Commit
•
079b1fd
1
Parent(s):
5f29eeb
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- pytorch_model.bin +1 -1
- train_results.json +8 -0
- trainer_state.json +1177 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.2650411128997803,
|
| 4 |
+
"eval_runtime": 208.9946,
|
| 5 |
+
"eval_samples": 4850,
|
| 6 |
+
"eval_samples_per_second": 23.206,
|
| 7 |
+
"eval_steps_per_second": 1.455,
|
| 8 |
+
"eval_wer": 0.27019870599341483,
|
| 9 |
+
"train_loss": 1.3658743798928183,
|
| 10 |
+
"train_runtime": 33116.2978,
|
| 11 |
+
"train_samples": 11695,
|
| 12 |
+
"train_samples_per_second": 17.657,
|
| 13 |
+
"train_steps_per_second": 0.553
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"eval_loss": 0.2650411128997803,
|
| 4 |
+
"eval_runtime": 208.9946,
|
| 5 |
+
"eval_samples": 4850,
|
| 6 |
+
"eval_samples_per_second": 23.206,
|
| 7 |
+
"eval_steps_per_second": 1.455,
|
| 8 |
+
"eval_wer": 0.27019870599341483
|
| 9 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262124593
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68c5c29efd8691197c1a6fedd29288acdbcd390bdcf91bc2bc706194e6965e76
|
| 3 |
size 1262124593
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 50.0,
|
| 3 |
+
"train_loss": 1.3658743798928183,
|
| 4 |
+
"train_runtime": 33116.2978,
|
| 5 |
+
"train_samples": 11695,
|
| 6 |
+
"train_samples_per_second": 17.657,
|
| 7 |
+
"train_steps_per_second": 0.553
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"global_step": 18300,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.27,
|
| 12 |
+
"learning_rate": 2.2866666666666662e-06,
|
| 13 |
+
"loss": 16.3914,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.55,
|
| 18 |
+
"learning_rate": 4.596666666666666e-06,
|
| 19 |
+
"loss": 10.3351,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.82,
|
| 24 |
+
"learning_rate": 6.93e-06,
|
| 25 |
+
"loss": 5.7909,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.09,
|
| 30 |
+
"learning_rate": 9.263333333333332e-06,
|
| 31 |
+
"loss": 4.7705,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.37,
|
| 36 |
+
"learning_rate": 1.1596666666666665e-05,
|
| 37 |
+
"loss": 4.1586,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.64,
|
| 42 |
+
"learning_rate": 1.3929999999999999e-05,
|
| 43 |
+
"loss": 3.7094,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.91,
|
| 48 |
+
"learning_rate": 1.626333333333333e-05,
|
| 49 |
+
"loss": 3.3499,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 2.19,
|
| 54 |
+
"learning_rate": 1.8596666666666664e-05,
|
| 55 |
+
"loss": 3.1161,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.46,
|
| 60 |
+
"learning_rate": 2.0929999999999998e-05,
|
| 61 |
+
"loss": 2.9926,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.73,
|
| 66 |
+
"learning_rate": 2.326333333333333e-05,
|
| 67 |
+
"loss": 2.9385,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 3.01,
|
| 72 |
+
"learning_rate": 2.559666666666666e-05,
|
| 73 |
+
"loss": 2.889,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 3.28,
|
| 78 |
+
"learning_rate": 2.793e-05,
|
| 79 |
+
"loss": 2.8583,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 3.55,
|
| 84 |
+
"learning_rate": 3.0263333333333332e-05,
|
| 85 |
+
"loss": 2.8445,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 3.83,
|
| 90 |
+
"learning_rate": 3.2596666666666666e-05,
|
| 91 |
+
"loss": 2.8281,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 4.1,
|
| 96 |
+
"learning_rate": 3.493e-05,
|
| 97 |
+
"loss": 2.8191,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 4.37,
|
| 102 |
+
"learning_rate": 3.7263333333333326e-05,
|
| 103 |
+
"loss": 2.7995,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.64,
|
| 108 |
+
"learning_rate": 3.959666666666666e-05,
|
| 109 |
+
"loss": 2.6081,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 4.92,
|
| 114 |
+
"learning_rate": 4.192999999999999e-05,
|
| 115 |
+
"loss": 2.0775,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 5.19,
|
| 120 |
+
"learning_rate": 4.426333333333333e-05,
|
| 121 |
+
"loss": 1.763,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 5.46,
|
| 126 |
+
"learning_rate": 4.659666666666666e-05,
|
| 127 |
+
"loss": 1.6346,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 5.74,
|
| 132 |
+
"learning_rate": 4.8929999999999994e-05,
|
| 133 |
+
"loss": 1.5687,
|
| 134 |
+
"step": 2100
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 6.01,
|
| 138 |
+
"learning_rate": 5.126333333333333e-05,
|
| 139 |
+
"loss": 1.5288,
|
| 140 |
+
"step": 2200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 6.28,
|
| 144 |
+
"learning_rate": 5.359666666666667e-05,
|
| 145 |
+
"loss": 1.4756,
|
| 146 |
+
"step": 2300
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 6.56,
|
| 150 |
+
"learning_rate": 5.593e-05,
|
| 151 |
+
"loss": 1.4463,
|
| 152 |
+
"step": 2400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 6.83,
|
| 156 |
+
"learning_rate": 5.826333333333333e-05,
|
| 157 |
+
"loss": 1.4291,
|
| 158 |
+
"step": 2500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 7.1,
|
| 162 |
+
"learning_rate": 6.059666666666666e-05,
|
| 163 |
+
"loss": 1.4114,
|
| 164 |
+
"step": 2600
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 7.38,
|
| 168 |
+
"learning_rate": 6.293e-05,
|
| 169 |
+
"loss": 1.3955,
|
| 170 |
+
"step": 2700
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 7.65,
|
| 174 |
+
"learning_rate": 6.526333333333332e-05,
|
| 175 |
+
"loss": 1.3691,
|
| 176 |
+
"step": 2800
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 7.92,
|
| 180 |
+
"learning_rate": 6.759666666666666e-05,
|
| 181 |
+
"loss": 1.3556,
|
| 182 |
+
"step": 2900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 8.2,
|
| 186 |
+
"learning_rate": 6.992999999999999e-05,
|
| 187 |
+
"loss": 1.3454,
|
| 188 |
+
"step": 3000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 8.2,
|
| 192 |
+
"eval_loss": 0.49262478947639465,
|
| 193 |
+
"eval_runtime": 209.1038,
|
| 194 |
+
"eval_samples_per_second": 23.194,
|
| 195 |
+
"eval_steps_per_second": 1.454,
|
| 196 |
+
"eval_wer": 0.5703299486541871,
|
| 197 |
+
"step": 3000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 8.47,
|
| 201 |
+
"learning_rate": 6.955620915032679e-05,
|
| 202 |
+
"loss": 1.326,
|
| 203 |
+
"step": 3100
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 8.74,
|
| 207 |
+
"learning_rate": 6.90986928104575e-05,
|
| 208 |
+
"loss": 1.318,
|
| 209 |
+
"step": 3200
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 9.02,
|
| 213 |
+
"learning_rate": 6.864117647058823e-05,
|
| 214 |
+
"loss": 1.3012,
|
| 215 |
+
"step": 3300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 9.29,
|
| 219 |
+
"learning_rate": 6.818366013071895e-05,
|
| 220 |
+
"loss": 1.2888,
|
| 221 |
+
"step": 3400
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 9.56,
|
| 225 |
+
"learning_rate": 6.772614379084966e-05,
|
| 226 |
+
"loss": 1.2874,
|
| 227 |
+
"step": 3500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 9.84,
|
| 231 |
+
"learning_rate": 6.726862745098038e-05,
|
| 232 |
+
"loss": 1.2699,
|
| 233 |
+
"step": 3600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 10.11,
|
| 237 |
+
"learning_rate": 6.681111111111111e-05,
|
| 238 |
+
"loss": 1.2558,
|
| 239 |
+
"step": 3700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 10.38,
|
| 243 |
+
"learning_rate": 6.635359477124183e-05,
|
| 244 |
+
"loss": 1.2512,
|
| 245 |
+
"step": 3800
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 10.66,
|
| 249 |
+
"learning_rate": 6.589607843137254e-05,
|
| 250 |
+
"loss": 1.2502,
|
| 251 |
+
"step": 3900
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 10.93,
|
| 255 |
+
"learning_rate": 6.543856209150327e-05,
|
| 256 |
+
"loss": 1.246,
|
| 257 |
+
"step": 4000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 11.2,
|
| 261 |
+
"learning_rate": 6.498104575163397e-05,
|
| 262 |
+
"loss": 1.2205,
|
| 263 |
+
"step": 4100
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 11.48,
|
| 267 |
+
"learning_rate": 6.45235294117647e-05,
|
| 268 |
+
"loss": 1.2192,
|
| 269 |
+
"step": 4200
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 11.75,
|
| 273 |
+
"learning_rate": 6.406601307189542e-05,
|
| 274 |
+
"loss": 1.2146,
|
| 275 |
+
"step": 4300
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 12.02,
|
| 279 |
+
"learning_rate": 6.360849673202614e-05,
|
| 280 |
+
"loss": 1.2143,
|
| 281 |
+
"step": 4400
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 12.3,
|
| 285 |
+
"learning_rate": 6.315098039215685e-05,
|
| 286 |
+
"loss": 1.205,
|
| 287 |
+
"step": 4500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 12.57,
|
| 291 |
+
"learning_rate": 6.269346405228757e-05,
|
| 292 |
+
"loss": 1.1777,
|
| 293 |
+
"step": 4600
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 12.84,
|
| 297 |
+
"learning_rate": 6.22359477124183e-05,
|
| 298 |
+
"loss": 1.1901,
|
| 299 |
+
"step": 4700
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 13.11,
|
| 303 |
+
"learning_rate": 6.177843137254901e-05,
|
| 304 |
+
"loss": 1.1729,
|
| 305 |
+
"step": 4800
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 13.39,
|
| 309 |
+
"learning_rate": 6.132091503267974e-05,
|
| 310 |
+
"loss": 1.1813,
|
| 311 |
+
"step": 4900
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 13.66,
|
| 315 |
+
"learning_rate": 6.086339869281045e-05,
|
| 316 |
+
"loss": 1.1773,
|
| 317 |
+
"step": 5000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 13.93,
|
| 321 |
+
"learning_rate": 6.040588235294117e-05,
|
| 322 |
+
"loss": 1.1517,
|
| 323 |
+
"step": 5100
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 14.21,
|
| 327 |
+
"learning_rate": 5.994836601307189e-05,
|
| 328 |
+
"loss": 1.1568,
|
| 329 |
+
"step": 5200
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 14.48,
|
| 333 |
+
"learning_rate": 5.9490849673202613e-05,
|
| 334 |
+
"loss": 1.1591,
|
| 335 |
+
"step": 5300
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 14.75,
|
| 339 |
+
"learning_rate": 5.903333333333333e-05,
|
| 340 |
+
"loss": 1.146,
|
| 341 |
+
"step": 5400
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 15.03,
|
| 345 |
+
"learning_rate": 5.8575816993464045e-05,
|
| 346 |
+
"loss": 1.1561,
|
| 347 |
+
"step": 5500
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 15.3,
|
| 351 |
+
"learning_rate": 5.811830065359477e-05,
|
| 352 |
+
"loss": 1.1304,
|
| 353 |
+
"step": 5600
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 15.57,
|
| 357 |
+
"learning_rate": 5.7660784313725484e-05,
|
| 358 |
+
"loss": 1.1355,
|
| 359 |
+
"step": 5700
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 15.85,
|
| 363 |
+
"learning_rate": 5.720326797385621e-05,
|
| 364 |
+
"loss": 1.137,
|
| 365 |
+
"step": 5800
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 16.12,
|
| 369 |
+
"learning_rate": 5.674575163398693e-05,
|
| 370 |
+
"loss": 1.1401,
|
| 371 |
+
"step": 5900
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 16.39,
|
| 375 |
+
"learning_rate": 5.628823529411764e-05,
|
| 376 |
+
"loss": 1.1202,
|
| 377 |
+
"step": 6000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 16.39,
|
| 381 |
+
"eval_loss": 0.35287711024284363,
|
| 382 |
+
"eval_runtime": 207.6346,
|
| 383 |
+
"eval_samples_per_second": 23.358,
|
| 384 |
+
"eval_steps_per_second": 1.464,
|
| 385 |
+
"eval_wer": 0.3944187331629481,
|
| 386 |
+
"step": 6000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 16.67,
|
| 390 |
+
"learning_rate": 5.583071895424836e-05,
|
| 391 |
+
"loss": 1.1132,
|
| 392 |
+
"step": 6100
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 16.94,
|
| 396 |
+
"learning_rate": 5.537777777777777e-05,
|
| 397 |
+
"loss": 1.1199,
|
| 398 |
+
"step": 6200
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 17.21,
|
| 402 |
+
"learning_rate": 5.4920261437908495e-05,
|
| 403 |
+
"loss": 1.1089,
|
| 404 |
+
"step": 6300
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 17.49,
|
| 408 |
+
"learning_rate": 5.446274509803921e-05,
|
| 409 |
+
"loss": 1.1038,
|
| 410 |
+
"step": 6400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 17.76,
|
| 414 |
+
"learning_rate": 5.400522875816993e-05,
|
| 415 |
+
"loss": 1.1022,
|
| 416 |
+
"step": 6500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 18.03,
|
| 420 |
+
"learning_rate": 5.354771241830065e-05,
|
| 421 |
+
"loss": 1.1059,
|
| 422 |
+
"step": 6600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 18.31,
|
| 426 |
+
"learning_rate": 5.3090196078431366e-05,
|
| 427 |
+
"loss": 1.097,
|
| 428 |
+
"step": 6700
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 18.58,
|
| 432 |
+
"learning_rate": 5.263267973856209e-05,
|
| 433 |
+
"loss": 1.0793,
|
| 434 |
+
"step": 6800
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 18.85,
|
| 438 |
+
"learning_rate": 5.217516339869281e-05,
|
| 439 |
+
"loss": 1.0814,
|
| 440 |
+
"step": 6900
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 19.13,
|
| 444 |
+
"learning_rate": 5.171764705882352e-05,
|
| 445 |
+
"loss": 1.087,
|
| 446 |
+
"step": 7000
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 19.4,
|
| 450 |
+
"learning_rate": 5.126013071895424e-05,
|
| 451 |
+
"loss": 1.0784,
|
| 452 |
+
"step": 7100
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 19.67,
|
| 456 |
+
"learning_rate": 5.0802614379084966e-05,
|
| 457 |
+
"loss": 1.0618,
|
| 458 |
+
"step": 7200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 19.95,
|
| 462 |
+
"learning_rate": 5.034509803921568e-05,
|
| 463 |
+
"loss": 1.069,
|
| 464 |
+
"step": 7300
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 20.22,
|
| 468 |
+
"learning_rate": 4.9887581699346405e-05,
|
| 469 |
+
"loss": 1.0555,
|
| 470 |
+
"step": 7400
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 20.49,
|
| 474 |
+
"learning_rate": 4.9430065359477114e-05,
|
| 475 |
+
"loss": 1.0559,
|
| 476 |
+
"step": 7500
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 20.77,
|
| 480 |
+
"learning_rate": 4.897254901960784e-05,
|
| 481 |
+
"loss": 1.0545,
|
| 482 |
+
"step": 7600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 21.04,
|
| 486 |
+
"learning_rate": 4.851503267973856e-05,
|
| 487 |
+
"loss": 1.0644,
|
| 488 |
+
"step": 7700
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 21.31,
|
| 492 |
+
"learning_rate": 4.8057516339869276e-05,
|
| 493 |
+
"loss": 1.0541,
|
| 494 |
+
"step": 7800
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 21.58,
|
| 498 |
+
"learning_rate": 4.76e-05,
|
| 499 |
+
"loss": 1.0467,
|
| 500 |
+
"step": 7900
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 21.86,
|
| 504 |
+
"learning_rate": 4.7142483660130714e-05,
|
| 505 |
+
"loss": 1.0309,
|
| 506 |
+
"step": 8000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 22.13,
|
| 510 |
+
"learning_rate": 4.668496732026143e-05,
|
| 511 |
+
"loss": 1.0453,
|
| 512 |
+
"step": 8100
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 22.4,
|
| 516 |
+
"learning_rate": 4.622745098039215e-05,
|
| 517 |
+
"loss": 1.0321,
|
| 518 |
+
"step": 8200
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 22.68,
|
| 522 |
+
"learning_rate": 4.5769934640522876e-05,
|
| 523 |
+
"loss": 1.0439,
|
| 524 |
+
"step": 8300
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 22.95,
|
| 528 |
+
"learning_rate": 4.5316993464052286e-05,
|
| 529 |
+
"loss": 1.0403,
|
| 530 |
+
"step": 8400
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 23.22,
|
| 534 |
+
"learning_rate": 4.485947712418301e-05,
|
| 535 |
+
"loss": 1.0145,
|
| 536 |
+
"step": 8500
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 23.5,
|
| 540 |
+
"learning_rate": 4.440196078431372e-05,
|
| 541 |
+
"loss": 1.0228,
|
| 542 |
+
"step": 8600
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 23.77,
|
| 546 |
+
"learning_rate": 4.394444444444444e-05,
|
| 547 |
+
"loss": 1.0236,
|
| 548 |
+
"step": 8700
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 24.04,
|
| 552 |
+
"learning_rate": 4.348692810457516e-05,
|
| 553 |
+
"loss": 1.0185,
|
| 554 |
+
"step": 8800
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 24.32,
|
| 558 |
+
"learning_rate": 4.302941176470588e-05,
|
| 559 |
+
"loss": 1.0165,
|
| 560 |
+
"step": 8900
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 24.59,
|
| 564 |
+
"learning_rate": 4.2571895424836596e-05,
|
| 565 |
+
"loss": 1.0058,
|
| 566 |
+
"step": 9000
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 24.59,
|
| 570 |
+
"eval_loss": 0.31432652473449707,
|
| 571 |
+
"eval_runtime": 204.5409,
|
| 572 |
+
"eval_samples_per_second": 23.712,
|
| 573 |
+
"eval_steps_per_second": 1.486,
|
| 574 |
+
"eval_wer": 0.3340931592641201,
|
| 575 |
+
"step": 9000
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 24.86,
|
| 579 |
+
"learning_rate": 4.211437908496731e-05,
|
| 580 |
+
"loss": 1.0177,
|
| 581 |
+
"step": 9100
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 25.14,
|
| 585 |
+
"learning_rate": 4.1656862745098035e-05,
|
| 586 |
+
"loss": 1.0175,
|
| 587 |
+
"step": 9200
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 25.41,
|
| 591 |
+
"learning_rate": 4.119934640522876e-05,
|
| 592 |
+
"loss": 1.0046,
|
| 593 |
+
"step": 9300
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 25.68,
|
| 597 |
+
"learning_rate": 4.0741830065359473e-05,
|
| 598 |
+
"loss": 0.9922,
|
| 599 |
+
"step": 9400
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 25.96,
|
| 603 |
+
"learning_rate": 4.028431372549019e-05,
|
| 604 |
+
"loss": 1.0011,
|
| 605 |
+
"step": 9500
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 26.23,
|
| 609 |
+
"learning_rate": 3.982679738562091e-05,
|
| 610 |
+
"loss": 0.9852,
|
| 611 |
+
"step": 9600
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 26.5,
|
| 615 |
+
"learning_rate": 3.936928104575163e-05,
|
| 616 |
+
"loss": 0.9901,
|
| 617 |
+
"step": 9700
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 26.78,
|
| 621 |
+
"learning_rate": 3.891176470588235e-05,
|
| 622 |
+
"loss": 0.9848,
|
| 623 |
+
"step": 9800
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 27.05,
|
| 627 |
+
"learning_rate": 3.8454248366013074e-05,
|
| 628 |
+
"loss": 1.0035,
|
| 629 |
+
"step": 9900
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 27.32,
|
| 633 |
+
"learning_rate": 3.799673202614378e-05,
|
| 634 |
+
"loss": 0.9707,
|
| 635 |
+
"step": 10000
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 27.6,
|
| 639 |
+
"learning_rate": 3.7539215686274506e-05,
|
| 640 |
+
"loss": 0.9894,
|
| 641 |
+
"step": 10100
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 27.87,
|
| 645 |
+
"learning_rate": 3.708169934640523e-05,
|
| 646 |
+
"loss": 0.9873,
|
| 647 |
+
"step": 10200
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 28.14,
|
| 651 |
+
"learning_rate": 3.6624183006535944e-05,
|
| 652 |
+
"loss": 0.9902,
|
| 653 |
+
"step": 10300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 28.42,
|
| 657 |
+
"learning_rate": 3.616666666666667e-05,
|
| 658 |
+
"loss": 0.97,
|
| 659 |
+
"step": 10400
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 28.69,
|
| 663 |
+
"learning_rate": 3.570915032679738e-05,
|
| 664 |
+
"loss": 0.9613,
|
| 665 |
+
"step": 10500
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 28.96,
|
| 669 |
+
"learning_rate": 3.52516339869281e-05,
|
| 670 |
+
"loss": 0.9734,
|
| 671 |
+
"step": 10600
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 29.23,
|
| 675 |
+
"learning_rate": 3.479411764705882e-05,
|
| 676 |
+
"loss": 0.9514,
|
| 677 |
+
"step": 10700
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 29.51,
|
| 681 |
+
"learning_rate": 3.434117647058823e-05,
|
| 682 |
+
"loss": 0.9566,
|
| 683 |
+
"step": 10800
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 29.78,
|
| 687 |
+
"learning_rate": 3.388366013071895e-05,
|
| 688 |
+
"loss": 0.9632,
|
| 689 |
+
"step": 10900
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 30.05,
|
| 693 |
+
"learning_rate": 3.343071895424836e-05,
|
| 694 |
+
"loss": 0.9635,
|
| 695 |
+
"step": 11000
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 30.33,
|
| 699 |
+
"learning_rate": 3.297320261437908e-05,
|
| 700 |
+
"loss": 0.9581,
|
| 701 |
+
"step": 11100
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 30.6,
|
| 705 |
+
"learning_rate": 3.2515686274509805e-05,
|
| 706 |
+
"loss": 0.9545,
|
| 707 |
+
"step": 11200
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 30.87,
|
| 711 |
+
"learning_rate": 3.205816993464052e-05,
|
| 712 |
+
"loss": 0.9501,
|
| 713 |
+
"step": 11300
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 31.15,
|
| 717 |
+
"learning_rate": 3.1600653594771244e-05,
|
| 718 |
+
"loss": 0.9506,
|
| 719 |
+
"step": 11400
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 31.42,
|
| 723 |
+
"learning_rate": 3.114313725490196e-05,
|
| 724 |
+
"loss": 0.9414,
|
| 725 |
+
"step": 11500
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 31.69,
|
| 729 |
+
"learning_rate": 3.0685620915032675e-05,
|
| 730 |
+
"loss": 0.9484,
|
| 731 |
+
"step": 11600
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 31.97,
|
| 735 |
+
"learning_rate": 3.0228104575163395e-05,
|
| 736 |
+
"loss": 0.9474,
|
| 737 |
+
"step": 11700
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 32.24,
|
| 741 |
+
"learning_rate": 2.9770588235294114e-05,
|
| 742 |
+
"loss": 0.9237,
|
| 743 |
+
"step": 11800
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 32.51,
|
| 747 |
+
"learning_rate": 2.9313071895424837e-05,
|
| 748 |
+
"loss": 0.9376,
|
| 749 |
+
"step": 11900
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 32.79,
|
| 753 |
+
"learning_rate": 2.8855555555555553e-05,
|
| 754 |
+
"loss": 0.9287,
|
| 755 |
+
"step": 12000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 32.79,
|
| 759 |
+
"eval_loss": 0.28960883617401123,
|
| 760 |
+
"eval_runtime": 208.0073,
|
| 761 |
+
"eval_samples_per_second": 23.316,
|
| 762 |
+
"eval_steps_per_second": 1.461,
|
| 763 |
+
"eval_wer": 0.29796689000944027,
|
| 764 |
+
"step": 12000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 33.06,
|
| 768 |
+
"learning_rate": 2.8398039215686272e-05,
|
| 769 |
+
"loss": 0.9325,
|
| 770 |
+
"step": 12100
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 33.33,
|
| 774 |
+
"learning_rate": 2.794052287581699e-05,
|
| 775 |
+
"loss": 0.9349,
|
| 776 |
+
"step": 12200
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 33.61,
|
| 780 |
+
"learning_rate": 2.748300653594771e-05,
|
| 781 |
+
"loss": 0.9342,
|
| 782 |
+
"step": 12300
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 33.88,
|
| 786 |
+
"learning_rate": 2.702549019607843e-05,
|
| 787 |
+
"loss": 0.9282,
|
| 788 |
+
"step": 12400
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 34.15,
|
| 792 |
+
"learning_rate": 2.6567973856209146e-05,
|
| 793 |
+
"loss": 0.9338,
|
| 794 |
+
"step": 12500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 34.43,
|
| 798 |
+
"learning_rate": 2.611045751633987e-05,
|
| 799 |
+
"loss": 0.9195,
|
| 800 |
+
"step": 12600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 34.7,
|
| 804 |
+
"learning_rate": 2.5652941176470585e-05,
|
| 805 |
+
"loss": 0.9146,
|
| 806 |
+
"step": 12700
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 34.97,
|
| 810 |
+
"learning_rate": 2.5195424836601305e-05,
|
| 811 |
+
"loss": 0.9249,
|
| 812 |
+
"step": 12800
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 35.25,
|
| 816 |
+
"learning_rate": 2.4737908496732024e-05,
|
| 817 |
+
"loss": 0.9137,
|
| 818 |
+
"step": 12900
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 35.52,
|
| 822 |
+
"learning_rate": 2.4280392156862743e-05,
|
| 823 |
+
"loss": 0.9147,
|
| 824 |
+
"step": 13000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 35.79,
|
| 828 |
+
"learning_rate": 2.3822875816993463e-05,
|
| 829 |
+
"loss": 0.9171,
|
| 830 |
+
"step": 13100
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 36.07,
|
| 834 |
+
"learning_rate": 2.336535947712418e-05,
|
| 835 |
+
"loss": 0.9178,
|
| 836 |
+
"step": 13200
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 36.34,
|
| 840 |
+
"learning_rate": 2.29078431372549e-05,
|
| 841 |
+
"loss": 0.9164,
|
| 842 |
+
"step": 13300
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 36.61,
|
| 846 |
+
"learning_rate": 2.2450326797385617e-05,
|
| 847 |
+
"loss": 0.909,
|
| 848 |
+
"step": 13400
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 36.89,
|
| 852 |
+
"learning_rate": 2.1997385620915028e-05,
|
| 853 |
+
"loss": 0.9106,
|
| 854 |
+
"step": 13500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 37.16,
|
| 858 |
+
"learning_rate": 2.153986928104575e-05,
|
| 859 |
+
"loss": 0.9059,
|
| 860 |
+
"step": 13600
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 37.43,
|
| 864 |
+
"learning_rate": 2.1082352941176467e-05,
|
| 865 |
+
"loss": 0.899,
|
| 866 |
+
"step": 13700
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 37.7,
|
| 870 |
+
"learning_rate": 2.0629411764705878e-05,
|
| 871 |
+
"loss": 0.9044,
|
| 872 |
+
"step": 13800
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 37.98,
|
| 876 |
+
"learning_rate": 2.01718954248366e-05,
|
| 877 |
+
"loss": 0.9037,
|
| 878 |
+
"step": 13900
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 38.25,
|
| 882 |
+
"learning_rate": 1.971437908496732e-05,
|
| 883 |
+
"loss": 0.9013,
|
| 884 |
+
"step": 14000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 38.52,
|
| 888 |
+
"learning_rate": 1.9256862745098036e-05,
|
| 889 |
+
"loss": 0.8963,
|
| 890 |
+
"step": 14100
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 38.8,
|
| 894 |
+
"learning_rate": 1.879934640522876e-05,
|
| 895 |
+
"loss": 0.8998,
|
| 896 |
+
"step": 14200
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 39.07,
|
| 900 |
+
"learning_rate": 1.8341830065359474e-05,
|
| 901 |
+
"loss": 0.8995,
|
| 902 |
+
"step": 14300
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 39.34,
|
| 906 |
+
"learning_rate": 1.7884313725490194e-05,
|
| 907 |
+
"loss": 0.9125,
|
| 908 |
+
"step": 14400
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 39.62,
|
| 912 |
+
"learning_rate": 1.7426797385620913e-05,
|
| 913 |
+
"loss": 0.8982,
|
| 914 |
+
"step": 14500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 39.89,
|
| 918 |
+
"learning_rate": 1.6969281045751633e-05,
|
| 919 |
+
"loss": 0.8959,
|
| 920 |
+
"step": 14600
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 40.16,
|
| 924 |
+
"learning_rate": 1.6511764705882352e-05,
|
| 925 |
+
"loss": 0.8826,
|
| 926 |
+
"step": 14700
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 40.44,
|
| 930 |
+
"learning_rate": 1.605424836601307e-05,
|
| 931 |
+
"loss": 0.8871,
|
| 932 |
+
"step": 14800
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 40.71,
|
| 936 |
+
"learning_rate": 1.559673202614379e-05,
|
| 937 |
+
"loss": 0.8815,
|
| 938 |
+
"step": 14900
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 40.98,
|
| 942 |
+
"learning_rate": 1.5139215686274508e-05,
|
| 943 |
+
"loss": 0.8849,
|
| 944 |
+
"step": 15000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 40.98,
|
| 948 |
+
"eval_loss": 0.2726878821849823,
|
| 949 |
+
"eval_runtime": 205.4369,
|
| 950 |
+
"eval_samples_per_second": 23.608,
|
| 951 |
+
"eval_steps_per_second": 1.48,
|
| 952 |
+
"eval_wer": 0.27975409269876356,
|
| 953 |
+
"step": 15000
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 41.26,
|
| 957 |
+
"learning_rate": 1.4681699346405228e-05,
|
| 958 |
+
"loss": 0.8804,
|
| 959 |
+
"step": 15100
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 41.53,
|
| 963 |
+
"learning_rate": 1.4224183006535945e-05,
|
| 964 |
+
"loss": 0.8861,
|
| 965 |
+
"step": 15200
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 41.8,
|
| 969 |
+
"learning_rate": 1.3766666666666665e-05,
|
| 970 |
+
"loss": 0.8913,
|
| 971 |
+
"step": 15300
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 42.08,
|
| 975 |
+
"learning_rate": 1.3309150326797386e-05,
|
| 976 |
+
"loss": 0.8769,
|
| 977 |
+
"step": 15400
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 42.35,
|
| 981 |
+
"learning_rate": 1.2851633986928103e-05,
|
| 982 |
+
"loss": 0.885,
|
| 983 |
+
"step": 15500
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 42.62,
|
| 987 |
+
"learning_rate": 1.2394117647058823e-05,
|
| 988 |
+
"loss": 0.8788,
|
| 989 |
+
"step": 15600
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 42.9,
|
| 993 |
+
"learning_rate": 1.1936601307189542e-05,
|
| 994 |
+
"loss": 0.8752,
|
| 995 |
+
"step": 15700
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 43.17,
|
| 999 |
+
"learning_rate": 1.147908496732026e-05,
|
| 1000 |
+
"loss": 0.869,
|
| 1001 |
+
"step": 15800
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 43.44,
|
| 1005 |
+
"learning_rate": 1.102156862745098e-05,
|
| 1006 |
+
"loss": 0.8773,
|
| 1007 |
+
"step": 15900
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 43.72,
|
| 1011 |
+
"learning_rate": 1.0564052287581699e-05,
|
| 1012 |
+
"loss": 0.8757,
|
| 1013 |
+
"step": 16000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 43.99,
|
| 1017 |
+
"learning_rate": 1.0106535947712418e-05,
|
| 1018 |
+
"loss": 0.8693,
|
| 1019 |
+
"step": 16100
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 44.26,
|
| 1023 |
+
"learning_rate": 9.649019607843137e-06,
|
| 1024 |
+
"loss": 0.863,
|
| 1025 |
+
"step": 16200
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 44.54,
|
| 1029 |
+
"learning_rate": 9.191503267973855e-06,
|
| 1030 |
+
"loss": 0.8634,
|
| 1031 |
+
"step": 16300
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 44.81,
|
| 1035 |
+
"learning_rate": 8.733986928104574e-06,
|
| 1036 |
+
"loss": 0.8549,
|
| 1037 |
+
"step": 16400
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 45.08,
|
| 1041 |
+
"learning_rate": 8.276470588235294e-06,
|
| 1042 |
+
"loss": 0.8528,
|
| 1043 |
+
"step": 16500
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 45.36,
|
| 1047 |
+
"learning_rate": 7.818954248366013e-06,
|
| 1048 |
+
"loss": 0.861,
|
| 1049 |
+
"step": 16600
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 45.63,
|
| 1053 |
+
"learning_rate": 7.361437908496731e-06,
|
| 1054 |
+
"loss": 0.8571,
|
| 1055 |
+
"step": 16700
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 45.9,
|
| 1059 |
+
"learning_rate": 6.903921568627451e-06,
|
| 1060 |
+
"loss": 0.8595,
|
| 1061 |
+
"step": 16800
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 46.17,
|
| 1065 |
+
"learning_rate": 6.44640522875817e-06,
|
| 1066 |
+
"loss": 0.8605,
|
| 1067 |
+
"step": 16900
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 46.45,
|
| 1071 |
+
"learning_rate": 5.988888888888888e-06,
|
| 1072 |
+
"loss": 0.8635,
|
| 1073 |
+
"step": 17000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 46.72,
|
| 1077 |
+
"learning_rate": 5.5313725490196075e-06,
|
| 1078 |
+
"loss": 0.8594,
|
| 1079 |
+
"step": 17100
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 46.99,
|
| 1083 |
+
"learning_rate": 5.073856209150326e-06,
|
| 1084 |
+
"loss": 0.8662,
|
| 1085 |
+
"step": 17200
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 47.27,
|
| 1089 |
+
"learning_rate": 4.616339869281045e-06,
|
| 1090 |
+
"loss": 0.8569,
|
| 1091 |
+
"step": 17300
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 47.54,
|
| 1095 |
+
"learning_rate": 4.158823529411764e-06,
|
| 1096 |
+
"loss": 0.8583,
|
| 1097 |
+
"step": 17400
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 47.81,
|
| 1101 |
+
"learning_rate": 3.7013071895424833e-06,
|
| 1102 |
+
"loss": 0.8585,
|
| 1103 |
+
"step": 17500
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 48.09,
|
| 1107 |
+
"learning_rate": 3.2437908496732026e-06,
|
| 1108 |
+
"loss": 0.8439,
|
| 1109 |
+
"step": 17600
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 48.36,
|
| 1113 |
+
"learning_rate": 2.786274509803921e-06,
|
| 1114 |
+
"loss": 0.8487,
|
| 1115 |
+
"step": 17700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 48.63,
|
| 1119 |
+
"learning_rate": 2.3287581699346405e-06,
|
| 1120 |
+
"loss": 0.8459,
|
| 1121 |
+
"step": 17800
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 48.91,
|
| 1125 |
+
"learning_rate": 1.8712418300653593e-06,
|
| 1126 |
+
"loss": 0.8576,
|
| 1127 |
+
"step": 17900
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 49.18,
|
| 1131 |
+
"learning_rate": 1.4137254901960782e-06,
|
| 1132 |
+
"loss": 0.8665,
|
| 1133 |
+
"step": 18000
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 49.18,
|
| 1137 |
+
"eval_loss": 0.2661794126033783,
|
| 1138 |
+
"eval_runtime": 208.8108,
|
| 1139 |
+
"eval_samples_per_second": 23.227,
|
| 1140 |
+
"eval_steps_per_second": 1.456,
|
| 1141 |
+
"eval_wer": 0.269577030231862,
|
| 1142 |
+
"step": 18000
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 49.45,
|
| 1146 |
+
"learning_rate": 9.562091503267974e-07,
|
| 1147 |
+
"loss": 0.8569,
|
| 1148 |
+
"step": 18100
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 49.73,
|
| 1152 |
+
"learning_rate": 4.986928104575163e-07,
|
| 1153 |
+
"loss": 0.8505,
|
| 1154 |
+
"step": 18200
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 50.0,
|
| 1158 |
+
"learning_rate": 4.117647058823529e-08,
|
| 1159 |
+
"loss": 0.863,
|
| 1160 |
+
"step": 18300
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 50.0,
|
| 1164 |
+
"step": 18300,
|
| 1165 |
+
"total_flos": 9.169878518581522e+19,
|
| 1166 |
+
"train_loss": 1.3658743798928183,
|
| 1167 |
+
"train_runtime": 33116.2978,
|
| 1168 |
+
"train_samples_per_second": 17.657,
|
| 1169 |
+
"train_steps_per_second": 0.553
|
| 1170 |
+
}
|
| 1171 |
+
],
|
| 1172 |
+
"max_steps": 18300,
|
| 1173 |
+
"num_train_epochs": 50,
|
| 1174 |
+
"total_flos": 9.169878518581522e+19,
|
| 1175 |
+
"trial_name": null,
|
| 1176 |
+
"trial_params": null
|
| 1177 |
+
}
|