Commit
·
5f73e5b
1
Parent(s):
8d717e5
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.56,
|
| 3 |
+
"eval_cer": 1.0571606245483118,
|
| 4 |
+
"eval_loss": 0.13207247853279114,
|
| 5 |
+
"eval_runtime": 939.5389,
|
| 6 |
+
"eval_samples_per_second": 1.064,
|
| 7 |
+
"eval_steps_per_second": 0.133,
|
| 8 |
+
"eval_wer": 3.6508096148043854,
|
| 9 |
+
"train_loss": 0.1500811086446047,
|
| 10 |
+
"train_runtime": 62196.6088,
|
| 11 |
+
"train_samples_per_second": 10.29,
|
| 12 |
+
"train_steps_per_second": 0.322
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.56,
|
| 3 |
+
"eval_cer": 1.0571606245483118,
|
| 4 |
+
"eval_loss": 0.13207247853279114,
|
| 5 |
+
"eval_runtime": 939.5389,
|
| 6 |
+
"eval_samples_per_second": 1.064,
|
| 7 |
+
"eval_steps_per_second": 0.133,
|
| 8 |
+
"eval_wer": 3.6508096148043854
|
| 9 |
+
}
|
runs/Dec21_20-41-09_129-146-177-31/events.out.tfevents.1671718880.129-146-177-31.1981645.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7232161df56912998f012f412c21873683a0890428dea8a05b053e53ce5d6f98
|
| 3 |
+
size 405
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.56,
|
| 3 |
+
"train_loss": 0.1500811086446047,
|
| 4 |
+
"train_runtime": 62196.6088,
|
| 5 |
+
"train_samples_per_second": 10.29,
|
| 6 |
+
"train_steps_per_second": 0.322
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2025 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.6508096148043854,
|
| 3 |
+
"best_model_checkpoint": "data/model/whisper-large-es-cv11-2/checkpoint-5000",
|
| 4 |
+
"epoch": 2.56,
|
| 5 |
+
"global_step": 8000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 1.1499999999999999e-08,
|
| 13 |
+
"loss": 0.8552,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 2.35e-08,
|
| 19 |
+
"loss": 0.8366,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 3.6e-08,
|
| 25 |
+
"loss": 0.8538,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 4.85e-08,
|
| 31 |
+
"loss": 0.842,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.04,
|
| 36 |
+
"learning_rate": 6.049999999999999e-08,
|
| 37 |
+
"loss": 0.8284,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.05,
|
| 42 |
+
"learning_rate": 7.299999999999999e-08,
|
| 43 |
+
"loss": 0.8163,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06,
|
| 48 |
+
"learning_rate": 8.55e-08,
|
| 49 |
+
"loss": 0.7738,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.06,
|
| 54 |
+
"learning_rate": 9.8e-08,
|
| 55 |
+
"loss": 0.7009,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.07,
|
| 60 |
+
"learning_rate": 1.0999999999999999e-07,
|
| 61 |
+
"loss": 0.5658,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.08,
|
| 66 |
+
"learning_rate": 1.225e-07,
|
| 67 |
+
"loss": 0.4291,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.09,
|
| 72 |
+
"learning_rate": 1.35e-07,
|
| 73 |
+
"loss": 0.3306,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.1,
|
| 78 |
+
"learning_rate": 1.475e-07,
|
| 79 |
+
"loss": 0.3119,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.1,
|
| 84 |
+
"learning_rate": 1.6e-07,
|
| 85 |
+
"loss": 0.3076,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.11,
|
| 90 |
+
"learning_rate": 1.725e-07,
|
| 91 |
+
"loss": 0.301,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.12,
|
| 96 |
+
"learning_rate": 1.85e-07,
|
| 97 |
+
"loss": 0.2942,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.13,
|
| 102 |
+
"learning_rate": 1.975e-07,
|
| 103 |
+
"loss": 0.279,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.14,
|
| 108 |
+
"learning_rate": 2.0999999999999997e-07,
|
| 109 |
+
"loss": 0.2694,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.14,
|
| 114 |
+
"learning_rate": 2.225e-07,
|
| 115 |
+
"loss": 0.2416,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.15,
|
| 120 |
+
"learning_rate": 2.3499999999999997e-07,
|
| 121 |
+
"loss": 0.2491,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.16,
|
| 126 |
+
"learning_rate": 2.475e-07,
|
| 127 |
+
"loss": 0.2518,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17,
|
| 132 |
+
"learning_rate": 2.6e-07,
|
| 133 |
+
"loss": 0.2497,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.18,
|
| 138 |
+
"learning_rate": 2.725e-07,
|
| 139 |
+
"loss": 0.2362,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.18,
|
| 144 |
+
"learning_rate": 2.8499999999999997e-07,
|
| 145 |
+
"loss": 0.24,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.19,
|
| 150 |
+
"learning_rate": 2.9749999999999996e-07,
|
| 151 |
+
"loss": 0.2215,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.2,
|
| 156 |
+
"learning_rate": 3.1e-07,
|
| 157 |
+
"loss": 0.2104,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.21,
|
| 162 |
+
"learning_rate": 3.225e-07,
|
| 163 |
+
"loss": 0.2115,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.22,
|
| 168 |
+
"learning_rate": 3.35e-07,
|
| 169 |
+
"loss": 0.2064,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.22,
|
| 174 |
+
"learning_rate": 3.4749999999999996e-07,
|
| 175 |
+
"loss": 0.2026,
|
| 176 |
+
"step": 700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.23,
|
| 180 |
+
"learning_rate": 3.6e-07,
|
| 181 |
+
"loss": 0.2048,
|
| 182 |
+
"step": 725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.24,
|
| 186 |
+
"learning_rate": 3.725e-07,
|
| 187 |
+
"loss": 0.1946,
|
| 188 |
+
"step": 750
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.25,
|
| 192 |
+
"learning_rate": 3.8499999999999997e-07,
|
| 193 |
+
"loss": 0.1872,
|
| 194 |
+
"step": 775
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.26,
|
| 198 |
+
"learning_rate": 3.975e-07,
|
| 199 |
+
"loss": 0.1851,
|
| 200 |
+
"step": 800
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.26,
|
| 204 |
+
"learning_rate": 4.0999999999999994e-07,
|
| 205 |
+
"loss": 0.1714,
|
| 206 |
+
"step": 825
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.27,
|
| 210 |
+
"learning_rate": 4.225e-07,
|
| 211 |
+
"loss": 0.1869,
|
| 212 |
+
"step": 850
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.28,
|
| 216 |
+
"learning_rate": 4.3499999999999996e-07,
|
| 217 |
+
"loss": 0.1847,
|
| 218 |
+
"step": 875
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.29,
|
| 222 |
+
"learning_rate": 4.475e-07,
|
| 223 |
+
"loss": 0.1931,
|
| 224 |
+
"step": 900
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.3,
|
| 228 |
+
"learning_rate": 4.6e-07,
|
| 229 |
+
"loss": 0.1776,
|
| 230 |
+
"step": 925
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.3,
|
| 234 |
+
"learning_rate": 4.725e-07,
|
| 235 |
+
"loss": 0.1803,
|
| 236 |
+
"step": 950
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.31,
|
| 240 |
+
"learning_rate": 4.85e-07,
|
| 241 |
+
"loss": 0.1721,
|
| 242 |
+
"step": 975
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.32,
|
| 246 |
+
"learning_rate": 4.975e-07,
|
| 247 |
+
"loss": 0.1837,
|
| 248 |
+
"step": 1000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.32,
|
| 252 |
+
"eval_cer": 1.2487604833694683,
|
| 253 |
+
"eval_loss": 0.16694751381874084,
|
| 254 |
+
"eval_runtime": 939.4915,
|
| 255 |
+
"eval_samples_per_second": 1.064,
|
| 256 |
+
"eval_steps_per_second": 0.133,
|
| 257 |
+
"eval_wer": 4.244191893794629,
|
| 258 |
+
"step": 1000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.33,
|
| 262 |
+
"learning_rate": 5.1e-07,
|
| 263 |
+
"loss": 0.1685,
|
| 264 |
+
"step": 1025
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.34,
|
| 268 |
+
"learning_rate": 5.225e-07,
|
| 269 |
+
"loss": 0.1667,
|
| 270 |
+
"step": 1050
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.34,
|
| 274 |
+
"learning_rate": 5.35e-07,
|
| 275 |
+
"loss": 0.1731,
|
| 276 |
+
"step": 1075
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.35,
|
| 280 |
+
"learning_rate": 5.474999999999999e-07,
|
| 281 |
+
"loss": 0.1667,
|
| 282 |
+
"step": 1100
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.36,
|
| 286 |
+
"learning_rate": 5.6e-07,
|
| 287 |
+
"loss": 0.1684,
|
| 288 |
+
"step": 1125
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.37,
|
| 292 |
+
"learning_rate": 5.725e-07,
|
| 293 |
+
"loss": 0.1713,
|
| 294 |
+
"step": 1150
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.38,
|
| 298 |
+
"learning_rate": 5.849999999999999e-07,
|
| 299 |
+
"loss": 0.1625,
|
| 300 |
+
"step": 1175
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.38,
|
| 304 |
+
"learning_rate": 5.975e-07,
|
| 305 |
+
"loss": 0.1578,
|
| 306 |
+
"step": 1200
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.39,
|
| 310 |
+
"learning_rate": 6.1e-07,
|
| 311 |
+
"loss": 0.1621,
|
| 312 |
+
"step": 1225
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.4,
|
| 316 |
+
"learning_rate": 6.225000000000001e-07,
|
| 317 |
+
"loss": 0.1652,
|
| 318 |
+
"step": 1250
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.41,
|
| 322 |
+
"learning_rate": 6.35e-07,
|
| 323 |
+
"loss": 0.159,
|
| 324 |
+
"step": 1275
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.42,
|
| 328 |
+
"learning_rate": 6.474999999999999e-07,
|
| 329 |
+
"loss": 0.1722,
|
| 330 |
+
"step": 1300
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.42,
|
| 334 |
+
"learning_rate": 6.6e-07,
|
| 335 |
+
"loss": 0.16,
|
| 336 |
+
"step": 1325
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.43,
|
| 340 |
+
"learning_rate": 6.724999999999999e-07,
|
| 341 |
+
"loss": 0.1582,
|
| 342 |
+
"step": 1350
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.44,
|
| 346 |
+
"learning_rate": 6.85e-07,
|
| 347 |
+
"loss": 0.1636,
|
| 348 |
+
"step": 1375
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.45,
|
| 352 |
+
"learning_rate": 6.975e-07,
|
| 353 |
+
"loss": 0.1555,
|
| 354 |
+
"step": 1400
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.46,
|
| 358 |
+
"learning_rate": 7.1e-07,
|
| 359 |
+
"loss": 0.1576,
|
| 360 |
+
"step": 1425
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.46,
|
| 364 |
+
"learning_rate": 7.225e-07,
|
| 365 |
+
"loss": 0.1605,
|
| 366 |
+
"step": 1450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.47,
|
| 370 |
+
"learning_rate": 7.345e-07,
|
| 371 |
+
"loss": 0.1614,
|
| 372 |
+
"step": 1475
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.48,
|
| 376 |
+
"learning_rate": 7.47e-07,
|
| 377 |
+
"loss": 0.1412,
|
| 378 |
+
"step": 1500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.49,
|
| 382 |
+
"learning_rate": 7.594999999999999e-07,
|
| 383 |
+
"loss": 0.1451,
|
| 384 |
+
"step": 1525
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5,
|
| 388 |
+
"learning_rate": 7.72e-07,
|
| 389 |
+
"loss": 0.1549,
|
| 390 |
+
"step": 1550
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 0.5,
|
| 394 |
+
"learning_rate": 7.845e-07,
|
| 395 |
+
"loss": 0.1528,
|
| 396 |
+
"step": 1575
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.51,
|
| 400 |
+
"learning_rate": 7.970000000000001e-07,
|
| 401 |
+
"loss": 0.1535,
|
| 402 |
+
"step": 1600
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.52,
|
| 406 |
+
"learning_rate": 8.094999999999999e-07,
|
| 407 |
+
"loss": 0.1578,
|
| 408 |
+
"step": 1625
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.53,
|
| 412 |
+
"learning_rate": 8.219999999999999e-07,
|
| 413 |
+
"loss": 0.1526,
|
| 414 |
+
"step": 1650
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.54,
|
| 418 |
+
"learning_rate": 8.345e-07,
|
| 419 |
+
"loss": 0.1493,
|
| 420 |
+
"step": 1675
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.54,
|
| 424 |
+
"learning_rate": 8.469999999999999e-07,
|
| 425 |
+
"loss": 0.1471,
|
| 426 |
+
"step": 1700
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.55,
|
| 430 |
+
"learning_rate": 8.595e-07,
|
| 431 |
+
"loss": 0.1359,
|
| 432 |
+
"step": 1725
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.56,
|
| 436 |
+
"learning_rate": 8.72e-07,
|
| 437 |
+
"loss": 0.1494,
|
| 438 |
+
"step": 1750
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.57,
|
| 442 |
+
"learning_rate": 8.845e-07,
|
| 443 |
+
"loss": 0.1521,
|
| 444 |
+
"step": 1775
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.58,
|
| 448 |
+
"learning_rate": 8.969999999999999e-07,
|
| 449 |
+
"loss": 0.158,
|
| 450 |
+
"step": 1800
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.58,
|
| 454 |
+
"learning_rate": 9.094999999999999e-07,
|
| 455 |
+
"loss": 0.1591,
|
| 456 |
+
"step": 1825
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.59,
|
| 460 |
+
"learning_rate": 9.22e-07,
|
| 461 |
+
"loss": 0.1408,
|
| 462 |
+
"step": 1850
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.6,
|
| 466 |
+
"learning_rate": 9.344999999999999e-07,
|
| 467 |
+
"loss": 0.1719,
|
| 468 |
+
"step": 1875
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.61,
|
| 472 |
+
"learning_rate": 9.469999999999999e-07,
|
| 473 |
+
"loss": 0.1466,
|
| 474 |
+
"step": 1900
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.62,
|
| 478 |
+
"learning_rate": 9.594999999999999e-07,
|
| 479 |
+
"loss": 0.1502,
|
| 480 |
+
"step": 1925
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.62,
|
| 484 |
+
"learning_rate": 9.72e-07,
|
| 485 |
+
"loss": 0.1487,
|
| 486 |
+
"step": 1950
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.63,
|
| 490 |
+
"learning_rate": 9.845e-07,
|
| 491 |
+
"loss": 0.1463,
|
| 492 |
+
"step": 1975
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.64,
|
| 496 |
+
"learning_rate": 9.97e-07,
|
| 497 |
+
"loss": 0.1343,
|
| 498 |
+
"step": 2000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.64,
|
| 502 |
+
"eval_cer": 1.208423670986067,
|
| 503 |
+
"eval_loss": 0.14436140656471252,
|
| 504 |
+
"eval_runtime": 941.5744,
|
| 505 |
+
"eval_samples_per_second": 1.062,
|
| 506 |
+
"eval_steps_per_second": 0.133,
|
| 507 |
+
"eval_wer": 4.083274665593885,
|
| 508 |
+
"step": 2000
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.65,
|
| 512 |
+
"learning_rate": 9.989444444444443e-07,
|
| 513 |
+
"loss": 0.1575,
|
| 514 |
+
"step": 2025
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.66,
|
| 518 |
+
"learning_rate": 9.975555555555556e-07,
|
| 519 |
+
"loss": 0.1468,
|
| 520 |
+
"step": 2050
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.66,
|
| 524 |
+
"learning_rate": 9.961666666666667e-07,
|
| 525 |
+
"loss": 0.1389,
|
| 526 |
+
"step": 2075
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.67,
|
| 530 |
+
"learning_rate": 9.947777777777778e-07,
|
| 531 |
+
"loss": 0.1412,
|
| 532 |
+
"step": 2100
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 0.68,
|
| 536 |
+
"learning_rate": 9.933888888888889e-07,
|
| 537 |
+
"loss": 0.1589,
|
| 538 |
+
"step": 2125
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.69,
|
| 542 |
+
"learning_rate": 9.92e-07,
|
| 543 |
+
"loss": 0.141,
|
| 544 |
+
"step": 2150
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 0.7,
|
| 548 |
+
"learning_rate": 9.90611111111111e-07,
|
| 549 |
+
"loss": 0.1374,
|
| 550 |
+
"step": 2175
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.7,
|
| 554 |
+
"learning_rate": 9.892222222222222e-07,
|
| 555 |
+
"loss": 0.1444,
|
| 556 |
+
"step": 2200
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.71,
|
| 560 |
+
"learning_rate": 9.878333333333333e-07,
|
| 561 |
+
"loss": 0.1583,
|
| 562 |
+
"step": 2225
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.72,
|
| 566 |
+
"learning_rate": 9.864444444444444e-07,
|
| 567 |
+
"loss": 0.1529,
|
| 568 |
+
"step": 2250
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.73,
|
| 572 |
+
"learning_rate": 9.850555555555555e-07,
|
| 573 |
+
"loss": 0.1455,
|
| 574 |
+
"step": 2275
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 0.74,
|
| 578 |
+
"learning_rate": 9.836666666666666e-07,
|
| 579 |
+
"loss": 0.1498,
|
| 580 |
+
"step": 2300
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.74,
|
| 584 |
+
"learning_rate": 9.822777777777777e-07,
|
| 585 |
+
"loss": 0.1464,
|
| 586 |
+
"step": 2325
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.75,
|
| 590 |
+
"learning_rate": 9.808888888888888e-07,
|
| 591 |
+
"loss": 0.1486,
|
| 592 |
+
"step": 2350
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.76,
|
| 596 |
+
"learning_rate": 9.795e-07,
|
| 597 |
+
"loss": 0.1449,
|
| 598 |
+
"step": 2375
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.77,
|
| 602 |
+
"learning_rate": 9.781111111111112e-07,
|
| 603 |
+
"loss": 0.1439,
|
| 604 |
+
"step": 2400
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.78,
|
| 608 |
+
"learning_rate": 9.767222222222223e-07,
|
| 609 |
+
"loss": 0.1454,
|
| 610 |
+
"step": 2425
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.78,
|
| 614 |
+
"learning_rate": 9.753333333333334e-07,
|
| 615 |
+
"loss": 0.1285,
|
| 616 |
+
"step": 2450
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.79,
|
| 620 |
+
"learning_rate": 9.739444444444443e-07,
|
| 621 |
+
"loss": 0.1495,
|
| 622 |
+
"step": 2475
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.8,
|
| 626 |
+
"learning_rate": 9.725555555555554e-07,
|
| 627 |
+
"loss": 0.1487,
|
| 628 |
+
"step": 2500
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 0.81,
|
| 632 |
+
"learning_rate": 9.711666666666665e-07,
|
| 633 |
+
"loss": 0.1284,
|
| 634 |
+
"step": 2525
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.82,
|
| 638 |
+
"learning_rate": 9.697777777777776e-07,
|
| 639 |
+
"loss": 0.14,
|
| 640 |
+
"step": 2550
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.82,
|
| 644 |
+
"learning_rate": 9.68388888888889e-07,
|
| 645 |
+
"loss": 0.137,
|
| 646 |
+
"step": 2575
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.83,
|
| 650 |
+
"learning_rate": 9.67e-07,
|
| 651 |
+
"loss": 0.1532,
|
| 652 |
+
"step": 2600
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"epoch": 0.84,
|
| 656 |
+
"learning_rate": 9.656111111111111e-07,
|
| 657 |
+
"loss": 0.1293,
|
| 658 |
+
"step": 2625
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 0.85,
|
| 662 |
+
"learning_rate": 9.642222222222222e-07,
|
| 663 |
+
"loss": 0.1386,
|
| 664 |
+
"step": 2650
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.86,
|
| 668 |
+
"learning_rate": 9.628333333333333e-07,
|
| 669 |
+
"loss": 0.1492,
|
| 670 |
+
"step": 2675
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.86,
|
| 674 |
+
"learning_rate": 9.614444444444444e-07,
|
| 675 |
+
"loss": 0.1413,
|
| 676 |
+
"step": 2700
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.87,
|
| 680 |
+
"learning_rate": 9.600555555555555e-07,
|
| 681 |
+
"loss": 0.1394,
|
| 682 |
+
"step": 2725
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.88,
|
| 686 |
+
"learning_rate": 9.586666666666666e-07,
|
| 687 |
+
"loss": 0.1377,
|
| 688 |
+
"step": 2750
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.89,
|
| 692 |
+
"learning_rate": 9.572777777777777e-07,
|
| 693 |
+
"loss": 0.1467,
|
| 694 |
+
"step": 2775
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.9,
|
| 698 |
+
"learning_rate": 9.558888888888888e-07,
|
| 699 |
+
"loss": 0.1341,
|
| 700 |
+
"step": 2800
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 0.9,
|
| 704 |
+
"learning_rate": 9.545e-07,
|
| 705 |
+
"loss": 0.1387,
|
| 706 |
+
"step": 2825
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.91,
|
| 710 |
+
"learning_rate": 9.531111111111111e-07,
|
| 711 |
+
"loss": 0.1358,
|
| 712 |
+
"step": 2850
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 0.92,
|
| 716 |
+
"learning_rate": 9.517222222222222e-07,
|
| 717 |
+
"loss": 0.1346,
|
| 718 |
+
"step": 2875
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.93,
|
| 722 |
+
"learning_rate": 9.503333333333333e-07,
|
| 723 |
+
"loss": 0.1438,
|
| 724 |
+
"step": 2900
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.94,
|
| 728 |
+
"learning_rate": 9.489444444444444e-07,
|
| 729 |
+
"loss": 0.149,
|
| 730 |
+
"step": 2925
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.94,
|
| 734 |
+
"learning_rate": 9.475555555555555e-07,
|
| 735 |
+
"loss": 0.138,
|
| 736 |
+
"step": 2950
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 0.95,
|
| 740 |
+
"learning_rate": 9.461666666666666e-07,
|
| 741 |
+
"loss": 0.1292,
|
| 742 |
+
"step": 2975
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 0.96,
|
| 746 |
+
"learning_rate": 9.447777777777777e-07,
|
| 747 |
+
"loss": 0.1312,
|
| 748 |
+
"step": 3000
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.96,
|
| 752 |
+
"eval_cer": 1.1932973663422914,
|
| 753 |
+
"eval_loss": 0.13624157011508942,
|
| 754 |
+
"eval_runtime": 939.634,
|
| 755 |
+
"eval_samples_per_second": 1.064,
|
| 756 |
+
"eval_steps_per_second": 0.133,
|
| 757 |
+
"eval_wer": 3.9324147641556872,
|
| 758 |
+
"step": 3000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.97,
|
| 762 |
+
"learning_rate": 9.433888888888889e-07,
|
| 763 |
+
"loss": 0.1414,
|
| 764 |
+
"step": 3025
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.98,
|
| 768 |
+
"learning_rate": 9.419999999999999e-07,
|
| 769 |
+
"loss": 0.1446,
|
| 770 |
+
"step": 3050
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.98,
|
| 774 |
+
"learning_rate": 9.40611111111111e-07,
|
| 775 |
+
"loss": 0.129,
|
| 776 |
+
"step": 3075
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.99,
|
| 780 |
+
"learning_rate": 9.392222222222221e-07,
|
| 781 |
+
"loss": 0.1378,
|
| 782 |
+
"step": 3100
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 1.0,
|
| 786 |
+
"learning_rate": 9.378333333333332e-07,
|
| 787 |
+
"loss": 0.1277,
|
| 788 |
+
"step": 3125
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.01,
|
| 792 |
+
"learning_rate": 9.364444444444443e-07,
|
| 793 |
+
"loss": 0.1301,
|
| 794 |
+
"step": 3150
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.02,
|
| 798 |
+
"learning_rate": 9.350555555555554e-07,
|
| 799 |
+
"loss": 0.1193,
|
| 800 |
+
"step": 3175
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.02,
|
| 804 |
+
"learning_rate": 9.336666666666666e-07,
|
| 805 |
+
"loss": 0.1223,
|
| 806 |
+
"step": 3200
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.03,
|
| 810 |
+
"learning_rate": 9.322777777777778e-07,
|
| 811 |
+
"loss": 0.1233,
|
| 812 |
+
"step": 3225
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 1.04,
|
| 816 |
+
"learning_rate": 9.308888888888889e-07,
|
| 817 |
+
"loss": 0.1176,
|
| 818 |
+
"step": 3250
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.05,
|
| 822 |
+
"learning_rate": 9.295e-07,
|
| 823 |
+
"loss": 0.1145,
|
| 824 |
+
"step": 3275
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.06,
|
| 828 |
+
"learning_rate": 9.281111111111111e-07,
|
| 829 |
+
"loss": 0.1225,
|
| 830 |
+
"step": 3300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.06,
|
| 834 |
+
"learning_rate": 9.267222222222222e-07,
|
| 835 |
+
"loss": 0.1251,
|
| 836 |
+
"step": 3325
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.07,
|
| 840 |
+
"learning_rate": 9.253333333333333e-07,
|
| 841 |
+
"loss": 0.1257,
|
| 842 |
+
"step": 3350
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.08,
|
| 846 |
+
"learning_rate": 9.239444444444445e-07,
|
| 847 |
+
"loss": 0.1075,
|
| 848 |
+
"step": 3375
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 1.09,
|
| 852 |
+
"learning_rate": 9.225555555555556e-07,
|
| 853 |
+
"loss": 0.122,
|
| 854 |
+
"step": 3400
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 1.1,
|
| 858 |
+
"learning_rate": 9.211666666666667e-07,
|
| 859 |
+
"loss": 0.1245,
|
| 860 |
+
"step": 3425
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.1,
|
| 864 |
+
"learning_rate": 9.197777777777778e-07,
|
| 865 |
+
"loss": 0.122,
|
| 866 |
+
"step": 3450
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.11,
|
| 870 |
+
"learning_rate": 9.183888888888889e-07,
|
| 871 |
+
"loss": 0.1219,
|
| 872 |
+
"step": 3475
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.12,
|
| 876 |
+
"learning_rate": 9.17e-07,
|
| 877 |
+
"loss": 0.1053,
|
| 878 |
+
"step": 3500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 1.13,
|
| 882 |
+
"learning_rate": 9.156111111111111e-07,
|
| 883 |
+
"loss": 0.1197,
|
| 884 |
+
"step": 3525
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.14,
|
| 888 |
+
"learning_rate": 9.142222222222223e-07,
|
| 889 |
+
"loss": 0.1162,
|
| 890 |
+
"step": 3550
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.14,
|
| 894 |
+
"learning_rate": 9.128333333333334e-07,
|
| 895 |
+
"loss": 0.1185,
|
| 896 |
+
"step": 3575
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 1.15,
|
| 900 |
+
"learning_rate": 9.114444444444444e-07,
|
| 901 |
+
"loss": 0.1148,
|
| 902 |
+
"step": 3600
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 1.16,
|
| 906 |
+
"learning_rate": 9.100555555555555e-07,
|
| 907 |
+
"loss": 0.1117,
|
| 908 |
+
"step": 3625
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 1.17,
|
| 912 |
+
"learning_rate": 9.086666666666666e-07,
|
| 913 |
+
"loss": 0.1211,
|
| 914 |
+
"step": 3650
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 1.18,
|
| 918 |
+
"learning_rate": 9.072777777777777e-07,
|
| 919 |
+
"loss": 0.1055,
|
| 920 |
+
"step": 3675
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 1.18,
|
| 924 |
+
"learning_rate": 9.058888888888888e-07,
|
| 925 |
+
"loss": 0.1113,
|
| 926 |
+
"step": 3700
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.19,
|
| 930 |
+
"learning_rate": 9.045e-07,
|
| 931 |
+
"loss": 0.1125,
|
| 932 |
+
"step": 3725
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.2,
|
| 936 |
+
"learning_rate": 9.031111111111111e-07,
|
| 937 |
+
"loss": 0.1146,
|
| 938 |
+
"step": 3750
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 1.21,
|
| 942 |
+
"learning_rate": 9.017222222222222e-07,
|
| 943 |
+
"loss": 0.1293,
|
| 944 |
+
"step": 3775
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 1.22,
|
| 948 |
+
"learning_rate": 9.003333333333333e-07,
|
| 949 |
+
"loss": 0.1306,
|
| 950 |
+
"step": 3800
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 1.22,
|
| 954 |
+
"learning_rate": 8.989444444444444e-07,
|
| 955 |
+
"loss": 0.107,
|
| 956 |
+
"step": 3825
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.23,
|
| 960 |
+
"learning_rate": 8.975555555555555e-07,
|
| 961 |
+
"loss": 0.1194,
|
| 962 |
+
"step": 3850
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.24,
|
| 966 |
+
"learning_rate": 8.961666666666666e-07,
|
| 967 |
+
"loss": 0.112,
|
| 968 |
+
"step": 3875
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 1.25,
|
| 972 |
+
"learning_rate": 8.947777777777778e-07,
|
| 973 |
+
"loss": 0.1173,
|
| 974 |
+
"step": 3900
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.26,
|
| 978 |
+
"learning_rate": 8.933888888888889e-07,
|
| 979 |
+
"loss": 0.1193,
|
| 980 |
+
"step": 3925
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 1.26,
|
| 984 |
+
"learning_rate": 8.92e-07,
|
| 985 |
+
"loss": 0.1148,
|
| 986 |
+
"step": 3950
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.27,
|
| 990 |
+
"learning_rate": 8.906111111111111e-07,
|
| 991 |
+
"loss": 0.1154,
|
| 992 |
+
"step": 3975
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.28,
|
| 996 |
+
"learning_rate": 8.892222222222222e-07,
|
| 997 |
+
"loss": 0.1206,
|
| 998 |
+
"step": 4000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.28,
|
| 1002 |
+
"eval_cer": 1.174809660666566,
|
| 1003 |
+
"eval_loss": 0.13329146802425385,
|
| 1004 |
+
"eval_runtime": 941.0652,
|
| 1005 |
+
"eval_samples_per_second": 1.063,
|
| 1006 |
+
"eval_steps_per_second": 0.133,
|
| 1007 |
+
"eval_wer": 3.8519561500553157,
|
| 1008 |
+
"step": 4000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.29,
|
| 1012 |
+
"learning_rate": 8.878333333333333e-07,
|
| 1013 |
+
"loss": 0.1156,
|
| 1014 |
+
"step": 4025
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.3,
|
| 1018 |
+
"learning_rate": 8.864444444444444e-07,
|
| 1019 |
+
"loss": 0.1223,
|
| 1020 |
+
"step": 4050
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.3,
|
| 1024 |
+
"learning_rate": 8.850555555555556e-07,
|
| 1025 |
+
"loss": 0.1176,
|
| 1026 |
+
"step": 4075
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.31,
|
| 1030 |
+
"learning_rate": 8.836666666666667e-07,
|
| 1031 |
+
"loss": 0.1134,
|
| 1032 |
+
"step": 4100
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.32,
|
| 1036 |
+
"learning_rate": 8.822777777777778e-07,
|
| 1037 |
+
"loss": 0.1195,
|
| 1038 |
+
"step": 4125
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 1.33,
|
| 1042 |
+
"learning_rate": 8.808888888888889e-07,
|
| 1043 |
+
"loss": 0.1131,
|
| 1044 |
+
"step": 4150
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.34,
|
| 1048 |
+
"learning_rate": 8.794999999999999e-07,
|
| 1049 |
+
"loss": 0.1251,
|
| 1050 |
+
"step": 4175
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.34,
|
| 1054 |
+
"learning_rate": 8.78111111111111e-07,
|
| 1055 |
+
"loss": 0.1102,
|
| 1056 |
+
"step": 4200
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 1.35,
|
| 1060 |
+
"learning_rate": 8.767222222222221e-07,
|
| 1061 |
+
"loss": 0.1223,
|
| 1062 |
+
"step": 4225
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 1.36,
|
| 1066 |
+
"learning_rate": 8.753333333333332e-07,
|
| 1067 |
+
"loss": 0.109,
|
| 1068 |
+
"step": 4250
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 1.37,
|
| 1072 |
+
"learning_rate": 8.739444444444444e-07,
|
| 1073 |
+
"loss": 0.1226,
|
| 1074 |
+
"step": 4275
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.38,
|
| 1078 |
+
"learning_rate": 8.725555555555555e-07,
|
| 1079 |
+
"loss": 0.1201,
|
| 1080 |
+
"step": 4300
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.38,
|
| 1084 |
+
"learning_rate": 8.711666666666666e-07,
|
| 1085 |
+
"loss": 0.1126,
|
| 1086 |
+
"step": 4325
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.39,
|
| 1090 |
+
"learning_rate": 8.697777777777777e-07,
|
| 1091 |
+
"loss": 0.113,
|
| 1092 |
+
"step": 4350
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 1.4,
|
| 1096 |
+
"learning_rate": 8.683888888888888e-07,
|
| 1097 |
+
"loss": 0.1132,
|
| 1098 |
+
"step": 4375
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 1.41,
|
| 1102 |
+
"learning_rate": 8.669999999999999e-07,
|
| 1103 |
+
"loss": 0.103,
|
| 1104 |
+
"step": 4400
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.42,
|
| 1108 |
+
"learning_rate": 8.656111111111111e-07,
|
| 1109 |
+
"loss": 0.1235,
|
| 1110 |
+
"step": 4425
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 1.42,
|
| 1114 |
+
"learning_rate": 8.642222222222222e-07,
|
| 1115 |
+
"loss": 0.1189,
|
| 1116 |
+
"step": 4450
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 1.43,
|
| 1120 |
+
"learning_rate": 8.628333333333333e-07,
|
| 1121 |
+
"loss": 0.1303,
|
| 1122 |
+
"step": 4475
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.44,
|
| 1126 |
+
"learning_rate": 8.614444444444444e-07,
|
| 1127 |
+
"loss": 0.1121,
|
| 1128 |
+
"step": 4500
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.45,
|
| 1132 |
+
"learning_rate": 8.600555555555555e-07,
|
| 1133 |
+
"loss": 0.1172,
|
| 1134 |
+
"step": 4525
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.46,
|
| 1138 |
+
"learning_rate": 8.586666666666666e-07,
|
| 1139 |
+
"loss": 0.1214,
|
| 1140 |
+
"step": 4550
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 1.46,
|
| 1144 |
+
"learning_rate": 8.572777777777777e-07,
|
| 1145 |
+
"loss": 0.1172,
|
| 1146 |
+
"step": 4575
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 1.47,
|
| 1150 |
+
"learning_rate": 8.558888888888889e-07,
|
| 1151 |
+
"loss": 0.1124,
|
| 1152 |
+
"step": 4600
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 1.48,
|
| 1156 |
+
"learning_rate": 8.545e-07,
|
| 1157 |
+
"loss": 0.116,
|
| 1158 |
+
"step": 4625
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 1.49,
|
| 1162 |
+
"learning_rate": 8.531111111111111e-07,
|
| 1163 |
+
"loss": 0.1147,
|
| 1164 |
+
"step": 4650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.5,
|
| 1168 |
+
"learning_rate": 8.517222222222222e-07,
|
| 1169 |
+
"loss": 0.1209,
|
| 1170 |
+
"step": 4675
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.5,
|
| 1174 |
+
"learning_rate": 8.503333333333333e-07,
|
| 1175 |
+
"loss": 0.1176,
|
| 1176 |
+
"step": 4700
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.51,
|
| 1180 |
+
"learning_rate": 8.489444444444443e-07,
|
| 1181 |
+
"loss": 0.1139,
|
| 1182 |
+
"step": 4725
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.52,
|
| 1186 |
+
"learning_rate": 8.475555555555554e-07,
|
| 1187 |
+
"loss": 0.1031,
|
| 1188 |
+
"step": 4750
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 1.53,
|
| 1192 |
+
"learning_rate": 8.461666666666665e-07,
|
| 1193 |
+
"loss": 0.1179,
|
| 1194 |
+
"step": 4775
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.54,
|
| 1198 |
+
"learning_rate": 8.447777777777777e-07,
|
| 1199 |
+
"loss": 0.1177,
|
| 1200 |
+
"step": 4800
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.54,
|
| 1204 |
+
"learning_rate": 8.433888888888888e-07,
|
| 1205 |
+
"loss": 0.1054,
|
| 1206 |
+
"step": 4825
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.55,
|
| 1210 |
+
"learning_rate": 8.419999999999999e-07,
|
| 1211 |
+
"loss": 0.1184,
|
| 1212 |
+
"step": 4850
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.56,
|
| 1216 |
+
"learning_rate": 8.40611111111111e-07,
|
| 1217 |
+
"loss": 0.1149,
|
| 1218 |
+
"step": 4875
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.57,
|
| 1222 |
+
"learning_rate": 8.392222222222221e-07,
|
| 1223 |
+
"loss": 0.1085,
|
| 1224 |
+
"step": 4900
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.58,
|
| 1228 |
+
"learning_rate": 8.378333333333332e-07,
|
| 1229 |
+
"loss": 0.1158,
|
| 1230 |
+
"step": 4925
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.58,
|
| 1234 |
+
"learning_rate": 8.364444444444443e-07,
|
| 1235 |
+
"loss": 0.1168,
|
| 1236 |
+
"step": 4950
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 1.59,
|
| 1240 |
+
"learning_rate": 8.350555555555556e-07,
|
| 1241 |
+
"loss": 0.1085,
|
| 1242 |
+
"step": 4975
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.6,
|
| 1246 |
+
"learning_rate": 8.336666666666667e-07,
|
| 1247 |
+
"loss": 0.1143,
|
| 1248 |
+
"step": 5000
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.6,
|
| 1252 |
+
"eval_cer": 1.0571606245483118,
|
| 1253 |
+
"eval_loss": 0.13207247853279114,
|
| 1254 |
+
"eval_runtime": 940.9316,
|
| 1255 |
+
"eval_samples_per_second": 1.063,
|
| 1256 |
+
"eval_steps_per_second": 0.133,
|
| 1257 |
+
"eval_wer": 3.6508096148043854,
|
| 1258 |
+
"step": 5000
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 1.61,
|
| 1262 |
+
"learning_rate": 8.322777777777778e-07,
|
| 1263 |
+
"loss": 0.1176,
|
| 1264 |
+
"step": 5025
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 1.62,
|
| 1268 |
+
"learning_rate": 8.308888888888889e-07,
|
| 1269 |
+
"loss": 0.1303,
|
| 1270 |
+
"step": 5050
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.62,
|
| 1274 |
+
"learning_rate": 8.295e-07,
|
| 1275 |
+
"loss": 0.1226,
|
| 1276 |
+
"step": 5075
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 1.63,
|
| 1280 |
+
"learning_rate": 8.281111111111111e-07,
|
| 1281 |
+
"loss": 0.1189,
|
| 1282 |
+
"step": 5100
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 1.64,
|
| 1286 |
+
"learning_rate": 8.267222222222222e-07,
|
| 1287 |
+
"loss": 0.118,
|
| 1288 |
+
"step": 5125
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 1.65,
|
| 1292 |
+
"learning_rate": 8.253333333333334e-07,
|
| 1293 |
+
"loss": 0.1264,
|
| 1294 |
+
"step": 5150
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 1.66,
|
| 1298 |
+
"learning_rate": 8.239444444444445e-07,
|
| 1299 |
+
"loss": 0.1162,
|
| 1300 |
+
"step": 5175
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 1.66,
|
| 1304 |
+
"learning_rate": 8.225555555555556e-07,
|
| 1305 |
+
"loss": 0.1081,
|
| 1306 |
+
"step": 5200
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.67,
|
| 1310 |
+
"learning_rate": 8.211666666666667e-07,
|
| 1311 |
+
"loss": 0.1205,
|
| 1312 |
+
"step": 5225
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 1.68,
|
| 1316 |
+
"learning_rate": 8.197777777777778e-07,
|
| 1317 |
+
"loss": 0.1191,
|
| 1318 |
+
"step": 5250
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 1.69,
|
| 1322 |
+
"learning_rate": 8.183888888888889e-07,
|
| 1323 |
+
"loss": 0.1166,
|
| 1324 |
+
"step": 5275
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 1.7,
|
| 1328 |
+
"learning_rate": 8.169999999999999e-07,
|
| 1329 |
+
"loss": 0.1173,
|
| 1330 |
+
"step": 5300
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 1.7,
|
| 1334 |
+
"learning_rate": 8.156111111111111e-07,
|
| 1335 |
+
"loss": 0.1138,
|
| 1336 |
+
"step": 5325
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 1.71,
|
| 1340 |
+
"learning_rate": 8.142222222222222e-07,
|
| 1341 |
+
"loss": 0.1261,
|
| 1342 |
+
"step": 5350
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 1.72,
|
| 1346 |
+
"learning_rate": 8.128333333333333e-07,
|
| 1347 |
+
"loss": 0.1213,
|
| 1348 |
+
"step": 5375
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 1.73,
|
| 1352 |
+
"learning_rate": 8.114444444444444e-07,
|
| 1353 |
+
"loss": 0.1097,
|
| 1354 |
+
"step": 5400
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.74,
|
| 1358 |
+
"learning_rate": 8.100555555555555e-07,
|
| 1359 |
+
"loss": 0.1169,
|
| 1360 |
+
"step": 5425
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.74,
|
| 1364 |
+
"learning_rate": 8.086666666666666e-07,
|
| 1365 |
+
"loss": 0.1145,
|
| 1366 |
+
"step": 5450
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 1.75,
|
| 1370 |
+
"learning_rate": 8.072777777777777e-07,
|
| 1371 |
+
"loss": 0.1165,
|
| 1372 |
+
"step": 5475
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 1.76,
|
| 1376 |
+
"learning_rate": 8.058888888888889e-07,
|
| 1377 |
+
"loss": 0.122,
|
| 1378 |
+
"step": 5500
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 1.77,
|
| 1382 |
+
"learning_rate": 8.045e-07,
|
| 1383 |
+
"loss": 0.1215,
|
| 1384 |
+
"step": 5525
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 1.78,
|
| 1388 |
+
"learning_rate": 8.031111111111111e-07,
|
| 1389 |
+
"loss": 0.1172,
|
| 1390 |
+
"step": 5550
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 1.78,
|
| 1394 |
+
"learning_rate": 8.017777777777778e-07,
|
| 1395 |
+
"loss": 0.1072,
|
| 1396 |
+
"step": 5575
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.79,
|
| 1400 |
+
"learning_rate": 8.003888888888889e-07,
|
| 1401 |
+
"loss": 0.1198,
|
| 1402 |
+
"step": 5600
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 1.8,
|
| 1406 |
+
"learning_rate": 7.99e-07,
|
| 1407 |
+
"loss": 0.1232,
|
| 1408 |
+
"step": 5625
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 1.81,
|
| 1412 |
+
"learning_rate": 7.976111111111111e-07,
|
| 1413 |
+
"loss": 0.1184,
|
| 1414 |
+
"step": 5650
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 1.82,
|
| 1418 |
+
"learning_rate": 7.962222222222222e-07,
|
| 1419 |
+
"loss": 0.1163,
|
| 1420 |
+
"step": 5675
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.82,
|
| 1424 |
+
"learning_rate": 7.948333333333333e-07,
|
| 1425 |
+
"loss": 0.1181,
|
| 1426 |
+
"step": 5700
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 1.83,
|
| 1430 |
+
"learning_rate": 7.934444444444444e-07,
|
| 1431 |
+
"loss": 0.1135,
|
| 1432 |
+
"step": 5725
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 1.84,
|
| 1436 |
+
"learning_rate": 7.920555555555555e-07,
|
| 1437 |
+
"loss": 0.1195,
|
| 1438 |
+
"step": 5750
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.85,
|
| 1442 |
+
"learning_rate": 7.906666666666666e-07,
|
| 1443 |
+
"loss": 0.1118,
|
| 1444 |
+
"step": 5775
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 1.86,
|
| 1448 |
+
"learning_rate": 7.892777777777777e-07,
|
| 1449 |
+
"loss": 0.1131,
|
| 1450 |
+
"step": 5800
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.86,
|
| 1454 |
+
"learning_rate": 7.878888888888888e-07,
|
| 1455 |
+
"loss": 0.1154,
|
| 1456 |
+
"step": 5825
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 1.87,
|
| 1460 |
+
"learning_rate": 7.864999999999999e-07,
|
| 1461 |
+
"loss": 0.1136,
|
| 1462 |
+
"step": 5850
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 1.88,
|
| 1466 |
+
"learning_rate": 7.851111111111111e-07,
|
| 1467 |
+
"loss": 0.1097,
|
| 1468 |
+
"step": 5875
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"epoch": 1.89,
|
| 1472 |
+
"learning_rate": 7.837222222222222e-07,
|
| 1473 |
+
"loss": 0.113,
|
| 1474 |
+
"step": 5900
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 1.9,
|
| 1478 |
+
"learning_rate": 7.823333333333333e-07,
|
| 1479 |
+
"loss": 0.1179,
|
| 1480 |
+
"step": 5925
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 1.9,
|
| 1484 |
+
"learning_rate": 7.809444444444444e-07,
|
| 1485 |
+
"loss": 0.1118,
|
| 1486 |
+
"step": 5950
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 1.91,
|
| 1490 |
+
"learning_rate": 7.795555555555555e-07,
|
| 1491 |
+
"loss": 0.1151,
|
| 1492 |
+
"step": 5975
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 1.92,
|
| 1496 |
+
"learning_rate": 7.781666666666666e-07,
|
| 1497 |
+
"loss": 0.1202,
|
| 1498 |
+
"step": 6000
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 1.92,
|
| 1502 |
+
"eval_cer": 1.1311114472512143,
|
| 1503 |
+
"eval_loss": 0.12914104759693146,
|
| 1504 |
+
"eval_runtime": 942.1178,
|
| 1505 |
+
"eval_samples_per_second": 1.061,
|
| 1506 |
+
"eval_steps_per_second": 0.133,
|
| 1507 |
+
"eval_wer": 3.8016695162425824,
|
| 1508 |
+
"step": 6000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 1.93,
|
| 1512 |
+
"learning_rate": 7.767777777777777e-07,
|
| 1513 |
+
"loss": 0.1144,
|
| 1514 |
+
"step": 6025
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 1.94,
|
| 1518 |
+
"learning_rate": 7.753888888888889e-07,
|
| 1519 |
+
"loss": 0.1077,
|
| 1520 |
+
"step": 6050
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.94,
|
| 1524 |
+
"learning_rate": 7.74e-07,
|
| 1525 |
+
"loss": 0.1064,
|
| 1526 |
+
"step": 6075
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 1.95,
|
| 1530 |
+
"learning_rate": 7.726111111111111e-07,
|
| 1531 |
+
"loss": 0.1191,
|
| 1532 |
+
"step": 6100
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 1.96,
|
| 1536 |
+
"learning_rate": 7.712222222222222e-07,
|
| 1537 |
+
"loss": 0.1176,
|
| 1538 |
+
"step": 6125
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 1.97,
|
| 1542 |
+
"learning_rate": 7.698333333333333e-07,
|
| 1543 |
+
"loss": 0.112,
|
| 1544 |
+
"step": 6150
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 1.98,
|
| 1548 |
+
"learning_rate": 7.684444444444444e-07,
|
| 1549 |
+
"loss": 0.1103,
|
| 1550 |
+
"step": 6175
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 1.98,
|
| 1554 |
+
"learning_rate": 7.670555555555555e-07,
|
| 1555 |
+
"loss": 0.1189,
|
| 1556 |
+
"step": 6200
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 1.99,
|
| 1560 |
+
"learning_rate": 7.656666666666667e-07,
|
| 1561 |
+
"loss": 0.1124,
|
| 1562 |
+
"step": 6225
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.0,
|
| 1566 |
+
"learning_rate": 7.642777777777777e-07,
|
| 1567 |
+
"loss": 0.1103,
|
| 1568 |
+
"step": 6250
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 2.01,
|
| 1572 |
+
"learning_rate": 7.628888888888888e-07,
|
| 1573 |
+
"loss": 0.0959,
|
| 1574 |
+
"step": 6275
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 2.02,
|
| 1578 |
+
"learning_rate": 7.614999999999999e-07,
|
| 1579 |
+
"loss": 0.0909,
|
| 1580 |
+
"step": 6300
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 2.02,
|
| 1584 |
+
"learning_rate": 7.60111111111111e-07,
|
| 1585 |
+
"loss": 0.0894,
|
| 1586 |
+
"step": 6325
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 2.03,
|
| 1590 |
+
"learning_rate": 7.587222222222221e-07,
|
| 1591 |
+
"loss": 0.0895,
|
| 1592 |
+
"step": 6350
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 2.04,
|
| 1596 |
+
"learning_rate": 7.573333333333332e-07,
|
| 1597 |
+
"loss": 0.0957,
|
| 1598 |
+
"step": 6375
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 2.05,
|
| 1602 |
+
"learning_rate": 7.559444444444443e-07,
|
| 1603 |
+
"loss": 0.0949,
|
| 1604 |
+
"step": 6400
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 2.06,
|
| 1608 |
+
"learning_rate": 7.545555555555555e-07,
|
| 1609 |
+
"loss": 0.0966,
|
| 1610 |
+
"step": 6425
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 2.06,
|
| 1614 |
+
"learning_rate": 7.531666666666666e-07,
|
| 1615 |
+
"loss": 0.0903,
|
| 1616 |
+
"step": 6450
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 2.07,
|
| 1620 |
+
"learning_rate": 7.517777777777777e-07,
|
| 1621 |
+
"loss": 0.0943,
|
| 1622 |
+
"step": 6475
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.08,
|
| 1626 |
+
"learning_rate": 7.503888888888888e-07,
|
| 1627 |
+
"loss": 0.0923,
|
| 1628 |
+
"step": 6500
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 2.09,
|
| 1632 |
+
"learning_rate": 7.489999999999999e-07,
|
| 1633 |
+
"loss": 0.0958,
|
| 1634 |
+
"step": 6525
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 2.1,
|
| 1638 |
+
"learning_rate": 7.47611111111111e-07,
|
| 1639 |
+
"loss": 0.0867,
|
| 1640 |
+
"step": 6550
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 2.1,
|
| 1644 |
+
"learning_rate": 7.462222222222221e-07,
|
| 1645 |
+
"loss": 0.0901,
|
| 1646 |
+
"step": 6575
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 2.11,
|
| 1650 |
+
"learning_rate": 7.448333333333334e-07,
|
| 1651 |
+
"loss": 0.0947,
|
| 1652 |
+
"step": 6600
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 2.12,
|
| 1656 |
+
"learning_rate": 7.434444444444445e-07,
|
| 1657 |
+
"loss": 0.0897,
|
| 1658 |
+
"step": 6625
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 2.13,
|
| 1662 |
+
"learning_rate": 7.420555555555556e-07,
|
| 1663 |
+
"loss": 0.0908,
|
| 1664 |
+
"step": 6650
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 2.14,
|
| 1668 |
+
"learning_rate": 7.406666666666667e-07,
|
| 1669 |
+
"loss": 0.0945,
|
| 1670 |
+
"step": 6675
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 2.14,
|
| 1674 |
+
"learning_rate": 7.392777777777778e-07,
|
| 1675 |
+
"loss": 0.0867,
|
| 1676 |
+
"step": 6700
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 2.15,
|
| 1680 |
+
"learning_rate": 7.378888888888889e-07,
|
| 1681 |
+
"loss": 0.0979,
|
| 1682 |
+
"step": 6725
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.16,
|
| 1686 |
+
"learning_rate": 7.365e-07,
|
| 1687 |
+
"loss": 0.098,
|
| 1688 |
+
"step": 6750
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 2.17,
|
| 1692 |
+
"learning_rate": 7.351111111111112e-07,
|
| 1693 |
+
"loss": 0.0915,
|
| 1694 |
+
"step": 6775
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 2.18,
|
| 1698 |
+
"learning_rate": 7.337222222222223e-07,
|
| 1699 |
+
"loss": 0.092,
|
| 1700 |
+
"step": 6800
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 2.18,
|
| 1704 |
+
"learning_rate": 7.323333333333333e-07,
|
| 1705 |
+
"loss": 0.0888,
|
| 1706 |
+
"step": 6825
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 2.19,
|
| 1710 |
+
"learning_rate": 7.309444444444444e-07,
|
| 1711 |
+
"loss": 0.0872,
|
| 1712 |
+
"step": 6850
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 2.2,
|
| 1716 |
+
"learning_rate": 7.295555555555555e-07,
|
| 1717 |
+
"loss": 0.0892,
|
| 1718 |
+
"step": 6875
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 2.21,
|
| 1722 |
+
"learning_rate": 7.281666666666666e-07,
|
| 1723 |
+
"loss": 0.0808,
|
| 1724 |
+
"step": 6900
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.22,
|
| 1728 |
+
"learning_rate": 7.267777777777777e-07,
|
| 1729 |
+
"loss": 0.0965,
|
| 1730 |
+
"step": 6925
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 2.22,
|
| 1734 |
+
"learning_rate": 7.253888888888889e-07,
|
| 1735 |
+
"loss": 0.0879,
|
| 1736 |
+
"step": 6950
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 2.23,
|
| 1740 |
+
"learning_rate": 7.24e-07,
|
| 1741 |
+
"loss": 0.0915,
|
| 1742 |
+
"step": 6975
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 2.24,
|
| 1746 |
+
"learning_rate": 7.226111111111111e-07,
|
| 1747 |
+
"loss": 0.0856,
|
| 1748 |
+
"step": 7000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 2.24,
|
| 1752 |
+
"eval_cer": 1.0840518328039128,
|
| 1753 |
+
"eval_loss": 0.13248993456363678,
|
| 1754 |
+
"eval_runtime": 939.5436,
|
| 1755 |
+
"eval_samples_per_second": 1.064,
|
| 1756 |
+
"eval_steps_per_second": 0.133,
|
| 1757 |
+
"eval_wer": 3.7010962486171173,
|
| 1758 |
+
"step": 7000
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.25,
|
| 1762 |
+
"learning_rate": 7.212222222222222e-07,
|
| 1763 |
+
"loss": 0.0921,
|
| 1764 |
+
"step": 7025
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.26,
|
| 1768 |
+
"learning_rate": 7.198333333333333e-07,
|
| 1769 |
+
"loss": 0.0928,
|
| 1770 |
+
"step": 7050
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.26,
|
| 1774 |
+
"learning_rate": 7.184444444444444e-07,
|
| 1775 |
+
"loss": 0.0895,
|
| 1776 |
+
"step": 7075
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 2.27,
|
| 1780 |
+
"learning_rate": 7.170555555555555e-07,
|
| 1781 |
+
"loss": 0.0949,
|
| 1782 |
+
"step": 7100
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 2.28,
|
| 1786 |
+
"learning_rate": 7.156666666666667e-07,
|
| 1787 |
+
"loss": 0.09,
|
| 1788 |
+
"step": 7125
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.29,
|
| 1792 |
+
"learning_rate": 7.142777777777778e-07,
|
| 1793 |
+
"loss": 0.0794,
|
| 1794 |
+
"step": 7150
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.3,
|
| 1798 |
+
"learning_rate": 7.128888888888889e-07,
|
| 1799 |
+
"loss": 0.0945,
|
| 1800 |
+
"step": 7175
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 2.3,
|
| 1804 |
+
"learning_rate": 7.115e-07,
|
| 1805 |
+
"loss": 0.0906,
|
| 1806 |
+
"step": 7200
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.31,
|
| 1810 |
+
"learning_rate": 7.101111111111111e-07,
|
| 1811 |
+
"loss": 0.0972,
|
| 1812 |
+
"step": 7225
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 2.32,
|
| 1816 |
+
"learning_rate": 7.087222222222222e-07,
|
| 1817 |
+
"loss": 0.0922,
|
| 1818 |
+
"step": 7250
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.33,
|
| 1822 |
+
"learning_rate": 7.073333333333333e-07,
|
| 1823 |
+
"loss": 0.091,
|
| 1824 |
+
"step": 7275
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 2.34,
|
| 1828 |
+
"learning_rate": 7.059444444444445e-07,
|
| 1829 |
+
"loss": 0.0981,
|
| 1830 |
+
"step": 7300
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 2.34,
|
| 1834 |
+
"learning_rate": 7.045555555555556e-07,
|
| 1835 |
+
"loss": 0.0913,
|
| 1836 |
+
"step": 7325
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.35,
|
| 1840 |
+
"learning_rate": 7.031666666666667e-07,
|
| 1841 |
+
"loss": 0.0955,
|
| 1842 |
+
"step": 7350
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 2.36,
|
| 1846 |
+
"learning_rate": 7.017777777777777e-07,
|
| 1847 |
+
"loss": 0.0918,
|
| 1848 |
+
"step": 7375
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 2.37,
|
| 1852 |
+
"learning_rate": 7.003888888888888e-07,
|
| 1853 |
+
"loss": 0.0991,
|
| 1854 |
+
"step": 7400
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 2.38,
|
| 1858 |
+
"learning_rate": 6.989999999999999e-07,
|
| 1859 |
+
"loss": 0.0926,
|
| 1860 |
+
"step": 7425
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 2.38,
|
| 1864 |
+
"learning_rate": 6.97611111111111e-07,
|
| 1865 |
+
"loss": 0.0999,
|
| 1866 |
+
"step": 7450
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 2.39,
|
| 1870 |
+
"learning_rate": 6.962222222222222e-07,
|
| 1871 |
+
"loss": 0.0857,
|
| 1872 |
+
"step": 7475
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 2.4,
|
| 1876 |
+
"learning_rate": 6.948333333333333e-07,
|
| 1877 |
+
"loss": 0.0948,
|
| 1878 |
+
"step": 7500
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 2.41,
|
| 1882 |
+
"learning_rate": 6.934444444444444e-07,
|
| 1883 |
+
"loss": 0.0961,
|
| 1884 |
+
"step": 7525
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 2.42,
|
| 1888 |
+
"learning_rate": 6.920555555555555e-07,
|
| 1889 |
+
"loss": 0.0926,
|
| 1890 |
+
"step": 7550
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.42,
|
| 1894 |
+
"learning_rate": 6.906666666666666e-07,
|
| 1895 |
+
"loss": 0.0936,
|
| 1896 |
+
"step": 7575
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 2.43,
|
| 1900 |
+
"learning_rate": 6.892777777777777e-07,
|
| 1901 |
+
"loss": 0.089,
|
| 1902 |
+
"step": 7600
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 2.44,
|
| 1906 |
+
"learning_rate": 6.879444444444444e-07,
|
| 1907 |
+
"loss": 0.0931,
|
| 1908 |
+
"step": 7625
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 2.45,
|
| 1912 |
+
"learning_rate": 6.865555555555555e-07,
|
| 1913 |
+
"loss": 0.0931,
|
| 1914 |
+
"step": 7650
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 2.46,
|
| 1918 |
+
"learning_rate": 6.851666666666667e-07,
|
| 1919 |
+
"loss": 0.0924,
|
| 1920 |
+
"step": 7675
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 2.46,
|
| 1924 |
+
"learning_rate": 6.837777777777778e-07,
|
| 1925 |
+
"loss": 0.0904,
|
| 1926 |
+
"step": 7700
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 2.47,
|
| 1930 |
+
"learning_rate": 6.823888888888889e-07,
|
| 1931 |
+
"loss": 0.1014,
|
| 1932 |
+
"step": 7725
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 2.48,
|
| 1936 |
+
"learning_rate": 6.81e-07,
|
| 1937 |
+
"loss": 0.0919,
|
| 1938 |
+
"step": 7750
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 2.49,
|
| 1942 |
+
"learning_rate": 6.796111111111111e-07,
|
| 1943 |
+
"loss": 0.0939,
|
| 1944 |
+
"step": 7775
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.5,
|
| 1948 |
+
"learning_rate": 6.782222222222221e-07,
|
| 1949 |
+
"loss": 0.0941,
|
| 1950 |
+
"step": 7800
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 2.5,
|
| 1954 |
+
"learning_rate": 6.768333333333332e-07,
|
| 1955 |
+
"loss": 0.0847,
|
| 1956 |
+
"step": 7825
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 2.51,
|
| 1960 |
+
"learning_rate": 6.754444444444443e-07,
|
| 1961 |
+
"loss": 0.0895,
|
| 1962 |
+
"step": 7850
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.52,
|
| 1966 |
+
"learning_rate": 6.740555555555555e-07,
|
| 1967 |
+
"loss": 0.0973,
|
| 1968 |
+
"step": 7875
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 2.53,
|
| 1972 |
+
"learning_rate": 6.726666666666666e-07,
|
| 1973 |
+
"loss": 0.0916,
|
| 1974 |
+
"step": 7900
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.54,
|
| 1978 |
+
"learning_rate": 6.712777777777777e-07,
|
| 1979 |
+
"loss": 0.0934,
|
| 1980 |
+
"step": 7925
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.54,
|
| 1984 |
+
"learning_rate": 6.698888888888888e-07,
|
| 1985 |
+
"loss": 0.0956,
|
| 1986 |
+
"step": 7950
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 2.55,
|
| 1990 |
+
"learning_rate": 6.684999999999999e-07,
|
| 1991 |
+
"loss": 0.0901,
|
| 1992 |
+
"step": 7975
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 2.56,
|
| 1996 |
+
"learning_rate": 6.67111111111111e-07,
|
| 1997 |
+
"loss": 0.1005,
|
| 1998 |
+
"step": 8000
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 2.56,
|
| 2002 |
+
"eval_cer": 1.0554799240323367,
|
| 2003 |
+
"eval_loss": 0.13198651373386383,
|
| 2004 |
+
"eval_runtime": 941.8725,
|
| 2005 |
+
"eval_samples_per_second": 1.062,
|
| 2006 |
+
"eval_steps_per_second": 0.133,
|
| 2007 |
+
"eval_wer": 3.7010962486171173,
|
| 2008 |
+
"step": 8000
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 2.56,
|
| 2012 |
+
"step": 8000,
|
| 2013 |
+
"total_flos": 5.435275349827584e+20,
|
| 2014 |
+
"train_loss": 0.1500811086446047,
|
| 2015 |
+
"train_runtime": 62196.6088,
|
| 2016 |
+
"train_samples_per_second": 10.29,
|
| 2017 |
+
"train_steps_per_second": 0.322
|
| 2018 |
+
}
|
| 2019 |
+
],
|
| 2020 |
+
"max_steps": 20000,
|
| 2021 |
+
"num_train_epochs": 7,
|
| 2022 |
+
"total_flos": 5.435275349827584e+20,
|
| 2023 |
+
"trial_name": null,
|
| 2024 |
+
"trial_params": null
|
| 2025 |
+
}
|