.78,
|
| 2955 |
+
"learning_rate": 1.970149253731343e-06,
|
| 2956 |
+
"loss": 0.6446,
|
| 2957 |
+
"step": 37900
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 48.91,
|
| 2961 |
+
"learning_rate": 1.7666214382632293e-06,
|
| 2962 |
+
"loss": 0.621,
|
| 2963 |
+
"step": 38000
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 48.91,
|
| 2967 |
+
"eval_loss": 0.1510026901960373,
|
| 2968 |
+
"eval_runtime": 664.8917,
|
| 2969 |
+
"eval_samples_per_second": 12.485,
|
| 2970 |
+
"eval_steps_per_second": 1.561,
|
| 2971 |
+
"eval_wer": 0.13566409299904572,
|
| 2972 |
+
"step": 38000
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 49.03,
|
| 2976 |
+
"learning_rate": 1.5630936227951152e-06,
|
| 2977 |
+
"loss": 0.6328,
|
| 2978 |
+
"step": 38100
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 49.16,
|
| 2982 |
+
"learning_rate": 1.3595658073270013e-06,
|
| 2983 |
+
"loss": 0.6267,
|
| 2984 |
+
"step": 38200
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 49.29,
|
| 2988 |
+
"learning_rate": 1.1560379918588873e-06,
|
| 2989 |
+
"loss": 0.6346,
|
| 2990 |
+
"step": 38300
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 49.42,
|
| 2994 |
+
"learning_rate": 9.525101763907733e-07,
|
| 2995 |
+
"loss": 0.647,
|
| 2996 |
+
"step": 38400
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 49.55,
|
| 3000 |
+
"learning_rate": 7.489823609226593e-07,
|
| 3001 |
+
"loss": 0.6259,
|
| 3002 |
+
"step": 38500
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 49.55,
|
| 3006 |
+
"eval_loss": 0.15063393115997314,
|
| 3007 |
+
"eval_runtime": 657.6897,
|
| 3008 |
+
"eval_samples_per_second": 12.621,
|
| 3009 |
+
"eval_steps_per_second": 1.578,
|
| 3010 |
+
"eval_wer": 0.13528238049796132,
|
| 3011 |
+
"step": 38500
|
| 3012 |
+
},
|
| 3013 |
+
{
|
| 3014 |
+
"epoch": 49.68,
|
| 3015 |
+
"learning_rate": 5.454545454545454e-07,
|
| 3016 |
+
"loss": 0.632,
|
| 3017 |
+
"step": 38600
|
| 3018 |
+
},
|
| 3019 |
+
{
|
| 3020 |
+
"epoch": 49.81,
|
| 3021 |
+
"learning_rate": 3.419267299864315e-07,
|
| 3022 |
+
"loss": 0.6375,
|
| 3023 |
+
"step": 38700
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 49.94,
|
| 3027 |
+
"learning_rate": 1.3839891451831749e-07,
|
| 3028 |
+
"loss": 0.6163,
|
| 3029 |
+
"step": 38800
|
| 3030 |
+
},
|
| 3031 |
+
{
|
| 3032 |
+
"epoch": 50.0,
|
| 3033 |
+
"step": 38850,
|
| 3034 |
+
"total_flos": 1.5592073449484252e+20,
|
| 3035 |
+
"train_loss": 0.9556344533304143,
|
| 3036 |
+
"train_runtime": 205033.5418,
|
| 3037 |
+
"train_samples_per_second": 6.067,
|
| 3038 |
+
"train_steps_per_second": 0.189
|
| 3039 |
+
}
|
| 3040 |
+
],
|
| 3041 |
+
"max_steps": 38850,
|
| 3042 |
+
"num_train_epochs": 50,
|
| 3043 |
+
"total_flos": 1.5592073449484252e+20,
|
| 3044 |
+
"trial_name": null,
|
| 3045 |
+
"trial_params": null
|
| 3046 |
+
}
|