End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.456202894449234,
|
| 4 |
+
"eval_runtime": 81.3467,
|
| 5 |
+
"eval_samples": 827,
|
| 6 |
+
"eval_samples_per_second": 10.166,
|
| 7 |
+
"eval_steps_per_second": 1.278,
|
| 8 |
+
"eval_wer": 0.15311909262759923,
|
| 9 |
+
"train_loss": 0.23119104407936014,
|
| 10 |
+
"train_runtime": 64754.7357,
|
| 11 |
+
"train_samples": 7439,
|
| 12 |
+
"train_samples_per_second": 5.744,
|
| 13 |
+
"train_steps_per_second": 0.359
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.456202894449234,
|
| 4 |
+
"eval_runtime": 81.3467,
|
| 5 |
+
"eval_samples": 827,
|
| 6 |
+
"eval_samples_per_second": 10.166,
|
| 7 |
+
"eval_steps_per_second": 1.278,
|
| 8 |
+
"eval_wer": 0.15311909262759923
|
| 9 |
+
}
|
runs/Feb04_05-15-53_job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba/events.out.tfevents.1644017044.job-c93f32d8-97c5-48e7-b5ec-c6c950f627ba.3825650.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e6b28ef1219f0ee47ac24fca77dd5658277accfabe9b1aaf4b4a0d71150ce753
|
| 3 |
+
size 364
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"train_loss": 0.23119104407936014,
|
| 4 |
+
"train_runtime": 64754.7357,
|
| 5 |
+
"train_samples": 7439,
|
| 6 |
+
"train_samples_per_second": 5.744,
|
| 7 |
+
"train_steps_per_second": 0.359
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1831 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"global_step": 23250,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.22,
|
| 12 |
+
"learning_rate": 2.8500000000000002e-06,
|
| 13 |
+
"loss": 7.8132,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.43,
|
| 18 |
+
"learning_rate": 5.82e-06,
|
| 19 |
+
"loss": 3.053,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.65,
|
| 24 |
+
"learning_rate": 8.82e-06,
|
| 25 |
+
"loss": 2.9223,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.86,
|
| 30 |
+
"learning_rate": 1.182e-05,
|
| 31 |
+
"loss": 2.6894,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.08,
|
| 36 |
+
"learning_rate": 1.482e-05,
|
| 37 |
+
"loss": 1.6894,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.08,
|
| 42 |
+
"eval_loss": 1.2422521114349365,
|
| 43 |
+
"eval_runtime": 81.2578,
|
| 44 |
+
"eval_samples_per_second": 10.177,
|
| 45 |
+
"eval_steps_per_second": 1.28,
|
| 46 |
+
"eval_wer": 0.8619092627599244,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 1.29,
|
| 51 |
+
"learning_rate": 1.782e-05,
|
| 52 |
+
"loss": 1.3281,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 1.51,
|
| 57 |
+
"learning_rate": 2.082e-05,
|
| 58 |
+
"loss": 1.1146,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.72,
|
| 63 |
+
"learning_rate": 2.3820000000000002e-05,
|
| 64 |
+
"loss": 0.9295,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 1.94,
|
| 69 |
+
"learning_rate": 2.682e-05,
|
| 70 |
+
"loss": 0.8392,
|
| 71 |
+
"step": 900
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 2.15,
|
| 75 |
+
"learning_rate": 2.982e-05,
|
| 76 |
+
"loss": 0.7543,
|
| 77 |
+
"step": 1000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 2.15,
|
| 81 |
+
"eval_loss": 0.5956054329872131,
|
| 82 |
+
"eval_runtime": 81.5082,
|
| 83 |
+
"eval_samples_per_second": 10.146,
|
| 84 |
+
"eval_steps_per_second": 1.276,
|
| 85 |
+
"eval_wer": 0.38166351606805293,
|
| 86 |
+
"step": 1000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.37,
|
| 90 |
+
"learning_rate": 3.282e-05,
|
| 91 |
+
"loss": 0.7128,
|
| 92 |
+
"step": 1100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.58,
|
| 96 |
+
"learning_rate": 3.582e-05,
|
| 97 |
+
"loss": 0.6859,
|
| 98 |
+
"step": 1200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.8,
|
| 102 |
+
"learning_rate": 3.8820000000000004e-05,
|
| 103 |
+
"loss": 0.6024,
|
| 104 |
+
"step": 1300
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 3.01,
|
| 108 |
+
"learning_rate": 4.1819999999999996e-05,
|
| 109 |
+
"loss": 0.6365,
|
| 110 |
+
"step": 1400
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 3.23,
|
| 114 |
+
"learning_rate": 4.482e-05,
|
| 115 |
+
"loss": 0.5481,
|
| 116 |
+
"step": 1500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 3.23,
|
| 120 |
+
"eval_loss": 0.5042839646339417,
|
| 121 |
+
"eval_runtime": 83.6557,
|
| 122 |
+
"eval_samples_per_second": 9.886,
|
| 123 |
+
"eval_steps_per_second": 1.243,
|
| 124 |
+
"eval_wer": 0.32457466918714556,
|
| 125 |
+
"step": 1500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 3.44,
|
| 129 |
+
"learning_rate": 4.7820000000000006e-05,
|
| 130 |
+
"loss": 0.5648,
|
| 131 |
+
"step": 1600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 3.66,
|
| 135 |
+
"learning_rate": 5.082e-05,
|
| 136 |
+
"loss": 0.555,
|
| 137 |
+
"step": 1700
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 3.87,
|
| 141 |
+
"learning_rate": 5.382e-05,
|
| 142 |
+
"loss": 0.5723,
|
| 143 |
+
"step": 1800
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 4.09,
|
| 147 |
+
"learning_rate": 5.682e-05,
|
| 148 |
+
"loss": 0.4716,
|
| 149 |
+
"step": 1900
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 4.3,
|
| 153 |
+
"learning_rate": 5.982e-05,
|
| 154 |
+
"loss": 0.4661,
|
| 155 |
+
"step": 2000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 4.3,
|
| 159 |
+
"eval_loss": 0.48127105832099915,
|
| 160 |
+
"eval_runtime": 81.5918,
|
| 161 |
+
"eval_samples_per_second": 10.136,
|
| 162 |
+
"eval_steps_per_second": 1.275,
|
| 163 |
+
"eval_wer": 0.2793005671077505,
|
| 164 |
+
"step": 2000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.52,
|
| 168 |
+
"learning_rate": 5.973458823529412e-05,
|
| 169 |
+
"loss": 0.4756,
|
| 170 |
+
"step": 2100
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 4.73,
|
| 174 |
+
"learning_rate": 5.945223529411765e-05,
|
| 175 |
+
"loss": 0.4022,
|
| 176 |
+
"step": 2200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 4.95,
|
| 180 |
+
"learning_rate": 5.9169882352941175e-05,
|
| 181 |
+
"loss": 0.4314,
|
| 182 |
+
"step": 2300
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 5.16,
|
| 186 |
+
"learning_rate": 5.8887529411764706e-05,
|
| 187 |
+
"loss": 0.4154,
|
| 188 |
+
"step": 2400
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 5.38,
|
| 192 |
+
"learning_rate": 5.8605176470588236e-05,
|
| 193 |
+
"loss": 0.3901,
|
| 194 |
+
"step": 2500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 5.38,
|
| 198 |
+
"eval_loss": 0.4370500147342682,
|
| 199 |
+
"eval_runtime": 81.2387,
|
| 200 |
+
"eval_samples_per_second": 10.18,
|
| 201 |
+
"eval_steps_per_second": 1.28,
|
| 202 |
+
"eval_wer": 0.25916824196597354,
|
| 203 |
+
"step": 2500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 5.59,
|
| 207 |
+
"learning_rate": 5.8322823529411767e-05,
|
| 208 |
+
"loss": 0.4014,
|
| 209 |
+
"step": 2600
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 5.81,
|
| 213 |
+
"learning_rate": 5.80404705882353e-05,
|
| 214 |
+
"loss": 0.3762,
|
| 215 |
+
"step": 2700
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 6.02,
|
| 219 |
+
"learning_rate": 5.775811764705882e-05,
|
| 220 |
+
"loss": 0.3777,
|
| 221 |
+
"step": 2800
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 6.24,
|
| 225 |
+
"learning_rate": 5.747576470588236e-05,
|
| 226 |
+
"loss": 0.3342,
|
| 227 |
+
"step": 2900
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 6.45,
|
| 231 |
+
"learning_rate": 5.719341176470588e-05,
|
| 232 |
+
"loss": 0.3512,
|
| 233 |
+
"step": 3000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 6.45,
|
| 237 |
+
"eval_loss": 0.4216073155403137,
|
| 238 |
+
"eval_runtime": 81.8015,
|
| 239 |
+
"eval_samples_per_second": 10.11,
|
| 240 |
+
"eval_steps_per_second": 1.271,
|
| 241 |
+
"eval_wer": 0.24584120982986768,
|
| 242 |
+
"step": 3000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 6.67,
|
| 246 |
+
"learning_rate": 5.691105882352942e-05,
|
| 247 |
+
"loss": 0.3204,
|
| 248 |
+
"step": 3100
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 6.88,
|
| 252 |
+
"learning_rate": 5.662870588235294e-05,
|
| 253 |
+
"loss": 0.3232,
|
| 254 |
+
"step": 3200
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 7.1,
|
| 258 |
+
"learning_rate": 5.634635294117647e-05,
|
| 259 |
+
"loss": 0.2976,
|
| 260 |
+
"step": 3300
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 7.31,
|
| 264 |
+
"learning_rate": 5.6064000000000004e-05,
|
| 265 |
+
"loss": 0.296,
|
| 266 |
+
"step": 3400
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 7.53,
|
| 270 |
+
"learning_rate": 5.578164705882353e-05,
|
| 271 |
+
"loss": 0.3016,
|
| 272 |
+
"step": 3500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 7.53,
|
| 276 |
+
"eval_loss": 0.3813554048538208,
|
| 277 |
+
"eval_runtime": 83.1204,
|
| 278 |
+
"eval_samples_per_second": 9.949,
|
| 279 |
+
"eval_steps_per_second": 1.251,
|
| 280 |
+
"eval_wer": 0.22570888468809075,
|
| 281 |
+
"step": 3500
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 7.74,
|
| 285 |
+
"learning_rate": 5.549929411764706e-05,
|
| 286 |
+
"loss": 0.2726,
|
| 287 |
+
"step": 3600
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 7.96,
|
| 291 |
+
"learning_rate": 5.521694117647059e-05,
|
| 292 |
+
"loss": 0.2953,
|
| 293 |
+
"step": 3700
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 8.17,
|
| 297 |
+
"learning_rate": 5.493458823529412e-05,
|
| 298 |
+
"loss": 0.2647,
|
| 299 |
+
"step": 3800
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 8.39,
|
| 303 |
+
"learning_rate": 5.465223529411765e-05,
|
| 304 |
+
"loss": 0.2524,
|
| 305 |
+
"step": 3900
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 8.6,
|
| 309 |
+
"learning_rate": 5.436988235294118e-05,
|
| 310 |
+
"loss": 0.278,
|
| 311 |
+
"step": 4000
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 8.6,
|
| 315 |
+
"eval_loss": 0.41506877541542053,
|
| 316 |
+
"eval_runtime": 81.9008,
|
| 317 |
+
"eval_samples_per_second": 10.098,
|
| 318 |
+
"eval_steps_per_second": 1.27,
|
| 319 |
+
"eval_wer": 0.21446124763705104,
|
| 320 |
+
"step": 4000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 8.82,
|
| 324 |
+
"learning_rate": 5.4087529411764704e-05,
|
| 325 |
+
"loss": 0.2559,
|
| 326 |
+
"step": 4100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 9.03,
|
| 330 |
+
"learning_rate": 5.380517647058824e-05,
|
| 331 |
+
"loss": 0.2669,
|
| 332 |
+
"step": 4200
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 9.25,
|
| 336 |
+
"learning_rate": 5.3522823529411765e-05,
|
| 337 |
+
"loss": 0.2189,
|
| 338 |
+
"step": 4300
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 9.46,
|
| 342 |
+
"learning_rate": 5.3240470588235296e-05,
|
| 343 |
+
"loss": 0.2321,
|
| 344 |
+
"step": 4400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 9.68,
|
| 348 |
+
"learning_rate": 5.2958117647058826e-05,
|
| 349 |
+
"loss": 0.2435,
|
| 350 |
+
"step": 4500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 9.68,
|
| 354 |
+
"eval_loss": 0.48163077235221863,
|
| 355 |
+
"eval_runtime": 80.9004,
|
| 356 |
+
"eval_samples_per_second": 10.222,
|
| 357 |
+
"eval_steps_per_second": 1.286,
|
| 358 |
+
"eval_wer": 0.21304347826086956,
|
| 359 |
+
"step": 4500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 9.89,
|
| 363 |
+
"learning_rate": 5.267576470588235e-05,
|
| 364 |
+
"loss": 0.2348,
|
| 365 |
+
"step": 4600
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 10.11,
|
| 369 |
+
"learning_rate": 5.239341176470589e-05,
|
| 370 |
+
"loss": 0.2266,
|
| 371 |
+
"step": 4700
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 10.32,
|
| 375 |
+
"learning_rate": 5.211105882352941e-05,
|
| 376 |
+
"loss": 0.2345,
|
| 377 |
+
"step": 4800
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 10.54,
|
| 381 |
+
"learning_rate": 5.182870588235294e-05,
|
| 382 |
+
"loss": 0.2266,
|
| 383 |
+
"step": 4900
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 10.75,
|
| 387 |
+
"learning_rate": 5.154635294117647e-05,
|
| 388 |
+
"loss": 0.2122,
|
| 389 |
+
"step": 5000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 10.75,
|
| 393 |
+
"eval_loss": 0.4489321708679199,
|
| 394 |
+
"eval_runtime": 81.0474,
|
| 395 |
+
"eval_samples_per_second": 10.204,
|
| 396 |
+
"eval_steps_per_second": 1.283,
|
| 397 |
+
"eval_wer": 0.21370510396975426,
|
| 398 |
+
"step": 5000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 10.97,
|
| 402 |
+
"learning_rate": 5.1264e-05,
|
| 403 |
+
"loss": 0.2038,
|
| 404 |
+
"step": 5100
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 11.18,
|
| 408 |
+
"learning_rate": 5.098164705882353e-05,
|
| 409 |
+
"loss": 0.1916,
|
| 410 |
+
"step": 5200
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 11.4,
|
| 414 |
+
"learning_rate": 5.069929411764706e-05,
|
| 415 |
+
"loss": 0.1896,
|
| 416 |
+
"step": 5300
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 11.61,
|
| 420 |
+
"learning_rate": 5.041694117647059e-05,
|
| 421 |
+
"loss": 0.207,
|
| 422 |
+
"step": 5400
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 11.83,
|
| 426 |
+
"learning_rate": 5.013458823529412e-05,
|
| 427 |
+
"loss": 0.1949,
|
| 428 |
+
"step": 5500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 11.83,
|
| 432 |
+
"eval_loss": 0.3977554738521576,
|
| 433 |
+
"eval_runtime": 80.708,
|
| 434 |
+
"eval_samples_per_second": 10.247,
|
| 435 |
+
"eval_steps_per_second": 1.289,
|
| 436 |
+
"eval_wer": 0.20633270321361058,
|
| 437 |
+
"step": 5500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 12.04,
|
| 441 |
+
"learning_rate": 4.985223529411765e-05,
|
| 442 |
+
"loss": 0.2177,
|
| 443 |
+
"step": 5600
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 12.26,
|
| 447 |
+
"learning_rate": 4.956988235294118e-05,
|
| 448 |
+
"loss": 0.182,
|
| 449 |
+
"step": 5700
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 12.47,
|
| 453 |
+
"learning_rate": 4.928752941176471e-05,
|
| 454 |
+
"loss": 0.1809,
|
| 455 |
+
"step": 5800
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 12.69,
|
| 459 |
+
"learning_rate": 4.900517647058823e-05,
|
| 460 |
+
"loss": 0.1876,
|
| 461 |
+
"step": 5900
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 12.9,
|
| 465 |
+
"learning_rate": 4.872282352941177e-05,
|
| 466 |
+
"loss": 0.1929,
|
| 467 |
+
"step": 6000
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 12.9,
|
| 471 |
+
"eval_loss": 0.38229823112487793,
|
| 472 |
+
"eval_runtime": 82.0352,
|
| 473 |
+
"eval_samples_per_second": 10.081,
|
| 474 |
+
"eval_steps_per_second": 1.268,
|
| 475 |
+
"eval_wer": 0.20264650283553876,
|
| 476 |
+
"step": 6000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 13.12,
|
| 480 |
+
"learning_rate": 4.8440470588235294e-05,
|
| 481 |
+
"loss": 0.1952,
|
| 482 |
+
"step": 6100
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 13.33,
|
| 486 |
+
"learning_rate": 4.8158117647058825e-05,
|
| 487 |
+
"loss": 0.1739,
|
| 488 |
+
"step": 6200
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 13.55,
|
| 492 |
+
"learning_rate": 4.7875764705882355e-05,
|
| 493 |
+
"loss": 0.1748,
|
| 494 |
+
"step": 6300
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 13.76,
|
| 498 |
+
"learning_rate": 4.759341176470588e-05,
|
| 499 |
+
"loss": 0.1625,
|
| 500 |
+
"step": 6400
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 13.98,
|
| 504 |
+
"learning_rate": 4.7311058823529416e-05,
|
| 505 |
+
"loss": 0.1757,
|
| 506 |
+
"step": 6500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 13.98,
|
| 510 |
+
"eval_loss": 0.3409084379673004,
|
| 511 |
+
"eval_runtime": 80.5311,
|
| 512 |
+
"eval_samples_per_second": 10.269,
|
| 513 |
+
"eval_steps_per_second": 1.291,
|
| 514 |
+
"eval_wer": 0.19650283553875236,
|
| 515 |
+
"step": 6500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 14.19,
|
| 519 |
+
"learning_rate": 4.702870588235294e-05,
|
| 520 |
+
"loss": 0.1422,
|
| 521 |
+
"step": 6600
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 14.41,
|
| 525 |
+
"learning_rate": 4.674635294117647e-05,
|
| 526 |
+
"loss": 0.1571,
|
| 527 |
+
"step": 6700
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 14.62,
|
| 531 |
+
"learning_rate": 4.6464e-05,
|
| 532 |
+
"loss": 0.1565,
|
| 533 |
+
"step": 6800
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 14.84,
|
| 537 |
+
"learning_rate": 4.618164705882353e-05,
|
| 538 |
+
"loss": 0.1717,
|
| 539 |
+
"step": 6900
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 15.05,
|
| 543 |
+
"learning_rate": 4.589929411764706e-05,
|
| 544 |
+
"loss": 0.1771,
|
| 545 |
+
"step": 7000
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 15.05,
|
| 549 |
+
"eval_loss": 0.3844490051269531,
|
| 550 |
+
"eval_runtime": 81.332,
|
| 551 |
+
"eval_samples_per_second": 10.168,
|
| 552 |
+
"eval_steps_per_second": 1.279,
|
| 553 |
+
"eval_wer": 0.19357277882797733,
|
| 554 |
+
"step": 7000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 15.27,
|
| 558 |
+
"learning_rate": 4.561694117647059e-05,
|
| 559 |
+
"loss": 0.1408,
|
| 560 |
+
"step": 7100
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 15.48,
|
| 564 |
+
"learning_rate": 4.5334588235294116e-05,
|
| 565 |
+
"loss": 0.1512,
|
| 566 |
+
"step": 7200
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 15.7,
|
| 570 |
+
"learning_rate": 4.505223529411765e-05,
|
| 571 |
+
"loss": 0.1505,
|
| 572 |
+
"step": 7300
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 15.91,
|
| 576 |
+
"learning_rate": 4.476988235294118e-05,
|
| 577 |
+
"loss": 0.1436,
|
| 578 |
+
"step": 7400
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 16.13,
|
| 582 |
+
"learning_rate": 4.448752941176471e-05,
|
| 583 |
+
"loss": 0.1452,
|
| 584 |
+
"step": 7500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 16.13,
|
| 588 |
+
"eval_loss": 0.37490135431289673,
|
| 589 |
+
"eval_runtime": 81.4843,
|
| 590 |
+
"eval_samples_per_second": 10.149,
|
| 591 |
+
"eval_steps_per_second": 1.276,
|
| 592 |
+
"eval_wer": 0.1899810964083176,
|
| 593 |
+
"step": 7500
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 16.34,
|
| 597 |
+
"learning_rate": 4.420517647058824e-05,
|
| 598 |
+
"loss": 0.1462,
|
| 599 |
+
"step": 7600
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 16.56,
|
| 603 |
+
"learning_rate": 4.392282352941176e-05,
|
| 604 |
+
"loss": 0.1425,
|
| 605 |
+
"step": 7700
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 16.77,
|
| 609 |
+
"learning_rate": 4.36404705882353e-05,
|
| 610 |
+
"loss": 0.144,
|
| 611 |
+
"step": 7800
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 16.99,
|
| 615 |
+
"learning_rate": 4.335811764705882e-05,
|
| 616 |
+
"loss": 0.1622,
|
| 617 |
+
"step": 7900
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 17.2,
|
| 621 |
+
"learning_rate": 4.3075764705882354e-05,
|
| 622 |
+
"loss": 0.1341,
|
| 623 |
+
"step": 8000
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 17.2,
|
| 627 |
+
"eval_loss": 0.44071856141090393,
|
| 628 |
+
"eval_runtime": 82.1207,
|
| 629 |
+
"eval_samples_per_second": 10.071,
|
| 630 |
+
"eval_steps_per_second": 1.266,
|
| 631 |
+
"eval_wer": 0.20264650283553876,
|
| 632 |
+
"step": 8000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 17.42,
|
| 636 |
+
"learning_rate": 4.2793411764705884e-05,
|
| 637 |
+
"loss": 0.1151,
|
| 638 |
+
"step": 8100
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 17.63,
|
| 642 |
+
"learning_rate": 4.251105882352941e-05,
|
| 643 |
+
"loss": 0.1477,
|
| 644 |
+
"step": 8200
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 17.85,
|
| 648 |
+
"learning_rate": 4.2228705882352945e-05,
|
| 649 |
+
"loss": 0.1377,
|
| 650 |
+
"step": 8300
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 18.06,
|
| 654 |
+
"learning_rate": 4.194635294117647e-05,
|
| 655 |
+
"loss": 0.129,
|
| 656 |
+
"step": 8400
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 18.28,
|
| 660 |
+
"learning_rate": 4.1664e-05,
|
| 661 |
+
"loss": 0.13,
|
| 662 |
+
"step": 8500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 18.28,
|
| 666 |
+
"eval_loss": 0.42531654238700867,
|
| 667 |
+
"eval_runtime": 82.3744,
|
| 668 |
+
"eval_samples_per_second": 10.04,
|
| 669 |
+
"eval_steps_per_second": 1.263,
|
| 670 |
+
"eval_wer": 0.1882797731568998,
|
| 671 |
+
"step": 8500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 18.49,
|
| 675 |
+
"learning_rate": 4.138164705882353e-05,
|
| 676 |
+
"loss": 0.1353,
|
| 677 |
+
"step": 8600
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 18.71,
|
| 681 |
+
"learning_rate": 4.109929411764706e-05,
|
| 682 |
+
"loss": 0.1216,
|
| 683 |
+
"step": 8700
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 18.92,
|
| 687 |
+
"learning_rate": 4.081694117647059e-05,
|
| 688 |
+
"loss": 0.1301,
|
| 689 |
+
"step": 8800
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 19.14,
|
| 693 |
+
"learning_rate": 4.053458823529412e-05,
|
| 694 |
+
"loss": 0.1195,
|
| 695 |
+
"step": 8900
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 19.35,
|
| 699 |
+
"learning_rate": 4.0252235294117645e-05,
|
| 700 |
+
"loss": 0.1183,
|
| 701 |
+
"step": 9000
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 19.35,
|
| 705 |
+
"eval_loss": 0.43106988072395325,
|
| 706 |
+
"eval_runtime": 80.8241,
|
| 707 |
+
"eval_samples_per_second": 10.232,
|
| 708 |
+
"eval_steps_per_second": 1.287,
|
| 709 |
+
"eval_wer": 0.18799621928166352,
|
| 710 |
+
"step": 9000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 19.57,
|
| 714 |
+
"learning_rate": 3.996988235294118e-05,
|
| 715 |
+
"loss": 0.1147,
|
| 716 |
+
"step": 9100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 19.78,
|
| 720 |
+
"learning_rate": 3.9687529411764706e-05,
|
| 721 |
+
"loss": 0.1178,
|
| 722 |
+
"step": 9200
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 20.0,
|
| 726 |
+
"learning_rate": 3.940517647058823e-05,
|
| 727 |
+
"loss": 0.1177,
|
| 728 |
+
"step": 9300
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 20.22,
|
| 732 |
+
"learning_rate": 3.912282352941177e-05,
|
| 733 |
+
"loss": 0.1142,
|
| 734 |
+
"step": 9400
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 20.43,
|
| 738 |
+
"learning_rate": 3.884047058823529e-05,
|
| 739 |
+
"loss": 0.118,
|
| 740 |
+
"step": 9500
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 20.43,
|
| 744 |
+
"eval_loss": 0.4430650472640991,
|
| 745 |
+
"eval_runtime": 81.2185,
|
| 746 |
+
"eval_samples_per_second": 10.182,
|
| 747 |
+
"eval_steps_per_second": 1.28,
|
| 748 |
+
"eval_wer": 0.18818525519848772,
|
| 749 |
+
"step": 9500
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 20.65,
|
| 753 |
+
"learning_rate": 3.855811764705883e-05,
|
| 754 |
+
"loss": 0.1125,
|
| 755 |
+
"step": 9600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 20.86,
|
| 759 |
+
"learning_rate": 3.827576470588235e-05,
|
| 760 |
+
"loss": 0.1231,
|
| 761 |
+
"step": 9700
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 21.08,
|
| 765 |
+
"learning_rate": 3.799341176470588e-05,
|
| 766 |
+
"loss": 0.1133,
|
| 767 |
+
"step": 9800
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 21.29,
|
| 771 |
+
"learning_rate": 3.771105882352941e-05,
|
| 772 |
+
"loss": 0.1152,
|
| 773 |
+
"step": 9900
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 21.51,
|
| 777 |
+
"learning_rate": 3.7428705882352944e-05,
|
| 778 |
+
"loss": 0.1123,
|
| 779 |
+
"step": 10000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 21.51,
|
| 783 |
+
"eval_loss": 0.4753414988517761,
|
| 784 |
+
"eval_runtime": 81.7131,
|
| 785 |
+
"eval_samples_per_second": 10.121,
|
| 786 |
+
"eval_steps_per_second": 1.273,
|
| 787 |
+
"eval_wer": 0.18204158790170133,
|
| 788 |
+
"step": 10000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 21.72,
|
| 792 |
+
"learning_rate": 3.7146352941176474e-05,
|
| 793 |
+
"loss": 0.1186,
|
| 794 |
+
"step": 10100
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 21.94,
|
| 798 |
+
"learning_rate": 3.6864e-05,
|
| 799 |
+
"loss": 0.1113,
|
| 800 |
+
"step": 10200
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 22.15,
|
| 804 |
+
"learning_rate": 3.658164705882353e-05,
|
| 805 |
+
"loss": 0.1243,
|
| 806 |
+
"step": 10300
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 22.37,
|
| 810 |
+
"learning_rate": 3.629929411764706e-05,
|
| 811 |
+
"loss": 0.1056,
|
| 812 |
+
"step": 10400
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 22.58,
|
| 816 |
+
"learning_rate": 3.601694117647059e-05,
|
| 817 |
+
"loss": 0.1037,
|
| 818 |
+
"step": 10500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 22.58,
|
| 822 |
+
"eval_loss": 0.40868785977363586,
|
| 823 |
+
"eval_runtime": 80.8904,
|
| 824 |
+
"eval_samples_per_second": 10.224,
|
| 825 |
+
"eval_steps_per_second": 1.286,
|
| 826 |
+
"eval_wer": 0.1833648393194707,
|
| 827 |
+
"step": 10500
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 22.8,
|
| 831 |
+
"learning_rate": 3.573458823529411e-05,
|
| 832 |
+
"loss": 0.1055,
|
| 833 |
+
"step": 10600
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 23.01,
|
| 837 |
+
"learning_rate": 3.545223529411765e-05,
|
| 838 |
+
"loss": 0.0977,
|
| 839 |
+
"step": 10700
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 23.23,
|
| 843 |
+
"learning_rate": 3.5169882352941174e-05,
|
| 844 |
+
"loss": 0.1076,
|
| 845 |
+
"step": 10800
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 23.44,
|
| 849 |
+
"learning_rate": 3.488752941176471e-05,
|
| 850 |
+
"loss": 0.1028,
|
| 851 |
+
"step": 10900
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 23.66,
|
| 855 |
+
"learning_rate": 3.4605176470588235e-05,
|
| 856 |
+
"loss": 0.1066,
|
| 857 |
+
"step": 11000
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 23.66,
|
| 861 |
+
"eval_loss": 0.415127694606781,
|
| 862 |
+
"eval_runtime": 81.0404,
|
| 863 |
+
"eval_samples_per_second": 10.205,
|
| 864 |
+
"eval_steps_per_second": 1.283,
|
| 865 |
+
"eval_wer": 0.18449905482041587,
|
| 866 |
+
"step": 11000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 23.87,
|
| 870 |
+
"learning_rate": 3.4322823529411766e-05,
|
| 871 |
+
"loss": 0.1091,
|
| 872 |
+
"step": 11100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 24.09,
|
| 876 |
+
"learning_rate": 3.4040470588235296e-05,
|
| 877 |
+
"loss": 0.089,
|
| 878 |
+
"step": 11200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 24.3,
|
| 882 |
+
"learning_rate": 3.375811764705882e-05,
|
| 883 |
+
"loss": 0.1014,
|
| 884 |
+
"step": 11300
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 24.52,
|
| 888 |
+
"learning_rate": 3.347576470588236e-05,
|
| 889 |
+
"loss": 0.0864,
|
| 890 |
+
"step": 11400
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 24.73,
|
| 894 |
+
"learning_rate": 3.319341176470588e-05,
|
| 895 |
+
"loss": 0.0977,
|
| 896 |
+
"step": 11500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 24.73,
|
| 900 |
+
"eval_loss": 0.43674391508102417,
|
| 901 |
+
"eval_runtime": 81.0756,
|
| 902 |
+
"eval_samples_per_second": 10.2,
|
| 903 |
+
"eval_steps_per_second": 1.283,
|
| 904 |
+
"eval_wer": 0.1782608695652174,
|
| 905 |
+
"step": 11500
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 24.95,
|
| 909 |
+
"learning_rate": 3.291105882352941e-05,
|
| 910 |
+
"loss": 0.0989,
|
| 911 |
+
"step": 11600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 25.16,
|
| 915 |
+
"learning_rate": 3.262870588235294e-05,
|
| 916 |
+
"loss": 0.0977,
|
| 917 |
+
"step": 11700
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 25.38,
|
| 921 |
+
"learning_rate": 3.234635294117647e-05,
|
| 922 |
+
"loss": 0.0869,
|
| 923 |
+
"step": 11800
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 25.59,
|
| 927 |
+
"learning_rate": 3.2064e-05,
|
| 928 |
+
"loss": 0.0851,
|
| 929 |
+
"step": 11900
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 25.81,
|
| 933 |
+
"learning_rate": 3.1781647058823534e-05,
|
| 934 |
+
"loss": 0.0968,
|
| 935 |
+
"step": 12000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 25.81,
|
| 939 |
+
"eval_loss": 0.4236660897731781,
|
| 940 |
+
"eval_runtime": 80.9504,
|
| 941 |
+
"eval_samples_per_second": 10.216,
|
| 942 |
+
"eval_steps_per_second": 1.285,
|
| 943 |
+
"eval_wer": 0.17561436672967864,
|
| 944 |
+
"step": 12000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 26.02,
|
| 948 |
+
"learning_rate": 3.149929411764706e-05,
|
| 949 |
+
"loss": 0.0866,
|
| 950 |
+
"step": 12100
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 26.24,
|
| 954 |
+
"learning_rate": 3.121694117647059e-05,
|
| 955 |
+
"loss": 0.0774,
|
| 956 |
+
"step": 12200
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 26.45,
|
| 960 |
+
"learning_rate": 3.093458823529412e-05,
|
| 961 |
+
"loss": 0.0974,
|
| 962 |
+
"step": 12300
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 26.67,
|
| 966 |
+
"learning_rate": 3.065223529411764e-05,
|
| 967 |
+
"loss": 0.0807,
|
| 968 |
+
"step": 12400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 26.88,
|
| 972 |
+
"learning_rate": 3.0372705882352942e-05,
|
| 973 |
+
"loss": 0.0835,
|
| 974 |
+
"step": 12500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 26.88,
|
| 978 |
+
"eval_loss": 0.4728855490684509,
|
| 979 |
+
"eval_runtime": 81.9776,
|
| 980 |
+
"eval_samples_per_second": 10.088,
|
| 981 |
+
"eval_steps_per_second": 1.269,
|
| 982 |
+
"eval_wer": 0.17807183364839319,
|
| 983 |
+
"step": 12500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 27.1,
|
| 987 |
+
"learning_rate": 3.009035294117647e-05,
|
| 988 |
+
"loss": 0.0888,
|
| 989 |
+
"step": 12600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 27.31,
|
| 993 |
+
"learning_rate": 2.9808000000000003e-05,
|
| 994 |
+
"loss": 0.0845,
|
| 995 |
+
"step": 12700
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 27.53,
|
| 999 |
+
"learning_rate": 2.952564705882353e-05,
|
| 1000 |
+
"loss": 0.0888,
|
| 1001 |
+
"step": 12800
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 27.74,
|
| 1005 |
+
"learning_rate": 2.9243294117647058e-05,
|
| 1006 |
+
"loss": 0.0816,
|
| 1007 |
+
"step": 12900
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 27.96,
|
| 1011 |
+
"learning_rate": 2.8960941176470588e-05,
|
| 1012 |
+
"loss": 0.0919,
|
| 1013 |
+
"step": 13000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 27.96,
|
| 1017 |
+
"eval_loss": 0.4152912497520447,
|
| 1018 |
+
"eval_runtime": 81.9103,
|
| 1019 |
+
"eval_samples_per_second": 10.096,
|
| 1020 |
+
"eval_steps_per_second": 1.27,
|
| 1021 |
+
"eval_wer": 0.17013232514177692,
|
| 1022 |
+
"step": 13000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 28.17,
|
| 1026 |
+
"learning_rate": 2.867858823529412e-05,
|
| 1027 |
+
"loss": 0.0886,
|
| 1028 |
+
"step": 13100
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 28.39,
|
| 1032 |
+
"learning_rate": 2.8396235294117646e-05,
|
| 1033 |
+
"loss": 0.0809,
|
| 1034 |
+
"step": 13200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 28.6,
|
| 1038 |
+
"learning_rate": 2.8113882352941176e-05,
|
| 1039 |
+
"loss": 0.0786,
|
| 1040 |
+
"step": 13300
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 28.82,
|
| 1044 |
+
"learning_rate": 2.7831529411764707e-05,
|
| 1045 |
+
"loss": 0.0766,
|
| 1046 |
+
"step": 13400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 29.03,
|
| 1050 |
+
"learning_rate": 2.7552e-05,
|
| 1051 |
+
"loss": 0.0677,
|
| 1052 |
+
"step": 13500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 29.03,
|
| 1056 |
+
"eval_loss": 0.431657612323761,
|
| 1057 |
+
"eval_runtime": 81.4905,
|
| 1058 |
+
"eval_samples_per_second": 10.148,
|
| 1059 |
+
"eval_steps_per_second": 1.276,
|
| 1060 |
+
"eval_wer": 0.16928166351606805,
|
| 1061 |
+
"step": 13500
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 29.25,
|
| 1065 |
+
"learning_rate": 2.726964705882353e-05,
|
| 1066 |
+
"loss": 0.0757,
|
| 1067 |
+
"step": 13600
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 29.46,
|
| 1071 |
+
"learning_rate": 2.698729411764706e-05,
|
| 1072 |
+
"loss": 0.0719,
|
| 1073 |
+
"step": 13700
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 29.68,
|
| 1077 |
+
"learning_rate": 2.670494117647059e-05,
|
| 1078 |
+
"loss": 0.0772,
|
| 1079 |
+
"step": 13800
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 29.89,
|
| 1083 |
+
"learning_rate": 2.642258823529412e-05,
|
| 1084 |
+
"loss": 0.0762,
|
| 1085 |
+
"step": 13900
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 30.11,
|
| 1089 |
+
"learning_rate": 2.614023529411765e-05,
|
| 1090 |
+
"loss": 0.0726,
|
| 1091 |
+
"step": 14000
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 30.11,
|
| 1095 |
+
"eval_loss": 0.43801796436309814,
|
| 1096 |
+
"eval_runtime": 83.6487,
|
| 1097 |
+
"eval_samples_per_second": 9.887,
|
| 1098 |
+
"eval_steps_per_second": 1.243,
|
| 1099 |
+
"eval_wer": 0.17362948960302457,
|
| 1100 |
+
"step": 14000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 30.32,
|
| 1104 |
+
"learning_rate": 2.5857882352941176e-05,
|
| 1105 |
+
"loss": 0.0654,
|
| 1106 |
+
"step": 14100
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 30.54,
|
| 1110 |
+
"learning_rate": 2.5575529411764707e-05,
|
| 1111 |
+
"loss": 0.0691,
|
| 1112 |
+
"step": 14200
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 30.75,
|
| 1116 |
+
"learning_rate": 2.5293176470588234e-05,
|
| 1117 |
+
"loss": 0.0677,
|
| 1118 |
+
"step": 14300
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 30.97,
|
| 1122 |
+
"learning_rate": 2.5010823529411764e-05,
|
| 1123 |
+
"loss": 0.0668,
|
| 1124 |
+
"step": 14400
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 31.18,
|
| 1128 |
+
"learning_rate": 2.4728470588235295e-05,
|
| 1129 |
+
"loss": 0.066,
|
| 1130 |
+
"step": 14500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 31.18,
|
| 1134 |
+
"eval_loss": 0.4384245276451111,
|
| 1135 |
+
"eval_runtime": 81.1759,
|
| 1136 |
+
"eval_samples_per_second": 10.188,
|
| 1137 |
+
"eval_steps_per_second": 1.281,
|
| 1138 |
+
"eval_wer": 0.16814744801512288,
|
| 1139 |
+
"step": 14500
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 31.4,
|
| 1143 |
+
"learning_rate": 2.4446117647058825e-05,
|
| 1144 |
+
"loss": 0.0663,
|
| 1145 |
+
"step": 14600
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 31.61,
|
| 1149 |
+
"learning_rate": 2.4163764705882352e-05,
|
| 1150 |
+
"loss": 0.0688,
|
| 1151 |
+
"step": 14700
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 31.83,
|
| 1155 |
+
"learning_rate": 2.3881411764705883e-05,
|
| 1156 |
+
"loss": 0.0677,
|
| 1157 |
+
"step": 14800
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 32.04,
|
| 1161 |
+
"learning_rate": 2.3599058823529414e-05,
|
| 1162 |
+
"loss": 0.0718,
|
| 1163 |
+
"step": 14900
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 32.26,
|
| 1167 |
+
"learning_rate": 2.3316705882352944e-05,
|
| 1168 |
+
"loss": 0.0713,
|
| 1169 |
+
"step": 15000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 32.26,
|
| 1173 |
+
"eval_loss": 0.4215048551559448,
|
| 1174 |
+
"eval_runtime": 80.5465,
|
| 1175 |
+
"eval_samples_per_second": 10.267,
|
| 1176 |
+
"eval_steps_per_second": 1.291,
|
| 1177 |
+
"eval_wer": 0.16294896030245748,
|
| 1178 |
+
"step": 15000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 32.47,
|
| 1182 |
+
"learning_rate": 2.303435294117647e-05,
|
| 1183 |
+
"loss": 0.065,
|
| 1184 |
+
"step": 15100
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 32.69,
|
| 1188 |
+
"learning_rate": 2.2752e-05,
|
| 1189 |
+
"loss": 0.0631,
|
| 1190 |
+
"step": 15200
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 32.9,
|
| 1194 |
+
"learning_rate": 2.246964705882353e-05,
|
| 1195 |
+
"loss": 0.0616,
|
| 1196 |
+
"step": 15300
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 33.12,
|
| 1200 |
+
"learning_rate": 2.218729411764706e-05,
|
| 1201 |
+
"loss": 0.0812,
|
| 1202 |
+
"step": 15400
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 33.33,
|
| 1206 |
+
"learning_rate": 2.190494117647059e-05,
|
| 1207 |
+
"loss": 0.0605,
|
| 1208 |
+
"step": 15500
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 33.33,
|
| 1212 |
+
"eval_loss": 0.457427054643631,
|
| 1213 |
+
"eval_runtime": 80.6078,
|
| 1214 |
+
"eval_samples_per_second": 10.26,
|
| 1215 |
+
"eval_steps_per_second": 1.29,
|
| 1216 |
+
"eval_wer": 0.1713610586011342,
|
| 1217 |
+
"step": 15500
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 33.55,
|
| 1221 |
+
"learning_rate": 2.1622588235294117e-05,
|
| 1222 |
+
"loss": 0.0566,
|
| 1223 |
+
"step": 15600
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 33.76,
|
| 1227 |
+
"learning_rate": 2.1340235294117648e-05,
|
| 1228 |
+
"loss": 0.064,
|
| 1229 |
+
"step": 15700
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 33.98,
|
| 1233 |
+
"learning_rate": 2.1057882352941178e-05,
|
| 1234 |
+
"loss": 0.0591,
|
| 1235 |
+
"step": 15800
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 34.19,
|
| 1239 |
+
"learning_rate": 2.077552941176471e-05,
|
| 1240 |
+
"loss": 0.0581,
|
| 1241 |
+
"step": 15900
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 34.41,
|
| 1245 |
+
"learning_rate": 2.0493176470588236e-05,
|
| 1246 |
+
"loss": 0.0632,
|
| 1247 |
+
"step": 16000
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 34.41,
|
| 1251 |
+
"eval_loss": 0.43431583046913147,
|
| 1252 |
+
"eval_runtime": 81.6162,
|
| 1253 |
+
"eval_samples_per_second": 10.133,
|
| 1254 |
+
"eval_steps_per_second": 1.274,
|
| 1255 |
+
"eval_wer": 0.16417769376181474,
|
| 1256 |
+
"step": 16000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 34.62,
|
| 1260 |
+
"learning_rate": 2.0210823529411763e-05,
|
| 1261 |
+
"loss": 0.0578,
|
| 1262 |
+
"step": 16100
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 34.84,
|
| 1266 |
+
"learning_rate": 1.9928470588235293e-05,
|
| 1267 |
+
"loss": 0.0616,
|
| 1268 |
+
"step": 16200
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 35.05,
|
| 1272 |
+
"learning_rate": 1.9648941176470586e-05,
|
| 1273 |
+
"loss": 0.0516,
|
| 1274 |
+
"step": 16300
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 35.27,
|
| 1278 |
+
"learning_rate": 1.9366588235294117e-05,
|
| 1279 |
+
"loss": 0.0533,
|
| 1280 |
+
"step": 16400
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 35.48,
|
| 1284 |
+
"learning_rate": 1.9084235294117647e-05,
|
| 1285 |
+
"loss": 0.0567,
|
| 1286 |
+
"step": 16500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 35.48,
|
| 1290 |
+
"eval_loss": 0.42308300733566284,
|
| 1291 |
+
"eval_runtime": 82.0264,
|
| 1292 |
+
"eval_samples_per_second": 10.082,
|
| 1293 |
+
"eval_steps_per_second": 1.268,
|
| 1294 |
+
"eval_wer": 0.16011342155009453,
|
| 1295 |
+
"step": 16500
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 35.7,
|
| 1299 |
+
"learning_rate": 1.8801882352941178e-05,
|
| 1300 |
+
"loss": 0.0562,
|
| 1301 |
+
"step": 16600
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 35.91,
|
| 1305 |
+
"learning_rate": 1.8519529411764705e-05,
|
| 1306 |
+
"loss": 0.0554,
|
| 1307 |
+
"step": 16700
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 36.13,
|
| 1311 |
+
"learning_rate": 1.8237176470588236e-05,
|
| 1312 |
+
"loss": 0.0564,
|
| 1313 |
+
"step": 16800
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 36.34,
|
| 1317 |
+
"learning_rate": 1.7954823529411766e-05,
|
| 1318 |
+
"loss": 0.0617,
|
| 1319 |
+
"step": 16900
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 36.56,
|
| 1323 |
+
"learning_rate": 1.7672470588235297e-05,
|
| 1324 |
+
"loss": 0.0556,
|
| 1325 |
+
"step": 17000
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 36.56,
|
| 1329 |
+
"eval_loss": 0.4404306411743164,
|
| 1330 |
+
"eval_runtime": 81.4874,
|
| 1331 |
+
"eval_samples_per_second": 10.149,
|
| 1332 |
+
"eval_steps_per_second": 1.276,
|
| 1333 |
+
"eval_wer": 0.1667296786389414,
|
| 1334 |
+
"step": 17000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 36.77,
|
| 1338 |
+
"learning_rate": 1.7390117647058824e-05,
|
| 1339 |
+
"loss": 0.0524,
|
| 1340 |
+
"step": 17100
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 36.99,
|
| 1344 |
+
"learning_rate": 1.7107764705882354e-05,
|
| 1345 |
+
"loss": 0.0602,
|
| 1346 |
+
"step": 17200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 37.2,
|
| 1350 |
+
"learning_rate": 1.682541176470588e-05,
|
| 1351 |
+
"loss": 0.0472,
|
| 1352 |
+
"step": 17300
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 37.42,
|
| 1356 |
+
"learning_rate": 1.6543058823529412e-05,
|
| 1357 |
+
"loss": 0.047,
|
| 1358 |
+
"step": 17400
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 37.63,
|
| 1362 |
+
"learning_rate": 1.6260705882352943e-05,
|
| 1363 |
+
"loss": 0.0426,
|
| 1364 |
+
"step": 17500
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 37.63,
|
| 1368 |
+
"eval_loss": 0.4458593428134918,
|
| 1369 |
+
"eval_runtime": 80.8169,
|
| 1370 |
+
"eval_samples_per_second": 10.233,
|
| 1371 |
+
"eval_steps_per_second": 1.287,
|
| 1372 |
+
"eval_wer": 0.16247637051039698,
|
| 1373 |
+
"step": 17500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 37.85,
|
| 1377 |
+
"learning_rate": 1.597835294117647e-05,
|
| 1378 |
+
"loss": 0.0513,
|
| 1379 |
+
"step": 17600
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 38.06,
|
| 1383 |
+
"learning_rate": 1.5696e-05,
|
| 1384 |
+
"loss": 0.056,
|
| 1385 |
+
"step": 17700
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 38.28,
|
| 1389 |
+
"learning_rate": 1.541364705882353e-05,
|
| 1390 |
+
"loss": 0.0498,
|
| 1391 |
+
"step": 17800
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 38.49,
|
| 1395 |
+
"learning_rate": 1.513129411764706e-05,
|
| 1396 |
+
"loss": 0.0473,
|
| 1397 |
+
"step": 17900
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 38.71,
|
| 1401 |
+
"learning_rate": 1.4848941176470588e-05,
|
| 1402 |
+
"loss": 0.0445,
|
| 1403 |
+
"step": 18000
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 38.71,
|
| 1407 |
+
"eval_loss": 0.4483908712863922,
|
| 1408 |
+
"eval_runtime": 82.388,
|
| 1409 |
+
"eval_samples_per_second": 10.038,
|
| 1410 |
+
"eval_steps_per_second": 1.262,
|
| 1411 |
+
"eval_wer": 0.16285444234404536,
|
| 1412 |
+
"step": 18000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 38.92,
|
| 1416 |
+
"learning_rate": 1.4566588235294117e-05,
|
| 1417 |
+
"loss": 0.0487,
|
| 1418 |
+
"step": 18100
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 39.14,
|
| 1422 |
+
"learning_rate": 1.4284235294117648e-05,
|
| 1423 |
+
"loss": 0.0491,
|
| 1424 |
+
"step": 18200
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 39.35,
|
| 1428 |
+
"learning_rate": 1.4001882352941177e-05,
|
| 1429 |
+
"loss": 0.044,
|
| 1430 |
+
"step": 18300
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 39.57,
|
| 1434 |
+
"learning_rate": 1.3722352941176471e-05,
|
| 1435 |
+
"loss": 0.0467,
|
| 1436 |
+
"step": 18400
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 39.78,
|
| 1440 |
+
"learning_rate": 1.344e-05,
|
| 1441 |
+
"loss": 0.0463,
|
| 1442 |
+
"step": 18500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 39.78,
|
| 1446 |
+
"eval_loss": 0.45076683163642883,
|
| 1447 |
+
"eval_runtime": 81.2197,
|
| 1448 |
+
"eval_samples_per_second": 10.182,
|
| 1449 |
+
"eval_steps_per_second": 1.28,
|
| 1450 |
+
"eval_wer": 0.15964083175803404,
|
| 1451 |
+
"step": 18500
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 40.0,
|
| 1455 |
+
"learning_rate": 1.315764705882353e-05,
|
| 1456 |
+
"loss": 0.0464,
|
| 1457 |
+
"step": 18600
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 40.22,
|
| 1461 |
+
"learning_rate": 1.2875294117647058e-05,
|
| 1462 |
+
"loss": 0.0395,
|
| 1463 |
+
"step": 18700
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 40.43,
|
| 1467 |
+
"learning_rate": 1.2592941176470588e-05,
|
| 1468 |
+
"loss": 0.0421,
|
| 1469 |
+
"step": 18800
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 40.65,
|
| 1473 |
+
"learning_rate": 1.2310588235294119e-05,
|
| 1474 |
+
"loss": 0.0463,
|
| 1475 |
+
"step": 18900
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 40.86,
|
| 1479 |
+
"learning_rate": 1.2028235294117648e-05,
|
| 1480 |
+
"loss": 0.0448,
|
| 1481 |
+
"step": 19000
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 40.86,
|
| 1485 |
+
"eval_loss": 0.4395386278629303,
|
| 1486 |
+
"eval_runtime": 82.035,
|
| 1487 |
+
"eval_samples_per_second": 10.081,
|
| 1488 |
+
"eval_steps_per_second": 1.268,
|
| 1489 |
+
"eval_wer": 0.1604914933837429,
|
| 1490 |
+
"step": 19000
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 41.08,
|
| 1494 |
+
"learning_rate": 1.1745882352941178e-05,
|
| 1495 |
+
"loss": 0.0483,
|
| 1496 |
+
"step": 19100
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 41.29,
|
| 1500 |
+
"learning_rate": 1.1463529411764705e-05,
|
| 1501 |
+
"loss": 0.0457,
|
| 1502 |
+
"step": 19200
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 41.51,
|
| 1506 |
+
"learning_rate": 1.1181176470588236e-05,
|
| 1507 |
+
"loss": 0.0363,
|
| 1508 |
+
"step": 19300
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 41.72,
|
| 1512 |
+
"learning_rate": 1.0898823529411765e-05,
|
| 1513 |
+
"loss": 0.0428,
|
| 1514 |
+
"step": 19400
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 41.94,
|
| 1518 |
+
"learning_rate": 1.0616470588235295e-05,
|
| 1519 |
+
"loss": 0.0434,
|
| 1520 |
+
"step": 19500
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 41.94,
|
| 1524 |
+
"eval_loss": 0.4489993453025818,
|
| 1525 |
+
"eval_runtime": 82.599,
|
| 1526 |
+
"eval_samples_per_second": 10.012,
|
| 1527 |
+
"eval_steps_per_second": 1.259,
|
| 1528 |
+
"eval_wer": 0.16068052930056712,
|
| 1529 |
+
"step": 19500
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 42.15,
|
| 1533 |
+
"learning_rate": 1.0334117647058824e-05,
|
| 1534 |
+
"loss": 0.0389,
|
| 1535 |
+
"step": 19600
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 42.37,
|
| 1539 |
+
"learning_rate": 1.0051764705882353e-05,
|
| 1540 |
+
"loss": 0.0482,
|
| 1541 |
+
"step": 19700
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 42.58,
|
| 1545 |
+
"learning_rate": 9.769411764705882e-06,
|
| 1546 |
+
"loss": 0.0467,
|
| 1547 |
+
"step": 19800
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 42.8,
|
| 1551 |
+
"learning_rate": 9.487058823529412e-06,
|
| 1552 |
+
"loss": 0.0379,
|
| 1553 |
+
"step": 19900
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 43.01,
|
| 1557 |
+
"learning_rate": 9.204705882352941e-06,
|
| 1558 |
+
"loss": 0.0347,
|
| 1559 |
+
"step": 20000
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 43.01,
|
| 1563 |
+
"eval_loss": 0.47717225551605225,
|
| 1564 |
+
"eval_runtime": 81.2937,
|
| 1565 |
+
"eval_samples_per_second": 10.173,
|
| 1566 |
+
"eval_steps_per_second": 1.279,
|
| 1567 |
+
"eval_wer": 0.15822306238185255,
|
| 1568 |
+
"step": 20000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 43.23,
|
| 1572 |
+
"learning_rate": 8.922352941176471e-06,
|
| 1573 |
+
"loss": 0.0327,
|
| 1574 |
+
"step": 20100
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 43.44,
|
| 1578 |
+
"learning_rate": 8.64e-06,
|
| 1579 |
+
"loss": 0.0407,
|
| 1580 |
+
"step": 20200
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 43.66,
|
| 1584 |
+
"learning_rate": 8.357647058823529e-06,
|
| 1585 |
+
"loss": 0.0368,
|
| 1586 |
+
"step": 20300
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 43.87,
|
| 1590 |
+
"learning_rate": 8.07529411764706e-06,
|
| 1591 |
+
"loss": 0.0365,
|
| 1592 |
+
"step": 20400
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 44.09,
|
| 1596 |
+
"learning_rate": 7.792941176470588e-06,
|
| 1597 |
+
"loss": 0.0332,
|
| 1598 |
+
"step": 20500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 44.09,
|
| 1602 |
+
"eval_loss": 0.4728728234767914,
|
| 1603 |
+
"eval_runtime": 81.4631,
|
| 1604 |
+
"eval_samples_per_second": 10.152,
|
| 1605 |
+
"eval_steps_per_second": 1.277,
|
| 1606 |
+
"eval_wer": 0.15822306238185255,
|
| 1607 |
+
"step": 20500
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 44.3,
|
| 1611 |
+
"learning_rate": 7.510588235294118e-06,
|
| 1612 |
+
"loss": 0.0353,
|
| 1613 |
+
"step": 20600
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 44.52,
|
| 1617 |
+
"learning_rate": 7.228235294117648e-06,
|
| 1618 |
+
"loss": 0.039,
|
| 1619 |
+
"step": 20700
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 44.73,
|
| 1623 |
+
"learning_rate": 6.945882352941177e-06,
|
| 1624 |
+
"loss": 0.0395,
|
| 1625 |
+
"step": 20800
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 44.95,
|
| 1629 |
+
"learning_rate": 6.663529411764706e-06,
|
| 1630 |
+
"loss": 0.0368,
|
| 1631 |
+
"step": 20900
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 45.16,
|
| 1635 |
+
"learning_rate": 6.381176470588236e-06,
|
| 1636 |
+
"loss": 0.037,
|
| 1637 |
+
"step": 21000
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 45.16,
|
| 1641 |
+
"eval_loss": 0.45591220259666443,
|
| 1642 |
+
"eval_runtime": 83.1729,
|
| 1643 |
+
"eval_samples_per_second": 9.943,
|
| 1644 |
+
"eval_steps_per_second": 1.25,
|
| 1645 |
+
"eval_wer": 0.15727788279773156,
|
| 1646 |
+
"step": 21000
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 45.38,
|
| 1650 |
+
"learning_rate": 6.098823529411765e-06,
|
| 1651 |
+
"loss": 0.0328,
|
| 1652 |
+
"step": 21100
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 45.59,
|
| 1656 |
+
"learning_rate": 5.8164705882352945e-06,
|
| 1657 |
+
"loss": 0.0541,
|
| 1658 |
+
"step": 21200
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 45.81,
|
| 1662 |
+
"learning_rate": 5.534117647058823e-06,
|
| 1663 |
+
"loss": 0.0413,
|
| 1664 |
+
"step": 21300
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 46.02,
|
| 1668 |
+
"learning_rate": 5.251764705882353e-06,
|
| 1669 |
+
"loss": 0.0403,
|
| 1670 |
+
"step": 21400
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 46.24,
|
| 1674 |
+
"learning_rate": 4.969411764705883e-06,
|
| 1675 |
+
"loss": 0.0328,
|
| 1676 |
+
"step": 21500
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 46.24,
|
| 1680 |
+
"eval_loss": 0.46641021966934204,
|
| 1681 |
+
"eval_runtime": 82.6563,
|
| 1682 |
+
"eval_samples_per_second": 10.005,
|
| 1683 |
+
"eval_steps_per_second": 1.258,
|
| 1684 |
+
"eval_wer": 0.1560491493383743,
|
| 1685 |
+
"step": 21500
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 46.45,
|
| 1689 |
+
"learning_rate": 4.6870588235294115e-06,
|
| 1690 |
+
"loss": 0.0334,
|
| 1691 |
+
"step": 21600
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 46.67,
|
| 1695 |
+
"learning_rate": 4.404705882352941e-06,
|
| 1696 |
+
"loss": 0.0334,
|
| 1697 |
+
"step": 21700
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 46.88,
|
| 1701 |
+
"learning_rate": 4.122352941176471e-06,
|
| 1702 |
+
"loss": 0.0324,
|
| 1703 |
+
"step": 21800
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 47.1,
|
| 1707 |
+
"learning_rate": 3.8400000000000005e-06,
|
| 1708 |
+
"loss": 0.0337,
|
| 1709 |
+
"step": 21900
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 47.31,
|
| 1713 |
+
"learning_rate": 3.5576470588235293e-06,
|
| 1714 |
+
"loss": 0.0366,
|
| 1715 |
+
"step": 22000
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 47.31,
|
| 1719 |
+
"eval_loss": 0.454330176115036,
|
| 1720 |
+
"eval_runtime": 81.2727,
|
| 1721 |
+
"eval_samples_per_second": 10.176,
|
| 1722 |
+
"eval_steps_per_second": 1.28,
|
| 1723 |
+
"eval_wer": 0.15434782608695652,
|
| 1724 |
+
"step": 22000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 47.53,
|
| 1728 |
+
"learning_rate": 3.275294117647059e-06,
|
| 1729 |
+
"loss": 0.0332,
|
| 1730 |
+
"step": 22100
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 47.74,
|
| 1734 |
+
"learning_rate": 2.9929411764705883e-06,
|
| 1735 |
+
"loss": 0.0295,
|
| 1736 |
+
"step": 22200
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 47.96,
|
| 1740 |
+
"learning_rate": 2.710588235294118e-06,
|
| 1741 |
+
"loss": 0.037,
|
| 1742 |
+
"step": 22300
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 48.17,
|
| 1746 |
+
"learning_rate": 2.428235294117647e-06,
|
| 1747 |
+
"loss": 0.0314,
|
| 1748 |
+
"step": 22400
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 48.39,
|
| 1752 |
+
"learning_rate": 2.1458823529411764e-06,
|
| 1753 |
+
"loss": 0.0377,
|
| 1754 |
+
"step": 22500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 48.39,
|
| 1758 |
+
"eval_loss": 0.45068359375,
|
| 1759 |
+
"eval_runtime": 80.8954,
|
| 1760 |
+
"eval_samples_per_second": 10.223,
|
| 1761 |
+
"eval_steps_per_second": 1.286,
|
| 1762 |
+
"eval_wer": 0.15595463137996218,
|
| 1763 |
+
"step": 22500
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 48.6,
|
| 1767 |
+
"learning_rate": 1.8635294117647059e-06,
|
| 1768 |
+
"loss": 0.0284,
|
| 1769 |
+
"step": 22600
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 48.82,
|
| 1773 |
+
"learning_rate": 1.5811764705882354e-06,
|
| 1774 |
+
"loss": 0.0326,
|
| 1775 |
+
"step": 22700
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 49.03,
|
| 1779 |
+
"learning_rate": 1.2988235294117648e-06,
|
| 1780 |
+
"loss": 0.034,
|
| 1781 |
+
"step": 22800
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 49.25,
|
| 1785 |
+
"learning_rate": 1.016470588235294e-06,
|
| 1786 |
+
"loss": 0.0319,
|
| 1787 |
+
"step": 22900
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 49.46,
|
| 1791 |
+
"learning_rate": 7.341176470588236e-07,
|
| 1792 |
+
"loss": 0.0331,
|
| 1793 |
+
"step": 23000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 49.46,
|
| 1797 |
+
"eval_loss": 0.4567229151725769,
|
| 1798 |
+
"eval_runtime": 80.8744,
|
| 1799 |
+
"eval_samples_per_second": 10.226,
|
| 1800 |
+
"eval_steps_per_second": 1.286,
|
| 1801 |
+
"eval_wer": 0.15330812854442344,
|
| 1802 |
+
"step": 23000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 49.68,
|
| 1806 |
+
"learning_rate": 4.51764705882353e-07,
|
| 1807 |
+
"loss": 0.0334,
|
| 1808 |
+
"step": 23100
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 49.89,
|
| 1812 |
+
"learning_rate": 1.6941176470588237e-07,
|
| 1813 |
+
"loss": 0.027,
|
| 1814 |
+
"step": 23200
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 50.0,
|
| 1818 |
+
"step": 23250,
|
| 1819 |
+
"total_flos": 2.4501256105279144e+20,
|
| 1820 |
+
"train_loss": 0.23119104407936014,
|
| 1821 |
+
"train_runtime": 64754.7357,
|
| 1822 |
+
"train_samples_per_second": 5.744,
|
| 1823 |
+
"train_steps_per_second": 0.359
|
| 1824 |
+
}
|
| 1825 |
+
],
|
| 1826 |
+
"max_steps": 23250,
|
| 1827 |
+
"num_train_epochs": 50,
|
| 1828 |
+
"total_flos": 2.4501256105279144e+20,
|
| 1829 |
+
"trial_name": null,
|
| 1830 |
+
"trial_params": null
|
| 1831 |
+
}
|